21 Oct
2007
21 Oct
'07
11:09 a.m.
On Friday 19 October 2007 20:04, Guy Sotomayor wrote:
Except that after going through a 3-phase bridge
rectifier, you have
360Hz (with a good DC component) vs 120Hz that drops to 0v so you
need a lot less filter caps with the 3-phase. And frankly the
situation I'm dealing with draws ~500A @12V - that's a butt load of
current over single phase (and that's just the stuff that *requires*
3-phase...there's a bunch more power that splits the phases to get
standard 120v single phase). Total load requirements are 70A across
all 3-phases (210A single phase...how the *hell* are you going to
wire that?).
Ok, well, I was specifically talking about JP's Onyx 2. It has been my
experience with more modern machines (like my IBM SPs and S/390s, and
like JP's Onyx2), that the power supplies are generally over-designed
enough that they won't really care if they're running off single phase
or three phase power. Either they're designed already to run off
either single or three phase (like the Onyx 2 I was referring to, and
the IBM SPs I've got), or they have an already over-designed amount of
power-soothing capacitors (which are going to be designed to work at
over the max possible loading of the computer, which it should
basically never reach, at least after the inital inrush current).
Also, good design dictates that the power supply should be designed to
be able to run for a cycle or two with lower or no input voltage, first
so that you can get a "useful" power fail interrupt, and second, so
that you can ride out short brownout or power blips from inrush
current, say from a disk drive or A/C compressor spinning up. :) I'll
further argue that if the equipment can tolerate a missing cycle of
A/C, then it should have no problem dealing with the difference between
rectified single and three phase power. Of course, not all machines
are designed "well", and some things won't appreciate the difference,
but I'd imagine that the problems will mostly be limited to much older
equipment, with linear, instead of switching, power supplies, and which
is fairly big in size (and might warrant 200A 208V 3ph service or even
480V 3ph service:).
I wasn't exactly trying to say that Big Computers like your PDP-10
should be run off single phase instead of 3-phase power*, it's
possible, but probably doesn't make sense. It probably makes more
sense to just buy a 3ph diesel (or LP/natural gas) generator to run it
if you're not going to do it frequently, than to convert the machine to
3 phase power. Or, just get 3phase power, of course. The cost of
installation will quickly get outweighed by the cost of running the
machine for any significant period of time. :)
* Though, honestly, you aren't 100% correct in your calculation; most
people would use 240V single phase instead of 208V three phase... 240V
1ph delivers 2/3 of the power of a 208V 3ph circuit of the same
amperage... so your 70A 208V feed is equivalent to a ~100A 240V 1ph
feed. I actually do have a 100A 240V "branch circuit" run in
my "computing facility", though it's to run a machine with more
mad-scientist appeal than computer-scientist appeal. :)
Pat
--
Purdue University ITAP/RCAC --- http://www.rcac.purdue.edu/
The Computer Refuge --- http://computer-refuge.org