Re: Steve Conner DRSSTC Model question
Steve Conner, Thu Jun 01 2006, 03:03PM

Yup, it's meters. I never did figure out which model was more suited, mine or Terry's 220k+1pF/ft, so if you have anything to add, I'd be glad to hear it

Re: Steve Conner DRSSTC Model question
HV Enthusiast, Thu Jun 01 2006, 06:00PM

Attached are some simulations and simulated data i took regarding this DRSSTC III system i am currently testing. Per the model, the arc lengths seem to be out of whack quite a bit, so voltage is too high on V6 i think. Perhaps, in practice, the geometry of the toroid will limit the maximum voltage rise.

Any suggestions???

Attached Data (PDF)

1. Frequency Response PSPICE Circuit
2. Steve Conner PSPICE Circuit
3. Primary Natural Frequency Plots (Cpri = 0.0825uF)
4. Primary Natural Frequency Plots (Cpri = 0.11uF)
5. Secondary Natural Frequency Plots
6. Primary Current Tuning Plots (Cpri = 0.0825uF)
7. FFT of Primary Current Tuning Plots (Cpri = 0.0825uF)
8. Primary Current Tuning Plots (Cpri = 0.11uF)
9. FFT of Primary Current Tuning Plots (Cpri = 0.11uF)
10. Spark Length Prediction (0.0825uF, 9uH)
11. Spark Length Prediction (0.0825uF, 12uH)
12. Spark Length Prediction (0.11uF, 9uH)
13. Spark Length Prediction (0.11uF, 12uH)
14. Spark Length Prediction Pri Capacitor size vs. Pri Tuning (EXCEL)
] 1149184855_15_FT10609_simulated_data_package4.pdf[ /file]

Re: Steve Conner DRSSTC Model question
Terry Fritz, Fri Jun 02 2006, 05:16AM

Hi,

Way COOL!! about Dan's new coils!!!! How does he get those "blowing up" pictures "timed" just right?? I use video cameras to catch such events (it works good!!), but Dan seems to have the "nice" camera ready to go at just the right split second!! Sound triggered shutter????

I have been working on streamer (leaders really) load models a lot the last two days (OK, "ALL" I have been doing ). For ScanTesla's new dynamic streamer load modeling. Many of the details are here:



But the streamer load is:

""""""""""""""
LengthTemp = 2.8 * sqrt(0.5 * VCsec_max * VCsec_max * (C2 + C3) * BPS); //Secondary energy length factor.

BreakoutLength = 2.8 * sqrt(0.5 * BreakoutVoltage * BreakoutVoltage * C2 * BPS); //Secondary breakout length.

if (LengthTemp < BreakoutLength) {C3 = 0.06666667 * LengthTemp * 1.0e-12;} //Streamer capacitive load.
else {C3 = (0.15 * LengthTemp - 0.08333333 * BreakoutLength ) * 1.0e-12;}
""""""""""""""

Now that is a mess... But basically, small streamers near the terminal have very little added capacitance. The secondary terminal voltage, and its trying to breakout, - dominate the "load". Once the secondary voltage breaks out, the capacitance is a nice linear line as a function of streamer length dictated by Freau's formula with a modified constant. The coil's terminal geometry dictates where the inflection or transition point is as "breakout voltage". It finally DOES predict bigger top loads are better )))

BreakoutLength is due to the fact that you really can't break a 1 inch streamer from a 30 inch sphere ) If you breakout, you have a "minimum" length...

The program's number are pretty close now, but could use some adjusting or at least verification... E-Tesla 6.20 http://hot-streamer.com/TeslaCoils/Programs/E-Tesla620.zip
can do that in like weeks of modeling on a 3GHz computer!! I will get that going now

But it is really "odd" now... There are so many "rules and constraints" that the models seem pretty "obvious"... There are just not a lot of "options" in streamer modeling now...

Streamer ground strikes are super odd!!! It takes "massive" currents to drain the system to match measured scope readings!! ScanTesla cannot handle the currents unless ulta small time steps are used... I am faking it by draing the system directly at the matrix level... The streamer model "fails" for ground strikes... There is "another" capacitance... But its displacment current might explain racing arcs

The future will tell all....

I think Dan, posted enough info on his new coil now... I will see if I can figure anything out...

My SISG coil with its very low primary loss gets rid of most of the "noise"!!! I can "see" things now )) Wonderful things

Cheers,

Terry

Re: Steve Conner DRSSTC Model question
Steve Conner, Fri Jun 02 2006, 09:19AM

OMG, Terry is hacking the system at the matrix level. Any minute now he's going to jump out of my TV and offer me red and blue pills. :P

Dan, I guess the problem is that my model doesn't limit the secondary voltage. It has no conception of breakout from the toroid or flashovers down the secondary. So you can generate impossibly long arcs from a small coil (of course my model has no idea what physical size the coil is, so that's not surprising)

I did experiment with zener diodes of 100kV or 500kV or whatever hooked between the "topload" and "ground" to model this, but they made the simulations take far longer or fail to converge at all, so I gave up. Basically, you just have to ignore any configuration that would give output voltage considerably higher tha the breakdown voltage of the bare toroid.

Re: Steve Conner DRSSTC Model question
HV Enthusiast, Fri Jun 02 2006, 11:34AM

Thanks guys. I might try incorporating your (Terry) new streamer model in PSPICE to see what comes up.

Thanks again, and we'll keep you posted on any new developments!!

Re: Steve Conner DRSSTC Model question
Terry Fritz, Fri Jun 02 2006, 11:09PM

Hi,

Steve said,

I did experiment with zener diodes of 100kV or 500kV or whatever hooked between the "topload" and "ground" to model this, but they made the simulations take far longer or fail to converge at all, so I gave up.

I have some streamer hit pics for the SISG here:



Only "6 inch hits at 108kV". But the terminal (31.1pF) is discharged in 20nS!! That works out to a streamer resistance of 129 ohms and a streamer peak current of "752 amps!!!" And that is just a 6 inch arc!

It is sort of interesting that the ring after the hit's frequency is 1 / (2 x pi x SQRT(Cs x Lp). But that is probably just chance in this case rather than any great truth...

The SISG fires consitantly enough that the scope's dealy function is easily able to catch hits. But the streamer hit times are just too short for MicroSim to deal with it well. There are model setups in the program "somewhere" that might be fiddled with to help.

Dan said,

Thanks guys. I might try incorporating your (Terry) new streamer model in PSPICE to see what comes up.

That sounds like it would make my brain hurt! If you know the arc length, put in 1.5pF/foot + 220k and be done with it. The "dynamic" stuff really does not vary the models that much. It is good for ScanTesla "trying things" since it detects configurations that are not real good and hurt the streamer length. But for a MicroSim model, I don't think you would gain anything.

There is a minimum breakout voltage and a minimum streamer length after breakout. But the streamers in the MicroSim world are basically established and the arc channel is solid, so it does not change during a streamer. You should just be able to set it and forget it.

Cheers,

Terry

Re: Steve Conner DRSSTC Model question
Steve Conner, Sat Jun 03 2006, 12:18PM

Hi Terry, all,

In streamer modelling, I always assumed arcs to ground to be more or less short circuits that discharge the topload instantly. So that agrees with what you measured. I discovered that it is possible to get ground arcs that stay dim and purple, when the coil is striking out just about as far as it can reach, and these don't discharge the topload instantly.

I think what is happening is this: When a streamer strikes ground, the current flow through it will increase because it now has a conductive path to ground as well as a capacitive one. This will make the streamer get hotter and lower its resistance, which increases the current even more. This process usually "runs away" and ends up with the streamer "transitioning" in a matter of nanoseconds, to a bright white arc that looks like a short circuit. However, it doesn't always run away. If the streamer is just about as long as the coil can produce (so its resistance is pretty high) and especially if it's a high frequency coil (so the capacitive currents are bigger compared to the conductive ones) it just might not happen.

As to the ringing after the strike that Terry saw: The quick discharge of the topload is a step function that could kick the system back into oscillation at any of its resonant modes.

Re: Steve Conner DRSSTC Model question
HV Enthusiast, Sat Jun 03 2006, 12:49PM

Terry, All,

Did you actually measure the peak current (i.e. 752A) in the output arc ground strikes or was this simulated value? In practice, i have never measured anything above 10A peak current (for my medium size DRSSTCs) with high current ground arcs.

Re: Steve Conner DRSSTC Model question
Terry Fritz, Sat Jun 03 2006, 05:25PM

Hi,

Did you actually measure the peak current (i.e. 752A) in the output arc ground strikes or was this simulated value? In practice, i have never measured anything above 10A peak current (for my medium size DRSSTCs) with high current ground arcs.

It might be the frequency reponse or rise time of the current monitor. I will look into that... Maybe the oscillation is messing with the monitor. The top terminal voltage was measured with planewave antenna good to 100MHz. Maybe a much better 100's of MHz CT is needed or a sheilded air core current transformer. My Pearson #101 100:1 CT is rated at 100nS rise time and 4MHz... The #110 is 20nS and 20MHz. Probably needs a 50ohm cable termination too at those speeds.

I used 108kV, 31.1pF, and 20nS for the RC time constant.

5RC = 20nS So R = 128.6 ohms.

If the terminal voltage is 108kV then the peak current is 108000 / 128.6 = 840 amps. (I had 97kV in my head yeasterday, but it is 108kV).

I discovered that it is possible to get ground arcs that stay dim and purple, when the coil is striking out just about as far as it can reach, and these don't discharge the topload instantly.

You can see the first low current light contacts in the scope pics in:



These are the leaders "tickling around" at the ground terminal. If you play it just right, those will be the only light arcing and you can avoid the "big one". Your other comments about this sound fine to me.

As to the ringing after the strike that Terry saw: The quick discharge of the topload is a step function that could kick the system back into oscillation at any of its resonant modes.

It could also be ground noise on the scope leads and such. Pumping 840 amps into the ground in 20nS might causee measurement errors Not sure... - Apparently not... Further testing seems to confirm tha the scope reading are good and pure.

Cheers,

Terry

Hi Again,

To follow up. Dan said:

Did you actually measure the peak current (i.e. 752A) in the output arc ground strikes or was this simulated value? In practice, i have never measured anything above 10A peak current (for my medium size DRSSTCs) with high current ground arcs.

I tested my Pearson #110 (10:1 20nS rise time 20MHz) and #101 (100:1 100nS rise time 4MHz) side by side on the wire from the grounded strike point to see how they compare. The coax cables from them were both terminated into 50 ohms at the scope inputs. Everything is as before.









The Yellow trace is the 4MHz #101 and the Blue trace is the 20MHz #110.

So the 4MHz current monitor is pretty usless and the 20MHz one is probably loosing a bunch of the spike...

So if one want's to measure streamer strike currents really accurately, you probably need 100MHz if not 1GHz 0f current probe bandwidth. And it needs to probably do 1000 amps for just "little" arcs. I don't know what kind of current monitor Dan was using, but maybe this is what caused to apparently very low readings?

Cheers,

Terry

Hi Yet Again,

I looked into using a Rogowski coil. But if v(t) = -m/RC i(t), the numbers just work out into impossible elements at say 20nS rise time and 1000 amps... They just don't do 50,000 A/uS stuff

So going to try a mutual inductance loop... Very insensitive, but it might pick this monster signal up with some accuracy

Cheers,

Terry


Hi Again Again

The inductance loop works!! I just made about a 5 inch loop of wire terminated to a 50 ohm low inductance resistor. The 50 ohm coax was terminated to 50 ohms at the scope too.



The loop (purple) is about 6 inches away from the ground wire for the strike target (yellow). It just inductively couples to whatever high current high speed signals happen to be around.

I found one!





It is still real "ringy". I may have run out of bandwidth on the scope now too. The loop could be much smaller since the signal almost pegs the scope!! The resistor could have been a much smaller 1 Watt surface mount too but I don't have one handy. The current is not to any scale. But the signal is somwwhat cleaner although far from "really" clean...

Basically, the discharge is very fast. 20nS or less. And the current is "high" probably like 1000 amps.

So I would say that Dan's reading of 10 amps is "low" It would take a small very well terminated for 1GHz loop and a 1GHz scope to really see it well I think. It is possible to "calibrate" such a thing too...

If one "touches" such a spark, it is odd to think about what say 100MW in 20nS might be doing to the atoms in ones body And this is just a "tiny" 6 inch spark...

Cheers,

Terry



Me again,

Here is a real interesting post on streamer arc to ground currents and times:



Here is a note on the "new" ScanTesla7.61 and how it does streamer loading:



Cheers,

Terry

Re: Steve Conner DRSSTC Model question
Terry Fritz, Mon Jun 05 2006, 05:56PM

Hi,

I just put up ScanTesla7.62 at:



It seems to work fine for conventional, SISG, and DRSSTC coils now. I included examples of all of them. The streamer load model is in the C file (plain text), but the meat looks like this:

=======================
LeaderLength = 27.19734 * sqrt(0.5 * VCsec_max * VCsec_max * (C2 + C3)); //Secondary energy length.

if (LeaderLength > StrikeDistance) LeaderLength = StrikeDistance;

if (T1 != T1_start && C3_inc1 < 0.0) //Do this if active streamer modeling is enabled and the model has run once.
{

C3 = (LeaderLength - TerminalDiameter) / 6 * 1.0e-12; //Streamer capacitive load.

if (C3 < 0.0) C3 = 0.0;
======================

Pretty simple now. I got the simulations and all done and added now so the constants are accurate. The program also has goal type 3 now to search for the best streamer length. Dwell time can also be used for conventional coil quench time too. The arcs to ground seem to work well now and match the scope. So, finally, the program seems to be doing its master's bidding

The default input file in the ZIP is for Dan's new DRSSTC-III. I was not sure for the toroid diameter and maybe a few of the inputs, but they are probably close. The program gave this:

================
ScanTesla V-7.62 June 4, 2006 Terry Fritz
Goal = 4.180500e+001 Maximum Leader Length
Model Number = 940
Goal Number = 125
Cprimary = 1.100000e-007
Lprimary = 9.100000e-006
Rprimary = 1.000000e-001
Coupling = 0.182500
Csecondary = 2.099000e-011
Lsecondary = 4.387000e-002
Rsecondary = 3.850000e+002
Cload = 6.300833e-012
Rload = 2.200000e+005
BPS = 120.000000
Dwell (T1) Time = 6.800000e-005
Ilprimay Maximum = 560.892753
ICprimary RMS (Arms) = 22.307666 ILsecondary RMS (Arms) = 0.391537
VCprimary Maximum = -5120.878882
VCsecondary Maximum = -416110.197722
Coil Power = 328.820961 Primary Bang Energy = 2.740175
Load Power = 220.028924 Load Bang Energy = 1.833574
Leader Length (in) = 41.804996 Streamer
Primary F0 = 159075.425254 Secondary F0 = 165855.582964
Load Energy Rise Time (uSec) = 50.600000

Models Tested = 1606 / 1606
========================

BTW - Dan's low strike current measurement I think is explained in the last post above here. I have been adding to it for the last two days

Cheers,

Terry

Re: Steve Conner DRSSTC Model question
Steve Conner, Tue Jun 06 2006, 10:01AM

Hi Terry,

Great stuff! I have some data that I captured from my own DRSSTC here: There is a DSO capture of the primary current during a ground strike. Is there any way we can get ScanTesla to model this and see how the simulated primary current matches up?

As regards probe bandwidth for capturing ground strikes: If the arc channel had very low resistance (which I believe it does) it would resonate like a metal antenna excited by a spark gap, much as in Hertz's original experiment. So, I imagine the ground current waveform would be a damped ringing whose frequency was such that the streamer was something like a quarter or a half wave long. That works out in the 50-100MHz range for the sizes of coils that most of us play with. The impedance of the connection to the ground target would affect this however.

Does your mutual inductance loop pick up i(t), or di/dt?

Re: Steve Conner DRSSTC Model question
Terry Fritz, Thu Jun 08 2006, 02:43AM

Great stuff! I have some data that I captured from my own DRSSTC here: There is a DSO capture of the primary current during a ground strike. Is there any way we can get ScanTesla to model this and see how the simulated primary current matches up?

The parameters I found for this situation did not match up right. I asked Steve if he still has the data on the coil measured. ScanTesla "should" be able to "do it" with the right input. If not, I will "fix it"

Does your mutual inductance loop pick up i(t), or di/dt?

That is a very good question??? I was just looking for "speed" data... All that "it" response data is unknown... I really have no idea other than the spark is "darn fast"!!!! and runs into the low thousands of amsps!!

Cheers,

Terry