Power for Power Series

[Robert G. Brown]

Hi, there.  I am a system administrator with a somewhat unusual
(but perhaps not uncommon) problem.  We just obtained an SG-220S
in the big rack mount.  I was somewhat surprised to learn, upon
delivery, that the rack requires 220V SINGLE PHASE power, which is
virtually non-existent in the United States.

We are trying to install it in the physics department, where we have
an abundance of three-phase 240 (which works out, if you connect across
any two "hot" leads, to be around 208V peak with a Pi/6 phase shift).
We might be able to find 240 two-phase (if we go back to the
transformer).  But there just is no 220 single phase (that is, one
220V hot, one current-carrying neutral, and one cold ground) around.

SG is "installing" it for us, but the installation man has gone off to
school (literally) for a week to learn how to install it, and I would
like to at least have power for it when he returns and when he left he
was clueless as to its real power needs.  If anyone out there has:

  a) installed a power series rack on a three phase line;
  b) installed a power series rack on a two phase line;
  c) blown up a power series rack while trying to do either one;

I'd love to hear from you.  A wiring diagram of any working solutions
would also be appreciated (the rack has a three prong plug -- hot,
neutral, cold).

I should note that we did try the three phase power (as a physicist, I
know that the system should NOT be able to see any real difference
between a hot-neutral pair at 208V and a hot-hot pair at 208V with a
phase difference -- unless it uses the current-carrying "neutral" as a
pseudo-ground, which is dumb).  

When we did it, the machine seemed to boot, but the prom monitor did
not come up on port 1 (or 2,3 or 4).  We have no graphics monitor, it
is a server--only configuration at this time.  When we put a scope on
the backplane and on the serial line itself we got very unusual
things.  The serial line showed a .15V, 60Hz triangular waveform on
ALL pins but 1 and 2.  1 had a more attenuated 60Hz signal, and 2 was
(relatively) quiet.  Obviously this is not at all like what we found
on a functioning serial line.

So:  Is our power supply bad?  Is our power bad?  Are any or all of
the boards bad?  Are the serial ports bad?  Or do we need to "flick
that little switch over there"?

      Thanks,
         Rob Brown
          rgb at physics.phy.duke.edu
          Duke University Physics Dept.
          Durham, NC 27706