"Matt" wrote in message
...
jeffc wrote:
"Conor" wrote in message
...

Then the heatsink wasn't applied right (at an angle, not just plopped
straight down so that it might trap air bubbles), or the heatsink was

not

clamped with much pressure. There should never be a "huge" gob of

compound

that is "quite thick".


Just got an old IBM desktop in. Its a P166MMX with fanless cooling.
Took the hetsink off and there's a big square of thermal compound.
You'll find all old P1 IBM desktops were like that....


Sounds like a "pad". Now a pad is *not* going to be squished as thin as
paste. Which just goes to show, even if the high pressure from today's
heat
sinks, and small "footprint" of the CPU core, weren't enough to make the
layer as thin as possible, it would still work with a (relatively) thick
pad.



Also the engineering of the HS mounting arrangement may assume that you
are using a pad, and that assumption affects the clamping pressure and
fit. You have to be careful when replacing a pad with paste.

It's also worth taking into account that those older CPUs were producing
much less heat flux (wattage) in general and per contact area... so the
thermal stress was not as large as more current CPUs. Back then, with simple
or no clamping mechanisms, a thermally conductive adhesive was used to bond
the HS to the CPU, being a compromise between an adhesive (with high bond
strength) and a thermal compound (with high heat conductivity). The
reuirements of today are much more demanding. I had a 486DX4-100 with a
malfunctioning fan that was spinning half of the time very slowly (bad
bearing i guess) and not at all the other half - for more than a year (in a
hot climate), no damage to the CPU. I could boot Pentium class CPU with just
stock HS, no fan, to POST, BIOS setting, and command prompt, with no
lockups. Would you do that with current CPUs ?