[Amps] Arctic Silver

Bryan Swadener via Amps amps at contesting.com
Sun Apr 12 09:26:52 EDT 2015


Hi Mike,
When I worked in the RF Lab for the worlds' largest OEM of RF-driven CO2 Lasers, I was one of the key players in developing the method of attaching APT (now Microsemi) ARF1500 FETs to our water-cooled heat spreader.

We used ordinary white "bird poop" thermal compound. We determined that, with a mounting surface that is flat and smooth (I don't recall the numbers for flatness and surface finish), it was critical to use the LEAST amount of thermal compound.  We didn't then just fasten the devices to the heat spreader. We "lapped" them into place using a figure-8 motion while applying pressure. We found that micromotion from a modified corner sander improved and sped-up the lapping process.  The corner sander output was modified to accept a 1/4" thick piece of rubber that was cut to fit the top of the ARF1500. It worked very well, to which Dick Frey (Microsemi Application Engineer) agreed.
The result of our work agreed with Kevin and the recommendation of others in industry -- using a layer of thermal compound greater than that needed to fill voids between the device and its mounting surface acts against cooling of the device. More goop is not better -- it's worse.
vy 73,Bryan WA7PRC
 
    Date: Sat, 11 Apr 2015 12:54:29 -0500
From: Mike Waters
Subject: Re: [Amps] Arctic Silver

This is generally good advice. However, as Manfred nicely pointed out,
"...applying an ultra thin, even layer is almost a guarantee for entrapping
air bubbles. And that's definitely worse than a thick grease layer!".

I fully agree with Manfred. It depends on the viscosity of the grease,
whether it's thin enough to be easily squeezed out by applying pressure.

73, Mike
www.w0btu.com

On Sat, Apr 11, 2015 at 9:21 AM, Kevin Stover
wrote:

>
> If after securing the transistor to the heat sink or heat sink to a
> processor you see heatsink compound oozing out the sides you've used waaay
> to [sic] much.
>
> The layer of heatsink compound needs to be an even layer almost thin
> enough to see through, with NO air bubbles.
   


More information about the Amps mailing list