Help with SSTC feedback circuits with 4046 IC and Schmitt Triggers

[AFreshLad]

I have built my first SSTC a while ago now and have been running it perfectly fine using an external clock signal at 100% duty cycle at high power. But now I want to add an interrupt circuit to be able to play music on it, and thus need the coil to run on a feedback signal. So I attached a CT at the primary and successfully got a clean enough feedback signal that was able to drive the coil. But the problem is that the moment it strikes an arc to ground, the coil shuts down because the power supply shuts off (its a switch mode PSU I used for low power testing). I'm not sure what the problem is, is it that the frequency of the feedback will decrease when an arc strikes ground and basically cause the primary to become a short?

Also, my feedback circuit consists of the CT output put across a shunt, and its voltage amplified by a Darlington array, which creates a good enough square wave, so why is a Schmitt trigger always used, does it have some other advantage I don't know of? And the other feedback circuit I've seen is the use of a 4046 IC, which I've researched a bit, but still don't really understand why its used. Is it because the 4046 will continue to output the default 'center' frequency if the tesla coil feedback crashes and goes outside of the lock range? I'm just looking to find the best and most stable way to get a feedback signal and any help will be appreciated.
Thanks!

[davekni]

But the problem is that the moment it strikes an arc to ground, the coil shuts down because the power supply shuts off (its a switch mode PSU I used for low power testing).

It may be electrical noise from the arc confusing the supply rather than overload.

[AFreshLad]

I'm not sure how that could occur though, because when I run the coil with an external signal, even when I strike an arc to ground its completely fine.

[Duane B]

why is a Schmitt trigger always used

A Schmitt trigger has hysteresis so it is somewhat immune to noise. Say you have 30 mv of noise on the signal you are switching on, the Schmitt trigger will only switch once at the crossover point, but a regular gate may switch several times in the 30 mv window where the noise is..

[AFreshLad]

why is a Schmitt trigger always used

A Schmitt trigger has hysteresis so it is somewhat immune to noise. Say you have 30 mv of noise on the signal you are switching on, the Schmitt trigger will only switch once at the crossover point, but a regular gate may switch several times in the 30 mv window where the noise is..

Thanks for the explanation of the Schmitt trigger but I'm still confused as to why a 4046 IC would be used? Does it also just provide filtering of some sort or does it have more features? Thanks.

[Duane B]

More info on the 4046:

https://highvoltageforum.net/index.php?topic=460.0

[AFreshLad]

More info on the 4046:

https://highvoltageforum.net/index.php?topic=460.0

From what I understand from reading the link is the 4046 only for providing a feedback signal within a limited frequency so that if the feedback goes beyond the limits, its return to the center default frequency?

[Magneticitist]

There are varying levels of intuitive design when it comes to PLL circuits but the ones on the simpler side just require you set that frequency range and hope you can find the secondary resonance within it. I believe fine tuning how small a window this range is will allow you to fine tune the phasing better. If you've ever looked at PLL SSTC's and thought "wow that's nice fat output" it's probably because of the tuned phase angle which is allowing maximum power throughput. When I played with the basic PLL circuit (Scopeboy's) I found it very easy to use an antenna and just change the timing capacitor to give me something around maybe 150-200khz window with my secondary resonance fairly close to the middle.   

The VCO on the PLL will output a 50% square wave at all times with a frequency depending on the timing components. Tuning that in til it finds a lock by getting resonant feedback will then allow you to use the same potentiometer for frequency tuning to then adjust the phase angle. If for whatever reason it drops out of lock it may find it again but not necessarily during the time you're loading it causing it to drop out. I think for me I had to play with the values a little til noticing I was able to load the coil various ways while maintaining lock and was then able to crank full voltage. Before that tuning I could bring my hand close or something and it would drop out. Steve Conner and Steve Ward I think have both designed more intuitive PLL circuits that are robust enough to handle ground strikes and various loading without difficulty but those seemed pretty complicated to me.

[AFreshLad]

By loading the coil do you mean just bringing objects close to the coil or just straight shorting it out with a grounded arc. I can understand how loading it by increasing the topload capacitance can keep the feedback inside the lock range, but I don't really understand what happens to the feedback frequency when a ground arc is made. From my experience I'm guessing the frequency decreases or increases way outside what it should be, and outside the lock range of a 4046. Could it also depend on the type and placement of the feedback, because as you said, you used an antenna which gets feedback from the secondary, whilst I used a CT on the primary, and as my tesla coil is not a DRSSTC, I'm guessing when the secondary is shorted with a ground arc, the primary doesn't really follow the resonant frequency of the secondary and rather follows its own resonance with whatever parasitic capacitances and inductances it has. I'm guessing this because the one tesla tesla coil is a DRSSTC and they got away with just a CT on the primary with no 4046, allowing them to use the resonance of the primary if the secondary were to short.

Also, concerning the phase angle, I'm guessing the feedback signal should be somewhat phase shifted from the signal I want to output, as the 4046 cannot give you a signal with the same frequency and phase as the output (I think) as the VCO needs some voltage from the phase comparator. So to just make things easier, as tuning a 4046 seems very finicky, should I just have a separate circuit that phase shifts the signal? I have already done this somewhat crudely with my current feedback circuit, by adding a capacitor and potentiometer in series across the feedback signal, allowing me to adjust the capacitive and resistive component and give me theoretically 0 - 90 degree adjustable phase shift, and it does work to an extent.

Thanks for all the help!!

[johnf]

Be very careful as to which 4046 you use
not all vendors ones are the same with some nasty quirks
I standardised on the Philips one as it had a very good datasheet

[AFreshLad]

I have already ordered mine from texas instruments. The datasheet isn't as extensive as the Phillips one but I'm hoping that the IC being from a reputable brand will be suitable.

[Magneticitist]

I'd heard from others that the PLL can be real finnicky but I was surprised when playing with some cheap CD4046's I got from ebay. I could pretty much do anything with the antenna, it was unremarkable the location or orientation etc. It's just a short piece of wire poking out randomly now. I had a long antenna fall over and touch the secondary a little then just lay flat and still, I saw no change in output at the time. I'd never pulled arcs directly to ground with it but that's what I heard was problematic, not so much that it should inherently have trouble getting a lock back but that it could cause the chip to try locking onto all types of dirty edges way outside of the f range so yes the secondary feedback seems to be problematic there. I suppose you could do the phase shifting externally but I sort of just find the CD4046 kind of handy if I just want to run 12v logic using just that IC and a couple gate drivers for example. If you just want reliable feedback, using a CT on the secondary seems reliable enough. How you are implementing primary feedback currently may be your problem though. I would think unless you've actually got the primary running as an LC tank at a frequency close to your secondary resonance then there could be all kinds of phase shift issues. Gate drivers with enables like the UCC seem to reliably allow me to run secondary feedback and interrupt without a startup oscillator because of the enable pulse on the driver outputs. Likewise when running CW I can switch the enable high and the pulse will start the feedback loop so long as there is voltage on the bus to do it. As for your dropouts it's hard to say as I guess it depends on the PSU but I can say I've never really seen issues with running secondary feedback, a decent CT, and something like a 74HC14.

[AFreshLad]

I would think unless you've actually got the primary running as an LC tank at a frequency close to your secondary resonance then there could be all kinds of phase shift issues.

Do you mean phase shift issues or frequency issues, because my guess was that the frequency of the primary just crashes too high or too low when the secondary is shorted with a ground arc. I have also been able to get the primary exactly in phase with the secondary by finely adjusting an external frequency source. Also, as my primary isn't an LC circuit, I can see that putting the CT on the secondary is probably better for feedback, but I don't know what would happen to this feedback if the secondary is shorted with a ground arc. Since this arc is basically shorting the capacitor that is the topload, wouldn't the LC circuit just fall apart, and stop resonating.

I'd never pulled arcs directly to ground with it but that's what I heard was problematic, not so much that it should inherently have trouble getting a lock back but that it could cause the chip to try locking onto all types of dirty edges way outside of the f range so yes the secondary feedback seems to be problematic there.

Also, I don't understand how your 4046 IC locked onto any signals outside the F range, as I thought it was supposed to only lock onto frequencies within the range you set.
Overall, I think I am going to go with the 4046 IC and not a 74HC14 although it seems reliable with feedback from the secondary as I want to be able to draw ground arcs with a grounded rod.
Thanks!

[AFreshLad]

I have been researching the 4046 more and looking into many people's 4046 circuit designs for tesla coils and I realized what you meant by the circuit locking onto random noise, as the datasheet says the phase comparator 1 has a tendency to lock onto harmonics, although I'm still not sure why it does. Also, I saw that the phase comparator 2 has many advantages over comparator 1, with no locking into harmonics, doesn't require 50% duty cycle input, and always has 0 degree phase difference, so why is phase comparator 2 never used. The only reason I can see is that the Vco of phase comparator 2 will go to 0 when no lock is made, which I'm guessing cannot work as the frequency of a tesla coil always decreases, not increases due to objects placed near it/ loading it. But I feel like this is such a small problem that has some hacky workaround, and I would really like to use comparator 2 as it seems so much better. Especially concerning the phase shift, comparator 1 shifts from 0 to 180 degrees as it goes from Fmin to Fmax, so it seems its not as suitable as comparator 2 in keeping the tesla coil perfectly in phase.

Also with the circuits that play audio, I see that the audio signal is injected into the signal feedback from the phase comparator to the VCO where the Low Pass Filter is, and I don't exactly get how that works. Does it just force the VCO signal into ground or VCC which forces the tesla coil out of resonance and 'turn off' the arc, and then return the tesla coil to its own feedback turning the arc 'on' again. Or is it actually modulating the VCO in with an analog signal from the audio, causing the tesla coil to shift off of resonance proportional to the audio wave amplitude and creating analog audio waves. If so, for a tesla coil capable of 100% duty cycle, an interrupt circuit wouldn't be necessary to play music right?

[AstRii]

Also with the circuits that play audio, I see that the audio signal is injected into the signal feedback from the phase comparator to the VCO where the Low Pass Filter is, and I don't exactly get how that works. Does it just force the VCO signal into ground or VCC which forces the tesla coil out of resonance and 'turn off' the arc, and then return the tesla coil to its own feedback turning the arc 'on' again. Or is it actually modulating the VCO in with an analog signal from the audio, causing the tesla coil to shift off of resonance proportional to the audio wave amplitude and creating analog audio waves. If so, for a tesla coil capable of 100% duty cycle, an interrupt circuit wouldn't be necessary to play music right?

Exactly as you stated, the VCO is modulated with analog audio which shifts the frequency off resonance. The output power is then modulated with the audio signal and that creates perfect analog audio playback.
I have made a few of these analog audio modulated coils, you can see them here:
https://www.uhvlab.org/sstc-ii
https://www.uhvlab.org/ctu-sstc

[Magneticitist]

My setup never locked onto bad frequencies but this is what I hear can happen as the IC looks to track the various dirty edges. It essentially confuses the chip as it tries to lock onto something it wont. Theoretically I suppose you don't need a dual resonant primary to use primary feedback but if you had some method of doing it by amplifying the feedback signal enough as well as filtering the really high frequencies the primary may want to ring at it CAN work. (right?) Assuming you had this ironed out I would imagine only phase shift issues could become evident if this shift became too much during loading.
I'm not really sure how you're going about this but I'd think secondary feedback would in fact do you a little better. Pulling arcs to ground has never seemed an issue to me with a good SSTC, more so a problem with a DRSSTC without any kind of OCD.

[AFreshLad]

Thanks for all the feedback!
After researching more into PLL and running a circuit sim on it, I think I have quite a good idea on how the XOR gate and VCO work together, and I sort of derived the equation for calculating F capture range of an XOR phase detector. I'm pretty sure the locking onto harmonics problem of an XOR phase detector is limited to the range of the VCO which is adjustable, and should be avoidable by making the VCO range small. But still, I want to try to find a way to use phase comparator 2 due to its ability to keep a 0 phase shift. From what I know, phase comparator 2 will not return to center if frequency lock is lost, but will still go all the way to the rails, either high or low depending on whether the frequency in is below or above center frequency. Thus, if I set the limits correctly and design a good feedback that keeps the frequency within these limits, then it should work fine, right? and even if it were to go outside the range, it should keep the frequency limited to a safe level at least. I have looked into steve's driver design from:
https://www.stevehv.4hv.org/CWcoil/PLLSchem.JPG
and he has used phase comparator 2, but I don't understand why he used an external delay circuit on the feedback from the VCO to the phase comparator to change the phase of the output waveform rather than biasing it like done on a type 1 phase comparator. To me it logically makes sense that it would still work with biasing, and I confirmed that using a circuit simulator. The only thing I noticed was that the phase shift varied quite a bit depending on if the feedback frequency proportional voltage for the VCO was closer or further from the biasing voltage, as that would determine how much the biasing voltage would effect it the circuit, but still it works. I think I will try to use phase comparator 2 just for the 0 phase shift ability and try to design a feedback signal from the secondary that won't drop out under any circumstance.

[davekni]

To me it logically makes sense that it would still work with biasing, and I confirmed that using a circuit simulator. The only thing I noticed was that the phase shift varied quite a bit depending on if the feedback frequency proportional voltage for the VCO was closer or further from the biasing voltage, as that would determine how much the biasing voltage would effect it the circuit, but still it works.

Yes, biasing phase comparator 2 does work.  I did that once decades ago, though not for a Tesla coil.  Ideal version would have phase comparator 2 output be + and - current sources rather than switches.  Then a DC current source would provide linear phase adjustment.  I did use a current source for biasing, even though 4046 phase comparator 2 output is not current source.

As the datasheet mentions, noise sensitivity is the primary disadvantage of phase comparator 2.  Input signal glitches count as valid edges, increasing input frequency.  I'd suggest a schmitt-trigger input buffer or inverter before the 4046.

[Magneticitist]

I agree about using both IC's. Teslista555 and others have seemed to benefit from adding the schmitt trigger stage. This was going to be how I went about another PLL if I got around to it.

[AFreshLad]

Yeah, I have ordered Schmit triggers alongside the 4046s will include them in the circuit between the PLL input and CT, as it doesn't really have any negative effects.