I used the Gboard he speaks of for years and can verify that 0 to 10% ontime was it. Anything above 10% and you are blowing things. I went full processor control but have struggled to build the right drive circuit. I know Matt is on to something with his universal switch.
putting things together Bob talks about germanium components (diode) needed to get the effect he was looking for ...
other interesting statements:
control over frequency, phase and pulselength
“42.8 21,4 10.7 kHz signals in the proper format” > phases locked or not?
For fraction of water it needs 3 phases
matching transformer mass for primary and secondary in the toroid
to find the "proper format" of 3 phases you have to know them or find them out by experiment. boyce used a µP (i assume he used a pic from microchip) to have better control over the phase lock of the 3 phase signals than by using those 555s he used in the 80s.
I used a parallax propeller and windows interface in PGen 2.0 for easiest manipulation of frequency, phase and pulselength for experimenters to mimic Bob´s PWM3H last development board.
that PWM3H board has a µC, voltage regulator, 3 driver ics and 3 germanium diodes and a capacitor ... simple straight forward design ;)
to me there seems to be no need for a bi-directional switch like Matt´s but usage of germanium diode seems to be mandatory.
using 3 phases to drive a toroid modulating dc current flow at femtosecond range means to me that you have to phase shift those 3 frequencies in a way that they add up or cut off each other by exact controlled phase shift.