[Home][2014 Index] Re: [TCML] First notch quenching? [Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [TCML] First notch quenching?


Again, that is very interesting information.

> > >
> > > Also agree, so its not quite the same as a spark gap clearing an
> arc.
> > >  Basically the energy will go to both the arc and back to the DC
> bus
> > > simultaneously.  You can control an H-bridge to "clamp" its output
> > > (to nearly 0V, really 2 junction drops) which will keep it from
> > > recharging the DC bus. This would be like a spark gap that doesn't
> > > quench, but is also very, very, low loss.
> >
> > I have previously given some thought to this.  Presumably, the way to
> > 'clamp' the output at the end of a burst would be to change the phase
> > of the drive to one side of a full-bridge so that the IGBTs are still
> > driven (and hence there is no need to try to run dc though GDTs), but
> > no additional energy is injected into the primary tank circuit.
> >
> > Does anyone do this?  Does it work 'better' or 'worse' than simply
> > ceasing to drive the IGBTs?
> >
> I do use this technique in my latest DRSSTC drivers (not published
> anywhere yet).  I use it mostly for current limiting, so that instead
> of terminating the gate drive upon current trip, i simply let the
> primary "coast" by leaving off 1 half-bridge (the diodes commutate).
> This allows the primary to ring with only the damping the secondary,
> and not regen to the DC bus.
>  I very much like the results of having this added control of the
> bridge.

I think I understand the diode commutating point, but don't you still need
to reverse the phase of the driven half-bridge?  Further, wouldn't you also
be better off continuing to drive the other half-bridge, to avoid leaving a
'high impedance' gate on the low-side of that half-bridge with possible
attendant Miller effect issues? (I feel quite out of my depth with that last

Turning back to the point of the original post (so this is not a complete
hijack), do I understand the (at least theoretical) approach you would
recommend for a DRSSTC then to be: (i) As a goal (but subject to component
limitations), to time the ending of each burst to correspond with the 'first
notch'?  (ii) To use your 'coasting' technique to implement current
limiting, but not end-of-burst?  (iii) To implement end-of-burst by not
driving the IGBTs?  (Why would that be better than 'coasting'?)


Tesla mailing list