Chared/burned thermal compound need help duron @ 1ghz

[frastini]

Hi, i got an overclocking combo from ocz, I couldnt wait for my heatsink to arrive in the mail, so i slapped on my oldschool alpha for socket 370.. (the one without copper in the base)

I threw on some silicon heatsink compound (have no idea what it is).. the chip was running quite hot, 120F and sometimes higher at times. I dont know if this was the spring for the alpha which i tightened once because the chip was approaching 140F at 800mhz.. tightening the spring dropped me down to the 120F range. on to my problem...

when i received my NEW heatsink in the mail, i took off my oldschool alpha, and noticed that the heatsink compound on the cpu was brown, and looked as if it might have discolored due to temperatures. Note: this compound was on the cpu for probably no more than 2-3 days.

my questions are:

1) what is specifically bad about silicon based heatsink compounds?
2) have silicon based heatink compounds ever been known to discolor to a fairly severe shade of brown?
3) i am currently running at the same frequency/mhz aprox 20F degrees less with a fresh coat of thermal compound (new heatsink would not fit motherboard) could the browned heatsink compound have dried up and become ineffective?
4) does anyone here need a alpha pep66 with quiet fan? wont fit the kt7 :/

anyhelp would be appreciated.

Frost

 

[DaddyG]

What you experienced is typical of silicone based compounds but not in such a short time. I'm guessing that your HSF is not fitting correctly. Do you know the core voltage that your cpu was set at ??

Why not give the TAISOL CEK733092 a try. AMD approved, cheap at about $11 and the CHOMERICS T725 PCTC that is pre-applied will never dry out.

 

[Nevin]

DaddyG,

You say,


<< What you experienced is typical of silicone based compounds but not in such a short time. >>


I have checked thermal compounds all the way back to some in a 1970s era power converter that runs hotter than hell. When can I expect the compound to char and turn brown since none of them have yet in 20+ years.

My guess is that the mystery paste he put on the chip was not thermal compound, but some sort of plain white grease.

Nevin

 

[Bartman39]

Hummm... Grease under normal conditions should not even discolor till around 300-400F (clear, white, otherwise) but if contaminated with something that would be a different story... Go for some Artic Silver... da good stuff!

 

[frastini]

well this thing says
type 44
silicone
Heatsink compound

on the back it says Meets western electric specifications
ks21343 and exceeds military performance Mil-c-47113

its manufactured by GC Electronics

after reapplying a coat its 750mhz @ 102F but what puzzles me
is why the silicon duscolored earlier.

i would get a taisol, but im not sure it can handle the heat of a duron @ 1ghz 1.85v this thing runs HOT.

 

[DaddyG]

Nevin,

Keep pimping your Arctic Silver !!. I have nothing to sell and nothing to gain. Silicone based compounds lose effectivity and dry out, if you havn't experienced this, then tuff. Submit your compound to AMD for testing !!. The plain facts are that most people don't know how to apply compound correctly, they assume that the pressure of the heatsink clip will squeeze out the excess and this is simply not true.
PCTC pads are almost impossible to install incorrectly, although show me something thats idiot proof and I'll show you a better idiot.

 

[Nevin]

DaddyG,

AMD will only consider approving thermal pads and PCTC as their extra thickness compared to thermal grease provides the cushioning the fragile AMD cores need to survive a heavy-handed heatsink mount. They are willing to sacrifice performance to cover themselves against having to replace the chips damaged by mechanically challenged people.

Those of us within the heatsink and thermal interface materials industries understand this. It is just another case of a bug being called a feature.

Nevin

 

[langour]

it may be possible that additives in the thermal compound may have reacted with the paint or the metal that you applied it to, especially with the increased heat.

most paints used by the manufactures will be fairly cheap, since the expected life of the electronics is short.

i doubt the discoloration is anything to worry about.

 

[frastini]

the really strange thing is a day later, and without many changes im running at 107F @ 750mhz thats a 5F degree jump from my previous post,
i intend to monitor this to see if it continues till 120F like it did earlier.

my two hypotheses are

1) the thermal compound is drying out / overheating / burning
2) the spring to my alpha becomes in effective after a few days of use

both are quite far fetched ideas but i dont know..

 

[DaddyG]

Nevin,

INTEL also uses PCTC on retail Pee3's and have stated that the warranty WILL BE VOID if the pad is removed. I doubt your explanation regarding the padding effect. TBIRD/DURON cores are crushed on the edge, not by vertical force, I doubt 5/1000 of PCTC provides that much protection. If it does provide protection, another good reason to use it !!. I believe the reason AMD will not recommend silicone based compounds is that most people apply the compound incorrectly, AND then expect AMD to be responsible. With the exception of the failure to remove the protective film from PCTC, its almost Fool Proof.

 

[Nevin]

DaddyG,

You're right.
I and the engineers I've spoken to at Dell, Ibm, Conpaq, Intel, Microforge, Swiftech, Globalwin, Millenium, and countless overseas heasink manufacturers are all wrong.

Intel says that overclocking will void the warranty. But they can't tell if you overclock.
Hmmm-
So Intel says that removing the stock thermal pad will void the warranty. And the stock thermal pad is attached to the stock heatsink. And you have to change the stock heatsink to overclock. And with a decent microscope, they can tell if you used anything other than the stock thermal pad/heatsink.

Overclocking problem solved!

Nevin

 

[frastini]

daddyg i never knew there was a &quot;correct way&quot; to apply thermal compound. However whatever meathod i use i seems to work,
the brownish tint puzzles me though.

What i've gathered from the responses is that this may just be the particular compound im using, I will try another type of heatsink compound.

 

[Mikewarrior2]

if you've used any socket A heatsink, and most p3 heatsink clips with new, good quality thermal grease...

then the pressure of the clip is more than enough to squeeze out excess grease. I'm not saying to glob a quarter sized glob on it, but a little bit extra is easily squeezed out. Want a way to find out? put a little bit extra grease on. Install hte heatsink. Pull heatsink off. NOtice some of the grease has been pushed onto the sides of the core and the corresponding non-core areas of the heatsink.

A little bit extra grease is far less detrimental than too little grease.


Mike

 

[frastini]

update, i was running at aprox 109-111 F with the unknown thermal compound, after a few days of running at 750mhz 1.6v and very little
temperature change (even after lots of quake3)

I decided to kick it up a notch 850mhz @ 1.7v

to my horror within 1 hour of just playing mp3s in win98 the chip temp
approached 140F... i smelled a strange odor.. i quickly powered
down once i saw the temp in my motherboard monitor and rebooted
back to 750mhz @ 1.6v hoping the temp would go back down to 110F

it didnt, it idled around 120F and wouldnt go back down, so i decided to investigate I took the heatsink off to check the paste, sure enough it had started discoloring again.

I cleaned off the discolored paste, and put a different brand of silicon paste on the heatsink, now its idling at 109-111F.

note: i have posted and ran some benchmarks of this chip at 1ghz,
i am extremely disappointed with having to run at this low speed
till my new heatsink gets here :/

I will try 850mhz once again with this new silicon paste.. maybe the old paste i had was just defective at higher temps :/

 

[DaddyG]

Frastini,

There is a correct way to apply the paste. The Arctic Silver site has a good example. It is NOT correct to assume the the clip pressure squezes out the excess correctly, you should apply only enough to get a translucent film, just a little more if your heatsink surface is not perfectly flat. Hopefully Nevin agrees with this.
TBIRDS get VERY hot VERY quickly when the thermal interface is messed up. 8secs to self destruct without cooling. I would check and RE-CHECK that Alpha, you probably should remove your mobo and look between the HSF and chip for any spaces. Your clip should provide a minimum of 12lbs, max 20lbs. Personally I would spend the $11 for a TAISOL CEK733092. Attach it with the PCTC pad after cleaning the cpu slug very well and have confidence that this solution will provide the required cooling.

 

[Nick Stone]

frastini

I think you have been neglecting your wife/GF and she replaced your heatsink compound with some vaseline just to get back at you

 

[HecDTec]

Daddy G is right on the money about the application of thermal grease. I bought Acrtic Silver and wrapped around thier tube was a note stating to go to thier website for instructions of proper application. I also read at another site that you should remove the &quot;mobo&quot; from the case when applying grease and/or heatsink/fan. The light in the background will help you see problems with the mating of the CPU and the HSF. The thermal grease is put on very thinly as to make up for small descrepencies in the mating materials and shortening the path from the Die to the HSF for effecient heat tranfer. I'm using a Chrome ORb which is better than stock, but worse than Globalwin FOP32 and Alpha 6035 and I'm reaching 1 Ghz comfortably at 35c. Difference is my L1 Bridges came intact.

 

[fast0]

Heheh, I have an interesting story about the golden orb...

I bought a golden orb (the revision 1) to put on my Tbird 850.
Put it on, got temperatures of 63Celsius. (That's like 180F I think)

Computer won't stay stable for very long.
Lower clock speed. Put oscillating fan beside computer. Runs stable at 48.

Buy Alpha PAL6035.

Computer runs at 850 @ 33C
Next step, overclocking.

 

[Shazam]

The Golden Orb is not meant for AMD procs.... You're lucky you didn't crack your TBird in half.

 

[Technonut]

DaddyG, Do you own stock in Taisol or Chomerics by any chance? j/k Seriously, I used the stock HSF with the thermal phase compound, made sure it was installed properly, and ran it hot enough for the phase change to take place. It ran hotter than hell at default settings. I removed that bazooka bubble gum crap and put Arctic Silver on it. It did not do much better. We are talking 50-60 degrees Cel. (800 T-Bird) I received an Alpha 6035, applied Arctic Silver, and now idle in the low 30's. I do have a 38 CFM Delta on it also. It has never gotten past 40 Cel, no matter how loaded. The Alpha installed like a dream, and never made me feel like I was going to crack the core. I tried an FOP32 today, and it was a nightmare to install, and contrary to reviews that I read, it came no where close to the Alpha in performance even with a 38 CFM Delta on it.

I admit I have not tried the Taisol, but with the performance, quality and ease of installation the Alpha offers, and a staggering cost of $24.00, (without the fan) why try it? I, like others put a lot of time and cash in my personal system, and with the money spent on other quality components why skimp on a quality unit that will cool and protect the heart of your system. IMHO it falls in the catagory of &quot;false economy&quot; $20-40 dollars should not break the bank compared to all the other cash spent. Just my opinion of course

 

[DaddyG]

Techno, no stock in either company
Nothing against Alpha either but the original post said something like
'oldschool Alpha for Socket 370'. Its a fact that many S370 just don't fit right on S462 due to the increased core height of TBIRD/Duron.

When I emailed AMD about PCTC they provided names of a couple of other manufactures beside CHOMERICS, but from my research, T725 seems to be the compound that many HSF manufacturers are using. Your results with the Alpha are great. I 'pimp' TAISOL cause they generally test to within a couple of degrees of Alpha, they install fairly easily and are readily available. The fan can be easily changed if you like the sound of jet engines when you apply the 'squeeze' to your baby !!

 

[frastini]

well, i believe most of my troubles were in this &quot;type 44&quot; heatsink
compound..

i have since switched to a different silicon compound that is not
as prone to such problems at high temperatures. my fop32 is en route,
and im sorta disappointed the alpha pep66 wont fit in my kt7.

the oldschool 370 cooler has a 27 cfm 60x60 fan on it, if i had
a delta, i might not have as many heat problems :] but i doubt i could live with the noise.

I hope i am not disaapointed with the globalwin :]