Do-it-yourself UPS

Giampiero writes "Over at dansdata.com some guy named Dan creates a UPS out of some spare parts. To sum it up, "if you're looking for an industrial-capacity UPS solution, and don't like the prices of the off-the-peg options, it might be easier than you think to roll your own."" Of course you can mentally substitute U.S. 110 volts for Australian 220 volts wherever necessary...

Comments

Be Careful

[TibbonZero]

Please, everyone take caution when working with high voltage and moreso, high ampre compenents. We don't want any fried ./ers

Re:Be Careful

[GigsVT]

People often misunderstand the electrocution hazards presented by electricity. Yes, technically it's the amps that hurt you, but the volts have to be there too.

I could hold a 1 volt 300000 amp power supply's leads all day and not be hurt. The reason is Ohms law.

Your body generally has a pretty high resistance. Ohms law states that amps=volts/resistance. Your body is probably between 20,000 and 300,000 ohms, depending on which part you are talking about. Wet or sweaty parts have lower resistance. Higher voltage is more dangerous, because more amps will flow through your body. A 500 volt at 1 amp power supply would probably be lethal, especially if you had wet hands.

A rule of thumb is that anything above 50 volts should be treated very carefully. This is about the threshold of where you will normally start to conduct possibly dangerous amounts of current. If your hands are sweaty or wet, or you are grounded well for some reason, better cut that down to 30 volts.

Here's a link

The parent post is correct though, be careful in any case. Don't try this stuff unless you know what you are doing around electricity.

Re:Be Careful

[lostchicken]

A lesser known danger is that, at least on my APC BackUPS 450, when the system is on backup after one has pulled the power main to the ups (unplugged it), the male plug that would go into the wall is still hot!

I haven't measured it, but it felt like 110VAC across the prongs to me when I discovered this effect by accident. Be very careful with these things...

Re:Be Careful

[stienman]

This indicates that something is seriously wrong with your UPS. If it did drive a current or voltage across the input power plugs it'd be powering the building grid which is not only illegal, but those doing so are liable for any damage to the power grid and associated line workers. They take just as much protection for a downed grid as a live one for stupid users who try to power a downed grid, but several die each year because someone's power went out and their generator/UPS/alternative power system was back feeding the utility company when the grid was down. Connecting two mismatched live grids together is fun for the whole family.

If your UPS is UL listed then there are several regulations which govern just this sort of action. Report the problem to APC, if they don't do anything about it (!!!) then report it to the UL and/or BBB.

I'm being completely serious. For it to go this bad there is certian to be more wrong with it. I wouldn't trust it to power anything worth more than $10.

-Adam

Re:Be Careful

[Alpha State]

It's quite possible that it was 110V! The mains AC goes through a transformer and rectifier to DC to charge the battery. The transformer can operate in reverse to generate high voltages! Although the rectifier will not work in reverse to generate AC, a voltage spike from changing the battery current could still drive the transformer. Of course, commercial UPSs should have safeguards to prevent this. Alternatively, the socket could have a loose connection creating resistance - I had one socket which gave PCs trouble - took ages to work it out as it would work when plugged into another socket. Working battery powered equipment like a UPS is not like working on a toaster - it is not safe once you pull the plug out of the wall!

Re:Be Careful

[compwizrd]

measure it, cause it's in the 30 volt range, if i remember right

Re:Be Careful

[lostchicken]

I'll do that.
A company like APC should have something in their product to keep from delivering 110VAC to my hand.

Thanks for the advice.

Re:Be Careful

[Fulcrum of Evil]

A more paranoid approach is to check wattages

Cue the story of the navy engineer that killed himself with a muultimeter (runs on a 9V). See, he wanted to measure his internal resistance (which was fairly low), so he held the leads, one in each hand, and punctured his thumbs. Long story short, he discovered that 80mA across the heart = 1 dead Seaman.

Re:Be Careful

[Phork]

Im guessing it would only do that once each time it's unplugged. Im guessing the capacitors dont have good enough bleeder resistors on them, so your hand is serving to discharge the capacitors, This could be giving 117v, it could be giving less, it could be giving 600, but more than likely it is a very low amperage. So i dont think it's much to worry about, but it is a bad design.

Re:Be Careful

[paxil]

I could hold a 1 volt 300000 amp power supply's leads all day and not be hurt. The reason is Ohms law.

Although what you say is mostly correct from a practical point of view, try not to forget that "ohms law" is an idealization and only applies to a small class of conductors. What G. S. Ohm noticed more than a hundred years ago what the in metalic conductors at a constant temperature the ratio of electric field to current density was approximately a constant, known as the resistivity. (note that this is a microscopic statement.) Working from this assumtion that rho=E/J, and assuming an isotropic, ohmic conductor, one easily arrives at the more familliar version of ohms law: R=V/I. Now this is a very usefull result, but please remember that it was drived only after making multiple assumptions regarding the nature of the conducting material and has noting to say with regard to what is going on at a smaller scale.

Anyways, just be carefull you don't read too much into "ohms law".

A tunnel diode is not a bad example of something you can hold in your hand which is most deffinately not ohmic and exhibits some interesting behavior because of this.

Electrochemical systems (such as collections of cells in a human body) are another good example of decidedly non-ohmic systems. Pick up a physical chemistry book if you want to learn more.

The take home point is: Yes, you could probably hold on to a 1v supply with no problem, but the effects of electricity on the body are not as simple as you make them seem.

And, No, I am not one of those people who think cell phones or power lines are harmfull :)

Re:Be Careful

[threephaseboy]

YES! Putting a diode across an AC line is a GREAT way to rectify the power, and thus turn AC into pulsed DC. Oh wait, i'm sorry, i forgot you are a flaming idiot

Re:Be Careful

[BlueUnderwear]

but I doubt it was 110V. If he touched the ground pin, he was grounding the entire circuit through his own body.

Back in the day, I had this happen occasionnally when unplugging my Commodore 64's transformer. (No UPS, just a normal transformer). Happened only when touching it right after unplugging it, seconds later it was ok. Could have been the inductance of the transformer doing funny things.

I'm pretty certain it's far less than line voltage too, as here in Europe line voltage is 220V which would be lethal if touched in such a way.

Re:Be Careful

[david.given]

Or even low voltage, high current devices. Car batteries are designed to supply 300A at 12V. Yes, that's three *hundred* amps. And it's unfused. They warn you to take off any jewellery while working on a battery. Do you know what happens if you accidentally short, say, a wedding ring between live and ground (such as the car chassis)? The ring flashes into vapour and neatly strips the flesh from your finger. Not pretty.

Treat 'em with respect, people. Big batteries are seriously kick-ass things.

Re:Be Careful

[GigsVT]

The UPS is working fine. The manual says to never unplug the UPS when it is operating. This is because it needs a proper reference ground, otherwise the ground and chassis float with a fairly low voltage, 50-90volts. Enough for you to feel it I'm sure.

Re:Be Careful

[jhines]

The conductor that is capable of handing 300000 amps would be heavy enough for you to cause bodily harm, should you drop it on any part of your self.

Getting crushed to prove the electrical point is a great Darwin canidate.

Re:Be Careful

[GigsVT]

Hehe, good point, lets assume they were wall mounted busbars. :)

Re:Be Careful

[markmoss]

I heard this long ago in Air Force tech school, only the victim was an Airman. I think it's a myth; as far as electricity goes, blood is dilute salt water, and most everything else is dilute salt water with various membranes in the way, and I don't think 9V could put that much current through it. But it is definitely worth remembering that most of the body's resistance is in the skin - and so the jolt you get depends very much on the quality of the skin contact.

Lightly touch a small exposed 110V wire with dry skin, and you just get a little tingle. Grab a big chunk of metal in each hand, and 50V might be more than enough to freeze your muscles with your hands clamped, so you stay there and fry. That's pretty well understood, but another thing that isn't so understood is what rings, etc., can do to you. If a live wire touches the ring, the electricity reaches a much larger area of skin, and the current is multiplied accordingly (if the other side of the circuit is also a large enough or wet enough area). And you might be sweating under the ring, too.

I worked as a repair tech for six years, and because you have to power the stuff up to troubleshoot, and because it's _broken_ so the power doesn't always stay where it belongs, I've gotten zapped many times. It never hurt me - aside from minor bruises from jumping away from the electricity. The rules: no rings, no watch or a cheap plastic one, plastic instead of wire-frame glasses, watch what you're doing always, keep one hand where it can't touch metal, and brush any metal you might have to touch with the back of your hand first (so reflex will break the contact if it's hot). And learn NOT to jump when you're up a ladder...

Re:Be Careful

[markmoss]

This might not actually be dangerous (unless you jump too far when you feel that tingle and crack your head). Solid state electronics always leaks a little bit. In most power supplies, the first thing the cord goes into is a bridge diode, which converts 120VAC to about 340VDC. If these were perfect diodes they would completely block the DC from the power prongs, but since they are real devices they have a few micro amps of reverse current. This might charge the prongs up enough to feel a shock, and it might be measurable with a voltmeter, but there's no way you could get enough current to be hurt. (Except if a diode was shorted, but in most cases the device would then be quite inoperative. Of course, if you work on "broken" electronics, it's hard not to sometimes find something like this the hard way.)

Another place I've seen this is with solid state relays. Is it really switched off? Hook it up to the line, touch the output terminals with a voltmeter, and it reads nearly line voltage (440VAC the time one of our machine repairmen came running to me). Put a 100K load across it, and there's about 5 volts. The relay is indeed switched off, but because it's a block of silicon, some electrons will sneak through it anyhow.

All this gives me a real liking for those big old electromechanical relays that physically separated the input and output when they were opened. Especially the ones with springs so heavy that welded contacts would either tear apart and open, or break open somewhere else... There's no question that when those are off (which often can be visually checked), they are really off. But they are expensive and heavy, and the big ones need solid metal mounts or everything shakes when they operate.

Re:Be Careful

[Fulcrum of Evil]

I don't think 9V could put that much current through it.

9V is irrelevant - it's Amps that kill, and 100mA across the heart will kill. As it turns out, the body's internal resistance is fairly low, so 9V is more than capable.

Do it yourself UPS?

[Anonymous Coward]

I for one do not have time to truck all my packages individually across the country!

Oh ... um ... never mind.

Do-it-yourself UPS? It's been done.

[TheFrood]

I've been doing this for years. I have a big brown truck, and whenever I want to send something to someone else, I just put it in a box, hop in the truck, and drive it over to them. I've even got a little portable touchpad for them to sign on, so it feels like the real thing.

TheFrood

I don't know...

[bravehamster]

..."rolling my own" world-wide shipping conglomerate sounds like a little too much work to save a few bucks on shipping. I'll just stick with Pullman Brown, thank you very much.

On the other hand, would I get to wear the little shorts? Chicks seem to dig those.

Cheap UPS

[delta407]

Or, then again, instead of building one you could scour your local office supply stores and just might happen to get lucky. I got a "last year's model", 650 VA UPS, new but in a beat-up box in the clearance bin. It really looked like trash and subsequently was repeatedly marked down from well over $100 to $10.

Somehow, I get the nagging feeling that this pristine condition UPS (that I'm using right now) was worth more than $10 :-)

Re:Cheap UPS

[Simon Garlick]

My story:

A well-known power accessory company which shall, for obvious reasons, remain nameless* recently released a new model of their networked rackmount UPS, thereby necessitating a price reduction for the previous model. After some calculations by representatives of the nameless power accessory company*, it became evident that the exorbitant "official testing and certification" taxes charged by the state govt would, when the price reduction on existing stock was taken into account, mean that the company would barely break even on sales of the older, but just as fantastic, UPS units.

Outcome: the units got "misplaced".

Some quick emails and phonecalls went out, and guys-who-knew-guys-who-knew-guys made a series of surreptitious visits to the warehouse. Along with many others, I walked off with a direct-from-factory, unopened, 1400VA, DHCP-addressable, http-serving, rackmount UPS. Insane.

God bless bureaucracy!

*APCC

Why convert DC to AC to DC?

[cyberformer]

A major part of the description on the site (and the cost of a "real" UPS) is how to convert the DC output of a UPS to the AC power required by most PCs.

Of course, PCs don't actually use AC power: there's a big box in the back that converts all the AC input back to DC. So why not save some money and bypass this, by running the PC straight off the battery (like a laptop)?

Re:Why convert DC to AC to DC?

[delta407]

Well, for one, your computer isn't simply running 12V -- you need 5V and in most cases 3V as well. (Possibly others.) Additionally, some components (AMD processors in particular) have very narrow operational ranges in terms of how much juice they get, and battery levels fluctuate.

Yes, you could work around this without going to AC, but it's easier and more flexible to just provide AC and let the power supply worry about the rest.

Re:Why convert DC to AC to DC?

[seanadams.com]

So why not save some money and bypass this, by running the PC straight off the battery (like a laptop)?

A latop doesn't run "straight off the battery". It has a switching power supply circuit which is not entirely unlike your desktop's AC->DC supply. Most of the stuff in there runs at 3.3V, whereas the battery is 18V or more. So you still need a power supply.

For desktop PCs, a 110V->5V supply is cheaper than a DC-DC supply.

Re:Why convert DC to AC to DC?

[Phosphor3k]

ATX Mobo connectors have to supply voltages of +12v, -12v, +5v, -5v and +3.3v. The molex connectors supply +5 and +12.

As a tip, if you need alot of voltage for a fan, you can splice the -12v going into the mobo, and put the fan between that line and a +5 or +12 to get a potential of 17v or 24v to run your fans off of. My Athlon almost runs at non-nuclear temps using this trick.

Re:Why convert DC to AC to DC?

[steveha]

I have, for years, wished for a DC input on the PC's power supply, and a standard for UPSes that would plug in to that DC input.

Taking DC, converting to AC so the PC power supply is happy, then the power supply converts to DC... it would be nice if it could just be DC all the way.

The problem is that your monitor, modem, etc. all expect AC power. It's easiest just to make a UPS that provides standard AC power, and plug everything in.

Someday, I think we will have "smart" plugs. Wall power outlets will not be live by default; they will only serve power when a proper coded request goes in over a smart plug. The device will be able to tell the wall outlet what kind of power it wants, and the outlet will be able to tell the device what kinds of power it can offer. Then little kids will stick butter knives into wall outlets and not get fried; PCs and monitors will ask for +5 and +12 volts DC and get it; and UPSes will be able to feed +5 and +12 volts DC to those PCs and monitors.

Actually, if you have the complicated smart power system I envision, there will probably be a UPS integral to the system. When your home loses power, the smart power system would broadcast a "power interrupted" signal and devices like your refrigerator and your laser printer will power themselves down; your PC will run for about 5 minutes and then power itself down (unless you are there and override it) and medical devices will run indefinitely. Maybe only the DC devices in your home will be on the UPS by default?

steveha

Re:Why convert DC to AC to DC?

[Oculus Habent]

The Power Supply in your computer may not mind square waveform because it's just making nice clean DC out of it, but I think the more sensitive components in your computer wouldn't like it so much.

Of course, you could go laptop auto-adapter style and have a clean DC-to-DC conversion.

Say, anyone feel like a UPS-in-a-case mod?

Re:Why convert DC to AC to DC?

[kfg]

Very simply because in most applications simply plugging things into the wall is easier and cheaper overall. This does end up with the typical home situation where every DC appliance in the house has its own AC/DC power supply.

The ideal situation would be where there was a "household" powersupply put in with the normal wiring harness, but we don't do it that way for a variety of reasons.

I *some* applications your idea is actually ideal though, those places off the grid, such as mountain cabins and marine installations.

To make it work you'll need a deep cycle battery, ("car" batteries are actually damaged by being allowed to run down), and a voltage regulator circuit, ( both the bits and instructions on how to make one available at Radio Shack). You'll also need to run a 12v LCD monitor. The whole rig will cost you about $100 US, and a couple hours of your time if you already know how to do it, and maybe 12 hours if you have to do the research.

Once upon a time this sort of electronics hack was just as popular a hobby passtime as software hacking is now and it's a quite doable project, but maybe a bit silly for straight home use.

KFG

Re:Why convert DC to AC to DC?

[Ig0r]

None of those ICs will be able to source enough current (15A for a cheap power supply) individually to economically replace the system's PS.

Re:Why convert DC to AC to DC?

[Fastball]

Even better would be cordless, plugless power. My apartment looks more like a heavily booby-trapped jungle from all of the tripwire power cords I've got strewn about.

Is it possible to transfer electrical power from a source to a device without a wire? If so, anybody know of any related work? I'd be interested to know.

I'm all about wireless. The more and sooner, the better.

Re:Why convert DC to AC to DC?

[mindstrm]

The reason for doing so is regulation. You would have to have a solid dc->dc supply inside the box in order to ensure everything worked well.

Your wall system sounds neat, but the problem arises that it is hard to regulate power on such large lines. You usually regulate DC, especially when dealing with low voltages in computer, as close to the destination as possible to avoid interference spiking the voltage.

In fact, to make it work, you'd need the power conversion circuitry at the wall plug.. otherwise you add in all kinds of resistance along the wiring as well.

Re:Why convert DC to AC to DC?

[Permission Denied]

These are some very interesting ideas: on the other hand, after some thought, I realized I wouldn't want a house and appliances like this.

This violates KISS: with the current setup, about the only thing that can go wrong is that I trip a breaker or blow a fuse. I can then go to my box and flip the breaker or go to the hardware store and get a new fuse. With a complex system like you imagine, almost any problem would require a professional electrician to fix and expensive components to replace. Also, batteries don't last forever, so you now have a recurring maintenance cost if your UPS system is "on-line" ("on-line" is what the UPS people call systems where the input AC feeds the battery and the battery feeds the output and "off-line" systems are where the AC passes through except when there's a problem, in which case there's some very fine circuitry which very quickly switches to the battery - obviously marketing jargon, an engineer would have come up with more descriptive terms).

This is sort of like what happened with automobiles: a friend of mine used to have this '67 VW Beetle. I flipped through the Chilton's for that car, and the great thing is that you could understand ALL of the car and fix most things yourself. I have a 1999 Pontiac Grand Am, and I can do very little work on this car since I don't have expensive diagnosis equipment.

Very interesting ideas, but I certainly don't want to be one of the first to have such an experimental house: I'll stick to simple breakers and fuses for now. However, if your ideas are implemented on a wide-spread basis, I'll jump in: as much as I like the idea of tinkering with a '67 Beetle, I enjoy driving my Grand Am much more.

Re:Why convert DC to AC to DC?

[steveha]

Your wall system sounds neat, but the problem arises that it is hard to regulate power on such large lines.

I don't know much about the practical details of electricity. I had sort of pictured several wires, one set with 110 Volts AC and one or more sets with DC, probably on smaller wires.

Would the hard-to-regulate problem go away if we had room-temperature superconducting wires?

to make it work, you'd need the power conversion circuitry at the wall plug

How bulky is the equipment needed to regulate DC and/or split multiple voltages off one DC line?

I assume this needs transformers, rectifiers, capacitors, and other bulky stuff that dissipates heat. Darn. I was hoping you could have everything in one box in the basement.

steveha

Re:Why convert DC to AC to DC?

[Gaccm]

as the other poster said, most fans can't do well beyond 15v if they are designed for 12v. Your gonna burn out your fans faster by the increased heat. I seriously suggest buying a high quality ($50) heatsink (if possible, 80mm) and put in a normal fan.

Do you have your comp on your desk? Mine is so loud i did the 7v trick (run the cpu fan at 7v [12v and 5v wires]) to make it quiet, only runs ~2C hotter too.

Re:Why convert DC to AC to DC?

[SmittyTheBold]

DC gets wonky when you run it long distances, which really isn't very far when you're talking about 12V.

That's a big reason the power grids all over the world use AC instead of DC.

Re:Why convert DC to AC to DC?

[mpe]

The problem is that your monitor, modem, etc. all expect AC power. It's easiest just to make a UPS that provides standard AC power, and plug everything in.

Modems typically require something in the region of 12V DC. It is prefectly possible to build a monitor which will run from a 12V supply. The problem with "standard AC power" is that it isn't any kind of standard. Different voltages and frequences are used throughout the world. COmbined with even more different types of connectors.

Re:Why convert DC to AC to DC?

[Octorian]

Last I checked, telco systems had to run off DC due to regulations on line noise, or something like that.

Thus, you can easily buy Sun Netra rackmount servers with DC power supplies (or AC, whatever you want). They take 48VDC, and regulate it down to whatever the guts of the computer wants to use.

Re:Why convert DC to AC to DC?

[bluGill]

No, DC actually works better than AC for long distances. However the longer the distance the higher the needed voltage.

AC however has the advantage that you can change voltages cheaply and efficantly. DC is much harder to work with from that standpoint. This ability to change voltages is why AC is used. (There are several other reasion, induction moters and problems switching DC are others).

AC works poorly for long distances because the peaks travel. Power may go around the earth 7 times in a second, but AC will reverse 60 times (US) in that second. If you have generators all over the country, then eventially you will have a place where power from one generator is positive, while the other is negative, and that doesn't work very well.

Re:Why convert DC to AC to DC?

[dr_db]

Friend of mine was a radio tech in the armed forces. To get the class to pay attention, the instructor loudly said "If you make a mistake and get in the beam, the first thing that will happen is your testicles will explode"

Immediately have quiet and attentive students.

Re:Why convert DC to AC to DC?

[tzanger]

Linear regulators like the 78xx/79xx and LM317 and co. are all fine and dandy but you'll be whoring a lot of power out as heat. I'm also not sure that they have the regulation required to meet the various power quality specs that motherboards require. Go switchmode and get your efficiencies (and step responses) up.

Or, do what everyone in the telco business does and buy DC AT/ATX power supplies. -48VDC in, regular switchmode supply for all your motherboard needs. And no UPS required.

Re:Why convert DC to AC to DC?

[cgleba]

LOL. . .

For those of you who don't know, Edison had a big battle with Westinghouse whether homes should be powered by AC or DC. Edison envisioned small DC power plants on every street corner because AC was dangerous. Westinghouse envisioned a large, central power plant that drives high-voltage electricity long distances and more efficiently powers the most common electrical appliances of the time -- light bulbs and electric motors.

The battle got UGLY -- to the point that Edison invented the electric chair to prove that AC was dangerous!

Basically what I've got for my emergency radio

[Beryllium Sphere(tm)]

Coupla caveats:

Put a "battery isolator", or at least a power diode with a heat sink, between the power supply and the battery. And allow for the fact that it will drop .3 or .7 volts.

Don't even think about wet cells inside the house. Sulfuric acid belongs out of doors. There's a reason you're allowed to ship the Concorde battery without declaring it hazardous: it won't spill. Further, a non-sealed battery will release hydrogen while it's being recharged. Is your computer room free of all ignition sources?

Fred KC7YRN

(I've seen a commercial 500 VA unit for US$40)

Re:Basically what I've got for my emergency radio

[morcheeba]

I'll second the diode - it's essential!!

Some power supplies (even some fancy lab power supplies - I've seen people kill them this way) don't have a diode on their outputs, so if you turn off the supply (or, say, the power goes out), then the battery will attempt to put energy into the power supply. If the power supply isn't made for this, it could smoke and/or catch fire. A diode ensures that power only goes *into* the battery. Put the cathode end (with the band, "negative") towards the + side of the battery, with the anode (the other end of the diode) towards the + side of the power supply. Make sure that the diode can handle the current (most diodes drop ~0.7v, so dissappated power=I*E=Charging current * 0.7v)

Second, I'd be really cautious about putting power supplies in parallel to achieve more current. While this will usually work, the same problem exists that exists: power supplies don't always behave well when connected to another supply. With one diode isolating each power supply and current-limiting supplies, this should work safely, but the voltage regulation may be poor and the supplies may not share equaly - things not of great importance here. The general problem in paralleling output transistors is that, depending on the circuit, increased temperature can lower the resistance of the transistor, which causes more current to flow through it (relative to the other transistors in parallel), which causes more heat, etc... until it blows (or the current limiting of the supply kicks in).

For another project, I wonder about UPS modding. I've got a UPS that puts out the right amount of power, but the battery is kindof small. It seems that I could replace it with a higher capacity car or motorcycle battery of the same voltage.

Re:Basically what I've got for my emergency radio

[evilviper]

a non-sealed battery will release hydrogen while it's being recharged. Is your computer room free of all ignition sources?

No, but my car engine sure as hell isn't either!

Re:Do-it-yourself UPS

[rmohr02]

What's a separate-boxes do-it-yourself UPS rig good for, besides making you look all technical and competent? ...You can use a bank of truck batteries to power your PC for a week without mains, if you like.

I fail to see a practical use for this, but that's gotta be one of the coolest things I've ever heard of.

Expanding batteries

[seanadams.com]

One of the main gripes I have about the offerings from APC et al is that only their expensive, high-current supplies have the connector for expanding the batteries. This is unfortunate for those of us who care more about running time than wattage rating.

The other thing that's irritating it that they rate these things in "VA" (watts?), when watt-hours would be a more useful to know.

Anyway, it turns out that it's not too difficult or expensive to jury rig your own UPS with extended run times. Pick up some 12AWG power cable, a couple of marine/RV deep-cycle batteries (don't waste your money on sealed or gell cell). Then take apart the UPS, and wire two of the 14V batteries in series with the internal 28V supply. Oh, and use a fuse. :)

For about $500 in all, I was able to build a UPS like this that could power six servers for over 24 hours.

Re:Expanding batteries

[El_Nofx]

When I did tech support for Gateway I got a guy who worked for APC, he had a UPS on his Fridge, his microwave, his TV, his stereo, and his pc but forgot to protect his phone line, Lightning struck by his house and took out his modem, along with his mobo. After he told me all that I laughed so hard he hung up, I wonder what ever happened to him, hehe

Re:Expanding batteries

[GigsVT]

Keep in mind the AC safety rating of 12 guage wire (for wiring houses) is 20 amps. Probably less for DC. If your UPS is more than a small one, better use 6 or 8 guage.

Also, running non-sealed batteries indoors can be dangerous. Putting them in a box outside and running wires in would be best. It also allows for more expansion, and solar charging systems. :)

Re:Expanding batteries

[tconnors]

Anyway, it turns out that it's not too difficult or expensive to jury rig your own UPS with extended run times. Pick up some 12AWG power cable, a couple of marine/RV deep-cycle batteries (don't waste your money on sealed or gell cell). Then take apart the UPS, and wire two of the 14V batteries in series with the internal 28V supply. Oh, and use a fuse. :)

For about $500 in all, I was able to build a UPS like this that could power six servers for over 24 hours.

I tried doing that, by plugging in two rather large 12 V batteries in series, but blew the UPS while testing, because it just wasn't expecting to be up for more than 15 minutes, so the transformer was seriously underrated and got very hot very quickly, before developing an open circuit. Because I couldn't be bothered re-winding the transformer, I just grabbed my 350VA inverter, whipped together a circuit that flipped a relay within 4 milliseconds of the mains failing (2 cycles @ 50Hz Australian), and using the 12V batteries in parallel. I haven't tested uptime from fully charged to empty yet, because I only just finished the charger circuit, but it seems to last for 5 hours of more. Oh, of course, because my stereo is "mission critical", I have it plugged in too :)

Re:Do-it-yourself UPS? It's been done.

[rmohr02]

Then you keep asking yourself "What can brown do for you?"

Energy drain.

[Renraku]

Where's your battery-power going to go when the mains fail? How is it going to be stopped from running back into the mains? Thats the same problem with smoke-detectors that plug into the wall and have a backup battery. When the power fails, the battery is pretty much instantly drained back into the main.

The value of your data

[ObviousGuy]

Aside from corporate networks where constant uptime is absolutely necessary, why in the world would someone want to shell out for a power supply? I can understand the hack value in piecing together your own UPS, but it's all pretty much worthless in the end. Any catastrophe that could knock out power to your outlet is going to be big enough to ruin the UPS in the process.

Unless you live in Myanmar with its unpredictable power producing capabilities, this kind of expensive toy is useless. Even in Myanmar, though, you'd expect that you'd be accustomed to backing up every once in a while.

Re:The value of your data

[TheOnlyCoolTim]

I don't believe it, I just thought the quote applied.

In context the guy saying it is some sort of nutjob....

And Godspeed You Black Emperor! is about as far away from rap as you can get.

Tim

Re:The value of your data

[Enigma2175]

Any catastrophe that could knock out power to your outlet is going to be big enough to ruin the UPS in the process. Unless you live in Myanmar with its unpredictable power producing capabilities, this kind of expensive toy is useless.

Since when is a "catastrophe" the only time the power goes out? Wind, rain, snow, tree branches and many other things can cause power outages. If you live in California, you may also get hit with rolling blackouts.

Most outages last only a few seconds. Instead of all my computers rebooting, they just continue running. Rarely does an outage last long enough for the computers to actually shut down, but when it does they shut down gracefully. No lost data, no waiting for filesystem checks, just pure computing enjoyment. I can't understand why anyone would not have backup power for their delicate equipment.

Re:The value of your data

[CrayDrygu]

However, in many modern countries (perhaps yours excluded) power is generated using modern techniques that are impervious to all but the most severe weather, and the transmission lines are underground

Are you some kind of idiot or something? That's not a flame, that's a serious question.

You obviously have never been on the eastern side of the USA, where the power grids were built before anyone thought to put them underground, and since the cost to bury them would be prohibitive, the power companies don't bother.

However, even that isn't an excuse, as above ground power lines can even be seen in Seattle (evidence: 1, 2), Salem, Oregon (1, 2), and Los Angeles (1, 2).

(Note: Most of those links aren't direct evidence, however there would be no need to mention staying away from downed power lines if they're all buried.)

Lacking some features

[BagOBones]

IMHO un attended shutdown is one of the best features of the higher end UPS systems.. It safely saves your work and turns the computer of in case of an extended outage.. also what kind of surge protection do you get from that getto unit? Remote battery status? I use a UPS to protect my self from bad wiring and surges.. Hooking up to a settup like that would just be to unstable IMHO.. To each thier own, I guess

I Don't Know If I'd Want To

[dupper]

The Speed of Business has been estimated, by a group of eminent physicists and economists, as roughly equivalent to the Ludicrous Speed. Seeing as how I have restraining belts in my garage workshop, stopping for lunch might destroy my beautiful helmet, thus rendering me unnattractive to that nice Druish girl I'm holding captive. No, I'll leave UPS to the proffessionals.

Note to MOCs (Moderators On Crack): not offtopic, as comment is based on a rather uncreative allusion to what I originally thought was the story's topic after first reading its subject line, before reading the body.

Re:I want a UPS!

[plover]

According to the article, his solution can be any amount of money you care to spend, from less than an off the shelf APC ($90 USD for a small one) to far more than your PC cost you.

You'd probably get more bang for the buck though if you already owned some of the components, such as a decent marine battery or a 25A power supply.

His article was more to say that if you roll your own, you can stack up a basement wall full of lead-acid batteries and run your PC for a week offline if you want. And you can do that cheaper than you can buy a real industrial power supply.

UPS maintenance

[PD]

These do-it-yourself UPS thingies are going to require maintenance. Car batteries are designed to deliver about 3400 amps directly into the alternator of a car, only during the time that you turn the key. They they trickle charge very slowly off the engine mains.

By contrast, UPS batteries are designed to discharge at a slower rate, but charge very very quickly to be able to work if the power suddenly cuts out two or more times in a row.

The problem with using one battery where another is required is that a memory effect can develop. That slow discharge on a car battery can cause it to lose capacity and in as little as 6 months it'll be useless for any power failures over about 5 seconds.

How much does a car battery cost? about $50. But, you're going to need one every 6 months or so, making a home built UPS actually many times more expensive than one you buy at Comp USA.

I know it's counter-intuitive, but many things about batteries are.

Re:UPS maintenance

[Thornae]

The problem with using one battery where another is required is that a memory effect can develop.

The author of the article might take issue with that statement. He recently penned a rant about the whole memory effect issue for the magazine he works for, which unfortunately isn't on line. I did find this paragraph, though:

Nor, by the way, do I intend to in order to "maintain the pack?s capacity", because I do not subscribe to the myth of "memory effect". If you think your camera, laptop, cellular phone or cordless drill battery suffers from memory effect, you are wrong. Check out http://www.repairfaq.org/ELE/ F_NiCd_Memory.html and the full NiCd Battery Frequently Asked Questions file at http://www.repairfaq.org/ELE/F_NiCd_Battery.html before you flame me about this.

I don't know enough about it myself to comment, but I'd suspect that Dan would be willing to indulge in a heartily technical discussion were you to email him about it...

Re:UPS maintenance

[Phork]

correct! this is what deep cycle batteries are meant for. You can buy them in auto shops uasualy, they use them for RVs and boats. They are designed to be used fo rlong periods of time, recharged, then used most of the way down again.

Re:UPS maintenance

[Compuser]

As a physicist I had to design a portable
experiment. Imagine an industrial table
full of physics equipment. It drew more power
than any desktop or small server but needed
to run for hours. We bought a single AGM sealed
valve regulated deep cycle battery for $80 and
it has worked well. It provided power continuosly
for more than 4 hrs and as far as we can tell
has not lost much capacity yet (i.e. after a few dozen fairly deep discharge cycles). It has also
worked for a couple of years now without a hickup.
So from personal experience, this scheme works
very well, but a quality setup will run you ~$300.
Look at boating and RV sites and newsgroups for
names of good battery manufacturers and compare
prices. We have nothing but good things to say
about Concorde's Chairman batteries.

Re:UPS maintenance

[Christopher Thomas]

Has anyone built a home-made UPS yet that relies on a flywheel for temporary power delivery? Not as crazy it sounds! See this article from IEEE Spectrum [scientecmatrix.com]. Not as crazy as it sounds!

I thought briefly about this, but 1) energy density *really* sucks compared to batteries unless you're buying a carbon-fiber flywheel for $lots, and 2) a catastrophic flywheel failure is even worse than a catastrophic battery failure. It takes surprisingly little energy to make a very effective bomb (it's the momentum that gets you).

Batteries are cheap; batteries work. Just get marine deep-discharge ones and be prepared to handle catastrophic failures down the road (batteries outdoors, and limestone gravel is your friend).

Re:UPS maintenance

[nels_tomlinson]

Car batteries are designed to deliver about 3400 amps directly into the alternator of a car, only during the time that you turn the key.

Whoops, a couple of boo-boos here. That would be 300 to 400 amps into the starting motor. The glow plugs will have cold current of around 400 amps for a little 4-jug diesel, but that falls off to 40+ after a few seconds, as they warm up.

I agree with most of your information about starting and deep cycle batteries.

Re:UPS maintenance

[mpe]

Alternators are just generators with electromagnets in place of real magnets(because electromagnets, for the same size, can produce FAR stronger magnetic fields.)

Most generators use electromagnets. Some have a small permenent magnet generator on the same shaft, powering the field coils.

Depending upon the vehicle, the output of the alternator is actually controlled by varying the current through the field coil.

Or by using a voltage regulator. Sicne the voltage needs to be kept constant, regardless of the engine RPM.

Oh...also...alternators produce AC which is converted to DC. Internally they have diode packs(they look like little buttons) which are arranged to give you DC from the AC.

An AC generator is easier and cheaper to build than a dynamo.

The guy sounds interesting.

[gooberguy]

(Note - driving around the city with a passenger who's pointing a 240 volt disco strobe at unsuspecting pedestrians is neither condoned, nor encouraged, by the author. And even though it makes rain look really cool, please do not stand in the rain holding the strobe.)

It looks like he's tried this before. I wonder if he'll have an article on how to take apart your UPS and use it to scare pedestrians with a strobe.

D/\ Gooberguy

Home brewed UPS

[ClimberTech]

My friend once made a UPS out of some furbies and a big-ass fishtank... it was sort of cool, although it didn't really work, at all. But it was cool, ie: when we cut the power, one of the furbies blew up, and the water started bubbling... but the computer still shut off, o well.

Re:Home brewed UPS

[mikefoley]

I'd highly reccomend, if you are in the Boston area, and you like Furbies, to check out the Furby Wall at the DeCordova Museum in Lincoln, MA. Furby Wall

Re:Australian monopoly money

[MaxVlast]

1.18 New Zealand Dollars =)

speaking of conversion...

[trb]

$1.00 AUD = $.56 USD = €.60 EUR

The Remains

[peterdaly]

Be careful with this stuff. Batteries can be nasty. At my previous job we had a "switch room" which housed out 50,000kva (yes "K"va) UPS. On the wall across from it were huge "grab the handle and yank" circuit breakers...which were covered in battery acid from the previous UPS.

Now this wasn't your home little ups box, this thing would blend in with three refrigerators side by side, and would run a 500 person electronics factory, and 500 person office (PC's at least) for 8+ hours. That was a kickass battery box.

Just remember, UPS's can go "BOOM" and I wouldn't want to try my hand at making my own and seeing it for myself. Some things are better left to APC and crue.

-Pete

Re:The Remains

[markmoss]

Also, anytime you've got a heavy battery, remember that when there's a short, the battery is a source of almost infinite DC current. At high DC currents, circuit breakers may weld shut when they ought to open, and if that's your only protection, your battery will pour energy into the wires and short until something burns away. It's a real fire hazard.

For proper protection where large batteries are involved, use fuses - best are those cylinders bigger than your thumb, like they used for the mains in houses 50 years ago, but the 1-1/4 by 1/4 inch glass fuses are apparently good enough for a car battery.

If you're looking for exandability...

[Corgha]

What's a separate-boxes do-it-yourself UPS rig good for, besides making you look all technical and competent?

Well, it lets you have monstrous battery capacity, if you like.

Why not buy the power supply and inverter in the same box, like this one, which sells for around USD$235.

Maybe it's a bit more expensive, but it can deliver 500 W (1000 W) peak, instead of 210 W (there are larger models available as well, up to 3600 W), you can use all the car batteries you want, and you may run less risk of electrocuting yourself.

Comment removed

[account_deleted]

Comment removed based on user account deletion

Re:Awesome!

[sharkman67]

I use TrippLite power inverters with marine deep cycle batteries.

I have a 24 and 36 volt inverters. To add additional capacity all I need to do is add more batteries. The only thing to make sure is that ther e is proper ventalation for the batteries.

The inverters have ports for remote management as well as status leds.

It is a much lower cost solution and not as risky as there is not really high voltages.

Re:Fixing APC Smart Rack UPS

[plover]

The author cautions about taking an existing UPS and trying to add more, bigger or extra batteries. The power supply inside is built (cheaply) to deliver only the amount of current required to charge the battery it was designed for, and no more.

However, you can probably replace the battery in your UPS with a similar model for less than the cost of a new UPS. Either find the manual for your UPS, search for your UPS model on line, or open up your UPS and see if it has any labelling regarding the battery capacity, or (better) a "replace battery with XXXXXX model or equivalent." Find an equivalent on line (thanks, Google!)

Check with a battery store (most major metropolitan areas will have several.) It may be worth the drive, as batteries are heavy, cost much to ship, and can only be delivered via ground (slowly.) And if you're uncomfortable working with live circuits or very hazardous materials, they'll probably be able to replace it for you for a small fee.

Finally, PLEASE RECYCLE YOUR OLD BATTERY! In most states, it is illegal to dispose of any lead acid batteries in the garbage or in a landfill. They are filled with corrosive toxic sludge. Please be responsible.

Re:Do-it-yourself UPS? It's been done.

[willybur]

Brown says, "I want some more pie... *chuckle* Brown didn't say that."

http://auction.apc.com/

[MattRog]

Or, you could visit:
http://auction.apc.com/

And get refurbished, and sometimes brand-new, warranteed APC power units for pennies on the dollar.

Two years ago I purchased two 420VA refurbished backups for around $50 each. They're currently protecting my three boxen, one of which I'm on now, and have been flawlessly through brown-outs, sags, surges, etc.

Typically the refurb'd units are from companies who have agreements with APC to buy many UPSs and return them when the batteries run out. Why? Customers used to return UPSs marked as 'defective' when in fact it was simply a dead battery. So, they tape over the battery bay and simply have the customers return the units with dead batteries and pick up a new one. They replace the battery and sell the ok unit again as refurbished.

Re:why australia?

[lostchicken]

It's the other way around.

The article is about building a 220VAC/50Hz device. It converting from Aussie voltage.

I love this guy

[nutbar]

A bench power supply is also a generally useful thing. I use this supply all the time when I'm building and testing things, or when I just feel like setting fire to a pencil.

I bet he used to set fire to ants with magnifying glasses too. Not that I'd ever do anything like that. No-sireee. Stop looking at me!

You mean "Exploding batteries"

[plover]

I hope those batteries are properly ventilated, keeping all the generated hydrogen away from ignition sources such as monitor switches, light switches, etc. (Of course, you wouldn't have that particular hazard with sealed or gel cell batteries, but you didn't "waste" your money that way.)

I also hope that they're on a concrete slab or on a floor over a beam designed to hold up the unusual amount of weight. If it's just a wooden joist construction room on an interior wall, you may find your floor sagging in a year or two as the nails slowly release.

The other thing is that your UPS may run fine until a deep discharge or two, and then blow out the charger because you're drawing more current than the power supply was designed to deliver.

Re:You mean "Exploding batteries"

[seanadams.com]

I hope those batteries are properly ventilated

Did I mention there were quite a few servers? This room was very well ventilated, with a large fan pointing outdoors on one side of the room, and an intake vent on the other side of the room.

keeping all the generated hydrogen

UPS supplies use a slow trickle charge, so the hydrogen from the electrolysis is produced at an incredibly slow rate. I might be worried if I had banks of hundreds of batteries in a sealed undergruond vault... but let's be reasonable here!


I also hope that they're on a concrete slab or on a floor over a beam designed to hold up the unusual amount of weight.


Yep, these sat on the bottom shelves of heavy duty racks, which were bolted 6" into the concrete.

your UPS may run fine until a deep discharge or two, and then blow out the charger because you're drawing more current than the power supply was designed to deliver.

Like I said, this was just to get more run-time out of the UPS. I wasn't running it anywhere near to it's current capacity. I did a test run and monitored it for overheating before I put it into production use.

Don't be so prophylactic. Some of us know what we're doing!

Re:why australia?

[Zeddicus_Z]

The Dans Data article was originally written by Dan for our magazine, Atomic Maximum Power Computing.

As we're an Australian mag with a (mostly - we sell in New Zealand, Singapore and a few other places) Australian audience, its setup for 240V.

The UPS D-I-Y article appeared in AtomicMPC Issue 13 (latest issue is 18)

Re:Awesome!

[glitch!]

I have a 24 and 36 volt inverters. To add additional capacity all I need to do is add more batteries. The only thing to make sure is that there is proper ventalation for the batteries.

Even though you specifically mention that you provide ventilation, I bet you will still get people yapping at you that your setup is dangerous because of the hydrogen :-)

Just out of curiousity, how do you connect the batteries together? Do you use ordinary automotive cables connected to two huge bus bars? Or do you have some fancy connection blocks? I thought about doing something like this myself, but I never saw any really elegant way of connecting the batteries together.

It is a much lower cost solution and not as risky as there is not really high voltages.

The current can be absolutely amazing, though! I might not be so worried about touching an exposed contact, but I would be paranoid about letting any kind of metal object anywhere near the battereies!

Re:I want a UPS!

[cscx]

I doubt that. You can pick up an APC Back-UPS 500 for $100 USD if you look hard enough. Places like Staples often have them on clearance.

Plus, APC adds filtering components and surge protection. I think there is a guarantee on the components hooked to it too. I had a Cisco router hooked to an APC Back-UPS once during a bad thunderstorm. People lost equipment all over the place during that storm, and I lost the UPS... it was DEAD. But it saved the router, which was more important!

Also, you get serial port communication for logging too. Sometimes it's nice to know that, well, damn, why DID the power go out at 4:30 AM?

Thanks but no thanks... I'll be sticking with my trusty APC. =) But this is a cool idea nonetheless.

This is dumb and dangerous.

[newerbob]

I would not recommend doing this. Batteries are dangerous and can overhead or explode.

Also the cost for these components is *much* higher than a $59 APS UPS.

However, if you wanted to roll your own you would ditch the inverter and simply regulate the gel-cel down to the +12, +5, +3.3, +2.2, etc that your computer needs, and use a DC-DC converter chip to get the -12 and -5volts.

That's much more efficient and somewhat safer.

You wouldn't want a person without a CS degree writing software; we shouldn't have folks without EE degrees designing power supplies.

Old Old Old Old

[TheDarkRogue]

Back in the day I use to have an account on a BBS that did this exact thing. The sysop then got others to do this too.

On another note, one of the nice things about this is the ability to expand the source of the 12v in. For instance, a small array of solar cells wired together and into the battery taking the load off of the house current/AC->DC thingy during some days. When we did the math for it, it paid for it's self in about 4 month of sunlight every day, or about a year in real life. 12v is a nice little voltage to work with when it comes to this and alternitive energy sources. Hell, wire a nice little 12v DC generator to an exercise bike for when your bored. Get a work out and save on your electric bill. Also, if you want to go real insane/creative, suck off of the telephone lines when not in use, they got something comming through them i'm sure. It's alot easier to get 12volts then it is 110ac.

110 vs 220

[Anne_Nonymous]

>> Of course you can mentally substitute U.S. 110 volts for Australian 220 volts wherever necessary...

Holy S#!t! And I thought the Canadian exchange rate was bad!

Wireless power?

[steveha]

You can beam power around as microwaves or a laser beam or something, but the equipment to collect the power and convert it will be large and bulky. And unless you want to turn your apartment into a large microwave oven with you inside it, you would need some kind of complicated aiming system to make sure the power only goes where you want it.

Simpler would be to have your gadgets run on battery power, with charging cradles.

If you imagine ultra-low-power technology combined with ultra-high-density batteries, you would have gadgets that don't need charging often and have no power leads. Then just have robots run around in the middle of the night, charging them while you sleep.

:-)

steveha

220Volts

[loddington]

Funny I thought we ran on 240 volts in Oz

Re:Do-it-yourself UPS? It's been done.

[Fulcrum of Evil]

I've even got a little portable touchpad for them to sign on, so it feels like the real thing.

If you really want it to feel like UPS Ground, shouldn't you kick the box around the back of the truck en route?

Myth.

[mindstrm]

Totaly myth.

Not possible.

Even if the resistance was only 1000 ohms, and it was most certainly a great deal higher, he would only have 9 mA, not 80.

Re:Myth.

[Uller-RM]

Not quite. Your skin's resistance is quite high... but once it pierces your skin (it takes an arc about 400V to do it, but if you stab yourself, tada), inner muscle and soft tissue racks up about 10 ohms an inch. And if applied across the heart, 9mA is enough to trigger fibrillation.

You don't need to be paranoid when working with lethal voltages... but you do need an ounce of common sense. Work with one hand.

(Cue web surfing while masturbating joke here - but I'm quite serious.)

Re:Myth.

[rot26]

As we used to say in rokit sientist skool "volt's jolt, mil's kill".

Re:Myth.

[tzanger]

You don't need to be paranoid when working with lethal voltages... but you do need an ounce of common sense. Work with one hand.

That's not even enough. 60Hz will capacitively couple quite nicely through an inch of rubber sole to concrete.

One hand helps a lot, but it doesn't mean you can go grabbing conductors. :-)

Don't-do-it-yourself UPS

[AlexA]

So, if I didn't want to build my own UPS, what are some good brands to get for affordable yet good quality UPSes (reliable, has computer interface that Linux supports, etc.)? I've heard not so great stories about APS...

Absolutely. And there is a reason we don't do it.

[mindstrm]

It's totally impractical. You would have to ensure that no metal objects were in the house, or they would pick up the power as well.

Read up no Tesla, he was a fan of this. Lots of wireless lights in his labs. High frequency, high voltage electric fields really.

Non-redundant UPS problems

[Animats]

One problem with this is that your inverter may be less reliable than your power line, in which case you've made things worse.

There are modular inverters with N+1 redundancy, but those are usually seen only in large units.

Some words of advice on UPSes

[NewtonsLaw]

I have two UPSes -- I used to have three but one died a horrible death for no apparent reason.

The smallest unit I have is a 600VA no-name Taiwanese box with two 7A SLAs. It has no fan and, although it's just an SBS, it still runs hot as hell.

As a result of this hot running, I discovered that the SLAs tend to dry out rather quickly such that you get a much shorter time than you'd expect when the mains power goes off (as I discovered just the other night :-(

My other box is a Siemens 1KVA full-time UPS with forced ventilation and a bunch of other cool features.

It's worth noting that even a fan-cooled UPS such as the Siemens can run way too hot -- as I discovered about a year after I installed it.

Without warning, the UPS started screeming at me with a continuous alarm buzzer. It was still working but it was not at all happy.

On touching the case I discovered why -- it was too hot to touch.

I shut things down really quickly and opened up the case to see what was wrong...

Dust!

The intake holes in the front panel have a fine wire mesh over them and, since this UPS (like most others) lives on the floor under my desk, enough dust had been sucked into the holes to totally block them.

So here's a DC-AC inverter busy delivering about 500W of power to several computers and monitors -- but without the benefit of any cooling. No wonder it wasn't happy.

I blew the filters out with compressed air, checked that the fan was okay and put it back together. It's been working fine for nearly two years since -- albeit that I check and clean the intake meshes every few months now.

That the Siemens box was smart enough to warn me it was in distress (rather than just failing) shows that nobody ever regretted buying quality.

So.. rules of thumb for UPSes...

If it runs hot-- expect the batteries to last just 18 months to 2 years -- and don't wait until the power fails to find out that you should have changed them already.

If you have a fan-cooled UPS mount it up off the floor or check that the cooling holes are clear at regular intervals.

And the idea scales well!

[zerofoo]

I did some network consulting at a lawfirm that had offices in a high-rise tower. They couldn't get the EPA permits granted for a diesel powered backup generator, so they built a "UPS" room.

The inverter came from these guys here.

They hooked up half a room full of 12 volt lead-acid batteries to charging systems and inverters and put the whole system under a vent hood. We never did figure out exactly how much runtime their server room had.

-ted

Re:Awesome!

[BLAG-blast]

Even though you specifically mention that you provide ventilation, I bet you will still get people yapping at you that your setup is dangerous because of the hydrogen :-)

To a large extent the danger of explosion can be reduced by using hydrogen catalyst battery caps on all cells of the lead acid battery. These can be purchased from Hydrocap, 975 N.W. 95 Street, Miami Florida, 33150,(305)696-2504.

do-it-yourself UPS is easy

[g4dget]

Get a laptop. They come with a built-in UPS, and the power management is tightly integrated with the BIOS. They also tend to be pretty energy efficient.

Standard Internet Horror Story

[_bobs.pizza_]

• some guy creates cool thing
• said guy puts directions on his homepage
• some bored /.er finds it, submits it to /.
• editor likes it, it gets posted
• INSTANT DEATH to the guy's homepage
• inventor guy looks at how many hits he's gotten, realizes why server is dead
• guy learns not to share innovations online

This is how we thank people that invent cool things in this day and age.

Re:Fixing APC Smart Rack UPS

[ncc74656]

If you got it for cheap, used, its batteries are probably shot. Like most laptops, replacing the batteries on an older UPS would often be more expensive than simply replacing the UPS.

UPS batteries tend not to be as expensive as notebook batteries...there aren't as many different varieties, so they tend to be somewhat standardized. I recently replaced the two 12V 7Ah batteries in a UPS at home. That UPS cost about $170 when I bought it. New batteries were about $40 for two. Given that the UPS is a 900VA unit that can keep a dual-P!!! server running for about half an hour, $40 isn't a bad deal to keep it running.

Re:Fixing APC Smart Rack UPS.. ---UPS Fire story..

[Mad Quacker]

The author cautions about taking an existing UPS and trying to add more, bigger or extra batteries. The power supply inside is built (cheaply) to deliver only the amount of current required to charge the battery it was designed for, and no more.

Actually the problem is that they were designed to run only for 5-10 minutes at a time. Instead of heatsinks they just use blocks of metal that have enough thermal mass to keep it within operation temperatures for that period of time.

I found out the hard way.. wired up an additional 25Ah in parallel with the 10Ah that came with it (rated for 500VA). Ran fine for 30 minutes. Then a burning smell and beeping for no reason (battery meter at ~40%). Unplugged it and immediately opened it up (30 seconds), as soon as I did one of the heatsinks (err lob of metal) fell onto the hardwood floor, with the semiconductor still attached glowing white hot. In about 1/2 sec it burst into flames (Oxygen starved from inside case!) and the whole block started to burn into the floor. Luckily I had some pliers to pick it up so it wouldn't burn the house down. There is still a big black square burnt into the floor. Branded by my ex-UPS.

I got damned lucky because I didn't drop the heatsink on any part of myself, I imagine any clothing would have caught fire instantly, etc.

In retrospect I suppose I could just add large heatsinks and a fan and it should work. The semiconductors have to be rated for the current, but the wasted energy just has no where to go.

12v DC input ATX power supplies...

[Radi-0-head]

I think these are what you're looking for:

http://www.keypower.com/DC_power/DX-250H.htm

Re:VA? Watts?

[threephaseboy]

W=P=E*I ONLY for purely resistive loads.
W == less then E*I for inductive loads, like computers, motors, etc... anything with a power factor less then 1. On the other hand, VA == E*I for any device, and gives the total amount of power required by the device, not how much it is using.
(E == volts, I == amps)

beowulf cluster batteries?

[thogard]

At work we bought a 3kva upsonic ups. Its good for about 10 minutes. That just would not do so we hooked up 16 deep cycle batteries to the thing. Now it will run for about 4 hours on a full load and a very long time when we start pulling things offline. It takes 4 days to recharge the thing.

We got a UPS because it has a good inverter and auto switching and a built in charger. An inverter would have cost more. I've got a snmp card so I can measure different things.

You have to be careful about these things beause you get a large voltage and massive amounts of current. Our system has two 96 volt banks and good for at least 300 amps. That can kill and it can also cause batteries to explode. Treat these with the respect they deserve. We have a small circut that goes between each battery that has 4 led's that let us know if any one of the batteries go bad. Its also hooked to a monitor system so we can get real time alarms if anything goes bad.

Re:Fixing APC Smart Rack UPS

[Agent Green]

Or, you can do what I did...and instead of paying $200 for an equivalent battery, you can spend that on a couple of Die Hard Deep-Cycle Marine batteries and 4-gauge cable hardware. ;)

This was a Smart UPS 1400 RM. The newer XL models actually have an expansion plug on the back which can support a few of these puppies. I'll get an XL model off of ebay someday...but the homegrown hack will have to work for now. ;)

Used / Refurbs / Hand-me-downs

[tweakt]

I'll stick with my APC Back-UPS Pro 1000 I got for $100 at a computer show. 670W of capacity, 1000VA, will power my workstation for several hours.

Look for people selling these from companies who've tossed them when upgrading to larger more centralized backup power (APC silicon, etc).

There was another rackmount 1400VA unit for $100 when I went back a few weeks later... I was a little upset but what the hell 1000 is enough for me, and I dont have a rack in my apartment (though I'd like to) ;-)

One word of warning...

[wirehead_rick]

that was not particularly covered in the article.

I used to design UPS's for a major American UPS mfr. The author slightly mentioned that you would not want to use a normal UPS for continuous operation due to overheating. He is correct but for the wrong reason. It is NOT overheating of the battery cables you have to worry about. It is the Transformer.

Since the Transformer is the biggest cost item next to the Battery they are highly optimized for the performance/price point. That means a transformer designed to run at 400VA for 3 minutes will probably melt into a glob of laquer, copper, and steel in about 20 minutes. Of course that also means it could short out, catch fire, or send unknown voltages to your computer possibly destroying your computer PS.

Just as a point of reference the industry has made a distinction between continuous duty and back-up supplies. UPS means temporary supply to allow you to save data. Inverter means continous duty.

If you wish to try this project I make two suggestions:

1. Use or purchase an Inverter. Tripp Lite sells the PV series inverters which were designed to be used exactly in the fashion of the unit in that article. They also sell APS's which are UPS's that _are_ continous duty.

2. If you MUST use a UPS inverter, then get one that is overrated by at least 10 to 1 in VA handling capability. Depending on the UPS you get this may not be enough, but it will get you close.

3. If you are really ambitious, replace the UPS transformer with a much bigger one. Keep in mind that the power handling of the drive circuitry is important too, but hey - you want to tinker, right? Just make sure you understand the proper step-up ratio and winding configuration of the original transformer.

Not only that

[Sycraft-fu]

But UPS battires are just simple Lead-Acid battires. Generally nice sealed ones, but none the less still a Lead-Acid battery which can be built fairly cheaply. They are, after all, a very old technology. Laptops, needing a much better power to weight ratio use more advanced battery technologies. At a minimum they use something like nickel metal hydride and generally all newer laptops use something like lithium ion battires. Well, while these provide great storage for their weight, they are much more expensive to construct and hence more expensive to buy.

Why not a DIY operating system?

[ReelOddeeo]

From the screen of the computer, I can see that he builds a DIY UPS, but he doesn't run a DIY operating system.

Parallel Extra Batteries

[Mr_Perl]

A friend of mine works for a hospital which recently changed out all their batteries in their UPS as they do every few years. He salvaged a few good ones, and gave me a couple.

I connected them in parallel to the power connectors on my APC UPS and can now expect many more minutes of power outage without interruption.

Your local hospital might be a good source of batteries for a project like this since they have to pay to dispose of the batteries anyway.

Re:UPS size twaddle (was Re:The Remains)

[sysadmn]

Indeed. Just last week I saw a 24,000kVA backup power system. It's 5 tractor trailers, one of which contains a modified jet engine. The bad news is that it puts out 13.8kV or more. CS majors - don't hook that to your PC. Rent one today!

Software support

[rossz]

What good is a homebrew UPS if it doesn't support decent management software such as NUT?

Having a UPS means nothing if you lose power and the server dies when the battery runs out. You want the system to gracefully shutdown when the battery level becomes critical.

Is it just me?

[Jay L]

A UPS is an emergency backup solution for mission-critical applications. Building a UPS out of spare parts based on instructions from a web site seems like sewing a parachute using recycled grocery bags with a pattern from McCall's.

DC light bulbs

[Frank T. Lofaro Jr.]

"Regular" light bulbs should run fine (and actually last longer) on DC.

Re:Fixing APC Smart Rack UPS

[AaronW]

I have repaired a couple of UPS. Every time it
just required that I replace the batteries.

I regularly see cheap =$20 UPS at the local Silicon Valley swaps where usually the only thing wrong is that they need new batteries.

I picked up a Tripplite Omnipro 1400VA UPS for $25 and a few months later picked up 5 batteries for $20. I also picked up a small Deltec UPS for $5 and a new battery for $12.

As for the Tripplite, it appeared totally dead when plugged in with the dead batteries that came with it.

A little more info...

[cr0sh]

A few people on here are bemoaning the fact that this "ghetto" UPS doesn't have monitoring. Even the author of the article notes this. However, building such a monitor wouldn't be too difficult.

Basically, you are looking at a relatively simple voltage window comparator circuit. Most of these circuits use op-amps arranged in a differential mode to provide 2 or 3 "windows", and typically light up LEDs to show "high", "medium", and "low" voltage levels. Simple digital signal probes do the same. Most of the time, these devices can be built using cheap op-amps or comparator ICs that house four comparators in a single DIP package. Hook the LED lines to a PIC/MAX232 system and monitor the data via RS-232. Or, simpler - monitor the status of the lines using the parallel port.

The "el-cheapo" solution is to just simple start the shut-down as soon as power is completely lost. Drop a relay across the 120 VAC input, then have a 5V source (from the battery and a 7805) switched "on" when the relay is disengaged when the voltage fails. Run the 5VDC to a opto-isolator circuit (to isolate the computer port from the mains at the relay), and the ouput from the opto-isolator to the parallel port or serial port (you could monitor this directly). Configure the software monitor to check every so often, and if the power is indicated to be "out" after 15 or so seconds, begin the shutdown sequence (this will allow for transients in power that might occur - you could up this to however long you wish).

Software would have to be written for both of these solutions, but it would be rather simple, for either *nix or 'doze.

In regard to the danger of this kind of "battery-box" solution - you definitely have to be careful with this, when it is on, and "off" (ie, doing the UPS thing). LA batteries can source a lot of current - you can easily make a cheapo stick welder with a couple of good 12V car batteries - that is serious current. Also, when an LA battery (actually, any battery) is under load, it will generate heat, and in the case of an LA battery, a lot of heat can be generated. This heat can cause the electrolyte to "boil", forming funky gasses. The batteries can also explode (heck, I have even personally witnessed dry cells explode). If you want to see an "interesting" site, go to an electronics junkyard and look for an old "heavy duty" UPS (look for refrigerator size boxes needing 440VAC input). If the batteries are still in the box, take a careful look at them - you might notice that they are "swelled", some may have even leaked or burst. This is because the UPS kicked on, and remained on for a VERY long time, and the batteries literally "cooked".

What the one poster said about having cleanup supplies, etc - yeah, that could be needed. If it was me, I would store the UPS outside my house, in a ventilated shed or something...