Re: Recent s.s.t.c work

["Tesla list" <tesla@xxxxxxxxxx>]

Original poster: "K. C. Herrick" <kchdlh@xxxxxxx>

No, there can be no shoot-thru there because the drivers are "totem 
pole" emitter followers.  Both of the transistors in a driver cannot 
be on at the same time.

Here's another nice thing about the shift-register scheme:  You can 
just choose a fixed tap on the register and vary the clock frequency 
to vary the delay.  I'm going to try for a 32-stage register with a 
clock of around 5 MHz. Varying the 5 MHz will change the delay proportionally.

My simulation will not accomodate a s-r longer than 24 stages.  32 
stages in the hardware, requiring four 74HC165Ns, will yield a delay 
of ~190 ns per stage, for a total available delay of ~6 us with a 5.3 
MHz clock.

Another thing I think likely:  Most of the delay in the loop would 
seem to be due to the turn-on times of the IGBTs, which will tend to 
be constant at a given h.v. level and independent of operating 
frequency.  Thus once set, the s-r's delay should remain suitable for 
somewhat differing Fr's.

Something else I might mention:  This scheme works only because the 
feedback loop includes the "pilot oscillator"; absent that, the 
feedback will never start.

And one last observation:  Recall from the several supply-current 
waveforms I've posted that, with phase-shift present, the supply 
current has both + and - components.  The ideal condition is to have 
no + component; the supply capacitors only discharging during the 
spark events.  Thus I think it may be feasible to incorporate a 
visual indicator in the supply circuit, e.g. a LED, that will show 
the amount of undesirable current present.  One could then merely 
tweak the clock frequency to minimize that indication.

Comments, anyone else...?

KCH

Tesla list wrote:
> Original poster: Steve Ward 
> <mailto:steve.ward@xxxxxxxxx><steve.ward@xxxxxxxxx>
>
>
>
> On 10/19/05, Tesla list <<mailto:tesla@xxxxxxxxxx>tesla@xxxxxxxxxx> wrote:
> Original poster: "K. C. Herrick" <<mailto:kchdlh@xxxxxxx>kchdlh@xxxxxxx>
>
> Nope, no shoot-thru in this instance because no H.V. is applied. I
> didn't apply H.V. because I'd need 4 probes to implement 2
> differential connections and I only have 3.
>
> KCH
>
>
> Hi Ken,
>
> I meant shoot through in the gate driver half-bridge that drives the 
> gate of each main IGBT.  No HV would have to present to have 
> shoot-through there, just the 60VDC present on the drivers.
>
> Anyway, the shift register is an interesting idea, and i will be 
> curious to see how it effects stability.  I would suspect that it 
> might be just fine working in that manner because 1) you should be 
> running for just a few cycles, and 2) the frequency of the primary 
> *shouldnt* shift a whole lot during those few cycles.  So chances 
> are the feedback will be slightly off, but im guessing that its 
> gonna be much closer to the zero cross than what you have now 
> ;-).  The nice thing about that shift register is that its 
> independent of your operating frequency (assuming i understand how 
> the shifter works).  Hope that all works out well, good luck!  Make 
> sure to allow for more than enough shift.
>
> Steve