Mounting remote FSD? Use compound??

[20050627|3|008327|000043|65.183.99.20]

Just wondering if anyone uses thermal compound when mounting the FSD to a remote cooler. I'm thinking of the stuff used for computer CPU's.

[BobND]

Here's what I've done on mine, and 2 customer vehicles.

I suspect the plastic covers over the two power transistors are to catch the nuts, if they fall off of the 4 transistor mounting screws.

I would think they would cut heat radiation to the heatsink at least a little, so I removed them.

I took the 4 nuts off of the power transistors, one at a time, put a drop of high strength LocTite on the threads, replaced the nuts and tightened the $#^@ out of them. Then, I put a dab of clear silicon sealant on each nut, and the protruding threads, to eliminate ANY chance of the nuts shaking loose. I let the silicon cure overnight.

Then, I took a tube of silicon dielectric grease and filled the cavity around each tranistor almost level full, thinking this would aid heat transfer to the heat sink.

I used computer CPU heat sink compound between the FSD and the heat sink. (The GOOD stuff, with powdered silver in it.)

Then, I carefully cleaned off any excess that squeezed out the edge, an ran a bead of clear silicon sealant around the parting line beween the FSD and the heat sink.

I used large extruded heat sinks, and mounted them behind the LH headlight.

No problems, whatever, so far, with any of the three.

[Hubert]

I would say yes use thermal conductive paste. I used it on both sides of the thin metal "gasket" that comes with a "GM Delco" FSD.

Also make sure the heat sink is flat and mates well to the FSD.

[20050627|3|008327|000043|65.183.99.20]

Followwed most of BobND's advice. Didn't touch the screws, but filled the voids with dielectric grease and used the good CPU stuff on the gaskets.

Started up and ran fine. Took a temp reading with one of those handheld temp tools and was around 125 after a good hill climb. Never checked what it was on the pump.

And, based on the idle and pickup, I think it was a good thing I did this. The truck had started idling rough a couple days ago and hesitated on me a couple of times. Seems to be better now.

[BobND]

I haven't been using the OEM "gaskets", just the good CPU heat transfer grease with the silver particles.

Anyone have a GOOD reason to use the "gasket", if the heat transfer compound is used, and the outside "parting line" where the FSD meets the heatsink is sealed with silicon for "weatherproofing"?

[DmaxMaverick]

I would not recommend using thermal conductive paste on the FSD. The GM "gasket" is far supperior for this application. Do not reuse the gasket once it has been tightened down. If you don't have an unused gasket, several layers of alumnum foil will do just as well. The object is to conduct heat away from the FSD and onto the heat sink or pump.

The reason paste is used on computer processors is specific to its use. Most OEM processors that are paired with a heat sink, have a heat sink that has the surface machined. Aftermarket heatsinks are seldom machined, but are simply extruded aluminum. Extruded aluminum has a porous, uneven surface, and needs something to take up the gaps. The processor's need for heat transfer is a much higher demand than the FSD. Although the processor's actual heat output is much less than an FSD, its microscopic hot spots are higher, in the 400

[JohnC]

The purpose of the gasket is "heat transfer". I seem to recall when I was in school, which was a loooong time ago, that every time heat has to transfer across a material boundary it looses efficiency. So, I suppose you could make an argument for using the gasket without the grease, or the grease without the gasket.

[BobND]

I've been having good luck with the conductive grease, but have not taken any heat measurements vs. the foil gasket, so I certainly won't defend the practice "to the death" without being able to positively back up my observations with actual temperature measurements

It certainly would be interesting to do so, though.

I would be EXTREMELY surprised if the conductive grease and filling the power transistor "pockets" with silicon dielectric grease does not do a MUCH better job of heat transfer than the way they were originally set up.

Using the foil just sets up another "interface" between surfaces that the rejected heat needs to cross to get to the heat sink. (The heat-sink units I have used have a very smooth, almost machined-looking surface, and I see no problems with the CPU heat transfer compound.)

[DmaxMaverick]

BobND

When it comes to the FSD's, the logic is pretty fuzzy. Some folks will pamper and baby them, to have them fail in short order. Others kick 'em, spit on 'em, beat 'em, cuss at 'em, etc, and they go forever. All this may be for not. The discussion is about heat transfer, not rolling the dice with an FSD. What works and what doesn't may not have anything to do with what we are discussing.

Filling voids with grease?? Very bad. Air space is a better heat conductor than a grease insulator. The grease will prevent the heat from escaping. You could use conductive potting compound, but not grease, if you want to fill the void with a conductant. If it is heat that is killing the transistor, the grease will accelerate its death.

This has been beaten to death for years.
Search the forum for posts by GMCTD. You may be surprised at what you will find.

[BobND]

So you feel that leaving the power transistors covered up with a plastic cover will keep them cooler than being surrounded with silicone thermal compound?

[Hubert]

Hmmmm microscopic? I think we all are saying the same thing. I think the "gasket" and/or paste are for the same reason to ensure conduction of heat not conduction convection conduction because of an air gap void in the mating. As long as the thermal conductivity of a paste and/or gasket is high either or both will work. I think both are safer. The backing of the FSD I used GM dealership bought looked like a lead material (I did not investigate too much, too expensive to scratch or play with) and really did not seem tremendously flat. At least not within a thousands or 3 of an inch. Neither was my heatsink. But they were flat within several thousands of an inch - close enough for me with paste and "gasket".

Definately don't want ANY voids between the two.

[moondoggie]

Good Day!

[i]

[johnny2can]

My two cents, based on my own (limited) research...

Generally speaking...

The purpose of a paste or pad is to eliminate the microscopic air gaps between the transistor and the sink, as the air will act as an insulator (the paste or pad will, also, but much less so than the air). Ideally, you would perfectly machine both surfaces and not use either paste or pad (but we're not idealists here... :cool: ).

It used to be that paste was MUCH preferable over pads because the paste has much better thermal conductivity (better heat transfer), but newer pad materials (last coupla years) have gotten closer (but not yet equal to) the conductivity of good pastes. Pads are also somewhat more tolerant should the clamping force be reduced for any reason.

But again, I read this. I'm playing armchair thermodynamic physicist based on information gotten off a heat sink mfr's web site. :D

[GMC Hauler]

So.....

What we need to do is some testing to settle this discussion.... I propose a evaluation, of both types of junctions to determine if there is any difference.

What would be needed:

2 identical heat sinks
2 fsd's
temperature probes
independent tester

I just happen to have: 2 identical heat sinks and thermocouple probe (cromel/allumel type). The thermocouples will have to have a junction made with a TIG welder.

Do we have any volunteers out there to prove their point with their spare FSD? I'll let whomever provides their FSD keep the heat sink they use. FSD's and mounts would be sent to a third party independent tester to prove their point.

If necessary, if I can get the FSD off my pump, I'll use it as a pad mounted version. I'd, however like to wait and see if there are any volunteers out there for both types of junctions.

http://pic14.picturetrail.com/VOL514/3320954/7024686/111019854.jpg