 I am trying to get 208V from a 208V transformer. I was under the
 assumption that a 75KVA 600  120/208V transformer would deliver 75Kva
 at 208V. With 600.00V input they are delivering more like 40  50 Kva
 at 200V.

 I have 2 of the same transformer that are doing this with similar
 loads. I took the cover off them and took some current readings. One
 has approx 150A on the highest leg and the other transformer has 180A.
 Both transformers are running unbalanced because there is a large
 single phase load on each 3 phase panel. There is a lot of heat being
 generated too.

 The loads are basically ballasts for high output UV lamps for curing
 inks on screen printing press's, with a few small motors for fans
 and conveyors etc. I do not know if these ballasts are electronic or
 not. I intend to find out tomorrow.
This could be the problem. If the load is very nonlinear, it can be
causing a substantial pulse load on the transformer if it is also a
major part of that load.
A normally resistive load is ideal. But when a load draws all of its
current only in a short pulse during each cycle, then it is causing
more of a demand on the transformer and wiring than what it would be
if averaged out.
A simple explanation works like this. Suppose you have 3 banks of
heaters that pull 90 amps of power each. You switch one bank on for
1 second at a time, and cycle through each so that only one bank is
on at one time. You get the same amount of heat as if you had just
one bank on steady. But now, suppose you turn all 3 banks on at
the same time for 1 second, then everything off for 2 seconds and
repeat. You get the same heat again, but this time you are drawing
270 amps 1/3 of the time. Consider that impedance loss affects the
power in proportion to the
_square_ of the current. At 3 times the
current you have 9 times the loss. Average that 9 times loss over
the 3 seconds in each cycle, and the average loss is 3 times as much.
When you have electronic loads that pull current in short pulses on
each AC cycle, it's still the same kind of problem. The math is more
complicated, but it puts more stress on the wiring and the transformer
windings, and results in more loss, more heat, and potentially even
damage or fire.
This problem is even worse with 3 phase power when using loads that
connect between any hot phase and neutral. That puts current pulses
on the neutral that do not balance out. You can end up with three
times the current on the neutral ... and that's just average ... the
pulsed current problem can make it worse if the pulses are narrow
enough.
Since your loads are connected phase to phase, the neutral overload
shouldn't be much of an issue. But you can still cause a 15% extra
overload just due to the pulses not matching between phases (they
don't average together like sine waves would). But you still could
have the I^2*R problem with deep pulses.
If you can get an oscilloscope reading of the current waveform on
each of the phases and the neutral, that might be more informative.


 Phil Howard KA9WGN  http://linuxhomepage.com/ http://ham.org/ 
Click to see the full signature.