Of notches, couplng and frequency splitting
Finn Hammer, Mon Apr 30 2007, 02:48PMAll,
In an attempt to find an explanation of the subject that I can understand, I modeled a tesla coil in Orcad.
It has Cpri=33nF, Lpri=77uH Csec=33pF Lsec=77mH and K12=0.0688836105 So that it should have it`s first notch after 7,5 primary cycles.
A plot of the primary and secondary currents looks like this:
and a FFT has the expected appearance:
IOW, both coil systems oscillate at a clean 100kHz if by them selves, and if coupled, they are, by popular convention, assumed to oscillate at 2 frequencies simultaneously.
I guess they do so, in the frequency domain, but I live in the time domain, so I measured the time of each peak , to see what comes up.
I get the list shown below, which shows that for the vast majority of the time, both coils are oscillating very close to 100kHz. Only at the times where there is little energy in the system, around the notches, do they oscillate faster, and I think this is because they have to reverse the phase, from giving to taking, but I am not sure about this.
I cannot see the low pole in these numbers.
I offer these numbers for all to ponder, because I feel it may help break the notion, that the coils oscillate in a magic unpredictable pattern. It may not be so complicated.
Of course, this is a disruptive coil, without breakout, with both coil systems tuned to the same frequency.
This is not the case with the typical DRSSTC, or so I have heard.
I have full understanding that people still guard their respective tuning procedures as a personal treasure.
But I also think, that the quest for the next generation of DRSSTC drivers has to rely on an openness about tuning, which at least "I" am not part of yet.
Explanation:
1row, time-sec-current = time for pos. peak and neg. peak of secondary current
2row, Incr = time period from peak to peak of secondary current
3row, "freq" =Inverse of time period, to imply a frequency relationship.
I cannot get the table to show up in proper format, so here it is:
Is this new, wrong or rehash?
Finn Hammer
Re: Of notches, couplng and frequency splitting
HV Enthusiast, Mon Apr 30 2007, 03:00PM
I don't think anyone is guarding their tuning procedures. Steve Ward, myself, and others have been fully open with "how" we tune our DRSSTCs. I have even published a short paper on a DRSSTC tuning study.
I personally think the book is nowhere near to be closed on the proper (or best) way to tune a DRSSTC, so perhaps you are merely confusing this uncertainty to "guarding."
HV Enthusiast, Mon Apr 30 2007, 03:00PM
I don't think anyone is guarding their tuning procedures. Steve Ward, myself, and others have been fully open with "how" we tune our DRSSTCs. I have even published a short paper on a DRSSTC tuning study.
I personally think the book is nowhere near to be closed on the proper (or best) way to tune a DRSSTC, so perhaps you are merely confusing this uncertainty to "guarding."
Re: Of notches, couplng and frequency splitting
Steve Conner, Mon Apr 30 2007, 04:14PM
No, they are oscillating at two frequencies simultaneously. It just so happens that when this is viewed in the time domain, the zero crossings of the resulting waveform are at intervals equivalent to a frequency (f1+f2)/2 - where f1 and f2 are the upper and lower pole frequencies.
If both coils were tuned to 100kHz to start with, then f1 and f2 will be equally spaced around 100kHz, and hence the zero crossings will come at a rate of 100kHz.
Under certain tuning conditions (which I could never be bothered doing the math to quantify) the waveform will reverse phase at the notch, giving one cycle that is twice the usual length.
I have documented the method I use to design and tune DRSSTCs several times in the past, in fact it may even be in the manual for my PLL driver board. The reason why I don't talk more about it is that I don't understand it well enough myself to say with confidence that it will work on any DRSSTC. Maybe it only works for mine. But anyway, here it is one more time:
Design resonator with a characteristic impedance between 50k and 100k Ohms
Use stubby secondary and 45 deg. conical primary to get tightest coupling possible without flashover
Choose primary inductance and capacitance to give a resonant frequency about 10% below the secondary, and a loaded Q of 10. In other words, when the inverter is delivering the maximum current you want to use, the tank capacitance should be such that the voltage across it is 10 times the DC bus voltage. This also tells you the tank cap voltage rating.
Set the current limit on your driver to the above maximum current. (This also guarantees that the tank capacitor won't be overvoltaged.)
If using a PLL driver, tune to the upper of the two pole frequencies.
Adjust the primary tapping point so that the current limit only just triggers when the coil is running at full DC bus voltage.
You should now have the size of sparks you expected when you started the design. If you didn't realise, the spark length was set by the bang energy you designed the inverter to deliver, which is roughly 2/Pi times the DC bus voltage times the inverter current limit times 200 microseconds (even if your burst is longer than that.) Or maybe it's 1/Pi or 4/Pi, I forget.
If you didn't size the resonator appropriately for that, too bad. If you made it too small (as I did) it will flash over and burn up, too big and your sparks will look wimpy.
Finding a universally valid tuning method (or proving that no optimum method exists) would be a major research project, which I couldn't even begin to be bothered doing unless someone paid me
I suspect any method that really worked would need to be a numerical simulation that incorporated a dynamic model of streamer loading, which we don't have. (I believe Terry Fritz is close to this with ScanTesla, though.)
Steve Conner, Mon Apr 30 2007, 04:14PM
No, they are oscillating at two frequencies simultaneously. It just so happens that when this is viewed in the time domain, the zero crossings of the resulting waveform are at intervals equivalent to a frequency (f1+f2)/2 - where f1 and f2 are the upper and lower pole frequencies.
If both coils were tuned to 100kHz to start with, then f1 and f2 will be equally spaced around 100kHz, and hence the zero crossings will come at a rate of 100kHz.
Under certain tuning conditions (which I could never be bothered doing the math to quantify) the waveform will reverse phase at the notch, giving one cycle that is twice the usual length.
I have documented the method I use to design and tune DRSSTCs several times in the past, in fact it may even be in the manual for my PLL driver board. The reason why I don't talk more about it is that I don't understand it well enough myself to say with confidence that it will work on any DRSSTC. Maybe it only works for mine. But anyway, here it is one more time:
Design resonator with a characteristic impedance between 50k and 100k Ohms
Use stubby secondary and 45 deg. conical primary to get tightest coupling possible without flashover
Choose primary inductance and capacitance to give a resonant frequency about 10% below the secondary, and a loaded Q of 10. In other words, when the inverter is delivering the maximum current you want to use, the tank capacitance should be such that the voltage across it is 10 times the DC bus voltage. This also tells you the tank cap voltage rating.
Set the current limit on your driver to the above maximum current. (This also guarantees that the tank capacitor won't be overvoltaged.)
If using a PLL driver, tune to the upper of the two pole frequencies.
Adjust the primary tapping point so that the current limit only just triggers when the coil is running at full DC bus voltage.
You should now have the size of sparks you expected when you started the design. If you didn't realise, the spark length was set by the bang energy you designed the inverter to deliver, which is roughly 2/Pi times the DC bus voltage times the inverter current limit times 200 microseconds (even if your burst is longer than that.) Or maybe it's 1/Pi or 4/Pi, I forget.
If you didn't size the resonator appropriately for that, too bad. If you made it too small (as I did) it will flash over and burn up, too big and your sparks will look wimpy.
Finding a universally valid tuning method (or proving that no optimum method exists) would be a major research project, which I couldn't even begin to be bothered doing unless someone paid me
I suspect any method that really worked would need to be a numerical simulation that incorporated a dynamic model of streamer loading, which we don't have. (I believe Terry Fritz is close to this with ScanTesla, though.)Re: Of notches, couplng and frequency splitting
Finn Hammer, Mon Apr 30 2007, 04:37PM
Steve,
The table shows the peaks only, what you say is that including the zero crossings will reveal the 2 frequencies, then?
I`l get a busy night to copy that into the table.
Dan, no offence. I may have been paying too little attention to the various posts about tuning the coils.
Cheers, Finn Hammer
Finn Hammer, Mon Apr 30 2007, 04:37PM
Steve,
The table shows the peaks only, what you say is that including the zero crossings will reveal the 2 frequencies, then?
I`l get a busy night to copy that into the table.
Dan, no offence. I may have been paying too little attention to the various posts about tuning the coils.
Cheers, Finn Hammer
Re: Of notches, couplng and frequency splitting
HV Enthusiast, Mon Apr 30 2007, 05:10PM
Here is a copy of the tuning document i did awhile back. i apologize for the quality, but i had to compress the images to reduce file size for attachment.
]1177953044_15_FT24628_tuning.pdf[/file]
HV Enthusiast, Mon Apr 30 2007, 05:10PM
Here is a copy of the tuning document i did awhile back. i apologize for the quality, but i had to compress the images to reduce file size for attachment.
]1177953044_15_FT24628_tuning.pdf[/file]
Re: Of notches, couplng and frequency splitting
Steve Ward, Mon Apr 30 2007, 05:59PM
Ive talked a few times about my tuning techniques (they certainly exist in other DRSSTC threads on this forum). What i have found from experimentation is that small coils need different tuning than larger coils. What defines small vs large? In my case, my DRSSTC-3 with its 12" tall secondary running at ~150khz (IIRC) was sufficiently small to require a different tuning technique than my DRSSTC-1 with its 22" tall secondary at 60khz. To further support my idea, all of my smaller DRSSTCs (one had a 6" tall secondary, and another with a 3" tall secondary operating at 1.2MHz) also required the "small coil" tuning technique to achieve good output spark length. On the other hand, my largest DRSSTC tunes rather well with the "large coil" tuning technique.
Large coil tuning: This tuning is straightforward and basically follows what we already know about notching. My larger coils usually have coupling near .18 or so, where a notch occurs sometime after 6-9 driving cycles. When a notch is seen (via CT on the primary circuit), the driving is ended. Notches may only be apparent at low power, and with proper tuning. You can also produce semi-notches when you tune slightly to either direction of this "notch producing" tuning point. I typically will tune a little bit lower to compensate for spark loading. So now you should still see a significant dip in primary current, this is the shut off point. As the sparks grow longer, the primary current envelope changes shape a bit, but there is still a significant dip in primary current. This dip indicates that most of the energy is now in the secondary coil, thus making it a good point to stop driving the system. You may find that driving the primary further will result in a second "bounce" in the current envelope, but this second bounce is usually of *very* high primary currents, but with not much benefit in spark length.
Small coil tuning: Small coils seem to require more cycles to produce really long sparks. If using the large coil tuning scheme on a small coil, you might end up dissappointed with sparks maybe half as long as otherwise possible given the size of the coil. So for small coils, i use a trick to build up energy in the primary circuit. I severely detune the primary circuit, so that it builds up energy without much transfer to the secondary. Eventually, though, there is a breaking point where the secondary develops enough voltage to begin breakout. Just after this point, a huge transformation takes place in the behavior of the coil. It may have to be seen in person to really get the idea, but the sparks jump from an inch in length, to maybe 3 feet in length, with a matter of 5% (sometimes less) voltage increase to the DC bus. What i believe happens at this transition point, is that spark loading (that 1" of corona) is enough to begin lowering the Q of the secondary circuit. At this point, energy transfer is much easier between the 2 circuits, and the primary (which now has a lot of energy in it) can dump it all into the secondary. From what ive seen here, ~3 cycles after the initial breakout begins, the primary is *drained* and a current notch is seen (so shut down the drive here). You probably wont be able to reproduce this in pspice too easily.. i couldnt. But, to further convince myself that this phenomena is real, i detuned my coil even more, to the point where it required 600A peak current (vs 400A normally) to achieve breakout. And as expected, the sparks grew even longer. I should mention that i always detune the primary to resonate lower than the secondary. At some point, you will likely exceed the current capability of your circuit, and not achieve any secondary streamers, so this is when you have gone too far.
There you have it, my tuning techniques. I have no idea if they work for anyone else but myself (5 coils may not be enough of a sample to validate or debunk my ideas).
Steve Ward, Mon Apr 30 2007, 05:59PM
Ive talked a few times about my tuning techniques (they certainly exist in other DRSSTC threads on this forum). What i have found from experimentation is that small coils need different tuning than larger coils. What defines small vs large? In my case, my DRSSTC-3 with its 12" tall secondary running at ~150khz (IIRC) was sufficiently small to require a different tuning technique than my DRSSTC-1 with its 22" tall secondary at 60khz. To further support my idea, all of my smaller DRSSTCs (one had a 6" tall secondary, and another with a 3" tall secondary operating at 1.2MHz) also required the "small coil" tuning technique to achieve good output spark length. On the other hand, my largest DRSSTC tunes rather well with the "large coil" tuning technique.
Large coil tuning: This tuning is straightforward and basically follows what we already know about notching. My larger coils usually have coupling near .18 or so, where a notch occurs sometime after 6-9 driving cycles. When a notch is seen (via CT on the primary circuit), the driving is ended. Notches may only be apparent at low power, and with proper tuning. You can also produce semi-notches when you tune slightly to either direction of this "notch producing" tuning point. I typically will tune a little bit lower to compensate for spark loading. So now you should still see a significant dip in primary current, this is the shut off point. As the sparks grow longer, the primary current envelope changes shape a bit, but there is still a significant dip in primary current. This dip indicates that most of the energy is now in the secondary coil, thus making it a good point to stop driving the system. You may find that driving the primary further will result in a second "bounce" in the current envelope, but this second bounce is usually of *very* high primary currents, but with not much benefit in spark length.
Small coil tuning: Small coils seem to require more cycles to produce really long sparks. If using the large coil tuning scheme on a small coil, you might end up dissappointed with sparks maybe half as long as otherwise possible given the size of the coil. So for small coils, i use a trick to build up energy in the primary circuit. I severely detune the primary circuit, so that it builds up energy without much transfer to the secondary. Eventually, though, there is a breaking point where the secondary develops enough voltage to begin breakout. Just after this point, a huge transformation takes place in the behavior of the coil. It may have to be seen in person to really get the idea, but the sparks jump from an inch in length, to maybe 3 feet in length, with a matter of 5% (sometimes less) voltage increase to the DC bus. What i believe happens at this transition point, is that spark loading (that 1" of corona) is enough to begin lowering the Q of the secondary circuit. At this point, energy transfer is much easier between the 2 circuits, and the primary (which now has a lot of energy in it) can dump it all into the secondary. From what ive seen here, ~3 cycles after the initial breakout begins, the primary is *drained* and a current notch is seen (so shut down the drive here). You probably wont be able to reproduce this in pspice too easily.. i couldnt. But, to further convince myself that this phenomena is real, i detuned my coil even more, to the point where it required 600A peak current (vs 400A normally) to achieve breakout. And as expected, the sparks grew even longer. I should mention that i always detune the primary to resonate lower than the secondary. At some point, you will likely exceed the current capability of your circuit, and not achieve any secondary streamers, so this is when you have gone too far.
There you have it, my tuning techniques. I have no idea if they work for anyone else but myself (5 coils may not be enough of a sample to validate or debunk my ideas).
Re: Of notches, couplng and frequency splitting
Finn Hammer, Mon Apr 30 2007, 06:50PM
Ok, so i plotted the zero crossings into the file, so that a record of the peaks and zero crossings are recorded.
what I see is basically a sawtooth waveform, with round tops.
Thanks to all for the tuning tips, I appreciate it!
Cheers, Finn Hammer
Finn Hammer, Mon Apr 30 2007, 06:50PM
Steve Conner wrote ...
No, they are oscillating at two frequencies simultaneously. It just so happens that when this is viewed in the time domain, the zero crossings of the resulting waveform are at intervals equivalent to a frequency (f1+f2)/2 - where f1 and f2 are the upper and lower pole frequencies.
If both coils were tuned to 100kHz to start with, then f1 and f2 will be equally spaced around 100kHz, and hence the zero crossings will come at a rate of 100kHz.
No, they are oscillating at two frequencies simultaneously. It just so happens that when this is viewed in the time domain, the zero crossings of the resulting waveform are at intervals equivalent to a frequency (f1+f2)/2 - where f1 and f2 are the upper and lower pole frequencies.
If both coils were tuned to 100kHz to start with, then f1 and f2 will be equally spaced around 100kHz, and hence the zero crossings will come at a rate of 100kHz.
Ok, so i plotted the zero crossings into the file, so that a record of the peaks and zero crossings are recorded.
what I see is basically a sawtooth waveform, with round tops.
Thanks to all for the tuning tips, I appreciate it!
Cheers, Finn Hammer
Re: Of notches, couplng and frequency splitting
Finn Hammer, Tue May 08 2007, 08:47PM
I just found a way to understand this:
I connected 2 function generators together via 100K isolating resistors, and observed the resulting waveform on the scope:
Viola! an intermediate frequency with notches, all nicely corresponding with the tables that Antonio posted on the pupman list back in 2003:
Cheers, Finn Hammer
Finn Hammer, Tue May 08 2007, 08:47PM
I just found a way to understand this:
I connected 2 function generators together via 100K isolating resistors, and observed the resulting waveform on the scope:
Viola! an intermediate frequency with notches, all nicely corresponding with the tables that Antonio posted on the pupman list back in 2003:
Cheers, Finn Hammer
Print this page