Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Exploding IGBT's

Status
Not open for further replies.

KingKrak

New Member
Hey,
I'm building a 3 phase variable frequency drive for small (less than 2kW) motors. I've pretty much finished building a prototype circuit and everything is just dandy, except...

My IGBT's keep exploding. Badly exploding...

To try and isolate the problem I built an independant circuit that consisted of a 3 phase rectifier with my IGBT's across the DC bus (as in the picture attached that I found on wikipedia). I had no motor connected, and I had nothing connected to the gate terminal of the IGBT (so they were not being turned on). The IGBT's were rated for 600V 10A, and I was rectifying 415V AC (so it ended up 580VDC). When I flicked the switch they exploded. I had assumed that 2 of these IGBT's in series gave me a voltage rating of 1200V. Maybe not...

So annoyed and with a burnt hand I decided to try a lower voltage. This time I rectified single phase 240VAC (so it ended up 340VDC). This time I tested between the rails of the DC bus to make sure I didn't have a short, and the multimeter read 0.8Mohms. When I flicked the switch the replaced IGBT's exploded again.

I've tested the entire circuit using a 14VDC power supply as the DC bus and it worked well. Anything useful seems to make the IGBT's explode.

Does someone know what's going on here?
 

Attachments

  • PWM_VFD_Diagram.png
    PWM_VFD_Diagram.png
    6.7 KB · Views: 2,857
how did you turned on and off the IGBTs? i suspect the short circuit occurs since your circuit turns the IGBTs on before the other arm breaks. you shuld have a driver circuit in order to do this perfectly. if you already have used a driver just post the complete diagram, some one can help you to simulate and tell the result.

also when you test it if you connect a low load on the drive side, and a series load (calculate it to the full out put voltage to give some medium current) between the DC source and inverter you can stop explotions, since the series load will limit the current in worse case if there will be a direct short circuit.
 
Last edited:
My guess is that you're either not allowing enough deadband delay between turning on/off the high and low side or you're getting a lot of inductive kickback which are popping the IGBTs
 
Hey,
I'm building a 3 phase variable frequency drive for small (less than 2kW) motors. I've pretty much finished building a prototype circuit and everything is just dandy, except...

My IGBT's keep exploding. Badly exploding...
When you say exploded what do you mean?
cracked case?
ded?

There are some simple DMM tests you can do on an IGBT (module) to give an idea of how it died
Also if you can open the module up and look at the die you can actually tell if it was over-voltage or over-current

To try and isolate the problem I built an independant circuit that consisted of a 3 phase rectifier with my IGBT's across the DC bus (as in the picture attached that I found on wikipedia). I had no motor connected, and I had nothing connected to the gate terminal of the IGBT (so they were not being turned on). The IGBT's were rated for 600V 10A, and I was rectifying 415V AC (so it ended up 580VDC). When I flicked the switch they exploded. I had assumed that 2 of these IGBT's in series gave me a voltage rating of 1200V. Maybe not...
While two series 600V IGBT will provide you will a static blocking of 1200V, it will not under dynamic/switching cases.

When you turn on an UPPER IGBT its emitter will be essentially at +VDC voltage, which mean the LOWER IGBT must be able to support the full DClink potential

For 415Vac 3ph this is 540V.
This in itself is bad, you really need to provide twice the blocking voltage for your nominal DClink. 600V IGBT's are fine for a 3phase 115Vac system (gives a DClink of 270V) but for 230Vac 3phase with a nominal DClink of 540Vdc its faar to close!

Once you actually start switching these devices you will get voltage overshoots on your IGBT's at turn-off, you can easily see 900Vdc peak voltage from a 540V DClink inverter. The thing about voltage and silicon is it is VERY unforgiving. While you can over-current silicon quite happily (and as long as you keep within the I^2t or the rule of 10 you are fine) BUT you a fraction over the voltage that the device will support (and a 600V IGBT may actually support 602V) them BAM dead!

I don't think this killed it (see next) BUT the use of 600V silicon in this cct and this voltage is a concern

So annoyed and with a burnt hand I decided to try a lower voltage. This time I rectified single phase 240VAC (so it ended up 340VDC). This time I tested between the rails of the DC bus to make sure I didn't have a short, and the multimeter read 0.8Mohms. When I flicked the switch the replaced IGBT's exploded again.

I've tested the entire circuit using a 14VDC power supply as the DC bus and it worked well. Anything useful seems to make the IGBT's explode.

Does someone know what's going on here?

Two things that could cause the issue

#1 do you have any deadtime between turning an UPPER IGBT {on,off} and turning the equiv LOWER IGBT {off,on} A safe starting position would be 500ns (I use 1us for 1200V, 800Amp IGBT's and they are actually ON/OFF in ~ 500ns)

#2 what is your gatedrive circuit (this is my specialty at work ;) ). The uppers need to be floating and your isolation method need to be able to handle the common-mode voltage you will see as well as your isolating drive signal being able to cope with being bounced by 1200V/us. A good OPTO can easily cope with that.
 
To properly drive your IGBT's you should be using driver IC's designed for your application. International Rectifier makes a number of FET/IGBT driver IC's that help greatly reduce the potential for IGBT damage. They are worth looking at.

I too would recommend using 1200 volt or higher IGBT's and with current capacities of around 4 times or more than your motors rated current.

The higher quality factory made VFD units tend to be leaning towards being grossly over built on the switching devices aspects. I have found that many 240 AC line source units have 1200 volt rated IGBT's and the bigger 480 volt line sourced units are running 1500 and even the newer 2000+ volt IGBT's.
Both have typically 4:1 and up to 8:1 IGBT current ratings over the units actual working current ratings.

The older and cheaper VFD units have far less voltage and current overhead built into them.

For what its worth you can buy a good used factory VFD unit for far less than what you can make one for and you get far better control and capabilities too! I have purchased many used but good condition 1.5 - 4 KW VFD units on line for under $120. I seriously doubt you could build one for that and have all the capabilities that go with a standard factory unit.

VFD units are like inverters. You can buy them cheaper than you can build one that would have the same functions and capacity as any typical low end factory unit.
 
Hey guys,
Thanks for the comments. I'll look into getting IGBT's with higher voltage ratings. I have a few factory made units sitting around and I'll pull them apart to see what voltage ratings they have. This drive is going to be for a low cost custom application, and there will probably be around 10,000 made. I can be a pretty lazy person, I would have just bought one if I needed it for something at home. I don't need the stress...

I'm currently driving the circuit with an IR2109, which is an IC designed to drive complementary IGBT's. But I feel I have to make this extra clear to everyone - the driving circuit was not connected when the IGBT's exploded. I had that part sitting on another bench across the room. There was nothing connected to the gate terminal of the IGBT. The IGBT's were never switched on. There was no load connected.

So most of my IGBT's no longer have their case intact. How can I look inside and tell if it was over-voltage?
 
There was nothing connected to the gate terminal of the IGBT.
Sounds like a good reason for them to blow up. If the gate is left floating, it could turn on by itself.
 
IGBT's have as sensitive of gate as a typical Mosfet. Any local static or even just the sudden increase in local voltage charges from the 540 VDC would be enough to get them to turn on far enough to have a melt down.
 
Hey guys,
Thanks for the comments. I'll look into getting IGBT's with higher voltage ratings. I have a few factory made units sitting around and I'll pull them apart to see what voltage ratings they have. This drive is going to be for a low cost custom application, and there will probably be around 10,000 made. I can be a pretty lazy person, I would have just bought one if I needed it for something at home. I don't need the stress...

I'm currently driving the circuit with an IR2109, which is an IC designed to drive complementary IGBT's. But I feel I have to make this extra clear to everyone - the driving circuit was not connected when the IGBT's exploded. I had that part sitting on another bench across the room. There was nothing connected to the gate terminal of the IGBT. The IGBT's were never switched on. There was no load connected.

So most of my IGBT's no longer have their case intact. How can I look inside and tell if it was over-voltage?

mmm un-driven IGBT gates could be the cause here.
The gate-threshold of IGBT's tend to be around 7V so having 540Vdc across a IGBT module with no gate cct or termination isn't a problem (I have had voltage on my DClink and accidentally turned off my gatedrive power...)

HOWEVER Depending on how fast the DClink came up you may have turned on some of the IGBT's via high dv/dt across them and charge entering the gate, via the millar capacitor, and turning the IGBT on.


IF you can open the module and remove the gunk have a look at the actual tiles. You will be able to easily tell the difference between an IGBT tile and a DIODE tile.

Now Overvoltage tends to leave lots of little cracks around the edge of the tile (its hard to explain but you will know it when you see it). The problem is an over-voltage always results in an overcurrent as well so you are going to have at least one tile with a MASSIVE crater in it. If you are lucky enough that the edges of the tiles are not obliterated as well you may see the cracks
 
Thanks guys. There's no cracks on the tiles, just craters. Oh, and some missing wires where the gate terminals used to be...

It was stupidity on my behalf for leaving the gate floating. I have some bad habits when it comes to things like that. It didn't even occur to me that I could get enough induced voltage there to maybe turn the IGBT on.

When my next batch of IGBT's turns up in the mail I will give it another go. I'm sure you will hear from me again...
 
FETS are unpredictable with the gates floating like that. Think of the trace from the gate going to the control circuitry as one plate of a capacitor and any nearby trace as the other plate. The gate will tend to follow the voltage of any nearby line like a sick little puppy.
 
If there's any way to do so, it would be a good idea to have a bleeder resistor (10k?) from gate to emitter, such that it doesn't ever get disconnected.
 
Hi All,

Can you tell me how does over current on the IGBT look like? Will it cause MELT down on over current? and Cause crater on over voltage?

Please advice?
 

Attachments

  • 2.JPG
    2.JPG
    52.8 KB · Views: 1,071
  • 1.JPG
    1.JPG
    34.3 KB · Views: 1,297
You should have started a new thread for your question.

If a transistor carries excess current from an overload or short, the small connecting wires will burn out and open. This creates an arc which destroys the surrounding circuit area. The higher the available current and voltage from the power supply, the greater the damage before the arc finally extinguishes. It's rather like a small arc welder.
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top