Author Topic: IGBTs keep dying (un-solved again)  (Read 2962 times)

Offline davekni

  • High Voltage Senior
  • *****
  • Posts: 553
  • Karma: +30/-0
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: IGBTs keep dying
« Reply #20 on: May 13, 2020, 03:52:05 AM »
Yes, that layout looks great!
David Knierim

Online TMaxElectronics

  • High Voltage Experimenter
  • **
  • Posts: 86
  • Karma: +3/-0
    • View Profile
    • My random (and very empty) electronics blog
Re: IGBTs keep dying
« Reply #21 on: June 18, 2020, 07:14:06 PM »
Ok now that everything has been re built, i have found some more issues. The CT transformer issues from my other post are now fixed and the controller seems to work, at least according to my measurements.
But unfortunately something is still wrong, and the IGBTs still die. The last two faults only occured on 2 out of 4 transistors though.

I now setup probes in a more permanent way and made a little shield for my scope from scheet metal, so i can test while the coil is running at full power.
And i might have found another issue:


It looks to me like the top transistor might be de-saturating at the high-ish current (here the peak was at around 330A), ignore the noisy peaks at the switching point, I was using the antenna earth lead.
I was probing from the emitter (directly on the screw) of the bottom transistor to the center point output, so the waveform should be symmetrical, but obviously isn't.
When this failure occurred, the top side transistor died more or less silently (with obvious signs of overheating of the IGBT die inside), and the bottom one blew up violently.

So what i think happened was: first the top transistor heated up from de-saturation, died and shorted out with not all that much energy dissipated, then the bottom transistor switched on and shorted VBus+ and VBus-, killing itself in the process and dumping all of the energy from the caps into itself.
Do you think this is a viable course of events?

Or might this just be a measurement error aswell?
« Last Edit: June 19, 2020, 01:57:12 PM by TmaxElectronics »

Offline davekni

  • High Voltage Senior
  • *****
  • Posts: 553
  • Karma: +30/-0
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: IGBTs keep dying
« Reply #22 on: June 19, 2020, 09:58:56 PM »
Yes, I'm confident you have figured out this failure accurately.  Notice the shape of the H-Bridge output voltage trace on top.  The dip isn't a negative half-sine-wave.  Instead, it starts dropping slowly, peaks in the middle, then ends with little drop.  That's exactly what to expect for desaturation, low resistance at low current and higher resistance at high current.

Just occurred to me now; that makes perfect sense for your boot-strap driver (L6494).  The boot-strap capacitors (C1 and C5 in your schematic) are charged when the corresponding H-Bridge output is low.  At high current, "low" isn't as low as normal.  Any low-side Vce drop gets subtracted from C1/C5 charging voltage, reducing Vge of the high-side parts.  Since you are starting with only 15V (minus diode-drop of L6494's internal bootstrap charging diode), any further drop will be quite problematic.

If switching away from current zero-crossing, either lots of phase lead or lag, diode conduction within the low-side IGBTs might allow C1 and C5 to charge fully.  IGBT diode conduction is negative Vce.  With well-adjusted just-enough phase lead, diode conduction will be minimal, so C1 and C5 may not get charged sufficiently.

Congratulations on your great analysis work!

The ideal solution would be floating gate supplies, but that's a major change.  Or transformer-driven gates.  For now, you could boost the gate supply from 15V to 20V if the L6494 can handle the additional power dissipation.  Even then, with 3-4Vce for the low-side IGBTs, you'll barely be at 15Vge on the high-side.  Most coilers here drive gates to ~+-20V to squeeze out better high-current performance.  (I use 0-18V, using isolated gate-drive chips and floating supplies, not boot-strap.)
David Knierim

Online TMaxElectronics

  • High Voltage Experimenter
  • **
  • Posts: 86
  • Karma: +3/-0
    • View Profile
    • My random (and very empty) electronics blog
Re: IGBTs keep dying
« Reply #23 on: June 19, 2020, 10:37:28 PM »
Quote
The boot-strap capacitors (C1 and C5 in your schematic) are charged when the corresponding H-Bridge output is low.  At high current, "low" isn't as low as normal.
That makes sense :D So I will probably try the 20V gate supply.

But even though I think that is in deed a problem, there is something else odd about my measurements. I just measured from VBus+ (top collector) to top emitter, with the ground reference being VBus+, and got this:

Here it looks like the bottom one is desaturating... So could this be an additional measurement error? I don't think the bottom transistor would actually do that. I also checked the top IGBT's gate voltage and it did get to 14.5V according to the scope.

My theory is that the offset at high current in this measurement would be from the interconnects (through the PCB and the standoffs, even if it seems way to big for that), and top side desaturation would only occur if the idle output of the driver of the particular side of the bridge happened to be positive, so the bootstrap capacitor had no chance to charge back up during the interrupter off period. Then when the signal comes in to start switching, the voltage would not be able to recover, the IGBT doesn't get enough gate drive and then goes boom.

Oh and I also filmed me taking apart the exploded IGBTs, I'll post a link here once thats on youtube ;)

Offline Hydron

  • Administrator
  • High Voltage Engineer
  • *****
  • Posts: 345
  • Karma: +11/-0
    • View Profile
Re: IGBTs keep dying
« Reply #24 on: June 20, 2020, 01:11:16 AM »
You might just be seeing the bus voltage drooping under load, though the amount is fairly large. What are you seeing after the interrupter signal stops and the tank energy is being dumped back into the bus? Does the shape of the waveform flip?

Online TMaxElectronics

  • High Voltage Experimenter
  • **
  • Posts: 86
  • Karma: +3/-0
    • View Profile
    • My random (and very empty) electronics blog
Re: IGBTs keep dying
« Reply #25 on: June 20, 2020, 01:49:11 AM »
Quote
You might just be seeing the bus voltage drooping under load
I was initially thinking that too, but then the voltage would recover more slowly, and not follow the current waveform almost perfectly would it?
I'll make sure to measure the bus voltage directly across the IGBTs tomorrow too (and also measure the CE voltage with a better probe connection) :D

Here is the Video if anybody is interested:

EDIT:
I just realised I had an isolated voltage source, with two channels, that is intended for gate driving:

That old induction cooker is really the gift that keeps on giving. I just had to expertly remove the part of the circuit board, so I can use it :P
« Last Edit: June 20, 2020, 05:03:08 PM by TmaxElectronics »

Offline davekni

  • High Voltage Senior
  • *****
  • Posts: 553
  • Karma: +30/-0
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: IGBTs keep dying
« Reply #26 on: June 21, 2020, 12:28:06 AM »
Looks like my last reply failed to post.  Appears I'm wrong about desaturation given your new scope traces.  Looking back at the previous H-Bridge output traces more carefully, only the start of the droop is rounded, not the end, so desacturation didn't quite fit even initially.

My thought now is that you have a resonance between local VBus film cap(s) and VBus wiring inductance.  That would be reasonable only if this scoping is being done without the bulk caps mounted locally in your nice mechanical design.  Did you remove the bulk electrolytic caps or connect them remotely for scoping?

Hopefully the cause will be more clear with a good scope trace of local VBus (low-side emitter for scope ground, high-side collector probed).
David Knierim

Online TMaxElectronics

  • High Voltage Experimenter
  • **
  • Posts: 86
  • Karma: +3/-0
    • View Profile
    • My random (and very empty) electronics blog
Re: IGBTs keep dying
« Reply #27 on: June 21, 2020, 12:58:26 AM »
Quote
Did you remove the bulk electrolytic caps or connect them remotely for scoping?
No they were connected normally.

Quote
Hopefully the cause will be more clear with a good scope trace of local VBus (low-side emitter for scope ground, high-side collector probed).
That will have to wait until monday, UPS didn't manage to get the replacement driver ICs in today. I burned out the last one while measuring the temperature while running at 20V, because i accidentally shorted some pins with my thermocouple  ::)
My idea with the isolated power-supply unfortunately won't work either, because as soon as the output start getting pushed up and down by the transistors, the circuit gets confused and shuts down... So for now I have made a small LM317 20V converter and will hook that up to a 32V power supply. The drivers didn't take more than 80mA in testing so a linear regulator should be good enough.
I still believe that desaturation while switching after an idle high period would be an issue. The RC time constant of the bootstrap network is 10u sF, so it would not have time to charge up much at all during the cycles where current is still low and end up at below 14V easily, at which point desaturation would start to occur, so the way I tried to explain the last failure might not be wrong after all.

Quote
My thought now is that you have a resonance between local VBus film cap(s) and VBus wiring inductance.
Do you mean similar to this? (excuse the incredible artistic skill)

That would mean though that Fres of the Bus would have to be very close to the Fres of the primary circuit though right? If the frequency was much higher (which i would expect) there would be ringing instead of an actual oscillation matching the drive frequency wouldn't it?

Offline davekni

  • High Voltage Senior
  • *****
  • Posts: 553
  • Karma: +30/-0
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: IGBTs keep dying
« Reply #28 on: June 21, 2020, 02:09:26 AM »
Yes, exactly as you sketched.  (Better than my sketching ability using a computer mouse.)  That's twice the coil frequency, which is the dominant ripple current frequency on VBus.  Yes, much too low since you have the bulk electrolytic caps connected with low inductance.  If that 2x frequency does show up between low-side emitter and high-side collector, then there must be some problem with high VBus impedance.  It's just hard to imagine what would cause that given your nice low-inductance layout.

Aren't both low-side IGBTs enabled during idle times between bursts?  If so, the bootstrap capacitors will be continuously charged when idle.  If not, then yes, low bootstrap voltage at the start of bursts could be causing failures.

Your 32V supply is likely to be problematic as well.  Most supplies are not designed to handle common-mode swings of a few hundred volts at 100kHz.  Even if it is an old linear supply with line-frequency transformer, the capacitance between transformer windings will add load to your H-Bridge and inject 100kHz common-mode into your power line making scoping difficult.  (The line signal will couple into lots of things.)  Floating gate supplies are almost always switching regulators with special transformers designed for low capacitance between primary and secondary windings.
David Knierim

Online TMaxElectronics

  • High Voltage Experimenter
  • **
  • Posts: 86
  • Karma: +3/-0
    • View Profile
    • My random (and very empty) electronics blog
Re: IGBTs keep dying
« Reply #29 on: June 21, 2020, 03:06:30 AM »
Quote
Your 32V supply is likely to be problematic as well
Sorry, I ment to say that the 20V supply replaces the 15V one for the ENTIRE drive circuitry. That way the 3-4V drop should still allow the capacitor to charge to >14V, as you said.

Quote
Aren't both low-side IGBTs enabled during idle times between bursts?
Actually no, the comparator has a transparent latch on the output which I use to stop switching. So the transistors remain in whatever state they were at the end of the burst, which is more or less random, especially when the interrupter frequency is variable.

When I re design the gate drive board I will make sure to get a proper isolated supply for the top gates. That will take a few months though as I already have some exams again next month  :'(

Offline davekni

  • High Voltage Senior
  • *****
  • Posts: 553
  • Karma: +30/-0
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: IGBTs keep dying
« Reply #30 on: June 21, 2020, 04:41:34 AM »
My mistake for presuming your 32V supply was for floating gate drive.  For DC into driver chips, that's fine.

Stopping bootstrap drivers in the high state is definitely a problem.  (There are a few driver chips with internal charge-pumps to keep the bootstrap supply charged, but those are rare and usually for ~60V operation, not 500V.)  I'd suggest fixing the logic before running any more tests with bootstrap driver chips.  (Or at least add a fast/schottky diode across the driver chip's internal bootstrap charging diode so that you have some hope of charging the bootstrap capacitor in one half-cycle at the start of each burst.)

That still leaves the puzzle of your odd droop in H-Bridge output.  That puzzle will need to wait for more scope traces.
David Knierim

Online TMaxElectronics

  • High Voltage Experimenter
  • **
  • Posts: 86
  • Karma: +3/-0
    • View Profile
    • My random (and very empty) electronics blog
Re: IGBTs keep dying
« Reply #31 on: June 21, 2020, 11:37:25 AM »
Quote
I'd suggest fixing the logic before running any more tests with bootstrap driver chips.
That would have to be done for the next version of the PCB, as it would required fairly complex modification of the logic signal path. I'm using the inverted and non inverted outputs of the comparator so I'd need at least two flip flops.
I have actually already got an external bootstrap diode, that's not in the schematic I posted here first (though it is in the updated one further down the thread). The chips have an internal resistance of >100ohm, which was waaay to high, so i added a 10ohm resistor and a schottky diode.

I will also change the bootstrap resistor to 1 ohm, then I have 5RC time constants in one low switching pulse, even after the recovery of the diode. Or is there a disadvantage to having it be so low? The pulse current will be fairly high, but that would only be there for a very short time, and is actually still withing the rating of the diode for a non repetitive event (or at least semi-non repetitive).

Offline davekni

  • High Voltage Senior
  • *****
  • Posts: 553
  • Karma: +30/-0
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: IGBTs keep dying
« Reply #32 on: June 21, 2020, 08:01:05 PM »
Yes, most diodes can handle the brief surge fine.  The potential issue is a momentary drop in the 15 or 20V supply with the sudden high current load.  If you have enough local bypassing, say 10x your 1uF bootstrap capacitance, with very low ESR, then it should be fine.  Many small ~10uF electrolytic caps have fairly high ESR, so wouldn't maintain 15/20V during that current spike.

Also, the peak current could possibly couple to logic signals depending on routing.  But it is certainly worth a try.
David Knierim

Online TMaxElectronics

  • High Voltage Experimenter
  • **
  • Posts: 86
  • Karma: +3/-0
    • View Profile
    • My random (and very empty) electronics blog
Re: IGBTs keep dying
« Reply #33 on: June 22, 2020, 03:34:59 PM »
I just measured the VBus and that is actually the source of those "dips". I did however forget to take a snapshot of the waveform ::) , so you will have to trust me once again
The weirdest thing is, that once the switching stops, the oscillations also die down to almost nothing, even though the current is still at almost the same amplitude...
I also checked with the probe shorted (in the same location) and the induced voltage was negligible too (kind of expectable with a signal of >100V)

Now my question is, why is the frequency so low ??? all of the caps were attached and the only loose wires were for the rectifier connection to the reservoir caps.

I don't think the oscillations are cause for any trouble at the reduced primary voltage though, as they remain well clear of the maximum of the components. So my biggest suspect for the last death is still insufficient gate drive.
The fix I made for that unfortunately ment that my resistive divider for the drive signal let out the magic smoke... guess I'll have to increase the resistor values, I was trying to go for the lowest possible impedance to avoid noise making its way in, but apparently an 8th watt resistor doesn't like dissipating 1W ;D.

Quote
The potential issue is a momentary drop in the 15 or 20V supply with the sudden high current load
The capacitors I have on the driver board are 100uF + 15uF tantalum and 100nF+4.7uF+10uF next to each driver chip, i think that should be good enough.
« Last Edit: June 22, 2020, 03:36:49 PM by TmaxElectronics »

Offline davekni

  • High Voltage Senior
  • *****
  • Posts: 553
  • Karma: +30/-0
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: IGBTs keep dying
« Reply #34 on: June 22, 2020, 10:08:10 PM »
I presume the frequency you measured on VBus is twice the primary oscillation frequency (about 170kHz on VBus).  If VBus impedance is low and measurement is good, then the only remaining possibility is that current is high.  Is there any chance that the driver chips aren't holding Vge low enough for the IGBTs that are supposed to be off?  That would make high current through one or both half-bridges, and fry IGBTs.  This theory could fit with your observation that the VBus ripple goes away after driving terminates while primary resonant current is still flowing.

Perhaps you could get some idea of this by monitoring VBus average (DC) current draw without the secondary coil in place.  Without the secondary making arcs, there should be fairly small total current draw, as much of the primary resonant energy from each burst is returned to VBus after the burst.

I'd also suggest scoping VBus at different points, across the local film capacitors and again at the bulk caps, to see how the ripple amplitude decreases farther back in the VBus supply circuitry.
David Knierim

Online TMaxElectronics

  • High Voltage Experimenter
  • **
  • Posts: 86
  • Karma: +3/-0
    • View Profile
    • My random (and very empty) electronics blog
Re: IGBTs keep dying
« Reply #35 on: June 23, 2020, 08:42:53 PM »
more measurement done and this time i remembered to take snapshots :P But i didn't manage to reach the film capacitor terminals directly
I put them on imgur, so the post size isn't crazy, since i have posted four: https://imgur.com/a/dmZyL4R

The three obvious observations are:
  • The highest amplitude oscillations seem to be across the IGBT terminals
  • There is almost no difference in amplitude between the two different sides of the bridge. If the current was high on only one I would expect a noticeable difference between the two.
  • With no secondary installed the power consumption remains at around 1A, after peaking while the capacitors are charging up.
    But I believe that is my meter getting confused by noise, as the current continues to flow, even after the power to the primary was turned off (the current was almost the same immediately after power off and then slowly dropped as the primary caps discharged and noise decreased)
    The other thing that makes me believe this, is the fact that even after power has been disconnected (of course only from the primary and not the controller), there continued to be current pulses for a while. I turned off the controller at ~20s after primary power off and there was still some audible clicking
    I was testing at only 10Hz  and 125us on time though
Was the list unnecessary? Probably but i wanted to try the feature out :P

The other thing is that the required inductance to reach 150 KHz is only about 40nH (i Have four 4uF film caps), but that still seems excessive for a very short path through a poured plane. According to the saturn PCB toolkit the a 25mm wide trace (smaller than it would probably be) has an inductance of <1nH/cm so that doesn't make sense. And inductance in the spacers would not show with such a dramatic voltage on the supply end of them would it?

Also I don't understand how the driving of the primary LC circuit could even trigger oscillations at 2x its own Fres. Perhaps the significant pre-switching causing the flowing current that now goes through the bridge "the other way around" to charge the capacitors, and them getting discharged when the current flows normally?
« Last Edit: June 23, 2020, 11:26:10 PM by TMaxElectronics »

Offline davekni

  • High Voltage Senior
  • *****
  • Posts: 553
  • Karma: +30/-0
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: IGBTs keep dying
« Reply #36 on: June 24, 2020, 06:01:14 AM »
Nice post with detailed data!

With 4 x 3.9uF and 5.83us period, I come up with 55nH for resonance.  Looking back at your mechanical design, I think that actually is reasonable for the standoffs and bulk-cap bus bars and bulk-cap internal inductance.  That resonance fits your scope traces perfectly.  (At first I was confused by the rapid change from 170kHz to 85kHz after enable ends.  Then recalled your drive design that leaves one IGBT on each half-bridge enabled.  That makes VBus current 1x resonant frequency after enable ends.  It also makes current decay more slowly.)

Hadn't previously paid attention to this detail, but I see that only one standoff at each end is conductive.  It's still lower inductance than many designs (including my own) that use wires/cables to connect the bulk caps.  (I compensate by having 360uF of film capacitance, twelve 30uF 4-lead caps.  Still had to redo wiring to bulk caps to pair +/- leads to reduce inductance.)

It's just bad luck that you happened to hit 2x primary resonance with your VBus layout.  Rarely happens.  It may not cause huge problems, but it's likely worth de-tuning some.  Longer stand-offs would lower frequency.  Perhaps target half way between 85kHz and 170kHz.  Or, you could use larger lower-voltage film caps in the same family, but that costs more money.  Or, you could lower inductance by making the bulk-cap bus bars into angle-pieces that bend down along the edges of the film caps to almost touching the ECB.  (Wouldn't need to redo the existing bars.  Could just add copper foil bent into an angle to the existing bars.)

Finally, in systems where any high-Q resonance needs to be avoided, Q is reduced by adding series resistors to some of the film caps.

BTW, the reason for 2x primary resonant frequency is that the H-Bridge is essentially a reverse bridge rectifier.  Ripple on a 50Hz line supply bridge is 100Hz.  In the same way, both the positive and negative current peaks on the primary are drawing current from VBus.  The zero-crossing points are drawing no current from VBus.

Edit:  Examined the traces after enable pulse end in more detail.  Fits 55nH quite well.  Voltage at the bulk caps is close to in-phase with current, at about 3Vpp for 300App, so about 0.010 ohms ESR for all four bulk caps in parallel.  Voltage at the IGBTs is almost 90 degrees leading current as would be expected for inductance.  It's about 12Vpp for 300App, or about 0.040 ohms.  55nH at 85kHz would be 0.030 ohms impedance.  However, since the caps have ~-0.12 ohms at that frequency and phase, the net parallel impedance should be 0.040 ohms.

If you do decide to replace film caps, you could change just two to B32776T5126K000 12uF 575V.  That would 32uF total, twice your existing 15.6uF.  Twice is about perfect.  Four times would make resonance hit 1x primary frequency, which might still cause some trouble.  Twice gets you nicely between 1x and 2x primary frequency, sqrt(2) times primary frequency.
« Last Edit: June 24, 2020, 06:57:07 AM by davekni »
David Knierim

Online TMaxElectronics

  • High Voltage Experimenter
  • **
  • Posts: 86
  • Karma: +3/-0
    • View Profile
    • My random (and very empty) electronics blog
Re: IGBTs keep dying
« Reply #37 on: June 24, 2020, 09:48:25 AM »
Quote
BTW, the reason for 2x primary resonant frequency is that the H-Bridge is essentially a reverse bridge rectifier.
That's what I was trying to explain at the end of my last post :)

So the frequency of that is fixed by the bridge, and the amplitude of the voltage swing is so high because we hit the resonant frequency of the VBus quite closely correct?

The standoffs for the Bulk caps are brass, so shouldn't have crazy amounts of inductance, but 55nH may well be the case.

Quote
If you do decide to replace film caps, you could change just two to B32776T5126K000 12uF 575V.
I will probably just do this. Its the easiest solution, since making new standoffs would mean I have to make them, and i don't have any more brass rods, so that would cost some money aswell...
But that will have to wait until next month, this months budget for experimentation has already been exhausted :P

I'll keep testing for other potential faults at 120V VPrim for now, and then go full power once those replacements arrive.

On another note (not sure if I should start another thread), has anybody used the Micsig DP10013 High Voltage Differential Probe? I might buy one, so I can measure things even when the primary has a main earth reference.

EDIT: I have now been playing music on the coil for probably an hour, and so far nothing (and luckily also nobody :P) has died :D I didn't push my luck though and stayed under 1kW of output power.
Here are some videos from the test, but absolute potato quality. I didn't take a proper camera into my workshop with me:

I have also made some better videos, they are on my youtube channel. Don't want to put even more stuff in this post :D
So even if it is not completely finished i wanted to say thanks to everyone for the help, without you I wouldn't have got it to work :)
« Last Edit: June 24, 2020, 09:18:54 PM by TMaxElectronics »

Offline davekni

  • High Voltage Senior
  • *****
  • Posts: 553
  • Karma: +30/-0
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: IGBTs keep dying
« Reply #38 on: June 24, 2020, 10:24:35 PM »
I've looked at that Micsig DP10013 probe listings, along with the occasional used Tektronix P5205 that show up on EBay.  Haven't personally used the Chinese versions, and it's been years since I used the Tek one (at work).  I've heard that the cheap ones have somewhat more ring to their step response, but likely would still work for most testing.  If they are like the P5205, it's a pair of well-matched single-ended high-impedance attenuators feeding a differential amplifier.

Very nice polyphonic sound response!

Edit:  Forgot that I was going to explain the inductance a bit more.  Assume the ECB is perfect, large planes with "zero" separation.  Also assume the bus bars on the bulk caps are perfect, "zero" gap between them, so no inductance for current going from one bar to the other.  That still leaves one relatively-large loop that is the dominant source of inductance.  Current between the bulk caps and ECB travels down one standoff, across the ECB, up the other standoff, and across the bus bars.  The area for that loop is the standoff length times the distance between standoffs.

Of course, converting geometry to inductance is a complex 3D problem.  For simple geometries, inductance is proportional to area / length, where length is the path length for the magnetic field.    In your case, path length varies from fairly short looping around the standoffs to longer around the bus bars or ECB towards the center of the loop area.  Although not possible to calculate precisely, 55nH seems about right for that loop between the two standoffs.
« Last Edit: June 25, 2020, 04:16:13 AM by davekni »
David Knierim

Offline Steve Ward

  • High Voltage Enthusiast
  • *
  • Posts: 30
  • Karma: +4/-0
    • View Profile
Re: IGBTs keep dying
« Reply #39 on: June 26, 2020, 07:10:28 PM »
Quote
On another note (not sure if I should start another thread), has anybody used the Micsig DP10013 High Voltage Differential Probe? I might buy one, so I can measure things even when the primary has a main earth reference.

Ive got 2 of them and think they're great.  That being said, its important to know their limits, but for IGBT switchings they should have enough bandwidth to give an accurate enough measurement.

Given the quasi-un-reliability, i would be investigating gate waveforms for voltage transients, especially at high voltage and high current. If your driver is lagging and you have hard diode commutation, this can really disrupt the operation of gate driver ICs because of the very high dv/dt seen at diode recovery time (though it looks like your driver now has phase lead to give switching before current zero crossing).

One more wild guess: With respect to boot-strap diodes, i have noticed that not all high speed silicon rectifiers are created equally.  I've had serious reliability problems with some diodes, especially ones rated higher than 600V, they appear to have much worse recovery properties that turn out to matter when switching at 100's of khz.  In my case i was using boot-strap drive for SiC mosfets, and I'd notice after some minutes of operation things would start to act funny and explode.  The fact that the bootstrap diodes were blown off the PCB was a clue...  I tested several diodes and found the US1J-13-F (600V) was perfectly fine coping with the high switching rates, but the US1M-13-F (1kV) would overheat and fail.  I tried a few other 1kV high speed diodes, all suffered from excessive heating.  My assumption is that whatever structural change required to give a diode higher blocking voltage is also hurting its ability to switch off.  I proved that the heating was not just conduction/supply current and was greatly related to switch frequency and bus voltage.  My eventual solution was to series 2 of the 600V diodes to handle the voltage, and bump up the power supply to compensate.   


High Voltage Forum

Re: IGBTs keep dying
« Reply #39 on: June 26, 2020, 07:10:28 PM »

 


* Recent Topics and Posts

post Re: Rogowski current probe
[Laboratories, Equipment and Tools]
Hydron
Today at 05:27:37 PM
post Re: DRSSTC Tuning for Music vs Big Sparks?
[Dual Resonant Solid State Tesla coils (DRSSTC)]
Zipdox
Today at 04:32:53 PM
post Re: DRSSTC Tuning for Music vs Big Sparks?
[Dual Resonant Solid State Tesla coils (DRSSTC)]
Mads Barnkob
Today at 04:17:51 PM
post Re: DRSSTC Tuning for Music vs Big Sparks?
[Dual Resonant Solid State Tesla coils (DRSSTC)]
GKnapp
Today at 02:06:04 PM
post Re: Rogowski current probe
[Laboratories, Equipment and Tools]
flyglas
Today at 10:57:58 AM
post Re: DRSSTC Tuning for Music vs Big Sparks?
[Dual Resonant Solid State Tesla coils (DRSSTC)]
Mads Barnkob
Today at 08:55:27 AM
post Re: DRSSTC Tuning for Music vs Big Sparks?
[Dual Resonant Solid State Tesla coils (DRSSTC)]
davekni
Today at 05:19:45 AM
post Re: Rogowski current probe
[Laboratories, Equipment and Tools]
klugesmith
Today at 04:46:12 AM
post Re: Easy to build Polyphonic MIDI Interrupter
[Computers, Microcontrollers, Programmable Logic, Interfaces and Displays]
Max
Today at 01:14:47 AM
post Re: Rogowski current probe
[Laboratories, Equipment and Tools]
flyglas
September 25, 2020, 09:41:10 PM
post Re: DRSSTC Tuning for Music vs Big Sparks?
[Dual Resonant Solid State Tesla coils (DRSSTC)]
Mads Barnkob
September 25, 2020, 08:40:39 PM
post Re: Easy to build Polyphonic MIDI Interrupter
[Computers, Microcontrollers, Programmable Logic, Interfaces and Displays]
Mads Barnkob
September 25, 2020, 08:26:52 PM
post Re: FPGA speeds these days?
[Computers, Microcontrollers, Programmable Logic, Interfaces and Displays]
SteveN87
September 25, 2020, 06:56:41 PM
post Re: FPGA speeds these days?
[Computers, Microcontrollers, Programmable Logic, Interfaces and Displays]
klugesmith
September 25, 2020, 06:15:47 PM
post Re: FPGA speeds these days?
[Computers, Microcontrollers, Programmable Logic, Interfaces and Displays]
SteveN87
September 25, 2020, 01:56:27 PM
post Re: FPGA speeds these days?
[Computers, Microcontrollers, Programmable Logic, Interfaces and Displays]
TMaxElectronics
September 25, 2020, 11:55:05 AM
post Re: FPGA speeds these days?
[Computers, Microcontrollers, Programmable Logic, Interfaces and Displays]
Weston
September 25, 2020, 09:18:51 AM
post One NOS ac flyback is nice. Finding its identical twin 3 months later is awesome
[Transformer (Ferrite Core)]
extremeodd
September 25, 2020, 03:43:26 AM
post Re: FPGA speeds these days?
[Computers, Microcontrollers, Programmable Logic, Interfaces and Displays]
davekni
September 25, 2020, 03:42:56 AM
post FPGA speeds these days?
[Computers, Microcontrollers, Programmable Logic, Interfaces and Displays]
klugesmith
September 25, 2020, 02:13:40 AM
post Re: Server PSU OPV Hack - Fujitsu DPS-800GB-5 A
[Electronic Circuits]
klugesmith
September 24, 2020, 03:02:00 PM
post Re: Server PSU OPV Hack - Fujitsu DPS-800GB-5 A
[Electronic Circuits]
v.coelho
September 24, 2020, 01:11:27 PM
post Re: Info needed: probably old military High Voltage Multiplier?
[Voltage Multipliers]
Zipdox
September 24, 2020, 10:25:42 AM
post Re: Info needed: probably old military High Voltage Multiplier?
[Voltage Multipliers]
johnnyzoo
September 24, 2020, 07:32:43 AM
post Re: Current transformer for narrow spaces
[Laboratories, Equipment and Tools]
klugesmith
September 23, 2020, 11:20:53 PM
post WTS [US]:"Terry filter" varistors for sale
[Sell / Buy / Trade]
MRMILSTAR
September 23, 2020, 10:41:54 PM
post Re: Info needed: probably old military High Voltage Multiplier?
[Voltage Multipliers]
Zipdox
September 23, 2020, 10:25:07 PM
post Info needed: probably old military High Voltage Multiplier?
[Voltage Multipliers]
HighVoltageRulezz
September 23, 2020, 10:06:30 PM
post Re: Server PSU OPV Hack - Fujitsu DPS-800GB-5 A
[Electronic Circuits]
petespaco
September 23, 2020, 04:26:34 PM
post Re: 6" Coil Update now 72" and better lighting
[Spark Gap Tesla Coils (SGTC)]
klugesmith
September 23, 2020, 03:06:28 PM
post Re: Server PSU OPV Hack - Fujitsu DPS-800GB-5 A
[Electronic Circuits]
Zipdox
September 23, 2020, 01:34:42 PM
post Server PSU OPV Hack - Fujitsu DPS-800GB-5 A
[Electronic Circuits]
v.coelho
September 23, 2020, 01:11:50 PM
post 6" Coil Update now 72" and better lighting
[Spark Gap Tesla Coils (SGTC)]
Bradselph
September 23, 2020, 09:03:56 AM
post Re: steam engine
[Capacitor Banks]
plasma
September 23, 2020, 12:53:49 AM
post Re: Help for people buying the "12-48 Volt 1800/2500 Watt ZVS induction Heater"
[Electronic Circuits]
extremeodd
September 23, 2020, 12:35:16 AM
post Re: Help for people buying the "12-48 Volt 1800/2500 Watt ZVS induction Heater"
[Electronic Circuits]
petespaco
September 22, 2020, 08:15:40 PM
post Re: Am I Misunderstanding of DC Blocking Capacitors
[Solid State Tesla Coils (SSTC)]
klugesmith
September 22, 2020, 05:59:21 PM
post Re: Help for people buying the "12-48 Volt 1800/2500 Watt ZVS induction Heater"
[Electronic Circuits]
klugesmith
September 22, 2020, 05:42:28 PM
post Re: Am I Misunderstanding of DC Blocking Capacitors
[Solid State Tesla Coils (SSTC)]
T3sl4co1l
September 22, 2020, 06:48:34 AM
post Re: Help for people buying the "12-48 Volt 1800/2500 Watt ZVS induction Heater"
[Electronic Circuits]
petespaco
September 22, 2020, 01:48:11 AM
post Re: steam engine
[Capacitor Banks]
klugesmith
September 21, 2020, 11:58:09 PM
post Re: steam engine
[Capacitor Banks]
plasma
September 21, 2020, 11:21:28 PM
post Re: steam engine
[Capacitor Banks]
HiVi
September 21, 2020, 08:30:22 PM
post Re: steam engine
[Capacitor Banks]
klugesmith
September 21, 2020, 07:11:30 PM
post Re: Am I Misunderstanding of DC Blocking Capacitors
[Solid State Tesla Coils (SSTC)]
klugesmith
September 21, 2020, 06:57:13 PM
post Re: DRSSTC Tuning for Music vs Big Sparks?
[Dual Resonant Solid State Tesla coils (DRSSTC)]
Uspring
September 21, 2020, 05:50:56 PM
post Re: Easy to build Polyphonic MIDI Interrupter
[Computers, Microcontrollers, Programmable Logic, Interfaces and Displays]
Max
September 21, 2020, 02:59:47 PM
post Re: VTTC on GU-81M
[Vacuum Tube Tesla Coils (VTTC)]
Continuum
September 21, 2020, 12:18:10 PM
post Re: steam engine
[Capacitor Banks]
plasma
September 21, 2020, 09:49:57 AM
post Re: steam engine
[Capacitor Banks]
Zipdox
September 21, 2020, 09:18:53 AM

Sitemap 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 
SimplePortal 2.3.6 © 2008-2014, SimplePortal