There was a German study published partly in Computer Shopper, Jim. 
I could look it up, but hopefully this will be clear enough... 
 
They started to look at fires in buildings which had large numbers 
of computer systems in simultaneous operation.  They noticed some 
increase in electrical fires in main switch boxes and associated 
transformers. 
 
What they found was that there was a harmonic reaction, where two 
or more PFC power supplies would get in sync, trying to modify 
what each perceived as distortion in the incoming AC sin wave as 
it was delivered to the computer load. 
 
Within milliseconds every PFC in the building would pick up on a 
ripple reflected back into the local grid, and the whole setup 
then avalanches into a feedback loop forcing the current/voltage 
peaks to blow the mains in the building. 
 
Your sin wave UPS is not all that smooth when you look at it on
a scope; there is a dozen or so steps or tics on each swing, and 
if the frequency of these gets in sync with the sample cycle of 
the switch mode power supply, resonance sets in, and things get 
very interesting very fast.  The UPS ticks occur around 7200 HZ 
on a couple I have looked at; the SMPS designs have been moving 
from 10K on up to 200k in laptop power supplies. 
 
The effort to reduce weight and expense has power supply, UPS, 
and inverter designs always trying to use cheaper, smaller, and 
lighter inductive windings, and if one increases the sample  
frequency, you can do that.  What you loose, is some slack in the  
system, as the ferrite core and the magnetic flux in a large low  
frequency transformer buffers these spikes and synchronicities. 
 
Before I hooked up a computer or peripheral to a UPS, I'd call 
the manufacturers of them if they will still guarantee function 
with UPS power.  In Germany, neither the UPS manufacturers nor 
the PFC outfits, nor the peripheral vendors were willing to 
admit that there really was a problem, and that the statistical 
correlation between the fires and the power supplies was a fluke. 
They offered no other explanation for the fires, which German 
engineers were able to replicate in the lab. 
 
The square wave inverters are gone from the market; they worked 
on the standard 60 cycles, and at that low frequency, the winding 
on the transformers for outputs over 100 watt was so extensive, 
that it is cheaper to make a 'sin' wave type which really runs on 
7200-10000HZ; the oscillator, MOSFETS, and diodes being cheaper. 
 
Granted that this is not a frequent problem; my concern is with a 
critical mission in which odd risks like these should understood. 
___ 
 * OFFLINE 1.58 
--- Maximus 3.01
---------------