There seems to be a problem with my12MHz HFSSTC

[曹靖]

Recently, I have encountered many problems using the newly purchased 75pF vacuum capacitor to make my 12MHz HFSSTC
I use an adjustable output voltage switch power supply to power him so that I can debug HFSSTC
But when I first started supplying power to HFSSTC, I found that the output voltage of the power supply could only be adjusted to 12V. Even if I continued to turn the voltage regulating knob, the voltage would not increase. Sometimes, it was accompanied by the output short-circuit protection action of the power supply
Afterwards, I thought it might be due to some harmonics between my HFSSTC and the power supply affecting the power supply voltage feedback network. So, I added a 20uH hollow inductor on the output side of the power supply to prevent the harmonic flow of HFSSTC from returning to the power supply voltage feedback network
The problem seems to have been solved. This time, the voltage can be increased to 80V~90V, but HFSDTC can only operate stably for a long time at 80V. After working at 90V for 30 seconds, the HFSSTC leakage waveform begins to shake violently, and the output voltage of the power supply is also unstable. The instability phenomenon mentioned earlier has also appeared
So far, this problem has been bothering me~I think there are many smart people here who may be able to see the root cause of the problem~Providing constructive suggestions to help me solve the problem (இ) д இ; )

[曹靖]

This red is the drain waveform of the FET~yellow is the gate
At the beginning of the circuit, I found that the Zener diode was reversed. I made a new sketch by hand. This time, there's no problem

[曹靖]

The digital display on the left shows the red voltage and blue current of the adjustable power switch source, which is the problem encountered above. Currently, the switch power supply can only output around 84.3V  3.94A

[曹靖]

These are the two discharge tips that I polished with a graphite rod

[davekni]

The problem seems to have been solved. This time, the voltage can be increased to 80V~90V, but HFSDTC can only operate stably for a long time at 80V. After working at 90V for 30 seconds, the HFSSTC leakage waveform begins to shake violently, and the output voltage of the power supply is also unstable. The instability phenomenon mentioned earlier has also appeared

Hard part is determining if instability is originating from HFSSTC drawing unstable current or from power supply providing unstable voltage.  Perhaps try more filtering between the two, such as an electrolytic cap across supply output.  My guess is that power supply is still the problem.  You could power from a variac and rectifier and filter caps (linear supply).

This red is the drain waveform of the FET~yellow is the gate

Is Vgs (voltage gate-to-source) waveform inverted or delayed by 180 degrees?  Vgs should be dropping as Vds rises.

These are the two discharge tips that I polished with a graphite rod

As you have likely read already, my 13.56MHz HFSSTC uses similar graphite rod tips, but sharp points burns to rounded at high power.

Nice clean mechanical construction!  Good luck with debug.  Nice to see another fixed-frequency HFSSTC.  Only other one I've seen besides mine.

BTW, it is easier to have single posts rather than a string of posts.  If you have more information to add, there is a "Modify" button just above your post towards the right side.  Click on "Modify".  That allows adding more information.  I do that very frequently.  (There is another "Modify" button below the post.  It does not work.  I have no idea why another non-functioning button exists.  Confused me for a long time.)

[曹靖]

The Vgs waveform is the reverse phase of my oscilloscope channel setting. So far, I can only add more filtering networks to solve the problem. I have a mechanical voltage regulator with a voltage range of 0-300V, which is linear. However, I don't have enough filtering capacitors in my hand to power it. Before this project, I worked on an 8MHz fixed frequency HFSSTC and never encountered this problem. It seems to work well under my switching power supply I don't need to add additional filtering network power supply, which can output full voltage. I am very confused about this (my power supply is made by myself, maybe this is also a problem. It doesn't have an aluminum box that is exposed. The input is 220V, 50Hz, and the output is 0-150V, adjustable 0-15A.) I registered for this forum a few days ago and I am not familiar with the operation. Thank you for helping me raise my operational issues

[曹靖]

I still suspect it's the problem with HFSSTC because I connected the drain and heat sink together. The parasitic capacitance and other parasitic parameters caused by the large heat sink and the copper layer covering the copper plate may affect the normal operation of the machine. I quickly corroded a new copper plate with an etching agent. I hope to verify my hypothesis

[davekni]

The Vgs waveform is the reverse phase of my oscilloscope channel setting.

By "reverse phase" do you mean inverted polarity (ie. negative of actual waveform)?  If so, does the actual waveform go from +6V to -15V (instead of the shown -6V to +15V)?  That wouldn't be consistent with SiC FET Vgs needs nor with zener values in your circuit (which look appropriate at +18V and -8.2V plus diode forward drop).

Before this project, I worked on an 8MHz fixed frequency HFSSTC and never encountered this problem. It seems to work well under my switching power supply I don't need to add additional filtering network power supply, which can output full voltage.

Great performance of your 8MHz coil!  It is possible that some resonance in supply or cabling happens to resonate close to 12MHz, so is sensitive to your new HFSSTC and not to previous coil.  Also possible as you think that instability is in HFSSTC itself.  Could even be an interaction between the two.  Arc impedance can have strange behavior.  Perhaps the combination of arc impedance and HFSSTC circuitry creates a negative resistance for DC input power (current decreases as voltage increases), causing supply regulation loop to go unstable.  Even if not negative resistance, reactive load impedance might be causing instability in supply regulation loop.

I still suspect it's the problem with HFSSTC because I connected the drain and heat sink together. The parasitic capacitance and other parasitic parameters caused by the large heat sink and the copper layer covering the copper plate may affect the normal operation of the machine.

My 13.56MHz HFSSTC also uses a live heatsink (SiC FET mounted directly to heatsink).  Thermal impedance is significantly lower that way.  I did need to add some grounded copper foil over the fan motor to keep its control electronics happy.  Fan is ~2mm from heatsink in my build.  Of course, your geometry is different, so 12MHz may be coupling to some control circuitry and causing issues for your build.

Good luck!  Looking forward to updates.

[曹靖]

There is a 180 degree phase difference, and when I have an oscilloscope to capture it, some of it is in advance, while others are not in advance, I have set up a new circuit and captured a new waveform. The new circuit has verified my hypothesis. The previous circuit was connected to the drain and heat sink, and the driver board is on the heat sink. They interfere with each other. Perhaps I can use the previous circuit drain to connect the heat sink to reduce thermal resistance. At the same time, I have taken shielding measures for the driver and continue the experiment

[davekni]

There is a 180 degree phase difference, and when I have an oscilloscope to capture it, some of it is in advance, while others are not in advance, I have set up a new circuit and captured a new waveform

OK, I think I understand.  Vgs and Vds were from different scope captures (captured at different times).  New capture looks appropriate.

The new circuit has verified my hypothesis. The previous circuit was connected to the drain and heat sink, and the driver board is on the heat sink. They interfere with each other. Perhaps I can use the previous circuit drain to connect the heat sink to reduce thermal resistance. At the same time, I have taken shielding measures for the driver and continue the experiment

This makes more sense now too.  I use live heat sinks (drain directly to sink), but never share that live heat sink with any other circuitry.  If any drive circuitry needs heat sinking, I'd suggest a separate heat sink for driver, ground that heat sink, and keep that sink away from live one.  (For my drivers, any needed heat sinking is usually copper area on circuit board, or occasionally TO220 devices mounted through insulating layer to grounded aluminum housing.)

[曹靖]

There is a 180 degree phase difference, and when I have an oscilloscope to capture it, some of it is in advance, while others are not in advance, I have set up a new circuit and captured a new waveform

OK, I think I understand.  Vgs and Vds were from different scope captures (captured at different times).  New capture looks appropriate.

The new circuit has verified my hypothesis. The previous circuit was connected to the drain and heat sink, and the driver board is on the heat sink. They interfere with each other. Perhaps I can use the previous circuit drain to connect the heat sink to reduce thermal resistance. At the same time, I have taken shielding measures for the driver and continue the experiment

This makes more sense now too.  I use live heat sinks (drain directly to sink), but never share that live heat sink with any other circuitry.  If any drive circuitry needs heat sinking, I'd suggest a separate heat sink for driver, ground that heat sink, and keep that sink away from live one.  (For my drivers, any needed heat sinking is usually copper area on circuit board, or occasionally TO220 devices mounted through insulating layer to grounded aluminum housing.)

   Thank you to David for providing the method. I have debugged the Class E part again and captured a waveform. The red Vds and yellow Vgs are the same as last time. As per your suggestion, I have grounded the cooling fin of the TO220 device in the driver circuit. However, the ground wire on my end is not very easy to find, I can only connect the heat sink to the negative pole of the driver power input. This time, I moved the driver board away from the live heat sink in the main circuit. The interference with the power source has never occurred again. I can continue the experiment again. I need to adjust it to the most perfect state to be satisfied

[davekni]

As per your suggestion, I have grounded the cooling fin of the TO220 device in the driver circuit. However, the ground wire on my end is not very easy to find, I can only connect the heat sink to the negative pole of the driver power input.

That sounds perfect.  Most important aspect is to have driver heatsink tied to local driver ground.  Connection to line ground isn't critical.

Waveforms look great!  Your coil's performance looks good too.

[曹靖]

As per your suggestion, I have grounded the cooling fin of the TO220 device in the driver circuit. However, the ground wire on my end is not very easy to find, I can only connect the heat sink to the negative pole of the driver power input.

That sounds perfect.  Most important aspect is to have driver heatsink tied to local driver ground.  Connection to line ground isn't critical.

Waveforms look great!  Your coil's performance looks good too.

I'm still not satisfied that the current input power is limited by the switching power supply~The input 100V consumes 7.5A, and the arc looks very thick~I want to input 200V~but my switching power supply power is insufficient~I want to wait until I have a large capacity electrolytic capacitor in my hand, and then I can use the mechanical voltage regulator to power it~In addition, I add several arc snapshots after the tip is covered by calcium sodium glass sheet~The arc becomes slender and bright yellow due to the increase of sodium ion concentration

[davekni]

Great performance already!  Air in room must be quite still.

The input 100V consumes 7.5A, and the arc looks very thick~I want to input 200V~but my switching power supply power is insufficient

What is your FET's current capability?  Peak current will be much higher than 7.5A DC input current.

[曹靖]

Great performance already!  Air in room must be quite still.

The input 100V consumes 7.5A, and the arc looks very thick~I want to input 200V~but my switching power supply power is insufficient

What is your FET's current capability?  Peak current will be much higher than 7.5A DC input current.

Yes~The air in the room should be calm and slightly disturbed, which can affect the straight arc~At the same time, I found that the current will decrease after dyeing the color arc~while using a graphite rod without dyeing the color arc will be between 7.5A and 8A~My FET withstand voltage and current parameters have margin~It allows me to do so~

[曹靖]

今天我用了一个机械调压器把功率调高到了近2KW，取得了不错的效果。我想和你分享。我没有YouTube帐户，所以我不能在上面发布视频。我不确定你在国内能不能看你的视频网站。我会把视频放在https://b23.tv/4NFtCpP哔哩哔哩平台上。如果有人能在中国的其他视频网站上观看，请告诉我。我会把视频发到那里。此外，我会附上一些渲染和DS波形在高功率。由于功率晶体管的耐压1200V，我还有机会把功率推得更高。目前，由于整流器滤波器组件的尺寸较小

[davekni]

I will post the videos on the Bilibili platform https://b23.tv/4NFtCpP

That link worked for me.  Impressive coil!
BTW, I had fun with a couple other breakout points for colored arcs.  Strontium chloride makes bright red.  Boron (boric acid or ammonium borate) makes a light green.  (Borax doesn't work because it is a sodium salt so the sodium yellow dominates.)  Steel wire (ie. steel spring) is also fun, but requires protecting everything around from the flying sparks.

[曹靖]

I will post the videos on the Bilibili platform https://b23.tv/4NFtCpP

That link worked for me.  Impressive coil!
BTW, I had fun with a couple other breakout points for colored arcs.  Strontium chloride makes bright red.  Boron (boric acid or ammonium borate) makes a light green.  (Borax doesn't work because it is a sodium salt so the sodium yellow dominates.)  Steel wire (ie. steel spring) is also fun, but requires protecting everything around from the flying sparks.

It would be great if you could access Chinese video websites and watch my videos! In one of my videos, there is also a discharge tip coated with strontium chloride. It does produce a bright red flame

[曹靖]

Due to internet restrictions in China, I finally uploaded a video about HFSSTC from a few months ago to YouTube through an accelerator yesterday. Here is the video link

[davekni]

Great to see you on youtube.  Hope that process continues to work for you.