RE: Series connection of Mosfets/IGBTs

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

Original poster: Greg Leyh <lod-at-pacbell-dot-net> 

Original poster: "Steve Conner" <steve.conner-at-optosci-dot-com>

>Greg, if anyone knows the answer to this question, it's gotta be you: How
>much voltage can I hit the IGBTs in my OLTC with? The datasheet rating is
>1.2kV. I'm running them at 850V at the moment but I'm hungry for more power,
>and the "voltage" knob on my power supply still has a way left to go. Is
>there anything I need to worry about as I start scraping the datasheet max
>voltage, apart from the obvious problem of not having much headroom for
>transients?

Transients and resonant rise are the main reasons for maintaining a certain 
operating 'headroom'.  Traction applications such as electric trains 
maintain a headroom of at least 2:1.  Typically a device will not withstand 
much more voltage than its rating.  Basically you have to size up your 
overall transient risk and ask the question, 'How lucky do you feel?'


>Finally, is it safe to break down the IGBTs with a small current (eg 10mA)
>so I can measure what their breakdown voltage actually is? I guess these
>things vary in manufacture, so I'm hoping mine might actually be more than
>1.2kV.

I've characterized C-E breakdown curves of IGBT packs and single dies 
before, using curve tracers in the mA range with no ill effects.  Small, 
[and brief] avalanche currents shouldn't be a problem for most 
devices.  You might be able to 'hand pick' a few units that have 10% to 30% 
higher C-E breakdown ratings.