[Prev][Next][Index][Thread]
Re: Optimal Quenching Tests
Tesla List wrote:
>
> Subscriber: FutureT-at-aol-dot-com Sat Jan 4 21:22:31 1997
> Date: Sat, 4 Jan 1997 02:20:41 -0500
> From: FutureT-at-aol-dot-com
> To: tesla-at-pupman-dot-com
> Subject: Re: Optimal Quenching Tests
>
<SNIP>
John, Richard, and all,
John, we must have been reading from the same scripts - our posts met
each other in passing! :^)
>
> Richard, All,
>
> I've been reviewing some of my notes:
>
> 1) I incorrectly posted that my 550 kHz , k = .09, TC quenched optimally
> (for best spark), at 4 uS, quench for best spark was actually at about 8 uS.
> It was a k = .22 test that gave best spark at quench time of around 4 uS.
Plugging in estimated values (derated for gap losses):
T = 0.85*(1/(2*550000*.22) = 3.5 uSec ~ 4 uSec observed
T = 0.80*(1/(2*550000*.09) = 8.1 uSec ~ 8 uSec observed
It _does_ hang together now!
>
> 2) There seems to be differing definitions of the theory of optimal
> quenching: It is my understanding, based on the work of the Corums, that
> the optimal time to quench is at the first beat frequency notch, which
> results from the split frequencies created by over-coupling. This notch does
> not occur until a certain amount of beating takes place, the exact amount is
> determined by the degree of coupling. I did the calculations, for the above
> (k = .09) TC, and the optimal quench comes out to be very close to the figure
> of 8 uS: The Corums' theory states that (for the low loss case), the
> optimal spark duration is approximately equal to: 1, divided by twice the
> difference between the high and low spectral components. In the above TC,
> there's a spectral split of about 50 kHz. 50 kHz times 2 = 100,000, and 1
> divided by 100,000 = 10 uS, losses will tend to shorten the required quench
> time, thus there is good agreement with the 8 uS quench time I obtained. I
> see no conflict between my (or your) results, and the Corum's theory of
> optimum quench times.
The really wierd thing, is that IF you quench at, or before, this point,
the frequency splitting in BOTH the primary and secondary goes away. Ed
Phillips and Richard Hull mentioned some time back that the grizzled
veterans of spark radio knew all about this phenomenon, and used it in
high power systems! The Corum's only rediscovered it - they probably
hadn't yet been born when this approach was being used! :^)
>
> 3) It seems to me that as the RF energy in the primary plunges downward
> toward the first beat freq. notch, the energy concurrently builds in the
> secondary. When the energy in the primary is at a minimum, and has been
> transfered to the secondary, which is now at a maximum, that is the time to
> quench. If the spark is quenched after only 1 RF cycle, at this low value of
> k, the energy will not yet be fully transfered to the secondary and spark
> output will be poor, as you found in your thyratron experiments.
Quenching after 1 RF cycle would be right-on for k=0.6 though! This
could be the best strategy for quenching the primary circuit of a
tightly-coupled driver on a maggie system...
> 4) Quenching too soon is not something that any coiler builder using a spark
> gap has to worry about. It is almost impossible to quench even at the firs
> beat frequency notch, (8 uS in the above case), Richard's thryratron system
> is probably one of the few quenching methods that is capable of quenching
> BEFORE the first beat-freq-notch.
>
But, if you design a rotary gap that has much _less_ mechanical "dwell"
than the optimum (i.e., like John's "0 dwell" offset-electrode
approach), and have a fast enough electrode seperation rate, mother
nature will help you quench it at the right time.
> 5) My results seem to indicate that spark output falls off rapidly as quench
> time moves outwards to the 2 nd or 3 rd beat freq. notch. The 2 nd notch
> corresponds to approximately the 17 th RF cycle of ringdown. The 3 rd beat
> freq. notch corresponds to approximately the 28 the RF cycle of ringdown, in
> above TC example. Most TCs quench somewhere between the 3 rd, and the 6 th,
> beat freq. notch--and that's with good quenching. Who knows where poor
> quenching TCs quench?
>
It does! Most secondary energy that gets transferred back to the primary
is lost in maintaining the plasma in the gaps. Quenching performance
also has a lot has to do with how much energy you're throwing off the
top in streamers/ground arcs (the good kind of loss!). You can actually
have a fairly poorly quenching gap and still pound _lots_ of energy into
the secondary during that first energy transfer cycle. Once you develop
heavy streamers, the secondary Q drops sharply, since it now appears to
be very "lossy". Energy expended in streamers will not make it back to
the primary for the next trip around. This has the effect of reducing
the energy available to reignite the gap. Since gap reignition is
prevented earlier, this leads to an "effective" improvement in quenching
performance. BTW, if you get an arc to ground during the first transfer,
_none_ of the secondary energy ever makes it back to the primary, and
you _will_ quench at the first notch.
This may be why a seemingly poorly-quenching gap, such as Cox's
Milwaukee Museum coil with the beefed-up, but reliable, rotary
electrodes, can still belt out 10 footers! Even on my lousy
static/vacuum gap system, once the top-end really starts cookin',
"effective" quenching performance improves, and I'll hit 1-2 primary
notches consistently at 360-420 BPS.
> 6) When I quenched at 8 us in the above example, I saw no notches in the
> secondary waveform; I saw only a build-up, followed by a nice high amplitude
> ring-down. But yes, I agree, more work needs to be done to verify the
> optimal quenching conditions needed for optimal spark output.
>
This makes sense - if we quench once the energy flow from
primary-secondary has completed, but before it can reverse, then we'll
never see a secondary notch! A real poser is why the spectral splitting
also seems to go away...
> Happy coiling!
>
> John Freau
And safe coiling to you!
-- Bert --