How hot do the heatsinks on IGBTs typically get with a DRSSTC?

[profdc9]

Hello,

I would like to add the ability to my DRSSTC driver to shut down if the IGBT heatsinks get too hot.  What is a typical safe temperature for IGBTs?  I would like to use the bimetallic thermal switches that are typically used as safety shutoffs for motors, but place them in contact with the heatsink.  I was thinking somewhere in the range of 45C-65C depending on how the temperature in the IGBT package would be related to the exterior temperature at the thermal switch.  I know there's probably not a single temperature choice that works for everyone, but perhaps if someone has run their coil, they could use a pyroelectric infrared sensor and test the temperature.

Thanks,

Dan

[Uspring]

Since DRSSTCs are usually run with low duty cycles, average power consumption is not the major issue. The die temperature jumps up and down during bursts and between them. Have a look at the transient thermal response in the IGBTs datasheet.

[Mads Barnkob]

I would say somewhere around "stone cold" if forced air cooling is used.

Zth(j-c) that I use in my calculation examples of calculating IGBT max switching speed ( http://kaizerpowerelectronics.dk/tesla-coils/drsstc-design-guide/igbts/ ) is only the temperature rise from junction to casing of the IGBT brick itself, and we want to keep that pulse jump below a certain limit, so that the junction can cool off between bursts. So if you look at the thermal impedance of the IGBT and calculate it in series with the thermal impedance of the heat sink, you could see how little the Kelvin/Watt gain is on the heat sink.

[Steve Ward]

As mentioned, transient heating of the IGBT dies is significant in this case.  However, to have a reliably functional system the transient heating simply can not be "too big" or else thermal fatigue effects will set in.  When i design my bridges, i aim for absolute maximum transient thermal rise of 30*C.  So this is 30*C hotter than the tab or baseplate of the device.  The tab/plate can then be somewhat hotter than the heatsink itself as there is some thermal resistance there, too.  I'm completely ignoring the question of how do i figure out the transient die temp, if you want to understand, read this:

http://www.pwrx.com/pwrx/app/IGBT-Intelligent-PwrMods.pdf

In real world case: I designed a system for a museum using CM300 modules and limit the heatsink temperature to 60*C with thermal shutdown feature.  The temperature really does climb into the 40-50*C range on a regular basis.  It's been in service for years so I can only assume I did ok.

If you insist on "cold" heatsinks, then you are putting too much heatsink in there .