Building the F2 - teach the newbie please! (one question at a time)

[DQ]

Hi. Since this is the great week for me, when I'll build my long awaited (2 1/2 years!) F2, a few questions are coming up. Please return to this thread even if you answered my first question as there will probably be at least another couple or so.

So, for a start, about mounting the FETs on the heatsink. I have two options, please tell me what would be the best:
1. I found some mica locally, looks pretty good (thin but I don't know how thin, not too many scratches) and I have some Arctic Silver Ceramique
2. I bought from Mouser some Laird Technologies T-gard 5000, they don't look like those in the picture, are rectangular instead and just a bit larger than the IRFPs. I suppose they should be used without thermal compound, right?

[DQ]

nobody?

[juma]

Originally Posted by DQ

nobody?

Nobody answers because it's pretty basic stuff - you'll be fine with either mica and white goop or with this new material (datasheet characteristics are fine) - I'd use the goop here too, it's always good to fill the small irregularities in material.
Now you have some answer.
Next guest in this thread will write about one material being better than the other, and the next one will claim the opposite, the third one will be nitpicking and revealing the mistakes of the previous two and soon you'll know the new meaning of the "being confused"

If I were you, I'd be curious to try this new material (since you already bought it) and share the experience

[DQ]

Lol! Thanks for your answer.

Well, just when you were writing your answer I took another look at the datasheet of the T-gard datasheet only to notice that "T-gardTM 5000 is ideal for applications that require a delta temperature across the interface of 2.0°C/watt or
higher on a TO-220 clip mounted @ 50 psi pressure." I would have thought a delta temperature of roughly 2 degrees / W (OK, it's TO-247, not 220) would not suffice for the 70W heat dissipation / 2 IRFP devices / channel of the F2...

Edit: and the mica appears to be about 0.05mm thick.

Re. the very basic question: I know it is, but this is my level of knowledge and it's not always so simple to find out what I need to know by my own. I usually use the "search" button on several forums, google it etc. before posting a question. I don't want to waste people's time, that's why I ask the questions in such a manner as to facilitate a simple answer like "Option 1." or "Option 2." I'm grateful for any help I may receive.

[juma]

You misinterpreted that. This is the part of the datasheet that is important:

[DQ]

You're right, I was calculating as if the thermal resistance would increase instead of decrease with increasing the surface! Stupid! In this case, for the ~2 square centimeters of the metallic part of the IRFPs I would get say ~0.7degrees/W which would roughly translate into a 25 degrees temperature difference between the mosfets and the heatsink. Should do, I guess...

Thanks again.

[juma]

Mistake again - please concentrate
The datasheet says:
0.28 degrees C / Watt on SQUARE inch (in2) of that material.
And 1 square inch = 6.5 square cm.
And that's when you use the screw and bolt to mount the transistor (not the clip).
OK?

[Rodeodave]

Mica and Goop has been and still is being discussed there.
What will you be mounting? The TO-247AC package has an isolated mounting hole, so no need to isolate the screw. Nelson is using washers to spread the force over the device, as can be seen in the pictures of the insides of the various F amplifiers.

[DQ]

Juma, why am I wrong again? In the datasheet it says 1.26 degrees/square cm/W (see attachment), so at about 2 square cm IRFP surface this would give about half or 0.63. I rounded it up to 0.7 just to be sure. 1 square inch is > 3 times the IRFP surface and that's why the figure / square inch is much lower. If the heatsinks I am building will be efficient enough, around 0.3 C/W that is, during the hot summer days when I get a bit over 30C in my room the contact surface of the mosfet should be at 70-75 Celsius. I hope this would be acceptable.

On the other hand, there comes Rodeodave just to fulfill your predictions. Well, you didn't actually make a recommendation, Rodeodave, and I'm not confused (yet) but the thread you pointed me at (thank you) seems to suggest that mica and thermal compound make a pretty good, tried and safe combination that should give around 0.4 degrees / W - but I didn't understand on what surface, perhaps you know? Is it on a square inch? Yes, I will use washers.
Another thing in this thread that popped into my attention is the recommendation to smooth sand the surface of the transistor as its plastic border could be higher than the metal surface and thus keep it from getting into contact with the heatsink. Unfortunately, 600 or 800 sandpaper has been mentioned and the best I have found so far was only 240... Not to mention that I have to smooth the surface of the diy heatsink (not such a good idea btw, but now it's already 70% done) because drilling the holes for the mounting screws resulted in an elevated rim around each hole.

I e-mailed Laird Tech. asking if their product should be used in conjunction with a thermal compound or not and I got the following answer:

Basically T-gard 5000 will fill micro undulations on the surface of the top/bottom heat transfer mediums, but as T-gard 5000
is only 0,005" = 0,127mm thick, the efficiency of this process is limited.
T-gards are actually just Thermal Interface Insulator materials.
Thermal greases will not decrease thermal performance and could be used to fill rougher top/bottom surfaces.

Thought about sharing it, perhaps could be useful to others as well.

[juma]

Why did you choose 1.26 figure ? It's for mounting devices with a clip. You will use screws and bolts.
Read the post #7 carefully, look again at the datasheet, and if you still don't get it, just use the */&;@#! thing and stop worrying