document updated 18 years ago, on Apr 5, 2006

Purely by experimentation, one 9v battery has noticable effects. Two 9v would likely be more than enough, but experimentation will have to confirm this. The very convenient result of this is that the voltage is easy to measure and verify with a standard multimeter.
Output voltage can be reduced either with a PIC-controlled DAC, or with a purely human-controlled potentiometer.
Output shape can be controlled either via PWM, or with a code-controlled output pin. If the PIC runs at 20MHz, and a max 500Hz output wave is needed, then the PIC has 20,000 instructions it can run between each state switch (eg. lo-to-hi or hi-to-lo), or clearly more than enough.
Multiple isolated output channels can be created by first charging a capacitor from the battery, then disconnecting it completely, and then connecting it to the output. (note that this can be potentially combined with polarity-reversing and voltage-doubling). Also, two isolated channels can be created if both have a 50% duty cycle, just be sure that when one is off, the other is on, and visa versa, so that neither is on at the same time (this reduces the need for extra DACs as well). If a 25% duty cycle would do just as well, then of course 4 channels could be had without capacitors.
Note that reversing the polarity on a channel is very useful... it makes both contacts recieve the stimulation (I believe only the one attached to the negative side gets the real jolt).

If we want a ton of channels, each with very short duty-cycles, a Mux would be a natural choice. Two muxes could be used (one positive, one ground) to allow for reversible polarities. An 8-channel mux would provide two upper-body channels and two lower-body channels.
If it's hard to find a mux with a 0-18v range, it would be possible to generate -9v, 0v, and 9v directly from the batteries (eg. no complicated voltage-inversion required) and drive one mux off of 9v and one off of -9v. This would probably require at least one extra DAC.

On the other hand, this bad boy puts out 100volts, so I might want to look at something closer to that range?? (also note that they manage to do voltage isolation with very small capacitors, so that if there's a short on the output side (or anything near a short through tissue), there will be minimal problems)
If I'm interested in current-limiters, here are a few starter pointers: one, two, three