Using a 240:32 transformer configured for 272:240, fed on the 272 side with 277 volts nominal, should get about 244.5 volts nominal out on the 240 side. That's 244.5 volts L-N.

272 volts is close enough to 277 volts that I'm assuming the transformer should be OK with it.

How often is this kind of thing done?

I'm just pondering the possibilities for running large numbers of computers directly on 240 volts. This is not a current project or such ... just "armchair thinking".

I'd think 244.5 should be OK with the 100-240 volt computer power supplies. The idea with the autotransformer is how to get close to the 240 volt level with minimal transformers (hence the buck-boost autotransformer) in cases where the utility won't provide anything else between 208/120 and 480/277 (which is probably most of them), such as the ideal 416/240.

Using a 240:48 transformer configured for 288:240 would step 277 volts down to 230 volts. But it would be somewhat less optimal than the previous configuration (5.6% more current on conductors and 50% more transformer capacity).

You are better off using an Isolated Transformer for this scenario.

The Auto Transformer derived from Buck/Boost arrangements are better suited for "Non-Critical Loads", like small Motors, Resistance Heating elements, etc.

A "Buck / Boost" Auto Transformer has poor regulation (+/- 20% at best), with an unstable output Voltage.

For a Project like you have described - driving large numbers of Computers at 240V, using an Isolated Transformer of either Single Phase or 3 Phase would be the better choice.

If Single Phase, go with a 480VAC Primary, to 120/240VAC Secondary.
Ground the Secondary's "Center Tap", but do not use it as an active Circuit Conductor (AKA "Neutral"). Just bond the Grounding Electrode System (GES) to the Center Tap, along with Equipment Grounding Conductors (EGC).
Bring an EGC to the first Panelboard, and use only 2 Pole devices.

If 3 Phase, go with a 480VAC Delta Primary to 240VAC Delta Secondary. "Corner Ground" the Secondary side, or "Center Tap Ground" one winding - but not both!!!

To get really tricked out, use a 480VAC Delta Primary, 240VAC Wye Secondary (Ground the common point), with an "Auxiliary" Delta Secondary for "killing" circulating currents.

Just my 2¢

Scott

I think the issue that will bite you, is using 240V 2 pole breakers will fill up you panels quick. Plus can you get breakers with LOW enough current ratings? 240V x 5A = 1200 VA.

Larry C

Are you sure this reputation is not the result of typical uses involving small transformers that have high impedance with widely varying loads? If a small BB transformer is driving a single device that goes on and off, I'd expect to see a significant voltage change because of the significant load change. One thing about a large data center is there is rarely such significant load changes.


This would mean much larger transformers. For example if the data center loads needed 375 kVA in an isolation transformer, a 50 kVA 240:32 transformer would do where its secondary were supplying 1/7.5 of the voltage.

Another option I had thought of was using 120/240 autotransformers to step 208/120 up to 416/240. But this would be larger transformers and larger service conductors.


These are disadvantages I'm trying to work around. I suspect one of the reasons (of many) that data centers still use 120 volts (in USA) to computers is because of the "two hots" problem of 240 volts. One of the goals is to avoid 2 pole OCPDs. This would be a lot of circuits of probably 20 amps. I'm also ready envisioning 3 pole OCPDs providing 20 amps of 416Y/240 to each rack. The densities of some of these data centers can get very high.


Well, obviously you can't ground both

This would present a lopsided load to the utility, since the loads would all be on 2 phases that are only 60 degrees apart. On the size scale where 3 phase is needed, this approach would have to be replicated 3 times at three different phase angles. When all grounds are bonded together, that triple system would effectively be a 6-star 480/416/240 system (a pair of 416Y/240 systems overlaid on each other at 180 degrees).


That would get the 416Y/240 system which is desired. But why not a WYE-WYE transformer (277 to 240 times 3)? If it were the case that computer power supplies can be had to run directly on 277 volts (I've actually seen them, but not in the numbers that would make it practical to plan a whole data center around them), then the circulating currents would be pushed upstream to the utility, assuming their transformer was DELTA-WYE.

Supposedly, switch-mode power supplies these days ... at least the better quality ones ... do not present a significant harmonic issue. That would, of course, still be something to be studied when building a large data center.

If the utility would provide 416Y/240 directly, that seems to be the way to go. I'm sure they would be concerned with the harmonics. If this were a data center in Europe, everything would be powered with 400Y/230 and nothing more would be done (unless harmonics is an issue).

Why would LOW current breakers be needed?

I obviously made an incorrect assumption. I was thinking individual breakers for individual power supplies. My mistake.

Larry C

Even with individual breakers, why would they need to be less then 15 amps? Nothing would need OCPDs any less. You couldn't use smaller than AWG 14 on any of the wiring.

I'm thinking about data centers ranging in size from 24 to 1200 rack cabinets, with 18 to 108 computers per cabinet (and lots and lots of air conditioning). Maybe ONE 240 volt 20 amp circuit would be enough per rack.

I'm assuming the air conditioning would generally be powered at 480.

One idea (if the UPS makers would do it) is for a UPS that takes 480 or 277 volts in on the AC-DC side, and produces 240 volts out on the DC-AC side. This would be a continuous online dual conversion UPS producing a separately derived system just for that one cabinet. If they can fit it in 4U to 6U of rack cabinet space, one of these could be put in each cabinet at the bottom.

They use 120v equipment because that is what they make for the consumer market. When a data center uses 240v it is to avoid having a neutral and the associated noise in the room. They want it all to be line to line. In fact we didn't even bring a neutral into the data center panels before 120v PCs became so common. Everything, including the monitors and small controllers was 240/208. There was a "convenience" transformer, 208/240 to 120v in most of the frames but they didn't ground either leg of the output.
When small switcher supplies came into the market most of those concerns went away and we started seeing 120v stuff all over the place.

The design assumes you will have a branch circuit with O/C protection based on the line plug and they have overload protection in the power supply so you won't need unusual breakers.

Every switching power supply I have seen can do 220-240 (setting the voltage selector) or 100-240 (autoranging). I can believe the desire for less noise in L-L connections, although I have never encountered it with computer equipment. I have encountered it when I worked in TV. So when commodity PCs came around, why the shift from 240 L-L to 120 L-N? Maybe because it was the only L-N choice below 277 and they wanted to do L-N instead of L-L?

240 L-L is what I'm planning to do at home once I find the right surge protection and UPS equipment that is compatible. But for 240 L-N all the European stuff becomes usable.

My guess as to why not many people make 277 compatible equipment is that for UL safety ratings. The spacing and clearance requirements for line voltage carrying components become larger above 265 VAC.

Larry C

Have you ever tried to find a 50kVA unit with a 32V output? Talk about special equipment.

PC were sold into the mass market where the 120v receptacle was the norm.
Noise tolerance got a whole lot better by the end of the 70s too. TTL logic had a lot to do with it and CMOS was even better than that. IBM removed all of the requirements for IG and such in the early 80s. We stopped "zap testing" machines on installation and they stopped worrying about star grounding. The modern switch mode power supplies have such a wide mouth that we don't even care that much about sags and spikes on the input voltage.
A lot of the things people "know" about noise and power problems in computer rooms aren't really urban legends but they are mostly ancient history. In my last 15 years at IBM I have to say, every time we "fixed" something by correcting a power or noise problem, we were back in there finding the real problem soon thereafter.

Right. Although it really isn't much more to go from the spacing needed for 240V (NEMA 6-15) to the spacing needed for 277V (NEMA 7-15). Even SOJ cable is rated for 300V. Switch mode power supplies, including those in lots of small wall warts, are safe as high as 240V and are listed as safe for use in places like Europe.

The big reason power supplies are not made in any quantity for 277V is the lack of market. The lack of market exists because 277V is not a domestic or even customary receptacle voltage. You MIGHT have 277V fluorescent lights in the office, but not 277V convenience outlets. If someone has a machine that needs 277V they call an electrician to run a new circuit and install a special outlet for it (assuming 277V is even available in the building).

But, I'm not trying to make computer run directly on 277V. Even if we could convince manufacturers to design them to work and be safe on 277V, flushing out the old ones already integrated in equipment would be years, if not a couple decades.

They do work on 240V. Since +/- 10% is considered an acceptable range, I'm guessing they should be fine on the 244.5V I'd get from the 240+32:240 buck-boost arrangement derived from 277V. That does limit my upper voltage swing to 7.9%. And it might be easier to get the utility to tap their transformers down to deliver around 272V, than to get them to deliver a 416Y/240 service (easy to do with common pole pigs, but for pad mount that would require a special order for three phase, or 3 separate single phase units).

If it turns out the 244.5V is an issue, one fall-back is to go with a 240+48:240 buck-boost arrangement, which would step 277V down to 231V. That would change the buck-boost capacity ratio from 7.5 down to just 6.

But that's a one time special order in the worst case. Try making special order power supplies that run on 277V for every computer, every network switch, every router, etc.

It's not practical to change out every piece of equipment to run on 277V. If it were, that's the direction I'd go. However, it is practical when building up a data center designed for around 240V L-N circuits to special order transformers needed to step the 277V down to around 240V.

What may be a useful question is weighing the cost of an off the shelf 277V to 240V isolation transformer at full capacity against the cost of a special order buck-boost transformer at the partial capacity buck-boost can work with. That's part of what I am pondering: could the buck boost approach be made practical. Several transformer manufacturers apparently will do a special order like that. The exact size would depend on the scale of the data center involved.

And there is the option to distribute 277V down to a group of rack cabinets and step it down to 240V or 244.5V on the scale that can utilize an off the shelf buck-boost transformer. One 7.5kVA buck boost transformer in 240:32 configuration could support 56.25kVA of load on that one phase (1/7.5 of the load is served by the 32V secondary in series on the 277V side, where the 1/7.5 is from the 240/32 ratio).

Maybe we here in the US should just give up and adopt 400/230V 50Hz like they have in europe. Wires would be smaller, coffee pots would heat up faster, everyone would be happy.

Is it too late to split the difference for the best of both worlds and make the whole world 400/230V 60Hz?

Changing to 50 Hz would be very hard unless there was a 16.7% voltage reduction (although going from 480 to 400 is that percentage). That kind of voltage reduction would mean more current and more wiring losses. Or all those transformers have to be changed out. Keeping 60 Hz would be fine. Keeping the voltages we already have would be fine. Adding to it would be desired.

We already have 240 V in homes. We just need to learn to use it, which includes tweaking NEC 210.6 a bit. We can run fluorescent lighting on the existing 240 V if we use double pole switches. Computers, and an increasing number of other electronics, will run fine on 240 V. There are NEMA-6-15P to IEC power cords available.

Making 400/230 or 416/240 (at 60 Hz) available on request, where 480/277 would be available, and making 240/480 available on request, where 120/240 would be available, would help. Allowing 2 different systems at the same time in electrical service (avoids the extra transformer losses where both 120/208 and a higher voltage at the same time, are desired) would help. They just need to bring it all in on a combined neutral to avoid the dual path issue across ground bonding. This would be a utility tariff issue.

The US will never abandon the good old NEMA 5-15 and 120/60hz, we have too much equipment out there.
Just look at how long it took to get rid of analog TV ... if it ever really goes away and that is a fairly minor amount of equipment, compared to everything else we plug in.

It still might be fun to put up a humorous web site about the coming electrical power transition

There is already a thread on alt.electrical.engineering about the upcoming switch to digital electricity.

That much is easy- all they have to do is slow the generators 20% and you get both a 20% decrease in voltage and frequency, and that 480V 60Hz generator is now putting out 400V 50Hz! Pretty much every 3-phase emergency generator in production can do this, too - drop from 1800rpm to 1500 and that's about it, it's reconfigured from US to Europe. Most feature 12-tap coils that can be configured for 208V 60Hz, 480V 60Hz and 400V 50Hz. Often they can be derated for 208V 50Hz if required.

The transition, especially in residential areas, would be difficult. People love their NEMA plugs. Fortunately, the trend nowdays is that even though the cords have NEMA plugs, the appliances don't really care WHAT they get. I can't tell you how many different countries I've plugged a NEMA plug into with just a plug adapter- 100V 50Hz, 230V 50Hz, it all works. Sure, it all has to be designed this way, since a 60Hz transformer would saturate and burn up and an unregulated device built for 120V getting 240V isn't happy either, but just about everything is built for easy manufacture and export anymore.

And how much power would you get out of those slowed down generators with the corresponding voltage reduction? Well, maybe that would be the way to "go green" and reduce the carbon emissions at those kinds of power plants

And we could lead a slower pace of life with the older synchronous motor clocks running slow.

We can schedule the switch to 50 Hz for February 17, 2017. Then when the new president comes into office, he/she can delay it to June 12, 2017.

It all works out well! The generators have to be derated 20% to compensate for core saturation at 50Hz, but the 20% lower voltage perfectly offsets it. If the coils are designed to handle full rated power at 208V (taps in parallel), they are generally sufficient for full rated power at 400V (taps in series) as well. There may be some derating for some equipment, though.

I don't want to switch to 50Hz, though- that's moving backwards; all our transformers and motors would need derated for saturation and it defeats the purpose of going with a higher voltage for lower costs. What we need is a compromise- 230Y/400V 60Hz for all