Help with SSTC feedback circuits with 4046 IC and Schmitt Triggers

[davekni]

That screen shot shows the results plot of two signals.  Can you make a screen capture of the schematic (circuit diagram) also?

[AFreshLad]

Hi, sorry I didn't reply sooner, I have just been trying to design and test a new circuit.
The problem about the phase shift stuff before I realized wasn't a problem, I must of had a brain fart for 3 days before I realized that of course the voltage would shift 180 degrees as the 'voltage source' for the primary is not the MOSFET side but the transformer side during the off time. I scoped the current on the primary and saw that it remained mostly in-phase, but did shift enough such that during the turn-on event, it would switch whilst a reasonable current was flowing through the primary, which I think is causing the voltage spikes.

My solution for this is what I've been working on, and I chose to put a CT on the primary as well, and use the feedback from that to time the turn-off and turn on events, whilst the frequency feedback is still driven by the CT on the secondary. But if I continued using the 74HC74 D type flip flop for the interrupt controller, it would be very unstable as if no current was in the primary, e.g. at startup of it the turn off time is long causing the coil to ring down to zero, the coil would just refuse to turn on as no clock edges were fed into the flip flop to tell it to turn on the coil. So I had to redesign the circuit and created feedback that pulled a signal low at zero current and high everywhere else, hence allowing the coil to startup even if the current in the primary is 0. This means I would need an SR latch rather than a D-type flip flop, and I don't have any latches so I couldn't test the circuit on the tesla coil, but the feedback circuit works and produces a usable signal. I also had to add a delay circuit to delay the pulses to match the delay in the driver circuit. My only problem is my circuit is quite 'bulky', uses a lot of components for what I think is a simple task. If you have any better ideas that would be helpful. I attached the schematic below:

The 74HC74 flip flop is for the delay, as I couldn't get enough delay by adjusting the pots on the CT feedback.

Also, I'm bothering with all this rather than just waiting for the coil to ring down as the interrupt generator I'm using in an Arduino running the TMRPCM library, which plays music by using a carrier wave of 32KHz to artificially produce the analog audio waves, and thus the off times are very low.

[AFreshLad]

I made a mistake with the schematic, the flip flop circuit for delay won't work as it's edge-triggered, defeating the whole purpose of the rest of the circuit. So I now need a different delay circuit design that can delay the pulse by about 800ns. Any ideas will be helpful.
Thanks

[AFreshLad]

I have just redesigned and built a new circuit which does successfully delay the signal whilst also keeping all the signal information by testing for both the rising and falling edge. However, the circuit is now very complex, and looking at other pulse delay circuits such as the 555 timer ones, its unnecessarily complex. The only problem is the 555 timer delay circuits I've seen delay by a large amount, about 100us, while I was a delay of about 800ns. Do you have any ideas on a simple delay circuit that keeps all the waveform information (as in if the signal in goes low, the signal out should go low and vise versa, rather than just using one rising edge to signal the pulses).
Thanks
I attached the new schematic below.

[davekni]

Schmitt inverter (HC14), series resistor, cap to ground, another schmitt inverter.  Rising edge delay will be a bit longer due to schmitt thresholds being a bit below VDD/2.  Fix that with a resistor in parallel with capacitor.

[AFreshLad]

I completely missed that idea. Thanks, I will try it, it should most like work much better than what I have now.

[Zbig]

Hi
I have just finished attaching and testing the PLL circuit with a Schmitt trigger and it works perfectly. It was not at all finicky and the phase comparator 2 works perfectly. It keeps a lock on the signal through any situation, ground arcs etc, and even when shorting the secondary to ground recovers instantly after removing the rod. Further, by fine-tuning the phase shift using a circuit similar to Steve Wards linked here:
https://www.stevehv.4hv.org/CWcoil/PLLSchem.JPG

using CP2 is the only way to get properly working PLL, with stable, adjustable switching phase, to compensate driver/MOSFET delays.   
This schematics is quite good, but not optimal.
Especially single-pole LPF may cause phase oscillations, so in my design I'm using two pole filter.

I was able to get 0 current switching to happen on my half-bridge, meaning I can finally attach the interrupt circuit without creating huge voltage spikes on my bridge. The 4046 IC was even run on a breadboard, with long looping wires connecting things at it still ran just fine (although the coil was running at very low power so less interference). The only problem I can see with phase comparator 2 is that I have set F min at ~200KHz whilst my coil resonates at ~296KHz which means the coil won't start by it self, instead I have to short the secondary and draw out an arc to start the coil. This problem might be solved at higher powers, but either way, it could also become a problem during interrupts if the secondary oscillations die out during the turn off period and doesn't restart oscillations again. This could also be solved by increasing F min.

My solution for proper (re)start is quite simple:  VCO voltage should not be kept constant during OFF period, but reset to value for nominal VCO frequency taken from low resistance voltage divider (1:1).
Also in my design I've added extra deadtime before gate drivers, and make automatic phase compensation for this dT and drivers delay by taking CP input signal from driver's output (and with one capacitor I also compensate for MOSFET turn on time). That's way whole tuning is to set VCO center frequency. 
 

 

[AFreshLad]

Hi, sorry for not responding in a while, it was our test period for school.
I decided not to use an interrupt circuit as trying to find a good way to restart the coil was turning out to be difficult because I wanted the coil to restart whilst the secondary was still resonating, causing the MOSFETs to be back driven by the secondary until I turn the MOSFETs back on, where a huge voltage spike would occur I'm guessing since it switched at a non - zero current position.

To recap, I do have a flip flop circuit to turn off and on the coil at zero current (timed by a CT on the secondary), which works perfectly when turning off, but when the coil is being back driven, it has a slight phase shift from when it's normally driven (relative to the secondary CTs timing) that causes the MOSFETs to switch whilst current is still flowing. I did try putting a CT on the primary so I would have the true current flow through the primary rather than an estimate created by the CT from the secondary, but trying to filter it and use it was either impossible or so unreliable that it would at some point fail and blow up the MOSFETs. I could also wait until the secondary ringing stops, but the interrupter I'm using, which is an Arduino running TMRpcm plays music using a 32KHz carrier wave, which is too quick to let the secondary ring down. I could find a different interrupter but at this point I decided to go with the simpler method of just biasing the VCO input with an audio signal like done here:

https://highvoltageforum.net/index.php?topic=1759.0

With this it works nicely and sounds pretty good, and with an additional class AB amp (as phone headphone output wasn't powerful enough) and increasing the bass on an equalizer app, the bass is more present and causes the arcs to violently jump in sync with the music which adds to the whole look. So now I'm just finishing up the coil and testing reliability as a whole package so I don't need to have an eye on every component in case something blows up, and so far on low power (60V) the driver seems fine. The secondary coil however is heating up quite a bit, softening the PVC tube its built on so much that it starts to become wobbly.

The Question part:

I can't yet test with higher power (since the high voltage MOSFETs are still in the mail), but with higher power, the secondary will definitely fall apart. I'm currently using 0.1mm wire with ~2200 turns and I now see that's way too thin, but my secondary coil has only around 25cm of winding area, and I would like to use at most 0.2mm wire so I can have a decent number of turns, which I'm pretty sure will still heat up quite a lot. So can I water cool the secondary by filling the inside of the PVC tube with water and circulating it through a radiator? Either way, I have to create a new secondary so sealing it at the bottom so I can fill it with water for experimenting wouldn't be that hard.

Thanks for all the help so far, the Tesla Coil is really looking good now.

[Zbig]

Water will cause some side efects based on water conductivity:
1) extra capacity to secondary coil
2) extra loss  by eddy currents and capacitance currents
3) possible insulation break to the water   

So in my opinion it is not good idea.

[AFreshLad]

Would using de-ionised water reduce its effect on capacitance and the eddy currents. Or could I use something like mineral oil or the stuff used for cooling transformers?

[Zbig]

Would using de-ionised water reduce its effect on capacitance and the eddy currents. Or could I use something like mineral oil or the stuff used for cooling transformers?

If any, try oil. But if secondary is on thick insulator, it will also reduce heat transfer to oil.  Transformer windings are immersed in oil.

[AFreshLad]

But if secondary is on thick insulator, it will also reduce heat transfer to oil.  Transformer windings are immersed in oil.

I think I might try the oil, but I don't think the insulation on the secondary is a problem because the immediate problem I'm facing right now is that the PVC pipe itself is heating so much that it is flopping around, the secondary coil wire itself doesn't seem to be too bad, but still will require cooling for higher power.

Also, I want to run my coil on rectified 240V AC but don't have a variac to try it out. I do have a transformer that'll halve the voltage and I've previously run my coil on around 120V AC and was able to "soft start" the system using one of these (BTA - 4000W):

https://www.ebay.com.au/itm/173878712854?chn=ps&_ul=AU&mkevt=1&mkcid=28

which I think is basically a beefed-up light dimmer, but this circuit isn't powerful enough to handle the current at 240V AC as it basically just shorts out the triac inside and just outputs the full sine wave. So is there any soft start circuit I could build for the huge filtering capacitors, or will I have to get a variac?
Thanks!

[AFreshLad]

Unfortunately, I haven't been able to try oil cooling my secondary coil yet as I've had to self-isolate (couldn't get oil), but my new MOSFETs arrived in the mail and I ran into a new problem.

The new MOSFETs are the Infineon 60R037P7. I was previously using the Vishay G47N60E MOSFETs. The G47N60E and the 60R037P7 only differ, as far as I can see, in rated drain currents, Vds voltage, and the turn on and off times. However, when I use the 60R037P7 without changing anything else about my tesla coil driver, the whole thing 'glitches' due to a huge amount of noise being introduced into the ground and power lines of the driver, which causes the coil to create even more noise. However using the G47N60E and even a cheap Chinese MOSFET TF68N80 (which has much worse turn on and off times), there is no such noise and the driver functions fine. So it must be a characteristic of the 60R037P7 that is causing the noise but I don't know what it is as my knowledge of MOSFETs is very basic.

Any ideas will be appreciated. Thanks!

[davekni]

Yes, Infineon 60R037P7 looks to be a very fast-switching FET.  It has low Crss (Miller capacitance), so will have rapid Vds rise and fall times.  That is generally great for low losses, but can cause problems, especially with GDT gate drive.  I've had two such projects with fast FET H-bridges fail for this issue, where the fast rise and fall times and GDT inter-winding capacitance cause high-frequency oscillation at each transition.

Several design details can help mitigate this issue:
1) Clean circuit layout with ground planes, for both driver and H-Bridge.  In case you haven't seen it, here's my recommendation for bridge layouts (half bridge initially, full bridge in second post):
https://highvoltageforum.net/index.php?topic=1324.msg9795#msg9795
2) Minimize GDT turns and twist lead wires.  In case it's helpful, my GDT tutorial so far:
https://highvoltageforum.net/index.php?topic=1854.msg13949#msg13949
3) Add common-mode chokes to GDT leads, especially the high-side FET output(s).  One high-side FET for a half-bridge, two for a full-bridge.  Make these GDT output leads longer and wind several turns around another ferrite core.  Two separate cores for a full-bridge, one for each high-side FET.
4) Add gate series resistors if you don't already have such.  May not help all that much given the low Crss of these FETs.

If the above is not sufficient, then pick slower FETs or change to fast IGBTs.  Or, with a bit more complexity, change to an isolated high-side gate driver IC instead of GDT.