[Prev][Next][Index][Thread]
Re: Double humpin'
Tesla List wrote:
>
> >From MALCOLM-at-directorate.wnp.ac.nzThu Nov 7 22:40:48 1996
> Date: Fri, 8 Nov 1996 09:42:15 +1200
> From: Malcolm Watts <MALCOLM-at-directorate.wnp.ac.nz>
> To: tesla-at-pupman-dot-com
> Subject: Re: Double humpin'
>
> Bert, Richard,
> A few observations from the scope...
>
> > > I get the impression that some of our folks think that you get a double
> > > hump or spliting of frequencies at tight coupling. Double humping only
> > > occurs in spark systems when we exceed the point known as "critical
> > > coupling". critical coupling has only a little to do with
> > > actual inductive coupling (about 50%)
> > >
> > > If we throw a fixed gap of a fixed dwell/quench in a system, critcal
> > > coupling occurs at some fixed coupling coefficient K=X. If we have a
> > > variable dwell/quench gap, and a fixed tight coupling, by varying the
> > > dwell we can make the system go from below critical coupling to well
> > > beyond. In short, critical coupling is a sliding point based on actual
> > > inductive coupling and dwell/quench time of the gap's realizable
> > > quenching ability. In theory we can have a single frequency output (no
> > > splitting) at k=.65. This was the struggle in the early days of spark
> > > transmitters in the 100KW-.5MW class.
> > >
> > > Richard Hull,TCBOR
> >
> >
> > Richard,
> >
> > Excellent point Richard! Seems like every time I think I'm beginning to
> > understand how these things work, mother nature, or you, slap me upside
> > the head with a dose of reality!
> >
> > A lot of coilers, myself included, have _lousy_ gaps which quench poorly
> > and non-repeatably. With longer-than-desired quench-times, most of us
> > seldom observe the phenomenon you describe! I needed to do a series of
> > PSPICE simulations to confirm this - I have NOT been able to confirm it
> > experimentally, since I don't have my high-speed rotary constructed as
> > yet (a Winter project...).
> >
> > The PSPICE simulation shows that, if the gap dwell-time is reduced to
> > the ideal time or somewhat less (i.e., the end of the first beat IF the
> > gap were to continue firing), the double frequency humps do, indeed,
> > disappear! Even if the primary is set to a somewhat lower frequency, the
> > secondary/toriod will ring up at its single natural frequency ONLY.
> >
> > This seems to imply that frequency splitting and secondary coil
> > flashover _could_ be reduced by consistently quenching quickly enough.
> > Previously I believed that secondary flashover was caoused by the
> > secondary being driven to the upper "hump" frequency, which in turn
> > caused the 1/4 Wave point to be lower on the coil. NOW, I'm no longer
> > sure that this is actually the case! Or maybe splitting only occurs
> > during subsequent beats (at lower energy??), but this doesn't sound
> > right either!! Hmmmm!! What seemed to be so simple before, turns out now
> > to be another layer to peel on the TC onion!
>
> Attached sparks produce practically the same result (ideal dwell
> timewise) as optimally quenching which is why it seems largely
> unnecessary for what we are doing. I guess if the topload ROC and
> capacitance is _just_ big enough, trapping the energy in the
> secondary using optimum quenching would produce a better result.
> That seems to be borne out by Richard"s experience with the huge
> toploads.
> While the amplitude of the oscillations is changing in the
> coupled system before quenching, a split spectra must IMO be produced
> (during ringup). Could you check the spectral output on SPICE Bert?
> I still think the top frequency is exciting the resonator in ways
> we don't want during ringup.
>
> Meanwhile, here are a couple of thoughts to consider on critical
> coupling. In a CW system, it is easy to arrange this by simply
> adjusting k=kc where kc is Q dependent and the primary and secondary
> impedances are well defined and fixed. At critical coupling, primary
> losses=secondary losses.
> How about a disruptive system where the Q's are high before
> breakout and k>>kc? The scope shows the ringdown/ringup is relatively
> lossless compared with the losses when a discharge breaks out. Energy
> transfer is a good deal more efficient than critical coupling. During
> this period and under these conditions the system is definitely
> overcoupled.
> Suppose we now quench at the end of ringup. We cannot say the
> system is critically coupled at this point because to all intents and
> purposes, coupling between the two circuits no longer exists.
> Moreover, secondary losses will be far higher than primary losses in
> toto.
> Suppose a secondary discharge starts while the system is still
> ringing up (scoping suggests this will be the case for most coils,
> particularly those without a very large topload). Under this
> condition, any secondary energy not participating in the discharge
> will, as usual, be reflected back down from the top, but it will now
> be with a phase shift as the top end impedance has changed (zero
> phase shift for the unloaded case). Now the still-coupled primary
> "sees" an impedance different than that which it saw when the
> secondary was unloaded. Under this condition, the primary can no
> longer unload its energy at the same rate as with the unloaded
> secondary case and dissipates more as a result. Here we are
> approaching critical coupling IMO.
> I've written this mainly to clarify my thinking. Any other
> thoughts on this most welcome.
>
> More melting pot stuff,
> Malcolm
Malcolm & All,
The version of PSPICE I'm using (6.3 Evaluation) has a nice feature as
part of the transient analysis - one can also do a Fast Fourier
Transform (FFT) on various voltages and currents. Once I set up the
model for my coil, I could then vary the dwell time leaving k fixed, and
once the transient analysis completed, do an FFT on the results. It was
using the FFT with shrinking dwell times that the lack of "double humps"
became evident. If we let the gap continue to fire, then we see the
humps begin to form, and spread. The longer the gap fires, the more the
humps spread, until they asymtotically reach the predicted Fhi and Flo
values.
In an attempt to better see this, I set up my storage scope with my coil
adjusted so that it didn't get break out, and measured Vout using a
pickup plate to give me a view of the waveform. During ring-up, I was
unable to see any evidence of any higher frequency present on the
secondary/toriod system - the frequency during ring-up stayed at the Fo
of the coil/toroid pair, just like the PSPICE simulation said it would.
However, this is different result from a set of measurements I took
earlier this summer, so I will need to do some further confirmation
measurements. I als have a hard time believing the results, and an even
harder time understanding the "why" of it...
Re: your other comments - for the most part I agree. When the I get
heavy corona, my secondary Q drops considerably (15 - 20), but it still
appears to be higher than the Primary Q (11). However, I'm looking to
make some better measurements under "fully cookin'" conditions...
Safe Coilin' to ya!
-- Bert --