IGBT keep failing with mains, but rated at 600V ???

Thread Starter

Robin66

Joined Jan 5, 2016
275
Hi. Below is my circuit. It is to be a motor controller but at the moment I'm testing with a 300W lamp in a domestic environment. I have marked the max. voltage ratings in blue. the problem is Q1 (FGD3N60LSD) keeps failing. C-E becomes permanently low impedance ~100ohm. The mains is 240VAC, so that's 330Vpk. To allow for spikes I have picked devices rated at 600V for BRIDGE1, D1 and Q1. Q2 is omitted at the moment. Fortunately the Q1 failure doesn't affected the rest of the circuit.

The failure happens immediately when the circuit is connected to the mains. I've programmed it to flash for 10ms every 500ms. A few times it has worked when first connected, then the next time I connect it to the mains, it fails. The duty cycle is low so nothing's getting hot when it's working correctly. When Q1 fails, Q1 quickly gets hot.

I tried a TVS rated at ~400V across Q1's C-E junction. This allowed the circuit to work for about 10 connections, then it failed on the 11th. so I think the failure is due to a voltage spike across Q1 C-E. But I don't understand why my overrating of 600V isn't sufficient to prevent this nor why the TVS (400V breakdown) didn't offer permanent protection. Also, why does it only fail when initially connected to the mains? I'm thinking it might have something to do with switch bounce and stray inductance. Any help/advice gratefully received.

IGBT failure.JPG
 

Beau Schwabe

Joined Nov 7, 2019
155
Only Q1 and not Q2? ... Ahh I see Q2 is omitted for the moment

It could be a Zero cross issue... you might need to turn ON the IGBT during the Zero Cross

Also, I don't see anything that limits the Gate voltage to +/-25 V max ... you could be exceeding that capacitively through the Collector contact and the Gate of the IGBT. A 15V Zener from each Gate to ground should be enough. I can't tell what the part number is on your gate controller is to tell what it is doing internally.

Try this ... remove both Q1 and Q2 temporarily and place a small cap 100pF across where pins 1 and 2 of the IGBT would be .... Scope pin 1 to GND and see what it looks like ... if it's greater than +/-25V then that is probably your problem
 
Last edited:

SamR

Joined Mar 19, 2019
5,031
The current and voltage ratings are not all of the story. You also have to look at the power rating. VI=W

Edit: yep the device maxs out at 40W and loading it at 300W will fry it
 

Deleted member 440916

Joined Dec 31, 1969
0
I think your problem is the cold resistance of the lamp, if in doubt measure it, combined with the SOA of the IGBT.
If you check the specification the latter is only ~100mA at 330V
 

MaxHeadRoom

Joined Jul 18, 2013
28,617
Appears you are controlling this with a Pic, Any reason you are not controlling the motor with a PWM output from the Pic with a ramp up?
Max
 

MisterBill2

Joined Jan 23, 2018
18,167
Hi. Below is my circuit. It is to be a motor controller but at the moment I'm testing with a 300W lamp in a domestic environment. I have marked the max. voltage ratings in blue. the problem is Q1 (FGD3N60LSD) keeps failing. C-E becomes permanently low impedance ~100ohm. The mains is 240VAC, so that's 330Vpk. To allow for spikes I have picked devices rated at 600V for BRIDGE1, D1 and Q1. Q2 is omitted at the moment. Fortunately the Q1 failure doesn't affected the rest of the circuit.

The failure happens immediately when the circuit is connected to the mains. I've programmed it to flash for 10ms every 500ms. A few times it has worked when first connected, then the next time I connect it to the mains, it fails. The duty cycle is low so nothing's getting hot when it's working correctly. When Q1 fails, Q1 quickly gets hot.

I tried a TVS rated at ~400V across Q1's C-E junction. This allowed the circuit to work for about 10 connections, then it failed on the 11th. so I think the failure is due to a voltage spike across Q1 C-E. But I don't understand why my overrating of 600V isn't sufficient to prevent this nor why the TVS (400V breakdown) didn't offer permanent protection. Also, why does it only fail when initially connected to the mains? I'm thinking it might have something to do with switch bounce and stray inductance. Any help/advice gratefully received.

View attachment 200107
If the circuit is actually wired as shown, any leakage current through that diode D1 will put the rectified mains voltage directly across the transistors. So perhaps moving the diode D1 connection to the other lamp terminal will solve the problem. IN ADDITION, that circuit is mostly tied to the mains, even though there is a bridge rectifier.
Excess current will destroy semiconductor devices quite rapidly.
 

Thread Starter

Robin66

Joined Jan 5, 2016
275
Only Q1 and not Q2? ... Ahh I see Q2 is omitted for the moment

It could be a Zero cross issue... you might need to turn ON the IGBT during the Zero Cross

Also, I don't see anything that limits the Gate voltage to +/-25 V max ... you could be exceeding that capacitively through the Collector contact and the Gate of the IGBT. A 15V Zener from each Gate to ground should be enough. I can't tell what the part number is on your gate controller is to tell what it is doing internally.

Try this ... remove both Q1 and Q2 temporarily and place a small cap 100pF across where pins 1 and 2 of the IGBT would be .... Scope pin 1 to GND and see what it looks like ... if it's greater than +/-25V then that is probably your problem
Yes ok, I should investigate the transients on the gate. You're right that depending on the internal circuitry of U2 (MCP1416) there isn't necessarily anything preventing 25V shoot thru, although the C-G capacitance should slow turn on and slow turn off of Q1 (miller effect), so I don't see how this could account for shoot thru. Defo worth investigating tho.
 

Thread Starter

Robin66

Joined Jan 5, 2016
275
The current and voltage ratings are not all of the story. You also have to look at the power rating. VI=W

Edit: yep the device maxs out at 40W and loading it at 300W will fry it
I'm using Q1 in switch mode (on or off) so it should not dissipate anything like 40W. when on it would be ~1.5V*2A = 3Watt, tops.
 

Thread Starter

Robin66

Joined Jan 5, 2016
275
Possible inrush current of the lamp? Incandescent lamps can draw at least 10 times their normal rating when first powered.
Yes this is possible. 10x inrush could put it beyond the 25A "absolute max" ratings for the pulsed collector current. I could do some trials with an ohmic load. I was using a lamp because it's obvious when it's ON. But an LED in series with a big wirewound resistor may work. When I eventually get to using the motor it'll have a big inductance which may avoid an inrush problem.
 

Thread Starter

Robin66

Joined Jan 5, 2016
275
I think your problem is the cold resistance of the lamp, if in doubt measure it, combined with the SOA of the IGBT.
If you check the specification the latter is only ~100mA at 330V
I'm using the IGBT in switch mode. When the C-E voltage is 330V (device OFF), the collector current is 0A. when the C-E voltage is ~1.5V (device ON) the collector current is 2A*

*in steady state, there are inrush issues which may be the problem
 

Thread Starter

Robin66

Joined Jan 5, 2016
275
Appears you are controlling this with a Pic, Any reason you are not controlling the motor with a PWM output from the Pic with a ramp up?
Max
yes, it's a PIC. The switching frequency is so low relative to the Mips that I don't need the PWM modules. I want to do some funky stuff with switching relative to the mains zero cross and getting cute with the duty cycle to reduce the natural 100Hz hum that mains motors produce. That stuff is way down the line though.
 

Thread Starter

Robin66

Joined Jan 5, 2016
275
If the circuit is actually wired as shown, any leakage current through that diode D1 will put the rectified mains voltage directly across the transistors. So perhaps moving the diode D1 connection to the other lamp terminal will solve the problem. IN ADDITION, that circuit is mostly tied to the mains, even though there is a bridge rectifier.
Excess current will destroy semiconductor devices quite rapidly.
yes the transistors are sufficiently rated to handle rectified mains (in an ideal world). Excess current (inrush) may be the issue. I'll do some more investigation around this. Perhaps I should ditch the incandescent lamp as a load
 
The IGBT SOA curve is 0.2A at 200V so I would expect an incandescent lamp to quickly kill it.
You should have a snubber or MOV to protect against inductive spikes, 600V is easy to exceed, even with a light bulb. They are spiral coil wound and inductive after all - not a resistive load.
 

Thread Starter

Robin66

Joined Jan 5, 2016
275
The IGBT SOA curve is 0.2A at 200V so I would expect an incandescent lamp to quickly kill it.
You should have a snubber or MOV to protect against inductive spikes, 600V is easy to exceed, even with a light bulb. They are spiral coil wound and inductive after all - not a resistive load.
I have a TVS now with a 400V rating. Is that not sufficient?
 
D1 is a TVS? It's across the load, not the IGBT. edit: I'm wrong, OP says it was added.
OK I'm saying your IGBT died from over current with the light bulb load.
But when you go to running a motor, a TVS across C-E will protect from over voltage, as long as it is a decent size and the bridge rectifier is not super slow recovery, as some are. If you are doing phase control, a TVS can cook due to the back EMF of the load.
 
Last edited:

Thread Starter

Robin66

Joined Jan 5, 2016
275
D1 is a TVS? It's across the load, not the IGBT.
No, D1 is a standard diode. I put a TVS across the C-E junction of Q1.
Ps. Sorry this isn't shown in the schematic. I tried this addition when I kept blowing Q1 (as mentioned in the OP). It helped a lot, but hasn't completely eliminated the problem
 
Last edited:

MisterBill2

Joined Jan 23, 2018
18,167
yes the transistors are sufficiently rated to handle rectified mains (in an ideal world). Excess current (inrush) may be the issue. I'll do some more investigation around this. Perhaps I should ditch the incandescent lamp as a load
Try it with a 40 watt bulb and see if that solves the problem.
 

DickCappels

Joined Aug 21, 2008
10,153
Your entire circuit is not shown. How do you power the unseen circuitry? Does the unseen circuitry connect to any other circuits or user controls?
 

Thread Starter

Robin66

Joined Jan 5, 2016
275
Your entire circuit is not shown. How do you power the unseen circuitry? Does the unseen circuitry connect to any other circuits or user controls?
I only posted the problematic part of the circuit so that it’s easy for people to identify the relevant components. The rest of the circuit is just a PIC, LCD display and a button, but only the former has been populated on the pcb and it’s acting as a fixed pulse generator for now. I wanted to ensure the high risk stuff (HV) works before debugging the fiddly stuff (LCD).
Ps. There’s also an AC-DC 12V power supply and a 3.5V voltage regulator. The 12V supply feeds the regulator and U2
 
Last edited:
Top