First test of the SSTC

[LoOdaK]

Hello everyone,
the first test of my first SSTC is coming up.
I'm waiting for a few more components that should arrive soon. I have a couple of questions before the first tests, I've usually seen other coilers test their SSTCs at a lower voltage first. The problem I have is that I don't have a variac. Is there any other way to lower the voltage during the test? What else should I pay attention to during the first tests ?

I hope my tests are going to be successful so I can share my results here

[davekni]

The problem I have is that I don't have a variac. Is there any other way to lower the voltage during the test?

If you have any lower-voltage supply (lab supply, old surplus laptop computer supply, transformer), that is a good first step.  I and many others start with a lab (bench) supply with adjustable voltage and current limit.  But I have also used old laptop supplies.  I'm presuming none of these are available to you.  Next best option is to use line power with an incandescent light bulb in series to limit power.

Presuming this is an interrupted SSTC, start testing with very-low duty cycle, such as 50us on-time at 1Hz repetition rate.  Slowly increase on-time before increasing repetition rate.  Even with a bulb in series, sparks should be possible at 1Hz.  1 second allows time for bulk caps to recharge between on-times even through light bulb.  (Before trying direct line power, use the highest wattage bulb available and/or multiple bulbs in parallel to increase available power while still providing some protection.)

What else should I pay attention to during the first tests ?

Do you have access to an oscilloscope?  That is the best way to see that circuitry is behaving as designed even at low power where no sparks are generated.

[Mads Barnkob]

As Dave mentions, very low BPS and then find a way to lower the input voltage. If you have spare electrolytic capacitors, you could make a capacitive voltage divider and with just 4 capacitors, you can get roughly 320, 240, 160 and 80 VDC from rectified 230 VAC.

I usually use a small fluorescent tube or bulb, near the secondary coil, to quickly see if there is a HF EMF emitted, a clear sign of a coil working at lower voltages than enough for break out.

[davekni]

If you have spare electrolytic capacitors, you could make a capacitive voltage divider and with just 4 capacitors, you can get roughly 320, 240, 160 and 80 VDC from rectified 230 VAC.

This sounds interesting.  Are you proposing a switched-capacitor divider with manual switching from series to parallel connection?  Or some other circuit I'm not imagining at the moment?

[LoOdaK]

Thanks a lot guys.

As for the capacitive divider, I have only polypropylene film capacitors lying around, can capacitive divider be made with them ? I could make about 50 VDC or something with them maybe.

[davekni]

As for the capacitive divider, I have only polypropylene film capacitors lying around, can capacitive divider be made with them ?

Depends on how high value your capacitors are.  If you have some large (ie. motor-run) capacitors an AC divider on line voltage could work.  For example, 30uF at 60Hz is 88 ohms reactance.  So if you had a 10uF and a 20uF capacitor, you could make 1/3rd of line voltage with 88 ohm source impedance.  10uF line hot to output, 20uF output to line neutral.  SSTC from output to neutral.  That should charge your SSTC bulk cap in reasonable time.  1uF and 2uF would be 880 ohms.  Might be usable, but you'd need very low repetition rate to allow time for bulk caps to charge between pulses.

[Mads Barnkob]

If you have spare electrolytic capacitors, you could make a capacitive voltage divider and with just 4 capacitors, you can get roughly 320, 240, 160 and 80 VDC from rectified 230 VAC.

This sounds interesting.  Are you proposing a switched-capacitor divider with manual switching from series to parallel connection?  Or some other circuit I'm not imagining at the moment?

Just series connected and you tap whatever point to ground that you want, its crude, but will give you a lower voltage from mains

[davekni]

Just series connected and you tap whatever point to ground that you want, its crude, but will give you a lower voltage from mains

That works only for AC (and AC-capable caps).  With DC, load will discharge bottom cap (or wherever load is connected) resulting in near-zero output voltage after a short while.

[klugesmith]

Dave's first reply suggested an incandescent light bulb in series with your apparatus.

In addition to that, you can put a resistive load (such as light bulb or heating appliance) in parallel with apparatus.  Now you have a resistive voltage divider, whose unloaded voltage can be as small as you like (even 10% of nominal mains voltage).  Voltage sensitivity to your load current is much smaller than it would be with light bulb alone. 

Divider losses might be more than 100 watts, but the components can handle that safely, and for most of us it's winter so extra heat is welcome.

[Mads Barnkob]

Just series connected and you tap whatever point to ground that you want, its crude, but will give you a lower voltage from mains

That works only for AC (and AC-capable caps).  With DC, load will discharge bottom cap (or wherever load is connected) resulting in near-zero output voltage after a short while.

But it will still work to charge up a larger DC bus capacitance for low BPS testing. I thought we were just talking about how to get a lower voltage for a small test and not a sustainable solution

[LoOdaK]

As for the capacitive divider, I have only polypropylene film capacitors lying around, can capacitive divider be made with them ?

Depends on how high value your capacitors are.  If you have some large (ie. motor-run) capacitors an AC divider on line voltage could work.  For example, 30uF at 60Hz is 88 ohms reactance.  So if you had a 10uF and a 20uF capacitor, you could make 1/3rd of line voltage with 88 ohm source impedance.  10uF line hot to output, 20uF output to line neutral.  SSTC from output to neutral.  That should charge your SSTC bulk cap in reasonable time.  1uF and 2uF would be 880 ohms.  Might be usable, but you'd need very low repetition rate to allow time for bulk caps to charge between pulses.

I went on an expedition to find the capacitors, I only have 680nF and similar lying around so that is not an option for me probably. Thanks for the explanation.

Dave's first reply suggested an incandescent light bulb in series with your apparatus.

In addition to that, you can put a resistive load (such as light bulb or heating appliance) in parallel with apparatus.  Now you have a resistive voltage divider, whose unloaded voltage can be as small as you like (even 10% of nominal mains voltage).  Voltage sensitivity to your load current is much smaller than it would be with light bulb alone. 

Divider losses might be more than 100 watts, but the components can handle that safely, and for most of us it's winter so extra heat is welcome.

Well I have light bulbs lying around so that is probably my way to go. Thanks for the info.

Also I'm looking at some variacs but I'm quite uncertain what amperage my variac needs to handle for a standard SSTC ? I'm never planning to run my SSTC in CW mode.

[davekni]

Also I'm looking at some variacs but I'm quite uncertain what amperage my variac needs to handle for a standard SSTC ? I'm never planning to run my SSTC in CW mode.

Don't know what your budget is, but you might want to consider saving for an oscilloscope before paying for a variac.  Incandescent light bulb voltage divider is less convenient than variac, but works for testing.  Even when using a variac, I often add a bulb in series with variac output for the first tests of a line-powered circuit.  Bulb limits current in case of a fault.

I think somewhere you mentioned that your SSTC design is somewhat close to one of Mads' coils.  Perhaps he has information on power draw vs. duty cycle.  In general, more power will make more impressive arcs.

However, you don't necessarily need a variac capable of full power to your SSTC.  Variac is needed mostly for low power testing.  Once your coil is working well at full voltage and low duty cycle, the variac can be replaced with a direct connection to line power.

One more caveat:  Some people always power their coils through variacs.  That provides one solution to limiting inrush current charging bulk capacitor.  For direct line power connection, you'll need some method to limit inrush current.  Large NTC thermistors designed for inrush current limiting are available.  Or some other resistor (such as incandescent light bulb) can be used in series with line power when first connected, then shunted by a switch after bulk cap is charged and before running the coil.

[LoOdaK]

Also I'm looking at some variacs but I'm quite uncertain what amperage my variac needs to handle for a standard SSTC ? I'm never planning to run my SSTC in CW mode.

Don't know what your budget is, but you might want to consider saving for an oscilloscope before paying for a variac.  Incandescent light bulb voltage divider is less convenient than variac, but works for testing.  Even when using a variac, I often add a bulb in series with variac output for the first tests of a line-powered circuit.  Bulb limits current in case of a fault.

I think somewhere you mentioned that your SSTC design is somewhat close to one of Mads' coils.  Perhaps he has information on power draw vs. duty cycle.  In general, more power will make more impressive arcs.

However, you don't necessarily need a variac capable of full power to your SSTC.  Variac is needed mostly for low power testing.  Once your coil is working well at full voltage and low duty cycle, the variac can be replaced with a direct connection to line power.

One more caveat:  Some people always power their coils through variacs.  That provides one solution to limiting inrush current charging bulk capacitor.  For direct line power connection, you'll need some method to limit inrush current.  Large NTC thermistors designed for inrush current limiting are available.  Or some other resistor (such as incandescent light bulb) can be used in series with line power when first connected, then shunted by a switch after bulk cap is charged and before running the coil.

Thanks a lot for the tips. I'm actually looking to buy some hantek DSO 2000 series scope soon, I think it should be more than adequate for my projects.

As for the inrush current limiting, I was thinking about designing a small circuit with the 555 timer and relay to limit the inrush current. After some time , the relay would switch from the inrush current limiting resistor to the "no resistance" path. Would that be a good solution also ?

[davekni]

I'm actually looking to buy some hantek DSO 2000 series scope soon, I think it should be more than adequate for my projects.

Expect that will be fine.  I have a bit-older Hantek model, DSO4102C.  Much more portable than my much-older Tektronix TDS754C.  Prefer my Tek scope for most use, but the Hantek one is good enough.  I do find its 2,4,8,20,... sequence for horizontal time/div to be annoying compared to the normal 1,2,5,10... sequence of Tek scopes.  BTW, had to put Hantek scope in a foil-lined box to use around my Marx generator.  Otherwise it got confused by large transient fields.  No such issue with Tek scope.

As for the inrush current limiting, I was thinking about designing a small circuit with the 555 timer and relay to limit the inrush current. After some time , the relay would switch from the inrush current limiting resistor to the "no resistance" path. Would that be a good solution also ?

Yes, that is a good solution, one I've seen before.  It's an automated version of the series resistor/bulb shunted by a manual switch.

Good luck with testing!

[LoOdaK]

Guys thanks a lot for the help.

I did my first test yesterday and it was kind of successful, at least nothing blew up
It is a great milestone for me, I got my "first light" 

I have one problem and I have a few "fixes" in my head but would like to hear from you guys.

I did my tests with a lightbulb in series. The coil was working, and it was producing arcs but the arcs were very disappointing. They were maybe about 1cm in length, I hoped for more.
I think my main problem is the coupling between the coils. I think I will need to increase it a little bit. Currently, the primary coil has 8 turns, I'm thinking about lowering it to 6 turns and seeing what happens.

The secondary is wounded on a 75mm tube (around 1000+ turns) and the primary coil is wounded on a 100mm tube.
One more thing which maybe draws a little bit of energy from the coil is the platform supports. I did my support with M10 rods (I don't know what you call them in English) and nuts which hold the lower platform (where are all the electronics) and upper platform (where are primary and secondary windings). They are resting below a lower platform on the ground so I'm assuming they could be isolated from the ground and maybe improve performance by a little bit.

One thing I could also do is move my antenna a little bit to see how that affects coil performance.

Below is the coil picture which I took yesterday (sorry for the bad quality, I will post better photos soon)



Full picture: https://i.imgur.com/R8BHRrQ.jpg

And yes I know, it is quite big for an sstc

[Mads Barnkob]

Congratulations on first light! It is a major milestone for anyone doing Tesla coils!

If a SSTC is working, but only giving minimal output at a fair input voltage, lets say above 150V DCbus, you should try to reverse the phasing of the primary coil. You might just be driving it out of phase.

[LoOdaK]

Congratulations on first light! It is a major milestone for anyone doing Tesla coils!

If a SSTC is working, but only giving minimal output at a fair input voltage, lets say above 150V DCbus, you should try to reverse the phasing of the primary coil. You might just be driving it out of phase.

Thanks Mads, yes I tried reversing the phasing and when I reverse it, I don't get any sparks at all so phasing is probably right.


Today I played with it the coil a little bit more, I elevated the primary coil and removed one turn from the primary winding (total 7). The performance was better but using 230VAC without a lightbulb and getting sparks about a 2cm max is really bad. I pushed it really hard in music mode with a random song from a phone and sparks grew to 3-4 cm and one of the MOSFETs popped.

I'm thinking about winding a new primary with a smaller diameter to improve coupling, I think that should help a little bit but I'm still not sure If that will improve performance to the level I was hoping for.

Below are a couple of photos of the coil and fets after destruction 









Album: https://imgur.com/a/A5PuwYO

[Rafft]

Cheer up bud!

Been in the same boat as you do (13yrs ago). Ive made my coils following from the web pages w/o ever using a -proper- scope, or any scope for that matter. I just relied in popping mosfets as my teachers.

If you live in.an area with 230v, get a 110v 500w step down transformer. (Or a variac if you have). Banging parts are less likely.

Eventually youll get it right as you progress

Here are some of my videos... SSTC from 110v to 220v

Dont loose hope. And get a scope 

/>

/>

/>

/>

[davekni]

I think my main problem is the coupling between the coils. I think I will need to increase it a little bit. Currently, the primary coil has 8 turns, I'm thinking about lowering it to 6 turns and seeing what happens.

Yes, looks like coupling might be low for an SSTC.  Entering your coil parameters into JaveTC will give you coupling, inductances, resonant frequency, etc.  Highly recommended.  Posting JavaTC results here will help too, as it could improve value of replies.  Raising primary does increase coupling.  Going as high as you did for the next post risks arcing from secondary to primary.
A couple other things that help to increase coupling:  Spread out primary turns, and pair primary lead wires close together (ie. fed through same hole in frame and twisted or taped together along their length.

I did my support with M10 rods (I don't know what you call them in English) and nuts which hold the lower platform (where are all the electronics) and upper platform (where are primary and secondary windings). They are resting below a lower platform on the ground so I'm assuming they could be isolated from the ground and maybe improve performance by a little bit.

I don't think those rods are close enough to cause any problem.  BTW, a counterpoise (ie. section of aluminum foil under entire coil frame and connected to ground) can help with performance.

The performance was better but using 230VAC without a lightbulb and getting sparks about a 2cm max is really bad. I pushed it really hard in music mode with a random song from a phone and sparks grew to 3-4 cm and one of the MOSFETs popped.

Probably better to focus on optimizing performance at lower duty cycle before pushing the coil hard.
Where did you purchase IRFP460 FETs?  There are plenty of counterfeits of that part on the market, mostly from China.  Counterfeit parts often have smaller die with lower current and avalanche energy capabilities.
Also, is this a purchased ECB or one you designed?  A half-bridge layout with high parasitic inductance adds stress to FETs, making failure more likely.  (Some purchased ECB's aren't that great either.  Wide variation in design quality exists.)
Whenever you do get a scope, that will make optimization easier, especially if there is some circuit issue limiting performance.

I'm thinking about winding a new primary with a smaller diameter to improve coupling, I think that should help a little bit but I'm still not sure If that will improve performance to the level I was hoping for.

JavaTC will help.  Enter your new ideas before constructing them.  Will also show if secondary impedance is way too high or low.  Conventional wisdom says ~50k impedance at resonant frequency is good.  Coils from 25k to 100k can work reasonably.

Good luck!  Nice to see progress.  Almost every project I tackle takes much longer than I initially think it should.

[LoOdaK]

Thanks a lot Rafft, those sparks are pretty good. I'm definitely saving for a scope.



Dave, as for the fets, they could be fake. I got them from a marketplace in my country. Below is the photo of the fet taken apart, I was curious about how it looks like on the inside



I compared it to the fets from this forum conversation: https://www.eevblog.com/forum/beginners/fake-irfp460-mosfets-from-aliexpress/

and the photo from it:


According to the post, real fet is on the left and fake on the right. I notice a smaller die size on the fake part. When I compare it to the my fet I'm somewhat confused, it doesn't look like a fake one but it doesn't look exactly like the real one also. Unfortunately, I don't have the equipment to test it and compare it to the datasheet values. So possibility of the fet being fake is high


As for the JavaTC results, I forgot to upload it. Here is the picture from the JavaTC:


The coupling coefficient is 0.307. What is the general rule of thumb about sstc coupling ? Is 0.4 a bit too much ?

JavaTC will help.  Enter your new ideas before constructing them.  Will also show if secondary impedance is way too high or low.  Conventional wisdom says ~50k impedance at resonant frequency is good.  Coils from 25k to 100k can work reasonably.

From the JavaTC I can see it is around 27k. Is that bad ? If it is, is there a way to fix it or I need to construct a new secondary ? Also, why is 50k a good thing, shouldn't lower values be better ?

Thanks a lot for the help and explanations Dave, this means a lot to me. I love learning new things and this forum is an awesome place to do so

[davekni]

Dave, as for the fets, they could be fake. I got them from a marketplace in my country. Below is the photo of the fet taken apart, I was curious about how it looks like on the inside

Since the die area looks similar to genuine parts, it could be just a newer version from Vishay.  Or it could be a better-quality counterfeit part.

From the JavaTC I can see it is around 27k. Is that bad ? If it is, is there a way to fix it or I need to construct a new secondary ? Also, why is 50k a good thing, shouldn't lower values be better ?

50k secondary impedance is something I learned from this forum.  Built my first three coils before finding this.  My SSTC happens to be 50k.  My larger DRSSTC is 34k.  It has a high peak power H-Bridge suitable for lower impedance impulsive arcs.  50k is not a critical magic number.  Here are a number of forum links including some explanations:
https://highvoltageforum.net/index.php?topic=115.msg722#msg722
https://highvoltageforum.net/index.php?topic=937.msg6293#msg6293
https://highvoltageforum.net/index.php?topic=1066.msg7682#msg7682
https://highvoltageforum.net/index.php?topic=2284.msg16721#msg16721
https://highvoltageforum.net/index.php?topic=1538.msg12055#msg12055
https://highvoltageforum.net/index.php?topic=1872.msg14039#msg14039
In general, it seems that >50k is better for smaller or lower-power coils and <50k is good for large high-power coils.  You may have trouble getting enough power into your 27k secondary for voltage to get high enough to create good arcs.  Perhaps you could try newer higher-power FETs.
A smaller top load will raise frequency and impedance, but also has disadvantages.  Winding a new secondary may be the best option.

Thanks a lot for the help and explanations Dave, this means a lot to me. I love learning new things and this forum is an awesome place to do so

I too have learned a lot from this forum in the 3.5 years since I found it.  Would have saved me some time and work in building my coils if I'd found it earlier.  Could have avoided reinventing things that were already known.

[LoOdaK]

In general, it seems that >50k is better for smaller or lower-power coils and <50k is good for large high-power coils.  You may have trouble getting enough power into your 27k secondary for voltage to get high enough to create good arcs.  Perhaps you could try newer higher-power FETs.
A smaller top load will raise frequency and impedance, but also has disadvantages.  Winding a new secondary may be the best option.

How could that improve my power transfer to the secondary ? Do you mean I could push more current through the primary ?

I'm quite uncertain what I should do regarding the number of turns of the primary. Should I increase them or decrease them? I read in one of the links above that I should also watch my primary impedance. If I got it right, lowering primary impedance is going to allow more current through the primary but will of course stress transistors more. I calculated my primary impedance to be around 14.4 ohms with the current setup.

Lowering the diameter of the primary coil to 8cm while keeping my number of turns the same gives me a coupling coefficient of 0.4+. It's a bit too much maybe but thing which concerns me the most is the secondary reactance. With the new primary, reactance drops to around 20k. Primary impedance also drops to around 9 ohms. So it means that I don't actually gain any performance benefits from this, or very little. If my hypothesis is right, this would allow more primary current but if my secondary reactance drops to 20k, the secondary "won't be accepting" that power actually and I even risk getting sparks from the secondary to the primary.

I'm pretty confused as to what I should do.

I could also get some K60H603 IGBTs, would they work ? They seem to be higher-power than IRFP460.


More precise measurements, new JavaTC values:



After modifying primary:





EDIT: After playing a little bit I think I could try this design:



I kept the diameter of the primary coil to ~11 cm, and increased spacing between turns, so the primary height should be around 94mm with the same number of turns (7). That should increase coupling to around 0.379. The decrease in secondary reactance is minimal, but the impedance of the primary is quite lower, resulting in around 7.1 ohms. If I take an approximation, that should result in ~23A of current through the primary according to Ohm's law (probably not accurate because I don't know the better formula, but good enough I think). I think that IGBT could drive this (maybe even IRFP460 ? )

This could work a little bit better I think.

[davekni]

How could that improve my power transfer to the secondary ? Do you mean I could push more current through the primary ?

Yes, more current without FETs failing.  Lower primary impedance will increase current.  Higher coupling will transfer more power to secondary.  Primary current phase will shift closer to voltage phase.  Real power will be higher for any given current.

I could also get some K60H603 IGBTs, would they work ? They seem to be higher-power than IRFP460.

IGBTs require 15Vge rather than 10Vgs of FETs.  IGBTs also generally have higher switching losses in hard-switched applications like most SSTCs.  (It is possible for an SSTC with high coupling to switch at zero current, but that is rare.)  Some SSTCs do use IGBTs.  Usually lower-frequency designs.

EDIT: After playing a little bit I think I could try this design:

Yes, something like this is much better.  SSTC primary should generally be at bottom of secondary.  Some insulation between primary and secondary is needed too.  If you can find any plastic pipe or tube a bit smaller than your existing primary form, but still large enough to go over secondary, that would be ideal.  Would get coupling up a bit farther, which improves output power for a given primary current.

The decrease in secondary reactance is minimal, but the impedance of the primary is quite lower, resulting in around 7.1 ohms. If I take an approximation, that should result in ~23A of current through the primary according to Ohm's law (probably not accurate because I don't know the better formula, but good enough I think). I think that IGBT could drive this (maybe even IRFP460 ? )

23A is close for primary current (ignoring effect of secondary).  Primary voltage is a square wave of +-163V, resulting in a triangle-wave for current.  RMS current is a bit lower, by a factor of PI / (2 * sqrt(3)) or 20.8A.  Peak current is higher by PI / 2, so 36A.  IRFP460 may handle this for short duty cycles.  36A is well within peak current capability, but switching at 36A adds more power dissipation to FETs, and even more so to IGBTs.  Probably worth a try at short duty cycles.  Or add one more turn initially.  One more turn and a little smaller diameter might be ideal.

Good luck with your refinements!

Edit2:  Had to change above sqrt(eight) to 2*sqrt(2) because sqrt(eight) using the digit 8 becomes an emoji.  Then realized my RMS calculation was incorrect, so changed above again.

[LoOdaK]

IGBTs require 15Vge rather than 10Vgs of FETs.  IGBTs also generally have higher switching losses in hard-switched applications like most SSTCs.  (It is possible for an SSTC with high coupling to switch at zero current, but that is rare.)  Some SSTCs do use IGBTs.  Usually lower-frequency designs.

My GDT output should be around 18V. So would it be a good idea to try it at least ?

Yes, something like this is much better.  SSTC primary should generally be at bottom of secondary.  Some insulation between primary and secondary is needed too.  If you can find any plastic pipe or tube a bit smaller than your existing primary form, but still large enough to go over secondary, that would be ideal.  Would get coupling up a bit farther, which improves output power for a given primary current.

I will try to make coupling ~0.4. As for the plastic for insulation, I don't think I would be able to find a tube but I will try to come up with something.

23A is close for primary current (ignoring effect of secondary).  Primary voltage is a square wave of +-163V, resulting in a triangle-wave for current.  RMS current is a bit lower, by a factor of PI / (2 * sqrt(3)) or 20.8A.  Peak current is higher by PI / 2, so 36A.  IRFP460 may handle this for short duty cycles.  36A is well within peak current capability, but switching at 36A adds more power dissipation to FETs, and even more so to IGBTs.  Probably worth a try at short duty cycles.  Or add one more turn initially.  One more turn and a little smaller diameter might be ideal.

Thanks for the explanations. I think I will add one more turn to the primary but I don't think I would be able to make my diameter smaller I will need to stick with 110mm for now and going to make my primary a little bit higher to compensate, I hope that won't cause any problems.


EDIT: Something like this:


If I did my calculations right, around ~27.5-28A peak current, which is lower than before. Although secondary reactance is a little bit lower I think it should work better.


Thanks a lot, I will play with the values a little bit to see what would be the best and easiest way. Will update this thread with new results

[davekni]

My GDT output should be around 18V. So would it be a good idea to try it at least ?

18V should be good for IGBTs.  Do you have gate resistors shunted by diodes?  Needed for IGBTs to get sufficient dead-time, as IGBT turn-off times are longer than turn-on times.

If I did my calculations right, around ~27.5-28A peak current, which is lower than before. Although secondary reactance is a little bit lower I think it should work better.

Yes, this looks like a good design.  I'm calculating 30.8A peak ignoring secondary.  The direct way to calculate square wave current into an inductor:  During positive half-cycle, inductor current ramps from -peak to +peak.  Current ramps from 0 to +peak in 1/4 cycle.  For 275.45kHz, 1/4 cycle is (1/4)/275.45kHz = 0.9076us.  Integrating the inductor equation of V=L*dI/dt gives I =  V*t/L.  V is 325V/2 for half-bridge.  L is 4.783uH per your latest JavaTC.  So, I = 325V/2 * 0.9076us / 4.783uH = 30.8A.  Of course, this ignores diode and FET/IGBT forward drops, switching times, etc.  Thus reality will be slightly lower peak current.

[LoOdaK]

18V should be good for IGBTs.  Do you have gate resistors shunted by diodes?  Needed for IGBTs to get sufficient dead-time, as IGBT turn-off times are longer than turn-on times.

Yes, I do have 10R on the gate shunted with 1N5819 diode.
Also about gate resistance, do you know any useful links where I could learn some simple methods on how to actually calculate gate resistors, at least approximation for tesla coil bridges ? I watched a few videos and browsed the Internet but it is kind of complicated to grasp for me and didn't quite find a simple way.

[davekni]

Also about gate resistance, do you know any useful links where I could learn some simple methods on how to actually calculate gate resistors, at least approximation for tesla coil bridges ? I watched a few videos and browsed the Internet but it is kind of complicated to grasp for me and didn't quite find a simple way.

I'm not aware of any simple methods to determine optimum gate resistance.  Most FETs and IGBTs are specified at 0-10Vgs or 0-15Vge drive.  Driving at +-12V or +-18V has a large effect on switching times.  Gate capacitance is rarely specified for negative voltages.  Best option is to check waveforms with a scope.

On the good side, gate resistance value (when bypassed by a diode) is less important for SSTC than for DRSSTC.  For SSTC, turn-off is (almost) always at relatively high current.  Gate diode makes turn-off fast, minimizing switching loss.  There is plenty of time for opposite FET/IGBT to turn on before current reverses.  So even a slow turn-on (large gate resistance) is generally fine.  10R is common for single TO220 or TO247 devices.  Should be fine for your SSTC.

Gate resistance serves a second purpose besides adding dead-time:  Reduces Vgs overshoot and ringing due to GDT leakage inductance.  Your GDT is well constructed with twisted pairs on the core and twisted lead wires.  That minimizes leakage inductance.  Even 5 ohms would likely be plenty for damping given your good GDT.

If you change to a TO247 IGBT, be careful to keep lead wires short even while bending to fit a TO220 ECB, especially emitter and collector.  Or, if you want to construct a half-bridge board for your IGBTs, here's my tutorial on making a simple low-parasitic-inductance board from either copper-clad board or copper tape:
https://highvoltageforum.net/index.php?topic=1324.msg9795#msg9795
Edit:  Oops.  Somehow was thinking IRFP460 was a TO220 device.  Actually TO247-style, so no issue with bending leads to change to IGBT.

Not likely an issue, but that IKW60N60 IGBT you selected has about twice the typical gate charge as IRFP460.  Presume your gate drive circuit can handle 2x power.  There are other IGBTs with gate charge similar to IRFP460 FET.

[LoOdaK]

So I got some time, I wound the new primary and tested the coil.
I was using K60H603 IGBTs.

Testing was successful, nothing blew up

Below is the video of the coil test and music test and a couple of photos:


New primary coil on the left and the old one on the right for comparison



Primary coils with the driver board in the middle.


Link to the test of the coil (can't include into post because it is short)
https://www.youtube.com/shorts/_DZXR-RVtt4

Music test:
https://www.youtube.com/shorts/41ZnYOcYUac


I needed to isolate secondary with some wrapping (I don't know how you call them in English). In the first video, there is a bit of corona around the primary when running almost full power, so coupling should probably be lowered slightly more. Also, there is some corona on the antenna, which should be fixed probably.

In the music test, I tried to play a faster song after this one but it was just a mess, the song is heavily distorted and song can't be recognized. What could cause that ? I'm thinking it is probably feedback and lack of shielding of the driver. I'm also considering winding the feedback transformer, how I should go about that ?

[davekni]

I needed to isolate secondary with some wrapping (I don't know how you call them in English). In the first video, there is a bit of corona around the primary when running almost full power, so coupling should probably be lowered slightly more.

Coats of polyurethane epoxy are common on secondary coils.  Makes them more robust against physical abuse (bumps etc.) and provides insulation.  I've seen electrical tape work too.  Insulation is best either tightly adhered to secondary (as with epoxy coating) or spaced away by at least several mm.  A tiny (but non-zero) gap can trap any remaining corona, increasing rate of damage to insulating layers (coil varnish and inside of added insulating layer).
Hopefully you will get some other answers and suggestions for insulation.

Also, there is some corona on the antenna, which should be fixed probably.

Make a loop at end of antenna, or better a smooth ball.  Reduces local field strength.  Antenna will work fine inside a capped plastic pipe or other insulation too.

In the music test, I tried to play a faster song after this one but it was just a mess, the song is heavily distorted and song can't be recognized. What could cause that ? I'm thinking it is probably feedback and lack of shielding of the driver.

Difficult to tell without scope captures.  Shielding issues seem unlikely to be different between song speeds.  Oscillation startup time might be an issue, or the remaining charge on capacitors connected to return side of secondary.

I'm also considering winding the feedback transformer, how I should go about that ?

There's plenty of DRSSTC feedback current transformer (CT) examples on the forum.  Most are two stages.  Only one stage is needed for SSTC secondary feedback.  Just wind CT secondary around core, 80 turns if you want to keep that ratio.  Then one turn for CT primary (just wire passing through core for toroid core).

[LoOdaK]

Coats of polyurethane epoxy are common on secondary coils.  Makes them more robust against physical abuse (bumps etc.) and provides insulation.  I've seen electrical tape work too.  Insulation is best either tightly adhered to secondary (as with epoxy coating) or spaced away by at least several mm.  A tiny (but non-zero) gap can trap any remaining corona, increasing rate of damage to insulating layers (coil varnish and inside of added insulating layer).
Hopefully you will get some other answers and suggestions for insulation.

I actually have two coats of polyurethane varnish on the secondary. When playing music I didn't see any corona forming.

Difficult to tell without scope captures.  Shielding issues seem unlikely to be different between song speeds.  Oscillation startup time might be an issue, or the remaining charge on capacitors connected to return side of secondary.

Hmm, probably oscillation startup, something tells me that's the reason I hope I will soon get a scope so I could see the waveforms.

There's plenty of DRSSTC feedback current transformer (CT) examples on the forum.  Most are two stages.  Only one stage is needed for SSTC secondary feedback.  Just wind CT secondary around core, 80 turns if you want to keep that ratio.  Then one turn for CT primary (just wire passing through core for toroid core).

Thanks, I have some spare N30 cores laying around so I will probably make one for the test. Transformer feedback seems like a better solution to me than antenna feedback.

[davekni]

I actually have two coats of polyurethane varnish on the secondary. When playing music I didn't see any corona forming.

Sounds good.  If air-dry varnish, that will be much thinner than can be obtained with polyurethane epoxy (as used for glossy table tops etc.).  My DRSSTC initially had racing spark issues.  Besides reducing coupling a little, I increased epoxy coating from 2 to 7 layers, about 1mm per layer.  Doubt you need to go anywhere near that extreme.