Smooth or Rough heatsink?

[paparazi]

I have my own views on this!! once I thought a smooth pollished base was the best for heat transfer and now I'm not so sure.
I read a highly technical and scientific artical and lab test that conclusively established that a certain amount of machine roughness helped in the bedding in of the thermal paste and surfaces to achieve the best heat transfer.
I have also noted that several heatsinks I have looked at although in certain areas are well pollished the contact patch is relatively rough.
You are not telling me these multi million Dollar companies would produce a heatsink that has a rough base unless they themselves wanted it to be.
Your views greatly appreciated. Martin

 

[Gloryhound]

Well roughness can make the heat conductive adhesive hold better and thus transfer heat better, but the real thing to keep in mind for heat sinks is the material they are made of and their design. The material is probably the most important. Caste Iron wouldn't be good abviously, but a highly conductive material like copper would be excellent, but then how do you keep the copper cool sence it melts at such low temps and holds it's heat. Ideally a material that absorbs and radiates heat at approximately the same rate is the most desirable. Undoubtedly the heat sinks could be much better designed. Only a tiny fraction of money or time is spent on the design and manufacturing of heat sinks I would think. Reguardless, most of the heatsinks available today are sufficient. The biggest problem I've notice is the fans, there has to be a better way they wear out quickly under adverse conditions and can be noisey. Perhaps a solution is a better heat sink that requires no fan? I've yet to see such a beast.

 

[Guest_imported]

Like Martin, I would have to query why a rough surface was supposedly better than a smooth one...

I've always believed that if it was possible to get the two surfaces so flat and smooth that there wasn't a micron out anywhere, then heatsink compound theoretically wouldn't be necessary. In practice, of course, engineering to these limits would be expensive and largely unnecessary, so we use the compound to "take up" any irregularities in the surfaces to be joined.

As regards the longevity of CPU and other fans, I think this is largely down to price. If we're prepared to pay for a more expensive product, then there is no reason why it shouldn't last a very long time. For example, the bearings for the main shaft in a hard disk drive rotate at a higher rpm than most computer cooling fans, and many of these drives will run continuously for years without problems. It's not usually the main bearings that fail in a hard disk. Good quality roller bearings and dust exclusion are, I believe, the key to long life for the cooling fan...


ROGER - GØAOZ.

 

[edfair]

Up to Pent 200 there were heatsinks with no fan required. The faster chips run , the more heat they generate. So fans became more neccessary. You can get ball bearing fans, but the prices are into the mid to high teens instead of $5.00 range.
Personal opinion on the surface finish. The smmother the finish, the less chance of voids, and the less you have to use to get 100% coverage. Less is better since it isn't as good a heat conductor as the heatsink. Ed Fair
efair@atlnet.com

Any advice I give is my best judgement based on my interpretation of the facts you supply.

Help increase my knowledge by providing some feedback, good or bad, on any advice I have given.