Rich> Some here seem to think that we are amateurs--if not Rich> idiots--working on a home system, and unaware of the Rich> precautions necessary before making changes to the power system Rich> of any large computer. Perhaps "some" are merely unaware of the background of the person(s) asking the questions. Yes, it's somewhat annoying to an expert to get an answer that assumes he's a novice. But it's potentially much nastier for a novice to get back an answer that assumes the knowledge of an expert. Personally, I tend to write as if the audience is not all that experienced (unless I specifically know otherwise, and even then there will be others reading the archives later) -- in the hopes that experts will simply skip over the "obvious" stuff without taking offense. Yes, an expert knows how to rewire electronics that uses 3-phase power so it uses several 1-phase feeds instead. A novice might overlook important details, such as the fact that the neutral in a 3-phase Y system is smaller than the phase wires, which means you have to substitute larger wires there for this rewire job. And finally, if you have a system that actually has a 3-phase input to the DC supply (for example, 6 diodes from 3 center-tapped windings feeding the ripple filter), then rewiring that input as 3 separate single phase circuits is not going to work quite right, because you will then have inadequate filtering (having replaced 360 Hz ripple by 120 Hz ripple). It doesn't sound like this shows up in VAXen; I don't know if it does in the KL-10; but I do know it is a consideration if you're working on a CDC 6000 series machine because it has such supplies. paul

Paul Koning wrote: Isn't that how the VAX 6000 is set up? I run a VAX 6000 on split-phase and it works just fine. Peace... Sridhar

On 20 Jan 2009 at 15:53, Paul Koning wrote: Yup, but there the ripple frequency is 2400 Hz (at least on the CPU supplies). I used to show movies at a drive-in theater where the arcs were fed configuration. Unlike the arcs that were fed from MG sets, these had a very noticeable 360 Hz whine when operating. Cheers, Chuck

Rich Alderson wrote: Just where do you eat Lunch? Ben.

While I can not say anything about the KL10s transformer, it should be noted that some three phase transformers can work on a single phase. It depends on how the windings are laid. Some of the guys that run old broadcast transmitters on the ham bands use this trick. -- Will

On Tue, 2009-01-20 at 13:06 -0800, Guy Sotomayor wrote: Since extremely high current switchers are relatively cheap (at least compared to getting three-phase installed, and massive aircon systems), is there any reason why you *wouldn't* go down this route? Particularly for a machine that's going to be on a lot? Gordon

On Jan 21, 2009, at 8:40 AM, Paul Koning wrote: Well you still need a lot of each. And finding -5.2v and -2v high current regulators isn't *that* easy. The 12v DC supply in the KL-10 supplies 500A. I think each -5.2v regulator supplies 30A and there's 5-7 of them (if memory serves...I haven't looked at the power supplies in my KL for a while). There's a similar number of -2v regulators but slightly less current. One of the big problems in dealing with this is the in-rush current which is *really* impressive. Go look at the KL10 installation manual on the specs. It's a bit scary to have an in-rush current that high (for up to 1-2 seconds). I don't know if those converters are set up for those kinds of in-rush current. My understanding was they were set up more for inductive loads (ie motors) but never having dealt with them, I'm working from ignorance. I just recall a friend who worked at KSC during Apollo where NASA had something like a 2-4 ton flywheel to be power the critical launch systems in the event of a power failure (to give time for the backup generators to kick in). They'd launched several Apollo missions when they decided to test a power failure during launch scenario (simulated). His description was that there was the sound of a bomb going off as the lights went out. Analysis afterwards indicated that the flywheel stopped in about 1/4 of a revolution! TTFN - Guy

You can say that again. A few years ago at work we had a vacuum motor which was plugged into a *normal* UK 3 pin 240v socket. When switching it on we often saw blue or green sparks from the socket and eventually got an electrician in to advise on various health & safety issues. Said vacuum pump is now connected to the mains via a special box (sorry don't know the name) that looks just like the start/stop box you get on tooling machinery (industrial drills, lathes etc.), with the green button for on and red for off. Sorry for going OT. Regards, Andrew B aliensrcooluk at yahoo.co.uk --- On Wed, 21/1/09, Paul Koning <Paul_Koning at dell.com> wrote: From: Paul Koning <Paul_Koning at dell.com> Subject: Re: Running 3 phase 780s on single phase power To: cctalk at classiccmp.org Date: Wednesday, 21 January, 2009, 5:08 PM Motors are the original "high starting current" devices. That's why protection devices for motors are different from conventional circuit breakers. A "locked rotor" motor draws a LOT more current than a spinning one -- some vague memory says it might be 10x or so. And the initial surge as the motor is accelerating looks like the locked rotor case, briefly. How briefly depends on the machinery that the motor is driving; if it's a high load it could take some time before the motor is spun up to speed. paul

It should not be a problem - the wiring in your building can certainly take a short overload like that. If you are concerned, you can put a motor starter in the line that drops a low value resistance in for a tiny amount of time just at startup. -- Will

Guy Sotomayor wrote: I saw one of these when my father took me to his workplace, Paxman Diesels in Colchester, in the 1970s. He explained that it was for an airport, to keep the landing lights on during a power failure. The flywheel was turned by an electric motor, and stored enough energy to keep the lights on just long enough to start the diesel engine. He called it a "no-break set", i.e. a generating set with no break in the supply. -- John Honniball coredump at gifford.co.uk

On 21 Jan 2009 at 22:00, John Honniball wrote: I think that an extreme example that I can think of was the magnet power supply on the NIMROD synchrotron in the UK during the 1960s. Huge flywheel spun up from mains power and then suddently discharged into an alternator as the beam came 'round. I tried to google images of it, but came up empty. Cheers, Chuck

For the record, the IBM 1401 CHM recently received ran off a rotary three phase converter for most of its life. -- Will

That's the basic design, though there are better ones that are somewhat more involved. Again, that depends on the design. I've read several long articles on the subject in various on-line and paper publications. At least one of them came with a lot of detailed testing including scope traces showing waveforms and balance. Both were excellent. I don't have the article at my fingertips; it probably was in Home Shop Machinist. The 70% number you quote seems low. Perhaps you are thinking of "static converters" which are not much more than a capacitor or two -- the machine's motor is connected as you describe above. Or maybe it's applicable to the most basic rotary converters, but not the more carefully tuned kind with multiple capacitors. In any case, the waveform for 3 phase power supplies isn't all that critical. Harmonic content is no big deal, the rectifiers cause a whole lot as they draw current only on the voltage peaks. The main concern would be that there are no high voltage spikes, but I don't see that being a risk with rotary converters. paul

Err, historical accuracy? The PSU is as much a part of the system design as the CPU (OK, maybe I exagerate, but not by very much). I know that if I owned a large adn rare machine (large VAX, PDP10, etc), I'd want to keep it as original as possile. -tony

For a living computer museum, certain tradeoffs *must* be made. Do you continue to spin 30 year old disk drives constantly, or do you build an emulator that looks to the CPU's I/O channels like the original disk? Do you upgrade power supplies with new aluminum electrolytic capacitors, or replace them altogether? If you replace them, do you do so in a way that is completely reversible, to maintain historicity? It is intended for the systems in this collection to run, but not to destroy their historical value in doing so. So we have alternate wiring harnesses in KL-10s, we have new caps in PDP-8s and PDP-11s, we have a Massbus disk emulator that talks to RH20s and RH11s and soon to RH780s, and we'll continue to develop ways to keep things running as we maintain their history. Rich Alderson Server Engineer, PDPplanet Project Vulcan, Inc. 505 5th Avenue S, Suite 900 Seattle, WA 98104 mailto:RichA at vulcan.com (206) 342-2239 (206) 465-2916 cell

Or, it's the difference between pragmatism and historical preservation. We try to preserve the system as closely to its history as possible, while recognizing that sometimes compromises are necessary. Sometimes, those compromises even reflect history: for instance, the replacement of linear supplies with switching supplies in PDP-10s, which seems to have become common practice as the model line matured. Our goal is always to have a system someone can _use_, as we work to develop a 'best practice' that addresses that while maintaining the historical integrity of the original machine. -- Ian

On Tuesday 20 January 2009, Rich Alderson wrote: I'll assume that I'm one of those people, and I'm sorry if I came off that way, I definitely do not feel that you and your group (PDPplanet) are idiots, I was trying to explain things for the benefit of other readers (also, Ian King made no mention of what group this was with, and seemed to me that he was missing the particular point I was making in my replies, which is my excuse for belaboring the points I was trying to make :). Pat -- Purdue University Research Computing --- http://www.rcac.purdue.edu/ The Computer Refuge --- http://computer-refuge.org