Greetings, This post is about thermal grease for processors. Many PC technicians think highly of silver thermal grease. What do people here think about it? Does it make sense to use bulk thermal grease? Or is a single use syringe the best way to go? Thank you.
--
( \_/ )
(='-'=) Don Kuenz
(")_(")
Don Kuenz wrote in news:m5i4q1$obk$ snipped-for-privacy@dont-email.me:
Just about any reputable brand of thermal transfer grease sold in bulk for general eectronics use will do fine if properly applied.
-- Ian Malcolm. London, ENGLAND. (NEWSGROUP REPLY PREFERRED) ianm[at]the[dash]malcolms[dot]freeserve[dot]co[dot]uk [at]=@, [dash]=- & [dot]=. *Warning* HTML & >32K emails --> NUL
Whichever you use, the absolute minimum you can deliver, not dollops of it by the handfull. Its only to fill any microscopic holes, any excess ,over that ,can only reduce thermal conduction if the excess is not squeezed out, so why place it there initially.
I currently use Arctic Silver because I inherited a box of the stuff from a shop that went out of business. My guess is that I have more than a lifetime supply because I use it VERY sparingly. The idea is to just fill in the cracks and gouges in the heat sink, not plaster the CPU or heatsink with a thick layer. More goop is NOT better. I'll guess that one tube lasts me about 12 Pentium 4 size CPUs. I apply a tiny blob of the stuff to the CPU, and smear it around with a plastic spatula until the entire surface is coated and that there are no lumps or blank areas.
I've also experimented with tooth paste, Vaseline, powdered aluminum suspended in vegetable oil, and other concoctions. Everything works with only subtle differences in CPU operating temperatures.
I also spent some time testing how a polished aluminum heat sink works against a polished CPU. The results were better than with Arctic Silver but had a problem. After a few months of operation, they would overheat. What was happening was the heat sink was expanding and contracting irregularly and would temporarily open a crack between the CPU and heatsink. Dust would get into this crack an create a tiny air gap, which was sufficient to ruin the cooling. With a heat sink that is tightly clamped down, this will not happen, but with the very loose cam clamps used by Dell in some of their early Optiplex motherboards, it became a problem.
I have some while silicon grease in a tooth paste size tube. I think it's at least 20 years old and is maybe 1/4th used. It really depends on how many CPU's will need grease. Adjust your quantities accordingly.
Incidentally, if you read the Arctic Silver instructions, they mumble something about the conductivity improving after the machine has been used for a while. That's true because the volatile oils, that make the silver dust flow, need to evaporate. If you leave the stuff open to the air, you'll end up with an unspreadable lump. Therefore, most such compounds need to tubes or syringes of some sort.
Don't forget to clean off the old grease and goo with some solvent before applying the new stuff.
-- 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
Do they? I don't know, but people who put together their own computers aren't "computer builders" they often follow instructions and have no deeper grasp. So once something falls into their view, they may become insistent, whether or not there is any value to a given product. They may not have even tried anything else, but since it works they soak in the marketing around the product, and repeat it.
They may be quite insistent on things, yet if things don't go well, they are lost. Something to cnsider.
Endless things are assembled with "everyday thermal grease" and no problems arise. That's an important thing to consider.
I think the single use packets exist because for most people, the only time they'll need it is when putting together a computer. it's convenient and you can throw the package away later.
I can't imagine it's different from a larger tube. But for someone who is only putting together a computer every few years, it may be cheaper to get the smaller package, and then they don't have the tube around to just sit there, or to figure out where they put it the next time they need it.
Michael
Most PC technicians aren't nearly as skilled or knowledgeable as electronics technicians.
The #1 maker/seller of silver grease, Arctic Silver, doesn't seem to know very much about it, not even the maximum safe voltage, but they did brag of it being used for IGBIs in windmill electronics that ran at high voltage, despite claiming that they didn't do any testing above 12 volts.
You can find performance comparisons of various kinds of thermal compounds online. I'd suggest a preference for tests *not* run by anyone who has a horse in the race.
You'll probably find, as I did, that the product-to-product variations are rather subtle.
Isaac
Similarly whenever you see test results for the related product silipads you never see the companies' products cross-compared with mica slips. Mica is non-patentable and dirt cheap with the technical advantage it does not creep , so does loose functionality over time and heat-cycling. What limited in-service testing I've done , mica does a better job in heat transfer, let alone longevity advantage
Mica also has the advantage of being non-conductive.
You would never want to use conductive thermal grease //by itself// when attaching a device directly to a metal heat sink. At least, I don't think so.
I want to thank everybody for their answers to my question. It looks like Dan put a lot of effort into creating the above web page, which compares thermal compounds. My takeaway from both that page and this discussion is that there is little measurable difference between commercial grade compounds.
Dan mentions Cooler Master on his page. Cooler Master offers a 200G tub of compound under the SKU of RG-ICFN-200G-B1. It seems to meet the needs of my business. It apparently comes with a credit card applicator to use instead of a syringe.
FWIW, all of the hundreds (thousands?) of motherboards and coolers that I've worked on over the years (decades) always used grey compound. Not one of them ever used zinc oxide.
There's the right way, the wrong way, and the Intel way of doing things. Doing it the Intel way works for me. Intel coolers come with a square of grey compound pre-applied. Intel warranty exchanges come with a syringe of grey compound when applicable.
Delta once had me swap out a few Dell motherboards at the local airport. The new motherboards each came with a syringe of grey compound. They also came with a pad coated with a chemical to remove the old compound in a single step process. Whatever was in that pad was corrosive enough to immediately start burning my bare skin.
Such are the hazards of service work. I always keep lots of vinyl, latex, nitrile gloves around these days. Never again will any chemical touch my bare skin.
In lieu of hazardous-to-your-health pads I use ArctiClean in my business. ArcticClean uses a two step process. In the first step a mild corrosive is applied. It smells of oranges and is probably acetic acid. In the second step a neutralizing agent is applied to stop the acidic reaction.
--
( \_/ )
(='-'=) Don Kuenz
(")_(")
The gray goo is some kind of metal (aluminum or silver) to provide the thermal conductivity. The white stuff is a ceramic (aluminum oxide or zinc oxide). Note the thermal conductivities: W/m*K Metallic Silver 406 - 430 Metallic Aluminum 205 - 250 Aluminum Oxide 26 - 40 Zinc Oxide 21 The huge differences don't translate into spectacular differences in final CPU temperature. That's because the thin layer of thermal compound is tiny part of the thermal circuit. You could use butter instead, and it will still sorta work. The thermal resistances are in the range of 0.01 to 0.10 C/W. It's like building a 120VAC resistance wall heater and then asking how much will the temperature drop if I insert a very small value resistor in series. It's barely noticeable.
80-way Thermal Interface Material Performance Test (14 pages)Gotta run, I'm late. Also think about packing density, heat sink warping, ductility, galling, viscosity, and re-usability.
-- 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
Thank you Jeff for making me rethink my choice. Upon reflection it now seems that Tgrease 2500 (
Intel may special order its grey compound directly from 3M or Dow. It wouldn't be the first Intel part that's not available through retail channels.
There's a controversy surrounding Arctic Silver 5's thermal conductivity. The company claims 8.7 W/mK but another lab measured it at
0.94 W/mK. So it seems that Tgrease works better than the storied Arctic Silver 5.
--
( \_/ )
(='-'=) Don Kuenz
(")_(")
I blew the clutch on my car. Wiped most of the day arranging for a clutch transplant. It's now 8:30PM and I'm in the office trying to catch up with promised work.
What I basically said was that it really doesn't matter. Like real resistors, the thermal resistances between a heat source and the ambient air add to yield the total thermal resistance between the CPU die and the ambient air. If you want real numbers, I can supply them. Suffice to say that the thermal resistance of the other components of the puzzle are much larger than the thermal resistance of the thermal grease. Off the top of my head, I think the typical P4 CPU is about
0.5C/W and the common heat sink (without the fan) about 2C/W. In between is the thermal goo, with optimistically about 0.05 C/W. Since they all add up together, the contribution of the 0.05 C/W is trivial. I could cut it to maybe 0.005 C/W by using diamond dust, and the total thermal resistance will have hardly changed.So, why all the fuss about the thermal grease? Because if there are any voids, air gaps, misalignment, warp, or gaps between the CPU and the heat sink, the thermal resistance goes from 0.05 C/W to something
10 or 100 times as large. Then it becomes really significant.Magic sauce. My guess is that it's more important for it to be environmentally correct, worker safe, and of course, cheap.
Dunno. It's tricky to measure. The solvent in the Arctic Silver goo does evaporate with time, which will certainly chance the thermal conductivity.
More insanity:
-- 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
It's not very flat and needs thermal goo on BOTH sides of the insulator.
I don't see why not. I commonly mount TO220 devices directly to a heat sink with only a little thermal goo in between. The thermal goo isn't going to act as a gap filler, but does help deal with uneven and rough surfaces.
-- 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
On Wednesday, December 3, 2014 12:00:21 AM UTC-5, Jeff Liebermann wrote: .
k so.
Probably referring to metal cased semis where the metal case is electricall y connected to the device and needs an electrical barrier (like mica and pl astic washers) to isolate it electrically. Years ago it was more common to find metal backed T0-220s or T0-3s that needed mica or silicon rubber. To day most semis are encapsulated and only need thermal goo to mount.
You used the mica washer to isolate the device from ground. You needed good heat connection to the heatsink, but didn't want electrical. The mica washer was to insulate, the thermal compound was to make sure things made good thermal contact.
Michael
I started fixing an amp somebody was trying to fix. Without insulators on the output transistors, it was never going to work.
Greg
This is fun,
How about some vegemite? :^)
George H.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.