Mosfet Temperature limit

[rocky79]

Hello,

I made a mosfet switch using part# NTMD4840NR2G. At 3 Amps the mofet drain temperature was 56ºC at 22ºC room temperature. ( Mosfet total power consumption is 0.24W)
As a designer who is looking for a reliable switch would you feel confortable operating the mosfet in a 40ºC ambient temperature?

I am estimating the drain temperature will be ~74ºC and the junction temperature to be about 84ºC. [Junction max temp=150ºC]
Junction temperature of 84ºC is based on Roj-a=184ºC/W and power dissipiation of 0.24W at TA=40ºC

 

[barry]

Something's not right. If you have 184 deg-C/W and .24 W, rise should be 44 degrees. You're showing 34 degree rise. I would go with worst case, so your junction would be at 94 worst case. Maybe a little high for my tastes, but it's within spec. If this is going in a space ship or medical equipment, I'd look to cool it more. For this device, you could add a bit of copper to your PCB and drop the Roj-a quite a bit.

 

[rocky79]

Something's not right. If you have 184 deg-C/W and .24 W, rise should be 44 degrees. You're showing 34 degree rise. I would go with worst case, so your junction would be at 94 worst case. Maybe a little high for my tastes, but it's within spec. If this is going in a space ship or medical equipment, I'd look to cool it more. For this device, you could add a bit of copper to your PCB and drop the Roj-a quite a bit.

Well, that thermal resistance is for using only the minimum PCB pads for the MOsfet however I have copper pour around the drain and some vias to help reduce that thermal resistance.
So the actual thermal resistance is lower. How about 85ºC for Junction temp?

 

[barry]

If I recall, with a 1" square copper area the Roj-a drops to about 110. That would give a rise of 24 degrees which seems reasonable. The hotter the device gets, the less reliable it will be, but I don't think there's any quantitative way to analyze that. At least, not any info the mfr will give.

 

[dick_freebird]

The manufacturer should be able to provide a rating (or
derating) for lifetime vs junction temperature. I've spent
most of my career on products rated for 125C or above.
I think most MOSFETs you can find will have at least an
industrial (85C) parametric performance rating and a
recommended operating temp range that extends at least
that far (implying an adequate sort of reliability).

 

[rocky79]

If I recall, with a 1" square copper area the Roj-a drops to about 110. That would give a rise of 24 degrees which seems reasonable. The hotter the device gets, the less reliable it will be, but I don't think there's any quantitative way to analyze that. At least, not any info the mfr will give.

Thank you Barry