Does the heatsink have to be pressed against the CPU to work?

hey guys, im screwing with myold x86 stuff. my dual mobo has both center tabs broken off of the socket. Now IMO its not worth it to go out and buy a new pair of HS when the rig mite not work at all

so i was wondering, does the HS have to be pressing upon the CPU really hard? or can i place it on top?

otherwise ill just bust out the electrical ties and fasten the HS down thru the holes lol

thanks in advance

Well, you wants some pressure on there so its making good contact.

I suppose if you didn't move it aroudn at all, it might do the job just sitting on there. Probably kinda of risky though.

How old is old? As long as it's making contact you'll be fine. Pressure would obviously help, but old equipment didn't run that hot back then, so I doubt you'd see a difference.

You should pressure it, just leaving it on won't be enough and you might risk burning the cpu. It just takes a few seconds until it burns so just make sure you put pressure on it before you power on .

I had a simmilar problem, when I first installed my 3.2 P4 I didnt notice that while taking it back up to my room from the basement one of the cornners of the pins on the HS came out... When I would start my computer, it would crash about 25 seconds later at over 200F!!! All just because one pin was off, and it was making about 75% contact! If there is any way to watch the temps, see how high it gets and if it goes past your liking look elsewhere -

If you dont have complete, and somewhat pressurized, then you wont have the proper cooling. If you've ever replaced an old hs with, lets say a Thermailright, you'll notice how much pressure it takes to get one on. There's a fair amount of pressure to get one of those puppies on. Get some of those ties to be on the safe side.

MadSkillzMan said:

hey guys, im screwing with myold x86 stuff. my dual mobo has both center tabs broken off of the socket. Now IMO its not worth it to go out and buy a new pair of HS when the rig mite not work at all

so i was wondering, does the HS have to be pressing upon the CPU really hard? or can i place it on top?

otherwise ill just bust out the electrical ties and fasten the HS down thru the holes lol

thanks in advance

Click to expand...

if its not worth going and getting a new HSF then you can just leave it sitting on top, the weight of the HSF might be enough pressure to do a good enough job.

How old IS it?

How much do you plan on running it?

How much is it worth to you?

Truly OLD cpus often did not HAVE heatsinks.

As for a cpu that DOES require a sink- well, many of us test cpus by merely placing a heatsink on top of the chip without fastening it down, and we have done it for years. But this is no way to run a system long-term: just the vibration from the fan will move it from the cpu core eventually.

Best performance from a heatsink or water block does require some amount of pressure, and some of us regularly exceed AMD and Intels specs and use far too much in order to get better temps (this DOES work, although not a huge gain.) Without giving the sink at least close to the specced pressure you will not get great cooling; it may not even be enough to run reliably.

MadSkillzMan said:

so i was wondering, does the HS have to be pressing upon the CPU really hard?

Click to expand...

The ideal situation would be that the surface of your IHS (the metal cap on your CPU) and the surface of your heatsink would be like .000001 flat. If you were able to squish such flat surfaces together and eliminate all the air, not only would you have a highly conductive thermal interface, but you would not be able to pull them apart by hand because there would be such a high surface adhesion. So what you are going for is a combination of thermal interface material (AS5 or whatever) to fill the microminature gaps, and pressure to force out all of the air. The issue really is whether you can apply the pressure evenly across the chip so as not to mash one or another corner and crack the chip.

There is no way that you can let the HSF just lay on top of the CPU with only its own weight holding it down and get good thermal transfer. You need to have springs or weights or levers or whatever mashing that heatsink down on the CPU. And you need high quality thermal interface material!

Didn't some of the old x86's run without a heatsink anyway? I have an old Pentium 133 with an epoxy'd HSF that had the same volume of Aluminum as 2 BGA ramsinks and fan that probably had like .5cfm on it. I highly doubt that thing would burn up in SECONDS. It would probably freeze/lockup but I don't believe they had the power to actually burn up.

Run it bear with no HSF but air blowing on, I bet it would work... and if it doesn't... please do not blame me for the 2 dollars you have lost

Bryan D.

I have never run a system even for a second with the heatsink just set on top of the CPU. I guess it was just the thought that it is safer to have the pressure for the proper thermal transfer. I think I would personally tie it down somehow. But the problem with that method is the possibility of not being able to get the proper contact.

it depends on the cpu. HP sold comps with K6 processors with just a fan blowing on the cpu, no heatsink, just a fan. but a pentium 1 200 requires some kind of HS. alot of older 486s ran bare just fine. but a P2 or anything newer need a decent hs. stick a rubber band or 6 around the clips that are still there, the outside ones. it should hold it decently, just be sure its not stopping the fan from spinning.

Here is a thread where we will find out who the oldies are.

What I would do, (and did not too long ago) is just set the biggest block of copper you have on top of the cpu. I used a copper shim with a tin can full of water sitting on top of it. The can didn't warm up, the cpu was fine. After shut down, I touched the cpu and it was only slightly warm. Now if you feel lucky, set a can of icewater on the copper block.

The question I have now is how are you going to overclock it? If I remember right, you will have to use jumpers and even at that you will be basically tricking the mobo to run as if a different cpu was in the slot. Also, if you could find one, (doubtfull) you could use one of those old clip-on frequency generators that overrides the signal generator crystal on the mobo. I'm not sure how they worked, never did it myself.

I have a machine that used to be my router. I'm not sure exactly what the processor is, but I believe it is 133-200 MHz. It has a little HSF on it that was just stuck on with thermal goo. One time I took the side panel off to clean the system out, and the heatsink had slid halfway down the processor. The machine was still running fine though.

I redid the setup with new thermal paste, and some thin gauge metal wire wrapped though the heatsink, and under the hooks on the CPU socket. Works great. The wire just keeps the heatsink from sliding down the processor again.

So if your processor is old enough, I don't think the amount of pressure against the CPU will really matter. As long as the heatsink makes contact, it will work.

hah crap i didnt mean THAT old, im talking SOCKET A 1.2ghz.

however i do remember those old 486's without HS's.

Anyways, atm im testing a different mobo, guy next door is gonna scrounge his scrap pile for a pair of HS clips

Those old 486's ran hot. we took one apart once and the CMOS battery had exploded and was all melted and gooey. It had like 3 cpus? 1 didnt have an HS, one did, the other was all melty.

some of the P1s i garbage pick have the tiniest HS's on them, something that would barely cool our southbridges.

IMO, if intel/AMD are making CPUs which require such extreme cooling, they botched something up. Notice, the NEWER 1.2s and 1.4s dont run as hot as earlier ones. Or my fav analogy, the 300mhz apple G3 runs WAY hotter than the 500mhz G3 upgrade, which can run without a fan. So its like as time progressses, the CPUs get cooler. only no one wants to spend time on cooling and just make a wicked fast CPU and brag theirs is sooo hot it requires a giant heatsink/water/phase ...like that alienware system. or apples dual g5, oh boy its watercooled.

dont get me wrong im all for the extreme cooling, fun bragging rights...just my .02

MadSkillzMan said:

IMO, if intel/AMD are making CPUs which require such extreme cooling, they botched something up. Notice, the NEWER 1.2s and 1.4s dont run as hot as earlier ones. Or my fav analogy, the 300mhz apple G3 runs WAY hotter than the 500mhz G3 upgrade, which can run without a fan. So its like as time progressses, the CPUs get cooler. only no one wants to spend time on cooling and just make a wicked fast CPU and brag theirs is sooo hot it requires a giant heatsink/water/phase ...like that alienware system. or apples dual g5, oh boy its watercooled.

Click to expand...

My guess as CPUs get faster (and hotter), the little electronics going on down there are getting smaller and smaller. So when they go back and make new, slower chips, they small electronics (someone help me out here) are more spaced out, since less are needed for a smaller job. If they are more spaced out, more breathing room.

I had an old Athlon 400 that had broken socket tabs I was turning into an e-mail box for the mother-in-law.

I scrounged for an HSF with a 2" x 2" base on it (so the socket lock lever would be exposed), layed on a thin even coat of thermal goo, cleaned the corners of the cpu off with q-tips, and mixed up some jbweld to dab on those clean corners.
I set the whole assembly in a clamp with a small block to avoid bending cpu pins overnight. Checking and double checking that the hsf would line up both over the socket cam box and leaving the socket lock lever exposed (so I could install the chip...duh).
So to make it short (too late), I had a permanent hsf on that chip. It never moved, and was in a normal vertical tower case.

There's also Arctic Silver/Alumina adhesive that you can use too. My 400 had a heatspreader, so jbweld worked fine in the unused corners. The thermal adhesives would work best for a nekid core cpu...just apply it thin to well cleaned parts.
Personally, I'd watercool 1.2's using the bolt holes...just 'cause I can

Things like 1.2 Athlons like to burn up if you play games with the heatsink. I'd recommend doing it right, or not doing it.

yea im probably gonna just takes larvas advice and do it right