Driverless Single Mosfet Relay interrupted SSTC? - [All kinds of problems]

[TiagoBS]

Hey guys, at the beginning of the year I finished my first half bridge sstc. The result was amazing thanks to your help!

Now I'm looking for a new project and I came across this video:

https://youtu.be/DpM4hGyOrm4


The proposed circuit is this one, much seen and used on the internet. However unlike the ones seen on the internet (eg Labcoatz Single Mosfet SSTC) it does not use a ballast.


I also found interesting the interrupter he implemented from a relay and an arduino. He also showed how the use of an adjustable voltage regulator affects the output format.


40cm sword sparks! I found this result very intriguing and impressive.

Well, driverless sstc with this circuit doesn't seem that complex to build (wrong  ), so I decided to build and see.

I used the same circuit proposed by him (without the snubber), but instead of using an Arduino I used a lm358 board with a 555 to trigger the relay.

Now comes my problem. I'm also not using a ballast and my circuit barely worked with 27vdc 15A from a power source, with that input voltage there was no output on the coil, but the coil lit a fluorescent lamp at close range.
The coil didn't work at 120vac, when I tried to use that voltage I blew the 5A fuse and damage my Mosfet.

Here is some pictures of my circuit and the respective component values.

In gray the Snubber that I'm thinking of using as I couldn't measure the inductance of the primary coil to calculate the ideal value.



The secondary coil has a frequency of 726kHz according to JavaTc. Approximately 650 turns of 32 awg wire. 150mm in length in a 75mm diameter PVC pipe. The primary coil has 4 turns and the Coupling Coefficient stood at 0.359.


BTW. I now have a cheap oscilloscope which should be enough to get some information about the circuit's operation and troubleshooting!
I just need to learn where to take measurements. 
When measuring the voltage divider with the potentiometer at 50k the oscilloscope showed 9.8v at 120vac.

Some guides I'm using:
Labcoatz single mosfet sstc
https://www.instructables.com/Simplest-POWERFUL-Solid-State-Tesla-Coil-SSTC/
And the teslista555 guide for single mosfet sstc with 31vdc input.
https://www.vn-experimenty.eu/teslov-transformator/sstc/mini-sstc.html

Questions:
Does it make sense to implement this type of interrupter?
Is it possible for this circuit to work at 120vac without a ballast?
Should the snubber be used together with 400v tvs?
Are the values ​​of the components used by me close enough?
How do I know if my Mosfet is suitable for the secondary frequency?(always wanted to know this)
What can I improve on my circuit for more reliable operation?(as soon as it works!)
How to measure primary coil inductance?
(I tried using an LCR Meter but it identified the primary as a resistor, is it possible to use a function generator for this?)

Thanks to everyone who can share knowledge and help!

[davekni]

Well, driverless sstc with this circuit doesn't seem that complex to build (wrong  ), so I decided to build and see.

Yes, these class-E circuits are simple in the sense of low part count.  Class-E Tesla coils also tend to be quite finicky.  If not precisely tuned, fried FET (or IGBT in this case) is common.  Most people build these because they look simple.  With lots of trial and error, some manage good results after some number of fried parts being replaced.
Such simple designs can also be a great learning project, especially when paired with a scope and analog simulation (ie. LTSpice or other).  Get simulated and scope waveforms to roughly match at low voltage (where parts are less likely to fry) before increasing voltage & power.
Inductance of lead wires can be significant.  Your build is better in that respect, shorter wires and paired (twisted or ran side-by-side such as to primary coil).

Does it make sense to implement this type of interrupter?

Simple.  Mechanical relay will have limited life, in case that matters.  Notice varying behavior in video based on when relay closes relative to line voltage (phase).

Is it possible for this circuit to work at 120vac without a ballast?

Yes, class-E oscillators can work without ballast.  Ballast can reduce failures due to mis-tuning, but also limits output power.

Should the snubber be used together with 400v tvs?

The "snubber" circuit in the video is really the primary tuning capacitor.  Notice that resistor is removed in that version.  Some class-E circuits use only "parasitic" drain capacitance, but most need additional capacitance to properly tune primary.

Are the values ​​of the components used by me close enough?

Analog simulation (LTSpice or ...) is the way to tell.

How do I know if my Mosfet is suitable for the secondary frequency?(always wanted to know this)

Power and voltage rating are more likely limiting than frequency here.  Counterfeit IRFP460s are very common, often with ~1/2 power capability of genuine parts.  Also note that the video shows changing to a 1200V IGBT with much higher power capability.  Depends on performance you want to achieve.

What can I improve on my circuit for more reliable operation?(as soon as it works!)

Simulate and measure.

How to measure primary coil inductance?

JavaTC is likely more accurate than any cheap meter.  If you want to measure, place a known capacitance in parallel and measure resonant frequency (w/o secondary).

Good luck with your project!

[TiagoBS]

Hey, guys
I did some tests and modifications in the circuit and managed to make it work with an interesting output.
I'm using a 27vdc 15A power supply.


I added a 1nF capacitor and a 22Ω resistor between DS (although I have calculated, with a little doubt, to avoid the spikes above 100v a resistor capacitor set of 8nF and 13Ω). I haven't measured the difference this might have made yet as I was unsure if I can use my oscilloscope probe to measure it (10:1 probe: 800Vpp (±400V)).
I also reduced the primary coil to two turns. I believe this change was the one that had the biggest impact.
For the next tests I'm thinking of using just one turn, but I'm uncertain if the low impedance could cause problems.

I also did simulations in LTspice and EasyEDA, but I'm having a hard time finding equivalent components like MOSFETs, however I was able to simulate the voltage divider and see the waveforms.

The interrupter works with a relay cutting the power supply voltage. The relay is being operated by an lm358 board.
Here I have already identified two problems. First is that when cutting the main voltage the relay is getting stuck on sometimes. The second problem is that the lm358 output is 5v and the relay uses 12v and this is preventing the relay from working properly.
During the tests, including in the gif, there were times when the relay worked fast enough to cause the switch effect on the output.

I'm going to change the circuit to interrupt the signal going to the Mosfet's Gate instead of the main voltage. I believe that this is the correct way and that I should have done it from the beginning.
To get around the problem of the malfunctioning relay I will try to use an L298N to activate it with 12v.

I intend to implement a switch using a 555, but I'm enjoying implementing one thing at a time, even if in a more archaic way because I'm starting to understand things better.
I'm also tending to use 30vdc on this coil instead of 120ac. A low voltage interrupted coil looks nice, maybe an irfz44n half bridge in the future.

Questions:
-Is there a relationship between the secondary diameter and the input voltage? For example, my secondary is 75mm in diameter, for a coil running at low voltage, would a smaller diameter coil be better?
-Would it be possible to use the lm358 to operate the Mosfet Gate directly? (Or maybe the GDT?)


This is the circuit of a low voltage coil proposed by Teslista555. I found it very interesting and I was left with a few questions:
-Why two diodes d1 and d3?
-Why was a resistor R3 used before the Gate?
-C2 is a resonant capacitor?
-Why is the circuit connected to the ground?

Thanks in advance for the help!

[davekni]

-Why two diodes d1 and d3?

Presume "d1 and d2".  Probably because he didn't have single diodes rated for high enough current.

-Why was a resistor R3 used before the Gate?

To adjust gate voltage so that FET draws just enough current to start oscillation, but not enough to burn out.

-C2 is a resonant capacitor?

Yes.  Primary resonant current flows through C1 as well, so both C1 and C2 need to be capable of handling primary resonant current.

-Why is the circuit connected to the ground?

Because secondary current needs to return to ground.  Most of the top-load capacitance is to ground.  Bottom needs to return to ground to complete secondary resonant circuit.

Not sure I understand the other questions.

[davekni]

I also did simulations in LTspice and EasyEDA, but I'm having a hard time finding equivalent components like MOSFETs, however I was able to simulate the voltage divider and see the waveforms.

Try a free program called "LTSpice_MOStool.exe".  I use it both at home and work.  Enter data-sheet parameters and it generates an LTSpice VDmos model.  A number of the parameters will require reading from data-sheet graphs, such as capacitances at different voltages and diode forward drop at different currents.  Fourth file down in this link is the executable to download:
https://groups.io/g/LTspice/files/z_yahoo/Util/Model%20Tools/Board%20Level%20MOSFET%20%28VDmos%29/Software

[TiagoBS]

Hey, guys!
Updates:
I'm learning a lot with this new project, every time I make a change I have a lot of questions.

The changes I made were as follows:



I added a 10k resistor before the Gate, changed the Snubber to 15nF 10ohm and primary with 2 turns.

I tested several combinations and also tested some calculations (I'm basing it on Java TC's primary coil inductance value) and came to some conclusions (Although not sure). The capacitance needed to limit voltage spikes increases 4 times when the desired voltage is divided by 2. For example, by limiting the spikes to 180v my calculations showed a value of 2.4nF. Reducing from 180v to 90v the required capacitance is 9.8nF.
I'm just using these calculations as an estimate of values ​​because I'm not sure about the values ​​of the variables used.
[This conclusion may be too clear for more experienced coilers, but I'm learning on the fly in this project  ]

In the tests performed I used 120vac mains voltage with a ballast instead of 27vdc as I was using.

I noticed that the output seems to be weaker, with thiner discharges.

As I am interrupting the circuit from a relay commanded by a 555 circuit I tried to use a 1000uF capacitor after the diode. In conjunction with this change I also started to interrupt the voltage going to the Mosfet Gate instead of the mains voltage. However, when doing this, the output was very weak, almost non-existent. I also noticed that with this change the coil now seems to work continuously with stronger pulses when the relay is activated (When a lamp is close, it remains lit and pulses with stronger luminosity when the relay is activated).

Questions:
I tried to add one more lamp in series with the one I used as a ballast and I didn't notice any difference in the result. Shouldn't the circuit show a difference in the output?

Can I wind up a transformer and use it as a ballast?
What measurements determine ballast in relation to output power? (Amps consumed in operation by ballast?)

I also did simulations in LTspice and EasyEDA, but I'm having a hard time finding equivalent components like MOSFETs, however I was able to simulate the voltage divider and see the waveforms.

Try a free program called "LTSpice_MOStool.exe".  I use it both at home and work.  Enter data-sheet parameters and it generates an LTSpice VDmos model.  A number of the parameters will require reading from data-sheet graphs, such as capacitances at different voltages and diode forward drop at different currents.  Fourth file down in this link is the executable to download:
https://groups.io/g/LTspice/files/z_yahoo/Util/Model%20Tools/Board%20Level%20MOSFET%20%28VDmos%29/Software

Thank you very much for the tip, I will test this software!

[davekni]

I added a 10k resistor before the Gate, changed the Snubber to 15nF 10ohm and primary with 2 turns.

10k gate resistor completely changes circuit operation.  Provides DC gate voltage, but no significant AC current flows through 10k.  The AC component of gate waveform is now only from gate-drain capacitance.  Thus feedback is from primary voltage, not secondary.
10ohm resistor in snubber circuit will dissipate most of the circuits power, leaving little for the coil.  As I said previously, class-E coils want a resonant capacitor here, not a snubber.  Best with no resistance.  Snubbers have uses.  This is the wrong circuit to use a snubber (R + C).

The capacitance needed to limit voltage spikes increases 4 times when the desired voltage is divided by 2. For example, by limiting the spikes to 180v my calculations showed a value of 2.4nF. Reducing from 180v to 90v the required capacitance is 9.8nF.

This is accurate, at least for one of the most common uses of snubbers.  That is for transformer coupled boost converters (non-resonant flyback converters).  Transformer leakage inductance is constant, as is max load current (current through leakage inductance).  To reduce spike voltage to 1/2 impedance must drop to 1/2.  For fixed inductance, this requires frequency to be 1/2 (spike width to double), requiring 4x capacitance.

Another way to look at class-E circuits such as this:  The drain (collector) waveform consists of only relatively-wide "spikes" that form the resonant waveform.  These "spikes" are the top part of a sine-wave.  The remainder of the sine wave is clamped to ~0V by FET/IGBT conduction.  If the "spike" voltage is too high, then the resonant capacitor (the drain-source or collector-emitter capacitor) needs to be increased to widen the "spike" and thus lower its peak voltage.  Or, operating current needs to be reduced to reduce spike voltage.

Can I wind up a transformer and use it as a ballast?

What you want to wind is an inductor, not a transformer.  A transformer core can be used, but an air gap should be added into the magnetic path.

What measurements determine ballast in relation to output power? (Amps consumed in operation by ballast?)

Simulation is a good way to experiment.  Use an inductor as ballast and try different inductance values.

Have fun learning!

[TiagoBS]

Good news!


-


I'm finally getting to an interesting result with this circuit!

I replaced the electromechanical relay with a solid state relay, and using 120vac with an adjustable voltage regulator I was able to change the shape and intensity of the discharges. Sword shape discharges with muffled sound!

When I'm measuring DS with my oscilloscope I'm not able to use the trigger function correctly I think, but I was able to identify times when the vmax was 198v (the top part of a sine-wave?).

Another way to look at class-E circuits such as this:  The drain (collector) waveform consists of only relatively-wide "spikes" that form the resonant waveform.  These "spikes" are the top part of a sine-wave.  The remainder of the sine wave is clamped to ~0V by FET/IGBT conduction.  If the "spike" voltage is too high, then the resonant capacitor (the drain-source or collector-emitter capacitor) needs to be increased to widen the "spike" and thus lower its peak voltage.  Or, operating current needs to be reduced to reduce spike voltage.

I will test larger capacitor values between DS. But I need to figure out a way to be able to see the still waveform on the oscilloscope screen.

By accident I discovered that the circuit is working without the ballast. In one of the tests, with a new transistor of a different brand I forgot to connect the ballast and the circuit worked at full mains power!
Although I haven't done more tests or used the relay to interrupt the signal because I think I have to optimize the values ​​of all components before that.

I also changed the primary from 2 to 3 turns and this greatly increased the size of the output, at the moment using 2 100w lamps in parallel I'm getting 10-11cm discharges.

Now I believe the next step would be to optimize the capacitor between DS.

Questions:
Why increasing the number of turns on the primary from 2 to 3 did the output increase and why increasing the height of the primary does the opposite?
I'm using two 100w lamps in parallel as a ballast, how do I calculate the value of an equivalent inductor?
For the solid state relay to work I had to connect a lamp to its output, along with the sstc circuit, is there any alternative to this problem?
So that I don't have to leave a light bulb plugged into it.

[davekni]

I will test larger capacitor values between DS. But I need to figure out a way to be able to see the still waveform on the oscilloscope screen.

Looks like you have a digital oscilloscope.  It likely has an acquisition mode for single-capture.  That mode stops acquisition after the first trigger event it sees.  Result will be stable.  Manually (scope menu/buttons) enable acquisition to make the next capture.

Why increasing the number of turns on the primary from 2 to 3 did the output increase and why increasing the height of the primary does the opposite

For my class-E coils (6.78MHz and 13.58MHz), required iterations of simulation and scope measurement to figure out optimum values.

I'm using two 100w lamps in parallel as a ballast, how do I calculate the value of an equivalent inductor?

No exact answer.  Incandescent lamp resistance changes about 15:1 between cold and hot (full voltage to lamp).  Exact resistance for your use is unknown, and changing some even during your half-cycles of enable.  First order:  Start with nominal lamp resistance at full power.  Calculate inductance that has same impedance as lamp at your line frequency (50 or 60Hz).  Use that inductance.

For the solid state relay to work I had to connect a lamp to its output, along with the sstc circuit, is there any alternative to this problem?

AC SSRs (solid state relays) generally use TRIACs internally, with gate pulses shortly after line voltage zero-crossings.  TRIACs need load current to remain enabled for the remainder of each line half-cycle.  Your class-E circuit must not be drawing sufficient current initially to keep TRIAC enabled.  I've used lamps for this purpose too.  Depending on the SSR, a small lamp of 5-10W may be sufficient.

[TiagoBS]

I added two more lamps in parallel, making a total of 4.
The result is already beyond what I could have expected!



I used a FGA60N65SMD in place of the IRFP460.
With that inductance as a ballast I was able to use the adjustable voltage regulator and get swords like sparks.

I was also able to connect the circuit to the mains voltage without ballasts and use the interrupter. However, the IGBT burned out when I tried to go to the limit of the adjustable voltage regulator (that was possible using the 4 lamps as ballast).
The last spark.




Some changes I will make:
Add a new toroid.
Increase the primary coil from 3 to 4 turns.
Perhaps further increase the capacitance of C between Collector and Emitter*.
I will also redo the entire circuit in a smaller package to reduce unwanted inductance.

Questions:
The first one is related to the capacitance between Collector and Emitter.

Here is the data related to my primary and secondary set.



I don't know if I'm doing the calculations correctly.
My secondary's frequency is around 649kHz and the primary's inductance is around 1,387 microhenry. I believe that the capacitance value to be considered is in relation to the capacitor between Collector and Emitter, right?
So, using a higher value for C would be bringing the primary impedance down, in this case, when using 68nF the impedance would be close to 2ohms, which I believe is bad for the IGBT, right?
Is there any way to calculate the primary impedance "ideal" range in relation to the IGBT/MOSFET used, or is it just from experimentation?

Second question.
I don't intend to use a bunch of light bulbs in my circuit. So I'm looking for an inductor to replace the bulbs.
So I performed the measurement as suggested to have a reference of an inductor that would give me an approximate result.

I'm using two 100w lamps in parallel as a ballast, how do I calculate the value of an equivalent inductor?

No exact answer.  Incandescent lamp resistance changes about 15:1 between cold and hot (full voltage to lamp).  Exact resistance for your use is unknown, and changing some even during your half-cycles of enable.  First order:  Start with nominal lamp resistance at full power.  Calculate inductance that has same impedance as lamp at your line frequency (50 or 60Hz).  Use that inductance.

And from the Inductor Impedance formula (XL= 2πfL) from this post:
https://highvoltageforum.net/index.php?topic=1739.0
I got the following result:

Current of the lamps in AC and DC

4 lamps in AC - 3.2A
4 lamps in DC (Rectified with 1 diode 10A10) - 1.7A

Z = U/I
Z= 120V / 3.2A (Current of 4 lamps in parallel)
Z= 37.5 Ohms

Inductor Impedance
Zl = 2*PI*f*L

L = Zl/(2*PI*f)
L = 37.5 Ohms / (2*3.14*60Hz)
L = 37.5 / 376.8
L = 0.09952229
L = 99.5 millihenry
or
L = 185 millihenry, since the current of the 4 lamps in DC (After a single 10A10 diode) is close to 1.6A.

Are the calculations correct?

Third question:
This was the waveform I was able to measure between Collector and Emitter. I increased the value of C, however I didn't notice a difference in the format or in the maximum voltage value. What is this ringing voltage at the end of the wave?
Is this elevation at the beginning the ramp that allows sword like output?

[davekni]

So, using a higher value for C would be bringing the primary impedance down, in this case, when using 68nF the impedance would be close to 2ohms, which I believe is bad for the IGBT, right?

Changing capacitance across IGBT changes both impedance and primary resonant frequency.  Frequency is defined by where impedance of L and C match.  Your calculations show mismatched impedances, so not at resonance.  Normal DRSSTCs usually start with primary frequency below secondary.  Many QCW coils start with primary frequency slightly above secondary to force initial upper-pole operation.  (Research coupled resonant circuits here or other internet places.  TC primary and secondary circuits are two coupled resonators.)

Is there any way to calculate the primary impedance "ideal" range in relation to the IGBT/MOSFET used, or is it just from experimentation?

Yes, 2 ohms is probably too low for your circuit.  For this class-E circuit, impedance is roughly peak Vce / peak Ice.
BTW, added Cce is in parallel with IGBT's internal Cce.  Use total for resonant calculations.

L = 99.5 millihenry
or
L = 185 millihenry, since the current of the 4 lamps in DC (After a single 10A10 diode) is close to 1.6A.

Are the calculations correct?

99.5mH looks correct.  The 1.6A value appears to be RMS current for half-wave rectified lamp drive.  That isn't relevant to your situation.  You want to control peak current.

This was the waveform I was able to measure between Collector and Emitter. I increased the value of C, however I didn't notice a difference in the format or in the maximum voltage value. What is this ringing voltage at the end of the wave?
Is this elevation at the beginning the ramp that allows sword like output?

At this low sample rate, all the 600kHz oscillation details are lost.  The initial plateau is probably where IGBT starts conducting but hasn't reached sufficient current to start oscillation.  IGBT may be dissipating high power during this plateau, so may be part of causing it to fail at higher voltage & current.  Doubt the plateau has any direct relation to sword sparks.

[TiagoBS]

So, using a higher value for C would be bringing the primary impedance down, in this case, when using 68nF the impedance would be close to 2ohms, which I believe is bad for the IGBT, right?

Changing capacitance across IGBT changes both impedance and primary resonant frequency.  Frequency is defined by where impedance of L and C match.  Your calculations show mismatched impedances, so not at resonance.  Normal DRSSTCs usually start with primary frequency below secondary.  Many QCW coils start with primary frequency slightly above secondary to force initial upper-pole operation.  (Research coupled resonant circuits here or other internet places.  TC primary and secondary circuits are two coupled resonators.)

Is there any way to calculate the primary impedance "ideal" range in relation to the IGBT/MOSFET used, or is it just from experimentation?

Yes, 2 ohms is probably too low for your circuit.  For this class-E circuit, impedance is roughly peak Vce / peak Ice.
BTW, added Cce is in parallel with IGBT's internal Cce.  Use total for resonant calculations.

Well, I redid the calculations and with 4 turns and 25nF Cce the frequency of Primary and Secondary is very close. However, the impedance gets very low, which I believe is the principle behind DRSSTC.
If I use my circuit with this configuration, wouldn't the low impedance cause problems?



When changing the primary to 5 turns and Cce to 12nF (JavaTC recommends ~15nF to match the frequencies) the impedance is close to 4.8Ω. The closer to 15nF, the lower the impedance.



Can ClassE circuits work with such low impedances in the primary?
Or would the best strategy be to find a combination of Cce and primary turns to have impedance not too low (>5Ω<10Ω?A) and the frequency of the primary and secondary set as close as possible?

For this class-E circuit, impedance is roughly peak Vce / peak Ice.

Could you explain that part please?

[davekni]

When changing the primary to 5 turns and Cce to 12nF (JavaTC recommends ~15nF to match the frequencies) the impedance is close to 4.8Ω. The closer to 15nF, the lower the impedance.

I think you would benefit from some experimenting with L/C resonant circuits, with a scope and signal generator and real parts and/or in analog simulation (LTSpice or other free simulators).  And perhaps some general reading and studying.
An L and C in series with no parasitic resistance will go to zero impedance at resonance.  Your class-E circuit is driving the L/C circuit in parallel.  IGBT is in parallel with C and L.  Parallel impedance of L/C circuit with no parasitics goes to infinite at resonance.

Quote from: davekni on October 05, 2022, 06:15:46 AM

    For this class-E circuit, impedance is roughly peak Vce / peak Ice.

Could you explain that part please?

The impedance I'm referring to is the impedance of L or C individually, 15.7 or 20.4 ohms in your last analysis.  Not the 4.8ohm difference between them (series resonant impedance).

[TiagoBS]

Hi, guys!
Updates on this project:

I made a lot of progress in building the case, the final structure is almost ready, I also made a new smaller board and changed the value of some components.
-

-


Now I can see the waveforms with my new 2-channel oscilloscope!
I would also like to better understand what I'm seeing here, especially in the Drain wave. I'm suspecting that my Coupling Coefficient is too high, 0.346 and that is causing these distortions in the Drain voltage
-

-


The coil is running at 479kHz, pretty close to what JavaTC calculated, 495kHz.

I studied a little the concept of Parallel impedance of L/C. Clearly I was going the wrong way. Now with the new adjusted values I came to this conclusion:

I decided to leave the primary with 3 turns, so the impedance was close to 5ohm, which I believe is within an acceptable value. JavaTC recommends 59nF for the resonant capacitor, however I am using a 68nF capacitor plus the DS MOSFET/IGBT capacitance. According to JavaTC this leaves me about 8% detuned. In order for the system to be tuned I am adjusting the height of the primary and observing the waveforms, although I don't quite understand what they say (Raising the height of the primary increases the necessary capacitance)
-

-
 [ Invalid Attachment ]

The problems I have now and how I think I can solve it:
Problem 1: The tric BTA12-600B adjustable voltage regulator does not seem to be handling the circuit power.
Solution 1: Use a BTA16.
Problem 2: I am using a Chinese square wave generator module. This board has frequency adjustment and Duty Cycle, however I had to change the Trimpots by two potentiometers. The board worked after this change, but it was complicated to make the adjustments and after a while of use it simply stopped working correctly.
Solution 2: I'm going to build my 555-Timer to the specifications I need.

What am I going to do differently in version 2.0 of this project.
What pleased me most about this project was the size of the case, only 10cm X 10cm!
-


Inside I have:
1 - Solid state relay
1 - 5cm x 5cm heat sink
1 - blower fan
1 - DC-DC Step Down
1 - Adjustable voltage regulator
1 - SSTC Board 5cm x 5cm
1 - 555-Timer Square Wave Generator Board
1 - Neon Lamp. (To Load the SSR)
-


However, I believe I can reduce the package even further.
With a 7cm x 7cm board I believe I can fit all the SSTC components, Adjustable Voltage Regulator, 555-Timer and DC-DC Step Down, besides I think I can reduce the size of the heat sink.

I'm still having problems with ballast, but I'm going to fix the problems I already have. I want a reliable coil that I can use without much chance of losing my IGBTs

[davekni]

I would also like to better understand what I'm seeing here, especially in the Drain wave.

I suspect the drain waveform ring is largely resonance of 68nF capacitor with wiring inductance from capacitor to IGBT.  For class-E, best to have resonant capacitor directly across collector and emitter (or drain and source for FET) with very-short lead length.  Or low-inductance parallel planes for interconnect.  Not specific to class-E, but I'd recommend taking a look at my post for low-inductance half-bridge construction:
https://highvoltageforum.net/index.php?topic=1324.msg9795#msg9795
Are you back to FET now?  Thought you'd changed to FGA60N65SMD IGBT.
Once you reduce inductance from IGBT/FET to 68nF resonant capacitor, capacitance value is likely to need changing.
The other issue with drain waveform is that drain isn't back close to 0V when gate voltage rises above threshold.  Turning on class-E IGBT/FET when drain voltage is high causes excess power dissipation.

In order for the system to be tuned I am adjusting the height of the primary and observing the waveforms, although I don't quite understand what they say (Raising the height of the primary increases the necessary capacitance)

Dual resonant systems have upper and lower resonant poles (frequencies).  Higher coupling splits those two frequencies farther apart.  Change in needed capacitance depends on whether operation is at upper or lower pole.  Not sure which pole your coil may be locking to.

1 - Neon Lamp. (To Load the SSR)

I'd be amazed if a neon lamp made any significant contribution to SSR load.  Generally requires an incandescent lamp of at least 5 watts.  You may be lucky enough that your coil circuit is drawing sufficient power at low voltage to keep SSR operating properly.

[TiagoBS]

I made some changes to the project.
Now all the components are in a PCP that I designed and almost everything works just as I planned.

-

-

-

-

Are you back to FET now?  Thought you'd changed to FGA60N65SMD IGBT.

I expressed myself badly, I'm using FGA60N65SMD and I have some K40T1202.

I'd be amazed if a neon lamp made any significant contribution to SSR load.  Generally requires an incandescent lamp of at least 5 watts.  You may be lucky enough that your coil circuit is drawing sufficient power at low voltage to keep SSR operating properly.

It's true, although the system works with a neon lamp, I noticed that this altered the functioning of the coil, even with two lamps in parallel, so for now I will continue using an incandescent lamp.

Regarding the waveforms, I think there was an improvement in the collector wave, but I can't say what those ripples are at the end.

I still haven't managed to make the transistor work in Zero voltage switching but I shouldn't be too far away, because the transistor heats up very little. The heat sink and cooler fan are more than enough to keep the transistor at low temperatures.


-

-


As suggested I did all the simulations beforehand and made sure everything was correct, then I planned the board and my 555 switch. Now I can vary Duty Cycle and frequency independently. Then I used a 10k potentiometer to regulate the pulse time (about 17ms). The frequency now ranges from about one pulse every four seconds to hundreds per second.

Now the last problems I need to solve to focus on designing the case and finalizing this project!

I still need a ballast to regulate and prevent my transistors from being damaged. The coil works without a ballast, but I can't go to the maximum of the voltage regulator because the IGBT burns out.
I'm using several incandescent bulbs in parallel, and I wanted to reduce this package to just one component, so I tried using one of those inductors that are inside fluorescent tube ballasts, like this one.



But it didn't make any difference, with ferrite core, with half or without the core, it looks like the coil was working at full power (and burning my transistors). On my last test however, it got hot and started to smoke.
Measuring without the core this inductor has 0.20mH and 1.37ohms.
What am I doing wrong here, shouldn't this be reducing coil power?
What can I use in place of light bulbs?
Or what can I do to connect the coil directly to the mains voltage but reduce the chance of the transistor burning? (reduce coupling???)

And what can I use instead of a light bulb as a SSR load?

[davekni]

Regarding the waveforms, I think there was an improvement in the collector wave, but I can't say what those ripples are at the end.

Probably due to IGBT lead inductance.  Even though now on an ECB, IGBT leads are quite long, and have added screw terminal connection.  Usually IGBTs are soldered to ECBs with the shoulder where IGBT leads thin down at ECB surface.  That is a key reason for the shoulder - to define how far IGBT (or other through-hole parts) insert into ECB holes.  BTW, what is on the back layer of ECB?  Or, is this single-sided?
It is possible that the lead length is contributing to IGBT failure without ballast, perhaps combined with non-ZVS conditions.  (Notice that ring is minimized when running ZVS in third scope capture.)

I still haven't managed to make the transistor work in Zero voltage switching but I shouldn't be too far away, because the transistor heats up very little.

Your third scope capture is ZVS with plenty of margin.  What did you change from first two scope captures to third?

The heat sink and cooler fan are more than enough to keep the transistor at low temperatures.

Low external IGBT temperature indicates average power dissipation is low.  Momentary peak internal die temperature will be higher, and perhaps too high when running without ballast.

Measuring without the core this inductor has 0.20mH and 1.37ohms.
What am I doing wrong here, shouldn't this be reducing coil power?

Simulate your ballast.  Run a line frequency simulation with a simple resistor load instead of TC circuit.  See what inductance is needed to limit current.  Even without simulation, you can get close by calculating inductor impedance at line frequency.

Or what can I do to connect the coil directly to the mains voltage but reduce the chance of the transistor burning?

First need to figure out why IGBTs are burning out.  Could be peak temperature internally.  Perhaps at high power operation is getting farther from ZVS.  Or perhaps at some larger arc conditions Q gets too low and oscillation stops.  (Would lose ZVS before oscillation halted.)  Or perhaps peak IGBT voltage gets too high, though that seems less likely, unless arc changes are causing a large shift in operating conditions.

And what can I use instead of a light bulb as a SSR load?

A resistor.  You will need to experiment with your particular SSR to see how high resistance will work (how low power dissipation is achievable for the resistor).  Probably SSR will need lower resistance (more load) when cold, so leave some margin when selecting resistance.

[TiagoBS]

I reached the 'end' of this project.

Since my first post in this thread it's been 209 days. I learned a lot of things, Tesla Coils are fascinating puzzles, solving the problems and understanding the principles used without having an electronics background was quite challenging!
I would like to thank davekni for all the help he gave me to complete this coil!

-The last changes were:
-A 1.8k resistor in place of the lamp to load the SSR.
-New Case to accommodate the PCB

Here's a picture of the final waveform (Although I'm still not sure if it's "ideal")
What does this purple marked overlay mean?
Would it be ideal not to have this overlap?
-



And some pictures of the complete build.
-

-

-

-

-

-

-

-

[davekni]

I would like to thank davekni for all the help he gave me to complete this coil!

You're certainly welcome.  Very nice coil construction.  Better mechanical build than anything I make.

Here's a picture of the final waveform (Although I'm still not sure if it's "ideal")
What does this purple marked overlay mean?
Would it be ideal not to have this overlap?

If you are interested, look up "Miller plateau".  When drain voltage starts rising, Cgd current is countering gate drive current, creating that pause in Vgs falling voltage slope.  (I'm presuming cyan trace is Vgs and yellow is Vds.  Labels in text are helpful.)
Waveforms look great.  Miller plateau does result in some power dissipation in FET.  Only way to reduce that is to provide more gate current (lower impedance gate drive).  Likely not possible without added circuit complexity (ie gate drive buffer of some sort).

Have fun with your coil!