For any of these solutions (auto-transformer, RC networks, etc) it would be nice to see some sort of before-and-after performance measure. For example, oscilloscope pictures of distorted waveforms before the solution is applied, and cleaned up waveforms after the solution is applied, if you can provide any. This would give us incentive to spend time and $$$ on that solution, or not.
Regarding the autotransformer, Variac etc make some units, seem to be optimized for power line conditioning, also capable of adjusting the output voltage.
Without going too deep into theory, the MSW smoothing effect seems to be the low-pass inductor/winding characteristic, which here may be able to filter our strong MSW harmonics and noise beyond 60 hz.
The fact that there is no isolation between input and output doesn't seem to matter, because I don't have any isolation to begin with.
A failure in the winding insulation should only have impact on the inverter, and would not endanger the load because I am using the auto-transformer as slight step-up and not step-down.
Also good benefit, if I have several taps in the secondary (or continuously variable secondary), then I can use it to regulate upward the RMS output voltage to the desired value with no-load, or make it load-dependent.
One thing I don't yet fully understand, there may be a phase difference between the voltage and the current at the auto-transformer output that powers the load (just like in any inductor with some resistance in the winding). This degraded power factor would cause reactive power to develop in the auto-transformer winding, which would be sent back to the inverter and not to the load. How much power loss would there be, and what this would do to the inverter, I don't know.
Seems like we are dealing here with a case where the power source itself, and not the load, is causing the reactive power component. Perhaps some capacitive power factor correction would be needed at the auto-transformer output to take care of this, and make this solution usable
Any comments?
Regarding the RC networks, I assume this would be R=5 ohm and C=0.7 uF elements connected in series. Then I am not exactly sure, two RC series circuits would be connected from each of the hot and neutral outputs of the MSW inverter to ground, or one RC series circuit connected between the hot and the neutral?
Which way is it?