28 Apr
2014
28 Apr
'14
2:12 p.m.
On Apr 28, 2014, at 4:55 PM, Eric Smith <spacewar at gmail.com> wrote:
On Sat, Apr 26, 2014 at 7:39 PM, Mark J. Blair
<nf6x at nf6x.net> wrote:
I thought I remembered seeing that. So, apart from not being 400 Hz, that matches the CDC
mainframe practice. With 3 phase input the ripple frequency is 6x the mains frequency (so
360 Hz in the US). If you feed the 3 inputs from single phase power, you?d get 120 Hz
ripple, and the ripple voltage would be 3x the normal value, which may (or may not) be a
problem.
I would expect either a rotary or a solid state converter to feed those just fine. Well,
a solid state one would if you could get it for that much power. I haven?t seen any
single phase input VFDs that go that high (about 5 HP). Rotary converters go that high
easily, and much beyond.
paul
So, for a computer that normally takes 3-phase power but actually just has
single-phase loads internally (i.e., it's made to take 3-phase power since
that's what's available in industrial/commercial buildings, but nothing in
the machine actually requires 3-phase power), I think that a single-phase
conversion might be preferable to using a rotary phase converter.
The US model KL10 contains many single-phase 120V loads, but the ECL power
supply front end (H760) actually uses three-phase, with three ferroresonant
transformers and a three-phase bridge rectifier to provide 12.6V at 500A.
In principle that could be replaced by some other 12V 500A power supply,
but if you're going to do that I think you'd want to go further and replace
the H761 linear regulators or the entire H761.