Re: Please help ! IGBT exploded ...

["Tesla list" <tesla-at-pupman-dot-com>]

Original poster: Sean Taylor <sstaylor-at-uiuc.edu> 

On Mon, 08 Mar 2004 19:52:16 -0700, Tesla list <tesla-at-pupman-dot-com> wrote:

>Original poster: "Rajesh Seenivasan" <rajeshkvs-at-hotmail-dot-com> Hi Mike,
>
>The bottom transistor on one leg of the H bridge exploded. (It shows short).
>The gate drive waveform of top and bottom igbts crossed each other at 0
>volts
>and increased from 0 to +/-5 volts in 800ns (very slow rising edge ?).
>I had 3.3 ohm resistor in series with gate.

800 nS is pretty slow . . . especially with no deadtime (depending on the 
actual waveforms, that kinda adds deadtime in), and using IGBTs since they 
are relatively slow to turn off.  Also, make sure if you're doing any sort 
of measurements with a scope while you have mains current on the IGBTs that 
one (preferable power to the IGBTs) is isolated, otherwise there won't be a 
common ground point, and you'll have some serious problems with current 
flowing in places you didn't expect or want!  Anyway, you could probably 
use a little more current in your gate driver - I use a pair of TC4421 and 
TC4422 devices for each gate with my IGBT bricks, and get around 200 to 300 
nS rise and fall times.

Here's a link to a deadband controller that works fairly well:
http://www.ews.uiuc.edu/~sstaylor/temp/deadband.jpg

I built it using 50kOhm pots for the resistors, 1N4148 diodes, and 0.001 uF 
caps, which gave me a pretty good adjustment range, though you'll have to 
experiment with the values.  You could also use a dual gang pot to have 
close to the same deadtimes for each half of the bridge.  When you use this 
circuit, just be sure to pair the IGBTs correctly - ie top left and bottom 
right get the same drive signal, and bottom left and top right get the 
other signal.

Sean Taylor