Magnifier coupling measurements, new (fwd)

---------- Forwarded message ----------
Date: Tue, 7 Jul 1998 09:10:48 EDT
From: FutureT-at-aol-dot-com
To: tesla-at-pupman-dot-com
Subject: Magnifier coupling measurements, new


In these tests, I looked at the primary current waveform by using
a current transformer attached to an oscilloscope.  These tests
provide further verification that despite the tight coupling of a 
magnifier driver, the overall coupling of a magnifier is loose, and is 
typically within the range of two-coil Tesla coils.  (In my previous
tests, I looked at the secondary or extra coil waveforms.)

I used the same old set up with pri, 19 turns, # 12 wire, sec, 1.5mH
6.5" dia by 3.4" high, # 24 pvc, driver k = .4, extra coil = 11.4 (13?)mH.
3" by 12" # 28 formvar, tank cap = 0.0015uF, Fo = ~500kHz, toroid is
1.5" by 4.5", synch gap 8 point series quench type, (4 points used),
powered by 10kV, 23ma oil burner ignition transformer.  A variable tune
cap (34pF to 167pF range) is installed across the secondary.

1)  The secondary variable tune cap was adjusted to give the longest
sparks into the air.  This occured at 167pF, and gave about a 4" spark.
The scope display of the primary current waveform showed the maximum
length beats at this tune point.  The first RF notch occurs at 14uS, and
the second occurs 28uS later as expected.  It is interesting that this 
second 28uS notch only occurs in its full duration at low powers with
almost no breakout.  When producing the 4" sparks, only about 14uS
of the second beat is seen, then the quench occurs.  The quench
occurs right in the middle of the beat.  This tune point also gives the 
best quenching which occurs at around 28uS (with 4" streamers). 
The notches are very deep.  I was surprised to see the quench occur
in the middle of a beat, especially with the close gap spacing I use,
but the system must run out of energy at this time.

Figuring the effective or actual coupling based on a 28uS beat, gives
k = 0.07 which is quite a low figure.  But it compares well with Malcolm's
results where his 0.54 driver k fell to k = 0.086 actual overall system k.

As the sec tune cap was varied, various beats occured in the primary,
and it was possible to create beats that were 20uS long, 15uS long,
etc., but the sparks were a little shorter, and quenching was poorer.
Sometimes the beats were very non-distinct, and blurry or not deep. 

It is interesting, that the adjustments that give shorter sparks, produce
numerous, short, bright sparks with a great deal of corona above them,
which almost makes the sparks appear very long, but it's an illusion
formed by the corona.  However, when the sec var tune cap is tuned 
for longest sparks, there are fewer streamers, and almost no corona
above, however more corona and sparks emit from the top turns of
the extra coil.  It would seem that the main tuning is not quite correct,
yet the sparks are longest when set this way.  I may need to adjust
my primary tap slightly, or maybe some other effect is occuring.

Also interesting was the appearance of the secondary (?) waveform.
It did not show normal energy transfers, but rather had a flat
shape (except at the beginning), but I may have been picking 
up some of the extra coil's energy, and this may have given a false
appearance to this waveform?

The fast quench that was possible with this arrangement created
a strong free ringdown after the quench.  The spark length was
maximized when this free ringdown was strongest, it seemed.  The
spark streamers also took on a crackly sound at best tune point,
and the gap and/or transformer became quieter indicating fast
quench and good energy transfer. 

2)  Next, I re-tuned one turn outwards on the primary.  Now is was
impossible to obtain the long primary beats, and the quenching was
worse.  The primary waveform looked a little confused, and the
notches were not deep.   Sparks were a little weaker than in (1)
above, but the corona did not move down to the top coil turns as the
cap was adjusted as it did in (1) above.  It was possible to obtain a
first RF notch at 10uS.  The second beat seemed to be about 15uS
long.  I cannot tell if all these notches represent true power transfers,
but I suppose they do.  If the gap could be quenched at 10uS, this
would suit a k = 0.1 or so scenario.

3)  I tuned outwards one more turn.  TC seemed stronger than in (2)
above, but weaker than at (1).  A first notch could be obtained at 7.5uS.
Best spark occured at 84pF for the var tune cap.  The beats looked
similar to beats that form without the variable tune cap.  This notch
position corresponds to a real coupling of 0.13, and also fits Antonio's
formula Ka = Kd sqrt Ls/(Lr+Ls).  Where Ka is the actual coupling,
Kd is the driver coupling, Ls is the secondary inductance, and Lr is
the inductance of the extra coil.

4)  I removed the var tune cap and re-tuned the pri at 19 turns.  The
sparks were similar to (2) above (kind of weak).  Beat duration was
12.5uS but notches were not deep.  This corresponds to k = 0.15 or so. 

When looking at the primary and secondary (?) waveforms, the
normal expected energy transfers were seen in this case.

None of the above tests showed amplitude variations within the free
(after-quench) ringdown, but some did show a varying DC offset
of the RF.

5)  Next, the synch-rotary was run as a static gap, by adjusting the 
position of the electrodes to a suitable spacing, and then not
energizing the gap motor during operation.  This permited the gap
to fire 6 or so times per AC cycle, instead of only once.  This made
the output sparks longer of course.  Quenching improved a little
due to the greater spark streamer loading, and perhaps also due to
the greater gap spacing during quenching.

The same kind of beats were seen in the primary and secondary
waveforms as before, but the critical sec var tune cap peaking effect
seemed to be mostly lost.  Lou Balint also noted in his work that this 
critical tuning effect disappeared at higher powers.  I suspect that it
is related to ionization conditions around the toroid, and is probably 
a red herring. 

The magnifier was also run again without the sec var tune cap, and
normal energy transfers were seen, as would be seen in a classic
two-coil TC.

The sec var tune cap seems capable of making the coupling looser,
but not tighter than without this capacitor.

Overall, I did not see any benefits to the magnifier design compared
with a standard two-coil TC in these tests.  Actual coupling and 
quenching demands seem the same as for a standard two-coil TC.
It is possible that some faster energy transfers are occuring, that I
was not able to see. 

If this work is valid, it has implications for H2 thyratron switched 
typical magnifiers; the thyratron may need to conduct for 8 RF 
cycles or so, for full energy transfer to occur, even with tightly 
coupled drivers. 

It would be quite interesting to try using the H2 thyratron switcher 
on a tightly coupled (k = 0.6) normal two-coil TC.  In this case, the 
fast switching capability of the thyratron would permit the full primary
energy to be trapped within the secondary within one RF cycle.  A 
dual cross thyratron (or thyratron with diode) setup would be needed 
to handle the reverse currents.  Providing sufficient primary to secondary
insulation may be a challenge.  In any case, a thyratron test on a
(k = 0.6) classic two-coil TC would provide a good cross check of
the energy transfer rates of magnifiers versus classic Tesla coils. 

A relatively tight coupling *can* be achieved in a magnifier, but it
requires extreme construction methods:
To achieve k = 0.6 (actual) in a magnifier, one would need to use
at least k = .85 for the driver, and make the secondary and extra
coil inductances equal or so (contrary to usual magnifier design).
    For example:

secondary = 6mH,  extra coil = 6mH.

Ka = .85 sqrt (6/12) = 0.6 (approx)

Voltage breakdown concerns may be as great as for a k = 0.6
two coil classic TC.

In any case, the benefits of close coupling in any Tesla coil will
probably be relatively minimal, but every little efficiency gain helps.
Tighter coupling may give us maybe 4% longer sparks or so in an
average case.
Comments welcomed,

John Freau