TSSP: List Archives

From: Terry Fritz
Date: Sat, 07 Oct 2000 18:50:47 -0600
Subject: Re: [TSSP] Progress report 28th Sep 2000

Hi Paul,

At 02:01 PM 10/7/2000 +0100, you wrote:
>Terrell W. Fritz wrote:
>
>> I would like to submit my paper at:
>> 
>> http://users.better.org/tfritz/site/papers/modact/modact.html
>> 
>> for perusal...
>
>Good stuff Terry, a few comments...
>
>On the equivalent circuit:
>
>You've not taken account of the proximity effect on Rs,
>so the 69 ohms should be more like 150 ohms.

That was the DC resistance plus a remarkably poor guess at the added AC
resistance I used when I wrote that paper.  I now use 391 ohms in that
model (remembering that my secondary seems to have poor Q).  The RMS
secondary current in this lumped model is about 300mA.  Thus, the loss is
35 watts.

>
>On primary waveforms:
>
>I guess the large current spike at the start of the primary current
>waveform is just circuit parasitic capacitance getting charged up
>very quickly. If so that pays a compliment to the bandwidth of your
>current pickup.

The probes are 40MHz bandwidth and those giant spikes are very real.  It is
believed that they are do to the capacitances of the gap suddenly
discharging.  Placing ferrite beads on the gap electrodes substantially
reduces the noise.  The zero crossing spikes are caused by the gap loosing
conduction as the direction of current flow reverses.  The gap actually
goes out briefly during these times.

>
>I'm puzzled by the apparent non-coincidence of the current and
>voltage notches.

I think they do coincide but there is a little noise that may confuse the
waveform a bit.

>
>DC offset to the primary ring voltage - the gap is rectifying?

Possibly.  The gap is symmetrical but the initial spike may cause the
conduction to be offset for a time.

>
>On secondary waveforms:
>
>The notches do coincide on this one. What's with the DC offset
>appearing in the secondary top voltage trace?  Probe artifact?
>I see no way for the secondary to sustain DC for so long.

That offset is very real.  Apparently, negative charges arc far more easily
than positive charges so the net DC effect is seen.  Richard Hull also
observed this effect.  My ignition coil TC's actually arc further in one
polarity than the other due to this effect.  The DC charge is only there
for a few hundred uS.

>
>On the V/I to the secondary arc:
>
>Looks like the I is very slightly leading the V, so there's
>some voltage dependent energy storage going on - some
>justification for a capacitive equivalent circuit. Terry,
>do you have data files (csv format?) containing these arc V/I
>curves as arrays of numbers? If so we can work out the effective
>R and C of the arc fairly accurately. If you repeat the run
>with different variac settings you'll soon get some evidence for
>arc Z as a function of length. Perhaps you've already done this?

Greg Leyh and I compared notes since we both had real time voltage and
current data (with good phse information) for the energy going into the
arcs of our coils.  My coil and his giant electrum both had an arc
impedance of 200,000 ohms in series with 1pF/foot of arc length.  A
surprising result considering the gross difference in size and power of the
two coils.  Apparently, people who study arcs are not very surprised by
this.  The numbers are a bit rough like +-50% but that impedance always
seems to give very good results.  In various test I have run in the two
years since this paper was written,  The 200K + 1pF/foot estimate of
streamer impedance has always held true.  The high bandwidth of the probes
are designed to preserve phase information just for this purpose.  Greg's
coil is very low frequency so his lab grade current taps were able to
provide good phase data too.

>
>Terry, I'm well impressed with your high current pickups - I'm
>assuming the current spikes are genuine and not spurious pickup.
>If so I'd like to consider using them for permanent instrumentation
>in a CW driver. Can you point me to some details?

The "plans" for those probes are at:

http://users.better.org/tfritz/Project_color.pdf

>
>Terry, I think I could do some number crunching on your trace
>data. They should reveal the effective average Z of the secondary
>arc, and, after a bit of shuffling, the dynamic V/I curve of the
>primary arc. Just confirm that upper and lower traces are always
>simultaneous sweeps and are not on alternate trigger.

The other papers at:

http://users.better.org/tfritz/site/papers/papers.html

May have such data.  However, the probes were not as sophisticated back
then so I would beware that data for critical analysis.  I was happy just
to see ANY waveforms back then :-))  I will have to search my records to
see if any of these raw data files still exist...  I can tell by the graphs
that they were digital imported files from the scope so the tabular data
was there at one time...

>
>Some arithmetic:
>
>Primary tank volts at trigger, 14kV, across 17nF gives
>a bang energy 1.67 Joules.
>
>Secondary peak top voltage 260kV across (16.7+10.3) pF
>gives 0.91 Joules.
>
>Therefore overall efficiency of energy transfer primary tank to
>secondary tank is 54%. Seems reasonable.

I have posted a few detailed studies of where all the power is going in
such systems.  This is close to our system of interest.

				Watts 	Percent 
Power into coil 		840 	100 
Power in 50 foot cable 		15.88 	1.89 
Power in neon primary 	30.47 	3.63 
Power in neon secondary 	38.2 	4.55 
Power in NST filter 		96.4 	11.48 
Power in gap 			156 	18.57 
Power in primary cap 		9.354 	1.11 (7.5C temp rise) 
Power in primary coil 		76.08 	9.06 
Power in secondary coil 	71.84 	8.55 
Power in self capacitance 	14.95 	1.78 
Power in terminal capacitance 	8.49 	1.01 
Power to arc 			322.0 	38.33 
TOTAL 			839.66 	99.96 

These models are improved over time as are the coils, so the numbers are a
bit different...

>
>Terry wrote:
>> ...and state that gap losses often account for ~50% of the loss
>> in Tesla coils.
>
>If you can probe for the V/I across the primary gap, and bottom any
>DC offset issues in the V probe, we can integrate the product of the
>two traces to yield the total loss in the gap, and then compare that
>with the total loss as calculated above. You would predict
>50% of (1.67 - 0.91) = 0.38 Joules. Also, calculating the correlation
>function of the two traces would lead to the gap V/I phase angle.
>
>(PS, How quickly does the 500V probe recover from a severe over-
> voltage? Quick enough for the above?)

If there is a powerful spike.  The fiber optical transmitters will saturate
and the recovery time is long.  The modern probes resist this effect very
well and I can always rescale as needed.  Some of the old tests in the
papers were discovered to have been affected by LED saturation.  Those
papers are all about two years old now and the equipment was in it's
infancy.  I need to go revisit all of them but time....

I have done such studies and the models seemed very close (+-30%).  My
modern probes systems are far far better and could give far more accurate
results today.  I have not repeated those tests since I had no reason. 

> 
>I suspect Malcolm may already have such a procedure in mind for his
>experiments.
>
>> These models seem very accurate but in this case are
>> "lumped parameter".
>
>I don't think thats a problem, providing we are aware of it and take
>due account when necessary (eg I avoided using the base current to work
>out (above) the stored secondary energy, we must use the top V instead).

MicroSim is more than capable of doing transmission line models too.  I
have tried this but the lumped models gave good results without the
dramatic calculation times needed for lossy coupled lines.  Many of the
small harmonic features can be reproduced with T-line models however.

>
>> I apologize for not testing my secondary outside yet to find the
>> free area Q but rain, snow, hail, and it is... yes, indeed... now
>> sleeting outside right now :-P   have prevent this experiment...
>
>No worries, its been raining here too, most days. Means I can't get my
>coil out to play either. I've plenty to do indoors - I've a big HF
>coupling transformer to wind if I get really bored!
>
>> My "200000 ohms plus 1pF/foot" value for a steamer impedance is
>> very rough but I think it is the best estimate out there at the
>> moment.
>
>Well we've got to start somewhere! Doesn't seem at all unreasonable
>for a first approximation.

Yes.  I am hoping the work of others, like Marco, and the many people who
have reproduced my probes will also provide info on other systems to pin
the numbers down better.

I hope this is all not straying too much from our topic here, but I wanted
to point these methods if they can be of help.

Cheers,

	Terry

>
>Cheers,
>--
>Paul Nicholson,
>Manchester, UK.
>--



Maintainer Paul Nicholson, paul@abelian.demon.co.uk.