[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Terry's DRSSTC actually hooked to a coil now >:-))



Original poster: Steve Ward <steve.ward@xxxxxxxxx>

Hi Terry,


> > >by following primary current, you just > > >"ride along" with the frequency... > > > >Yup, this is exactly what happens. But you have to be careful because there > >are two possible "rides" the coil can go on. If you design the system to > >give the right voltages and currents on one of these paths, they may be all > >wrong if it switches to the other one. > > I guess I am not understanding what the "two paths" are here. The two coil > system has primary and secondary resonances which beat against each > other. One small and the next larger "notches" due to this beating can be > seen here:

The 2 "rides" are the upper and lower pole of the frequency split that
occures with all TCs.  The idea as i understand it is that you want to
tune the primary to be "suggestive" of what frequency the system
should lock onto.  If you find the lower pole more desireable for some
reason, it helps to tune the primary extra low so that you minimize
the chance of it changing to the upper pole.  I think the problem with
tuning in the middle is that it might flip one way or the other,
perhaps unpredictably, and i think the problem with that is that its
not really optimized for either frequency of operation.  It may also
change poles during the burst, which might lead to some problems (this
just gave me an idea of something i need to test).  So its best to
pick a pole and stick with it i think.

Better let Steve C. come and clean up my partial understanding here ;-)

>
> http://drsstc.com/~terrell/pictures/FullSystem-01.gif
>
> I should have let it go longer and you could see lots of beats.  Between
> the two "notches" you can see where the current (yellow) shifts some.  The
> frequency certainly is not a stable value but bouncing around.  But these
> are perfectly normal for a primary current driven system.  The secondary
> current driven system like Dan's is much different in that the beating is
> in the drive signals where I think it "beats up" the IGBTs more since the
> zero current crossing miss in that case.

Well, possibly.  Its hard to say what is better yet until someone
makes primary feedback work properly with no explosions ;-).  Ive
tried primary feedback on several occasions, but each time found some
small quirk that scared me off.  Most recently i found that using
primary FB and making sparks to ground causes something in my
controller freak out, turning the coil ON for many mS!!  This never
happens with the secondary feedback... i need to investigate what the
cause of this was, might just be noise ;-).  But the thing that always
scared me about primary feedback is that if you dont have a limiter
(well, your IGBTs are always the limiter), the current will ring up
without hesitation during a heavy streamer to ground (i watched the
current nearly double in this case!).

  The primary current is a powerful
> fourth order system but it seems very "stable" as it beat between the two
> frequencies.  If there is other some weird mode, "I" am not aware of
> it.  Perhaps those that have studied the system's transforms have found
> something there?

I think you should be able to tune the primary system to stay at just
one frequency, but i can not be certain of this.  I dont have
measuring equipment capable of accurately detecting if the primary
current was jumping from F1 to F2 like you do ;-).

>
>
> >Primary current feedback can't select between the two paths, it will just go
> >to the one whose gain is highest.
>
> Yes. Is that possibly bad?
>
> >If they are about equal, then streamer
> >loading may tip the balance and make it switch mid-burst. You can prevent
> >that by deliberately mistuning- to ensure lower pole operation, tune the
> >primary lower than the secondary, and for upper pole, tune it higher.
>
> Right now primary and secondary are tuned to 85.1kHz and tuned to within
> 0.1kHz of each other. It is trivial to mistune, but I just don't see
> "why"? If the frequency want's to shift, the driver does not care. I have
> no local oscillators or anything controlling the H-bridge. It is totally
> timed from the primary current. It could go from 1Hz to 100kHz and then to
> 200hZ and the driver would not care (1Hz would probably drain the filter
> caps or hit current limits). Perhaps in the case of an oscillator
> controller like your PLL controller, such shifts are far more of an issue?


I think you just want to generally avoid this because it seems
unstable.  I would also like to hear exactly why this is a "bad"
thing.  But i think its been shown that DRSSTCs produce the best
sparks when operating at one of the pole's and staying there.

>
> If the secondary took a heavy hit, it's oscillations could certainly be
> shifted out of sync with the primary or totally disrupted.

Hehehe... my nightmare.  Those CM300s sure didnt react well to these
extra chaotic situations!  Problem with ground strikes and my
particular setup was the secondary jumping to a harmonic (3X Fres
roughly) and getting a mess of "bad" switching transitions fed into my
driver... lots of hard switching and at 3X the normal operating
frequency... just terrible really.

  But i would
> think once that load is gone it would all snap right back in place.

Indeed it does nicely, except now you have a nice ionized path
allowing for maybe 10-15 more bursts being delivered once again into
that heavy strike, same craziness all over again and again.

  But
> the primary would not care much...

Well, the primary might ring up to some higher than normal currents,
but hopefully your over-current detector catches that.

I can imagine if the primary were
> driven from a local oscillator, such disruptions could do odd things in
> that six pole feedback system case.

Hmm.  I keep having this continuing debate in my head over what is
more reliable... secondary FB, primary FB, some sort of PLL, or just a
plain old VCO.  Jimmy's first coil ran beautifully from a VCO and had
no problems with ground strikes as far as i know.  He also pushed his
IGBT bricks way beyond what i ever managed.  He used the 150A 600V
bricks and said he got them up to something like 1700A in a test.  Im
using the 300A 1200V bricks and have had them pop at maybe 1200-1400A,
but i think i was hard switching (read above) and they latched :(.

>
>
> >If you reverse the phase of the primary feedback you get a third possibility
> >which is the one described by Antonio de Queiroz and Jimmy Hynes. The system
> >oscillates at the "zero" midway between the two "poles". This oscillation is
> >not stable in the steady state but it might persist long enough to get
> >breakout.
>
> It will just shut off. The logic will not allow the IGBTs to turn on out
> of phase. If you reverse the CT, the driver will just counter the start
> pulse current and massively damp the oscillation off. In the case of a
> secondary current driven coil, you can reverse the CT however.
>
>
> >It's hard to tell from the trace which mode you're on but I would guess the
> >lower pole. The upper pole never shows notches, and the "Antonio" mode won't
> >oscillate at all if the feedback phasing is set for driving a single
> >resonant circuit.
>
> I "think" it is riding both poles and the resulting primary current
> regardless of frequency or poles and zeros. The driver just does not care
> what the frequency is or how much it shifts from cycle to cycle.


Dunno what to think of that yet.

When the theory gets too hard for me to understand, thats when i go
build something ;-).  Still lots of experimentation to be done with
the DRSSTC stuff!

Steve W.


> Cheers, > > Terry > > >etc > > > > > > > >Steve C. > >