Problems with my simple DRSSTC

[stoppi_physics]

Hello!

My name is Christoph (often called stoppi by friends) and I'm a teacher for mathematics and physics. At the moment I'm trying to run my simple DRSSTC.

It has a fullbridge with four HGTG20N60 IGBTs (40A/600V) and I run the tesla coil with a 10A-variac.

The driver is very simple and works with secondary feedback without over current protection.

The Gate-signals seem to be okay apart from the relativ slow edges (around 300 ns rise-time). For the interrupter I use the schematics from Steve Ward (https://www.stevehv.4hv.org/interrupter/intwbmsch.JPG)

At the moment I have the following problems:

If I operate the Tesla coil in normal interrupter mode, the current requirement is very low (0.7A at 55V, 70µs on-time and 150 bps), but the volume is already very high. The flashes are very finely branched. If I increase the on-time further, the volume becomes unbearable even at low voltages of around 80V.

In burst mode I have different problems. Here I get relatively nice discharges, which in my opinion are atypical for a DRSSTC. They are more like a VTTC. Strangely enough, the volume is more tolerable in burst mode. Rather a dull hum and not the extremely loud single tone. The power requirement is enormous in burst mode. With low voltages on the full bridge of 30V, the current is already 10A and more and my 2000W Variac is already overloaded. I can't go any further with the tension here either.

I've reversed the feedback from the secondary coil, but then the Tesla coil doesn't work at all. The interrupter housing is grounded.

The frequency of my DRSSTC is around 300 kHz.

Since I'm absolutely not a Tesla professional, maybe one of you has a tip for me. I would just like to get the discharges typical of a DRSSTC, whereby 30-40 cm long flashes would be enough for me. It would also be nice if I could operate the Tesla coil directly at 230 Vac without a variable transformer. For this I would also throttle the Tesla coil and, for example, only use low on-times.

Thank you in advance for your efforts, best regards

[GrantV]

Hi Stoppi, I do not have enough knowledge to answer your questions properly so am hoping the others will jump in and help :-)

This looks more like an interrupted SSTC than a DRSSTC to me? (which I have never played with )

The 'high tone' of your spark output is a combination of the fres of your secondary and the 150hz from your interrupter.

Considering that you are getting output from your coil at lower voltages and current, I would probably think that your coil coupling is way too tight. Have you tried a larger diameter primary form and/or moving the primary lower?

BTW: Congrats on your build 

[Mike]

It's normal for Tesla coil discharges to be very loud, I always wear earmuffs when using mine. Did you make any measurements of your primary and secondary resonant frequencies? Initially is start with the primary tuned just a few % lower han your secondary. You can increase the amount of detuning for better performance once you're happy that everything else is working properly.

As Grant mentioned you have high coupling with the primary wrapped right around the secondary. I'd recommend inputting all your dimensions into JavaTC to see where that lands you. Most DRSSTC are running with coupling between 0.1 to 0.2. This could also be helpful for estimating your resonant frequencies to see how far apart they are.

Mike

[davekni]

As Grant mentioned you have high coupling with the primary wrapped right around the secondary. I'd recommend inputting all your dimensions into JavaTC to see where that lands you. Most DRSSTC are running with coupling between 0.1 to 0.2. This could also be helpful for estimating your resonant frequencies to see how far apart they are.

Yes, I too recommend JavaTC.  However, I suspect coupling isn't that high as-is.  Looks like only three primary turns closely spaced at bottom of coil form, which isn't likely to couple too strongly.  JavaTC will estimate (reasonably accurately) the coupling factor along with resonant frequencies.

The driver is very simple and works with secondary feedback without over current protection.

This is the first example I've seen of a DRSSTC using secondary feedback.  Not surprising that performance is unique.  Without running JavaTC myself on your coil, I'd guess primary frequency is a bit above your 310kHz operating frequency.  If so, that would keep IGBTs running with phase lead (switching before primary current zero-crossing), which may explain in-part why your IGBTs haven't died yet.

Secondary feedback may also explain why you are getting away without overcurrent protection.  At secondary frequency, arc loading will provide damping.  As arcs get longer causing secondary frequency to drop, primary is farther out-of-tune with operating frequency, so primary current doesn't increase arbitrarily high as happens with primary feedback.

Would be interesting to see primary current and voltage waveforms during operation!  Not so easy unless you have an isolated supply for H-bridge or a differential scope probe.

I would just like to get the discharges typical of a DRSSTC, whereby 30-40 cm long flashes would be enough for me. It would also be nice if I could operate the Tesla coil directly at 230 Vac without a variable transformer. For this I would also throttle the Tesla coil and, for example, only use low on-times.

Not sure how to achieve normal DRSSTC performance without a normal DRSSTC circuit.  That would be primary feedback and current limit.  Most DRSSTC drivers also include phase lead to keep IGBTs switching slightly before current zero-crossing.  Some small coils omit phase lead.  Without phase lead, IGBT stress is higher and switching spikes higher, so require more margin between bus voltage and IGBT voltage capability.  (TVS diodes across IGBT C-E can be worse than none.  IGBTs often have higher avalanche energy capability than the TVS diodes.  If a TVS diodes fries, that can fry other IGBTs.  Gate-emitter TVS diodes are a good idea.)

The Gate-signals seem to be okay apart from the relativ slow edges (around 300 ns rise-time).

Since you do not have diodes across gate resistors to provide dead-time, slow edges are necessary.  Your 300ns edges are about right for providing needed dead-time for your IGBTs.

BTW, have you checked temperature of your UCC driver chips?  Especially when running continuous mode (not interrupted), I'd suspect they may run very hot.  Unusual to drive four IGBTs directly from such driver chips, especially with the 1:2 GDT ratio.

Hope your project goes well!  Feel free to ask followup questions here on the forum.

[Rafft]

 Antenna and secondary feedback was how it was done 10+ years ago 

[stoppi_physics]

Thank's a lot to everyone for your comments and support, I much appreciate it...

Yes, the design of my DRSSTC is very old school. This is because I made a 10 year break to test it... 

If I'm right I shouldn't use diodes across the gate-resistors?

I have done measurements concerning the primary and secondary feedback but I will do it once again. For the secondary coil I got a resonance frequency around 312 kHz. The primary capacitor has 220/3 = 73 nF and I use 3-5 turns for the primary coil.

I have noticed that my on-time was way too high in burst mode. This was the reason for the extremely high current (more than 10A at just 20V). Now the current is lower (3-4A at 40V from the variac). This still seems too high for me...

Today I replaced the 27V zener diodes against 24V and also changed the GDT. Now 15 turns secondary side instead of the 17 turns before. I hoped to decrease the rise time of the edges. But they are still in the 300 ns range. But as you mentioned this could be positive and the reason, why my IGBTs are still alive 

My coupling seems way too high, around 30%. This is because the diameter of the secondary coil is 11 cm and the diameter of the primary just 13 cm. Should I switch to a horizontal primary coil?

And I have forgotten to mention that I use MUR460 and 440V TVS-diodes across each EC-path of the IGBTs.

Thank's once again for your support 

[davekni]

Antenna and secondary feedback was how it was done 10+ years ago 

I'd thought those were all SSTCs, not DRSSTCs.  Were there other DRSSTC examples like this?  My first coil was 9 years ago, so no personal experience that far back.  (And at the time I didn't know of other designs or any of the acronyms such as SSTC or DRSSTC.)

If I'm right I shouldn't use diodes across the gate-resistors?

Yes, that's my suggestion, no diodes.  Most designs do use diodes, but also have lower impedance gate drive (FET buffer after driver chips, 1:1 GDT with lower leakage inductance).  Your rise time looks good for sufficient dead time.  Also, if you were to add diodes, gate resistor power dissipation would go down and driver chip power dissipation go up.  Might be more heat than driver chips can handle if diodes are added.

My coupling seems way too high, around 30%. This is because the diameter of the secondary coil is 11 cm and the diameter of the primary just 13 cm. Should I switch to a horizontal primary coil?

Coupling is higher than most DRSSTC designs.  However, I have no idea what ideal coupling is for a secondary-feedback DRSSTC.  Anyone else have experience with secondary-feedback DRSSTC?  I'd love to see more information about these designs and how they perform.

And I have forgotten to mention that I use MUR460 and 440V TVS-diodes across each EC-path of the IGBTs.

Don't see any avalanche energy specification for this old IGBT.  No way to tell if TVS diodes are helpful or harmful.  Also hard to tell if MUR460 is a problem or not.  It will share current with IGBT's internal anti-parallel diode.  Both diodes have negative Vf vs. temperature curves.  If MUR460 conducts enough current to get hot, then it will take even more of the current and get even hotter.  Also don't know how much current the TVS diodes may share in forward conduction.  Since your coil has been working so far, perhaps leave it alone.  For anyone using newer IGBTs with avalanche energy rating, I'd recommend no external C-E diodes, only bidirectional TVS for gate protection.

Good luck with your coil!

[stoppi_physics]

Thank's dave for your help 

I have started to build this DRSSTC 10 years ago but then left the project until now. Therefore I oversplept the new designs with OCP and primary feedback...

The reason why I still try to use this design is that I want a very simple setup. And my expectations concerning the output are not very high. As I mentioned before I would be very happy with 30-40 cm sparks.

Here you can see exactly the same setup running:

I'll make a horizontal primary coil to lower the coupling, hoping to reduce the current consumption in burst mode. If I have new informations I'll inform you.

Thank's for your great help, cheers stoppi

[Mads Barnkob]

Antenna and secondary feedback was how it was done 10+ years ago 

I'd thought those were all SSTCs, not DRSSTCs.  Were there other DRSSTC examples like this?  My first coil was 9 years ago, so no personal experience that far back.  (And at the time I didn't know of other designs or any of the acronyms such as SSTC or DRSSTC.)

Steve Wards and Jimmy Hynes first work from interrupted SSTC to DRSSTC was using secondary base current feedback: https://stevehv.4hv.org/DRSSTC1.htm

[davekni]

Steve Wards and Jimmy Hynes first work from interrupted SSTC to DRSSTC was using secondary base current feedback: https://stevehv.4hv.org/DRSSTC1.htm

Thank you for the link.  Looks like secondary feedback lasted for about a year.  That schematic is dated 4/2004.  Next version with primary feedback is dated 3/2005.

[stoppi_physics]

Hi!

I have changed some things. First I now use a larger Topload to decrease the resonance frequency, it's now around 280 kHz. Second I use a flat primary coil. For 3 turns I get maximum output. The coupling is now at 20.1 percent. With the vertical primary coil it has been over 30 percent.

But the problem now is, that I just get an output in Burst-mode of the interrupter. The current is then between 2-4 A at 100 V from the variac.
When I change to "normal"-mode, I don't get any sparks in air. Only with a screwdriver there are tiny sparks of 5 mm length. When I change the on-time and frequency of the interrupter in "normal"-mode up and down, there are sometimes single sparks in air, but only like a single eruption.

What do you think, is the reason for this behaviour? What should I do next? I'm absolutely not the Tesla-expert 

Thank's for your help in advance, cheers Christoph (stoppi)

[davekni]

What do you think, is the reason for this behaviour? What should I do next? I'm absolutely not the Tesla-expert 

If anyone on the forum has experience with secondary-feedback DRSSTC, that will be more helpful than anything I suggest.  I'd personally love to understand more, and perhaps then be able to assist.  However, that path may be long, involving more scope measurements.
Do you have any ferrite cores left?  If so, you could make a current transformer for measuring primary current.  It would help me understand if you can measure three signals:  One GDT primary lead (or 74HC14 output), primary current (new current transformer into a burden resistor), and a scope probe as antenna to detect secondary voltage (top load voltage).  Overall shape across one interrupter enable pulse and a few zoomed-in images to show relative phase at start, middle, and end of enable pulse.  If only two scope channels are available, separate captures of each set of two of the above three signals is fine.  Of course, this effort is only if you are interested in teaching me.  If you get other helpful responses, try those first.

Edit: Decided to modify my simplified DRSSTC simulation model for secondary feedback, mostly for my own understanding.  Operation is more similar to primary feedback than I'd expected.  Startup seemed a bit slower once I added some delay to approximate driver and IGBT delays.  I attempted to roughly match parameters of your coil.



Noticed one possible issue.  Requires about 240A peak primary current to hit 100kV peak secondary voltage.  My 100kHz coil needs about 150kV peak for breakout.  At 280kHz, breakout voltage should be lower, but I'm not sure how much lower.  I also don't know what current your IGBTs can handle for short durations, especially in this semi-hard switched mode (switching after current zero-crossing) as your simple driver does.

Two options come to mind:
One is to try short on-times and higher bus voltage (higher variac setting).  If IGBTs can handle high current briefly, you might get primary current high enough fast enough to breakout well.  Keeping duty cycle low would keep total line power reasonable.  Risk is that IGBTs might not survive.
Other option is to increase impedance of primary.  Connect 37.5nF primary resonant caps in series instead of parallel (for 1/4th capacitance).  Double the turn count of primary coil (with turns closer together so that the updated coil fits in roughly the same space).  That makes primary inductance 4x higher, so same primary resonant frequency.  Then you can run at higher bus voltage and lower primary current.  Will require only half the current for a given secondary voltage.

Good luck with your testing and modifications!

[stoppi_physics]

Thank you very much, David, for your support and thoughts 

I will try to get an oscilloscope working with batteries to look at the signals during the run. Maybe I see some abnormalities and problems.

Cheers stoppi

[davekni]

I will try to get an oscilloscope working with batteries to look at the signals during the run.

The specific scoping I'd suggested shouldn't require scope battery power.  That's why I avoided requesting H-bridge signals directly.  Driver should be isolated from line, so allow scoping one GDT input wire or HC14 output pin.  That will show phase of H-bridge closely enough, missing only the ~300ns IGBT delay.  Antenna probe is just a probe in the vicinity with unconnected tip.  Current transformer isolates primary current from line voltage.  That's the only item needing construction, a new CT to measure primary current, along with a burden resistor across CT secondary.  There are many examples of DRSSTC primary CTs on the forum, typically used for feedback and OCD.  Many are two stages, so require two cores (or three cores to cover both feedback and OCD, but you don't need that).  For your small coil, a single stage may suffice, especially if you have magnet wire and patience to wind 100+ turns.

Might make more sense try higher primary impedance rather than the effort for more detailed scoping.  (Or try higher primary current for short pulses.)  After simulating, I'd guess low primary impedance (relative to H-bridge capability) is more likely the issue than any subtly related to secondary feedback.

[stoppi_physics]

Hello again!

My DRSSTC-project has been sleeping for a while, after I ruined my secondary coil because of discharges between windings. Therefore I had to make another secondary with the help of my daughter. I choose a PVC-pipe with 7.5 cm diametre and a coil-length of around 50 cm with 0.3 mm enameled copper wire. What do you think? Do I have to shorten the pipe at the top to get the torus closer to the end of the secondary? At the moment there is a gap of around 6 cm between the upper end of the secondary and the end of the pipe...

Thank's for your help, cheers stoppi

[davekni]

Do I have to shorten the pipe at the top to get the torus closer to the end of the secondary? At the moment there is a gap of around 6 cm between the upper end of the secondary and the end of the pipe...

Some gap between top end of secondary and top load is good.  Allows space for magnetic field from secondary to return (to escape top of winding) without being shorted by conductive top load.  Secondary radius (3.75cm) should be enough for that.  However, I doubt 6cm causes any issues.

I choose a PVC-pipe with 7.5 cm diametre and a coil-length of around 50 cm with 0.3 mm enameled copper wire. What do you think?

Fairly high aspect ratio (height / diameter).  Similar ratio to my DRSSTC is (110cm / 16cm).  One down-side of high aspect ratios is that to achieve good coupling factor, bottom of secondary gets high volts/turn, so can cause arcs between secondary windings as I gather you have already experienced.  I had to reduce coupling a bit (lower primary height a bit) to avoid such racing sparks on my coil, to 0.141 if I'm recalling correctly for my coil.

[stoppi_physics]

Thank's David for your help and the information. Then I will try a test run with this setup without shortening the gap at the top of the secondary. Cheers stoppi

[Dominic@TWHV]

Hi Christoph:

Judging from the spark size, if you are 100% sure this is a DRSSTC setup, there are two possibilities that I can think of regarding the performances.

1. Badly tuned primary resonance circuitry, it needs to match (10% lower detune) the resonance frequency of the secondary coil and topload. With this coil size, you could expect the secondary frequency to be around 300-500khz (judging from the sizes and using my personal experience).

2. Poor feedback. Poor feedback signals could disrupt the driver from properly resonating and generating sparks at the resonance frequency, confirm your driver has good feedbacks, and your feedback transformer (we don't often use antenna for a DRSSTC, if that's what you used, try switching to a CT) is picking up the right signal and not some random signal enough to disrupt the feedback (rare, but possible).

These are just my personal opinions, and they may **not** be correct. Take it as a suggestion. Best of luck fixing your problem!

Dominic

[stoppi_physics]

Thank you very much for your help, Dominic, i really appreciate it...

Today I determined the resonance frequency of the new secondary coil. I got 336 kHz. This isn't very low but i used a 0.3 mm thick wire and the diametre of the secondary is only 7.5 cm.

For the feedback I use a CT with one winding for the primary and 60 windings for the secondary.

The next step will be testing it with my variac. I will report updates here 

[Dominic@TWHV]

336khz is a weird number judging from previous experience and the coil size in the images that you provided.

I would be expecting ~200khz (+-50k) from that size, so something may not be right here.

As always, good luck! Feel free to reach out if you have more problems!

Cheers, Dominic