Re: Homegrown servo motor driver?

If you want an impressive target, Vicor is routinely using 1MHz PWM to
drive it's high end DC-DC converters in the 95 - 99% efficiency range.
This translates into MOSFETs switching times around 25ns.

About 15 years ago, I designed and built a circuit to switch 400v in
20ns (it was an extraction pulse driver in a mass spectrometer). At
the time, I learned that you can make even relatively crappy mosfets
switch very quickly if you can move charge in and out of the gate fast
enough. To do that I stole a play from the CMOS book and used a
matched set of P and N mosfets at the gate of the primary switching
transistor.

You could put such a gate driver at each of the four gates in an
H-bridge and be able to craft basically an arbitrary waveform at the
output. With a bit different configuration, you could even make the
output a constant current source with an arbitrary waveform rather
than a CV output, making your control loop easier to work out.

Also I don't see any reason this drive circuit couldn't be used for
servos as well as micro-stepping a stepper with controlled torque.
Then, to get creative, you could pre-compensate the drive waveform to
the stepper to account for winding inductance.

Emissions would be a real pain, though unless you used a dog-cage
around the whole thing.

On 6/9/06, codeSuidae <codesuidae@...> wrote:
>
> I'm also interested in how one should go
> about choosing a PWM frequency and the benefits/trade offs of higher and
> lower frequencies and how motor characteristics affect the decision.