Experiment: Which mosfet thermal interface material is best?

Status
Not open for further replies.
Jump to Latest
#21
oreo: See e) above in Methods section. Screw head both measured highest and provided best consistency.

kasey: you are quite welcome!

rookakoma: black kept the mosfet the coolest, resulting in slightly higher sink temp. This makes sense: heat from the mosfet is dissipating through the sink the best. If I had a DMM that reported temps with decimals (rather than integers), there might be more interesting things to see here.

Mark: Wow- I didn't know such a device existed! This looks like it would provide better data. Next time I place an order, I'll have to grab a few.
 
#25
Eric said:
rookakoma: black kept the mosfet the coolest, resulting in slightly higher sink temp. This makes sense: heat from the mosfet is dissipating through the sink the best. If I had a DMM that reported temps with decimals (rather than integers), there might be more interesting things to see here.


Eric, Wrong conclusion!

Better heatsink emission means better dissipation and / or better radiation.
At your experiment the dissipation factor is constant (because you used the same heatsink), so better radiation makes cooler heatsink.
1°C is not too much at heatsink... It makes 1°C cooler device (when the thermal resistance between the device and sink is constant).
The reason of the different device temperatures at your experiment is th. res. between the device and heatsink.

This conclusion based on the lows of thermodynamic.
Click to expand...
 
#27
I would have expected a good mica insulator to be better than the others, but with serious fussiness with getting a very thin paste layer. Your mica must be thick or something was not flat.
If you use so much paste that is squeezes out, you have too much and may have big trouble later as it dries out.
 
#28
rookakoma: fair enough - thank you for the additional details.

davidsrsb: it is possible that something was not flat. The mounting hole was drilled on my drillpress, so I'd expect it to be reasonably perpendicular to the surface of the sink, though it is possible that something is off. The same hole was used for all measurements, though some of the thermal interfaces were more rigid than others. If the hole is not perpendicular and this disadvantaged the mica & grease, then the AlOx performs even better than measured as it is also rigid.

One of my motivations was to actually see what happens since we all have certain expectations. I like data.
 
#30
MEPER said:
How would only thermal grease (the white stuff) compare?
Click to expand...

Some of the comments that were posted made me wonder about about how perpendicular my hole was in the surface of my sink. I double checked my drill press - the table is as perpendicular as I can make it (I ended up not actually adjusting it). I drilled a new hole and was as careful as I could be to make sure the hole was true. I then re-ran my experiment as described in my first post. The results were the same (ambient temp may have drifted a bit, but I compensated in my table to provide the same starting point).

I added a new measurement: only thermal grease was used between the mosfet and the sink. This only works for a single channel of the MoFo amp, DON'T DO THIS WITH ANY OTHER AMP HERE - you'll create an electrical short.

I expected thermal paste alone to perform the best, but the magnitude of the difference was a bit of a surprise to me. I have attached an updated chart with the new test condition.

Finally, the thermal compound that was used for all tests is Ceramique2, Tri-Linear Thermal Compound by Arctic Silver.
 

Attachments

  • Thermal Interface.jpg
    Thermal Interface.jpg
    82 KB · Views: 554
Last edited:
#32
Eric said:
I expected thermal paste alone to perform the best, but the magnitude of the difference was a bit of a surprise to me. I have attached an updated chart with the new test condition.
Click to expand...



With thermal paste only the temp of the screw is significant lower than the other options. What can be concluded from this observation? …...that the temp of the chip/junction is lower also using thermal paste only?
 
#33
david: I will measure the mica thickness tonight and let you know what I find. I used the mica that was supplied in the VFET2 kit just because it was available.

MEPER: My interpretation is that lower temps of the mounting screw indicate lower transistor temps because there is better heat flow from the transistor into the sink. As the performance of the thermal interfaces improved, the mosfet temps were successively cooler. Thus, the best thermal interface is the minimal one: just goop. If I'm wrong, someone please correct me...
 
#38
Goop only...

Hi Eric,

When trying to get the very best thermal contact between a CPU and heatsink I have lapped the CPU and heatsink with fine “wet & dry” to ensure they are as flat as possible (thus maximising the contact area). Could you do the same and report relative results from your test rig please? I’m asking because I will be directly mounting my FETs to the heatsink with goop...
 
Status
Not open for further replies.
Top Bottom