How much of a temp diff does thermal paste make?


Recommended Posts

How much of a difference in temperature would there be between a cpu with an ideal amount/spread of thermal paste, and not enough thermal paste?  

 

I ask because every time I vacuum out my computer, I usually get an 8-10*c drop in cpu idle temp (when dusty, my AMD Athlon II X4 620 w/stock cooler idles at 47-49*c, when clean it idles at 39*C, but I've got it to drop as low as 34*C).  

This time, I also decided to remove and reapply the thermal paste on the cpu.  I'm using a no-name brand thermal paste (7.5W/m-k, 0.06 C-in2/W, 10% silver), I completely removed it from the cpu using 100% pure acetone (did my best to remove it from the heatsink, but after 10 minutes of scrubbing, the paper towel was still coming up with dirt, so I gave up), and then applied a half-grain-of-rice amount on the cpu and spread it out with a latex glove.  I remember hearing that the ideal amount is enough so that you can just barely see the writing/logo on the cpu, but not completely cover it up, and not completely visible.  That is the exact amount my cpu now has on it.

 

But after doing that, and vacuuming, I only got a temp drop of 2*C, much less than my normal 10*C.  Did I apply the thermal paste poorly, resulting in a cpu temp INCREASE of ~7*C?  Or did I apply it just fine, and there was just less dust in there than I thought there was?

Link to comment
Share on other sites

probably less dust. thermal paste as long as it's there and not to much won't affect much. if your CPU and cooler was polished perfectly flat, you would optimally have no thermal paste(they used to d this a few years back, but today the thermal paste is better and the risks in machine polishing is to high to bother) 

Link to comment
Share on other sites

Half a grain of rice seems like a small amount. It also seems strange that you had so much trouble removing the old thermal paste. I don't know about acetone, but I use 99% isopropyl alcohol and it comes off very easily, no scrubbing required at all. It also could be that the brand of thermal paste you used is simply inferior to the old one.

 

Unfortunately there's no real way to tell if the thermal paste is applied properly other than removing the heatsink and having to do it again. I hate thermal paste. 

 

In any case, your CPU temperatures are nothing to worry about.

Link to comment
Share on other sites

Most likely you didn't use enough.  You also may have created a problem by spreading it out by hand.  Use a little more and use the natural pressure of the heat sink screws tightening to spread the TIM.  TIM facilitates the heat transfer much better than bare metal.  It is very possible to run a CPU w/o TIM, but the real question is would it be worth trying.  A good paste with good application will greatly reduce temps and keep todays Ivys and Haswells from overheating.

 

http://www.hardwaresecrets.com/article/Thermal-Compound-Roundup-January-2012/1468/5

Link to comment
Share on other sites

I have cats, and so a lot of hair, and with a decent cooler it still used to get so clogged that my PC would eventually overheat if I taxed it too much. This would happen about every 6 months, at which point I'd have to vacuum it out like you did.

 

If your system, especially your heat sink fan, gets clogged with a lot of dust it can make your heat sink very hot because the dust and hair holds the heat in.

 

In the new system I built I went with a very good Cooler Master and, for the first time, case fans with dust covers (just some foam attached). The system remains very clean inside after 9 months.

Link to comment
Share on other sites

I can almost never not use enough thermalpaste, his amount s fine.

 

remember you're not supposed to have a "layer" between the cpu and cooler. they're both already nearly perfectly flat. the paste is there only to fill in microscopic  scratches in the surface, since otherwise they would have air in them. and Air is a great thermal insulator, meaning it would prevent heat transfer. the best is metal to metal, but since that's hard to achieve thermal paste is the next best thing to fill out those microscopic voids.

Link to comment
Share on other sites

Never spread thermal paste on a CPU. It's not necessary. Take a look at this video:

 

 

You might also want to look into dust filters for the intake fans, as well as improving your case's airflow. That much dust shouldn't be getting in there at all, and even then, it shouldn't cause that much of a temperature difference.

  • Like 2
Link to comment
Share on other sites

I can almost never not use enough thermalpaste

 

.....what?

 

lol, thanks for the help everyone!  It is a very hot summer here in Toronto this year (~35*c every day so far), so that must be why I didn't get the temp drop I expected.  But as for what HawkMan said, that is what I've heard before; use as little paste as possible because you only need to fill in the micropores.  Thermal paste doesn't conduct heat as well as the aluminum heat sink, so the more paste you use, the less heat you transfer to the heatsink.

Link to comment
Share on other sites

Never spread thermal paste on a CPU. It's not necessary. Take a look at this video:

 

 

You might also want to look into dust filters for the intake fans, as well as improving your case's airflow. That much dust shouldn't be getting in there at all, and even then, it shouldn't cause that much of a temperature difference.

 

...everything about that video seems horribly horribly wrong.  First of all, that is NOT how I spread mine out.  I didn't leave lots of ridges and lumps that would create air bubbles; I spread it perfectly smooth, and I used about 1/10th the amount of paste he used.  Second, when he finally gets to his idea of the 'correct' method (the "pea"), it doesn't reach the corners, which is where the cpu cores actually are.  It covers maybe 50% of each cpu core.  

Obviously stats don't lie, and if the pea method results in a lower temp, I'll stick with it.  But when I think about it, everything he did just seems like it should result in temperature INcreases.

 

I used to have dust filters, but my case is so old that the dust filters were no longer working.  They had accumulated so much dust that it was starting to turn into crude oil (literally, the dust was turning into black sticky goo), so I had to get rid of them.  And yes, the dust gets right inbetween the cpu heatsink blades.  Don't get me wrong, its not a ridiculous amount of dust, it's just a little, but it's like forcing the heatsink to wear a tshirt in the summer instead of letting him go nude.

Link to comment
Share on other sites

Never spread thermal paste on a CPU. It's not necessary. Take a look at this video:

 

 

You might also want to look into dust filters for the intake fans, as well as improving your case's airflow. That much dust shouldn't be getting in there at all, and even then, it shouldn't cause that much of a temperature difference.

 

The pea/rice grain method is what they recommend because it's easier to apply. but spreading it out is in fact the better option. what he claims is air bubbles is in fact areas with metal to metal. provided you're not a total amateur when spreading it out. 

 

notice how the pea method gives you a thick layer of paste, notice it especially after removing the paste = yeah, that's a lot of paste the heat has to transfer through, instead of metal to metal, the paste is just supposed to fill in the microgaps instead o air, not a layer over the whole thing. also notice how it only covers the center of the CPU. leaving a good 40% of the surface unable to transfer any heat. 

 

so those methods while simple to apply, gives you a much thicker layer, and a much smaller area for heat transfer, since it's so thick there's zero heat trasnfer on the areas without paste. 

 

granted modern paste is so good it doesn't matter much outside of a couple of degrees, and that's why they started reccomending it. it doesn't make much difference with modern paste and it's easier to apply. but if you're after optimal heat transfer, you spread it out. personally I just use a tiny little rize grain in the middle and don't spread it out because I Don't care that much about heat and it doesn't make much of a difference, but that massive 4 times as much as you need that dude puts on... OMG...

Link to comment
Share on other sites

All you need is a pea dot in the center.  If you de-lid your CPU, you'll see that the heat is primarily given off in the center of the metal heatsink lid.  You don't need a perfect square, what matters most is the dot that spreads to fill all the microscopic spaces between the metal lid and the metal of the heatsink.

Link to comment
Share on other sites

...everything about that video seems horribly horribly wrong.  First of all, that is NOT how I spread mine out.  I didn't leave lots of ridges and lumps that would create air bubbles; I spread it perfectly smooth, and I used about 1/10th the amount of paste he used.  Second, when he finally gets to his idea of the 'correct' method (the "pea"), it doesn't reach the corners, which is where the cpu cores actually are.  It covers maybe 50% of each cpu core.

 

The cores are located on a die in the center of the CPU, not the far corners. As long as you cover that, you're fine. Just make sure you press the heatsink on properly, and it'll spread into a nice, thin layer across all of the important parts. It'll spread out even more when it heats up too.

Link to comment
Share on other sites

Well you definitely don't want big gobs of paste or it'll do more harm than good. I've had a couple of Macbooks and an iMac where they applied paste like it was mayonnaise. Also, make sure you're using a quality thermal paste that's actually a thermal paste. Some of the cheap stuff I've seen is practically a rebranded Elmer's Glue.

Link to comment
Share on other sites

http://skinneelabs.com/2011-thermal-paste-review-comparison/

 

the video method is fine, you don't need to spread it the heatsink does that when you tighten it paste is paste IMO

 

 

 

also just to correct something you mentioned earlier, the cores are in the center not the edges you don't have to spread it thin out to the edges

 

If you do the video method, you do a tiny little rice grain in the center, a quarter of the blob the idiot in the video did. 

 

And no, paste is not paste, and the pressure of the heat sink can only do so much. 

Link to comment
Share on other sites

There is very little difference in compounds, look at the link or check out TPUs, its all within a few degrees. Again paste is paste. Regarding method of installing paste I will not argue a out, you have the pea method, the line method, the x method it depends on the compound you choose to use

 

TIMBestTemp5-535x600.png

 

 

I also suggest everyone look at this video http://www.youtube.com/watch?v=EyXLu1Ms-q4&NR=1 x method seems to be the best not pea, but it also depends on type of paste and type of heatsink

 

http://www.youtube.com/watch?v=ffK7L0Qj13Q&feature=iv&annotation_id=annotation_461720 here is with mx2

Link to comment
Share on other sites

Well, the only component I ever had die of overheating was a video card that had gobs and gobs of thermal paste on it from the factory. 

 

If you look at a factory cooler (and trust me, Intel / AMD engineers probably know more about thermal transfer than you do), it's always a small square in the center of the heatsink, and fairly thin. 

Link to comment
Share on other sites

again im backing my info with sources; obviously they know more but if you read anything I typed it depends on sink what type is best and its a circle pattern not square on the intel sinks, its also pre applied of course they wont pre apply with a drop but a thin layer

Link to comment
Share on other sites

I guess it depends on your definition of "center" and "corners"

 

300px-AMD_Phenom_die_equalized.png

but okay, tonight I'm gonna try reapplying it using a centre pea like in the video.  I think it's too much, as I believe the resulting extra thickness will add too much thermal resistance, but maybe I'm wrong.  I only have so much paste left in the tube.

Link to comment
Share on other sites

I guess it depends on your definition of "center" and "corners"

 

That only shows a shot of the die, not the entire chip. Here's what a delidded CPU will look like:

 

cpu-delidded.jpeg

 

Granted, that's a different chip from what you have, but it's the same idea. That reddish part in the middle is what generates all of the heat. As long as the thermal paste covers that particular area, you're fine.

Link to comment
Share on other sites

I guess it depends on your definition of "center" and "corners"

 

300px-AMD_Phenom_die_equalized.png

but okay, tonight I'm gonna try reapplying it using a centre pea like in the video.  I think it's too much, as I believe the resulting extra thickness will add too much thermal resistance, but maybe I'm wrong.  I only have so much paste left in the tube.

 

 

you do realize that isn't the entire processor and only the die, with the IHS on the cores are in the middle ....

edit someone beat me to it

Link to comment
Share on other sites

I always make an x, 2 diagonal lines from corner to corner. I tried different methods from pea sized dot, to a happy face. X gave me the best results.

 

I REALLY hope you use quality paste if you use the x method and especially if you use that large an x, because that's a LOT of paste and will result in at least a 1mm thick layer of paste. 

Link to comment
Share on other sites

I REALLY hope you use quality paste if you use the x method and especially if you use that large an x, because that's a LOT of paste and will result in at least a 1mm thick layer of paste. 

 

well if youre doing an x, of course you're going to use a fine tip syringe so you don't get too much on there. anyone whos put thermal paste will know if there is too much on there when applying it. I might get a little overshoot sometimes, but overall,it gives the most consistent spread IMO. I always use mx-4

Link to comment
Share on other sites

Well you definitely don't want big gobs of paste or it'll do more harm than good. I've had a couple of Macbooks and an iMac where they applied paste like it was mayonnaise. Also, make sure you're using a quality thermal paste that's actually a thermal paste. Some of the cheap stuff I've seen is practically a rebranded Elmer's Glue.

 

 

Does Mayonnaise Last as a Thermal Compound? (hardwaresecrets.com) :shiftyninja:

 

Yes and no.

Link to comment
Share on other sites

This topic is now closed to further replies.
  • Recently Browsing   0 members

    • No registered users viewing this page.