DaClan Review: Thermal Interface Shootout

[DonE]

On a slightly related note, would you by any chance include some notes on thermal gloop cleaners in your guide when you get around to it? I've been wondering if the different types of gloop require different cleaning products, and which ones are reliable.

Thanks for the great work.

 

[dasickninja]

Always over 90% isopropyl alcohol, and lintless cloths.

 

[Talon]

As DaSickNinja said, alcohol is fine. Its certainly the cheapest. Some thicker "goop" as you say can take a bit of diligence to remove with alcohol though. I prefer elbow grease over more $ for a cleaner though.

If you really want to try a TIM cleaner then I might recommend Akasa TIM cleaner. Breaks stuff down fast and makes it even easier. Still higher priced than good old alcohol however.

 

[tool_462]

Lapping doesn't negate the need for thermal compound. I have a lapped Apogee block sitting on a lapped D805 and I use about a quarter grain of rice of Ceramique and it covers the whole surface easily. There are still minor imperfections that can't easily be seen that the TIM fills up.

 

[dasickninja]

There is a point when lapping that you have both the heatsink base and the IHS so smooth that using normal thermal compound will actually inhibit the transfer of heat. Then you have to use some really viscous stuff.

 

[tool_462]

Secret trick: Keep your ceramique sitting with the applicator end up. It will seperate ever so slightly and become a little thinner. It isn't a huge difference but it is noticeable when applying it. AS5 seperates more dramatically but the resulting liquid isn't very thermally conductive, so don't use that technique on AS5.

 

[flasher702]

Did you have any problems getting the Shin-Estu off? The first time I encountered that stuff I thought the techs at Monarch Computers were morons. I could not get it off without doing things I thought would damage teh processor. I could not scrap the excess Shin-Estu off the sides of the heatspreader. I thought they had used thermal epoxy by mistake. Turns out they have pretty knowlegable techs who very quickly put me in my place when I called them to bitch but I still remember using a flathead screwdriver against the heatspreader to try and push the CPU off of the HSF and it wouldn't budge effectively making both CPU and HSF useless as I couldn't put it back into the computer.

Why do the temps vary so widely durring the 24hour run? Did the ambient temp change? I wouldn't have expected such a large variation within just 24hours. Wouldn't expect the TIM to "set" and bring temps back down. Were these systems installed in a case or sitting on a bench?

 

[tool_462]

Usually a twist of the HSF breaks the bond and you can remove it easily.

Temps vary (at equal times if you didn't notice) because certain portions of the program used to put load on the CPU end up taxing the CPU more than others.

It's discussed fairly thoroughly in the thread so read back through it once.

 

[flasher702]

Usually a twist of the HSF breaks the bond and you can remove it easily.

A heat gun was also suggested, but you're not supposed to put torque between the PCB and the Heatspreader and the Shin-Estu is the only one I've ever experienced that problem with. It was all still under warrenty so I just sent it back since I didn't feel like being responsible for it if it broke.

Temps vary (at equal times if you didn't notice) because certain portions of the program used to put load on the CPU end up taxing the CPU more than others.

It's discussed fairly thoroughly in the thread so read back through it once.

At eight pages I'm assuming there's a lot of off-topicness and other banality. I'll pass. Ty though

 

[flasher702]

There is a point when lapping that you have both the heatsink base and the IHS so smooth that using normal thermal compound will actually inhibit the transfer of heat. Then you have to use some really viscous stuff.

It might be prudent to point out that such a high degree of flatness could not be obtained using clandestine methods. You should really only lap if the base of the HSF is noticibly grooved and, as you say, a polished base can lead to all of your TIM leaking out which is bad.

Personally, I use the chewing gum stuff I got tired of re-applying grease and wanted something that would stay properly mated when frequently moving the PC for LANing. I'm sure the Shin-Estu would work quite well for that as well, but the chewing gum stuff actually comes off after it's been on there for a year xD

 

[tool_462]

Chewing gum stuff?

 

[HotFoot]

I thought I should post an experience of my own. I'm using AS5 and (unfortunately) the Intel stock cooler on a PD820. This is a temporary solution, as I will be getting both a cooler CPU and better cooler eventually, but for now, I do with what I can.

Earlier this week I applied the AS5 using the grain of rice sized amount of grease and then squashed it with the HSF. My temps were higher than I was expecting, but then I'm coming down from a nice tower-style cooler with heatpipes. Anyway, I began to wonder about my temps, which easily exceeded 65C at stock settings. I checked that the infamous Intel mounting mechansim was set properly, and it was. I removed the HSF and found that I hadn't gotten a neat circle of grease from the pressing action. It was more over to one side than the other, but it still covered the centre of the heat spreader, which is most important. Anyway, I went to the AS online instructions and used the thin-line method, again spreading the grease only by pressing down with the HSF. My load temperatures are now 5C better than before. I'm much happier now, and I'll manage to wait until a better cooler comes my way.

I'll see how much the temps improve over the break-in period. AS claims the temps may drop 2-5C. Another 5C over the next week or two would be great. I could increase the clock a little more.

 

[deathbybubba]

I read through all 8 pages and had a couple of questions on lapping.

First off, what is lapping and how is it achieved?

Second is it worth the time and effort to lap both surfaces?

Thanks
BUBBA

 

[flasher702]

Chewing gum stuff?

It looks like chewing gum and has about the same consistancy. It's highly stable. Comes pre-applied on a lot of low-end HSFs. AMD had a brief marketing phase where they were recommending it because it didn't pump out leaving gaps instead of TIM. I think it came with my s939 x2 4200+ (I don't really remember). Takes a good bit if heat and time to melt the stuff down and get good contact though.

Having replaced HSFs on a number of older computers when I worked as a bench tech (replacing the entire HSF was often cheaper and always easier then replacing just the fan) I've seen a lot of examples of how thermal grease doesn't age well.

 

[tool_462]

Yeah, I see what you are talking about. AS5 ages really well it seems, though I usually end up taking my heat sinks off every few months. But I have put it on Intel stock HSFs and seen it age well for a year or more.

@ Bubba: Here is a decent lapping guide.

Unless you are really into overclocking you won't want to spend the time and (obviously) void your warranty on your HSF and CPU if you choose to lap them.

I have my IHS on my D805 lapped and my Apogee is lapped. I originally just lapped the block and it dropped about 2C under load but after I lapped my IHS (which was really concave) temps dropped a total of 5-6C under load. Pretty good if you ask me

 

[clue69less]

First off, what is lapping and how is it achieved?

Second is it worth the time and effort to lap both surfaces?

The Wiki blurb on lapping is reasonably accurate in its historical treatment but it's not current about materials and methodology. WRT lapping HSFs and CPU IHSs, just realize that any lack of flatness or surface scratches can result in poor thermal contact. So if you grind and polish the two surfaces perfectly, you wouldn't even need TIM. It's been done (avoiding TIM that is). So if you have one surface that is significantly rougher or less flat than the other, lapping can make a difference. Lapping both makes more sense and if you do, a TIM with small solids is a good idea. It's fun to do a surface roughness measurement of an expert's hand lap - a mirror finish can be far from flat and scratch free.

 

[clue69less]

I have my IHS on my D805 lapped and my Apogee is lapped. I originally just lapped the block and it dropped about 2C under load but after I lapped my IHS (which was really concave) temps dropped a total of 5-6C under load. Pretty good if you ask me

Yea, that's an excellent improvement. Enough to make a performance difference.

 

[tool_462]

Gave me an extra voltage level to keep it under 35C load. Not that I can't go higher in temps but at 4.4Ghz I was loading in the 44C range and it kept me under 40C after lapping. At 3.9-4.0 (where I keep my CPU for day to day use) I hit 33ish and was 38 load before. Not a bad improvement.

 

[deathbybubba]

When you lapped your IHS, what did you use for weight? HSF's can be heavy, but proc's weigh nothing.

 

[tool_462]

I just used my finger with very very light pressure in the center of the CPU with the wet sandpaper on a flat pane of glass. You dont want to push hard and you dont need to. It just takes a little time is all.

 

[Storm1234]

Just now read this thread today, so my question is.

So if I have a PD805 Zalman 9500 HSF, the Shin-Etsu X23 is the best to get for it, including over the Shin-Etsu 751 stuff.


Thanks alot for help. Storm

 

[Storm1234]

Anybody, is this correct..


Very much appreciated. Storm~

 

[dasickninja]

Sorry for the late response. Haven't gotten to this thread in a bit. With a Pentium D, you'll need every degree that you can get, so th Shin-etsu X23 is the best bet especially when combined with a good lap.

 

[Drez143]

Did newegg just get it in? Cause it only has 2 reviews. Shin-Estu Paste

 

[deathbybubba]

That's a different paste then the one Ninja reviewed. I believe the one he reviewed is the x23

x23 is in stock here.