The proposed circuit was breadboarded and tested by CM on his ST5. He did a great job and I really enjoyed working with him.
Bench Testing:
CM did his bread board bench test with a fixed 120VAC line voltage applied, and a light bulb attached to the field coil connections.
The regulated voltage trim pot (R2) is set so that the bulb barely glows. This does adequately test the entire circuit, and upon hookup to the generator, only a minor adjustment will be required.
Testing on CM's ST-5:
CM found the circuit worked fine as a harmonic regulator, <1 volt variation with a range of loads. For this configuration the harmonic winding is connected to the D3 bridge diode instead of L1, N. This configuration has merit for those who plan on loading their ST head very heavily, especially with poor PF reactive loads. Voltage will fall off with extreme loads, but the field coil can't be overloaded. (Thanks Bill Rogers for the test results and discussion of this issue via email.) The down side is less flicker reduction, poorer AC waveform (the "harmonic hump"). If you are using the last 30% of our rated ST output (even if just by poor PF loads), you should use this configuration with the basic regulator as shown. However, you should change to a 600volt IGBT transistor and a good sized heatsink, as it does get hotter using this configuration, since the switching time isn't synch'd to the the low voltage time, and harmonic peak voltages are higher.
With the circuit using line voltage (L1, N) as excitation source, the unit will have better flicker reduction, tighter regulation, and a much nicer looking AC waveform. The slow blow fuse, F, should be carefully chosen and tested to protect the field windings in case of a protracted overload.
An interesting oscillation:
We had to reduce the capacitance of the AC to DC filter greatly to eliminate an LCR oscillation of his ST5 regulator-field coil which caused a very noticeable flicker. I'm guessing this is pretty much the same situation that Jeff M saw on his original (non digital) regulator/ST-5. To detune the oscillator, we reduced the capacitance at C8, C7. Ultimately, less proved best for lowest flicker, and instead of the original 200uF we are now at 10uF. Regulation tightness did not suffer. With this change, flicker is now noticeably better than on the stock harmonic setup, instead of worse. The schematic has been updated for this change. Stay tuned for some measurements.
Other Changes:
A hysteresis resistor shown in the original schematic as 1M has been corrected to 10M. CM caught that in the bench test stage.
The optional small pcb mount AC relay to cut out the start button as AC voltage rises is now not optional. It's too easy to get an over voltage situation without it. R7 is now not critical, 20 to 50 ohms, 10 watt, is a good choice. The schematic shows 50 ohms, 10 watts. (Edit- both the AC relay and resistor can be eliminated for harmonic regulation.)
Here's the updated schematic in pdf:
http://www.driveway.com/mdqde12615Here's the udated Digikey parts list (.txt)
http://www.driveway.com/likde03565Here's the schematic in the .sch format of the free schematic editor of Expresspcb.com:
http://www.driveway.com/mfhdw58912CM may design a PCB, too! Hat's off to him for his great work!
Bruce