Hello,
I'm interested in doing some remote temperature sensing of some high power mosfets on a PCB.
I've never used this pnp thermal diode method before, but it seems like the MAX6622 would be a good part for my application. Now my question is, do they have any pnp transistors with an isolated heatsink (tab) of some sort, preferably small surface mount packaged? if I can mount that pnp on the heatsink as the mosfet's drain tab, i'd think that'd be the best thermal transfer method w/o using glues or other messy sticky methods.
Thanks!

 

Hello,
I'm interested in doing some remote temperature sensing of some high power mosfets on a PCB.
I've never used this pnp thermal diode method before, but it seems like the MAX6622 would be a good part for my application. Now my question is, do they have any pnp transistors with an isolated heatsink (tab) of some sort, preferably small surface mount packaged? if I can mount that pnp on the heatsink as the mosfet's drain tab, i'd think that'd be the best thermal transfer method w/o using glues or other messy sticky methods.
Thanks!

The data sheet - see page15: does not recommend "large" power transistors. What they suggest are small-signal devices, and such things do not normally have isolated heatsink tabs. http://datasheets.maxim-ic.com/en/ds/MAX6622.pdf

 

Another thought: you could use a thermocouple. The reason I mention it is because I think of the heat sink as an aluminum extrusion, which means it can have a tapped hole. All that it takes to mount a thermocouple is a small screw to clamp the thermocouple junction. And you don't need to weld the TC joint like most folks think -- just give it a couple of twists and clamp under the screw.

A downside is that this will measure temperatures relative to ambient unless you put in some kind of zero point reference. But I'd imagine you're after temperature changes, so you might not need the reference. I've done this with type K thermocouples in a car that lost the temperature gauge and I had to haul our trailer across a couple of states on a hot summer day. I stuck a TC under a hose clamp on the cast aluminum of the water pump/thermostat cover and measured the TC voltage with my Fluke meter while driving. Crude, but it worked and got me there safely.

 

Thanks for the help guys,
the thing about the FETs is they're D2PAKs and surface mounted. I think I'll run some high current through them to heat them up, measure their case temperature w/ a handheld temp sensor, and compare it w/ the readout from MAX6622, and try and determine an estimate of the thermal difference between the FETs themselves and the IC if I mount is closeby them.

 

Odd; they talk about using PNP transistors all through the MAX6622 datasheet, but on page 15 table 10, they're recommending NPN transistors - go figure. Looks like the proofreader was asleep that day.