Thermal resistance through plastic case top?

Hi,

Heat is generated on the chip:

In one case the way for heat spreading is:
* chip --> metal sheet --> thin isolation plastics --> external heatsink
Where the chip is directly mounted on the metal heat spreader, and the metal provides a rather large area to transport the heat to the external heat sink

in the other case
* [metal sheet] --> chip --> thick mold plastics
Here the metal sheet is on the opposite direction of your intended heat flow and the mold is rather thick and not optimized for heat flow.
The chip is designed for heat flow in driection of metal sheet and PCB .. it is not optimized for TOP heatsink. So this is only a (bad) option.

There are other IC´s that truely are designed for heat transport to the TOP side of the IC .. and TOP_mounted heatsink.

Klaus

cupoftea said:

Why are these figures so different?
They should be the same.
Plastic is plastic.

Click to expand...

You are just misinterpreting the data. "Junction-to-case" refers to metal tab in case of TO-220. IRFH7440 datasheet gives separate thermal resistance to case top, the value clarifies that only a small share of rated power can be dissipated through case top.

You are just misinterpreting the data. "Junction-to-case" refers to metal tab in case of TO-220.

Click to expand...

..Oops sorry, my apologies... i meant the TO220FP (plastic TO220) package.
Ie the plastic TO220.
It has Rth(jc) = 3.6degc/W

..and thats through plastic...so why is it so different for the "thru plastic " of the IRFH7440.?.....or is it a different kind of plastic...and if so, "why is that so" , is what i am wondering?

Rth difference between T0-220 and TO-220FP is about 2.5 K/W, that's the thermal resistance of thin insulation layer between internal metal tab and bottom surface. KlausST already discussed the details.

As yes, thanks to Klaus and yourself FvM, i now see that the internal structure is totally different.
I must confess i didnt realise that inside the plastic tab of a TO220FP, is actually a metal tab.

The case has 6 sides.
Measuring the bottom side metal case, the Rj-case is given. (not the top)
When there is no heatsink other than the minimal pads on FR4, the Rj-pcb thermal resistance is much higher.
This reduces the resistance in half.

So if testing the thermal resistance of the PCB layout, one would use the thermocouple on the PCB next to the case rather than the top case which may have more thermal gradient than the PCB close as possible, (check and possibly on the other side of the board)

The point of the OPs confusion seems to be where on the case this applies as there are many variations of case design including Al, Cu and Steel framework inside the plastic and some have heatsink interfaces to the PCB interface. There are obviously 6 sides of the case and only 5 accessible when mounted, so the T.pcb is a convenient reference point used by some but not all OEM's.

When in doubt, use the PN junction voltage drift in an oven to calibrate your chip in oven, chip on board powered up at room temp. The oven with a known low DC current in low power <<1 mW will be an accurate thermometer for the junction.

But this should always be true.
θJA = θJC + θCA.

https://www.analog.com/media/en/training-seminars/tutorials/mt-093.pdf