Ramped SSTC build, some questions

[Manz]

Hello everyone,
I started building a classic ramped/staccato fullbridge SSTC. The entire circuitry will be enclosed in a ATX power supply and the goal of the project is to have it as a small plug and play coil.
I've already run the coil with a fgh60n60smd fullbridge but it blew up. so before running it again I'm making some changes.
I've switched the antenna for a 1:45 secondary CT, added zeners to the gates of the IGBTs, lowered the Fres from 420 to 360Khz and I'm now shielding the driver just to be  sure it won't pick up any interference.

Now some doubts.

I have some salvaged FGL40N120AND IGBTs and I'm not sure if they are suitable for operation in a SSTC. here's the datasheet. https://www.onsemi.com/download/data-sheet/pdf/fgl40n120and-d.pdf
I think they should be fine as the only "problem" I see is a higher gate charge compared to the 60n60 I've used but it should be manageable.

The other question is about operation once the coil is not longer in the "testing phase":

 Is it a problem if I connect the fullbridge directly to mains voltage (after the rectifier)? in this case unlike DRSSTCs there's not a huge inrush current as the capacitance on the bridge will be around 1uF (two 0.47uF mkp capacitor in parallel as snubbers). But maybe the IGBTs won't like going from 0V to >300V in a split second? I don't know if thet could cause some kind of failure.

here's a video of the coil running at 120Vac before it blew up  :/

https://www.youtube.com/shorts/bwDrzU9l52Q

I will scope everything before running the coil at high power to make sure it's all good and I'm also thinking about adding more turns to the primary so the current doesn't rise too much.

[Mads Barnkob]

What is the reason for using IGBTs, over MOSFETs?

In SSTC operation, you have a higher average current and not so much peak currents, where the IGBT excel.

Using unfiltered mains AC through a full-bridge will give a different spark effect, Richie Burnett made some experiments with different mains input filtering and waveforms: https://www.richieburnett.co.uk/sstate.html

[Manz]

I chose to go with IGBTs just because I have lots of them on hand that I've scrapped from old electronics and they should be able to do the job.
If they aren't good enough I will buy some MOSFETs

[davekni]

I chose to go with IGBTs just because I have lots of them on hand that I've scrapped from old electronics and they should be able to do the job.
If they aren't good enough I will buy some MOSFETs

Since you have the IGBTs, no harm in trying them.  However, standard SSTC driver circuits use secondary feedback.  Relative phase of H-Bridge current (primary current) and H-Bridge voltage (switching times) is not explicitly controlled.  360kHz is quite high for IGBTs.  Is possible if phase and dead time are well controlled.  I'm using IGBTs for my 450kHz QCW coil.
My guess is that you will have some success in staccato mode with on-times <1ms.  Long on-times of ramped use are likely to overheat IGBT die within 8-10ms of a half line cycle.

Is it a problem if I connect the fullbridge directly to mains voltage (after the rectifier)?

Adding TVS diodes across rectified line may be helpful.  If line is switched on at max voltage, line wiring inductance combined with 1uF capacitance will make a ring that can theoretically hit twice peak line voltage (if no losses).  Other need for TVS diodes is turn-off in staccato mode.  H-bridge current suddenly ends.  Again, line inductance causes voltage to spike.  Your 1200V IGBTs may be fine for these issues, as spikes may not get that high.

[ZakW]

Since you have the IGBTs, no harm in trying them.  However, standard SSTC driver circuits use secondary feedback.  Relative phase of H-Bridge current (primary current) and H-Bridge voltage (switching times) is not explicitly controlled.  360kHz is quite high for IGBTs.  Is possible if phase and dead time are well controlled.  I'm using IGBTs for my 450kHz QCW coil.

I noticed a significant bump in performance when replacing my old IRFP460 MOSFETS with IGBT's. I have used FGH60N60S's in the past at around 350-500kHz. I upgraded to the FGH75T65SHD for more current capability. They also work just fine in that range.

Having said that I have killed A LOT of IGBT's working on my ramped coil but I think that is caused by other factors.

My guess is that you will have some success in staccato mode with on-times <1ms.  Long on-times of ramped use are likely to overheat IGBT die within 8-10ms of a half line cycle.

Dave, can you elaborate? The half cycle of the mains ramp (120v) is ~8.33ms, how can a ramped coil run at or below 1ms? My coil is tuned to be on for the entire 8.33ms cycle. If it shuts off around ~6ms it makes a loud snapping sound so I have the pulse width adjusted for the full cycle. Am I confusing pulse width with on-times?

For reference - scope shot with my probe hanging in the air. The coil is 'on' for at least 8ms running at 562kHz as well.

[Manz]

My guess is that you will have some success in staccato mode with on-times <1ms.  Long on-times of ramped use are likely to overheat IGBT die within 8-10ms of a half line cycle.

I was able to run the coil at 10ms ontime for quite a bit before the IGBTs gave up, they were room temperature (running at around 3BPS) even without using a heatsink but I guess this doesn't mean the die temperature was also low.

Adding TVS diodes across rectified line may be helpful.  If line is switched on at max voltage, line wiring inductance combined with 1uF capacitance will make a ring that can theoretically hit twice peak line voltage (if no losses).  Other need for TVS diodes is turn-off in staccato mode.  H-bridge current suddenly ends.  Again, line inductance causes voltage to spike.  Your 1200V IGBTs may be fine for these issues, as spikes may not get that high.

Thank you, I didn't really think about that. I'm adding some TVS diodes as soon as possible.

One last thing.
https://www.onsemi.com/pdf/datasheet/fgh40n60smd-d.pdf
This is the fgh40n60smd datasheet and in page 6, figure 15 "Load current vs frequency"  it seems like they can hardswitch well over 25A @ 350Khz and 400V Vce. Is this a good parameter to look for to have some kind of comparison to SSTC use?
As the primary current in a sstc is limited by the primary impedance it shouldn't rise over 30A even at peak line voltage (in my case) and the average current during a pulse will be lower.

[davekni]

I was able to run the coil at 10ms ontime for quite a bit before the IGBTs gave up, they were room temperature (running at around 3BPS) even without using a heatsink but I guess this doesn't mean the die temperature was also low.

Do you have any measurements or estimates of primary current?  Perhaps I'm thinking more of designs with high primary current compared to yours.  Also, perhaps with luck, primary voltage and current phase are aligned well.

This is the fgh40n60smd datasheet and in page 6, figure 15 "Load current vs frequency"  it seems like they can hardswitch well over 25A @ 350Khz and 400V Vce. Is this a good parameter to look for to have some kind of comparison to SSTC use?

I'd been looking at FGL40N120AND based on your initial post.  Do you have fgh40n60smd also?
That graph should be quite useful.  However, I wonder if it is accurate for this particular part.  The graph for 100C case crosses 40A at roughly 190kHz.  Allowed power at 100C case is 174W.  Conduction losses would be 42W (2.1Vce * 40A * 50% duty cycle), leaving 132W for switching losses.  132W / 190kHz = 0.675mJ.  However, typical total switching loss at 175C die is listed as 1.57mJ, more than 2x higher.
Typical SSTC use is better than above case.  Presuming sufficient dead time, Eon is zero, as voltage transition is entirely due to primary current at turn-off time.

As the primary current in a sstc is limited by the primary impedance it shouldn't rise over 30A even at peak line voltage (in my case) and the average current during a pulse will be lower.

I'd been making this same presumption.  However, other discussions on the forum and some simulations indicate otherwise.  Especially with high coupling, resonant secondary current can cause higher primary current.  On the good side, this resonant current is usually close to in phase with voltage.  Can increase peak current while decreasing current at switching times.

I noticed a significant bump in performance when replacing my old IRFP460 MOSFETS with IGBT's.

Does sound familiar, but also unexpected.  IGBTs may handle more current before burning out.  Unless there is some other change, nothing would be expected to cause current to increase.  Best guess I have is that the increased delay of IGBTs happened to make primary voltage more in-phase with secondary current, so increased primary current and therefore output power.

I have used FGH60N60S's in the past at around 350-500kHz. I upgraded to the FGH75T65SHD for more current capability. They also work just fine in that range.

All depends on current, especially current at H-bridge switching times.  With primary current feedback (DRSSTC) and a driver such as UD2.7, phase lead can be adjusted to minimize switching current.  With secondary feedback (SSTC), relative phase of H-bridge switching (primary voltage) and primary current is not explicitly controlled.  Any given design may get lucky and have great phasing, with H-bridge voltage slightly leading phase of primary current.  Per a bit of simulation I ran, good primary phasing with secondary feedback may be more likely than I'd thought.  That may be the case for your coil.

Having said that I have killed A LOT of IGBT's working on my ramped coil but I think that is caused by other factors.

Perhaps some of the "other" factors were changes that happened to change primary phasing.

Dave, can you elaborate? The half cycle of the mains ramp (120v) is ~8.33ms, how can a ramped coil run at or below 1ms? My coil is tuned to be on for the entire 8.33ms cycle. If it shuts off around ~6ms it makes a loud snapping sound so I have the pulse width adjusted for the full cycle. Am I confusing pulse width with on-times?

Manz' initial post said this coil was to run in both ramped and staccato modes (unless I'm misinterpreting that initial post wording).  My comment was that staccato mode with <1ms on times may work well.  Ramped mode may work well too if primary phasing ends up good.

[Manz]

Manz' initial post said this coil was to run in both ramped and staccato modes (unless I'm misinterpreting that initial post wording).  My comment was that staccato mode with <1ms on times may work well.  Ramped mode may work well too if primary phasing ends up good.

I'm sorry for the confusion, I thought staccato and ramped sstc were the same thing.

What I'm trying to build is a sstc that detects the mains voltage zero crossing to turn on (for around 0-10ms adjustable with a potentiometer). This way it takes advantage of the rising line voltage to make straight and silent arcs.
I used a schematic I found online to detect the zero crossing (I think it's the same schematic labcoatz uses) and then the interrupter signal is fed to the driver. The driver has a diode clamp, a 74hc14 to clean the CT signal and a pair of UCC3732x.
I hope everything is clear now.

I have some used IGBTs on hand that I can use for the SSTC:

K40H1203
FGH40T120SMD
FGL40N120AND

The fgh40t120smd seems promising so I will try them out and see how they perform in a fullbridge.

[davekni]

I'm sorry for the confusion, I thought staccato and ramped sstc were the same thing.

I'm not sure how the TC community typically uses the word "staccato".  For music, it describes notes with abrupt pauses between.  I see how that could fit as a qualifier to ramped SSTC, where ramps last for less than a half line cycle with gaps between ramps.  That would differentiate it from something like my SSTC that follows rectified line voltage continually with no gaps.  Per dictionary definitions of "staccato", seems like it could apply to any interrupted SSTC or DRSSTC.  Perhaps ramped SSTCs are the only ones where a distinction is needed.

Anyone else know how "staccato" is usually used for TC description?  I'm curious.

[ZakW]

I found this definition on 4HV a while ago https://4hv.org/e107_plugins/forum/forum_viewtopic.php?id=155167



More discussion on it here - https://4hv.org/e107_plugins/forum/forum_viewtopic.php?id=156908