Have you ever tripped a breaker during normal operation ?

[panjaksli]

Well,

today I've tried running my medium DRSSTC (300A pk)from mains directly (with softstart ofc).
I've played some songs and I have tried raising the duty cycle incrementally, but all of a sudden the "house" breaker popped (16A 230V) and I thought the bridge went poof, but it didn't, to my surprise it is still working exceptionally well.
(I know for a fact, that it was not caused by continuous overload, because both my own 10A "outlet" breaker and the 16A "house" breaker popped both at the same time in a inrushy fashion.
Also worth mentioning; both breakers are analog (just normal overcurrent protection).

So my question is: is it normal for a medium sized DRSSTC (50*11cm) with 300A pk and 10mF bus capacitor to pop a breaker with inrush current during normal operation ?
My music is playing in 100ms "beats" for each tone, so is it possible, that one of the pulse discharged the bus capacitor too much and then the breakers were popped by inrush current charging the caps from let's say 150V ?

[panjaksli]

Well, I have measured the continuous power draw at 500bps, seems to be around 8A  - 10A at 230V, so my theory is probably right ?
2kW is exceptional for a simple TO-247 IGBT full bridge, right ?

Funny thing is that when I used to power the bridge from a transformer, the IGBTS blew up a lot, maybe because of the much higher on times ?

[buchtawill]

A continuous draw of 8-10 A at 230v is a LOT of power, especially for TO-247. How old is your circuit breaker? Normal breakers (heat detection) usually don't trip from sudden spikes of current, they require maybe 2-3 seconds well beyond their rated limits.

Try to determine if your breaker is functioning properly. Regardless, you should add a breaker on the input of your coil before the rectifier. I use W57-XB7A4A10-10, which is 10A @ 250VAC. Could you attach photos of your coil?

[panjaksli]

A continuous draw of 8-10 A at 230v is a LOT of power, especially for TO-247. How old is your circuit breaker? Normal breakers (heat detection) usually don't trip from sudden spikes of current, they require maybe 2-3 seconds well beyond their rated limits.

Try to determine if your breaker is functioning properly. Regardless, you should add a breaker on the input of your coil before the rectifier. I use W57-XB7A4A10-10, which is 10A @ 250VAC. Could you attach photos of your coil?

I'm using ixxh150n60c3 igbts with external diodes dsei60-06a, weird no one is talking about those IGBTS, as they have rating for 700A pulsed and the die size is 10*8mm or smth like that and they cost close to nothing in my country and are pretty fast and easy to drive. Whole bridge cost me about 20-30$.

Actually there are two breakers, one is for the outlets in the "central fusebox" or how do you call it and its 16A, and the other one I have on my PSU which is DZ30-32-10C which is 10A, also I'm pretty sure they have both a shortcircuit protection (is that electromagnetic ?), basically they disconnect within 0,1s if the current is 2x the rated value, I've managed to pop both of them at the same time. Also the inrush had to be lower than 100A or so, because the main house breaker stayed on.

Also I don't see that as a problem, it's either a breaker tripping or IGBT explosion, I would choose the breaker every time and I see this more as an curiosity.

[profdc9]

I have a full bridge of FGH60N60 running on doubled 120 VAC (330V bus voltage).  I use pulse skip mode so that the peak current is only 120 A and the pulses are somewhat longer than ordinary interrupter mode which ends the pulse when the overcurrent limit is exceeded.

Since this mode stays pretty close to the SOA for the IGBTs, and the duty cycle is pretty low, I have not had a problem with blowing the IGBTs.  However, I can easily drive the full bridge hard enough to trigger a 20A breaker.  In fact, I usually do just that when I play Bach's Toccata and Fugue.  However, it has never caused a problem because there is undervoltage protection on the driver board so it shuts off and does not allow shoot-through of the IGBTs as the bridge is powering down.  All that is required is to reset the breaker, allow the overload trip to cool, and it's back in business.  At least so far. 

So no problems here.

[Max]

Normal circuit breakers (at least here in europe) allow you to draw a pretty decent amount of overcurrent:

(Click to get to source).

So as long as you stay below 3x rated current you should be fine for quite some time. And sure enough I've drawn average currents of 25A with 40A peaks over a 16A breaker without problems.

While I only have experience with a significantly larger coil I think it kinda lies in the nature of DRSSTCs to easily blow breakers. They don't have any issues with high pulse currents, and in most cases thermals are "cool" enough to support a decent burst that draws way more current than normally. Most coils that may be designed to draw 10A continuously have no problems whatsoever to draw 20, 30 or even more amps for a few seconds. Correct me if I'm wrong but I think earlier DRSSTC designs were more "highest peak currents, lowest duty cycle", while modern coils tend to have less extreme peak currents, supporting much more than the "usual" 10% duty.


Kind regards,
Max

[panjaksli]

I have a full bridge of FGH60N60 running on doubled 120 VAC (330V bus voltage).  I use pulse skip mode so that the peak current is only 120 A and the pulses are somewhat longer than ordinary interrupter mode which ends the pulse when the overcurrent limit is exceeded.

Since this mode stays pretty close to the SOA for the IGBTs, and the duty cycle is pretty low, I have not had a problem with blowing the IGBTs.  However, I can easily drive the full bridge hard enough to trigger a 20A breaker.  In fact, I usually do just that when I play Bach's Toccata and Fugue.  However, it has never caused a problem because there is undervoltage protection on the driver board so it shuts off and does not allow shoot-through of the IGBTs as the bridge is powering down.  All that is required is to reset the breaker, allow the overload trip to cool, and it's back in business.  At least so far. 

So no problems here.

Well, youve reminded me one of my many mistakes,

I used to power my board from simple linear supply and the board doesnt have UVLO, and as I disconnected the power and tried to fire just like that, 4 IGBTs blew up in double H bridge
I've somewhat fixed this problem by using SMPS for the driver board, which supplies full power for 1-2s after being disconnected, so thats plenty of time for the bus caps to discharge safely, if I don't manage to turn off interrupter in time.
Also SMPS seems to be the better option for driver overall, since linear supply's voltage tends to really sag under load even with big filter caps and big transformer.Stable 24V+ driving voltage is a must.

And I've learned the really hard way that you cannot simply use 1ms+ pulses with drsstc and 300A limit in SKP mode, even the best TO 247 IGBTs won't handle that.

[davekni]

ixxh150n60c3 spec's look great except for the lack of internal diode.  A similar part packaged with a diode would be a real winner.

How are you measuring the 8-10A draw at 500BPS?  Is that AC current on the line or DC after rectifier?  If AC, are you using an analog or digital meter?  Many digital meters measure AC as rectified average, scaled by 1.11 to display what would be RMS for a sine-wave.  Breakers respond to true RMS, which may be twice as high (50% power factor) for a DRSSTC load.

Another possibility for breaker tripping is an interrupter malfunction causing a momentary very-long on-time.  If the arc dissipates enough energy to keep primary current under OCD, the line power draw would look like the sudden high-current event you expect occurred.

[panjaksli]

ixxh150n60c3 spec's look great except for the lack of internal diode.  A similar part packaged with a diode would be a real winner.

How are you measuring the 8-10A draw at 500BPS?  Is that AC current on the line or DC after rectifier?  If AC, are you using an analog or digital meter?  Many digital meters measure AC as rectified average, scaled by 1.11 to display what would be RMS for a sine-wave.  Breakers respond to true RMS, which may be twice as high (50% power factor) for a DRSSTC load.

Another possibility for breaker tripping is an interrupter malfunction causing a momentary very-long on-time.  If the arc dissipates enough energy to keep primary current under OCD, the line power draw would look like the sudden high-current event you expect occurred.

Well, getting cheap diodes is the key, I'm doing the to-247 igbt + diode sandwich (diode is mounted on top of a igbt with sticky thermal tape), so there is really small trace lenght, also heat transfer seems to be fine and I never had a single diode fail. Might not be suitable for higher powers...

I'm measuring AC current draw with a MT-5211 meter, I don't think it has true RMS. (also 8A was for continous use, like normal interrupter single tone all the time)

But when playing music with high pitched tones (above 1kHz) the arcs are much thicker, which is logical, since more pulses are in single note, thus are drawing much more power.
Also each note is set to last exactly 100ms, if I had managed to discharge the capacitors enough during the 0 rectifier voltage, I could have basically made an artificial inrush.

My interrupter is kind of dumb; midi player with a sound font with 4% duty cycle square wave; then fed into simple RC filter with time constant of 100us, then it's converted by "optical remote" from aliexpress, which can lenghten/shorten the on time even further.

But afterall I can't really complain, my gear consists of soldering iron and 50V usb oscilloscope, that's all I have.

[davekni]

With diodes mounted on top, they have little heat-sinking.  That makes me think of your previous quote:

"And I've learned the really hard way that you cannot simply use 1ms+ pulses with drsstc and 300A limit in SKP mode, even the best TO 247 IGBTs won't handle that."

With SKP mode, the diodes conduct significant current.  Are you sure the diodes didn't fry first, then the shorted diodes fried IGBTs?

MT-5211 doesn't claim "true-RMS", so likely isn't.  So, when you measure 8-10A, it is more likely 16-20A RMS.  Breaker tripping isn't too surprising.

[panjaksli]

Diodes are always fine after bridge explosion, I haven't even really noticed that they are heating up, the recovery time doesn't change much with temperature and saturation voltage is also lower with temperature. I think I can cool 5W per diode with that mounting.
The main culprit really is that those IGBTs cant handle 300A for 10ms straight (its qcw capable controller, thus using 10ms bangs for normal drsstc should and will result in explosion in SKP mode)

[panjaksli]

I don't think it performs that bad tho...