I know this is a really old thread, but I have discovered a problem with the Valhalla circuit implementation. It also affects other power supplies based on the Valhalla design, such as the Hercules II.
The problem is poor starting ability due to clipping of the power supply output waveform for the first few seconds after every start button press. This is caused by an inappropriate time constant in the coupling of the last filter stage to the amplifier driver stage.
The PCAS 022 circuit uses a resistor ladder to bias the U2 driver stage, with a feed resistor value of effectively ~220k (R23). The coupling capacitor from the previous stage is 22uF (C15). This gives a time constant of 4.8 seconds. The problem arrises because the DC level from the preceding stage varies when the start button is pressed, and the DC bias point for the output stage takes ~5 seconds to settle down. During this time the output waveform may clip and oscillate.
The fix is to change R23 to 10k and C15 to 1uF, giving a time constant of 0.01 seconds for the DC conditions to stabilise after switch on. That way the motor has full, undistorted voltage right from the button press and produces full torque immediately. Check that the output voltage is still in specification after doing this modification and adjust if necessary.
