Slashdot Mirror
University Switches To DC Workstations
An anonymous reader writes "Researchers at the University of Bath, UK are undertaking an in-depth study of energy consumption within the new network, with the aim of demonstrating that running a large network of devices on DC rather than AC is both more secure and more energy efficient. AC electric power from the grid is converted to DC and runs 50 specially adapted computers in the University Library. Students using the system have noticed that the new computers are more compact and much quieter than the previous systems. The immediate advantages of the new system are not only for the user but for the energy bill payer and the environment."
Tesla is giggling in his grave
which is totally what she said
There is no evidence or reason for DC to be more "secure". If some lame argument about it being harder to bring your own power source / utilise their outlets when you have a custom system is put forward, then, well... no.
I can understand the efficiency argument to a certain extent, although if a workstation needs enough power that a fanless AC PSU is unsuitable then the more efficient AC PSUs will be enjoying enough load to reach over 80% efficiency. Are the centralised rectifiers + wires + in-computer DC-to-DC converters as efficient?
The only thing inside a computer that actually runs on AC is the computer's powersupply. The powersupply regulates this to DC voltages! The powersupply is also quite bulky and noisy compared to the other components.
"Initial tests show that the system in Bath emits approximately half as much energy as heat than the previous AC powered system while running much faster."
Yes, I'm sure it'll generate less heat when most of that heat comes from converting AC to DC, but why the hell would it run faster when everything else in the computer is still the same?
Selective quotes from TFA:
Researchers at the University are undertaking an in-depth study of energy consumption within the new network, with the aim of demonstrating that running a large network of devices on DC rather than AC is both more secure and more energy efficient.
The new DC network also offers greater security. DC power supply units have a simpler design, with fewer parts that could fail and need replacing. The system at the University also charges a number of batteries when usage levels are low to allow the system to run independently from the grid for up to eight hours should a cut in power be experienced.
The above two paragraphs are the only I could find in TFA that mention security. I gotta ask -- can anyone speculate how centralizing the PSU would lead to a more secure system? Is it possible that there is a regional definition of "secure" to mean "very reliabile" or "very available." As in, we have "secured" a constant municipal water supply?
I guess Con Edison should have waited just a few more years. Apparently 125 was not quite enough.
I guess I could swing either way on this one.
Todd: I hope it proves as delicious as the farmers that grew them
Too bad, if he was rotating it would probably generate some power.
We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.
running much faster
With DC power, the electrons get to run laps, and every time they get to your computer, they can do a little bit of work, spreading it out among all the electrons. With AC power you got those electron thingies racing back and forth and back and forth, but never getting anywhere. Only the few electrons near the computer actually do any helpful work, and they get worn down really quickly, so they stop working as efficiently, and the CPU slows down, and it's just generally bad.
Edison: "Genius is one percent inspiration and ninety-nine percent perspiration."
Tesla: "If Edison had a needle to find in a haystack, he would proceed at once with the diligence of the bee to examine straw after straw until he found the object of his search. I was a sorry witness of such doings, knowing that a little theory and calculation would have saved him ninety percent of his labor."
I'm not sure how you came to that conclusion. AC suffers from several effects that make it less efficient and/or more expensive over long distances.
For DC, the power delivered is V*I. For AC, it's similar except the V is really Vrms - you must insulate for Vpeak, but you only get Vrms * I power. For sinusoidal AC, the difference is a factor of 1.414.
With AC circuits that have non-zero reactance, you must choose a conductor that can carry Imax, but the power delivered to the load is only Vrms * Imax * cos(phi), phi being the phase angle between the voltage and current.
AC circuits suffer from the skin effect where the power travels more on the surface of the conductor rather than equally throughout its cross-section. This requires a larger solid or stranded conductor than would be required for DC.
AC has a few things going for it - the ease with which voltage can be transformed, the ease of generation with rotating generators, and ability to drive large, multiphase motors efficiently.
Tiller's Rule: Never use a word in written form that you've only heard and never read. You will end up looking foolish.
Truthfully both approaches are valuable, and we would be a poorer planet without either of these men. It's a shame they disliked each other so much.
Edison: "Genius is one percent inspiration and ninety-nine percent perspiration."
Edison contracted out all his perspiration. That's genius.
Today if you blow a power supply (one of the most common computer failures) you lose one computer. If you blow the power supply for the office floor you might lose 100 or 200 computers.
For reasons which are a long story, I have had several servers up and running on 12V for many years now. The powerstream guys are pretty much the gold standard of ATX 12 volt power supplies, as far as I know:
http://www.powerstream.com/DC-PC-12V.htm
Note that these are "honest wattages" not the "marketing wattages" seen in the AC power industry. The price of a 300 watt DC supply seems high compared to a 100 watt AC supply from China that has a sticker claiming 300 watts. However its not too bad compared to a AC supply that actually only provides 300 watts despite having a sticker labeled 800 watts or a million watts or whatever marketing felt necessary. Also the powerstream supplies, to the best of my knowledge, are some of the few computer power supplies you can buy that do not have forged FCC and UL registries, which is worth something to me. In summary, expensive, but strongly recommend based on years of experience.
Anyway, what happens when the primary rectifier goes down, is my battery bank will run the asterisk PBX and friends for something like half a day, during which time I can source a generator and charger, or perhaps casually purchase a new supply, etc. Also I have multiple supplies any of which could theoretically power the whole works (at a cost of high heat and much shorter capacitor lifetimes, etc). So you Y-cable them to run multiple plants off one supply. Guess what, the same Y cable can be used to run multiple plants off one battery, if one fails. Etc.
Theoretically, I could run the entire phone system off an idling car, assuming you have enough gas in the tank. Unfortunately my entire plant draws just a little too much for the cigarette lighter plug, probably 15 amps total. If I could invest in new phones / new servers / etc and get total plant draw down to 5 amps, not only would my batteries be 1/3 cheaper or last 3 times longer in an outage, but I could also run the works conveniently off a car cig lighter port.
Obviously if you have zero battery capacity then you are instantly in deep doo doo, but given three or so figures of amp-hours you're good to go for a very long time.
Wire everything in Amphenol power poles, exactly like the ham radio guys so you can use their DC products, and keep a stock of extension cords and Y cables and other gadgets. Use fuses, and as a subset of that rule, only use automotive fuses because they are infinitely available. Use 12 volts as your standard because you probably own a mobile 12 volt generator (aka your car). Perhaps if you're in the .mil and have a 24 volt humvee, do 24v instead, whatever.
A DC powered system is frankly pretty straightforward and simple.
"Science flies us to the moon. Religion flies us into buildings." - Victor Stenger
Tesla was the one who advocated for AC power. Edison was the one who argued for DC as the safer choice, and publicly electrocuted horses and other animals using AC to show how unsafe it was.
You see? You see? Your stupid minds! Stupid! Stupid!
The major drawback to DC power is in the wiring. Direct current requires larger gauge wiring than AC power, which increases material costs considerably. In general, DC power is economical only if the wiring between the computers and the DC source is less than 35 feet in length. More than that, AC power becomes more economical.
FTFA:
Somehow, I suspect that the cable run to the individual machines is more than 35 feet.
His figure of 35 feet is of course completely made up, or at best applies only to one very specific situation. Its a rather complex non-linear solution that depends on current level, local union labor contracts, price of copper wire, UPS and battery capacity, etc.
Obviously, if you are charging an ipod at a zillionth an amp after a 12V to 5V converter, you can run that thru thousands of feet of small gauge (cheap) speaker wire before the voltage drop will matter. And if you're doing the thousand watt gamer PC or NAS farm you'll need something approaching welding cable to keep the voltage drop low enough. In between, well, its in between. But by no means as simple as a 35 foot cutoff.
"Science flies us to the moon. Religion flies us into buildings." - Victor Stenger
When I first had to deal with telephone equipment, I came across the -48 VDC power standard for things like SONET nodes, digital cross connects, channel banks, and telephone switches. I believe this is due to cathodic protection of buried copper cables.
You can find -48 VDC rectifiers, AB fuse panels (think redundant DC power supplies) and lots of telecom gear in racks that is powered with -48 VDC.
"We can't solve problems by using the same kind of thinking we used when we created them." -- Albert Einstein
AC is still the prime motive force in electrical generation and always will be.
No, not necessarily.
We've already started moving away from AC for long-distance power transmission, using "HVDC" instead for things like 2MV transmission lines.
The main advantage of AC is that, with no semiconductor technology available, you can easily step it up and down between different voltages using an iron-core transformer, nothing more than a bunch of iron and some copper wire wrapped around it. High voltage is absolutely necessary for power transmission, because I^2*R losses are too high at lower voltages, but high voltage isn't usable by end-users because of safety and other concerns.
Nowadays, with power electronics (giant power transistors capable of handling thousands of volts and amps) and high-frequency switch-mode power conversion, that stuff is mainly obsolete, so it's fully possible to eliminate AC for power transmission, and even get better conversion efficiency than transformers. The only reason it's really still used is 1) our infrastructure already uses AC, so you can only replace it in certain places where it won't be too disruptive (like long-distance links), and 2) iron-core transformers are still much cheaper than electronic alternatives, so it's only economically feasible to switch to DC for certain large-scale projects, not for every transformer in a subdivision.
There's no technical reason that, in the future, DC couldn't become the standard, with electronic "transformers" stepping the voltage up and down as necessary.
You're missing the point. Much of our electronics run on 5V DC. If you put a big 5V converter on the side of your house, and a bus system to connect this to all your 5V gadgets, the losses would be greater than just having all those wall-warts. Even worse, your gadgets wouldn't work, because the voltage drop between your converter and your gadgets would be so great, you'd end up getting 4V at the gadget, and worse, the drop would vary depending on how much current that gadget is drawing (and other gadgets on that run).
If you want to be more efficient, the answer is simple: throw away all those wasteful transformer-based wall-warts, and replace them with high-quality switching wall-warts instead. They're lighter and also have better efficiency, both when under load and when not loaded. The problem is that switching wall-warts cost more than the crappy Indian and Chinese-made transformer-based ones, so gadget makers don't usually bother to include them.
Switching to DC isn't a magic bullet to solve energy efficiency problems, despite this idiotic (and just plain wrong) article.