Negative Heat Sink thermal resistance

[TeddyLu]

I'm currently developing a heat sink for a product that produces up to 150 W max power dissipation. I'm currently getting a negative thermal resistance for the heat sink. I don't think that is possible..

I'm not sure how to move on at this point to properly design a heat sink to meet the heat dissipation needed by the chips used. I don't want to blindly design a heat sink and hope for the best.

The chip has 3 layers of material to go through before the heat gets to the heat sink. All stated temperatures are based off the 150 W max power dissipation, so they're the max temperature of that level at that max power dissipation

Max Channel Temp of the chip is (275.00 °C)
thermal resistance of the junction is (1.13 °C/W)
Temp of the copper moly copper under the chip is (41 °C)
thermal resistance of the copper moly copper is (0.0427 °C/W)
Temp of the housing is (34.5897 °C)
thermal resistance of the copper housing (0.0277 °C/W)
temp of the thermal compound (30.4294 °C)
thermal resistance of the thermal compound (0.0029°C/W)
temp of the heat sink (29.9976 °C)

Using 35 °C as my ambient air

I'm getting a thermal resistance of -1.619670533 °C/W for my heat sink.

Are my calculations off? I noticed that if I relieved the power dissipation to more nominal working levels, the thermal resistance makes more sense with a positive value. Any suggestions is appreciated!

 

[Rive]

temp of the heat sink (29.9976 °C)

Using 35 °C as my ambient air

One quite interesting heatsink which can remain at 30°C in 35°C ambient air

Ps.: as a side remark - the temp of the housing (34.5897 °C) is just not a good match for the dissipation (150W). We are fighting with 50W dissipation, and the copper housing is around 100°C already, with a quite decent active water cooling. I don't know the details of your project, but that 35°C is just not OK. You need some serious redesign around the junction. Maybe your build is not OK for power devices?
Or something is off with your math.
Check out some custom "IGBT bare die" or "IGBT direct bond" solutions to see how it is done when dissipation matters.

Ps.: I found you a reference for the required thermal resistance values: https://www.electronics-cooling.com...allenges-of-high-power-semiconductor-modules/

 

[TeddyLu]

Hi Rive,

Thank you for the link, I realized that I made an error on my math as I used (T_sink - T_ambient)/Pd_max instead of (T_jc - T_ambient)/Pd_max to get the thermal resistance of the heat sink needed. Most of my thermal resistances are positive now but there are a few that are a bit suspicious as they are still negative. (I'm calculating various die chips at this point)

I'm moving ahead and designing a heat sink to see if my calculations for the thermal resistance that are positive, to see if they are somewhat accurate to fit the heat dissipation needed for those chips.

 

[Rive]

Most of my thermal resistances are positive now but there are a few that are a bit suspicious as they are still negative.

You missed the hint, then... In this setup, the heat goes one way (it should) , so the temperatures of the different stages goes down all the way from the channel to the heatsink.
A negative thermal resistance is like a negative speed in solution for a mechanical problem: it means that the math is OK, just the stuff wanna' go the other way.
In your original problem - there is no way that the heatsink has lower temperature than ambient - since the heat comes from the heatsink, to the environment... With reversed temperature of course it gave a negative resistance.