VOGONS


First post, by ultra_code

User metadata
Rank Oldbie
Rank
Oldbie

(Note: I have come to the conclusion that this question is actually kind of stupid, and upon a quick Google search (which I was apparently too lazy to do before 😀 ) I found a zillion answer posts to my question; check out my second post on this topic to find great examples of what I am talking about. However, some of the experiences shared here may be interesting or useful.)

Hello VOGONS members!

I was wondering: What is the best TIM to use with a SECC2 heatsink for a Pentium III (both in terms of effectiveness in heat dissipation and heatsink/CPU contact), just your usual thermal paste or thermal pads?

Backstory

You see, for my first retro PC build (which you can find here), I bought a new, larger passive Foxconn SECC2 heatsink (that uses that usual sliding latch mechanism for attaching it to the CPU) to replace the original heatsink on my Pentium III 600E CPU:

I ended getting the original heatsink that came with my CPU off by following goninanbl00d's suggestion found on this topic:

ZTBlQxkl.jpg

However, look at this photo of the new heatsink (https://i.imgur.com/Y3lNkD8.jpg). Do you see that little "extrusion" sticking out at the top of the side of the heatsink (the part of the heatsink that slides down the sides of the retention clips on the motherboard)? That prevented me from slotting the processor into the slot, as the motherboard's CPU retention clips do not have the recesses on the tips of them that this heatsink was obviously designed for. What a waste of money!

Anywho, wanting to put some heatsink on my processor, I ended up putting some Arctic MX-4 thermal paste onto the CPU and reattaching the original copper-colored heatsink pictured above with some replacement retention clips.

Main Issue

After reapplying the original heatsink with that new thermal paste, I ran 3DMark99 with HWMonitor running in the background recording CPU temps to a .csv file to compare the cooling performance of the thermal paste vs. the original thermal paste/metallic thermal pad (whatever it was) (I conducted the same test before removing the original heatsink). The results surprised me, though.

It seems as though my CPU got hotter (by a few degrees Celcius) and got hotter faster than with the original TIM. It could be that the original TIM was better than the Arctic MX-4, or that I used too much thermal paste, which, according to Linus from LinusTechTips, could cause an insulating-effect. It could also be that the heatsink is not 100% as tight as it could be against the CPU, as I do not own the OEM clamp/press used to press the clips tightly down along with the heatsink against the CPU to attach the whole thing, and so ended up using my hands to press the clips down until they stopped popping back up (yikes!). So, finally, back to my original question.

The Question to Begin with

I just bought one of these SECC2 Startech heatsinks to replace the original passive heatsink I am still using (mainly because of its easy-to-service sliding retention clip mechanism) from eBay, and this time, I want to apply the right TIM the first time. So, again, what TIM should I use?

I noticed this SECC2 Cooler Master heatsink comes with a pre-applied thermal pad. Was it considered back in the day a good idea to use thermal pads with slot 1 CPU heatsinks vs. thermal paste?

Sorry for the potentially pointlessly long post, but hopefully this provides the complete context for my question. Thanks!

Last edited by ultra_code on 2018-11-05, 17:26. Edited 2 times in total.

Builds
ttgwnt-6.png
kcxlg9-6.png

Reply 1 of 4, by cyclone3d

User metadata
Rank l33t++
Rank
l33t++

MX-4 should be fine. You probably don't have it seated on there properly. It probably is not sitting flat.

With a CPU like that where the small die is directly contacting the cooler, any extra TIM would be squished out.

The other thing with those old heatsinks is that the surface is usually horribly machined and not flat at all which hurts heat transfer a lot.

Same goes for CPUs with heatspreaders unless there is barely any clamping force. If there is a bit too much to begin with, once the CPU and heatsink heat up, it should thin out the TIM and let it squish out so there is a thin layer. Now if you stick a massively thick layer on there it will end up making a huge mess but will still thin out after a few heat/cooling cycles.

Linus is generally ok-ish for general tips, but either doesn't know stuff well enough or is just really bad at explaining stuff.... I've watched a few of his videos and was never impressed one bit.
That is the general consensus on the other forums I frequent.

Yamaha modified setupds and drivers
Yamaha XG repository
YMF7x4 Guide
Aopen AW744L II SB-LINK

Reply 3 of 4, by ultra_code

User metadata
Rank Oldbie
Rank
Oldbie

Thanks for your responses!

Some updates:

First of all, I now realize this question was pretty dumb of me to ask because there are a zillion answers to the same question visible after doing a simple Google search (examples of such answers include http://www.pcgamer.com/whats-the-difference-b … rease-and-pads/ , https://linustechtips.com/main/topic/805664-t … -thermal-paste/ , and https://www.youtube.com/watch?v=9mgv6V5n5BA). I think I was a little too lazy to Google search 😀 .

Second, to quote myself:

the_ultra_code wrote:

It could also be that the heatsink is not 100% as tight as it could be against the CPU, as I do not own the OEM clamp/press used to press the clips tightly down along with the heatsink against the CPU to attach the whole thing, and so ended up using my hands to press the clips down until they stopped popping back up (yikes!).

It turns out that the "replacement clips" I bought (which you can find here) were meant for certain Dell PC heatsinks (which they say in their "compatibility" tab) (if I remember correctly, my processor and its heatsink came from an HP PC), and there's a reason why. Take a look at the following retention clips:
vDrGXOch.jpg
The one on the left is one of the "replacement" clips I bought and the one on the right is one of the original. Looking at the original, its legs are a bit longer than the replacement's ones.

To make sure that is was just because the difference in leg length that prevented the legs from fully "locking into" place, I threw one of the thermal pads I bought for my Voodoo3 3000 onto the CPU (for just this testing) and attached the heatsink with the original clips. To my delight, the clips were able to completely click into place and "retain" the heatsink without the need of that OEM clamp-thingy, and to my disappointment, because now I have two clips I cannot use. What another waste of money!

cyclone3d wrote:

If there is a bit too much to begin with, once the CPU and heatsink heat up, it should thin out the TIM and let it squish out so there is a thin layer. Now if you stick a massively thick layer on there it will end up making a huge mess but will still thin out after a few heat/cooling cycles.

b8koDJil.jpg
If the lines of thermal paste trailing from the CPU die that I created when I removed the heatsink are an indication of how I used too much thermal paste, I made a "huge mess". Lesson learned. I will now apply less thermal paste and use a spreader tool I have to get a better layer of paste for between the CPU and that StarTech heatsink I will be using.

Finally, with the help of someone who PM-ed me, I found out I could find CPU retention "brackets" for my motherboard to support that large Foxconn heatsink.
4KOjYvOl.jpg
I decided to buy this Foxconn CPU bracket off of eBay (which you can find here) to complement the heatsink, even if I do not plan to use the heatsink anytime soon because of my new, "easier" plan of using that StarTech heatsink. (On a side note, if anyone wants both the heatsink and the bracket to match it, just PM me and we could work out a sale 😀 ).

Last edited by ultra_code on 2018-11-05, 17:31. Edited 1 time in total.

Builds
ttgwnt-6.png
kcxlg9-6.png