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

Re: Spark Gap Replacements >Subject: Re: Spark Gap Replacements >> Subject: Spark Gap Replacements



Tesla List wrote:
> 
> Subscriber: FutureT-at-aol-dot-com Fri Jan  3 22:12:34 1997
> Date: Fri, 3 Jan 1997 18:20:19 -0500
> From: FutureT-at-aol-dot-com
> To: tesla-at-pupman-dot-com
> Subject: Re: Spark Gap Replacements
> 
> <<
>  >Subscriber: bert.hickman-at-aquila-dot-com Wed Jan  1 21:42:58 1997
> >Date: Wed, 01 Jan 1997 10:06:26 -0800
> > From: Bert Hickman <bert.hickman-at-aquila-dot-com>
>  >To: tesla-at-pupman-dot-com
>  >Subject: Re: Spark Gap Replacements
> 
>  Tesla List wrote:
>  >
>  >> Subscriber: FutureT-at-aol-dot-com Tue Dec 31 22:56:47 1996
>  >> Date: Tue, 31 Dec 1996 16:02:10 -0500
>  >> From: FutureT-at-aol-dot-com
>  >> To: tesla-at-pupman-dot-com
>  >> Subject: Spark Gap Replacements
>  >
>  >> Bert,  All,
>  >
>  >> Bert, you also mentioned (in your TCBA article) your interest in using
> solid
>  > state switches to replace a spark gap at low voltages.  Lou Balint and I,
> >> (working independently), have done some very preliminary work in this
> area.
>  >  We both used horizontal output transistors to replace a spark gap in the
>  >> tank circuit of a TC.  This was not a typical solid-state TC, --- there
> was
>  > no oscillator circuit operating at the resonant frequency.  The design was
>  >> exactly like a typical spark-gap TC, except that the spark gap was
> replaced
>  > by a transistor.  The capacitor was charged by the power supply, then the
>  >> transistor was turned on using a pulse generator, transistor "on" time
> could
>  > be kept on for various time periods to simulate a quench time.  When the
> >> transistor was turned on, the tank resonated until the energy was depleted
>  > (damped waves).  I was unable to obtain any spark output from the
> secondary
>  >> coil, although I was able to draw off a 1/2" spark using a screwdriver.
>  >  Power input was low, and DC power supply voltage was about 150 volts.
>  Best
> > > spark occured when "quench" occured at the first RF notch.  A small neon
> bulb
>  > placed at the top of the secondary coil lighted most brightly when
> "quenched"
>  >> at the the first RF notch.  Resonant frequency = 500 kHz, Pulse rep. rate
> =
>  > 200 to 1000 PPS.
>  >
>  >> Lou Balint of PA, found the same results as I did, however he was able to
>  > obtain a 1/16" spark from the top of the secondary (spark emitted without
> >> being "drawn off" ).
>  > Unfortunately, at this slightly higher power level, the transistors ( 2)
> > > burned out in a couple of seconds.
>  >
> > > In our work, we did not use back to back reversed transistors, I used one
>  > transistor, Lou used two in parallel.
>  >
>  >> The purpose of this work was to explore the benefits of fast "quenching",
> and
>  > avoid the difficulties of 1st notch quenching in actual spark-gap TCs.
>  >
>   >> John Freau
> 
>  >John and all,
> 
> > Thanks for the update! This is an area I'd like to further explore when
> >time permits, though probably with power MOSFETS or IGBT's instead of
>  >straight bipolars, for the same reason - investigating quenching. In
> >your work, did you see any surprises, or did experiment seem to match
> > theory?
> 
>  >-- Bert,  Richard --
>   >>
> No surprises, but I really need to improve the system before real conclusions
> can be drawn.   Richard Hull's hydrogen thyratron method may be  the "ticket"
> to answering these quenching questions.
> 

I completely agree! Way to go Richard!! Sometime down the road, maybe
he'll provide us with some further info on how he controls this
beastie...  :^)

>  BTW, Bert, did you measure the quench-time of your system when it quenched
> on the 1st notch?   More details such as  k value, frequency, and
>  quench-time, would be appreciated.  I guess what I'm asking is the same
> question you asked me, were there any surprises?  

No surprises, but the best I've ever done is quench (intermittently) at
the END of the first primary->secondary energy transfer. However, when
this DID occur, it occurred at the predicted time, and peak secondary
output always corresponded to this case. I've never been able to quench
BEFORE the end of the first energy transfer from primary to secondary.
Since my "k" has always been in the range of 0.153 to 0.22, this means
I'll always see at least 5 to 7 half-cycles of primary current while
energy transfers between primary to secondary.

> There seems to be a
> question in my mind about what's really happening waveform-wise, during
> pre-first notch times.  My quench time  was about 4 uS at k = .09, at 550
> kHz, or somewhat longer than TC tutor advocates, yet I saw no "wave packets"
> in the secondary RF.  This quench time gave me the longest sparks.
> 

Excellent question!! Lets make sure we're talking precisely about the
same thing. Assume that the gap does not quench, and "fires" at time
t=0, and that the uncoupled resonant frequency of the primary and
secondary is Fr. If we look at the secondary's RF output, we'll see an
envelope of oscillations of frequency Fr starting from 0 and hitting a
maximum at t=T1, going back down to 0 (T2), climbing to another
lower-amplitude maximum (T3), going back to 0 (T4), and so on. If we
looked at the envelope of the primary current oscillations, we'd see it
rise to a maximum 1/4 cycle (of Fr) after t=0, go to a minimum at about
T1, rise to another lower-amplitude maximum at T2, a minimum at T3, and
a still lower maximum at T4, and so on. As you are aware, the so-called
"magic k" values are those which result in time T1 being equal to an
_integral number_ of half-cycles of Fr. If you are looking at the
secondary waveform, the first "notch" would be at t=T2; if you're
looking at the primary current, it would be at t=T1. Lets assume you're
looking at the secondary RF output.

If you quench anytime between 0 and t=T1, you'll only see a single ring
up of energy into the secondary, and a longer-period ring down. This is
the time where energy is being transferred, ONE WAY, from the primary to
the secondary. If you quench sometime between T1 and T3, you'll transfer
energy back, ONE WAY, from secondary to primary, and will be past the
point of ideal quench. Except for when k=0.6, the ideal quenchtime
always be at a point _after_ the first primary current zero-crossing.
The great majority of us have "lousily" quenching gaps - we can't quench
at T1, and are forced to quench at the second, or sometimes even the
third, minimum primary energy points (i.e.,  T3, T5...). Much of the
secondary energy now gets transferred back into the primary and expended
in the gap! By the time we begin to again transfer energy back into the
secondary (between T4 and T5), gap losses have eaten our lunch! 

I now believe the quenchtime design objective should be to hit somewhere
between 80-90% of predicted T1 - quenching a little early is much better
than letting a good portion of it transfer back into the primary! As
Richard Hull has pointed out earlier, this ALSO seems to avoid problems
with actually seeing the effects of frequency splitting. However, you
CAN get too much of a good thing if you cut it much too short,
particularly for lower k values. Unless you take heroic measures, or use
an electronic switch/thyratron, you'll have extreme difficulty quenching
at any earlier zero crossings of primary current (Ip). You wouldn't want
(and probably couldn't) quench at times where Ip >> 0. And, from an
energy transfer standpoint, there are no clear benefits in doing so.

Now, with this in mind, lets look at your example above. With k = 0.09
and Fr = 550 kHz, the approximate "beat" frequency (Fb) would be about
k*Fr or about 49.5 kHz. The corresponding time between "notches" would
be 1/Fb or about 20 uSec, and the theoretical "ideal" quench time (T1)
would be about half of this, or 10 uSec. If you were _actually_
quenching at 4 uSec, then this would have been before the first
primary-secondary energy transfer had gone to completion. This would
explain why you observed no "wave packets" in the secondary RF - you
quenched before T1, so there _was_ no energy transfer back to the
primary! If TC tutor was advocating an even shorter time, then it was in
error!
 
However, this does not explain why a 4 uSec dwell gave you the longest
sparks. Only about 5 half-cycle energy transfers would have ocurred,
when the low k value suggests that at least. Could gap arcing have
increased your "effective" dwell to a longer time?. If this was truly
the effective quenchtime, then  something very interesting is going on!
How did you actually measure the quenchtime? TAlso, this would be about
"on the money" if your k was 0.19...

> Bert, did you try "degrading" your quench to the 3rd notch (down from the
> alternating 1st or 2nd), to gauge the effect on the spark length?

Yes, and beyond. By switching off the air-flow and vacuum to the gaps, I
can "saturate" them as I increase input power - thereby causing them to
quench at later transfer cycles. Secondary streamer length climbs with
increasing power, peaks out, and then begins significantly shrinking to
perhaps no more than half the maximum length. The gaps also "brighten"
significantly from the additional energy being dissipated, and they make
a distinctly "duller" sound. I've always focussed on trying to get from
two energy transfers down to one. Quenching later consistently gives
poorer output performance, both theoretically and practically.
> 
> Happy coiling,
> 
> John  Freau

Safe coiling to you as well, John!

-- Bert --