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

Optimal quench tests -- round 2]




-- BEGIN included message

Bert, Richard, All,

I did some tests and measurements today using a transistor to replace a spark
gap in a TC.  These tests were performed at a very low power and no sparks
were actually produced.  The transistor was controlled by an adjustable pulse
generator. The purpose of the tests was to verify the theoretical benefits of
optimal "quench" times, and their effects on spectral splitting or beating.

In the first test the center frequency was 240 kHz and the optimal quench
time for monochromatic maximum amplitude output was 4.2 uS.  Using a longer
quench time of 12 to 20 or more uS, the center frequency was reduced to about
the same amplitude as the sidebands which were beginning to form.  As the
quench time was lengthened, the center frequency evenually disappeared and
the sidebands increased in strength.  The sidebands were measured to be ~95
kHz apart, with the lower at ~205 kHz and the upper at about ~300 kHz.  I
used the amount of sideband splitting to "backfigure" the coupling at k =
.36, using the TC Tutor program.  I didn't measure the coupling in the more
traditional ways.   I made an attempt to measure the Q of the primary by
first removing the secondary, and letting the primary ring down to the 10%
amplitude level.  This took 36 RF cycles resulting in a Q value of 49
 (1.364*N = Q).  However, my set up was plagued by noise in the scope display
which may have invalidated the Q measurements.  I installed the secondary and
measured its Q in the same fashion, results were the same (Q = 49). 

For the second test, the coupling was reduced and measurements performed
again.
The center frequency was 227 kHz, spectral split was 55 kHz, lower freq.
~205, upper freq. ~ 260 kHz, again the degree of splitting was used to
backfigure the coupling resulting in k = .22.  Optimal (max. amplitude
output) quench STILL occurred at ~4 uS, and was monochromatic.   This did not
agree with the projected ~9 uS which results from the formula 1/2delta f.
 These results suggest that maximum power transfer occurs with only one cycle
of RF in the primary since any longer quench time reduced the output
amplitude.   However, I do not fully trust the results because there was a
lot of distortion in the primary waveform which may be affecting the results.
 Also the cut-off action of the transistor switch may be less than ideal.
 Also, no sparks are produced in this system, which may change things.  At 12
uS quench, sidebands could be seen.  Many of these phenomena are difficult to
see precisely using my equipment.  The "next best" quench time after 4 uS was
around 27 uS which agrees with the above formula (splitting is now occuring).


These results show only a tantalizing glimpse into the world of quenching and
frequency beating effects, clearly more work is needed.  I also have to
verify the above measurements.  Yet, certain observations stand out; 1)
sufficiently fast quench times completely eliminate spectral splitting
(except of course for the minor bandspreading that is inherent to damped
waves).  2) as the quench time lengthens, the center frequency weakens and
finally disappears from view.   Again, this system is unloaded and produces
no sparks.  The "rules of the game" for a sparking TC may differ greatly.  A
lot depends on how much of the total system energy is released into spark
production BEFORE the quench.  

Perhaps Richard Hull's upcoming hydrogen thyratron experiments can answer
these questions in a spark-producing system.

Happy and quenchy coiling,

John Freau

-- END included message