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

Re: Optimal Quenching

Subject: 
            Re: Optimal Quenching
       Date: 
            Sat, 15 Mar 1997 21:55:48 -0800
       From: 
            Bert Hickman <bert.hickman-at-aquila-dot-com>
Organization: 
            Stoneridge Engineering
         To: 
            Tesla List <tesla-at-pupman-dot-com>
 References: 
            1


John and Malcolm,

Sorry about the delayed resonse to your earlier post - been a little
busy lately. Interesting that the upper and lower sidebands are still
relatively asymmetric about the center frequency even in the later set
of measurements.

             Fc  Delta
            ---  ------- 
   Fl: 205  240  35  kHz
   Fu: 260  240  20  kHz
                              
Anyway, the optimal dwelltime seems to be in the right ballpark now. I
also concur with Malcolm's interpretation of why we get the sidebands
after the ideal quenchtime. I'm also interested in the type of switching
circuit used to make a variable "gap" - a properly biased MOSFET pair
sounds like it would be close to an ideal low power experimental tool. 

Safe coilin' to you!

-- Bert H --      


Tesla List wrote:
> 
> Subject:
>         Re: Optimal Quenching
>   Date:
>         Fri, 14 Mar 1997 14:34:15 -0500 (EST)
>   From:
>         FutureT-at-aol-dot-com
>     To:
>         tesla-at-pupman-dot-com
> 
> Malcolm, All,
> 
> I agree that maximum energy should be transfered at ~1/2kF, and this is
> why I
> didn't trust my original results.  I did some more work last night on my
> transistor "gap" system, and my original concerns were well grounded,
> indeed
> various flaws were uncovered. The tuning of the system was incorrect in
> the k
> = .22 arrangement, and this helped to throw off the results.  Also, the
> "biasing" of the bipolar transistor was incorrect, causing weird voltage
> spikes in the primary.  Under these conditions, the generator may have
> been
> driving the coil directly, to some degree, instead of allowing the
> primary to
> ring at its natural frequency.   Another problem was that too much of
> the
> primary energy was bypassing the (turned off) transistor and ringing
> down
> through the power supply -- I had to add a larger choke to stop this,
> but
> there's still some leakage.  After making these changes, the system
> behaved a
> lot better, but I still don't trust it too much.  But at least now, I
> got the
> best secondary output using a ~9 uS quench which is much more in line
> with
> expectations.  Center frequency is 240 kHz, and side frequencies (which
> show
> up at longer quench times) are as reported previously.
> 
> Malcolm, thanks for your comments, it sounds as if your experiments were
> more
> sophisticated than mine.  I'd be interested to hear a description of the
> test
> set-up you used for your experiments.  Some time ago, I had a discussion
> with
> Bert H. regarding just when splitting occurs.  His PSPICE models showed
> NO
> splitting before ~1/2kF.  Are you saying that splitting occurs at or
> beyond
> 1/2 cycle, but that it does not become "troublesome" (noticeable, or
> possible
> racing sparks, etc.) until after ~1/2kF?  Is it a matter of degree?  Is
> PSPICE correct?  I'd be most interested to hear your views.
> 
> Towards optimal coiling,
> 
> John Freau
> 
> << Hi all,
> >          Having done experiments with a MOSFET "gap" system last
> > year, I'd like to make a few comments on John Freau's post and ask a
> > few questions.
> >     First, why do sidebands occur (and increase as quench time is
> > increased)? If you look at a quarter cycle of ring, it has a well
> > defined amplitude, right? No change in amplitude = no sideband
> > production. If you look at a half cycle, you now see a change in
> > amplitude. Spectrally decomposing this there are now several
> > frequencies present, the centre frequency and two sidebands of rather
> > low amplitude. As the time over which you examine the
> > amplitude-changing-waveform increases, the sideband amplitudes
> > increase at the expense of centre frequency amplitude until you
> > decompose a complete beat envelope at which point, the centre
> > frequency is entirely suppressed and the sideband amplitude hits a
> > maximum. This spectrum is not intuitively obvious when you look at
> > the waveform in the time domain. What you see there is the centre
> > frequency whose amplitude is changing.
> 
> >     I did exactly what John did and increased the dwell time
> > progessively. What I saw was the secondary amplitude increase in
> > proportion to the dwell until the point (1/2kF approx) at which no
> > energy remained in the primary to be transferred. At no stage did I
> > see the secondary reach that amplitude when cutting the dwell time
> > shorter. I didn't expect to because loose coupling kept a high
> > proportion of the primary energy more closely coupled to the primary
> > than the secondary. On cutting dwell time to that of a quarter cycle,
> > I got enormous spikes across the parasitic primary capacitances as
> > the energy still coupled to the primary coil dumped itself into those
> > strays when the cutoff gap disconnected Cp from the primary coil. It
> >never coupled into the secondary for the most part. Cutting
> > conduction time off at half a cycle left a reversal on the
> > cap as I described in the SCR post.
> 
> >     Question: Did you (John) manage to get all primary energy into
> > the secondary in just 1/4 cycle of oscillation despite the loose
> > I think I'll repeat those expts over the Easter break when I'll
> > have time.
> 
> > Malcolm
>   >>