Spreading thermal compound or not?

[Texun]

Originally posted by: Shadrack

Originally posted by: Gamingphreek
I get a plastic zip lock bag, put my finger in it and spread it out evenly across the core. Then i get any excess off with the edge of a paper, leaving the surface extremely smooth. Then i put the HS/F unit on and enjoy.

Just plopping the HS/F on is a good way to trap air and not get an even spread across the CPU (unless your CPU is circular). Always spread it yourself.

-Kevin

has no clue what he is talking about....rofl

Then I don't know what I'm doing either. I use the "finger in the ziplock bag" trick and my temp right now is 30C, although I don't push it all the way out to the edge. It gets squeezed out by the HS.

 

[Gamingphreek]

Originally posted by: Texun

Originally posted by: Shadrack

Originally posted by: Gamingphreek
I get a plastic zip lock bag, put my finger in it and spread it out evenly across the core. Then i get any excess off with the edge of a paper, leaving the surface extremely smooth. Then i put the HS/F unit on and enjoy.

Just plopping the HS/F on is a good way to trap air and not get an even spread across the CPU (unless your CPU is circular). Always spread it yourself.

-Kevin

has no clue what he is talking about....rofl

Then I don't know what I'm doing either. I use the "finger in the ziplock bag" trick and my temp right now is 30C, although I don't push it all the way out to the edge. It gets squeezed out by the HS.

Just to give you a heads up...Shadrack was banned

 

[DasFox]

Ok so if everyone is saying a DOT in the middle is the best way, then how big is this dot suppose to be, I mean looks like we're saying use your best judgement on this.

And how would you really know that dot is going to spread out all over the entire thing covering every area of it.

I personally have been using AS5 spreading it with a plastic card for the past 6 years with no problems ever. This way I know the entire core is covered in a thin layer.

I have seen to many dots like those made for a HSF in a video card, north and south bridge chip on a motherboard leave so much of the surface not covered to make me trust this method, UNLESS you practice it over and over to see what it takes to make it work right.

Now granted the CPU core is alot smaller, but I want to make sure it's completely covered.

 

[mindless1]

Originally posted by: DasFox
Ok so if everyone is saying a DOT in the middle is the best way, then how big is this dot suppose to be, I mean looks like we're saying use your best judgement on this.

And how would you really know that dot is going to spread out all over the entire thing covering every area of it.

The key is not that it covers every square millimeter of it. The key is the thinnest junction between the die in the middle and the large mass of metal above it (the heatsink).

Heat radiates outward from (close enough to, for the purpose of this discussion) dead-center. Trying to promote maximum interface between the edges of a heat spreader and 'sink, ultimately results in excess compound being applied, because even though the CPU spreader and 'sink "look" very flat, they're not perfect. You ideally want the shortest, best conductive path to the heatsink fins so the air cools them. This path is directly upwards, so the best interface only allows the least possible thermal compound in the middle.

That doesn't necessarily mean you don't want the compound to spread towards the edges, but not at the cost of having too much in the middle or just as bad, air pockets do to aforementioned issue of the mating surfaces not being perfectly flat, parallel mating together.

I personally have been using AS5 spreading it with a plastic card for the past 6 years with no problems ever. This way I know the entire core is covered in a thin layer.

There are a lot of ways it can be applied and keep "Most" CPUs, video GPU, etc, cool enough. Even putting half a tube of AS on it, after enough time the heatsink retention force would squeeze most of it out, but ultimately it's not quite as thin in the middle unless you happened upon a (random and typically uncommon) heatsink that was convex in exact center over the CPU die. Carefully spreading is usually better than gross overapplication, but it doesn't account for unflat surfaces, especially if spread very thin.

I have seen to many dots like those made for a HSF in a video card, north and south bridge chip on a motherboard leave so much of the surface not covered to make me trust this method, UNLESS you practice it over and over to see what it takes to make it work right.

Sure, practice if it makes you comfortable, though a low-enough temp is evidence enough, one need not try for some ultimate record-breaking low temp as a couple degrees either way should never be a factor, system isn't set up right if it's that close to instability or thermal throttling, etc. Even so, the whole thread was about some ideal, so there it is... a target application method.

Now granted the CPU core is alot smaller, but I want to make sure it's completely covered.

Take some (pre-flipchip) GPUs for example, one does not want the compound to cover the whole thing if it has a metal heat-spreader in the middle. The outer edges of those chips, the black epoxy casing, was rasied higher than the center. Mad overclockers (I plead guilty) sometimes even took sandpaper to those chips and sanded down the epoxy just to have the center spreader level with the border to further decrease amount of compound necessary to fill any (then vastly reduced) voids. If the epoxy casing had thermal compound on it too (unsanded, as shipped) the 'sink would just sit that much further (thickness of compound on edges) away from the center of the chip, while the epoxy casing is a relatively terrible heat conductor.

Granted, it's not quite the same as on a flipchip with an entirely metal spreader, but even when temps look good, there can be differences in temp at different points on the core. For example, I was troubleshooting an Athlon XP2000 box, CPU temps weren't alarmingly high but there was instability. After removing heatsink I saw degraded silicone based compound that had dried up and left small islands between the CPU and 'sink.

Apparently the region with the thermal diode in it was still sheding heat being near an island but other areas weren't. After reappling a synthetic based compound, CPU was then stable at even higher temps than it had registered previously (how so, why higher temp? Recall that I'm a mad overclocker... had to see what it'd do. Returning to stock speed it was only 5C cooler, IOW, 49C before, 44C after @ stock, and stable at 55C o'c (loaded not idle temps)). Lesson here- you may have same temp report in two different heatsink interface scenarios but that doesn't necessarily mean ALL regions of the CPU are equally cool (enough) in both.

 

[oddyager]

Originally posted by: DasFox
Ok so if everyone is saying a DOT in the middle is the best way, then how big is this dot suppose to be, I mean looks like we're saying use your best judgement on this.

Depends on the surface area of the chip but in general, I put a dab about the size of the lead on a pencil head and let the weight of the heatsink and the heat to help evenly spread it out. You want to avoid a thick layer which is what everyone here is saying.

One other point I want to touch up on. There really isn't a need to spread the mixture to cover the entire chip itself. What you DO want to do though is spread it over the base of the heatsink and rub it down so that it fills every microscopic groove and valley. Again, you do not want to create a visible layer here. The surface of this base is not perfect, which is why some people like to sand it down for a flatter, smoother area. If you don't want to do that, you can use the compound instead, and as people suggested, can use the card, razor blade, or finger through a ziplock bag to apply it.

 

[TheBearingEdge]

This thread is kind of dumb.

You guys are arguing over points that have little or no ground outside of personal experience.

A good tip... get a bottle of pure isopropyl alcohol and TRY SOME DIFFERENT STRATEGIES!!!! The one that works best for you is the correct one... DUH. If you've only tried one way 'cause that's what someone said was the RIGHT way or you haven't tried improving your own approach, then you really have NO perspective.

A couple no-brainers:

-Thermal compounds have a break-in period.
-Finer surfaced heatsinks need a thinner layer than something less polished.
-If you wonder if your "BB-sized dot" is too much or not enough or not
spreading on its own then CHECK IT OUT. Take a freakin' look!

I tried the BB approach and it didn't work for ME. It just ended up being a smashed BB. Like 50% of my die surface got any compound and I WASN'T about to apply MORE pressure or end up with a destructively thick layer of compound.

A major factor is that my compound is very viscous. I had to adjust and be dynamic.

Trial and Error people, c'mon...

 

[mindless1]

Originally posted by: TheBearingEdge
This thread is kind of dumb.

Not at all, it's necessary because we have folks like yourself spreading nonsense that has to be dispelled.

You guys are arguing over points that have little or no ground outside of personal experience.

Your post is so incredibly backwards that I"m not even going to bother picking it apart. NO. This is not ego-centric, not subjective, it is an objective discussion of best thermal compound application.

If your "me" scenario finds that the compound is too viscous, then the being "dynamic" is to dynamically decide that the compound is not suited for the use. Same with most other situations... have an AT case but want to build an ATX system? Dynamically decide to get an ATX case. Have an AGP video card but need a PCI? Dynamically decide that this too is not subjective, you have the wrong part for the job and need pick what is appropriate instead.

There really isn't any "me" to it at all. Now having said that, much ado is made about lowest CPU temps, as if it's a contest. While there's no reason to WANT an inferior thermal junction, on the other hand if the 'sink is installed, CPU is staying well below the thermal threshold, there's no need to pull off the 'sink and check it, because at the very specific target application itself, the job was sufficient enough. No matter what "me" is thinking, the subjective interpretation of the situation matters not relative to the objective state of the CPU temp.

 

[TheBearingEdge]

my bad... I was up too late and was very grouchy.... text discussions are so easily misinterpreted.

I think I meant to post this reply in another thread!!! I'm a dork!!!

I'm not trying to infect anyone with rationalist-thinking. Clearly this is a scienctific and empirical realm. But I just read so many posts that use hugely broad terms or questions and I think wow... isn't each case determined by some underlying factor? Not to say that they are "invisible" or "indeterminable" factors; that's just SILLY and, upon reading my original post I sounded like a total boob.

Mindless1 I actually like your posts. They pretty much compensate for the presentation of the questions. I just wish you didn't have to remind everyone of the concepts found numerous other places online. Whatevs.

Peace, everyone.