Icy78,
For the part that you cant believe, and don't understand, go back and read post #2. Jeff spelled it out pretty succinctly.
The chiller specifications calls the voltage deviation very tightly to cover their behind, an ideal condition not always achievable in the field, the tight spec makes it your problem if something should go pear shaped.
You would never be able to hold that, with any rotary converter, with the type of variable load you are running. The only converter that could hold that spec in your situation, would be the Phase Perfect mentioned earlier.
As you see, the voltage deviation is the least of the concern. The current deviation is normally multiples of the voltage differences. This is the major problem, the current is what does the real work. The main thing to watch for is that none of the line currents run at above the marked FLA rating, of each respective motor. If the currents remain at or lower than the FLA, then burnout is not too likely. As Jeff mentioned, with non symmetrical currents, you might get some circulating currents producing heat, but that's not usually a major factor.
Then the next consideration is, do you have enough torque to start and run the load with the reduced voltages and currents. It appears that you do, cause it runs OK. Not as strong as it could be if well balanced, but good enough to get er done. The heat of summer will be the acid test to see how well she hold's up to extreme loading. Having hermetic compressors as the main loads is beneficial, as the refrigerant helps keep the temps of the windings in check.
MTW
