FET case temp

Status
Not open for further replies.

Mosaic

Well-Known Member
Hey,
What's a safe Case temperature for POWER FETS like the IRFP3306?

Doing a thermistor protection circuit for 'em.

thx
 
Depends upon the power it is dissipating. You have to use the thermal resistance from junction to case and the maximum allowed junction temperature to determine the limit. Is the FET on a heatsink?
 
With good thermal contact to a 0.5°C/W heatsink and for an ambient temperature of 50°C you should not allow the transistor case temperature to exceed about 90°C
 
We use a higher voltage International Rectifier FET in the same package with the same rated operating junction temperature and have done quite a bit of abnormal and abusive testing with it. They share a large heat sink that likely dissipates about 1.5°C/W with a couple of other devices that don't dissipate near as much power. At 6¢ an ounce for aluminum, they like trim as much as they can get away with for mass production of consumer goods.

Once the FET reaches about a 160°C case temperature during our typical abnormal overload test, they will usually blow within the hour and sometimes within minutes. At about 140°C, they usually survive this test but their reliability gets compromised to the point where they will not complete a 1000 hour life test.
 
I think crutschow's 90°C case limit is probably appropriate if reliability is a priority. If it's just a personal gadget that I can fix myself without involving a customer service agent, UPS, and a field service tech, I'd push it harder.
 
Thanks,

I am using a thermistor to maintain reliability. I had estimated 75C as the hottest for the case temp for longevity at a 30 C ambient.
 
It makes a big difference where you put the thermistor. If you put it on the plastic of the FET it is measuring close to the junction temperature. If you mount it on the metal tab or the heat sink it is closer to case temperature.
 
We use thermocouples glued directly to the case with a cyanoacrylate and an accelerator so the temperature difference between the case and the thermocouple junction is pretty small. OTOH, the limited accuracy of the thermcouple itself means the FET's lasting an hour might actually be 155°C instead of the measured 160°C and those lasting minutes might actually be 165°C. It might be the parts themselves too. We don't really have the proper instrumentation to tell for sure.
 
We put it on the plastic. The metal is not accessible on every part and there are a few applications where the metal is also at a potential (>300V) that could damage our dataloggers. The temperature of the metal bit connected to the drain (not the heat sink) is closer to the junction temperature but in the name of uniform test procedures...

IIRC, some past test data indicated the difference between the two locations was pretty small at "normal" temperatures and only slightly higher than the thermocouple accuracy (probably ≈10°C) at "abnormal" temperatures.
 
Thanks Cachehiker,
In another thread we have a temperature sensor on a FET mounted to a water cooled heatsink so the metal part of the FET stays at 25C (or a little less) but the plastice was almost what we expected the junction temperature to be. Some reading kind of confirmed this, but it is always nice to have some "real world" numbers.
 
Our application is also electrically isolated and the thermal resistance between the junction and the drain is much lower than the thermal resistance between the drain and the heat sink.

I may have to dig into some of the older test data as I'm not entirely sure whether it was the drain or the plastic case that was cooler. In the end though we concluded it didn't much matter as both measurements had to be at least 30°C cooler than the junction under overload conditions. They could only be used for comparison and we needed to be comparing apples to apples.
 
Status
Not open for further replies.
Cookies are required to use this site. You must accept them to continue using the site. Learn more…