Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

IGBT Short Circuit problem

Status
Not open for further replies.

kratosrazor

Junior Member level 2
Joined
Nov 12, 2018
Messages
23
Helped
0
Reputation
0
Reaction score
0
Trophy points
1
Location
Mazanaran, Iran
Activity points
200
Hello,

I am designing a 390 DC to 220 AC 1KW single phase inverter.
I am using a proportional resonance control with LPC1768 Microcontroller.
in low output voltage, I get a pure sine wave and I have no problem, but when the input increases above 150 Volts, my IGBT get short circuit.
these are the schematic and PCB designs.
3.PNG
4.PNG
1.PNG
2.PNG
 

There's a number of dubious circuit details, e.g. using low voltage 1N4148 as bootstrap diode, 1k gate resistors. Can we trust the schematic? Also what's the switching pattern? Do you provide sufficient dead time to avoid shoot through?
 

Hi,

I assume the high value R22, R24, R25, R26 cause problems. Especially the slow switch OFF, while the high side has fast switch ON.

Don´t you have a scope?
Or an simulation software?

Klaus

Added:
the driver schematic is really hard to read. Please follow these rules:
* signal flow: from left to right
* positive supply on top
* negative supply (GND) on bottom of the page
* use GND symbols.
* avoid 45° in schematic
* avoid unnecessary bends

C10 is the VCC reservoir capacitor. It should keep VCC stable.
Now with every falling edge C9 is charged. but C9 is 4700 times bigger than the reservior... makes no sense to me...
--> I recommend: C10 should be bigger than C9. Both should be fast and LOW ESR. Maybe ceramics ones.

I doubt that an 1N4148 is suitable to charge a 47uF capacitor in very short time. Expect high current peaks... while you want low voltage drop.
--> I recommend a 1A schottky diode.

***
Your GND plane (red layer) is good. But you use that big isolation that many important GND connections are made with the blue layer. This causes unnecessary high line impedance. Don´t do this, especially not with switching power applications.
 

I doubt that an 1N4148 is suitable to charge a 47uF capacitor in very short time. Expect high current peaks... while you want low voltage drop.
--> I recommend a 1A schottky diode.
The diode must be 400V rated in the first place.
 

Hi,

I assume the high value R22, R24, R25, R26 cause problems. Especially the slow switch OFF, while the high side has fast switch ON.

Don´t you have a scope?
Or an simulation software?

Klaus

Added:
the driver schematic is really hard to read. Please follow these rules:
* signal flow: from left to right
* positive supply on top
* negative supply (GND) on bottom of the page
* use GND symbols.
* avoid 45° in schematic
* avoid unnecessary bends

C10 is the VCC reservoir capacitor. It should keep VCC stable.
Now with every falling edge C9 is charged. but C9 is 4700 times bigger than the reservior... makes no sense to me...
--> I recommend: C10 should be bigger than C9. Both should be fast and LOW ESR. Maybe ceramics ones.

I doubt that an 1N4148 is suitable to charge a 47uF capacitor in very short time. Expect high current peaks... while you want low voltage drop.
--> I recommend a 1A schottky diode.

***
Your GND plane (red layer) is good. But you use that big isolation that many important GND connections are made with the blue layer. This causes unnecessary high line impedance. Don´t do this, especially not with switching power applications.

sorry, I missed changing some values.
here is the schematic.

3.PNG
4.PNG
I should say that I have no problem with open loop control with Vin=200 V (I didn't test higher voltages). but the IGBTs get short with closed loop control. for input 130 Volt, I have no problem and the output is just fine.
 

if you are using 4148's as bootstrap - at some point above 75V they will go bang ... you need 10E turn off res and 1k turn on to provide extra dead time

above about 200V you will start to make RFI which can affect every thing ...
 

if you are using 4148's as bootstrap - at some point above 75V they will go bang ... you need 10E turn off res and 1k turn on to provide extra dead time

above about 200V you will start to make RFI which can affect every thing ...

the bootstrap diode is 1n4937.
all gate resistors are 15 ohms
I corrected the schematic in post #5.

- - - Updated - - -

There's a number of dubious circuit details, e.g. using low voltage 1N4148 as bootstrap diode, 1k gate resistors. Can we trust the schematic? Also the swhat's the switching pattern? Do you provide sufficient dead time to avoid shoot through?

I corrected the schematic at post #5.
I use unipolar switching pattern.
I use IR2111 as gate drive and its deadtime is 650ns.
 

Schematic circuit faults, next round. Vs to high-side emitter connection seems to be missing for both drivers. Also 10 nF bypass capacitor for low-side driver is ridiculuously small.

- - - Updated - - -

O.K., the first point is fixed in the new schematic. The 10 nF point is more in the line of Easy peasy's guess, false switching, maybe self oscillation due to self generated RFI.

- - - Updated - - -

The closed loop problem observation also suggests a possibility of output overcurrent, e.g. inductor saturation.
 

The closed loop problem observation also suggests a possibility of output overcurrent, e.g. inductor saturation.

Oh, I forgot to say that my inductor gets warm even at no load. is it ok?
inductor model is t184-26 iron powder and the switching frequency is 15Khz.
on more thing, I didn't use the current sensor. wouldn't that be a problem?
 

no surprises the Fe powder gets warm at no load ... no over current protection? yes likely will be a problem ...
 

Why not test the inverter with load, increasing supply voltage, increasing output current but no "closed loop control"? Monitor output waveform, output and DC bus current.
 

I tested inverter with both open loop and closed loop control.
I attached the result.
the DC Link voltage is 105 Volt.

this is the VCE with open loop control.

DSC_1780.jpgphoto_2019-06-30_21-45-51.jpg


this is the VCE with closed-loop control.

DSC_1777.jpgDSC_1778.jpg



this is ADC input for closed loop control.

ADC_Closed loop.jpg
 

Where are those > 100V narrow spikes coming from? another converter - there appears to be a lot you are not showing ...
 

Where are those > 100V narrow spikes coming from?
Switching of the other bridge branch?

We can see that the "closed loop control" is somehow behaving strange, but we don't the control algorithm, how it exactly steers the PWM, what are its constraints, e.g. manipulated value limits? That's why I suggested to verify that the output stage can handle rated voltage and current with static set points.
 

I have used 7n40 MOSFET transistor instead of Fg25n120 IGBT transistor and the output was just fine.
DC link voltage was 200*sqrt(2), the maximum VDS was 350V and output was 170 V RMS. But when I use Fg25n120 again, IGBTs got short circuit in one leg.
So does my IGBT has a problem? under what condition a 1200 V 50 A IGBT get short?!

thank you for helping.
 

the total turn off time for the IGBT can be as high as 354nS ( see data sheet ) so it it likely that minor shoot thru is killing them, try 500nS dead time for the IGBT ...
 

I have used 7n40 MOSFET transistor instead of Fg25n120 IGBT transistor and the output was just fine.
DC link voltage was 200*sqrt(2), the maximum VDS was 350V and output was 170 V RMS. But when I use Fg25n120 again, IGBTs got short circuit in one leg.
So does my IGBT has a problem? under what condition a 1200 V 50 A IGBT get short?!

thank you for helping.

I have asked the same question, I'm using STGW25H120F2 and are experiencing problems too, my working theory is that somehow collector and emitter gets "shorted" when it's off. I was about to prepare a current measurement of Ic but now you gave me a good idea to try with a MOSFET to confirm my suspicion. But under which conditions an IGBT will do that? Nobody really knows yet.
 

please specify exactly which mosfet you used?

- - - Updated - - -

IGBT's can turn on very fast ( < 20nS ) I see you have no turn on gate resistors - this likely means that when a top device turns on, for example, the dv/dt across the lower device is so high the capacitive coupling to the lower gate is pulling the gate up and then you have both devices on briefly - with enough peak current to kill them after a short time - this is common for power electronics engineers with little experience of driving IGBT's ...

easy to check with a high speed scope ( 200MHz ) across the lower gate ...
 

I have asked the same question, I'm using STGW25H120F2 and are experiencing problems too, my working theory is that somehow collector and emitter gets "shorted" when it's off. I was about to prepare a current measurement of Ic but now you gave me a good idea to try with a MOSFET to confirm my suspicion. But under which conditions an IGBT will do that? Nobody really knows yet.

do you mean your igbt gets short circuit even without giving any pulse?
In this case, put a 1k resistor on Gate emitter across each IGBT. maybe your problem would fix.

Easy peasy said:
the total turn off time for the IGBT can be as high as 354nS ( see data sheet ) so it it likely that minor shoot thru is killing them, try 500nS dead time for the IGBT ...
I have used IR2111 as gate driver and it gives 650ns deadtime. isn't that enough?

Easy peasy said:
please specify exactly which mosfet you used?

seller gave me this datasheet.
https://www.farnell.com/datasheets/1730531.pdf

Easy peasy said:
IGBT's can turn on very fast ( < 20nS ) I see you have no turn on gate resistors - this likely means that when a top device turns on, for example, the dv/dt across the lower device is so high the capacitive coupling to the lower gate is pulling the gate up and then you have both devices on briefly - with enough peak current to kill them after a short time - this is common for power electronics engineers with little experience of driving IGBT's ...

I have used turn on gate resistor. you can see the schematic at 5th post .

Easy peasy said:
easy to check with a high speed scope ( 200MHz ) across the lower gate ...
Good idea. I will check that and post soon.
 

OK so the mosfet was a 17NN40 not a 7N40, the devices are similar on paper - it must be gate drive peaking when it is supposed to be off - or - fake parts for the igbt's ...

- - - Updated - - -

the turn off R is approx 7.5 ohm - try just a 40V 1A schotty in reverse across a 47 ohm turn on resistor - this will pull the gate down harder at turn off and slow the turn on just a little to reduce dv/dt across the device that is supposed to be off ...
 

Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top