Streamer loading
Finn Hammer, Wed Nov 29 2006, 08:36PMI would like to invite to an exchange of ideas of how the streamers load the coil.
In the other thread, Stephen Conner advised me to load the coil harder, by halving the frequency and also lowering the breakrate.
This helped a lot.
What I have observed in real life, and now also with scantesla is, that by increasing the voltage to the coil, the ring down jumps from 3rd. to 2nd. and sometimes to 1st. notch quench.
When this happens, the coil really screams, so it would be nice to predict this operating point by educated guess rather than trial and error or "poking around in the dark".
Is it a matter of matching the impedance of the streamer to that of the coil, and if so: how do i calculate these impedances.
Any thoughts on this subject are greatly appreciated.
Cheers, Finn Hammer
Re: Streamer loading
Hazmatt_(The Underdog), Thu Nov 30 2006, 01:43AM
I'm going to be looking into this once the build phase of the CSULB Tesla Coil is finished.
One method we are going to conduct the measurements is with Terry's antenna.
The other method will be with impedance probes. They will be passive and constructed from small RF connectors, chip resistors and capacitors. There are some details on the web about these.
Then of course, there is PSpice and related variants, which I will be looking at how the spectrum notches. So I will post what I can and hopefully in a relavent manner.
Hazmatt_(The Underdog), Thu Nov 30 2006, 01:43AM
I'm going to be looking into this once the build phase of the CSULB Tesla Coil is finished.
One method we are going to conduct the measurements is with Terry's antenna.
The other method will be with impedance probes. They will be passive and constructed from small RF connectors, chip resistors and capacitors. There are some details on the web about these.
Then of course, there is PSpice and related variants, which I will be looking at how the spectrum notches. So I will post what I can and hopefully in a relavent manner.
Re: Streamer loading
teravolt, Thu Nov 30 2006, 05:58AM
Finn this might be stupid questons but doesn't the secondary's resonoant frequency go down as the spark load gets bigger? Maby the frequency splitting changes with more of a load and pole amplitudes are dynamic with spark load and cupling. How much does tuning low help? Maby a parmertic spice progam could be made. It will interesting to see where this thread will go
teravolt, Thu Nov 30 2006, 05:58AM
Finn this might be stupid questons but doesn't the secondary's resonoant frequency go down as the spark load gets bigger? Maby the frequency splitting changes with more of a load and pole amplitudes are dynamic with spark load and cupling. How much does tuning low help? Maby a parmertic spice progam could be made. It will interesting to see where this thread will go
Re: Streamer loading
Finn Hammer, Thu Nov 30 2006, 06:45AM
All,
When I turned up the voltage to the coil and watched the quench progress from 3rd. to first notch I thought that I saw the effect of tuning: The streamer capacitance adding to the top load capacitance, and thus bringing the coil into better tuning alignment.
But Scantesla corrects for this.
In the following text file you can see how doubling the voltage makes scantesla detune the primary to the new resonant frequency with longer streamer, something the real tesla coil could not do.
and on this picture you can see how putting more power trough the coil makes it quench faster:
It seems obvious that the coil is happier at the high powered operating point to the right.
But how to predict this operating point without at first making some general guidelines which can lead towards a more rigid method.
Is all this at all understood?
Cheers, Finn Hammer
Finn Hammer, Thu Nov 30 2006, 06:45AM
All,
When I turned up the voltage to the coil and watched the quench progress from 3rd. to first notch I thought that I saw the effect of tuning: The streamer capacitance adding to the top load capacitance, and thus bringing the coil into better tuning alignment.
But Scantesla corrects for this.
In the following text file you can see how doubling the voltage makes scantesla detune the primary to the new resonant frequency with longer streamer, something the real tesla coil could not do.
and on this picture you can see how putting more power trough the coil makes it quench faster:
It seems obvious that the coil is happier at the high powered operating point to the right.
But how to predict this operating point without at first making some general guidelines which can lead towards a more rigid method.
Is all this at all understood?
Cheers, Finn Hammer
Re: Streamer loading
Finn Hammer, Thu Nov 30 2006, 06:06PM
I hope I have found enough out, to warrant a double post.
Poking around on the web, I stumbled upon Derek`s site, he has a Zo calculation in one of his coil calculators.
Dropping some numbers into it, I noticed it had the same feel as the surge impedance calculator on my spreadsheet.
So it would appear that the Zout of the Tesla Coil secondary resonator is simply the surge impedance of a LC circuit, which is Z = SQRT L/C
This can be rewritten as:
L = C*Z^2
Now, if the output impedance of the Tesla coil secondary should match that of the streamer (which Terry has just recently determined to be 125Kohm), then it becomes very easy to figure things out.
Choose a coil size that looks about right for the power input and expected streamer length.
Put a topload on it to get a Csec number.
Then calculate Lsec.according to the formula above.
One thing I noticed was that I get awfully many turns on the secondary coil, let me make an example:
A coil, 10" by 45" with a 8" by 32" toroid has a total Csec of 54.75pF (Wintesla)
This adds up to a Lsec. of 855mH which *I* think is a lot. It takes 4100 turns of No.31 wire. and drops the res. freq. down to 20.5kHz.
Not the numbers we are accustomed to.
And I guess an indication that the idea of matching impedances for most efficient energy transfer is not the whole truth.
I am determined to find this out, but I cannot do it alone.
Cheers, Finn Hammer
Finn Hammer, Thu Nov 30 2006, 06:06PM
I hope I have found enough out, to warrant a double post.
Poking around on the web, I stumbled upon Derek`s site, he has a Zo calculation in one of his coil calculators.
Dropping some numbers into it, I noticed it had the same feel as the surge impedance calculator on my spreadsheet.
So it would appear that the Zout of the Tesla Coil secondary resonator is simply the surge impedance of a LC circuit, which is Z = SQRT L/C
This can be rewritten as:
L = C*Z^2
Now, if the output impedance of the Tesla coil secondary should match that of the streamer (which Terry has just recently determined to be 125Kohm), then it becomes very easy to figure things out.
Choose a coil size that looks about right for the power input and expected streamer length.
Put a topload on it to get a Csec number.
Then calculate Lsec.according to the formula above.
One thing I noticed was that I get awfully many turns on the secondary coil, let me make an example:
A coil, 10" by 45" with a 8" by 32" toroid has a total Csec of 54.75pF (Wintesla)
This adds up to a Lsec. of 855mH which *I* think is a lot. It takes 4100 turns of No.31 wire. and drops the res. freq. down to 20.5kHz.
Not the numbers we are accustomed to.
And I guess an indication that the idea of matching impedances for most efficient energy transfer is not the whole truth.
I am determined to find this out, but I cannot do it alone.
Cheers, Finn Hammer
Re: Streamer loading
Steve Conner, Thu Nov 30 2006, 06:19PM
Hi Finn
From my DRSSTC work, and Richie's CW SSTC work and equations, and a bunch of articles on RF power amp design, I figured out the following stuff:
1) The loaded Q of a Tesla coil should be less than 10 for efficient operation.
2) Even though the concept of loaded Q is meant for application to CW systems, it still seems to be applicable to DRSSTC style pulsed systems, though there could be an error of up to a factor of 2 with short bursts.
3) The loaded Q is roughly equal to the shunt equivalent of the streamer load resistance, divided by the characteristic impedance of the resonator. (which is just the surge impedance- sqrt(L/C)- like you figured out)
Terry's streamer load model at the time was a series RC of 220k + 1pF/foot, and when transformed to the shunt equivalent, it gave resistances of about 500k to 1M Ohm, depending on the streamer length. So I started recommending that people build resonators with a surge impedance of 50 to 100k Ohms. Yours was towards the bottom end of that range, hence I recommended increasing it. Interestingly my OLTC2 was right at the top, at 130k.
I managed to show that with DRSSTCs the effective output impedance, hence loaded Q, can be adjusted over a fairly broad range by just moving the primary tapping point, so a more accurate analysis wasn't needed. If I picked an operating frequency, set the loaded Q of both primary and secondary circuits to 10, and specified an output voltage greater than the breakout voltage of the bare toroid, this was enough to completely constrain the design of a DRSSTC and tell me how many primary turns and what size tank capacitor to use. This was how I designed my last DRSSTC, and I was more than happy, terrified even, with the output from it.
I've not managed to stay up to speed with Terry's newer streamer models and their implications, but I think his new stuff suggests that bigger coils want higher surge impedances.
Steve Conner, Thu Nov 30 2006, 06:19PM
Hi Finn
From my DRSSTC work, and Richie's CW SSTC work and equations, and a bunch of articles on RF power amp design, I figured out the following stuff:
1) The loaded Q of a Tesla coil should be less than 10 for efficient operation.
2) Even though the concept of loaded Q is meant for application to CW systems, it still seems to be applicable to DRSSTC style pulsed systems, though there could be an error of up to a factor of 2 with short bursts.
3) The loaded Q is roughly equal to the shunt equivalent of the streamer load resistance, divided by the characteristic impedance of the resonator. (which is just the surge impedance- sqrt(L/C)- like you figured out)
Terry's streamer load model at the time was a series RC of 220k + 1pF/foot, and when transformed to the shunt equivalent, it gave resistances of about 500k to 1M Ohm, depending on the streamer length. So I started recommending that people build resonators with a surge impedance of 50 to 100k Ohms. Yours was towards the bottom end of that range, hence I recommended increasing it. Interestingly my OLTC2 was right at the top, at 130k.
I managed to show that with DRSSTCs the effective output impedance, hence loaded Q, can be adjusted over a fairly broad range by just moving the primary tapping point, so a more accurate analysis wasn't needed. If I picked an operating frequency, set the loaded Q of both primary and secondary circuits to 10, and specified an output voltage greater than the breakout voltage of the bare toroid, this was enough to completely constrain the design of a DRSSTC and tell me how many primary turns and what size tank capacitor to use. This was how I designed my last DRSSTC, and I was more than happy, terrified even, with the output from it.
I've not managed to stay up to speed with Terry's newer streamer models and their implications, but I think his new stuff suggests that bigger coils want higher surge impedances.
Print this page