Re: Recent s.s.t.c work

["Tesla list" <tesla@xxxxxxxxxx>]

Original poster: "Sean Taylor" <sstaylor@xxxxxxxx>

On Sat, 15 Oct 2005 22:43:18 -0500, Tesla list <tesla@xxxxxxxxxx> wrote:

> Original poster: Steve Conner <steve@xxxxxxxxxxxx>
>
>
>
> > If anyone has some ideas as to how to maintain zero-current switching
> > in a simple feedback-system, I'd like to hear them!
>
> I explored a whole bunch of different ideas, but I kept coming back to
> the PLL. It doesn't give perfect zero current switching under all
> conditions but it gets close most of the time.

I have thought about a couple different systems, similar to a PLL but on a
microcontroller.  The idea is this:

Sample the primary current (or secondary base current, whatever your
favorite source of feedback may be, thought primary current seems to work
the best) at about 40 to 50 times the expected resonant frequency.

At each sample (except the first) find the first derivative and calculate
an expected zero crossing time (assuming a linear waveform).  Continuously
re-calculate this time with subsequent samples, and make a switching
decision based on this - ie if the zero crossing is expected to happen
before the next sample, switch at that time (or perhaps start increasing
your sample rate).  Just in case you have made an error in the expected
zero-crossing time, simultaneously check for a zero crossing by looking
for a sign change in the current.

As with the PLL scheme, one could tell ahead of time and switch earlier
for on/off time lags in the large IGBTs.  This could be accomplished with
a fairly common microcontroller with a built in ADC.  One would just have
to shield the thing pretty well, and perhaps make a simple LPF for the
incoming current.  The LPF could either have a cutoff frequency high
enough to not affect the fundamental hardly at all (or be really steep),
or you could find the frequency of the waveform in the microcontroller and
comensate for the phase response at the given frequency :P

Sean Taylor
Urbana, IL