CPU Thermal paste for old machines

[slomacuser]

What paste do you use? I have 7100, 6100, G3 and some other Mac's that needs new thermal paste. How many grams for 2-4 CPU's. Method of applying: dot, line, cross, ...

I have found this paste Akasa Termalna pasta akasa AK-455 - 1,5G -

just for 1,50 EUR in discount offer, what do you think would it be good for old Macs?

I am new to this so any tip welcome, thanks!

 

[Cory5412]

I would probably go get regular modern thermal paste, same stuff you'd use when building your PC, the main reason for that being that there's no good reason not to.

I hop on over to my local generic electronics retailer (in this case, Best Buy) and have whatever they have for cheap.

I don't think the amount you apply matters as long as it looks like enough to cover the area that needs to contact. For a 601, that's probably about the size of a small grain of white rice. For something like an '040 with a larger contact area, that could be a little bit more paste.

These older CPUs don't really have the idea of throttling, nor do they have a lot in the way of controls. Either they are cooled well enough that they run and do not suffer thermal faults and overheat, or they are not and the machine glitches or turns off.

So, my advice would probably be to wait for bad performance in one of these ways, then reapply, then if it doesn't do the bad thing anymore, it worked.

 

[carguyty]

When you're staring at the business end of a 20 year old machine, I reccon anything will be better than the dried out and brittle gunk hiding under any paltry heatsink it may have (I'm looking at you 630CD  > ). I get plenty of G4 'Books my way and every one of them gets new paste if I have to get in the case for any reason. Arctic Silver Ceramique is pretty stellar for everything ESPECIALLY when there's no fan involved. 

Just a dot, a little smudge on the CPU cap, is all it needs. Replace the heatsink straight down and don't wiggle it none. Tighten the screws or straps and don't look back. 

I don't have any experience with the brand you mention but mine is "high quality" and only about 2 or 3 EUR, so I'm sure you got a deal. Paying for quality (however that is measured) is a good idea. Some require refrigeration, some are in single use packets. Read all the manufacturer instructions and you'll be alright. Remember that the most of these are passive cooled and some don't have heatsinks at all, so whatever you do will be an improvement over stock.

 

[VMSZealot]

In my experience, it doesn't matter if the paste is dried out and brittle.  You aren't using it for its lubricating properties - and the only reason that it's fluid when you first apply it is so that it can flow properly under the heatsink when you cinch it down.  Once tight, it can dry out all it likes and be as brittle as anything, it isn't going anywhere and its thermal properties will be unaffected.

I think that (unlike capacitors) this falls into the old adage, if it ain't broke…

 

[Cory5412]

On modern computers (and of course by modern I mean anything with SpeedStep, so Pentium Ms and Pentium 4s) dried out and brittle thermal paste actually does have a material impact on computer performance. Basically, at least in the dry southwest, at a certain point in its life, a computer will start to overheat, and will continue until you take off the heatsink and replace that material.

It loses its ability to transfer heat. EDIT: Which is the entire reason you had a heat-sink to begin with. To transfer heat away from the component you put it on.

This matters a lot less when you're, on an overclocked '040, outputting 9 watts of heat, than when you're putting out today's wattages.

 

[trag]

I don't think the amount you apply matters as long as it looks like enough to cover the area that needs to contact.

In my experience, it doesn't matter if the paste is dried out and brittle.

[NITPICK]The amount does matter, if you use too much.[/NITPICK]  And too much is enough to squish out from under the heat sink, run over the side of the chip and get on the pins.   I killed a Power Computing Power 120 that way, before I knew better.    As you wrote later, a little dab.   You're just filling in the imperfections between two nominally flat pieces of metal.   So it doesn't take very much.

And it does matter if it is dried out and brittle, because in most cases, the "fluid" was part of the heat conduction matrix.  The grainy, dried residue does not form a good continuous heat conduction surface.   I've seen a lot of 7100s whose only problem was dried out crusty heat sink grease.

 

[Unknown_K]

I just buy a cheap tube on ebay (from China) and make sure I spread it around with a Q-tip before putting the heatsink on. You just need enough to fill in air pockets between the two metals. From what I read the different brands are not that much different in heat transfer when new, but some turn to powder after the liquid/solvent evaporates and others don't.

 

[TheWhiteFalcon]

Paste condition definitely does matter. As I've been preparing them for retirement I've been redoing the paste on some E5520s with i5-2520M chips - CPU temp drops an average of 12C with new paste. 

And as has been said, don't use too much. You can insulate the chip that way and do more harm than good. 

 

[trag]

At least some of the greases are electrically conductive, and if you get them on the pins, they will cause shorts.  That's why using too much killed my Power 120.  And I had a NewerTech NuBus PDS G3 upgrade whose cache did not work.   Took a look under the heat sink and the assembler had dribbled a thin line of heat sink grease across the cache chip pins.  Got a warranty replacement, but clearly, the heat sink grease was shorting the pins on the cache chips.

 

[slomacuser]

I have orderd that paste and will try to apply it today, thanks for tips

 

[rsolberg]

I've been using Artic Silver Ceramique for a few years with good results. I follow the directions on Artic Silver's web site- they suggest different application methods for different model CPUs based on the heat spreader. (or lack thereof). I usually "tint" both surfaces as suggested to help ensure uniform mating surfaces before applying the little blob or line. A tube goes a very, very long way when applied as directed.