OT: Has anyone tried this product for heat-sink compound?

htm

What is it like when it ends up on your clothing if assemblers use handfulls of the stuff rather than ,all that is necessary, pin-head amounts. Let alone large area bolted together Al to Al , H/S to chassis etc - can anyone explain why its necessary there at all let alone handfulls of it

er.

lls

one

It handles the same (consistency wise) as any thermal grease. It's gel- like but "hardens" to a flexible but non-bonding constituent.

I've never tried it. However, looking at the MSDS sheet at:

there doesn't seem to be any copper in the mix. I can't even figure out which of the ingredients give it the copper color. Also, the data sheet includes all kinds of interesting characteristics, but conveniently leaves out thermal conductivity.

Since 80% of the contents are silicon greases, and the rest is a mix of silicon dioxide (ceramic) and mica, it would appear that the thermal conductivity is primarily controlled by these additives. Both are good thermal conductors, but in order to be really superior, the flakes or gains need to touch and overlap in order to conduct the heat. With less than 20% concentration, that's not going to happen.

Note the thermal conductivity of silicon grease as compared to various metals, which are about 100 times more thermally conductive:

--
Jeff Liebermann     jeffl@cruzio.com
150 Felker St #D    http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann     AE6KS    831-336-2558

The reason I think it's a possible a good substitute is...it squeezes down to an extremely thin film. If surfaces are nearly perfect...there would be almost nothing between them. Thanks for your input.

Wow....really? You actually *want* to glue the parts to the heatsink? Whatta mess that would make.

This is a non-bonding type of silicone used for sealing instead of a gasket. Parts can be easily disassembled.

I haven't tried it. It obviously was formulated as a substiture for or improvement to gaskets in applications like automobile thermostat housings. As such, it was designed more for stability at high temperatures rather than for good thermal transfer properties.

Perhaps by using it between an older CPU and heat sink you can determine it's heat transfer properties. Let's face it, if someone had said "I tried it on my now computer, and the CPU runs 20°C cooler." would you use it wthout confirming the performance?

PlainBill

Have you ever measured the thermal transmission characteristics of various heat sink mounting methods and the effects of thermal grease? I have (in about 1993) using a home made fixture[1]. I have the report buried somewhere, but as I vaguely recall, the best thermal conductivity was achieved when both the device and the aluminum heat sink were polished flat until they looked like mirrors, with no thermal grease in between.

Second best was the usual rough and warped heat sink surface, with a sheet of silver leaf in between and no grease. Silver is very very very malleable and will fill the cracks, gouges, and small pits very nicely. However, it's rather expensive mostly because it can't be beaten into a thin enough sheet. Gold leaf can do this, was available in very thin sheets, and worked almost as well as silver. Oddly, gold leaf on polished heat sinks make things worse. Some RF xsistors had gold plated copper bases, but as these were rough and warped, the gold did little to improve thermal conductivity.

The absolute worst was a massive heap of silicon grease, as found on most CPU tops. This prevents metal to metal contact, which is really what you're trying to achieve. The grease just fills in the cracks, evaporates somewhat, and leaves aluminum oxide dust in the crack to provide some additional heat sinking. By itself, aluminum oxide doped silicon grease is not very heat conductive. I didn't have any Arctic Silver available in 1993 to test.

This test was to see what could be done to get more heat out of commonly available RF power transistors. The use of thermal grease resulted in extremely erratic xsistor temperatures, as different assemblers would apply radically different amounts. One surprise was that the most important consideration was RF transistor base warpage. The final procedure was to polish the xsistor bases on fine sand paper until the gold was gone and the copper base was bright and shinny. Same with the heat sink except it was done on a buffer.

The same method will probably be useful for CPU cooling. The newer CPU packages are already have a shiny flat mating surface. All that would be necessary is to polish the mating heat sink. However, unlike the RF power xsistors I was testing, most CPU sockets do not have the luxury of compression screws, which will squeeze the gold into the cracks and crevasses.

[1] I didn't use these methods. I was only interested in the temperature differential across the power transistor to heat sink junction. I heated up the transistor, and measured the differential temperature across the gap after the heat sink temperature had stabilized. To get a decent reading, the test heat sinks were identical solid blocks of aluminum, cooled with chilled water running through the block. [2] Gold leaf is not as expensive as you might thing, but is a major handling problem. Touch it with anything and it will stick. You can't manually handle it or it will tear. Watch out that you're not buying "imitation" gold leaf (Dutch gold), which is thicker, and is mostly brass, but not as thermally conductive.

--
Jeff Liebermann     jeffl@cruzio.com
150 Felker St #D    http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann     AE6KS    831-336-2558

gask...

The best compound is one of micronized silver and ceramic particles. After all you are trying to fill microscopic voids between device and heatsink so they need to fit.

--
Live Fast Die Young, Leave A Pretty Corpse

Did you try silipads in your comparison study ? where they around in 93 ?

Memory fault and I goofed. It was about 1983. I thought I had scanned and posted the manual for the Intech M3600 radio, but I guess not.

Yes. They were a total disaster. Berquest and Thermalloy were making them in one thickness and material type at the time. I don't recall what it was, but might be able to find the data in my "archives". What Sil-pads were good for was the ease of handling, consistency, and cleanliness. I would have used them if they were suitable. Well, I did use them for TO-3 and other non-RF power devices. However, this was a performance problem, not a manufacturing exercise. The price you pay for the benefits of silpads is a modest thermal resistance, which was unacceptable.

We were trying to squeeze as much power out of a pair of transistors, that would normally dissipate about 175 watts each. This had to pass through a 30 mm^2 base to a much larger heat sink. Almost any thermal resistance between the xsistor base and heat sink would raise the xsistor temperature enough to initiate a thermal shutdown. Under ideal circumstances, the heat sink should be very close to the temperature of the transistor base. It was, but when a few degrees of differential temperature translates into many watts LESS of RF output, any difference was considered a bad thing. The transistors could produce almost any RF power output desired, if only they could be kept sufficiently cool. The other incentive was above a certain junction temperature (which I've forgotten), thermal runaway is a real problem.

There was another problem with silpads. The base of the xistor was a round disk with ears. We had several bases, but this was the problem maker:

When we torqued down on the mounting screws, the sil-pad material tended to bunch together near the mounting holes, resulting in the base taking the shape of a "bow" with less pressure in the middle. Of course, the place where we needed the most compression was in the middle. I contrived a layered spacer that sorta worked, but it increased the thermal resistance even more by increasing the gap.

Incidentally, I also tried beryllium oxide insulators, mica insulators, aluminum foil, brass shims, and various home made mixes of aluminum oxide dust. All of them worked when properly applied and with proper compression. However, direct metal to metal polished contact worked best.

--
Jeff Liebermann     jeffl@cruzio.com
150 Felker St #D    http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann     AE6KS    831-336-2558

If you must use a compound, I suggest a mix of some kind of grease that will vaporize when the heat is applied (hopefully leaving little residue), and diamond dust:

Silver is good, but diamonds are much better.

However, the ultimate in thermal conductivity is still metal to metal. A good example are the diodes used in automotive alternators and industrial power equipment. These diodes are press fitted with an arbor press into the heat sink. Metal to metal at it best with not even microscopic gaps. I tried to get the RF power xsistor vendors interested in an interferenc fit package, but because I didn't have a sufficiently large purchase order, they weren't interested.

--
Jeff Liebermann     jeffl@cruzio.com
150 Felker St #D    http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann     AE6KS    831-336-2558

This stuff is NOT a silicone grease but rather a room temperature vulcanizing silicone rubber. There is no copper in it, and it is a poor heat conductor. General Electric used to be the manufacturer, but it appears that they have sold out to a Chinese company. If memory serves it used to be either RTV-60, RTV-650, or RTV-106. I cannot remember which is the two-part and which is the one-part material. MG Chemicals is a distributor for Momentive Performance Materials, the _obviously_ Chinese company that makes it now. They do make two higher thermal conductivity silicones known as TSE3331 and TSE3941. The former is a grey potting and encapsulating compound while the latter is a white adhesive.

As Jeff Liebermann pointed out, the smoother the surfaces are and the closer they fit, the better the heat transfer will be. All the silicone heat transfer grease does is fill the microscopic void spaces (dead air space is a good thermal insulator).

Working around chemists, I had easy access to silicone vacuum stopcock grease. I have used it in a pinch for thermal grease and it worked well. You can find a similar grease in the plumbing department of Lowes and Home Depot. It is used to lubricate valve stems, rubber seals, and those ungodly expensive ceramic Price-Pfister faucets. It will work far better than any Permatex product as a heat transfer grease.

Oh, and one other reason to not use Permatex in this application. This silicone is an acetoxy cure which means it releases corrosive acetic acid when it cures.

Dr. Barry L. Ornitz WA4VZQ ham_call_letters at live.com

r

l.

se

er

s at live.com

Thank you, Dr.

Overclockers used to (and may still; I don't play any more) spend a fair amount of time on this for CPU heatsinks. I gave it a spin once, using a polished marble building tile for the reference surface, successive grades of wet-work polishing paper, and a lot of water. Much work for just a few degrees change in delta-T but it was instructive. Also, it was pretty cool to see just how well-polished one could get a copper surface.

--
Rich Webb     Norfolk, VA

Yep the best is no need for any compound. But in an imperfect world it's not possible on much consumer grade electronics. I switched to the Arctic Silver paste after using a generic white compound on my new PC CPU. It's a 120 watt quad core AMD overclocked from 3.2 to 4 ghz. I noticed the CPU running very near its upper limit so I removed the compound and applied the Arctic paste. The results were a 5-7 degree Celsius drop in idle and load temperatures. The cooling apparatus was packaged with the CPU so I know it was designed specifically for it. No it's not a drastic drop in temp but enough to convince me that for my purposes it's the best choice of pastes.

--
Live Fast Die Young, Leave A Pretty Corpse

r

l.

se

er

s at live.com

The tube I have says, "Made in USA".

..

..

oor

es

ls

ic

ell.

hose

tter

id

ers at live.com

The Dr. is right, except for possibly one point: Ultra Copper does NOT release acetic acid when it cures because it's instead alcohol- based to be safe for use in cars equipped with exhaust oxygen sensors. All of the Permatex silicone RTVs with the word "Ultra" in their names are made for oxygen sensor compatibility, and all of them smell like alcohol before they're cured. Also "Copper" refers only to the color (actually reddish brown) and physical characteristics and is not an indication of any copper content (none). BTW Permatex doesn't produce silicone RTV in different colors merely for cosmetic purposes, and many if not all of the colors indicate different physical characteristics. For example, Ultra Copper is made to withstand higher temperatures than any of their other RTVs, while Ultra Grey is for gaskets with bolts that are close together, and they have others made for withstanding gear oil or antifreeze:

~The tube I have says, "Made in USA".

I suspect Permatex used to buy from GE because their sales would be too small to manufacture it themselves. If you look for GE silicones, many companies still stock them. However if it is made in China and sold in bulk and Permatex packages it for sale, I think they can get away with calling it "Made in the USA."

If you go to a gardening supply and buy genuine Holland tulip bulbs, you will likely get bulbs grown in Washington state. These bulbs are shipped to the Netherlands and packaged. This allows them to be called "imported from Holland."

Most pharmaceutical intermediates are today made in India or China, and many drugs are totally made there and shipped to U.S. manufacturers who compound the drugs with fillers and then stamp out the pills. They then qualify as U.S. made pills.

As you no doubt can tell, I am not very enamored with the pharmaceutical industry. I take colchicine for ankylosing spondylitis. As a drug, it has been around for over 2000 years and is normally used to treat gout. My usage is "off-label" meaning it was prescribed for other than its approved use. My rheumatologist gave it to me for its anti-inflammatory properties since my congestive heart failure prevents me from using NSAIDs (non-steroidal anti-inflammatory drugs reduce your kidney function). A number of generic drug manufacturers made it and a three month supply cost less than $10. One "unnamed" pharmaceutical manufacturer decided to go through the FDA testing and got it approved. They then were able to get the FDA to force all the generic houses to stop selling it. That three month supply is now well over $400. By the way, these numbers are AFTER insurance.

"Made in the United States of America" somehow doesn't mean what it once did.

73, Dr. Barry L. Ornitz WA4VZQ ham_call_letters at live.com