Are Data Centers Finally Ready For DC Power?

1sockchuck writes "It's been five years since a landmark study outlined the potential benefits of DC power distribution in data centers. But adoption of DC in data centers remains limited, even as the industry aggressively pursues a wide array of other energy savings strategies. Advocates of DC distribution are hoping a new study will jump start the conversation about DC distribution, which can save energy by eliminating several wasteful AC-to-DC conversions within a data center. Meanwhile, an industry association for DC power adoption, the EMerge Alliance, has formed a new technical standards committee for data centers, and is advancing a 380-volt DC power standard. Will DC distribution ever gain momentum in data centers?"

Comments

Edison reaching out from beyond the grave

[ravenspear]

I told you bitches I would prevail one day!

Re:Edison reaching out from beyond the grave

[rolfwind]

Not really. AC was the answer to how to transport electricity long distances.

Currently, it is still converted to DC in a huge amount of devices, in the computer at the PSU. Few devices use AC iirc, something like a fan/ceiling fan probably has an AC motor because a DC motor would slice your finger off if you decided to play with the blades. So, the question then just remains how to optimize the point of conversion. It's rather like the electric car-fossil-fuel-electric-plant/gasoline car debate: have a bunch of small inefficient combustion engines or a large efficient one but deal with transport losses (along with a bunch of other issues).

In this case, just where along the line do you convert the AC to DC. Since DC can't really travel far at all without significant losses, I guess that would be at the rack level?

Re:Edison reaching out from beyond the grave

[Wonko the Sane]

AC was the answer to how to transport electricity long distances

AC was used because using transformers to convert between voltage levels was more efficient than motor-generators and solid state electronics hadn't been invented yet. All other things being equal, DC is always more efficient than AC for long distance transmission.

Re:Edison reaching out from beyond the grave

[Elder Entropist]

Very high voltage was the answer to how to transport electricity long distances. AC was the answer to how to convert that high voltage to safer/useful low voltages cheaply. Very high voltage DC can lose less power over distance than AC. On smaller/cheaper wires too due to the AC skin effect.

Re:Edison reaching out from beyond the grave

[Mr Z]

The problem with DC back in Edison's day was that you couldn't easily step it up or down. DC doesn't have higher losses than AC at the same voltage. In fact, DC radiates less energy away than AC does, and is therefore more efficient.

Ohmic losses all come down to I^2 * R. R is the resistance of the cable, and I is current. To deliver a given amount of power, you have to have a certain V*I. To reduce Ohmic losses, then, you have to reduce the amount of current, which means going up in voltage.

Incidentally, that's also what's driving automobile manufacturers toward 48v instead of 12v, since it would cut the current from the battery by a factor of 4, thereby reducing the amount of loss in the wiring by a factor of 16. That means you can use smaller wires to deliver the same amount of power, safely.

Re:Edison reaching out from beyond the grave

[khallow]

Since DC can't really travel far at all without significant losses, I guess that would be at the rack level?

Transmission losses are actually less than for AC. They don't lose energy to inductance with nearby conducting loops and impedance losses are about the same as for a three phase line with same RMS voltage. The real problems are conversion to and from AC, and the fact that DC operates at a much lower voltage (low voltage results in high losses, whether AC or DC) when in actual servers.

The idea behind DC powered centers is that the AC to DC conversion is done in one place, away from the servers so that a) the heating load of the center is lower, and b) it can be done in one place with a relatively efficient converter rather than in a thousand places with less efficient designs. The accompanying baggage as I gather is that you're either running a lot of power-losing low voltage lines or doing some sort of power-losing DC voltage step down inside the center.

Re:Edison reaching out from beyond the grave

[Anonymous Coward]

Edison was a hack. Most of "his" good ideas were from Tesla, without attribution.

However, like in most things, both Westinghouse and Edison/Tesla were correct. DC makes more sense for short distances, but AC makes more sense for longer distances. Local DC power makes more sense in a datacenter than pretty much anywhere else. I wouldn't be surprised that most other 'industrial' processes already utilize in-house AC/DC conversion. IT is just behind due to the depth of the problem of moving over to DC - there is a huge supply chain which needs to change, first, and then you've got a LOT of old AC equipment you've got to phase out.

Re:Edison reaching out from beyond the grave

[Anonymous Coward]

Way to take a completely normal, productive conversation and shit all over it.

Try being less of an asshole for a day or two. You might find that people hate you less.

Re:Edison reaching out from beyond the grave

[Anonymous Coward]

It's not only the radiation that makes AC less efficent. It is also how you can build conductors.

If you built just a simple very long cylinder out of copper, it is the perfect conductor for DC. For AC only the border is used, as the electric field presses electrons there. So with AC you have to use complex cables working around this, while with DC you get better behaviour with a simpler and less expensive design.

You won't get that for small voltages. But for the big power lines going long distances, this is significant. (And the reason why newer lines are DC nowadays, in the rare cases where there are newer lines).

Re:Edison reaching out from beyond the grave

[Anonymous Coward]

Tesla replying "Yep if you need to move electricity 100 feet across an indoor facility you are the best but I move it across miles and counties and countries"....bitches indeed ;P

Re:Edison reaching out from beyond the grave

[Anonymous Coward]

a DC motor would slice your finger off if you decided to play with the blades.

Really? Why? I've accidentally stopped computer fans without losing fingers several times, and they run on DC.

Re:Edison reaching out from beyond the grave

[Anonymous Coward]

I believe that this is on his tombstone.

"DC 4 ever Bitches!"

Re:Edison reaching out from beyond the grave

[EETech1]

That's why they use 380 volts! One big splice goes to all the 12 volt stuff, then another splice comes off of that splice to do the 5 volt stuff. It is not run through regulators, it just happens automatically due to the superior characteristics of DC power! They also tap into the ground wire at various places to get the -5 and -12. Magic I tell ya!

It sees it's best efficiencies running near 100% utilization through so you want to plan your workloads accordingly, or you risk watching your $#!+ let out it's magic smoke! So all in all it should drive down the price of "the cloud" by forcing competition!

Win, Win!

Re:Edison reaching out from beyond the grave

[Anonymous Coward]

I sliced by thumb pretty bad accidentally sticking it into the blade of a 120x38mm Panaflo CPU fan.

Re:Edison reaching out from beyond the grave

[Relayman]

At 60 Hz, the skin effect is virtually nonexistent.

Re:Edison reaching out from beyond the grave

[Anne Thwacks]

The skin effect is only significant at frequencies WAY above 60Hz, so this is not relevant. AC is used because transformers can convert huge amounts of power from one voltage to another reliably, efficiently and cheaply.

Its true you can convert voltages DC to DC using electronics, but reliably at 132kV? It aint easy, cheap and certainly neither if you want efficiency.

(Blow up a 132kV IGBT, then come back and tell me about the damage :-)

Re:Edison reaching out from beyond the grave

[budgenator]

The typical computer power supply and almost everything else anymore uses a rather efficient switching power supply, it takes the input power, rectifies it, runs it through a transformer by switching the DC on and off at high frequency, then rectifying the output from the transformer to supply the rest of the computer. The power supplies change from 115 VAC to 230 VAC input by a simple switch, the 230 VAC , which has a peak to peak voltage of 325 volts, just a stones throw from 380 VDC. The only reason that switching power supplies aren't used on DC is because one side of the input rectifier bridge will probably burn out from not having a rest cycle to cool off. Beef up that bridge rectifier and the power supply will not care what the input frequency is. It would be trivial to convert power supplies to 380 VDC.

Re:Edison reaching out from beyond the grave

[afidel]

That's only true with very expensive solid state components that didn't exist when the decision to go AC was made and it also ignores the fact that DC branching is extremely complex so it only works when you are moving power from generation site to large single use site, a wider grid is much harder to do with DC.

Re:Edison reaching out from beyond the grave

[icebike]

Why would you need a rectifier bridge on DC at all?

Re:Edison reaching out from beyond the grave

[garlicbready]

When it comes to high voltages one of the other advantages AC has is safety
with High Voltage DC the muscles in your hand would tend to grip / contract or hold onto a conductor
AC on the other-hand just tends to throw you off as it's alternating back and forth at 50 or 60 times a second
The longer your in contact the more chance you have of suffering burns, or your heart stopping

I'd guess 12V / 24V is probably the best, as it's low enough not to be a safety risk (think car battery)
and at the same time most PC Hardware tends to run on a combination of 12V / 5V anyway (converting from 12 to 5 is trivial)
Having an AC to DC conversion on the side of the wall then running a cable to each rack perhaps in a ring configuration might be the way to go
probably depends on the amount / size / thickness of cabling you can fit in to each rack / under the flooring

When they talk about losses, usually it's over long distance over a few miles or between substations
although I have heard that you can get losses of a few volts of AC between opposite sides of a hanger for example
but that's likely to be less controlled / more dirty than regulated DC within a Datacenter
Also with 12V there may be a lot less cost in terms of replacing the server PSU's as it's already closer to what you already need

Of course the most important thing is that we'd need some form of new standard plug
perhaps a different colour and some extra pins for future expansion
perhaps google should ask apple, I hear they're good with plug design :)

Re:Edison reaching out from beyond the grave

[mirix]

They are used on DC all the time, the input rectifier/doubler is dropped, but everything else is the same.

They often go by the name DC-DC converter, though. So really, a computer SMPS is a DC-DC SMPS, with a rectifier on the front to make the DC.

The closest thing to a SMPS that actually runs on AC is the thyristor fired rigs, can't recall how you call that topology now. Technology is very similar to that of a light dimmer, it fires on part of the wave and passes current for that portion of the wave... but then it's rectified anyway, for use in the device. (note, these are awful for power factor).

Re:Edison reaching out from beyond the grave

[Anonymous Coward]

Wow I was always taught that AC was more efficient over long distances, but after reading up on it you are definitely correct. TIL something new.

http://en.wikipedia.org/wiki/Electric_power_transmission
"As of 1980, the longest cost-effective distance for DC electricity was determined to be 7,000 km (4,300 mi). For AC it was 4,000 km (2,500 mi), "

Re:Edison reaching out from beyond the grave

[budgenator]

1. so you don't get a nasty surprise when some idiot manages to wire or push a plug in backwards
2. you can use the same supply on both 125VAC, 240VAC and 380VDC, they'll cost the same to make but Probably charge more for the 380VDC model

Re:Edison reaching out from beyond the grave

[Anonymous Coward]

no at 60hz the skin depth is 8.4mm! it is already pretty thin conductor.

Re:Edison reaching out from beyond the grave

[colinrichardday]

Wouldn't that be Westinghouse/Tesla and Edison? Tesla worked for Westinghouse (although he had previously worked with Edison).

Re:Edison reaching out from beyond the grave

[Anonymous Coward]

12V or 24V would be a horrible idea. A single rack would require many hundreds of amperes.

Re:Edison reaching out from beyond the grave

[WaffleMonster]

I told you bitches I would prevail one day!

There seems to be a popular/fundemental misunderstanding of the tesla/edison debate.

DC is MORE effecient on the wire than AC given the same voltage, amperage and wire gauge.

The reason for this is in AC systems eddy currents induced by changing electric fields at 50/60hz cause electrons to migrate away from the core effectivly reducing wire size.

Why AC has been the choice for so long is an engineering problem.

Building rectifiers to convert AC to DC from huge AC generators which produce virtually all of our electricity with the kinds of voltages needed to carry massive quantities of volumes of energy is difficult, unreliable and ineffecient..even today.

Back then it was practically impossible. The choice between Tesla and Edison really boiled down to high vs low voltage. Low voltage transmission required impossible quantities of copper or decentralized generation.

Tesla wanted larger more centralized generation which given what we use for fuel these days is an exceedingly smart move.

Re:Edison reaching out from beyond the grave

[ssyladin]

As an aside, car manufacturers are also moving towards higher voltage because that gives them easier access to drive high-power systems like power steering. Currently most power steering is pneumatically driven, which comes with a hefty overhead cost in terms of manufacturing and maintenance. With a high-voltage bus to drive it, the complex machinery simplifies it to an electric motor and some gears.

Re:Edison reaching out from beyond the grave

[aXis100]

I would argue that #2 is becoming less relevant too.

Many industrial 3 phase motors are being driven by variable frequency drives for improved control and huge efficiency savings. The first stage of a VFD would be rectifying the 3 phase AC back to a DC bus (same as a UPS) so DC distribution would work fine for that too.

Re:Edison reaching out from beyond the grave

[Vegemeister]

No.

1. Only very stupid engineers design power connectors that can fit both ways.
2. Doubling the voltage in rectification relies on the AC polarity reversal. Also, modern power supplies use a boost converter input stage (which corrects power factor) rather than an optional voltage doubler.

Re:Edison reaching out from beyond the grave

[Vegemeister]

You have a fundamental misunderstanding either of power conversion, or of prudent safety practices. You should change your username.

Re:Edison reaching out from beyond the grave

[lgw]

Electric power steering feels like crap, though - even less feedback from the road than normal power steering. It's one of the problems with hybrids today, to the point that the Infiniti hybrid I recenty test drove had a "hybrid power steering system", just to give it proper feel (so the complexity of both systems added together, yay!). The complexity of power steering comes from trying to make it feel like a simple rack and pinion.

Re:Edison reaching out from beyond the grave

[tombeard]

All my power steering systems have been hydraulic.

Re:Edison reaching out from beyond the grave

[tyrione]

You have a fundamental misunderstanding either of power conversion, or of prudent safety practices. You should change your username.

Agreed.

Re:Edison reaching out from beyond the grave

[lgw]

That "whooshing" sound you hear? Yeah, the guy who modded that post "informative" heard it too.

Re:Edison reaching out from beyond the grave

[Bright Apollo]

Oh, you mean a wider grid like, say, the entire Northeast Corridor of Amtrak?

Re:Edison reaching out from beyond the grave

[Anonymous Coward]

At 60 Hz, the skin effect is virtually nonexistent.

Monster Cable begs to differ!

Re:Edison reaching out from beyond the grave

[ktbaia]

I beg to differ. The Electronic Power Steering in my 1995 NSX decidedly does NOT feel like crap. When EPS systems were first put on production cars (the NSX was the first), it was on sports cars, where good steering feel is very important. See this link for information on how a decent EPS system should work. Just because Toyota can't figure out how to do a decent EPS doesn't mean that it hasn't been done. http://www.nsxprime.com/FAQ/Technical/eps.htm

Re:Edison reaching out from beyond the grave

[Anonymous Coward]

At 60 Hz, the skin effect is virtually nonexistent.

That's why I prefer European porn at 50Hz.

Re:Edison reaching out from beyond the grave

[Wonko the Sane]

but reliably at 132kV? It aint easy, cheap and certainly neither if you want efficiency.

Nothing that involves large amounts of power at 123 kV is cheap.

Even if a solid state substation is initially more expensive to build than a traditional substation that's a sunk cost while the efficiency gains of using DC accrue every day.

Re:Edison reaching out from beyond the grave

[soundscape]

I wish I had mod points for you.

Re:Edison reaching out from beyond the grave

[Falconhell]

3 things worry me about DC transmission.

1. Corrosion is a much bigger problem, cars are usually negative eart so that corrsion occours on the
device rather than the chassis side, maiking it reapirable.

2. Switching. To switch the same current of DC as AC the swithch needs to be a LOT bigger as arcing can and will occour with DC switches.

Have a look at the specs of switches, often the switch will handle 5 or more times current with AC than DC.

3. When touched one muscles tend to lock on to the item with DC.

Re:Edison reaching out from beyond the grave

[DCFusor]

The electric power steering on my Cruze and on my Volt is very nice, thanks, and less bump-steer than my Camaro, which has the old style. I think they're made by Bosch, for a lot of cars...no complaints so fa

Re:Edison reaching out from beyond the grave

[petermgreen]

When they talk about losses, usually it's over long distance over a few miles or between substations

Losses can be quite significant even within a building, especially at lower voltages.

The problem is (assuming single phase AC or single ended DC here for simplicity, other systems are a little more complex but the main principle holds) for a given cable size and power consumption power losses go as the square of the (RMS) supply voltage. Conversely for a given power consumption and acceptable level of power loss the cable size required goes as the inverse square of voltage.

So a 48V system ends up needing 25 times as much copper as a 240V system to achieve the same losses. A 12V system would need 400 times as much copper as a 240V system to achieve the same losses. Of course using that much copper isn't very p

Also with 12V there may be a lot less cost in terms of replacing the server PSU's as it's already closer to what you already need

I doubt it's significant, you are going to want isolating PSUs in the servers to make sure DC return currents don't accidently go where they shouldn't.

*running a rackful of high end servers off 12V you could be talking THOUSANDS of amps per rack. So unless you want signal connectors welded in their sockets you need to keep VERY careful control of where that current goes.

Re:Edison reaching out from beyond the grave

[LinuxIsGarbage]

A brushless squirrel cage AC motor itself is better than a DC motor, regardless of where the power source for the VSD comes from, so the parent's original statement is still correct. Incidentally while the AC motors are more trouble free, I've had more problems with high horsepower AC drives than DC drives.

Re:Edison reaching out from beyond the grave

[Anonymous Coward]

hook up a 12v battery directly to your motherboard. Come back and tell us what happens after the smoke clears.

Re:Edison reaching out from beyond the grave

[Anonymous Coward]

obviously you have never drove a mazda 2/demio (the best feedback i ever got from a steering wheel, i have drove a lots of car and i own an altezza)

Re:Edison reaching out from beyond the grave

[icebike]

Been there, done that. 5volts too.
I've run computers complete from batteries.
Its not a problem.

So if you had a point, its totally obscure.

Re:Edison reaching out from beyond the grave

[JoeMerchant]

Since DC can't really travel far at all without significant losses

AC/DC doesn't matter wrt distance, voltage does. High voltage transmits the same power with less current (more current = more loss), thus the 380VDC system proposed in the article, even though the machines themselves use 12, 5, and lower DC voltages.

Re:Edison reaching out from beyond the grave

[JoeMerchant]

The problem with DC back in Edison's day was that you couldn't easily step it up or down.

There's also the problem that, at reasonable transmission voltages, if you shock yourself on 60Hz AC, you're likely to spasm off the terminals within a few cycles, but a DC shock is more likely to cause your muscles to lock you onto the source where you will stay until fried extra crispy, unless helped off by a bystander.

Re:Edison reaching out from beyond the grave

[sfm]

I sometimes wonder if there is much difference between using Monster Cable and coat hanger wire as far, as sound quality goes.

Re:Edison reaching out from beyond the grave

[Anonymous Coward]

Nonsense. My audiophile friend (who also happens to work at Best Buy, a fine purveyor of these wares) tells me all the time how sound quality is improved by the increased thickness that only Monster Cables sport. ^_^

Re:Edison reaching out from beyond the grave

[bkcallahan]

A/C skin effect applies *based on frequency*; I assure you skin effect at 50/60Hz, even at high voltages, is negligible -- Doesn't matter if it's 420kV or 4.2V. A circuit at 29.350 MHz at low voltages has to worry a LOT more about it than a 50/60Hz line voltage. (And it really starts kicking off at the start of the microwave range, 300MHz, and is exceptionally important by the end of the microwave range 3GHz.) The reason A/C is used is based off of Ohm's law and is based on current and resistance; Jack up the voltage 1,000x and you can reduce the current (and therefore, heat loss) 1,000x.

Re:Edison reaching out from beyond the grave

[bkcallahan]

+1

Re:Edison reaching out from beyond the grave

[JWSmythe]

Well....

    As most coat hangers sold now are plastic, I'd say that's a problem..

    And a single coat hanger is a single conductor. It would take 2 metal coat hangers that don't touch (or one at half the distance).

    I'd be willing to bet, with properly insulated and terminated coat hangers, they'd be superior to monster cables. :)

Re:Edison reaching out from beyond the grave

[Anonymous Coward]

skin effect isn't the reason, look up the skin depth at 60Hz.

The reason is that peak, not RMS, voltage and current are limited, and the integral of cos^2 is half the integral of 1.

Re:Edison reaching out from beyond the grave

[jones_supa]

In one blind test the audiophile audience couldn't point the difference between a "hifi scam" cable and a lamp wire which was connected to the amplifier with only one thread of copper.

Re:Edison reaching out from beyond the grave

[evilviper]

the 230 VAC , which has a peak to peak voltage of 325 volts, just a stones throw from 380 VDC.

I have a much better idea... let's just run datacenters at 230VAC. 90%+ efficient PSUs these days means you're throwing away standard components to save a couple percentage points on efficiency... not a good trade off.

The DC datacenter idea is a holdover from the days of 60% efficient PSUs, which is long past.

BTW, APC agrees with me:
  http://www.apcmedia.com/salestools/NRAN-76TTJY_R2_EN.pdf

Re:Edison reaching out from beyond the grave

[The Master Control P]

Ohm's law applies just as much to AC as DC. A sine wave of a given RMS voltage will put the same power into a resistive load as DC of the same voltage.

You're right that high voltage leads to lower current and lower heat loss (quadratically, not linearly. P=I^2R). The reason AC was originally chosen is that there's no way to effectively transform DC voltage to take advantage of the above without modern semiconductors: mogens are inefficient and require constant maintainence, while MARs are inefficient, and huge, and fragile, and full of lots of hot mercury.

Re:Edison reaching out from beyond the grave

[SuricouRaven]

The skin effect at 60Hz isn't really that significant.

Re:Edison reaching out from beyond the grave

[SuricouRaven]

Tesla was the genius inventor, but Westinghouse was the business mind that turned inventions into money - and paid for more inventions. Edison was both inventor and businessman, he didn't need to work with anyone.

Re:Edison reaching out from beyond the grave

[locofungus]

It's of the order of a centimetre at 50Hz in aluminium.

Re:Edison reaching out from beyond the grave

[Anonymous Coward]

At 60 Hz, the skin effect is virtually nonexistent.

At 60Hz, the skin depth in Aluminium is about 11mm, and the skin depth of Copper is about 8.5mm

Re:Edison reaching out from beyond the grave

[mwvdlee]

Yes there is!
Mainly because you can't afford decent speakers after paying for the Monster Cable wires.

Re:Edison reaching out from beyond the grave

[adolf]

Hmm. APC, being in the business of selling AC UPS systems, probably isn't too keen on the concept of DC power distribution to begin with. (Which is not to say that they're lying, but just that they're certainly biased.)

We use DC quite a lot in small(ish) commercial 2-way radio systems. One of the really neat things about it is that integrating a "UPS" can be just as simple as adding an appropriate lead acid battery in parallel with the load and calling it a day.

Not surprisingly, these batteries don't say "APC" on them....

Re:Edison reaching out from beyond the grave

[PIC16F628]

Totally agree with you. I have few times touched live 230V AC in the house and you can feel your hand/fingers vibrating (at 50Hz). Everytime we are cycling through 0V, it is easy to pull back your hand from the wire. With DC, your hand will get stuck. In my opinion this itself should be the determining point for choosing AC. After all, as we progress, power consumption of devices will keep falling.

Re:Edison reaching out from beyond the grave

[sourcerror]

Is there solid state technology to convert up to the kV level?

Re:Edison reaching out from beyond the grave

[Anonymous Coward]

s/pneumatically/hydraulically/

Most power steering is hydraulically driven. Pneumatic == air.

Re:Edison reaching out from beyond the grave

[fnj]

Most power steering is hydraulic, not pneumatic. Most power brakes are pneumatic.

Re:Edison reaching out from beyond the grave

[Stuarticus]

but reliably at 132kV? It aint easy, cheap and certainly neither if you want efficiency.

Nothing that involves large amounts of power at 123 kV is cheap.

You've already made your first mistake and blown up your first test unit!

Re:Edison reaching out from beyond the grave

[TheTurtlesMoves]

Transformers are still the cheapest way to change voltages by a long shot. Especially for very high currents and voltages.

Re:Edison reaching out from beyond the grave

[lazybeam]

Since when is a corridor a grid?

Re:Edison reaching out from beyond the grave

[GameboyRMH]

Currently most power steering is pneumatically driven

LOLWUT? You mean hydraulically right?

Also to make a direct-drive electric (vs. electro-hydraulic) PS system that doesn't feel like crap requires a lot of advanced tech in the car. It can be cheaper in mass production but it isn't simpler by any means.

Re:Edison reaching out from beyond the grave

[TheTurtlesMoves]

Don't forget that transformers typical efficiency are higher than switch mode power conversion. Quite a bit higher, and then add the fact they are a lot cheaper too.

Re:Edison reaching out from beyond the grave

[TheTurtlesMoves]

Switched mode power supplies like in PCs have no transformer. They also work better with DC input since they don't need to be nice about power factors so much.

Re:Edison reaching out from beyond the grave

[AmiMoJo]

Indeed, the EU is looking at building large solar thermal plants in north Africa and sending the power back to Europe via DC.

Re:Edison reaching out from beyond the grave

[AmiMoJo]

Oh god, don't get them started! The next big snakeoil-fest in hifi will be DC power distribution.

Re:Edison reaching out from beyond the grave

[Anonymous Coward]

It isn't so much AC or DC that matters, but the voltage. The higher the voltage, the less electrons have to actually move, which is where the savings come from. It's very hard to increase the voltage on a DC line (the easiest way to do it is converting it to AC first).

Second problem is what happens in case of a disaster. While I'm not going to claim that a loose high-voltage AC wire is a good thing, it'll basically blow, vaporize a part of the wire in a very short time and that will be it. If a high voltage DC wire were to short-circuit it'll probably relatively slowly (a few dozen seconds) melt the entire substation into the ground, fusing itself to whatever is near. It is *much* harder to secure DC equipment against short-circuit than AC.

There's another obvious problem with 380V DC, or any significant DC voltage, one of the reasons we stopped using DC (the other being that most generators produce AC, and it being very freaking hard to get high-voltage efficiencies using AC) ... and it is this : with 240V AC, or even 480V AC you will very likely survive contact between your skin and a current carrying cable. With 380V DC you won't (not that you would with 120V DC, it's not the voltage that does it, it's really that DC is much more dangerous).

This isn't because of the power per se, but rather because of the dumb fact that a 380V DC cable that starts discharging through your skin will melt itself into your skin, heat it up and damaging your skin until there's blood. Once there's blood, conductivity will vastly increase and it's all over. Whereas 380V AC will change your body into a capacitor, which will kill you, but it will take far longer to actually do it, the heat up of the skin will be far more spread around, which means the skin won't get immediately damaged and the current will not be able to disperse into your blood. (short version : touch a high voltage DC wire, you're dead. With a high-voltage AC wire if you don't seriously scratch yourself on the wire, you'll probably walk away with a cool story)

Re:Edison reaching out from beyond the grave

[Anonymous Coward]

Skin effect isn't much of a problem at 50Hz. It is reactive power that needs to be pushed back and forth without any productive outcome except of heating wires. The next problem with AC are higher harmonics produced, to some degree by every step down power supply. On the other hand distributing power with DC in data centres poses the same problems as hundred years ago. Large power requires thick wires. Let's look at some numbers.

42U cabinet with 24*60W (what about HPC clusters with 200W+ GPUs?) is 1440W at 48V requires 30A at 12V it is 120A. With a rule-o-thumb 10A/mm^2 of copper this gives 3-12mm^2 wires. Multiply this by tens or hundreds of cabinets.

Re:Edison reaching out from beyond the grave

[OeLeWaPpErKe]

For AC, yes. For DC, no. Which is the problem ...

Problem with DC : upscaling the voltage is hard, complex and inefficient (but you can convert AC directly into high-voltage DC)
Problem with AC : there does not exist such a thing as an AC battery. You want battery backup, you need DC

Only in datacenters, which work with battery backup and thus usually have multiple AC->DC->AC conversion points, with big losses, is this even a factor.

How it works today is :
BIG high-voltage AC input to the datacenter, which is converted to DC (inefficient) and fed into the power system, which charges the batteries (inefficient) and powers several huge inverters which convert back to AC (inefficient), then this AC is transmitted over a normal power distribution system into the server racks, and each server takes in AC and ... converts it into 5V and 12V DC (inefficient).

So the theory is, let's change this into :
BIG high-voltage AC input to the datacenter, converted to DC (inefficient), which charges the batteries (inefficient), is distributed troughout the building and powers the servers.

The idea being that a huge extremely high-voltage power supply is more efficient than all the ping-pong conversions going on today.

The alternative to this is what google does. Instead of eleminating the AC-DC conversions at the headend, eliminate them at the tail end.

Google datacenters:
BIG high-voltage AC input to the datacenter, which is directly distributed into the datacenter and fed into the servers, which convert it to DC, and with that DC line, charge a battery specific to that one server, and use the same source for the actual server.

It has the same efficiency gains, obvious massive redundancy and maintenance gains. You have transformed the datacenter from a big centralized SPOF into a much more resilient and more efficient beast. BUT it requires people to maintain their own battery backup.

Re:Edison reaching out from beyond the grave

[Wonko the Sane]

Is there solid state technology to convert up to the kV level?

You need to combine solid state components with reactive components (inductors and/or capacitors) but can be done.

The easiest and most logical place to do that conversion would be at the point of generation as part of the rectification process since most forms of creating power involve rotating prime movers and naturally produce AC anyway.

Re:Edison reaching out from beyond the grave

[The_Crisis]

In this case, just where along the line do you convert the AC to DC. ?

FTFA:

Most data centers use power distribution systems in which AC power from the grid is converted into DC power to charge the UPS batteries, and then converted back to AC for the equipment. The loss of power through multiple AC/DC conversions has been cited as an argument for using DC power distribution. ”DC distribution can eliminate at least three conversions,” said Symanski.

It sounds like the flow will be having your 120/208/240/other VAC coming into your data center area to one transformer that will send your VDC power through a UPS to the PDU's and so on down the line. It sounds to me like the main point they are trying to make is the most significant gains in efficiency is by eliminating unnecessary conversions and so eliminating the cause of a lot of lost/wasted energy. Assuming your data center is set up with a fairly reasonable footprint (read: in the same room) then arguments of power loss over distance are pretty much moot. It doesn't sound like they're wanting something as radical as changing the standard of the entire grid, instead focusing on making the most of the juice you're getting from it.

Re:Edison reaching out from beyond the grave

[Anonymous Coward]

Yes, but it was Tesla's idea vs Edison's idea. Westinghouse was Tesla's investor.

Re:Edison reaching out from beyond the grave

[Anonymous Coward]

It might be non-existant at 220V. That is only the voltage for the last line. A normal power line (not thosel little wires for the last mile) have at least 10kV, often up to 1000kV (though most common is 380kV around here).

Re:Edison reaching out from beyond the grave

[Smallpond]

Arcing is a big problem. AC goes through 0 volts twice each cycle, so arcs die quickly. A faulted DC switch contact can just sit there and arc until the human in the circuit stops conducting.

Re:Edison reaching out from beyond the grave

[Smallpond]

This is not true. Loss is Watts = I * R where I is the current and R is the resistance of the conductor. If I cut the supply voltage in half, I need twice the current to supply the same power to the device. To keep losses constant I therefore need half the resistance. Since R is proportional to the area of the conductor, I need twice the area which means twice the copper. There is no squared term.

Re:Edison reaching out from beyond the grave

[Smallpond]

n/m I'm totally wrong.

Re:Edison reaching out from beyond the grave

[budgenator]

1. Only very stupid engineers design power connectors that can fit both ways.

or only use one level of idiot proofing

Re:Edison reaching out from beyond the grave

[PhinMak]

My 2004 Mini S had electric power steering... and Mini Cooper's brand is built on nimble driving, including responsiveness and steering feel.

The one drawback that I have to point out is that the electronic pump is a one of the weaker parts in the car. Anecdotal evidence (my experience plus discussions with dealer service managers and wrench turners) suggests that they fail at least once every 5 years... To add insult to injury, the part ALONE costs $900 at the dealer. Compare to less than $100 for a comprable hydrolic pump in a domestic car. This suggests serious design/material costs or price gouging by the manufacturer.

Re:Edison reaching out from beyond the grave

[budgenator]

Yes they do, without a transformer, they would be limited to the 15 amps the mains supply and 12 VDC @ 15 amps is only 180 watts, that isn't enough to run some CPUs. The power supply I'm holding right now has at least two visible without opening the case

Re:Edison reaching out from beyond the grave

[budgenator]

A transformer also allows AC neutral to be bonded to ground on the transformer's secondary winding, a floating AC neutral can easy destroy a lot of equipment.

Re:Edison reaching out from beyond the grave

[swb]

Why would APC care? They are primarily in the business of selling backup power systems to handle the loss of utility power, which will probably stay AC for the foreseeable future.

Even if there was some mass migration to DC power distribution for data centers or even in large office buildings (building managers sometimes make rules about data centers in their buildings to keep building operation costs low), why would APC not want to be involved in that business as well?

You still have a risk of loss of utility power and need some kind of backup power system, and while the old parallel lead acid battery is fine for stone-axe simple systems like radios or alarm panels, I doubt anyone will be willing to "just" hardwire a battery array the size of fridge to their data center without any kind of monitoring, switching or charge maintenance system associated with it.

Re:Edison reaching out from beyond the grave

[Bright Apollo]

Since it has its own power generation plants, for one.

Re:Edison reaching out from beyond the grave

[Elder Entropist]

I was speaking of long distance transmission, where very large conductors are used and skin effect does play a significant role in limiting the max effective size of an individual conductor and in the composition of the conductor as many long distance transmission wires use a steel core with an aluminum sheath because the skin effect causes most of the current to ride the edges. For less industrial-level conductors, yes, the skin effect is negligible at 50-60Hz.

Re:Edison reaching out from beyond the grave

[alfredos]

[citation needed]

Re:Edison reaching out from beyond the grave

[Grishnakh]

Totally wrong. You seem to have very little knowledge of electricity.

Not really. AC was the answer to how to transport electricity long distances.

Wrong. High voltage is the answer to how to transport electricity long distances with low losses, according to Ohm's Law. For a given power, the higher the voltage, the lower the current. The lower the current, the lower the losses (which are calculated by I^2 * R; the losses go up with the square of the current). AC was the answer for a while because with Tesla's ingenious inventions of the synchronous AC generator and the transformer, it was very easy to generate power in AC and then convert it to very high voltage, and then convert it back down to more useful low voltages for use at customer sites. There was no way to do this with DC for some time. Eventually (1930s I believe), someone invented HVDC, a way of converting generated power to a very high voltage DC. I believe this was used in a northern Quebec generating station for a long time. HVDC reached higher voltages than even regular AC distribution, so it was favored for very long distances, plus being DC, you didn't have to worry about synchronizing it, so it was very useful for connecting separate AC distribution networks.

Currently, it is still converted to DC in a huge amount of devices

It's converted to DC in any electronic device. The only place it's not converted to DC is in non-electronic, electrical devices such as fans, stovetops, ovens, electric dryers, etc. (And even in these, many times it now is partially converted to DC because many such devices are using electronic controls these days rather than just simple switches.)

Few devices use AC iirc, something like a fan/ceiling fan probably has an AC motor because a DC motor would slice your finger off if you decided to play with the blades.

Wrong. AC vs. DC has nothing to do with "slicing off fingers". Fans use AC motors because they're dirt cheap, as AC motors are mechanically much simpler than DC motors and have better performance and don't have to worry about wearing out brushes. (Well, better performance for fans at least; they use less power, and it's easy to make them multi-speed with a few capacitors. Many power tools use universal motors, which are really DC motors that run on AC; these have high speed and power, but they wear out relatively quickly and make a lot of noise, but for an occasional-use home power tool that's not a big problem.) A fan can only slice off your fingers if it's moving really fast, and that's a function of motor design, not AC vs. DC. It's easy to make a low-speed DC motor, but it's not as cheap as an AC motor, given the fact that your house is already powered by AC. If your house were powered by 120VDC, then your fans would use some kind of DC motor too, but they'd probably cost more.

Since DC can't really travel far at all without significant losses

Totally wrong. As I said before, low voltage power can't travel far without significant losses. There's tons of HVDC links in power distribution systems around the world. Look it up.

The problem with this 380VDC idea I see in the summary (no, I haven't bothered reading the article yet) is that 380V is useless for powering electronics, so it'd have to be stepped down to 12V, 5V, 3.3V, 1.8V, etc for use in the electronics. That requires a DC-to-DC converter, which just like any power supply also has losses (which are no less significant than in a regular 120VAC PSU). This idea seems downright stupid on the face of it: convert 480VAC (standard industrial distribution power level) to 380VDC with a giant AC-to-DC converter for the whole datacenter, then in every computer use a DC-to-DC converter to convert this to 12V, 5V, and 3.3V? This will only incur more losses than what they're doing now. What would make much more sense is to simply run 480V to every rack, and in every rack have a 480VAC-to-12VDC converter. Most devices in a computer run on 12V, and convert it down to lower

Re:Edison reaching out from beyond the grave

[alfredos]

Electric power steering feels like crap, though

As a 3rd gen Prius user, I beg to differ. Granted, it's not the addictive feel of a BMW coupe, but it feels just fine and if I have to find something needing improvement it's the initial response. But I am rather pleased with this car's handling and especially its power steering, all things considered.

Re:Edison reaching out from beyond the grave

[Grishnakh]

Wrong. ALL power brakes in passenger cars and other lightweight vehicles are hydraulic. Most power brakes in heavy vehicles (over-the-road trucks, etc.) are pneumatic.

Re:Edison reaching out from beyond the grave

[Grishnakh]

It would be trivial to convert power supplies to 380 VDC.

Yes, but then why bother having a big 480VAC-to-380VDC converter for the datacenter? You're just exchanging losses in one place for another.

What's make more sense is to run 480VAC (standard industrial distribution voltage) to each rack of servers, and in each rack have a single (or dual for redundancy) 480VAC-to-12VDC converter, then run that with bus bars to each server.

Re:Edison reaching out from beyond the grave

[Anonymous Coward]

It's a blatantly obvious joke post, you humorless fuck. That goes for you too, parent post.

Re:Edison reaching out from beyond the grave

[Grishnakh]

Given that 120VAC is mainly a residential thing, I'd be surprised if any datacenters really ran their PCs off 120VAC instead of 240VAC, which is easily available in any commercial environment (in 3-phase form).

Furthermore, three-phase 480VAC is the standard in industrial environments, so it'd make more sense to use that.

Re:Edison reaching out from beyond the grave

[colinrichardday]

Yes, I was trying to say that Tesla should have been paired with Westinghouse, not Edison.

Re:Edison reaching out from beyond the grave

[colinrichardday]

Yes, but why did the poster to whom I responded pair Tesla with Edison instead of Westinghouse?

Re:Edison reaching out from beyond the grave

[SuricouRaven]

Possibly because he was half-right. Edison *was* a hack. A good inventor, yes - but an even better businessman, quite ruthless, and not above stealing credit. He ran a research lab with a sizeable team of inventors, but personally took the credit for everything they came up with in order to create the image of a genius inventor. Better for business to have good publicity like that.

Re:Edison reaching out from beyond the grave

[Skapare]

That depends on the sourced voltage. The if the source voltage is X relative to ground, the floating system cannot be greater than X+V where V is the voltage across the secondary terminals. And there are also ways to hold the float near ground, as well (e.g. winding shielding, or a double stage low voltage grounding). But the simple case of a plain transformer with an ungrounded secondary is an issue at X+V.

Re:Edison reaching out from beyond the grave

[Vegemeister]

Diode drops are bad, 'mmkay?

Re:Edison reaching out from beyond the grave

[colinrichardday]

The way it the OP phrased it put Edison and Tesla on one side of the AC/DC question, and Westinghouse on the other.

Re:Edison reaching out from beyond the grave

[adolf]

When the systems keep chugging away without even noticing that the power dipped because there's a simple and massive battery bank hanging on the DC rail, there's little need for AC backup. Even the emergency lighting can run from DC.

Soon after the utility power drops, the genset will kick on, anyway. There's very little need for a traditional, large UPS in a datacenter when everything important runs on DC.

And I'm sure that APC would certainly want to be involved in that business (who wouldn't?), but their existing product line just doesn't fit very well with DC power distribution: Most of their profit currently comes from the inverters that they make, which just aren't necessary in this context.

Meanwhile: As long as big, fresh, reliable AC Delco deep-cycle lead acid batteries continue to be stocked and sold at NAPA (with free delivery, even), having APC-blessed batteries shipped out from wherever is going to be a non-starter from the standpoint of any customer with a DC-powered datacenter and an eye on their wallet.

Re:Edison reaching out from beyond the grave

[evilviper]

Which is not to say that they're lying, but just that they're certainly biased

Yes, they obviously might be a little biased. However, dismissing a report because of the source, without actually trying to refute anything they've said, is incredibly lazy and dishonest, and just makes you look biased.

One of the really neat things about it is that integrating a "UPS" can be just as simple as adding an appropriate lead acid battery in parallel with the load and calling it a day.

That's a great strategy for small and unimportant loads. For critical servers, not having the wiring to do routine battery tests is a complete non-starter.

And just because it's "neat" doesn't mean it's considerably more efficient. Switching power supplies have gotten extremely efficient. Total up the losses in the most efficient PSU and UPS you can find, and tell me how much we're hoping to gain even if DC was 100% efficient, and how much it's going to cost to get there. The numbers aren't appealing.

Re:Edison reaching out from beyond the grave

[TheTurtlesMoves]

Pull em apart yourself if you like. I have 2 here on my desk. I have build several myself. They don't have a transformer in em. Inductors, yes. Transformer no. There is no 15amp limit without one. Not sure why you would think that. Its not in my "SMPS for dummies" ;).

Re:Edison reaching out from beyond the grave

[Anonymous Coward]

The reason A/C is used is based off of Ohm's law and is based on current and resistance

Yes, quite true.

Jack up the voltage 1,000x and you can reduce the current (and therefore, heat loss) 1,000x.

Nope! Gotcha!

I don't work in the power industry, so I never thought about this one enough. But a while ago I was confused that neighborhood power distribution was a few kV. That didn't seem to be enough voltage increase, because I, like you, had in my mind the 1000x voltage increase means 1000x current decrease means 1000x power loss decrease.

But that's wrong, wrong, wrong. Power loss equation is I-squared-R. Which means if you reduce the current by 1000x, you reduce power loss by the square of the current difference, i.e. 1,000,000x. Now all of a sudden all those high voltage distribution systems make a lot more sense. The grid losses become very manageable!!!

Re:Edison reaching out from beyond the grave

[Anonymous Coward]

Tell that to AC power line manufacturers. Or try to build a hight voltage power line using regular wires. And see how negligible things are then....

Re:Edison reaching out from beyond the grave

[swb]

And I'm sure that it corporate data centers everywhere they will just be fine with bare NAPA batteries on the floor with no charge status or load status, especially the kinds of places that have installed generators.

None of them will be interested in enclosures for those batteries that ensure they don't leak, overcharge, get shorted or do anything else nasty, or do something useful like supply some AC power for other devices or enable monitoring and alarming.

And APC won't want to be in that business.

I get your beef, APC is in the overpriced battery business. But batteries plus and the like have been selling batteries that work fine in APC systems forever, you certainly have not had a gun to your head to buy their batteries.

Re:Edison reaching out from beyond the grave

[budgenator]

Well OK maybe the limit isn't always 15 Amp, they usually have a 20 Amp circuit supplying a typical home's kitchen, and then there is always the 240 Volt circuit to the drier and the range in the kitchen which will get you 30 Amps, the reason for this (for non-isolating PS's without transformers) is that a switching mode power supply is basically an electronically controlled switch and the current out cannot exceed the current in. Wikipedia shows where the transformer is located very clearly in this diagram. More information on SWITCHING MODE POWER SUPPLY (SMPS) TOPOLOGIES is found on the preceding link. If I found out that one of my computer's power supplies was non-isolating, the power supply would be replaced immediately and it would be the last computer I'd ever buy from that manufacturer.

Re:Edison reaching out from beyond the grave

[TheTurtlesMoves]

I know what they are. I have made them. Built them. Look inside a RCD and you see that you can have safety without a transformer. I know that many designs in PCs don't, they are inductors, for the simple reason that transformers that are at the 90%+ efficiency range put the price up too much.

SMPS can and do convert with high current and voltage ratios with great power factors if designed right. The last one i did was a 50amp +-24V and earth rail for a stepper motor controller from a 240V mains. It had no transformer and it had a good power factor. I did it that way because its was cheaper than a transformer with that kind of rating. Look up the price of a 500VA transformer sometime, or even better look at the size of the dam thing.

Just pulled one of my old ones apart and checked it. It does not have a transformer, Every core/ferrite is a two terminal device only. DC isolation does not make a device safe either btw. In the countries where i have read the codes, its not needed, and RCD/isolation is only needed for a class of "outdoor" use devices. Most of the code deals with proper insulation. Do you really think those $20 hair dryers are DC isolated?

also needed for houses

[Anonymous Coward]

How many little wall-warts does the average house have? Tens? We need low voltage DC in our houses, and standardize all the little widgets on one of (say) two voltages. Each outlet could supply them in a dedicated connector alongside the current AC.

Re:also needed for houses

[oh_my_080980980]

You've apparently never ran electrical wire in the home nor understand why that would be a completely insane idea.

Re:also needed for houses

[Anonymous Coward]

If we're talking about 5V/1A USB or whatever it really wouldn't be that big of a deal, if we're talking about the 1000W you use to power your gaming rig, yeah, it's stupid.

Re:also needed for houses

[LehiNephi]

I do/have done both (run electrical and understand the implications), and GP has a point. When I think of the things in our house that *must* run on AC, it's only our fridge, freezer, and HVAC. Everything else in the house either converts it to DC or could run quite happily on DC. For certain you'd want to have a different kind of plug for DC devices, but even that would give us an opportunity to 1) standardize on one global plug standard, at least for DC, and 2) allow us to design a small, rugged, safe type of plug.

Or is there some implication that I'm missing, and that you decided not to point out, in favor of flaming GP?

Re:also needed for houses

[stanlyb]

And dangerous. And not practical. And expensive too.

Re:also needed for houses

[Anonymous Coward]

the issue isn't running the electrical wire in the home, the issue is distribution over long distances. You can't distribute DC at low voltages due to massive resistive losses, and DC doesn't play well with voltage changes. Once in the home, there's no problem running DC per say, but that being said, you'd probably still want to distribute in the home at 100+ volts, which means that all the problems of changing voltages around would in macro would still hold for household size.

And for those who don't know what I'm talking about, math Vloss = I*R where I is the number of amps being pulled and R is the resistance of wire. I is determined by I=P/V where V is the line voltage, R is determined by the width of the wire and the length, there's an equation that I can't remember off the top of my head and can't be bothered to look up. But basically, the narrower the wire, the higher the resistance. The longer the wire, the higher the resistance.

Re:also needed for houses

[suutar]

expensive yes, practical maybe not. I don't see offhand how it's more dangerous than having AC all over, but I'm not an EE. Would you care to explain?

Re:also needed for houses

[vlm]

For certain you'd want to have a different kind of plug for DC devices, but even that would give us an opportunity to 1) standardize on one global plug standard, at least for DC, and 2) allow us to design a small, rugged, safe type of plug.

Aka the famous (in some circles) Anderson Power Pole. Go ask a ham radio guy.

The thing I love about in house DC distribution, which I have in my house, is it forces at least a token effect at "green power reduction". Suddenly given the choice of a 12 volt 6 watt LED fed by $2 of small gauge wire vs something resembling welding cable wire to run a 200 watt halogen, you make the ecologically correct choice.

I used to use cast off surplus 200 watt desktops for my mythtv frontends. Unholy pain to run on 12 V. Now I use 5 watt Zotac boxes. Good for everyone in every way.

Re:also needed for houses

[Wonko the Sane]

You can't distribute DC at low voltages due to massive resistive losses

This is equally true of both AC and DC.

DC doesn't play well with voltage changes

We have these things now called semiconductors. You may have heard of them.

Re:also needed for houses

[nschubach]

What bothers me is all the new LED bulbs that have transformers in them (guessing, because they get hot! ... feels like wasted energy) to power the LED bulbs when you could have just run a 12VDC line and powered them all on a central transformer like garden lights. I'd think it would be more efficient to run DC to lighting and certain outlets like those where small devices would sit (with standard plugs as you mention) and keep 112-120VAC for things like the appliances.

You could also centralize a backup battery (easily) to keep your lights on if the power suddenly surges.

Re:also needed for houses

[David_Hart]

All AC appliances and DC power adapters use a standard voltage of 110v in US/Canada/etc., 220v in other parts of the world (UK/etc.). To implement DC in the house, you would either have to standardize on a specific DC voltage or create a smart power standard (i.e. similar to POE for network gear). Today, most devices that require AC/DC adapters, convert to different DC voltages. That being said, most mobile devices have been standardized to the USB standard of 5V.

Re:also needed for houses

[AK Marc]

Then can you explain why running 12VDC along-side 120VAC (assuming they are sufficiently separated/shielded to not induce 120VAC on the DC line) is a "completely insane idea." Just saying "if you knew what you were talking about, you'd know why it's a bad idea, but I'm too stupid to explain why, I can just mock" isn't very useful

Re:also needed for houses

[snowraver1]

The Alternating Current allows your muscles to have a chance to let go when you grab on to the wire, while DC will freeze your muscles up. That is one theory anyways... I have never personally been shocked good enough with DC to confirm this.

Re:also needed for houses

[Wonko the Sane]

It's a matter of how many watts you expect to supply and how large the wires must be to supply that power at 12 volts while maintaining the voltage at the outlets within an acceptable range.

Re:also needed for houses

[amorsen]

When I think of the things in our house that *must* run on AC, it's only our fridge, freezer, and HVAC

If your fridge, freezer, and HVAC are halfway decent, they convert to variable-frequency AC. Fixed-frequency AC motors are inefficient unless the load is constant (and load isn't constant in those applications).

Re:also needed for houses

[nschubach]

There are a lot of different voltages for DC transformers, but taking a look around my house this past year I spotted mostly 12VDC. I had one 20VDC (Laptop) but the rest were 12. In the past, I've seen 6 and 9 volt transformers, but I rarely see these anymore.

Re:also needed for houses

[Anonymous Coward]

"You've apparently never ran electrical wire in the home nor understand why that would be a completely insane idea."

Actually, if you stop and think instead of flaming for a sec, the post you reply to isn't a bad idea. It would centralize the AC->DC conversion in a single more efficient place, and it would allow all those little devices to be cheaper and less of a pain in the butt since they wouldn't all have to include their own transformer. Sure, low voltage power distribution is inefficient over long distances, but for the small ranges inside a house it wouldn't be a problem to have 12VDC everywhere.

Re:also needed for houses

[PimpDawg]

I have a 7805 voltage regulator next to me right now that does a fine job of stepping 12V into 5V at up to 1A without a heat sink. And that was a very cheap component. I imagine that one can do a lot more with more money or volume.

Re:also needed for houses

[ShavedOrangutan]

It's a lot harder to let go if you grab a hot DC wire. And I'm guessing, but I'll bet a DC shock to the heart would be worse than AC.

Re:also needed for houses

[sjames]

I've heard it said many times that when you accidentally contact the wires, DC tends to seize up your muscles so that you freeze in place and cook while AC tends to throw you off. I have no idea if that's actually true or not but I can see how it might be.

Re:also needed for houses

[FaxeTheCat]

Yes, and in the process it wastes 7/12 (58%) of the energy. not what you would call "green".

Re:also needed for houses

[Imagix]

I usually saw the 6 and 9 V transformers for devices which are normally battery powered. 2 x AA batteries would be 6 V, and the 9 V battery (of course).

Re:also needed for houses

[Anonymous Coward]

If you're lucky, the sudden muscle contraction with DC will yank you free before serious damage occurs. With AC, that doesn't happen.

Re:also needed for houses

[Anne Thwacks]

I promise they are not half-way decent, and there is no way in hell they convert to variable frequency AC - the washing machine possibly does.

The efficiency of most fridge and A/C motors is truely appalling - 30% would be considered good.

Re:also needed for houses

[stewartjm]

A few years back, I had my HVAC blower motor replaced with a GE Evergreen ECM motor. AFAIK, ECM motors are basically DC motors with AC conversion circuitry. Since I run the blower all the time, to help equalize temperature throughout the house, it's probably already paid for itself. It's roughly 30% more efficient than a comparable AC motor. In addition, it is significantly quieter than the motor it replaced.

It seems likely similar DC motors would work in pretty much all household appliances, though they may not have the same efficiency gains in all applications. So, it just comes down to economics, if it becomes cheaper to go to DC distribution in homes, then it will happen.

Re:also needed for houses

[Obfuscant]

What bothers me is all the new LED bulbs that have transformers in them (guessing, because they get hot! ... feels like wasted energy)

High power LEDs get hot because you are running good amounts of current through them, not because there is a transformer. Transformers are pretty much useless with the DC current that runs LEDs.

I'd think it would be more efficient to run DC to lighting and certain outlets like those where small devices would sit ...

The problem comes in deciding what voltage to use. 12V means you need rather hefty wires to get the required current for some devices. A 6W LED needs half an amp at 12V. If you use a voltage that makes the current resonable, then you need to convert that voltage to what your device needs, every place you have a device.

Sending 380V means you can use the same or smaller wires than you'd use for 120V systems, but you'll be busy converting that 380V DC to 12V DC or 5V DC or 1.2V DC -- and while DC-DC conversion has gotten a lot better, it is still more complicated than a simple transformer.

Re:also needed for houses

[Anonymous Coward]

You mean 4x AA batteries. They're 1.5V each.

Re:also needed for houses

[Anonymous Coward]

It would also be dangerous to have two electricity standards in the home. How many homeowners do you think might accidentally wire the wrong plug to the wrong electricity source? Having 120 VAC unexpectedly come out of your 12V DC source could be seriously dangerous.

Re:also needed for houses

[budgenator]

Compare the typical 12 gauge domestic power wire for (125 VAC @15 amps) in your house to a jumper cable for your car for (13.8 VDC @135 Amps) and you would see why.

Re:also needed for houses

[Anne Thwacks]

In the room where I am now, (my sitting room) there are 14 wall warts - most (say 10) are 5V phone chargers, etc. capable of over an amp. Five are laptop chargers rate at about 20V at about 5A. (Yes, there are five laptops in the room at present)

So you want 10A cabling for a 5V supply - so loss needs to be less than 0.2V - that is some hefty wire. And you want a 20V supply capable of more than 25A - say 30A. That is some more big fat wire! (6mm2 wire costs money and is a pig to connect to)

I think I prefer the 240V supply, thin wires and wall warts.

I can, and have, designed 240V off line switchers with losses of less than 1W - and they don't cost very much to make in million-off quantities. I am not sure the losses are much less with a single SMPS, and I certainly dont like putting all my eggs in one basket (Look what happened to Bernie Madoff's investors whio tried that optimisation)

Re:also needed for houses

[Anonymous Coward]

Apparently you don't know how that component actually works.

If you draw 1A at 5V on the output the component has to source 1A of current through the input. So it has to drop 7 volts at 1 amp. For your 5 watt draw it will be giving off 7 watts of heat.

Yes, if you only draw 10 mA through it, it's not going to get hot. Draw 1 Amp and you will fry it (w/o a heat sink).

Re:also needed for houses

[Anonymous Coward]

My house is dual-wired. It uses standard 3-core flex for both 240VAC and 24VDC circuits, i.e. live+neutral+earth for a 240VAC circuit, pos and neg for 24VDC (with the earth core unused). The lighting and refrigeration run off 24VDC, everything else is 240VAC. The DC circuits were designed to carry safe loads, so there are more of them - less outlets or sockets per circuit means more circuits. The advantage of using the same wiring is that if I wish to change a circuit from one to the other, I can move one end of the cable from the DC bus to the AC bus, and change the outlet/socket as appropriate. That's happening gradually - some of the 24VDC lighting circuits (currently running quartz halogens) will be cut over to 240VAC to run cheaper CFLs and/or drop-in LED lamps.

Re:also needed for houses

[c 0xFF]

Well, it's not exactly a transformer, but you're essentially correct -- a high-power LED bulb needs to be supplied its forward voltage at a constant current, which means rectifying and bucking the voltage down. There's lots of schemes to do this, some more efficient than others. The current limiting resistor scheme most of us are familiar with is about the most inefficient way, but it's also very inexpensive. LDOs and linear regulators are still very inefficient. Buck converters are better, but very expensive.

In the end, that conversion has to happen no matter what, whether at the bulb or at a central box somewhere else, so there's always going to be some wasted energy. Hopefully you'll get improved efficiency by centralizing everything.

But, even if the overall efficiency is the same, there's still the benefit of moving the heat to a different location. One major problem with LED bulbs is that they dim and can handle less current as they heat up. Notice the heat sinks they're putting on the really high power LEDs these days. Performing the voltage conversion elsewhere means that this heat is moved elsewhere.

And if you're smart about where "elsewhere" is located, you're also reducing the strain on the air conditioners! Put the conversion in a basement or outside the building, for example.

I wonder how much of the heat produced in a data center is actually produced by the power supplies for each server. 10% to 20% feels about right. Ignoring any other benefits of moving to DC, I would think that the 20% reduction on the air conditioning bill would be justification enough.

Re:also needed for houses

[Anonymous Coward]

I usually saw the 6 and 9 V transformers for devices which are normally battery powered. 2 x AA batteries would be 6 V, and the 9 V battery (of course).

2x 1.5v is not 6v

Re:also needed for houses

[c 0xFF]

Linear regulators (like the 7805) are stupidly inefficient. Not too bad if you're regulating down only a bit, which is why they're used to supply the variety of voltages on a single motherboard. The problem is that an LDO supplying 1A at 5V will draw >1A at 12V. About 40% efficient (5/12).

Better are DC/DC buck converters. These cost less than a buck in bulk and are about 85% efficient or better (I've heard 95%). I'm not sure what the efficiency would be for high-current applications.

The biggest disadvantage with buck converters, in my opinion, is the extra design effort that goes into using one. A 7805 just needs a few capacitors (and even those aren't really a big deal for many applications). Buck converters need a Schottky diode, an inductor, and a couple capacitors, and sometimes more. Kinda keeps them outside the realm of hobbyists.

Re:also needed for houses

[mirix]

Linear regulators are horribly inefficient. If you're drawing more than a 100mA at that much drop, it will need a heatsink - otherwise it will overheat and go into thermal shutdown.

Switch mode is what you are looking for, though. Plenty efficient, however more complicated than a linear reg.

Re:also needed for houses

[Cramer]

They have a transformer in them because they are d***ed cheap. A solid state (electronic) regulator is much more efficient, smaller, and lighter, but requires some engineering and a few pennies more to build.

All of them have some sort of power regualtor (transformer, transistors, etc.) that will generate heat in the process. You cannot put 120V (AC or DC) right into an LED. (not for long...)

Re:also needed for houses

[Anonymous Coward]

There are already two electricity standards in houses. The plugs are different. You cannot plug your electric oven or dryer into the 115 outlet.

Re:also needed for houses

[Anonymous Coward]

ohh, buck converters are only expensive compared to throwing a resistor in it's place. the price of integrated control ic's is really dropping quite fast. it's at the point where you can throw together 30+ amp buck converters for a couple bucks (pun!). and that's for a one off. an sr flip flop, an op amp, diode, oscillator, inductor, mosfet, resistor, and a couple of bjt's for good measure, and you got yourself any kind of converter you want.

Re:also needed for houses

[Khyber]

You can drop 120V DC down a serial string of LED (specced to closely match that voltage in series) all day long.

Re:also needed for houses

[AK Marc]

Why are you delivering 135 amps to charge your cell phone?

Re:also needed for houses

[tombeard]

Something like this:
http://www.thinkgeek.com/gadgets/travelpower/e81a/?srp=4
But centrally wired?

Re:also needed for houses

[AK Marc]

Which doesn't address the "completely insane" comment. I understand it may not be ideal for everything, but there's a difference between "not ideal" and "completely insane."

Re:also needed for houses

[lgw]

I'd think that if you ran it at 12v, and limited devices that plug in to 1A, you be fine. Sure, the TV and PC would still convert AC power, but the 100 other widgets in my house could plug in DC.

Re:also needed for houses

[c 0xFF]

When compared to the cost of a light bulb, a couple of bucks is still a lot. An incandescent costs less than a buck, a fluorescent (the real competitor that LED has to beat) about $3 or less. LED bulbs are still very expensive, and price varies a lot. Let's say $10 for the cheapest. I'm sure the price will come down even more.

In the end, though, you're right. It's easy to forget just how cheap electronics can be. And it really only has to supply 10A or less.

Re:also needed for houses

[MightyYar]

when you could have just run a 12VDC line and powered them all on a central transformer like garden lights.

I had to chuckle a bit... the 5-year-old kitchen in the house we built has an under-counter DC system installed. I think they are quite common. In my case there are a couple of transformers in the basement. No backup battery, though :)

Re:also needed for houses

[MightyYar]

I fail at the internet. Try this link instead :)

Re:also needed for houses

[lgw]

So supply 12V DC with the same 15A (really 12A) limit as AC. Problem solved, and 95 of the 100 non-appliance devices in my house would be fine on DC under that limit. My TV, PCs, and Amps could stay AC.

Re:also needed for houses

[lgw]

High power stuff can stay AC. All my little devices have little powe requirements: cell phone charger, digital clock, DSL modem, KVM switchbox, yadda yadada. Very few devices need more than an amp, if you're not some crazed geek who walpapering his house with laptops.

Re:also needed for houses

[rocketPack]

Indeed, I've done it myself... and if you throw in a cheap-o bridge rectifier, you get twice the brightness!

Re:also needed for houses

[Tapewolf]

I do/have done both (run electrical and understand the implications), and GP has a point. When I think of the things in our house that *must* run on AC, it's only our fridge, freezer, and HVAC. Everything else in the house either converts it to DC or could run quite happily on DC.

I bet they don't all run at the same voltage, though.

Of the things I have that you'd expect to find in a normal home - my monitors and TV are all CRTs. For the monitors it's because I prefer the colour reproduction (want OLED!) and the TV hasn't died yet. I have a number of fans, a microwave oven and a condensing boiler - all of which are liable to die horribly if fed on DC. I'm not sure the printer would be too thrilled with it either since it probably uses a CW voltage multiplier or similar to drive the corona wire and/or laser.

On the more fringe side (a small home studio), half of my wall-warts actually output AC because the device needs to be able to step up to 48v to drive a microphone (I think they use CW multipliers, but I'm not sure). Even some of my synthesizers want 9v or 12v AC - why I really can't imagine since they're basically just computers anyway. I'll ignore the MSR-24 because it's PSU is the size of a small desktop PC and your emphasis is on eliminating small DC loads, I believe. However, even the normal, DC wall-warts output many and varied voltages from 3v-15v - even 36v for the mixing desk. Best case you'd probably be looking at a 12v supply for the house and using regulator ICs to drop it down, but I can't imagine that would be much more efficient than a myriad of switch-mode wall wart supplies.

Re:also needed for houses

[Wonko the Sane]

The last washing machine motor I looked at had multiple windings to achieve different speeds rather than using a VFD.

Re:also needed for houses

[Obfuscant]

but the 100 other widgets in my house could plug in DC.

That's 100A of 12V. From here, you'd need at least number 2 or number 1 wire to carry that current. Ballpark figure.

Same reference, you lose 3 volts for every ten feet of 12 gage at that current (one wire for supply, one for return.) Can your 12V device run happily on 9V?

Sure, the TV and PC would still convert AC power,

The costs of running two power systems in a house would swamp any savings you think you'd make by using DC.

Re:also needed for houses

[Cramer]

Which is cheaper... 100 LEDs or a tiny 10:1 transformer? Answer: the transformer. That's why they do it.

(Even the CREE ultra-high output LEDs only sink 3V. That'd be 40. And they are not even remotely cheap.)

Re:also needed for houses

[Khyber]

No, they do a transformer because their situation is 100% different from mine.

(Protip: Look at my signature.)

It's far cheaper in *MY* industry to do huge banks of diodes.

Re:also needed for houses

[petermgreen]

That is true but while accidental connection (either through miswiring or damaged wiring) of a 120V appliance to a 240V supply may destroy the appliance it is unlikely to cause serious injury.

OTOH connection of a 12V DC system (where noone cares about stopping people touching the live parts of the connectors) to a 120V or 240V system could be a lot more dangerous.

But I don't think that is the biggest issue, you would just have to treat 12V wiring like data/telecoms wiring and keep it seperate from mains wiring.

The real issues with running 12V arround the home are resitive losses (remember these are much worse in lower voltage systems unless you use massive cables), and losses from the extra conversions steps (you have to convert from mains to 12V to whatever the devices want) and the fact that it would cost a lot of money to do for no real benefit.

Wall warts have improved a LOT in recent years, I just looked up a cheap wall wart on rapid and saw "no load current less than 0.5W" as an advertised feature.

Re:also needed for houses

[skids]

If you want that for your house, CAT5 and newer PoE gear will get you almost 30A per device, and do it safely. (PoE being one of the only "X over ethernet" standards I actually have any respect for whatsoever. It's well designed; you'd have to work pretty hard at it to cause a hazard.)

That's not enough for your gaming computer, or these days even your laptop, but it's certainly enough for an efficient lightbulb, APs, cameras, speakers, and a good number of those other "wall wart" devices.

Problem is though, since consumer devices other than APs and cameras don't usually have a PoE port with which to USE the power, you have to convert it -- with a wart-ike device. Back to square one.

Re:also needed for houses

[bkcallahan]

Actually, the voltage drop off of a LED is ~1.2V (This is a physics issue; use germanium, and the drop is 0.8V.) When you run an LED off of a 5VDC line, let's not forget the 330 ohm resistor in series with it, so that the 1.2V appears across the LED. And you guessed it! That resistor has the exact same current running through it, and it's wasted as heat. Bump it to run off of 12V, and you're putting an even larger voltage drop across the resistor (10.8V) wasting heat -- so now you're wasting 90% at 12V.

Re:also needed for houses

[Anne Thwacks]

I have a few relatives staying over and they want to compute while they watch X-factor!

Re:also needed for houses

[Tixover]

If I remember right (and it was a long time ago that I abandoned power system work) the current required to stop the heart varies with frequency - if you graph it you get a sort of upside down bell curve with the lowest current required at about 50Hz. At that time - not sure if it is still the same, several areas in the London underground where DC gear was sited had wooden hand rails that were supposed to be so that you could "drag yourself off" anything that was shocking you and locking up your muscles. My devotion to science didn't run deep enough for me to see if it was actually a practical option...

Re:also needed for houses

[Anonymous Coward]

You can't run high-power LEDs for lighting with a resistor, you need a proper constant-current source. So those waste numbers are nonsense.

Re:also needed for houses

[Hank the Lion]

Voltage drop across a white LED is 3-4 V. Who ever heard of germanium LEDs with 0.8V drop?
And if you have a power LED used for lighting, you don't use a resistor to limit the current; you use a small switchmode power supply.
If you design a good one, you can have electrical efficiency of >90% running from 12V, wasting less than 10%.

Re:also needed for houses

[TheRaven64]

At 48V it isn't too bad though. I'm interested in using DC for lighting because I want to put solar panels on my roof in a few years. I'll be generating DC and storing DC and if I use LEDs for lighting then I'll be using DC as well (and I am for a lot of things anyway) so transforming it to AC in the middle seems a bit pointless. I'd love to switch the lighting circuits entirely over to DC, and maybe run a few extra wires for things that are currently using wall warts.

Re:also needed for houses

[TheRaven64]

Or you have it running at 48V inside the walls and then step it down to 12 or 5V in the sockets...

Re:also needed for houses

[fnj]

LEDs have an extremely sharp current vs voltage knee, and tend to run away thermally. You have to feed them using current regulation, not voltage regulation. That regulation would be wastefully lossy if done using linear regulators, so we're looking at switching regulators. Ta da - the current is gonna have to be AC at the regulation point anyway. There would be a saving, because any design worth its salt would use high frequency AC, not 60 Hz, but the saving wouldn't be as great as you might think.

The native voltage of a white LED junction is around 3.5v. You feed them from as little as 4 volts without limiting the current and they will instantly burn out. Or if you drop it down to 3v they will practically go out.

Re:also needed for houses

[cynyr]

I'll give you it is different... buut you can get a 80% efficient switch mode dc-dc converter on a chip for a few bucks (including the few extra bits you need to run it) in single lots. Imagine what GE can get price wise in their 10,000,000 part order? Heck i have a a dc-dc switch-mode PSU in my cigarette lighter to USB power thing in the car. It costed me $1 including the shipping.

seeing as the AC->DC wall worts seem to be getting much better and smaller i'm not really sure which is better any more.

Re:also needed for houses

[fnj]

You're guessing wrong. AC at around 50-60 Hz is the worst possible form of electrical current for causing ventricular fibrillation, which is essentially 100% loss of consciousness within seconds and 100% fatal within minutes if not met rapidly with intervention in the form of both removal from contact and cardioversion equipment. Microamps of 60 Hz AC directly through the heart for one second will kill you if there is no one to save you. Milliamps through the torso for one second can do the same.

It takes much more DC than that to induce fibrillation, and it is going to be hit or miss with ANY amount of current. Most of the time, brief contact with enough DC passing through it will spasm the heart until the contact is broken, but you have a good chance the heart will restart on its own within seconds of breaking contact unless it literally gets fried. If you stay connected to enough DC, then yeah, the continuing spasm in the heart keeps the circulation at zero for the duration and you will be a goner, but it's nowhere near as likely this will happen as it is that AC will kill you by inducing VF.

And I'm not at all convinced that AC is easier to let go of than a similar amount of DC. With the current going positive to negative and back to positive 60 times per second, it's not like the muscle has any chance whatsoever to relax. Both positive and negative current cause it to contract in a spasm. It's gonna stay gripping the wire if there is enough current.

Re:also needed for houses

[cynyr]

I would think the load of a fridge while running the compressor would be about the same all the time. It turns on, runs at 100% cooling, and turns off again.

Changing the speed of the compressor motor adds all sorts of extra hardware to the refrigeration circuit. I'm willing to bet the single speed fridge has a fixed port expansion device. Swapping to the variable speed setup would at least add a "thermal expansion valve", if not an "Electronic expansion valve" and the EEV controller board, and probably a receiver in the high side as well. It may not actually cool the plate evaporator evenly (meaning freezing food in one part, and warm food in the other.

anyways, just my $0.02 and an HVAC guy, not a refrigeration guy.

Re:also needed for houses

[fnj]

I don't think that's how it works. The AC is going positive to negative and back to positive 60 times a second. Both the positive and the negative current flow cause the muscle to contract. It's not as if half the cycle causes contraction and the other half causes relaxation. And there is nowhere near enough time as the current crosses zero so rapidly for the muscle to even begin relaxing.

One thing about which there is no possible disagreement: 50-60 Hz AC is far more likely to cause lethal ventricular fibrillation if contacted even briefly, than either DC or any other frequency of AC. Essentially nobody spontaneously recovers from VF ever. It takes cardioversion with appropriate equipment to recover.

Re:also needed for houses

[Anonymous Coward]

AC is actually more dangerous, since it will cause repeated contractions. It can induce heart fibrillation easier than DC. Getting "stuck" holding a live AC line is a known phenomenon. My brother had a friend in EE who had to kick out his own leg out from under himself in order to be able to let go.

Re:also needed for houses

[Anonymous Coward]

Go ask a ham radio guy.

No thanks.

Re:also needed for houses

[Anonymous Coward]

One at a time, or all at once? Needing to unplug the switch to plug in the router would suck.

Most wall warts are under 1A, I believe, though I have seen 4A. At 1A, you get 15 devices, at 4A you get 3 devices running on 12A.

Re:also needed for houses

[Cid Highwind]

Fixed-frequency AC motors are inefficient unless the load is constant (and load isn't constant in those applications).

But the load for a refrigerator/freezer/AC should be constant, or close to it. You have a condenser that needs refrigerant delivered at a pressure where it's a gas at inlet temperature and liquid at outlet temperature. Pressure beyond what's required to keep the refrigerant flowing out of the condenser in the liquid phase would be wasted (or harmful because liquid would back up into the compressor). So you've got constant pressure in the condenser with a fixed-volume pump on one and and a fixed orifice on the other... optimum speed for the pump would be constant, right?

Unless you want to get fancy and use a valve to throttle refrigerant flow into the evaporator instead of using a fixed orifice...

Re:also needed for houses

[lgw]

I have more than one wire in my house. Most of my widgets take less than the 1A limit - some mA. Using the same wire gague for AC and DC would obviously be worthwhile, and the same 15A breaker on both would work fine. Are you just being contrary for the sake of being contrary?

Re:also needed for houses

[lgw]

"under 1A" != "1A"

And in any case, 15 devices on a single circuit would be pretty good, though the reality would likely be 50 or so.

Re:also needed for houses

[lgw]

Reality TV, huh? You have my sympathies.

Re:also needed for houses

[Muad'Dave]

Maxim (not the magazine) has nice but expensive buck and boost converters with decent documentation and reference implementations. I used the MAX-756 boost converter on a single solar-cell-powered XBee data node. It boosted 1.2V to 3.3V, no sweat.
Naturally they came out with the MAX1674/MAX1675/MAX1676 parts the day after I purchased the 756.

Re:also needed for houses

[Muad'Dave]

...newer PoE gear will get you almost 30A per device...

You probably mean 30W, not 30A. I agree that it's a pretty good standard wrt safety. I assumed it was just 48V on the pair with no communication. Boy was I wrong!

Re:also needed for houses

[amorsen]

I'm fairly sure that fridges rated A+++ use variable speed compressors in order to achieve sufficient efficiency for that rating. Panasonic makes some.

Re:also needed for houses

[amorsen]

Start/stop is inefficient and should be avoided. E.g "Energiforbrug i kølehuset" which is admittedly in Danish but claims 20% improvements in efficiency when using a frequency controlled compressor in the air conditioning rather than a traditional thermostat start/stop control.

Re:also needed for houses

[sjwt]

From experience, I can say both AC and DC can throw you.. maybe not always, but with two shocks, one from each, both where throw back events..

One from a worn 240 volt stranded power cable, and the other from a 1990 circa Apple monitors' main capacitor.. Both sent me back about 2 meters with a real jolt, though the DC hurt more, but it was one hell of a large capacitor.

Telecom's been doing this for many, many years.

[Anonymous Coward]

DC power is the standard in the telecom industry.

I design systems based around HP's BladeSystem, and the DC power modules just drop in and go. It's very easy, works great, and most of all, my telecom customers love them.

Re:Telecom's been doing this for many, many years.

[Shatrat]

I work with DC power in Telecom and it has 3 huge advantages I can think of off the top of my head:

1) You centralize your rectification. Instead of having hundreds of power supplies running at 80% efficiency, you can have a large rectifier system running at up to 96%.
2) Lead Acid batteries are hugely more reliable and less expensive than equivalent UPS systems, and provide more holdover time. They're still expensive and finicky, but many times less so than a UPS.
3) Any old technician with a brain in their head can run DC power feeds to equipment relatively safely due to the low voltages involved. AC power work of any kind should have a qualified electrician involved.

Re:Telecom's been doing this for many, many years.

[elkto]

I would Mod you Up, but Slashdot is behaving strangely. The only bizarre part of the 48 volt telecom standard is the positive ground. From an Engineers perspective, no stranger than a negative one though.

Re:Telecom's been doing this for many, many years.

[the linux geek]

Yeah, I've never quite figured out why telecoms have standardized on 48VDC while everyone else completely ignores its existence. Most midrange servers (HP Integrity and Nonstop, iirc most smaller SPARC Enterprise boxes, some commodity stuff) are available in 48VDC configurations, so it's not like there's a lack of hardware for it.

Re:Telecom's been doing this for many, many years.

[Anonymous Coward]

Our Uni did a DC battery protected power install for one half of our network core many years ago.

It's amazing how simple DC is to get right compared with more complex AC UPS systems.

The only issue is complacency as it was "low voltage" (48V) system the current draw was quite high.

If you do DC make sure an appropriately qualified electrician does your cabling etc.

Re:Telecom's been doing this for many, many years.

[Anonymous Coward]

The positive ground in telco systems is not bizarre at all: one end of the twisted pair is grounded, and, being at zero volts does not suffer galvanic corrosion. The other end is at -48V and benefits from cathodic protection: it's the anode that generally gets corroded in a galvanic cell.

Re:Telecom's been doing this for many, many years.

[vlm]

4) Done right with a positive ground system, leads to less corrosion problems with outside plant. Admittedly "inside" the data center, if you're got corrosion, you're doin it wrong.

5) Less AC hum. We had some microwave site to site short hop gear back in ye olde NTSC days that could only be run off battery without 60 hz interference bars on the screen. Not technologically relevant anymore, but the point remains that DC is always going to be cleaner than AC.

6) Better lightning protection. I'm sure its happened, but I've never heard of losing a telco DC bus. Big conductors, giant batteries across them, lightning is just not an issue anymore at the power level (still need to ground feedlines / waveguide / whatever you've got at home like that)

7) dump most of the power conversion heat in the battery room where its all built to handle high temp and no one visits (other than occasional battery maint). Cheaper cooling in the data center, data center is somewhat more habitable, etc.

Re:Telecom's been doing this for many, many years.

[uncledrax]

^ This ^
Although we're a ISP grown from a Telecom.. so we have large DC plants at most of our sites anyway, so for us, it's just asking our vendors for DC options... most vendors DO have DC PSU options.. it's more of a pain to adapt to the annoying 21" 2-post telecom racks we have.. :/

Re:Telecom's been doing this for many, many years.

[mjwalshe]

because telcos have much much more stringent requirements for up time - a major switch failure is a once in a lifetime event for most people - we have had 3 failures for amazon in the last 4 months at work.

Re:Telecom's been doing this for many, many years.

[the linux geek]

Oh, certainly. My post was more a question of why everyone else has ignored 48VDC, not why telcos use it.

Re:Telecom's been doing this for many, many years.

[Anonymous Coward]

Yeah, except DC in telecom (20+ year veteran here) is neg 48 voltage. Always fun when some outside electrician wires stuff up backwards.

Re:Telecom's been doing this for many, many years.

[Anonymous Coward]

2) Lead Acid batteries are hugely more reliable and less expensive than equivalent UPS systems, and provide more holdover time.

Ummm, what do you think UPS systems use? Lead-acid batteries.

You probably mean that 48V DC UPS systems are much simpler than AC UPS systems since you don't have to convert from AC to charge the battery, then discharge the battery & convert to AC to power your equipment.

Re:Telecom's been doing this for many, many years.

[Shatrat]

UPS systems do use lead acid batteries...PLUS large input filter capacitors, and rectifier capacitors which have to be serviced every 5-7 years and can potentially fail catastrophically (boom) if you do not. I also work with large UPS systems for a few of our data centers, up to 300 KVA, and they are much more expensive, require more frequent maintenance, and don't provide the hours of holdover time you can get out of the same footprint of large VRLA or flooded batteries.

Re:Telecom's been doing this for many, many years.

[markhahn]

alas, PSUs have been at 94% for several years now.

also, what kind of batteries do you think are _in_ a UPS?

anyone can do low-current DC. you want a fairly specialized, fully-qualified electrician doing high-current DC, which is what we're talking about.

Re:Telecom's been doing this for many, many years.

[michael_cain]

Decades ago, I was in a telco central office where multiple electromechanical switching systems (and associated relay logic) were powered off 48VDC. The main power buses down in the basement near the battery farm consisted of multiple copper 2x6s strapped together, with about 18 inches between ground and -48VDC. At one point I noticed that there were two little stubs sticking an inch or two out from the copper and asked about them. The craft got a big grin and explained that a newbie had been standing on those power runs, doing something with a steel pry bar, and had dropped it -- and it landed across the conductors. He said it was like a giant flash bulb going off. 48 volts, about a bazillion amps, and a one- or two-ohm conductor shorted across it...

Re:Telecom's been doing this for many, many years.

[mirix]

1) You centralize your rectification. Instead of having hundreds of power supplies running at 80% efficiency, you can have a large rectifier system running at up to 96%.

However each workstation still has a PSU in it. instead of converting from 120V AC to 12/5/3.3V, it now converts from 48V to 12/5/3.3V. And still suffers losses - you have to add these to your figure.

Also, 48V needs heavier wire to pass the same amount of power. Copper is bloody expensive these days.

Re:Telecom's been doing this for many, many years.

[Khyber]

"I've never quite figured out why telecoms have standardized on 48VDC while everyone else completely ignores its existence"

I think the music industry would like to have a talk with you regarding their 48V phantom power for XLR microphones that is built into many mixer boards.

Re:Telecom's been doing this for many, many years.

[WaffleMonster]

I work with DC power in Telecom and it has 3 huge advantages I can think of off the top of my head:
3) Any old technician with a brain in their head can run DC power feeds to equipment relatively safely due to the low voltages involved. AC power work of any kind should have a qualified electrician involved.

Short the posts of a car battery, count the number of milliseconds it takes the pliers to be welded to the leads... Then come back here and tell us all about how low voltage is "safe" and requires no qualifications.

Re:Telecom's been doing this for many, many years.

[OrangeTide]

a UPS has an inverter in it. but banks of batteries for DC power is tough, like an old german u-boot.

Re:Telecom's been doing this for many, many years.

[Agripa]

Oh, certainly. My post was more a question of why everyone else has ignored 48VDC, not why telcos use it.

There are at least two reasons.

Internally, 120/240 volt AC switching power supplies rectify to 340 volts or slightly higher if boost power factor correction is used. They will already work with 380 volts DC although you would want to remove the rectifier or boost PFC circuit since they are not needed.

The other reason is for smaller copper distribution requirements than for 48 volts.

Re:Telecom's been doing this for many, many years.

[GumphMaster]

Having witnessed the demise of a mere 5kVA UPS in the bottom of an HP server (in a telephone switch building) I can certainly attest to the boom-ness of the large capacitors. It was a great light show that would have been better had it not been the new replacement unit (for the already failed original) that went boom. The noise drew attention from all over the floor. I can also attest to the need for spare underwear on the part of the HP tech :)

I am quite sure that accidentally shorting the DC bus bars running about the ceiling would have been a _very_ lively show.

Re:Telecom's been doing this for many, many years.

[Tomato42]

...that's one huge market. I assure you that most technicians working with stage equipment don't know that XLR connections are XLR connections, let alone general population.

Re:Telecom's been doing this for many, many years.

[Tomato42]

I'm no electrician, I'd guess the low voltage DC is safe because you can grab uninsulated leads in your hands and nothing will happen. Hell, I touched live wires in 48VDC with my tongue without any long-term effects!

Re:Telecom's been doing this for many, many years.

[Anonymous Coward]

Not to take away from your point but the new servers I just purchased have a 94% efficient power supply (supermicro platinum). With the cost of power, cooling and density the days of 80% efficient power supplies are over, at least for us.

Re:Telecom's been doing this for many, many years.

[Falconhell]

Back in my early days as a telecom apprentice, one of the tricks we got up to was removing the carbon center
conductor from a big dry cell, mounting it on 1 50V DC busbar via a large nut then screwing it out until an arc was drawn, instant carbon arc lighting. Dropping a shifting spanner on the busbars gauranteed a HUGE BOOM, and a vapourised spanner.

Re:Telecom's been doing this for many, many years.

[Falconhell]

I think the telecoms industry would like to have a word with you about the invention of phantom power. Phantom power cicuits were orginally designed for putting 2 48V phones down 1 set of lines!

Re:Telecom's been doing this for many, many years.

[Falconhell]

As an experienced sound engineer I disagree, every one I know in the industry is well aware of what XLR's are.

We fix the damn things all the time!

Re:Telecom's been doing this for many, many years.

[bill_mcgonigle]

We fix the damn things all the time!

That reminds me, I have an XLR cable sitting in a bucket that's all crackly. Just a straightforward re-solder?

Re:Telecom's been doing this for many, many years.

[petermgreen]

3) Any old technician with a brain in their head can run DC power feeds to equipment relatively safely due to the low voltages involved. AC power work of any kind should have a qualified electrician involved.

I have to disagree here the lower voltage does mean the risk of electric shock is (mostly) removed but it also increases the current involved and hence increases the risk of fire. DC is also far more prone to arcing than AC. Lead acid batteries can be especially nasty things due to high short circuit currents.

It may be that in some jurisdictions there is less regulatory overhead to working on a DC system due to the fact that less people are doing it (regulations are generally created as a result of people fucking up) but I wouldn't really say it is safer to work on.

Re:Telecom's been doing this for many, many years.

[LinuxIsGarbage]

What's bizarre is that original theory was electricity was the movement of positively charged particles. It ended up being negatively charged particles moving around. With most science when you make a mistake you correct it. Circuits? Nope, we continue to pretend it's positively charged particles moving from positive potential to negative potential.

Re:Telecom's been doing this for many, many years.

[Wonko the Sane]

The thing is that it doesn't really matter which way you conceptualize it, except for some specific situations like trying to understand how a vacuum tube or CRT works.

Re:Telecom's been doing this for many, many years.

[petermgreen]

Afaict there are a few issues with 48V DC for a dataceventer.

1: it's nonstandard (in the computer industry) so you pay a premium for equipment that runs off it and reduce your choice of equipment.
2: It's lower voltage so for a given level of tolerable loss your cables have to be much bigger
3: It's DC so it's more prone to arcing making all your switches and protective devices more expensive and basically ruling out the use of plug and socket connections for anything other than final connection of individual devices

The combination of these factors make a 48V DC system expensive.

Re:Telecom's been doing this for many, many years.

[LinuxIsGarbage]

if it was a one ohm conductor there would be 48 amps through it.

I've twice seen a socket wrench that was welded to 460V. One was across the busbars on the secondary side of a transformer.

Re:Telecom's been doing this for many, many years.

[petermgreen]

Safe to touch? yes.
Safe to drop metal objects on? HELL NO

Re:Telecom's been doing this for many, many years.

[Anonymous Coward]

8) All the problems you get with the finicky batteries in #2 above, are present, and WORSE in a large AC UPS because you're draining more current over 15 minutes instead of a couple hours.
9) Easier scaling - most people miss this one totally. Want to add another battery bank? at 48VDC, just set it up, charge it and plug it in. Want to add another 200Amps of rectifier power? Same deal. With AC, if you add another generator you need to match voltage, frequency, phase and rotation all at once in order to have it not catch fire when it connects. That takes matching electronics that often limit you to using the same vendor across all your AC plant for generators and UPS units. DC power is just incredibly easy to add redundancy and extra capacity versus AC.

One major downside to DC power is it's not as common, and so it's often misunderstood. Within a building, your runs are usually short enough that the larger copper conductors are less expensive than the conversion back to AC in the large UPS.

Re:Telecom's been doing this for many, many years.

[Engeekneer]

1) You centralize your rectification. Instead of having hundreds of power supplies running at 80% efficiency, you can have a large rectifier system running at up to 96%.

We actually just got new server hardware. The power supplies are rated for 92%+ and in practice, we measured the efficiency to be almost 95%. So while our other points are valid, I think the losses from power supplies have gone down a lot.

Re:Telecom's been doing this for many, many years.

[TheRaven64]

Telcos use it and have very high uptime requirements. Therefore, all of the stuff available for it is designed for very high uptimes and therefore very expensive. Therefore it's cheaper for anyone who doesn't have such strict requirements to use cheap commodity crap, which isn't available in 48V DC form. Get a few big datacentres to run 48V DC and charge less to customers who use it and you'll see servers start to be available...

Re:Telecom's been doing this for many, many years.

[Electricity Likes Me]

Exactly. For most things it's irrelevant since the physical movement of the electrons is a secondary effect - it's the electric field which does all the work. The drift rate of electrons in a wire tends to be on the order of centimeters per second.

Re:Telecom's been doing this for many, many years.

[Electricity Likes Me]

3: It's DC so it's more prone to arcing making all your switches and protective devices more expensive and basically ruling out the use of plug and socket connections for anything other than final connection of individual devices

The combination of these factors make a 48V DC system expensive.

This is the exact opposite from a comment on arcing that I saw earlier on this topic. Thinking about the physics of it I'm not quite sure it makes sense: AC has a much higher local difference in the electric field (due to the oscillation), which would be more likely to cause ionization of gases (which would lead to arc break down).

Re:Telecom's been doing this for many, many years.

[Khyber]

If it crackles near the connector, it's likely that you can fix it with a simple solder job. You might lose a few inches of cabling but that's better than shelling out for a new one, especially if you're proficient in soldering.

Re:Telecom's been doing this for many, many years.

[Anonymous Coward]

#1 used to be true, but isn't so much anymore.

I can tell you - for a fact - that HP's DC offerings (centered around BladeSystem) carry NO price penalty for DC power - or even NEBS compliance, if that's your cup of tea.

It's the same exact BladeSystem chassis that everyone else uses; the power input module is swapped from a 3-phase AC (or, egads, 1-phase AC) to a DC input module, and that's that. Same price, same components, no penalty - price or performance - for using DC power.

It's great stuff. Should be MUCH more highly publicized than it is.

Re:Telecom's been doing this for many, many years.

[lanner]

ANother advantage of DC power cable is that it is almost always cut to need, on the spot, and it's potentially as easy to plug in as a speaker wire. The cable management aspect is awesome if you work in high-density server environments. Anyone who has racked 42 servers in a 42U rack knows how awesome cable management is. Specifically, it's about slack management. No slack = no problem.

Re:Telecom's been doing this for many, many years.

[petermgreen]

My understanding is that is is much harder to sustain an arc with AC because the current drops to zero twice per cycle.

And in related news...

[Anonymous Coward]

Tesla rolls over in his grave

Re:And in related news...

[nurb432]

Nah, he will just be proven right, again.

Re:And in related news...

[DickBreath]

Would an AC current be generated by Tesla spinning in his grave?

Re:And in related news...

[Bucky24]

Depends on how you harness the energy I would imagine... Though I'm not very familiar with electrical systems.

Re:And in related news...

[blair1q]

At what rate? And AC or DC? If AC, he could be inducing an earthquake.

Google 12VDC proposal better.

[Animats]

There's no particular reason that 380 VDC distribution should help efficiency. You still need about two more levels of switching power supply before power reaches the ICs.

Google's proposal that motherboards should need only 12VDC made more sense. Drives already run on 12VDC, and there's already a level of power conversion near the CPU to get the desired CPU voltage. The USB devices do need +5, but a 12VDC to 5VDC switching converter can handle that. And single-voltage power supplies are more efficient and simpler than multi-voltage ones.

Re:Google 12VDC proposal better.

[Anonymous Coward]

This is my biggest complaint with the USB standard. Why the hell not 12v?!

Re:Google 12VDC proposal better.

[Mr Z]

The current carrying capacity of the wires would need to be about 30 times larger, though, to deliver the same amount of power. That's pretty huge. To go to 12v everywhere, you'd need huge current-carrying wires everywhere (think "as big as your car battery cables or bigger"). To carry 1kW through a 380V line, you only need to handle 2.6A. To carry 1kW through a 12v line, you need to handle 83A. And that's just one beefy server.

Now think of your house wiring. Outside of your major appliances, where do you see runs higher than 15A or maybe 30A? There's a reason high voltage is good.

Re:Google 12VDC proposal better.

[Anonymous Coward]

what about salt water filled copper pipes for the conductors?

Re:Google 12VDC proposal better.

[mjwalshe]

bollocks do you not understand ohms law - running any thing non trivial on 12v dc is a non starter - any telecom engineer will(and thats on 48V have amusing horror stories about near death (if they where lucky) with big dc systems and that is using 48V

Re:Google 12VDC proposal better.

[vlm]

The original low speed USB electrical spec was pretty much classical 5 volt NRZI TTL. So insisting on 12 volt supply would mean every USB device would require a 5 volt regulator inside it to talk to the data lines, and the data lines would need protection circuitry on both momma boards and all USB devices because TTL traditionally gets really pissed off when an input voltage rises about its power voltage in case of a short. CMOS gets pissed off too at over voltage. It would just be a bad scene.

Something like RS-485 but really faster would have been "better", but ...

Re:Google 12VDC proposal better.

[hpa]

USB is designed to enable inexpensive devices. 5 V is so that when cable losses are counted in, you can use a LDO linear regulator to obtain 3.3 V (Vcc) without excessive losses. 12 V would require a switching regulator.

Re:Google 12VDC proposal better.

[Wonko the Sane]

what about salt water filled copper pipes for the conductors?

Excellent idea! Hydrogen gas, oxygen gas, chlorine and an ignition source all in the same package. What could possibly go wrong?

Re:Google 12VDC proposal better.

[amorsen]

There's no particular reason that 380 VDC distribution should help efficiency. You still need about two more levels of switching power supply before power reaches the ICs.

It is easier to step down DC than AC. You don't have to contend with keeping current up during the parts of the wave where AC supplies close to zero power. Instead you just chop at a sufficiently high frequency that your ripple current gets sufficiently low.

Re:Google 12VDC proposal better.

[Forbman]

car battery cables are thick because they carry 500+ amps when you're starting your car. None of the other 12V cables in your car are nearly that thick; they're 18-20 gauge wires. Even the wires going to the headlights are pretty reasonable.

Re:Google 12VDC proposal better.

[Anonymous Coward]

There's no particular reason that 380 VDC distribution should help efficiency. You still need about two more levels of switching power supply before power reaches the ICs.

Google's proposal that motherboards should need only 12VDC made more sense. Drives already run on 12VDC, and there's already a level of power conversion near the CPU to get the desired CPU voltage. The USB devices do need +5, but a 12VDC to 5VDC switching converter can handle that. And single-voltage power supplies are more efficient and simpler than multi-voltage ones.

The higher the voltage, the lower the current. The lower the current the smaller the wires can be and the lower the losses due to resistance in the wire.

And regulating a higher voltage down to a lower one involves negligible losses compared to converting AC to DC.

Re:Google 12VDC proposal better.

[Anne Thwacks]

When USB was invented, most devices were still 5V anyway.

Re:Google 12VDC proposal better.

[hawguy]

car battery cables are thick because they carry 500+ amps when you're starting your car. None of the other 12V cables in your car are nearly that thick; they're 18-20 gauge wires. Even the wires going to the headlights are pretty reasonable.

A 360 watt server needs 30 amps @ 12V. Put 30 of them in a rack and you need 900 amps @ 12VDC.

Divide by 10 for 120 Volts (AC or DC, doesn't matter): 3 amps for one server, 90 amps for 30 servers.

Divide by 2 for 240 Volts: 1.5 amps for one server, 45 amps for 30 servers.

Re:Google 12VDC proposal better.

[OrangeTide]

it's your fault for not buying firewire, you can pull 45W of power over a firewire cable.

Re:Google 12VDC proposal better.

[petermgreen]

You can't easily get a solid 5V supply out of a USB port. Some cheap and nasty devices do run parts of their electronics directly off the USB 5V line (indeed i've done it myself) but doing so is not how you make a solid reliable device unless you are sure all your parts can work down to 4V or so.

USB itself is a 3.3V interface (not 5V as vlm asserts) so pretty much any device needs a 3.3V regulator even if it's just to power the USB physical interface.

Re:Google 12VDC proposal better.

[thsths]

> Google's proposal that motherboards should need only 12VDC made more sense.

True, but you cannot run a data-centre on 12V, especially not with current copper prices. For distribution, you need x00 V.

I would say: get a three phase rectifier (or two phase in the US?), and see whatever near constant DC voltage comes out of that. In Europe that will give you about 550V with a slight 150 Hz distortion. Step it down to 12V near the board, and down to 0.8V or whatever on the board.

The key advantage of three phase rectification is that you automatically get a very high power factor, and you can avoid costly (and expensive) power factor compensation.

Ideally you would run the board with more than 12V, too. But that requires different hard disks, different board, different fans etc, and there is no clear competing standard. 28V, 35V, 42V, 48V - they all advantages and niches.

Re:Google 12VDC proposal better.

[TheRaven64]

you can pull 45W of power over a firewire cable

Well, in theory. The cable will support it, but every peer on the chain is allowed to produce or consume up to 45W and there's no guarantee that any of them actually will produce the power. The first device I owned that had a FireWire port was a laptop with a 65W PSU - no chance of that providing 45W to the FireWire port...

Re:Google 12VDC proposal better.

[OrangeTide]

yes, it's variable, but I fail to see your point. USB ability to provide power is also variable. so is power-over-ethernet. at least these devices actively report that they cannot meet a peripheral's power demands. with a simple wall transformer it just gives you the wrong voltage and overheats if you try to draw too much power from it.

Re:Google 12VDC proposal better.

[Lennie]

Kind of related I came across this video today:

http://cmstudio.coolermaster.com/video_play.php?lang=en&cid=3&vid=24

Death

[Anonymous Coward]

But won't more geeks die due to more harmful shocks, as DC (high voltage/amperage) becomes more common? I thought DC was more likely to burn your nerves when you compare the same AC to DC potential.

why 380v?

[wierd_w]

Wouldn't it make more sense to drive at 12v with an insane amperage behind it, than to drive at 380v and garantee the necessity of a voltage regulator rated for high voltages?

I mean, the whole reason for doing away with ac current was to eliminate the rectifier and regulator circuits, which belch heat into the data center. Using 380v, which no datacenter device that I know of uses natively (well, maybe the innards of a crt, but that's actually much higher than 380v... AND a deadend tech.), seems kinda... well.... unproductive.

Is it because of impedence problems or something?

Re:why 380v?

[DerPflanz]

High current ('insane amperage') needs very thick cabling. Not very cheap or efficient. For transport, high voltage AC is the best choice. That's why transport networks use that.

I didn't do the math on DC transport/distribution in datacenters, but it at first glance it does need high voltage for transport, just to keep the cabling anywhere near affordable. Change to lower voltages when needed.

Re:why 380v?

[Anonymous Coward]

I hope the backlight in my LCD isn't deadend tech. These things are hard to see without the 700V inverter keeping the light on.

Re:Why 380v?

[Anonymous Coward]

http://hightech.lbl.gov/dc-powering/faq.html

6. Why have you chosen 380 VDC?

We have chosen 380 VDC because losses are lower than 48 VDC and server power supplies today convert AC to 380 VDC. It would be a simple process to manufacture power supplies to accept 380 VDC.

Re:why 380v?

[Urban Garlic]

Basic Ohm's law -- the resistive loss through a DC wire is the voltage drop across the wire, times the current through the wire. But the voltage drop across the wire is proportional to the current, it's just I*R, so the total power dissipated in the wire itself (i.e. not transferred to the load) is I*I*R. So, you want the current going to the load to be as small as possible. But, of course, the load still needs to get all the power it needs, so the operating voltage (which is distinct from the through-the-wire voltage *drop*, of course) needs to be higher if the current is lower.

So, high operating voltages reduce distribution losses.

The same analysis works for AC too, and is the reason that trans-continental transmission wires have such crazy-high voltages. AC has additional losses due to radiation and induction, of course.

Re:Why 380v?

[RichMan]

440 * sin(120) = 381.05 ....

3 phase has 2 ways of looking at the voltages, Y or delta.
The 3 phase delta is 440v when you measure between any pair of the 3 wires. The center point is ground. You don't see that in delta, but you do when measuring it in Y form. The same signals that are 440v when measured as a pair are 3 x 380v when looked at in the Y configuration.

So 3 phase 440v gives you 3x 380v to ground.

As to the 12v/5v/1.5v/ whatever you are going to have to do DC to DC all over the place. Better to have as high a voltage as possible for less current and less losses.

Re:Why 380v?

[Mr Z]

Well, it would allow for about a 10% sag in the AC source, if that was 440v. 90% of 440v is 396v, which gives you some margin for conversion losses.

Re:Why 380v?

[Anonymous Coward]

You can just look up HVDC for the reasons why.

Re:why 380v?

[Anonymous Coward]

That is basically Google's proposal for a standard in power.

As for why they want a high voltage, it's simple. Low voltage dc is terribly inefficient when traveling even moderate distances. There was one article i remember, the gains you got from having a central more efficient adapter, you lost due to requiring, at minimum, several feet of wires between the adapter and the various server racks.

It might be possible that the efficiency of ac to dc saves you quite a bit as dc to dc (especially when dropping voltage) is relatively simple and easy (more efficient basically).

Basically:
A 120v ac to 380v dc adaptor for sections/all of the datacenter
380 dc to 12v/5v/3.3v dc adaptor for each mb

vs

120v ac to 12v/5v/3.3v dc adaptor for each mb

While the above seems more complex, dc to dc conversion is simple when it's just dropping the voltage meaning you the redundant parts are also the more simpler parts while the more inefficient conversion is centralized to be more efficient. This basically is a system designed to avoid power loss through the distance between the adapter and servers.

Re:why 380v?

[BZWingZero]

Sorry to tell you this, but it is. Newer LCDs have LED backlights which don't require that pesky inverter.

Re:Why 380v?

[russotto]

The article says that 380v DC is the sweet spot, but why? Here in the US 440v (3 phase) AC is pretty common, as is 220v AC. I realize there's a world of difference between AC and DC, but that's about all I can think of.

They've probably figured that common power supplies designed for 240VAC can be run off 380VDC by bypassing the rectifier diodes. Doing the math gets you 340V, but maybe they've looked at the actual devices available or in common use.

Re:why 380v?

[mjwalshe]

its the Amps that kill - and 12v at insane amps would require very very thick wires

Re:why 380v?

[hpa]

380 V is presumably because it is the phase-to-phase voltage in 220 V 3-phase wiring, so there is plenty of distribution products already rated for that voltage. Why 380 V and not 400 V (230 V 3-phase, which is the modern standard) is a good question; it might have been what they actually intended.

Re:why 380v?

[amorsen]

For transport, high voltage AC is the best choice.

For transport, high voltage DC is the best choice. No impedance worries, no phase to keep in sync over thousands of km. AC is legacy.

Re:Why 380v?

[vlm]

The article says that 380v DC is the sweet spot, but why? Here in the US 440v (3 phase) AC is pretty common, as is 220v AC. I realize there's a world of difference between AC and DC, but that's about all I can think of. 380/4=95 x 4v rails I suppose? Someone with an EE degree or master electrician jump in here and explain this to me please.

OK

Look at the input stage of a stereotypical switchmode power supply. in 120 volt mode you see a voltage doubler config. In 220 volt config you see a plain ole straight rectifier. (If you ever wondered why you can run a switcher configured for 220 on 120 with no fireworks, but config for 120 and plug into 220 and it blows up, now you know) Your DC voltage to the input of the switch mode chopper is gonna hover right around......... 380 volts. No point getting overly precise because line voltage and component tolerance is not overly precise, but pretty much you take an off the shelf switcher, rip out the input rectifier and filter, dump in raw 380 VDC and off ye go.

Now the "pretty much" relates to all kinds of fun like inrush current limiting for the chopper may have relied on the resistance of the rectifier and filter, or maybe your chopper controller runs off a tiny little AC transformer.

But to a first approximation, if you're reading this on a desktop PC, you are already using a more or less 380 volt DC power supply, so its pretty easy to bodge anything using 380 VDC.

Re:why 380v?

[Anonymous Coward]

To transport the same power like an AC line a low voltage DC line needs to carry a lot more current (P = U * I). That requires very thick cables (expensive and cumbersome to handle). BTW there is already a DC standard in telecom's equipment (-48V DC). I think the idea behind the 380V DC is to use the same wiring as AC and to have only small design changes for the PSUs. An input of 230V AC gives you about 320V DC after rectifying. So the standard AC PSUs would need only a small design change to take 380V DC. To improve efficiency a little bit more the rectifier may be removed. Or the PSU could be designed to take 230V AC and 380V DC.

Re:Why 380v?

[vlm]

Here's a google provided whitepaper on the exact topic:

http://www.directpowertech.com/docs/Whitepaper_AC_units_on_DC_rev%20A.pdf

Why they put something titled "not for public distribution" on a google accessible server is a mystery. If your country has an extradition treaty with Sweden then google for another whitepaper instead of this one. Whatever.

Re:why 380v?

[doesnothingwell]

At the pc power supply's first conversion stage the 110 vac is rectified and or doubled then smoothed to (~250-380 vdc) then chopped and transformed to 12v an 5v. Further filtering and more conversion to 3.3 v and lower depending on the motherboard. Lower dc voltages and higher currents are harder to switch on and off in case of a malfunction. There is much to consider about where, when, and how much dc voltage you want. My home solution would differ from that of a large data center, but 380 vdc sounds ok for this application. Standard pc power supplies can consume this with little or no modifications.

Re:why 380v?

[Anonymous Coward]

Yes.... but even if your source resistance is zero, your body still presents a resistance and it's the applied voltage that dictates what the current flowing through you would be

Re:why 380v?

[ferongr]

And subsequently have a bluish white-point (non-optimal), smaller gamut (because of the light spectrum produced by the LED) and can be tiring at low brightness settings for some people (including me) because of the PWM brightness control. No professional-grade monitor uses white-LED backlighting.

Change is not always for the better.

Re:why 380v?

[folderol]

I would hate to see a 380V DC system. It would be extremely dangerous for two reasons. Under fault conditions the slightest spark would rapidly become a very hot maintained arc, whereas AC is snubbed every half cycle so needs a much greater failure for a plasma based arc to be maintained. An electric shock would be guaranteed death. DC locks your muscles. AC makes them vibrate and often actually causes you shake away from the source. DC also sets up damaging electrochemical reactions in the body. AC doesn't.

Re:why 380v?

[Anonymous Coward]

The trick with 380V DC is that you can run it directly into most 240V AC power supplies and it works fine. (most AC power supplies start off with a bridge rectifier to convert the 240V AC into 380V DC).

You can still float a string of batteries across the 380V DC feed, so you get all the advantages of DC, but get to use cheap 240V AC power supplies in the servers. As an rexample, for HP blade servers, the DC power supply nodules are four times the price of the AC power supply modules.

Re:why 380v?

[WaffleMonster]

Wouldn't it make more sense to drive at 12v with an insane amperage behind it, than to drive at 380v and garantee the necessity of a voltage regulator rated for high voltages?

I mean, the whole reason for doing away with ac current was to eliminate the rectifier and regulator circuits, which belch heat into the data center

Only if you run a copper mine :)

Re:why 380v?

[Agripa]

Using 380v, which no datacenter device that I know of uses natively (well, maybe the innards of a crt, but that's actually much higher than 380v... AND a deadend tech.), seems kinda... well.... unproductive.

Any datacenter using 120/240 AC power supplies has 340 to 380 volt DC all over the place but it is internal to the power supplies. Most of them will run on 340 to 380 volts DC already without changes.

Re:why 380v?

[Vegemeister]

The problem is not PWM, but PWM at low frequency.

Re:why 380v?

[lgw]

DC is better for connecting single endpoints at a distance(especially superconducting DC). It sucks for a grid, as splitting a cable off a pole becomes non-trivial.

Re:why 380v?

[lgw]

It's all going OLED eventually anyway. But for cheap non-IPS monitors that don't even try to keep colors the same at different viewing angles (i.e., most monitors sold), led backlighting is better, simply because it doesn't burn out after 2 years!

Re:why 380v?

[Wonko the Sane]

It sucks for a grid, as splitting a cable off a pole becomes non-trivial

Why? Because there are only two wires instead of three?

Re:why 380v?

[ferongr]

At very low duty cycles(5-15%) a very high frequency is required for flicker-free operation and since most LED backlit displays are either low-end TNs or TVs, manufacturers find it hard to justify the a better IC that would increase the cost of BOM significantly. Dell's 2*12 IPS series doesn't suffer from flickering, but most other LED-backlit monitors (and even many mobile phones) do.

Re:Why 380v?

[ran-o-matic]

440VAC 3 phase is not common in the US. 480VAC is common in the US. 380VDC is a common intermediate voltage for switching power supplies. This means that a 380VDC input supply can be made using a relatively small mod to an existing supply.

Re:why 380v?

[ferongr]

I haven't had CCFL monitors fail on me. CCFL lifespan (like home fluorescent lighting) is more influenced from on/off cycles than power on hours. Setting the monitor's at sleep at 30 minutes (as opposed to 5 minutes) significantly improves the lifetime of the backlight.

Re:Why 380v?

[Wonko the Sane]

440VAC 3 phase is not common in the US. 480VAC is common in the US

It's really the same thing.

When you're talking about industrial power distribution numbers like 440 or 480 should be thought of as referring to a category rather than a specific voltage. The devices that are made to work with that kind of power assume that actual values could range anywhere from about 430-490 since in practice there is a large amount of variation, especially in older facilities.

Re:Why 380v?

[Hadlock]

I was under the impression that 440v 3 phase is pretty common in industrial areas, particularly when using heavy duty electric motors, etc. Is this not the case? Another commented pointed out that they're essentially a "class" of industrial power.

Re:why 380v?

[petermgreen]

Afaict the main issue is not the actual splitting of the wiring it's what you do with the connection afterwards. Dealing with high voltage DC requires far more equipment than dealing with high voltage AC. So DC works well for systems with long lines connecting a small number of points, but would be prohibitively expensive for the regular grid (which connects many points through a series of transformers that gradually increase/reduce the voltage).

Re:why 380v?

[petermgreen]

The trick with 380V DC is that you can run it directly into most 240V AC power supplies and it works fine. (most AC power supplies start off with a bridge rectifier to convert the 240V AC into 380V DC).

Be that as it may I think you would still want a DC certified PSU for a few reasons

1: Some power factor correction systems may respond badly to DC supply
2: Some bridge rectifiers may not be specified to handle having the current running through the same pair of diodes all the time
3: IEC connectors are only designed for AC, not for the (far more arc prone) DC, you would probably want some form of locking connector instead

Re:why 380v?

[Wonko the Sane]

If the ~10,000 V lines that you see bringing the power into your neighborhood carried DC instead of AC the only difference it would make is that instead of a transformer feeding wires that supply your house it would be an inverter.

Re:why 380v?

[lgw]

Well, running DC for the "high tension" lines would probably make sense (there's often more than one step down before it gets to your house), and simplify phase matching over really large power grids.

Re:why 380v?

[lgw]

I've never had on last more than 2 years, gone through 4 or 5 now. But I've switched to IPS monitors, which I suspect will have much higher quality parts. For TV, of course plasma is the One True Display (until OLED matures).

Re:why 380v?

[petermgreen]

instead of a transformer feeding wires that supply your house it would be an inverter.

does an inverter with 10KV DC input and 240V/415V 100A three phase "true sinewave" AC output (or whatever your local standard is) even exist?
how much does/would it cost?
how reliable would it be?
how bulky would it be?
how would it handle sitting on a pole for years exposed to wide temperature swings?
would it be able to handle power flow in both directions?

Re:why 380v?

[Wonko the Sane]

does an inverter with 10KV DC input and 240V/415V 100A three phase "true sinewave" AC output (or whatever your local standard is) even exist?

That's a ridiculous question.

Why 380v?

[Hadlock]

The article says that 380v DC is the sweet spot, but why? Here in the US 440v (3 phase) AC is pretty common, as is 220v AC. I realize there's a world of difference between AC and DC, but that's about all I can think of. 380/4=95 x 4v rails I suppose? Someone with an EE degree or master electrician jump in here and explain this to me please.

conversion on the motherboard?

[bigtrike]

Wouldn't you just end up putting a switching power supplies elsewhere and create heat problems, then?

Re:conversion on the motherboard?

[compro01]

You already do that anyway. Motherboards take most of their power at 12V and regulate it down to what's needed.

A modern CPU can use up to 65-ish watts, but runs at about 1.5V, so you're needing 43.3 amps of current. You're not going to be running that little voltage and that much current down a reasonably sized wire of any useful length.

Videocards do the same thing.

You'd need much larger conductors

[bigtrike]

Lower voltages require larger conductors to carry the same current. Copper isn't that cheap.

Re:You'd need much larger conductors

[Anonymous Coward]

Lower voltages require higher current to carry the same power. More current requires a larger conductor.

Re:You'd need much larger conductors

[wierd_w]

Skin effect.. that's right.

But you can get around that with multistranded wire, right? A bundle of 7 small conductors netting the same approximate volume as 1 big conductor has substantially more conduction surface.

Re:You'd need much larger conductors

[Anonymous Coward]

Yup. If you need to provide 200 watts, at 380V you can do it with only about .5 amps, easily carried on 26- or 28-gauge wire. But at 12V, you'd need to allow for 16 amps, for which you need 12-gauge or larger.

Re:You'd need much larger conductors

[nzac]

Lower voltages require larger conductors to carry the same power, (due to I^2*R losses). Copper isn't that cheap.

FTFY

Losses in a wire are determined by the current alone.

Re:You'd need much larger conductors

[nzac]

No the skin effect is for (High frequency) AC.

For DC impedance is determined by the material and the cross section area.

It does make the cables easier to bend though.

Re:You'd need much larger conductors

[pclminion]

It's not skin effect, it's the fact that P = VI. If the machine draws 500 watts at 12 volts, the current in the conductor must be 42 amps. The power dissipated in the conductor is I^2*R. Suppose you're willing to dissipate 10 watts in the conductor. That means 10 watt = (42 amp)^2*R, which implies the conductor resistance must be 0.005 ohms.

Suppose that instead you supply the power at 380 volts. The current is now only 1.3 amps. 10 watt = (1.3 amp)^2*R, which means a conductor resistance of 6 ohms, which is over *1000 times* greater than at 12 volts. What this means is you don't need a huge freakin' wire.

Re:You'd need much larger conductors

[wierd_w]

*smacks forehead*

[Note to self: Avoid asking questions about EE, a field you DID NOT take in college, while recovering from a head cold and while under the influence of medication. Seriously, you'll thank me later. There is a reason why the bottle says not to operate heavy machinery. Hint: it also applies to high energy electronic devices.]

Re:You'd need much larger conductors

[sjames]

No, not skin effect. Power = I^2R, so the more current, the more power loss in your cables. You can significantly cut the losses by doubling the voltage and halving the current so long as the device being powered can handle it.

Consider, your 1200 Watt heater can run on 100 Amp @ 12V or 12A @ 100V. A wire that can handle 100A is rather heavy.

Re:You'd need much larger conductors

[some1001]

Umm... That is just one form (I^2*R=P).

The other form is to say... P=VI while I=V/R so you get... GASP!!!!!!!!!!!! P=V^2/R

Current and voltage are related to each other, so don't say none of this them there "determined by the current alone."

Re:You'd need much larger conductors

[grim4593]

Multi-stranded wire does not help when dealing with the skin effect. Since the skin effect is caused by an electro-magnetic field and the strands are in contact with each other a multi-stranded wire will experience the same effects as a solid conductor. To get around the skin effect you need to use something called a Litz wire. A Litz wire is a mult-stranded wire in which each conductor is smaller than the skin depth and each strand is insulated from the others. Each strand is routed from the center of the bundle to the edges of the bundle throughout the length of the bundle to ensure there are no "center-strands" which do not contribute to the conductivity of the wire.

Re:You'd need much larger conductors

[vlm]

Lower voltages require larger conductors to carry the same current. Copper isn't that cheap.

And you're limited on the high end by state electrician licensing boards who require "high voltage license" instead of regular license somewhere around 440 volts to 600 volts. So however cool you think it might be to design your entire infrastructure around 1024 volt DC, the cost of electricians would be much cheaper if you can keep it under 400 volts.

Obviously there are some states where this doesn't matter, all I can say is I've heard of cutover points of 440 volts, 600 volts, in some weird combination of TN , WI, and a couple other states.

I am told the main thing they teach you in high voltage continuing education class is that for all practical purposes, all accidents are fatal above 500 volts. That's pretty much what I've heard from the grapevine... 120 volt deaths are always two parters, like grayhair's heart couldn't take it or the shock made him fall off the ladder and the broken neck is what killed him, whereas 440 threephase and above its always "condolence cards can be sent to..." except for true miracles. That and they put you thru the ringer about threephase and wye vs delta and figuring out rotation directions without destroying industrial equipment, none of which matters in a data center.

Re:You'd need much larger conductors

[nzac]

I^2*R is by far the best way to look at this problem.

Its easy to measure, calculate and in general work with. It gets around needing to use the voltage drop from one end if the wire to the other.

V is dependent on wire and load impedance and supplied voltage to obtain a required power.

You will find the power dissipated in wire can be calculated purely from the impedance (somewhat fixed) and the supplied current. The voltage drop across a wire is related to the current though it and the wires impedance and can therefore be factored out..

Re:You'd need much larger conductors

[Anonymous Coward]

It's acualy lower voltages require larger conductors to carry the same ammount of watts (power). 400 Watts power supply could be supplied from 400V and 1A or 12V and 33.3A (Watts = Voltage x Amps).

While copper isn't cheap one of the issues with running things at low voltage is the voltage drop. The only way to counter act this is to use larger copper conductor or increase the voltage. If you were running a server that required 400W to run and had a 12V system it would require a 10mm/15AWG conductor (which can handle maximum current of 35A) the cost would skyrocket when you think of how many servers a data center holds. If you were to increase the voltage to 24V system it would drop amps to ~17A and only require a 2.5mm conductor which is a huge cost saver. The other option is to run two cables in paralelle to split the load. Voltage drop is a factor when considering that work around due to it being a static value that increases as the conductor gets smaller.

Most of the telecom gear in NZ runs off of a 48V System and uses gear that runs at those voltages (fibre switches, ISAM's, etc.). I believe to make DC power supplies cost effective in data centers they will need to create servers that can run at higher voltages.

Re:You'd need much larger conductors

[Falconhell]

DC has resistance but NOT impedance. Geez!

Re:You'd need much larger conductors

[petermgreen]

Current and voltage are related to each other,

You just have to be careful WHICH voltage you are talking about ;).

In particular when talking about a source supplying power via a cable to a load there are three (interrelated) voltages being considered. Vsupply, Vload and Vdrop (Vload=Vsupply-Vdrop) but only one current (assuming leakage is negliable) is being considered. Further in any reasonable system it is likely that Vsupply >> Vdrop. As a result of all this it's generally best to think of losses as a function of current..

The total power lost in a cable is a function of the current through the cable and the resistance of the cable. It is only indirectly (once you introduce the characteristics of the load) a function of the supply voltage of the system.

Re:You'd need much larger conductors

[AzN1337c0d3r]

Actually skin effect does come into play with DC. Just not steady-state DC currents. If you PWM a DC signal, you will also get skin effect.

We have 48VDC as one standard...

[mlts]

If one has worked in a telco, we already have a standard, and that is 48VDC. This is the domain of the Sun Netras of yore.

If I were to recommend a voltage, why not plain old 12VDC? Yes, the amps have to be high, but we already have a connector for this (beats wiring up things by hand and throwing a breaker), and it is not hard to find off the shelf hardware to support this, be it batteries, power distribution units, inverters/converters, solar panels with MPPT controllers, and so on. We have two large markets (RV/marine) that are dedicated to 12VDC.

Why not just use an established standard? 12VDC works and has a lot of support, or if a higher voltage is needed, then 48VDC.

384VDC just seems to be asking for trouble. It would require yet another separate connector that can't be plugged into 120VAC or 240VAC, generators would have to have an adapter for it. It would require a complete retooling to get to that standard.

Making another voltage level is throwing the baby out with the bathwater. Why not just go with an established DC voltage level?

Take 12VDC. Most generators, from the expensive inverters by Honda or Yamaha can generate that, as well as the construction grade open-framed ones.

Re:We have 48VDC as one standard...

[Wonko the Sane]

why not plain old 12VDC?

Look up the rating of the power supply currently operating your computer then calculate how many amps would be required to deliver that power at 12 volts. Look up the gauge of wire that would be required to supply that much current without melting the insulation. Then multiply by the number of computers in a typical data center.

Re:We have 48VDC as one standard...

[Anonymous Coward]

480watts/12V = 40amps

40amps requires a minimum of 12 gauge wire.

That's beefy.

Re:We have 48VDC as one standard...

[geekmux]

384VDC just seems to be asking for trouble. It would require yet another separate connector that can't be plugged into 120VAC or 240VAC, generators would have to have an adapter for it. It would require a complete retooling to get to that standard....

I'm sorry, but according to the 1%, you've exceeded the allowed threshold for Common Sense with your remarks here. Such atrocities against Greed will not be allowed or tolerated.

Anytime a new standard is being proposed, you can bet there are several people standing behind it poised to make money off it. And based on your suggested necessary changes, they stand to make a lot of money off this new standard.

Who ever said the creation of new standards needed to make sense anymore? We're here to feed the 1% and "create" jobs, which is yet another purpose of new standards. Nobody ever said anything about current standards and interoperability, and you sure as hell aren't going to hear it coming from the guy looking to build and sell 384VDC widgets.

Re:We have 48VDC as one standard...

[Anonymous Coward]

You seem to have some lack of basic understanding. Yes, the current will be higher at lower voltage. This does NOT correlate to needing thicker wires, as the wire has to withstand not current but power which is the result of multiplying voltage with current.

If you have half the voltage but twice the power, the loss in the cable will be the same and therefore the cable needs the same rating.

He, this is really basic electric theory. Maybe you want to learn about Ohms law.

Re:We have 48VDC as one standard...

[mkiwi]

384 VDC is a common voltage used in "Boost" regulators that are universal input. That's probably why that voltage was used. You obviously can't power a microprocessor off of a 384V line (SiC would be necessary to withstand the high voltage). There has to be some sort of switching going on.

I have not RTFA, but what would make sense to me would be to use a boost converter to get 384V, send that 384V all around the building, then convert it down to 12V, 5V, 3.3V, and 1.2V using a few large DC-DC converters. Most enterprise computer power supplies have a PFC boost converter already, so doing all that in one massive stage would save space inside the PSUs and improve efficiency (although the current Boost converters are generally 90%+ efficient).

12VDC and 5VDC would easily cook a modern microprocessor. I guess it's a question of doing the PFC boost for everything at the start, or breaking it up into tiny sections (as is already the case). Safety shouldn't be a big issue, and you need the 384V to supply enough power to all your DC-DC converters. To get the same power with lower voltages you'd need to use expensive superconductors.

Re:We have 48VDC as one standard...

[Anonymous Coward]

Why not just use an established standard?

It's not so profitable.

Since time immemorial vendors have been creating incompatible standards to encourage people to buy stuff.

Re:We have 48VDC as one standard...

[Anonymous Coward]

Of course this needs to read "If you have half the voltage but twice the current".

Re:We have 48VDC as one standard...

[Wonko the Sane]

Yes, the current will be higher at lower voltage. This does NOT correlate to needing thicker wires, as the wire has to withstand not current but power which is the result of multiplying voltage with current.

You've managed to be right while also being wrong at the same time.

You could use voltage*current to calculate the thermal losses in a conductor but what you've done incorrectly is assume that "voltage" in this equation is the voltage between the conductor and ground.

The correct way to calculate losses in a conductor is current * end-to-end voltage difference

The end-to-end voltage difference is directly proportional to the current so the most efficient way to calculate the losses is current squared times resistance.

Since the surface area of a wire is proportional to the square of the wire diameter and the conductivity required is proportional to the square of the current carried it ends up that wire diameter is directly proportional to the current.

Re:We have 48VDC as one standard...

[Wonko the Sane]

40amps requires a minimum of 12 gauge wire.

Now for even more fun figure out how much deviation is acceptable from the nominal value of 12V with it comes to what a motherboard will accept then consider how long the longest wire run is from your central power supply to the furthest computer.

Now consider that the voltage must be maintained within the tolerance band with the load ranging anywhere from 1 computer operating to all of them operating. From that you can calculate the highest acceptable resistivity of the wires. Remember since you're distributing from one point to many you need to keep voltage in tolerance for any combination of loads operating - you can't just raise the output of the power supply to compensate for the line losses and call it good.

Once you've figured out this maximum acceptable resistivity you can figure out how large your wire must be (hint: it's going to be a lot larger than 12 AWG).

Re:We have 48VDC as one standard...

[CCarrot]

You could use voltage*current to calculate the thermal losses in a conductor but what you've done incorrectly is assume that "voltage" in this equation is the voltage between the conductor and ground.

The correct way to calculate losses in a conductor is current * end-to-end voltage difference

The end-to-end voltage difference is directly proportional to the current so the most efficient way to calculate the losses is current squared times resistance.

Since the surface area of a wire is proportional to the square of the wire diameter and the conductivity required is proportional to the square of the current carried it ends up that wire diameter is directly proportional to the current.

A concise explanation of the principles of voltage drop, thank you! I was looking for someone who had posted that!

Re:We have 48VDC as one standard...

[Anonymous Coward]

384V is actually not a horrible choice, being just over the peak voltage of a nominally 240V AC supply (~340V, remember 240V is an RMS value so you need to multiply by sqrt(2)), The advantage here is that the headroom gives space for the boost converter used for the PFC to work.
Large 380V dc supplies are standard fare in the motor speed control industry.

Cable size for a given percentage voltage drop falls as the square of the voltage, so relatively high voltage heavy power makes sense (380V is still low voltage in power industry terms).
High voltage DC has one major issue however, it is very difficult to build a circuit breaker or load break switch for it in a cost effective way, because the arc does not self quench (unlike the AC case).
It can be done, but a breaker at say 400V DC is orders of magnitude larger (and more expensive) then the same unit for 400V AC, and looks more like something that would be seen in medium voltage AC service (> 1000V AC), think air blast or oil immersion arc quenching.

Thus heavy power DC distribution is a four way optimization problem (Wire size, losses, switching and conversion circuitry complexity), and while 48V is probably reasonable within a rack, it will be too low by far for main heavy power distribution.
12V is just stupid, a 1Kw load is ~100A at that voltage and the cables you need to maintain a reasonable voltage drop get really silly over more then a few meters.

Re:We have 48VDC as one standard...

[Anonymous Coward]

Why not 12vDC? Voltage drop over distance. For datacentre PSUs, you're going to need runs of ~100m+.

At 100m, you're looking at a voltage drop of ~20%, vs ~5% for 48v or 1% for 380v

Re:We have 48VDC as one standard...

[DewDude]

Hey...... 384VDC is handy if you wanna power tube amps directly. Just sayin'

Re:We have 48VDC as one standard...

[IceFoot]

why not plain old 12VDC? ... we already have a connector for this

Because a cigarette-lighter connector is noisy and unreliable! Sheesh, I hate them.

Re:We have 48VDC as one standard...

[Anarchduke]

I wonder why there isn't a mod category for cyncical

Re:We have 48VDC as one standard...

[Anonymous Coward]

> Since the surface area of a wire is proportional to the square of the wire diameter

no it isn't, asshole. Just because it has units of length squared doesn't mean you can approximate it by taking any random thing with units of length and squaring it. It goes with diameter times wire length.

Re:We have 48VDC as one standard...

[Wonko the Sane]

So I'm an asshole because I mistyped cross-section as surface area?

You don't think there's a chance that you may be over reacting?

Re:We have 48VDC as one standard...

[petermgreen]

384VDC just seems to be asking for trouble. It would require yet another separate connector that can't be plugged into 120VAC or 240VAC,

True there would be some cost involved in choosing appropriate connectors and certifying power supplies for it.

generators would have to have an adapter for it. It would require a complete retooling to get to that standard.

Afaict in a datacenter the typical structure is you have grid and generator power coming in, those go to an automatic transfer switch and then to a UPS. The UPS (assuming a high quality online unit) converts the AC to DC, takes it past the battery bank and then converts it back to AC. American datacenters also often have transformer based PDUs that step down the voltage between the UPS and the computers. Since generators are placed before the UPS system they do not need to change.

The advantage of going DC distribution inside the datacenter include
1: it takes out the conversion step from DC at the battery bank to AC for distribution.
2: It takes out all the issues of power factor management,
3: It's much easier to paralell up DC busses than AC busses (AC busses have to be brought into phase before they can be paralelled).

Downsides
1: you need special power supplies (in theory a cheap and nasty SMPSU can actually run off DC but in reality i'd want vendor certification before actually doing it)
2: arcing is far more of an issue with DC than AC because the zero crossings of AC tend to break arcs. This means that switches, circuit breakers, connectors etc need to be explicitly designed for DC.
3: there is little standardisation of connectors ETC for DC (afaict the telco guys tend to just bolt stuff to bussbars which is tolerable at 48V but not really appropriate at higher voltages)
4: normal electricians aren't used to working with DC

Making another voltage level is throwing the baby out with the bathwater. Why not just go with an established DC voltage level?

Because the "established DC voltage levels" are too low to be distributed efficiently and economically arround a datacenter. 12V is way too low, even telco 48V is rather low for the power use in a modern datacenter. Remember for a given acceptable loss the copper requirements go as the inverse square of the voltage.

Also the new voltage is not an accident, it's set at a level to make reuse of existing PC PSU designs easier.

Take 12VDC. Most generators, from the expensive inverters by Honda or Yamaha can generate that, as well as the construction grade open-framed ones.

Yes small generators often have a 12V outlet for running car accessories but afaict it can't supply much power.

Safety concerns for regular customers

[Anonymous Coward]

I barely trust my first level tech guys to work on DC power let alone customers who have a hard enough time configuring their routers.

We use DC all the time for our telephone gear, core and border routers (so we only have one type of power supply to stock), and other odds and ends. This mentality comes from us being a telephone company as well as an ISP/colocation center. Otherwise we would be all AC I imagine.

So the first concern I have would be safety if one expects customers to be around DC power plants, power distribution panels, and possibly pulling out a -48V wire
from the back of a server because they failed to tighten down the screw enough holding it in. DC, when it shocks you, does not let go!

Re:Safety concerns for regular customers

[amorsen]

DC, when it shocks you, does not let go!

On the other hand DC also doesn't make your heart fibrillate.

Why would you use power distribution panels and screws? Plain old power cables and connectors work fine for AC, they'll work fine for DC too. Polarized connectors can be an advantage, but for this purpose it isn't even necessary, you can make a power supply which will work for reversed current. C13 and similar are polarized anyway, so they would be a good start, although you would obviously have to make the connectors incompatible with standard AC connectors.

Re:Safety concerns for regular customers

[Anonymous Coward]

I have never seen power cables for DC power. Sure I have seen different kinds of connectors for going into servers and routers but the other end has always been red and black wires that get screwed down into a smaller power distribution panel. I imagine you are probably correct if that DC power became more mainstream that things would become easier and safer to work with. Right now though, if you go into a Central Office you will see many dangerous situations for those who do not understand what is going on.

Ten years?

[Anonymous Coward]

Ten years ago I worked for a startup that was making 1U and 2U appliances for streaming video distribution. Back then the hardware guys were saying that DC was the way to go.

Now it's ten years later and who uses DC? I guess I'll believe it when I see it.

Re:Am I the only one that thinks this is a step ba

[LehiNephi]

I think you failed to mention how much current is being pushed down that pair of 16AWG wires. The power loss comes from (current * resistance (of the conductors)), not from the voltage.

What do you mean?

[Anonymous Coward]

Where are all the Tesla leg humpers when we need them!!??!?

It's complicated

[Anonymous Coward]

It's a complicated subject that is mostly immune to a simple analysis.

The goal is to increase efficiency, by eliminating a few levels of voltage conversion.

One simple plan would be to convert then incoming AC to a medium Dc voltage, a few hundred volts, then distribute that down to the board level, at which you'd have your basic switching converter/regulator to drop it down, in one step, to the 12 and 5 and 1.X volts that the disks, logic, and CPU need.

You can't drop it down any earlier than the board level, as the cost of copper quickly limits that. For instance a 5 volt supply at the rack level would require fire-hose size copper conductors from rack to boards.

Problem is, 300 volts DC is kinda hard to handle. Dc likes to arc and does not like to stop arcing, making 300 VDC switches and circuit breakers very large and complex. Also running 300VDC to the board level increases the cost of wiring, as wires at that voltage have to meet a higher code level. it's a very delicate thing to find the right topology of AC, DC, and voltage levels.

Re:It's complicated

[budgenator]

Problem is, 300 volts DC is kinda hard to handle. Dc likes to arc and does not like to stop arcing, making 300 VDC switches and circuit breakers very large and complex. Also running 300VDC to the board level increases the cost of wiring, as wires at that voltage have to meet a higher code level. it's a very delicate thing to find the right topology of AC, DC, and voltage levels.

As far as the wire, low voltage for Electrical mains is 600 volts or less so that isn't a problem, 60 Amp 380VDC
DC Rack Mount Power Strip are off the shelf equipment, and if memory serves me correctly ConEd just stopped supplying 600VDC service just a year or two ago in NYC so the technology has had plenty of time to mature since Tom Edison started it.

Actually no generator change required

[Anonymous Coward]

As mentioned above, a Y configuration on a 440V (delta) 3-phase generator is 381.05V. once rectified, this is almost exactly what you need.

3-phase diesels for this type of thing are incredibly common and in industrial electrician would find setting up this configuration trivial.

As mentioned by many, high amps mean large gauge wire, and copper prices continue to rise.

Other server optimizations

[John.P.Jones]

Along similar lines but related (in a way)...

I have been wondering why data centers don't use more optimized hardware that basically packages CPU & memory on a single chip (multiple dies until they can fit on one) so that basically all the pins go to either power or to a network adapter, no local disk or video adapter or anything else, just a black box with power, network i/o and heat dissipation. The thing would boot up over the local network by a controller and use other data nodes (dumb but fast NAS front end to disk drives) for storage beyond the locally cached memory.

are you kidding? try better than 95%

[Chirs]

A capacitive switching DC-to-DC converter is capable of up to 98% efficiency depending on how it's configured and what voltages it's operating at.

battling a strawman

[markhahn]

unfortunately, DC studies like this tend to argue their case against a strawman. namely, it's now quite easy to find servers with 94% efficient PSUs. and many datacenters can operate without dual-conversion UPSs (consider Goggle's design of simply sticking a small battery onto each server.)

Re:battling a strawman

[whois]

The PSU is still a point of failure, a heat source, and is costing 6% efficiency per server at the start. DC has a bunch of advantages in this type of environment, one being that redundant power is easy.

The only real downside to DC is the lack of standardized plugs, and the lack of jumpers. Wiring harnesses eliminate some of this, but having to run and terminate wires of various sizes tends to turn off everyone but electricians.

Re:battling a strawman

[Anonymous Coward]

Oh, well then. If the only thing lacking is standardized plugs and the lack of jumpers, then everyone get off your a** and start building DC everything.

Except now you have to use plugs designed for and implying another purpose. Or else hard wire the connections and watch your bills for electricians time skyrocket. Or else begin a years-long process of developing, and getting multiple standards bodies & code approvals for a whole set of new plug designs.

Of course it can all be done. However the time, effort and dollars needed to do so are significant and are the precise reasons why you don't see much of all-DC infrastructure installations.

Re:battling a strawman

[SuperQ]

You can get equally good AC/DC on small PSUs and big central PSUs these days. 6% on 1000 servers is the same if you have 1000 PSUs or 1.

The Parent post is correct and your reply didn't answer it.

Re:battling a strawman

[petermgreen]

The PSU is still a point of failure, a heat source, and is costing 6% efficiency per server at the start.

The thing is even with DC systems you still end up needing a PSU. The voltages needed to efficiently move power round a datacenter are higher than you would reasonablly want to put on a computer motherboard and since computer equipment isn't generally designed to be isolated from the chassis it's bolted to you really want your PSU to be isolating (large return currents running through signal cabling would be BAD).

Re:are you kidding? try better than 95%

[FaxeTheCat]

I commented specifically on the use of a 7805 voltage regulator which is not a switching converter.

Re:are you kidding? try better than 95%

[Anonymous Coward]

A switched cap DC-DC converter on the other hand are limited to the CURRENT it can handle and also the ratios of voltage convertion. it doesn't do very well outside of those two. I would be very surprise if someone shows me one that do 10+W in a small size.

But what about the poor switches???

[Oryn]

Has anyone considered all the arcing and sparking that simple on / off / circuit breakers will have to deal with? At least with AC you have a chance that the switch will be opened or closed at the zero crossing period and that AC makes it harder to draw arcs when breaking a circuit.

I'm not sure its really going to present any more effeciancy. I guess if you use a 3 phase site rectifier then it will save you the cost of 3 phase copper and make the server power supplies cheaper. Not forgetting that you could actually plug an unmodified server directly into 384V DC (so long as the psu is set to 240v).
I guess the days of computers using 60hz as a reference are over :)

Re:But what about the poor switches???

[DerekLyons]

Has anyone considered all the arcing and sparking that simple on / off / circuit breakers will have to deal with?

You just design the breaker appropriately. It's not exactly rocket surgery.
 
Telcos have used 48VDC for decades. Submarines have used circuit breakers on high voltage/high amperage DC for just about as long. (Even modern nuclear boats have a big ass 12VDC (in the US) battery for backup power.)

Re:But what about the poor switches???

[Wonko the Sane]

Even modern nuclear boats have a big ass 12VDC (in the US) battery for backup power.

That's only true for large values of "12" that approach 300. It is a very big battery though, and they will reliably handle discharges at over 1 MW.

Re:But what about the poor switches???

[Oryn]

I'm not sure you are getting my point. 48vdc != 384vdc

Re:But what about the poor switches???

[DerekLyons]

I'm not sure you're getting *my* point - so I'll repeat it.

High voltage/high amperage DC circuit breakers are old hat. The methods to deal with the arcs have been known for the better part of a century.

Don't forget the elephant in the room

[UnoriginalBoringNick]

http://www.wired.com/science/discoveries/news/2008/01/dayintech_0104

Power monitoring

[CCarrot]

Hmmm...I am only familiar with AC power monitoring. How would current monitoring work for a DC system? DC current can't induce, so you couldn't use simple current transformers like you do for AC systems...I suppose some sort of inline resistive/voltage monitoring would work, but it makes a retrofit a bit of a biatch, and you'd see additional losses over the monitoring resistor.

Thoughts?

Re:Power monitoring

[blair1q]

Hall effect.

In the presence of a static magnetic field (as around a conductor carrying a constant current), electrons in the clamp circuit, which also carries a DC current, will be pushed to one side of the clamp conductor, inducing a voltage relative to the other side. Measure the voltage and you know the current in the wire it's clamped around.

Re:Power monitoring

[gboss]

You make a large connection with a known (low) resistance (i.e. known copper cross section) and then measure the voltage drop across it.

Re:Power monitoring

[CCarrot]

Hall effect.

In the presence of a static magnetic field (as around a conductor carrying a constant current), electrons in the clamp circuit, which also carries a DC current, will be pushed to one side of the clamp conductor, inducing a voltage relative to the other side. Measure the voltage and you know the current in the wire it's clamped around.

Oh, that's clever! Thanks, I knew there must be a way, I'm just used to working with AC...;)

Telco Central Office Anyone?

[Anonymous Coward]

Where have you been? Telco sites have been DC power for the last 50 years or more. People in the the data center world generally don't get involved because it is more expensive to deploy and support than A/C power. In other words people are generally rational consumers when they aren't a monopoly backed by the government.

Maritime Appliances

[Anonymous Coward]

Most maritime appliances work on 12, 24, or 48 vdc, because of the extensive use of batteries on-board. You only see plentiful AC power on vessels with large diesel drive. Mah boat has solar panels and sails, no diesel.

DC power distribution is awesome

[wezelboy]

At least 25% power savings. Higher reliability. Lots of copper though.

not unless...

[Anonymous Coward]

It will never happen unless an industry standard connector(s), cables, PDUs, and other components can be quickly agreed upon, and that are as easy and flexible in usage as the current AC stuff.

Hardware mfgs will also need a standard to create DC-to-DC power supplies for all their servers. Companies are not going to build entirely new server lines to accommodate DC. They will need to be able to make a DC power supply that can slide into the existing PSU bay(s) in place of AC power supplies.

380v sounds extremely excessive, since PCs only need 12v. You have to be careful with the trade-offs though - voltage/amperage. Low voltage would require huge wires to supply hundreds/thousands of servers, but although a higher voltage helps keep the wiring under control, you then have to have switching power supplies to lower the voltage in the server itself back down to 12v. I don't see how moving to DC would create any efficiency, there's enough politicians there already. ;->

No Savings for SMB?

[Anonymous Coward]

OK... So I manage a datacenter with about 120 servers that pulls about $17,000 of power per year. The study suggests a 15% savings on power by running DC over AC, so I could cut my annual power bill by $2,500.00.

I am actually planning a new expansion datacenter with about 80% of the capacity of the current one. If I outfit this new Datacenter with DC instead of AC I can save around $2,000 per year in power. The new Datacenter is going to cost me about $200K to build.

So, Since far fewer datacenters are built with DC power, how much more is it going to cost me to build it with a DC infrastructure as opposed to a traditional AC one? Even if I could get all the power hardware swapped over at a one to one cost (not likely), I'm still going to need to pay some consultant a pretty penny to assist with the design and implementation, which will easily wipe out 5-10 years of savings at my $2000 per year clip.

So... When they tout 'big savings' you really ned to scale up pretty big before it starts to convert to 'real' savings. Thats why DC has yet to create any interest for small and medium businesses like mine. Power is a hell of a lot cheaper than a DC infrastructure.

Re:No Savings for SMB?

[petermgreen]

Also at least according to APC you can get a lot of the advantages of a DC system without the hassle by going for a european style AC system (240/415 three phase without transformer based PDUs) rather than an american style AC system (transformer based PDUs). And the vast majority of server and network hardware is perfectly happy on a European style system (after all the vendors want to sell it in europe too).

Datacenters?! Who gives a sh.. about datacenters?

[Pf0tzenpfritz]

Seriously. Ok ...halfways: Datacenters have got double floors and ceilings, lanes between rows and rows of racks where the wiring (and the heat) are going. They've got highly efficient power-sources, optimized air-flow concepts that make use of thermal differences instead of fans etc. pp. They can handle AC or DC, copper and fiber and whatever. Now have a look at my desk, where my "mobile" computer is - spun into a bizarre, organic -or parasitic- fabric of USB-cables, CAT6 (wireless sucks, when you are trying to keep latency low and MIDI-sync intact) and AC adapters. Incredible lots of these shitty little bastards that will block two or three outlets, produce heat, static,AC-buzz, chaos, madness and most likely deseases, terror and tiny electro-slums. I've got an Akai LPD8 connected by USB, an Akai APC20 (1xUSB, 1xDX that will run into another direction and fraternize with the wiring invasion from...) a notorious self-powered USB-Hub (1x6V DC, 1xUSB to Computer, 6 Ports used by...) a Rane SL1 (4xAudio IN, 8xAudio Out, 1x -guess what- 9V DC with it's own, ugly little adapter), a Yamaha UX-16 (bus powered, but USB+2xMIDI connecting with a) Boss DR660 (2xMIDI, 4xAudio and it's own #+*&%! 7.5V DC) all that on a battle mixer (with it's own 10V DC adapter) and -of course- the source of all evil, my Wire Notebook - with it's own 20V DC. Oh - and I have a 19" rack, too. Inside are an Aphex Type C (24V AC(!)), compressor/limiter (9V AC) light (12 DC), flanger (9V DC) and a fan (12V DC) to cool all the other fricking little adapters, all with their own funny little incompatibe connectors, way too long cables, crummy voltages. And year after year this is getting worse. Now look at your own setup. Maybe you've got just an all-in-one box, netbook or the like with next to no wires connected, maybe you're cursed with even more low-voltage-jungle - and now multiply to millions of homes light years of LV wiring, swarms of nasty little connectors... What the world needs is some damn standard to get these devices powered by a single source to get the wires away from where they never were meant to be. Who cares if datacenters run AC or DC inside as long as they have one common mains line? They might run 12V or a kV, it really doesn't make that hell of a loss that the common households are producing with no change -at least none to the better- in sight...

AC/DC Rawkz!

[TiggertheMad]

When I think of the things in our house that *must* run on AC, it's only our fridge, freezer, and HVAC

For those of us with a modest knowledge of electrical devices, what is it that *prevents* these devices from running DC? My understanding is that AC was chosen over DC so many years ago based off of long distance transmission requirements.

Re:AC/DC Rawkz!

[Wonko the Sane]

AC motors are more efficient than DC motors, but then again that really means three phase AC motors instead of single phase kludges.

Re:AC/DC Rawkz!

[petermgreen]

For those of us with a modest knowledge of electrical devices, what is it that *prevents* these devices from running DC?

induction motors are cheap and reliable but they need AC to run. Conventional DC motors are less reliable because they have brush wear.

Of course the high end stuff uses inverter drives anyway because that allows greater efficiency over a range of operating speeds (especially as the inverter output can be three phase whereas in most domestic situations a direct mains fed motor would have to be single phase) so they could probablly be designed for either AC or DC incoming.

AC was chosen over DC so many years ago based off of long distance transmission requirements.

AC was chosen over DC because it was much easier to perform voltage conversion on it. Back in the day the only way to convert DC voltages was motor-generator sets which were expensive and inefficient. Nowadays we tend to use switched mode power supplies anyway for small conversion requirements within the home because small transformers are not particually efficient but still for larger requirements and higher voltages transformers are hard to beat on efficiency, cost and reliability.

My GTI's steering is wonderful.

[zerofoo]

My 2008 VW GTI has an electric assist rack, and it is one of the best systems I've ever felt in terms of feedback and heft. I've also driven overboosted hydraulic systems that feel like mush.

Electric assist steering can be done well, and hydraulics can be done poorly. The technology isn't to blame, it's the engineering that matters.

Re:My GTI's steering is wonderful.

[DanielArdelian]

The 2008 VW GTI has an electric motor that drives a hydraulic pump that does the steering assist.
(it's not a true electric-only assist)

Re:My GTI's steering is wonderful.

[zerofoo]

Actually,no it doesn't. There is no hydraulic pump or reservoir for the steering system in my car. I know, I've looked for it, and I've actually seen the steering rack pulled from another 2008 GTI - no hydraulic lines at all.

It is a pure electric system - not a hybrid system as in the 2011/2012 Jeep Grand Cherokee.

Plus brushless motors

[NotSoHeavyD3]

The other reason for AC current is that you can make brushless motors far more easily, at least when the AC/DC war was going on. Of course now it's not hard to get DC brushless motors. (Hell, all the decent motors for RC electric planes and copters are BLDC.)

Re:Plus brushless motors

[LinuxIsGarbage]

Computer fan motors are brushless DC. But really they are permanent magnet AC motors with a simple VFD (variable frequency drive) in it. I suspect RC planes are the same.

In the industrial world VFDs are very popular. On anything from a 2HP conveyor to a 1000HP+ piece of equipment. They rectify three phase input to DC, then convert it back to AC at the desired speed. Some are setup so you can have one central rectifier, and multiple inverter sections for your different loads.

Re:Plus brushless motors

[Anonymous Coward]

> But really they are permanent magnet AC motors with a simple VFD (variable frequency drive) in it.

Don't get me wrong, I don't consider this a controversial or fantastic claim, but what's your source on this? What would the circuit diagram look like?

Re:Plus brushless motors

[narftrek]

http://en.wikipedia.org/wiki/Brushless_DC_electric_motor

There is a sample schematic in the sidebar.

Re:Plus brushless motors

[Anonymous Coward]

I guess my question then becomes when is a brushless DC electric motor actually not really a permanent magnet AC motor

Amps, not volts

[DragonHawk]

Any old technician with a brain in their head can run DC power feeds to equipment relatively safely due to the low voltages involved.

Voltage only determines if it can overcome the resistance of your skin (and maybe clothing). Beyond that, it doesn't matter. Amperage, on the other hand, determines the power -- the amount of damage the current will cause.

10 milliamps can kill you. But without at least several dozen volts behind it, it won't make it through your body.

But. Put something nice and conductive (like a tool) across a low-voltage circuit and you'll get an arc from the short. You don't need high voltage with that conductive material. And the arc itself can be dangerous. Temperature of the sun, chunks of hot metal flying around, etc.

Now consider that the battery plant in a typical telco CO is the size of a small one-bedroom apartment. The amount of power in that battery string is truly frightening. The main bus bars are often *several inches thick*.

As one guy put it, "Drop a wrench, learn braille."

Re:Amps, not volts

[Wonko the Sane]

Voltage only determines if it can overcome the resistance of your skin (and maybe clothing). Beyond that, it doesn't matter. Amperage, on the other hand, determines the power -- the amount of damage the current will cause. 10 milliamps can kill you. But without at least several dozen volts behind it, it won't make it through your body.

Maybe you do understand voltage, current and resistance but they way you are using your words makes it sound you are making a very common mistake. Amperage is an effect of two things: voltage and resistance.

If I touch a 1000 volt wire that is carrying 100 amps and resistance of the return path (including my body) is 1 megaohm then exactly 1 milliamp will flow through my body.

It wouldn't matter if the wire was carrying 10 amps, 1,000 amps or a 1 million amps. 1000 amps / 1 megaohm creates 1 milliamp of current regardless how large the wire is or how much current it carrying through other paths.

In practice, the size of the wire does matter because you can create fault currents that can not be sustained by a given wire and the actual voltage falls off when you try but regardless the point stands - amperage is a function of voltage and resistance (impedance) in a given situation, not an independent quantity.

Re:Amps, not volts

[rev0lt]

I'm not shure 10mA is enough to kill an average person. Where I live, electrical installations are required by law to have a differential switch that, for domestic use, should fire at 30mA. what this means is that the switch should cut power automatically if detects a current equal or greater than 30mA on the ground wire, and that value is commonly considered the human threshold limit.

Re:Amps, not volts

[Electricity Likes Me]

Technically 1 mA can kill a person - it depends on the path through the body. 1 mA going through your heart is enough to trigger fibrillations. AC and DC also makes a difference according to wikipedia.

Re:Amps, not volts

[rev0lt]

Considering the technical specs of defibrilators, and assuming that a 300J charge can kill you (probably a much smaller charge can do it, but anyway), you'd need 300W/s - at 1mA, that's a discharge of 300/0.001A = 300,000V applied in a path trough the heart. At that voltage, you probably can create a 10 inch arch trough the air - applied to a person, the arch itself would burn a whole trough the path.

Re:Amps, not volts

[Electricity Likes Me]

Defibrillators are intended to you know, defibrillate.

Human resistance varies by quite a bit (largely based on how wet our skin is). Don't need a 300W/s shock to put the heart into fibrillation.

Re:Amps, not volts

[rev0lt]

The point remains that for 1mA to cause real damage, the voltage by itself would burn a hole through the skin.

Re:Amps, not volts

[Electricity Likes Me]

No it doesn't at all.

The resistance of the human body can drop as low as a few tens of ohms if your skin is wet. 1 mA of current through the heart muscle does not take tens of thousands of volts. It depends on the resistance.

Pre-supposing that 300W/s is clearly the needed voltage is pulling a number out of your ass. If "clearly the voltage would be so high as to be dangerous itself" then what do you think is happening in the more common case of someone touching 240V live and getting a 60 mA shock?

Re:Amps, not volts

[rev0lt]

I guess you've never measured water resistance, then :) If you have wet skin, the current will flow trough your skin and not necessarily trough your heart (you know, electrons tend to choose the easiest/least resistant path). But IF you applied it in a way it would flow trough your heart, how would it be? Well, we have the example of the tried and true defibrilators, and the pads using to kickstart a heart in heart transplants. 300J is usually the max rated energy discharge marked as "safe" for the equipment usage, so there you go - you have a threshold.
Given that 300J = 300Ws, that's 300W of power applied per one second of use. To get 300W of power at a current of 1mA, you'd need a very high voltage figure - enough to cause arching. Even if you don't like the 300J figure, with 50J the result would be the same, but it would take more time to reach the same damage level.
In the more common case of touching a 240V line at 60mA, we're not talking about 1mA. Also, if considering a common 240V line(AC), we're not talking about resistance, but impedance, which is a rather different concept.
So, technically, the situations where 1mA can kill you, the voltage that you'd need would, by itself, cause much more damage than the current, when mantaining the energy transfer.

Re:Amps, not volts

[rev0lt]

Transcribing from http://emedicine.medscape.com/article/780639-overview#aw2aab6b3 , regarding an external pacemaker:

The average current necessary for external pacing ranges from about 50 to 100 milliamperes (mA); 100 mAs applied to an average chest with 50-ohm () resistance for 20 msec delivers 0.1 Joules (J). This is well below the 1-2 J required to cause an uncomfortable tingling sensation in the skin.

See how they account energy transfer(J), and not current by itself? And see how far away are the values of current, from the 1mA you're talking about?

Re:Amps, not volts

[Electricity Likes Me]

Transcribing from http://emedicine.medscape.com/article/780639-overview#aw2aab6b3 , regarding an external pacemaker:

The average current necessary for external pacing ranges from about 50 to 100 milliamperes (mA); 100 mAs applied to an average chest with 50-ohm () resistance for 20 msec delivers 0.1 Joules (J). This is well below the 1-2 J required to cause an uncomfortable tingling sensation in the skin.

See how they account energy transfer(J), and not current by itself? And see how far away are the values of current, from the 1mA you're talking about?

Note also, how all the examples you're talking about, are discussing the medical uses for electrical stimulation, wherein they're highly concerned with not killing the patient?

The point was never "1 mA *will* kill you" the point was that even 1 mA, if it finds a current path through the heart muscle, can induce fibrillation.

Which you've been saying clearly can't happen because, there are medical uses for this in defibrillators?

Moreover, again, why limit the shock time to such a low number? 1 mA is not a lot of current. Amongst other things there's a serious danger someone would be shocked for considerably longer time then 20 milliseconds.

Telco power connectors

[DragonHawk]

Only very stupid engineers design power connectors that can fit both ways.

The DC power supply connections in telco equipment is generally screw terminals and spade connectors.

Re:Telco power connectors

[skids]

Yep. Many fond memories straining my hands crimping on a spade connector that was a gauge lower than the crimps, and applying conductive anti-corrosive gel.

Lack of an idiot-proof standard connector is another problem that has limited DC use in less specialized environments.

Not sure how I feel about 380V. -48V DC is dangerous enough as is.

Re:Telco power connectors

[lazybeam]

I wouldn't want to be handling 380V with screw terminals and spade connectors!

Reminds me of a place where I used to work (pre-2005): each store had an ADSL modem and ethernet router, same brand. Same connector for the DC power supply. Opposite polarities. So if they were ever unplugged at the same time there was a 50% chance of letting the magic smoke out, which did happen.

Re:Telco power connectors

[alfredos]

ADSL modem and ethernet router, same brand. Same connector for the DC power supply. Opposite polarities. So if they were ever unplugged at the same time there was a 50% chance of letting the magic smoke out, which did happen.

A pair of devices so brokenly designed can't, of course, include a series diode. Oh no, that would cost one more cent per device - prohibitive!

Re:Telco power connectors

[Grishnakh]

Yes, and that's because telco equipment makers keep trying to conform to the way things were done 100 years ago. There's no reason for a computer datacenter to use 100-year-old standards and methods. A few decades ago, someone invented something called a "connector", consisting of several electrical connections in one unit, which could be keyed so that it could only be inserted one way to prevent reversed connections. Let's use this instead of going back to the bad old days of screw terminals.

Re:Telco power connectors

[Vegemeister]

A series diode would substantially reduce efficiency.

Re:Telco power connectors

[alfredos]

What efficiency? That of the manufacturer's sales team?

A series diode is only a 0.6V drop and, for low-power devices like these, its power use is comparable to all those fancy LEDs in the front. Only that none of these LEDs will protect the device from plugging the wrong supply.

Re:Telco power connectors

[Vegemeister]

It is not at all comparable to the indicator LEDs. The LEDs only have 15mA or so flowing in them. A series diode would have to pass the entire supply current of the device. Assuming the modems are running on 12V, a series diode would make the modem consume 5% more power than it otherwise would. 'Only 5%' you say, but that's every modem in the US you're talking about. No, I think a connector that only fits one way is good enough.

Arc flash hazard

[DragonHawk]

If I touch a 1000 volt wire that is carrying 100 amps and resistance of the return path (including my body) is 1 megaohm then exactly 1 milliamp will flow through my body.

Riiight. I=V/R, not just a good idea, it's the law.

Remember where I talked about shorting a wire with a tool? That's the danger in telco power system. Not you touching the wire -- your body is a lousy conductor, compared to copper. But if you short a bus bar with a screw driver, or something like that, the resulting arc flash will really ruin your day. The arc converts the electrical energy to thermal and kinetic energy, which is perfectly capable of burning your face off.

380V DC makes sense

[amoeba1911]

This actually makes perfect sense... almost every new power supply is designed to accept 100V to 240V AC. Fully rectified 240V AC RMS is already very close to 380V DC, so their proposed system would simply be feeding pre-rectified DC electricity to the power supplies.

Before you go home and try plugging your own computer power supply into a 380V DC power supply, I'll tell you it will not work. The computer power supplies you get have an active power factor correction circuit which will not work with DC input, so you would need to dissect the power supply to remove these parts. After that it should work just fine with DC input.

The good thing about going DC: you need less parts in the switching power supplies, meaning cheaper power supplies.
Also with 380V, you can use a cable diameter 1/5th of the size to carry the same power (compared to 110V RMS).
And backup devices (UPS) use batteries to store electricity in DC form, in case of power failure, heavy and expensive MOSFET circuits have to drive transformers to convert the DC back into AC, so changing the computers to DC would completely eliminate the need for DC to AC converters, you could store 380V directly in the batteries.

Re:380V DC makes sense

[CrazyBusError]

Fully rectified 240V AC RMS is already very close to 380V DC

Uh, no. No it isn't. Not even vaguely, in fact. 240V effectively *is* the DC equivalent (ignoring rectification losses), the PMPO (ie. peak voltages your 50 or 60hz sine wave actually hits) will probably be around 380v though. If you want 380VDC, then you need the same or more AC.

DC is more expensive

[gavron]

Sure, it's more efficient because we don't need all those switching PSUs.

Of course nobody wants to buy more expensive servers with DC PSUs. So, that's out.

And Data Centers don't want to waste 1/4 of their space on battery rooms, so that's out.

And 380V is not any standard anybody uses to deliver -48VDC (the real standard for data center DC voltage).

So overall, I'd say GOOD JOB APRIL 1 and this is all a pile of stuff that will never happen.

E

Re:DC is more expensive

[gavron]

FYI those of you who are thinking "Oh but 380VDC could be used in a 240VAC PSU if we take the rectifier out" you are RIGHT except for the fact that these are switching PSUs so... no... it wouldn't work at all.

Back to -48VDC we go.

Try making up new standards all day long. This has been a standard in data centers for over forty years for a reason. And if battery rooms aren't going to be funded by data centers, and expensive PSUs won't be funded by server owners, the warlock tiger blood winner here is "failure".

Re:DC is more expensive

[skids]

I don't think switching to DC would require a "battery room" any larger than the current UPS system. But you do pinpoint the very precise reason why DC is limited to telcom and hasn't taken off in the run of the mill IT computer room: vendors refuse to manufacture and sell cheap DC PSUs for their equipment. The minute they see someone ask for DC, the dollar signs start lighting up behind their eyeballs and they think "ooh boy, specialty need, lets get some extra cash out of this one."

Car battery safety

[DragonHawk]

I'd guess 12V / 24V is probably the best, as it's low enough not to be a safety risk (think car battery)

Put the metal shank of a screw driver across a car battery's terminals sometime and let me know if your opinion is the same.

(More here: #38222188)

Er, but all things are NOT equal

[Anonymous Coward]

I = V / R (AKA R = V / I ). This the key here.

AC power is easily transformed to extremely high voltages with low current to allow it to travel long distances with low resistance.

You can not do this with DC power. End of story. This is why DC power failed, because traveling even city-wide distances needed repeater stations all over the place to compensate for the resistance.

Re:Er, but all things are NOT equal

[Wonko the Sane]

You can not do this with DC power. End of story.

Sorry, while that may have been true at the beginning of the 20th century it certainly isn't true at the beginning of the 21st.

Re:Er, but all things are NOT equal

[atisss]

Obligatory wikipedia article: http://en.wikipedia.org/wiki/High_voltage_direct_current

By increasing voltage you can decrease loss by resistance.

If most of devices today are using DC, it could be more efficient to use DC for transmission, however you would still need a step-down converter. Linear converters can be very inefficient, but switched converters are more complex and expensive. However most computer PSU's anyway have switched converters, so it would be worth.

Re:Er, but all things are NOT equal

[OeLeWaPpErKe]

Try building a circuit to increase the voltage of a DC power source, then you'll understand.

Yes, we know that once you magically have huge voltage DC, there is no more problem. Getting to that point, however, is the problem.

Re:Er, but all things are NOT equal

[atisss]

How about not decreasing it in the first time?

European power lines do deliver 380V three phase AC from substation to building, data-centers have their own substations.
So, initially it would require high efficiency AC-DC rectifier at substation, which can be better managed than hundreds of small AC-DC rectifiers at device.
That would also have an upgrade path allowing to deliver DC directly to substation, so eliminating rectifier.

Re:Er, but all things are NOT equal

[OeLeWaPpErKe]

True ... but only true if you
1) trust the service provider to have a trustworthy DC network (the damage a DC power surge will do is much higher than an AC surge)
2) assume safely isolating servers from this DC net does not carry any cost either in efficiency, or in money
3) assume that you can convince server makers to build & support such servers (not a problem if you are google or facebook, but ...)

Re:Er, but all things are NOT equal

[Grishnakh]

It wasn't true even as far back as 1940. HVDC links have been around for ages.

Re:Er, but all things are NOT equal

[Anonymous Coward]

Okay: http://en.wikipedia.org/wiki/Voltage_multiplier

Re:Er, but all things are NOT equal

[OeLeWaPpErKe]

The first line of that page :

A voltage multiplier is an electrical circuit that converts AC electrical power from a lower voltage to a higher DC voltage, typically by means of a network of capacitors and diodes.

This is, by the way, a circuit that you might as well describe as "a cute AC voltage multiplier met a very nice rectifier gentleman one night and ...".

And the DC -> DC circuits there always convert to AC, then change the voltage, then convert back to DC. This works, but it's very inefficient. It is so inefficient that you probably wouldn't use them if there is any other option.

Cost.

[bill_mcgonigle]

Yeah, I've never quite figured out why telecoms have standardized on 48VDC while everyone else completely ignores its existence.

I went through this with a client. There was a 'green' data center opening up an hour north of them that insisted on all-DC gear. They charged per-Amp, all you can eat data and rack were included.

We spend probably 20 hours spec'ing an all-DC set of servers, switches, firewalls, smart-PDU's, serial consoles, etc.

By time we were done, the cost of all that DC gear was 1.5x the cost of the COTS AC gear, and that difference paid for almost two years of standard colo outside Boston. Plus we could get onsite service contracts on all of the AC gear.

In theory, all-DC is great.

Lightning & cooling

[DragonHawk]

Better lightning protection. I'm sure its happened, but I've never heard of losing a telco DC bus. Big conductors, giant batteries across them, lightning is just not an issue anymore at the power level (still need to ground feedlines / waveguide / whatever you've got at home like that)

I'm not sure what the power supply type has to do with lightning protection?

Telco DC power supply systems are very robust, but that's because they're highly redundant and the telcos put lightning arrestors on just about everything that enters a CO.

dump most of the power conversion heat in the battery room where its all built to handle high temp and no one visits (other than occasional battery maint). Cheaper cooling in the data center, data center is somewhat more habitable, etc.

While it is certainly true that moving those heat sources out of the computer rooms is a big win, keep in mind that practically all the power that goes into a computer is dissipated as heat. So while you do get some cooling savings -- as you note, it's cheaper to cool power equipment than delicate electronics -- it's not as much as you might expect.

It's the arc right next to you that's dangerous

[DragonHawk]

I'm no electrician, I'd guess the low voltage DC is safe because you can grab uninsulated leads in your hands and nothing will happen.

You're right, that is relatively safe.

As the parent post says, the danger comes when you put something *else* that's *highly conductive* across the bus bars. The arc flash can be deadly, if you're right next to it.

Re:It's the arc right next to you that's dangerous

[Wonko the Sane]

To be more specific the danger comes from the the fault current capability.

1 millivolt could could vaporize an (unrealistically) low resistance wrench that fell across some bus bars as long as the power source could maintain 1 millivolt output with a billion amps of fault current for a few milliseconds.

On the other hand, shorting a couple 26 AWG wires together that are hooked up to a 12 volt power supply aren't going to create a situation anything like shorting the terminals of a battery because those little wires have too much resistance to allow a dangerous level of fault current (It still might break the power supply though).

I've seen this before

[DragonHawk]

By time we were done, the cost of all that DC gear was 1.5x the cost of the COTS AC gear ...

So, the standard technology adoption catch-22. Almost nobody uses it because it's expensive, and it's expensive because almost nobody uses it.

Foo.

Re:I've seen this before

[bill_mcgonigle]

So, the standard technology adoption catch-22. Almost nobody uses it because it's expensive, and it's expensive because almost nobody uses it.

Which basically drives at the fact that it's optimizing something that doesn't really matter enough.

Real estate, infrastructure, and bandwidth are much more expensive, in comparison, per U, than electricity. If electric costs go up, people will probably care more about electric costs.

"Dave made me say that?"

[Falconhell]

No I didn't. (:

Re:"Dave made me say that?"

[rdebath]

No not Other Dave, he meant Proper Dave.

AC has arcs too

[DragonHawk]

Has anyone considered all the arcing and sparking that simple on / off / circuit breakers will have to deal with? At least with AC you have a chance that the switch will be opened or closed at the zero crossing period and that AC makes it harder to draw arcs when breaking a circuit.

I've seen enough videos of AC arcs from switching failures that I suspect AC isn't that much better in practice. All the 440+ volt switchgear at work has big "ARC FLASH HAZARD -- PERSONAL PROTECTIVE EQUIPMENT REQUIRED" signs on it, too, come to think of it.

Meditate on "ohm"

[Anonymous Coward]

Would replacing your desktop power cord with jumper cables make sense?

There is a very good reason why low voltages are not used for long-distance or high-current power distribution! Someone figured this out a long time ago, his name was "ohm"

Who the hell modded the parent up?

Back in the days

[SpaghettiPattern]

Back in the days DC was converted down by "burning" the surplus. That's what a voltage regulator like the 7905 does for low powers. For higher power you'd use a huge transistor and a Zener diode or a (79??). A truly stable source of DC that wouldn't cause the horrible 50/60 noise was the holy grail.

How did that change? (Makes for discussion. And I'm too lazy too look it up myself.)

Re:Back in the days

[Wonko the Sane]

How did that change?

You can still regulate voltage that way but instead of being 95% efficient your computer's power supply would be about 60%, with a huge increase in power consumption, size and heat generation.

Switching power supplies do create high frequency noise that must be filtered out but that's the price you pay for the increased efficiency.

business case?

[dutchwhizzman]

I'd like to see the figures in your business case on this. DC-DC conversion of anything commercial that holds just a little power (over 1 Ampere) still uses DC-AC-transformer-AC-DC as a path. The trick is that the new transformers are a lot smaller because of the IGBTs switching to a much higher AC frequency.

Now tell me, where can I get IGBTs that can switch 132 kV at thousands of Amperes, transformers that will efficiently convert these voltages, and capacitors that will reliably smoothen out the rectified high frequency voltages so it's true DC again? How much will they cost? How much more efficient will your installation be? If you have the answers to this, you could be the great leader of the next big company, just like Ford, Gates and Jobs, to name a few.

Re:business case?

[Muad'Dave]

Ask these folk.

More to the point,

[mbstone]

Why not build the entire data center out of CMOS so it uses no power?

Cirrascale racks can already take 48VDC power

[Anonymous Coward]

What's the big deal? Cirrascale.com already has racks that can take 48VDC. as well as 480VAC, 400VAC, 208VAC all in 3phase.

Re:Cirrascale racks can already take 48VDC power

[Skapare]

Their "solution" still needs work. The optimization required is to get all the AC to DC conversion done in one step AND parallel the backup batteries. A blade enclosure taking ANY of those voltages you list is going to ALSO have to do more conversion steps since the final board in the blade needs other voltages. 48 volts would just be an intermediate step, so that's not practical. Perhaps what they can do is convert the AC to board level DC in the blade enclosure AND ALSO have a DC connector from the battery (at the same voltage), and do the parallel/transfer operation in the enclosure. But I don't see anything like that in their description.

A better idea might be to make battery packs AS BLADES and have the enclosure smart enough to handle either a CPU board or a battery pack in ANY slot (then we can decide what balance of compute power vs. backup power we want).

My 2 cents

[sdk4777]

First, a lot of people here claim that AC is safer because of muscle reaction i.e. that one has some time to let go, whereas with DC this wouldn't be possible. This is true, but having a current running through your body for some time isn't directly going to kill you. It depends on the time and magnitude of the current and on the path it takes. Generally, the body can manage DC currents for longer than AC currents with the same peak value. Why? Because of ventricular fibrillation. At 60Hz, this is very likely to occur, and also the cause of death when looking at current levels far below the 'instant fry' level. Have a look at IEC60479. For much higher (kHz) range frequencies, the situation changes in favor of AC, due to the fact that the current will choose to flow on the surface of the body and not through it. It will still give you nasty burns, but you're much more likely to survive. ................. For the 380V choice, it is advantageous wrt to 48V because of efficiency; Modern medium to high power supplies are mostly made with a boost PFC, upconverting the voltage to a DC bus of around 400-500V, followed by some kind of bridge converter (forward, ahb, LLC). Now, we can take out the PFC stage and rectifier, and also simplify the EMI filter stage - this will take out somewhere around 5% of the loss power while reduce power supply cost and increasing reliability.

Re:My 2 cents

[sdk4777]

I meant to say 50%, not five.

You FAIL

[Anonymous Coward]

"DC can't really travel far at all without significant losses"

What utter bullshit. Power utilities are using DC for long distance transmission because it's *better* than AC. You fail HARD.

Electricity Flame War!

[plerner]

Is AC better than DC? Or is DC better? Which do you prefer for home or work? Is it good for Home Office? How about if I am using a Mac? Or perhaps a virtual desktop? There is one and only one way to go AC/DC, and that is a one way highway... http://www.youtube.com/watch?v=oxOpz-feD-g

Re:Electricity Flame War!

[Skapare]

I found this one to be more fun: http://www.youtube.com/watch?v=FivDcIHfguU

12vdc is the way to go

[Cyfun]

And in a power outage, you could partly power your house from your car!

Not saying it's not a good idea

[DragonHawk]

Yes, and that's because telco equipment makers keep trying to conform to the way things were done 100 years ago. There's no reason for a computer datacenter to use 100-year-old standards and methods.

Not saying it's not a good idea. For that matter, I bet you could come up with a backwards and forwards compatible solution if you tried hard enough, although it would be big and bulky and thus suboptimal on that front. But AT&T stopped doing that kind of development work after the divestiture. One of the few drawbacks to that decision.

Re:Not saying it's not a good idea

[Grishnakh]

You don't need backwards compatibility. This is a datacenter, not a telco installation. Datacenters aren't run by telcos, and have absolutely nothing to do with telcos. So there's no reason to have any kinds of backwards compatibility or any relation at all to the way telcos do things. Backwards compatibility in actual telco equipment probably would make some sense, so that telcos have an easier time integrating new equipment into older locations (plus their technicians are used to their crazy standards), but again, we're talking about datacenters here, which have absolutely nothing to do with telcos, so there's no reason to even consider them or the way they do things. Just because these datacenters are thinking about switching to DC power distribution doesn't mean they should even give one thought to the way telcos handle it. Cars use DC power distribution too, and they certainly don't use spade connectors and screw terminals any more; they use rather sophisticated connectors, and also waterproof connectors under the hood. The military also uses DC power distribution in places too, but that doesn't mean these datacenters need to use $10,000 circular connectors made of Inconel.

10 mA

[DragonHawk]

I'm not shure 10mA is enough to kill an average person.

That was the figure quoted during the safety course I had to take when I took the job at Cabletron. I have no idea where they got it from. They could have made it up, for all I know.

But assuming that 30 mA figure has anything to do with what's actually safe is just that -- an assumption. It might have been driven by something else entirely. For example, in the US, under NFPA NEC rules, anything below 50 volts is considered to be harmless and outside of their jurisdiction. This figure was arrived at not because of some careful medical study or safety data, but because the US phone system runs at 48 volts DC, and NFPA didn't have the political clout to take over that.

Standards are written by the people who pay for them, don't forget. :)

Re:10 mA

[rev0lt]

Standards are written by the people who pay for them, don't forget. :)

Maybe in the US, not here :)
I think there is some scientific background to the 30mA value, but I'm too lazy to search for it. But the 30mA value is used in most residential differential switches across EU, so probably is an international value.
Please note that the 30mA is from the ground wire, possibly originating from a chassis shortcut, so someone touching the equipment would only be affected in the worst case scenario (direct contact with the electrified area and a clear shorter path to the ground). Btw, in industrial installations, the maximum ground current tolerance is 300mA, because many industrial equipment can charge the ground path with more than 30mA.

I hope they get this right

[Skapare]

Your power comes in from the grid (and the backup generators) at AC. So you must do an AC to DC conversion somewhere. The issue is that doing so in the computer's own PSU is bad because there are issues getting the AC to the computer efficiently. Plus, there is also some power loss in doing the voltage conversions once in the PSU for "standard" DC voltages, then again on the motherboard for specific voltages it needs.

The best ideas basically run a single DC voltage to the computer and motherboard, and let the motherboard convert to what it needs. They get this single DC voltage in bulk from a single conversion from AC, and parallel the single DC with batteries. How large those batteries needs to be depends on whether you have generators. If you do have generators, smaller batteries are needed to carry the equipment long enough to get the generators started. If not, you probably want longer batteries so you can defer a shutdown long enough to see if the power will come back quickly (this generally needs to be 5 to 10 minutes), plus the time it takes to be sure all shutdowns are done gracefully (10 to 15 minutes total). Generators are usually good to go within 2 minutes and many a lot faster than that (in a few seconds).

The idea is to avoid those dual conversion online UPSes which result in AC to DC, DC back to AC, and that AC back to DC (and another DC to DC with common computers).

The big issue to be answered is just where to make the conversion from AC to DC. At one extreme is one big conversion for the whole data center, and run DC all over the place. This begs the question of what voltage to use. Too low and you have to way oversize the conductors and also have a massive fault current issue (arcs that go BOOM instead of just sizzle). Too high and it cannot be safely handled due voltages that kill people, arcs that don't stop flowing across switch gaps, and hugely expensive circuit breakers to interrupt without the benefit of a zero crossover that AC has.

AC and DC are both unsafe. But we've come a longer way with means to make AC safe in the hands of untrained people. The big hazard to the masses has been getting an electric shock. And we have a lot of solutions that work on AC. Most of them can also be used on DC. But DC, especially at these levels, has some extreme hazards that do not involve people getting an electric shock. DC arcs are harder to snuff out. DC is harder to make ground fault detectors.

IMHO, the safer system involves bringing AC all the way to the cabinet and do the AC to DC conversion here. This does open the option to use somewhat higher voltages like three phase 480/277 as found in North America. Europe and the rest of the world should just stick with 400/230 (but I do also suggest North America learn to work with 416/240 since it offers a direct 240 volt AC connection that can be directly used on equipment not able to use the DC system). That DC in the cabinet will now be paralleled with batteries and fed to all those single DC voltage motherboards that Google wanted.

My point is that it is safer to not mess with DC at such a large scale. 48 volts DC is about the same as 240 volts AC in regards to safely. But a data center wide 48 volt system is not optimal due to huge conductors, and 380 volts DC is overly dangerous and that danger goes all the way to the cabinets where people untrained on how to handle high power circuits would be working. (This is not an issue on the Klingon home world where they probably run everything at 2400 volts)

We are waiting for manufacturers to make the equipment that can handle this. Motherboards with a single DC input (though I'm not sure 12 volts is best, it is at least common and offers some advantages) are needed. Power supplies for older motherboards with that same DC input are also needed. Network switches and other devices used in these cabinets also need to take this DC voltage input. Then we need a "UPS" system that has the AC to DC conversion, DC output at the decided voltage, and blade-style battery packs so we can keep them small enough for one lanky sysadmin to replace (with integrated safety circuit so they don't arc when contacts touch, but have to be activated once plugged in).

WTF??

[Anonymous Coward]

"FYI those of you who are thinking "Oh but 380VDC could be used in a 240VAC PSU if we take the rectifier out" you are RIGHT except for the fact that these are switching PSUs so... no... it wouldn't work at all."

Hullo? Are you clueless?

Throwing the baby out with the bathwater

[DragonHawk]

Datacenters aren't run by telcos, and have absolutely nothing to do with telcos.

A "central office" and a "data center" are basically the same thing. And indeed, there's quite a bit of overlap between the two worlds. So "absolutely nothing" is a bit strong. And if you do throw away everything and start from scratch, you loose compatibility with all the existing equipment. You also potentially discard lots of proven equipment, standards, and technologies. Maybe that's worth it, maybe it isn't. There are often good reasons to change things up. At the same time, though, gratuitous incompatibility doesn't do anyone any favors.

Re:Throwing the baby out with the bathwater

[Grishnakh]

Again, there's no existing equipment that you need from a telco that you need in a data center. Datacenters use racks of server computers, and right now they don't use -48VDC to power them. I've worked with telco equipment; it sucks. There's no reason to adopt those 100-year-old standards, especially the idiotic screw terminals they use for everything, unless you're trying to make things more difficult for the technicians who install and maintain stuff (or you're trying to create make-work for union workers). What "existing equipment" do you think a datacenter would want to use from a telco? Some landline switches? DSL switches? PON switches? Why would a datacenter need any of this stuff?

And no, there's no real overlap between the two worlds, except that they both involve electricity, rack-mounted equipment, and electronics. Datacenters use fairly standard rack-mount computers; telcos use specialized telecom equipment. They don't even use the same size racks. Telcos use -48V because that's what they've always used, and they used to use it because they had giant banks of batteries powering everything in case the power went out. Datacenters don't do that now (they might have some kind of backup generator, maybe), and they're not going to switch to that.

If you think exposed terminals are so great, do this: modify all the 120V (and 240V too) equipment in your house so that instead of using standard 3-prong connectors where there's no exposed conductors, change them so they all have 3 screw terminals on the outside, and whack off the ends of all your power cords, crimp some ring terminals on each wire, and connect them. Make sure you get all three wires right too; you wouldn't want to swap hot and neutral, and you certainly wouldn't want to swap hot and ground. Make sure to do this on your dryer and oven too, and keep the terminals exposed and near the floor so your kids or pets can touch them easily.

"Not known to be bad" != Good != Bad

[DragonHawk]

Again, there's no existing equipment that you need from a telco that you need in a data center.

There's quite a few "data center" products on the market that are designed to run off 48 VDC, including computer power supplies, servers, Ethernet switches, management equipment, etc. And then there's the power equipment itself -- rectifiers, batteries, wiring, etc.

Now, as I said, maybe the advantages of better designs would make it worth getting rid of that stuff, but maybe not. I'm not assuming just because the telcos invented it that it has no application to the datacenter.

I might remind you, the transistor was invented by AT&T to switch phone calls. Unix? Came out of AT&T. Information theory? AT&T. If we get rid of everything the telcos invented, there's not going to be much usable stuff left.

Since you seem to have some trouble with logic, I'll be explict: This doesn't automatically mean everything the telcos do is a good idea, either.

... 100-year-old standards ...

You keep saying that telco standards are old, like that automatically makes them inferior. "New" is not automatically better. When I hear about a standard that's been around for 100 years, that makes me think there may well be a good reasons it has survived so long. There is certainly going to be a large body of knowledge, experience, and products catering to it.

For this discussion in particular, the telcos do reliability better than just about anyone, so yes, I think it makes sense to at least look at what they're doing to see if one can learn from it.

Telcos use -48V because ... they had giant banks of batteries powering everything in case the power went out. Datacenters don't do that now (they might have some kind of backup generator, maybe), and they're not going to switch to that.

Um. I'm not sure what data centers you've been in, but I can't imagine a real data center without backup generators. Many of them also provide short-term supplies to carry the load until the generators start.[1] Sometimes it's capacitor banks or flywheels, but sometimes it's old-fashioned lead acid batteries. So, yes, giant banks of batteries very much are a part of some data centers.

The whole discussion is about DC in the data center. The telcos do what they do because it *makes sense*. You run all your equipment off of on-site batteries. You convert from city power to site DC once, when it comes into the building, and use that to keep the batteries charged. If city power fails, the batteries just start discharging for a bit. There's no switching delay, no inverter loss, no transition artifacts. If power's out for longer, you start your generators up.

It makes a heck of a lot more sense than converting AC to DC, then back to AC, then back to DC again, which is what many other UPS systems do.

[1] Some leave that to the customer, so you get lots of smaller UPSes in the individual racks instead. Inefficient.

If you think exposed terminals are so great...

I don't. In fact, I stated rather close to the exact opposite. I'd appreciate it if you didn't build straw-man arguments.

VW's literature on the EPS

[zerofoo]

http://www.volkswagen.com/vwcms/master_public/virtualmaster/en2/experience/innovation/technical_glossary/electromechanical.index.html