[Prev][Next][Index][Thread]
The Ideas to Date
----------
From: terryf-at-verinet-dot-com [SMTP:terryf-at-verinet-dot-com]
Sent: Monday, June 01, 1998 10:19 PM
To: tesla-at-pupman-dot-com
Subject: The Ideas to Date
The following is a thread between myself and David Sharp of TCBOR. The
material is rather heavy reading but we felt it should be shared with everyone.
Terry Fritz
______________________________________________________
terryf-at-verinet-dot-com wrote:
>
> Hi Dave,
>
> At 08:38 PM 5/31/98 -0400, you wrote:
> >Terry
> >
> >OUTSTANDING WORK!!! The lumped parameter model had been
> >suggested as early as 1992 by Duane Byland, and now by
> >your measurements validated the theory!! We reviewed
> >your posting at the monthly TCBOR meeting, it floored
> >a bunch of folks, and made us realize these following
> >ideas we kicked around...
>
> Thanks for the compliment! I was studying the current levels when I
> suddenly noticed the phases where the same. I rechecked and recalibrated
> everything and then realized that something profound was being demonstrated.
> I had always believed the 1/4 wave and standing wave theories, so I knew
> that this was very significant.
>
> >
> >This leads to a hypothesis about "classical" Tesla Coils
> >that would suggest:
> >
> >1. Vo (resonator) is a function of sqrt(Ls/Lp)
>
> Or Vsec=SQRT(Cpri/Csec) :-) Csec seems to be composed of two capacitances.
> The top terminal capacitance and the self capacitance that appears to be
> drawing current bellow the top of the resonator (perhaps I should say
> inductor now!). I think the winding capacitance to ground is acting to
> produce Cself. This is interesting because it would suggest that in order
> to discharge Cself, the arc current would have to pass through a portion of
> the secondary's top windings. Much to work on there........
>
> >2. Vo is a function of resonator base current.
>
> Like Vo = Ibase * 2*pi*Lsec This equation is probably very near to
> reality. Even though the resonating capacitance is split between Ctop and
> Cself, the voltage on each is the same.
>
> >3. Resonator base current is a function of k, Vin (HVAC),
> > and primary surge impedance [Zp = sqrt(Lp/Cp)]
>
> I assume you have seen my paper "A Comparison Study of Multi-Gap and Simple
> Tesla Coil Spark Gaps with Various Primary to Secondary Coupling
> Coefficients" (I have to start using shorter names :-)) This paper shows
> the relationship well. The numbers calculate and match the results well. I
> was very intrigued by the sudden jump in base current for the multi-gap case
> just before the gap started to loose it's ability to quench (k=0.1753).
> There is more to be learned there too......
>
> >4. Spark gap erosion "fiddle factor" is an inverse function
> > tank current; which is a function of Vin (HVAC)/Zp
>
> There is much to be learned still about the gap characteristics and what
> governs the loss of quenching at a certain point. Probably temperature and
> cooling rate but I think the lead parasitic inductances and capacitances may
> contribute to poor quenching as well. I suspect there is some way to play
> with the gap and get much better quenching in general. My other paper
> "Tesla Coil Primary Circuit Behavior Analyzed at High Bandwidth" Shows that
> the arc is stopping at the zero current crossings. It should be possible to
> shift the tuning of the primary at that point. This would prevent the
> primary from absorbing energy back from the secondary due to the tuning
> shift in the primary. I have notice a number of strange things in the
> primary current waveforms. At one point I could get the gap to stop
> conducting for 90 degrees at the third peak (it conducted normally on the
> peaks after that. If whatever was causing that can be exploited it would
> be fantastic. Probably some oscillation or something was responsible. It
> is also interesting just how repeatable all the high frequency
> characteristics of the gap are. This indicates that it is not random but
> rather controlled by yet unknown factors.
>
> >
> >To build a high performance machine (as R Hull's Nemesis system):
> >1. Ls/Lp > 1000
>
> Obviously we want high voltage :-) But the higher the voltage the less
> current we can drive too (given the same input power). Hopefully, I and
> others will be able to determine more about how much energy and what kind of
> impedance the arcs present to the system. Then we can design to optimally
> drive the arc's impedance for maximum power transfer.
>
> >2. Zp should be relatively high (not 20 ohms, more like 40-50 ohms)
>
> The optimal value will depend on all the calculations when the data is all
> in. Perhaps the best value is higher than most people design for now? This
> would seem to be suggested.
>
> >3. Lp should be > 75uH at tune point (100-150 optimum)
>
> Again, It will depend on the real numbers that come out of all this. If one
> takes and sets up a guess work system, there are optimal numbers that come
> up all over the place. However, if you change the base numbers, the optimal
> values shift all over the place. I have not been able to set back and say
> "Eureka! it is all now so clear!" I still look at zillion's of waveforms
> and computer models and still don't see the wonderful over all view of how
> these coils need to be designed. The nature, power, impedance, etc. of the
> output arcs is key to all this. They are what we want to supply power to.
> Without knowing what form of power they need, we can't design the optimal
> system to deliver it. I am hot on this now!
>
> >4. Use as high a voltage (Vin HVAC) to maximize impulse power,
> > while limiting tank current, and attendant gap losses.
> > Richard is currently using 20kV, Alex and I have hardware
> > to go 20, 34.5, 40 and 69kV!!! Gaps at higher voltages will
> > shocking (no pun intended)
>
> Oh yes!! Since the energy in the system is proportional to V^2 ,
> Increasing the primary voltage will do wonders for getting energy into the
> system. It will be very interesting to see how your high voltage systems
work.
>
> >5. Use multiple series gaps (rotary or static) to spread thermal
> > losses and reduce material erosion rates.
>
> Very important the get good quenching and allow high coupling coefficients.
> The above paper suggests that tuning for the optimal k is very very
> important! Multi-gaps seem to be much better at this. Unfortunately, I
> have never really looked into rotary gaps much. I like the multi-gaps
> better because they are easy to make and deal with. They don't throw stuff
> at me either :-)
>
> >6. This allows use of smaller higher voltage capacitors in tank
> > circuit.
>
> Perhaps a good trade off. Remembering that the lower Cpri is the lower Vo
> is too. Ie. Vo = SQRT(Cp/Cs)*Vi .
>
> >7. Fo of system should be less than 50kHz.
>
> Many advantages to running at low frequency. Sparks get a bit more
> dangerous! But at this level, we aren't exactly trying to make our hair
> stand on end with our coils :-)) One very interesting implication of the
> standing wave theory going away is that Q may not be very important!! Let
> me say that again... Q is not very important! We always wanted high Q to
> help the standing waves ring up to a high value. If there are no standing
> waves then Q is not a big deal. I believe is is still very very important
> to use good RF design to reduce parasitic impedances that mess up the
> primary system and I suspect hurt the quenching. See the paper "Tesla Coil
> Primary Circuit Behavior Analyzed at High Bandwidth".
>
> >8. Loaded Fo of system should be < 1/3 resonator Fo with no load
> > (i.e. capacitive load is =>10X Cint of resonator)
>
> Big top capacitance... Probably true. The Cself (or Cint) my need to
> discharge through the top windings of the secondary to power the arc. This
> added impedance my vastly reduce Cself's ability to make nice arcs. This
> may explain much as to why big top terminals can produce lager sparks.
> Still more to learn there.....
>
> >
> >FWIW, Again nice work!!!
> >
> >Regards
> >
> >DAVE SHARPE, TCBOR
> >
> >
>
> I have to ask my self "Why doesn't the secondary have standing waves?" and
> "What about those coils with all the LEDs on them hooked to a signal
> generator. They seem to show standing wave effects?" Well, I have two
> theories there.
>
> 1. A standing wave system needs a point source to produce standing
> waves. Thus, our loosely coupled inductive coils have energy being injected
> into too large of an area to be able to support the point source needed for
> standing waves. However, a signal generator can inject power at the very
> base so the standing waves are produced. This implies that "tube" coils and
> the extra coil in a maggie may have standing waves. However, for energy to
> be transferred those standing waves must not be maximum, thus these systems
> work at some VSWR between 1 and some value. (I should have the range
> covered there :-))
>
> 2. The dimensions of our coils are too small compared with the
> wavelength of Fo to set up standing waves. The signal generator driven
> effect is caused by the distributed capacitance acting with the distributed
> inductance. Not a standing wave, but more of a distributed impedance effect.
>
> If I could determine the base and top current phases of a tube or
> magnifier coil driven solely at the base, it would answer this question. If
> they are out of phase, theory #1 is suggested. If they are in phase theory
> #2 is suggested.
> There is a way to do this without too much equipment if one can
> measure the base current. By looking at the phase of the base current and
> the phase of the top voltage (with an antenna) of a point driven coil, the
> question can be answered. A coil hooked to a signal generator would work
> fine. Of course, I can go do that right now and I will
> :-).................... Well guess what! The base current and top voltage
> are in phase. That means the base current and top current are 90 out of
> phase. That means theory number one wins!!! Right on the bleeding edge of
> science here Dave! Good, I was having a hard time finding a way to explain
> the details of #2 anyway :-) BTW here is an interesting tidbit. If you
> have a source and a load impedance with a full standing wave between them,
> the power delivery to the load is zero! If I had thought that up in the
> right way years ago this all would have been academic now.... Oh well, when
> it all falls into place it all comes together at once!
>
> Thanks very much for your interest and comments. I feel we are making some
> real progress in our understanding of coils. There apparently are many
> things we can do to really make them better. Today we have wonderful tools
> that can see into areas we have never seen before. This knowledge puts us
> on this right track to move forward with a solid understanding of what we
> are doing.
>
> Please keep me up to date on what is going on there. The work of the TCBOR
> is legendary and it is an honor to have gained the attention of your group.
>
> All the best!
>
> Terry Fritz May 31, 1998
>
> BTW - With your permission, I would like to re-post this letter to Chip's
> mail list. Please let me know if that is OK.
>
> BBTW - If you repeat my base current and top voltage test. Be sure to use a
> quality antenna terminated into 50 ohms. Otherwise the scope's and other
> impedances may mess with the true phases. Don't want any standing waves or
> anything :-))
Terry
BTW#1 Post it...Its too important not to!!!
BTW#2 Additional information......HOLD THE PRESSES :^)
1. R. Hull, A. Tajnsek and I have measured the output voltage of
IMPULSE DRIVEN (Spark Gap) coils using electrostatic voltmeters.
Voltage is an inverse square law function to distance from coil
(radially). ES measurement was UNSUCCESSFUL with CW coils of
ANY KIND (VTTC, or SSTC), or BIPOLAR IMPULSE DRIVEN coils
"half-wave length". If a FARADAY MECHANICAL ES Voltmeter is
used, scaling to 1' distance from discharge toroid will give a
good "normalized" output voltage (ie if 10kV ES is measured 4'
from discharge electrode, would normalize to 160 kV [4^2 * 10kV]).
ES voltages follow this relationship out at least 9' from magnifier
checked, and should be capable of much further (SAFER) distances,
is a function of ES voltmeter sensitivity.
2. I've mapped the E-field around small VTTC's and the relative
E-fields
match what is predicted for a 1/4 wavelength antenna. With CW
excitation
this matches and validates your experiment! Small gas filled glass
tubulations were used.
3. To validate my earlier musing, below are models of 3 operating and 1
proposed VTTC systems I've been involved with. The spark lengths
observed are in very good agreement with what is predicted by this
model. More empirical testing is necessary, but as a "first order"
model has excellent agreement:
Assumptions:
a. Streamer length ~ 8.7kV / in
b. Push-pull tank voltage: Tank Vin = (Vin - Tube drop) * 2
NOTE: Vin = Vpeak of input wave [DC, 1.414 Vrms]
c. Vo = Tank Vin * sqrt(Ls/Lp)
d. All L values uH, all V kV, spark length inches
e. Esg of 300W exciter 250VDC -at- 800VDC (proportional
control via VARIAC/automatic feedback current control)
----------------------------------------------------------
Chris Allmond VTTC Laser Exciter (300W)
6-807WA -- 3VT|| x 2 (push-pull)
Lp 215
Ls 20000
SQRT (Ls / Lp) 9.65
DC Vin 0.80
Tube Vdrop 0.30
Tank Vin 1.00
Vo predicted 9.6
Spark Length (predicted) 1.1
Spark Length (observed) 1.2-1.5
----------------------------------------------------------
Dave Sharpe VTTC AC powered (0.5kW)
811A -- x 2 (push-pull)
Lp 244
Ls 30000
SQRT(Ls / Lp) 11.09
Vin 2.55 (1.8kV*1.414)
Tube drop (kV) 0.15
Tank Vin 4.60
Vo predicted 53.2
Spark Length (predicted) 6.1
Spark Length (observed) 6.5-7.0
----------------------------------------------------------
Dave Sharpe VTTC LSPS powered (2.0kW)
572B/T160L -- x 2 (push-pull)
Lp 244
Ls 30000
SQRT(Ls / Lp) 11.09
Vin 6.8 (2.4kV*2.828)
Tube drop (kV) 0.2
Tank Vin 13.2
Vo predicted 146
Spark Length (predicted) 16.8
Spark Length (observed) 17-18
----------------------------------------------------------
Dave Sharpe VTTC LSPS powered (4.0kW input--proposed)
4-400C -- x 2 (push-pull)
Lp 200
Ls 100000
SQRT(Ls / Lp) 22.36
Vin 6.8 (2.4kV*2.828)
Tube drop (kV) 0.7
Tank Vin 12.4
Vo predicted 272
Spark Length (predicted) 31.4
----------------------------------------------------------
This study needs to be repeated with impulse driven coils but
looks very convincing to me, with typical values observed on
three independent VTTC systems. The 4-400C as a side note, John
Freau has suggested that 2 4-400C in properly configured circuit
should be capable of 30" discharges.
Regards
DAVE SHARPE, TCBOR