Arctic Silver Stuff made my CPU temp Skyrocket... WTF?

[Krueger81]

Hello,

I applied some Arctic Silver Compund to my Duron 600 today and as soon as I turned the PC on the CPU temp skyrocketed to like 60C+. I turned it off and reapplied it ... Still the same. Threw some Radio Shack stuff on there and its stable at 49C and 900MhZ now

Any suggestions?

Krueger81

P.S.: Yes it was a very thin layer.

 

[pdo]

what kind of mobo do you have? Cuz I have an Asus A7V rev1.01 which I mounted the thermistor to the HS. Maybe the Arctic is better at transfering heat from the Processor core to the HS that's why there's seems to be a higher reading on your temp..

 

[Mikewarrior2]

You can't use any SOCKET A mb's cpu temp readings seriously. THey measure air underneath cpu temp. Not cpu CORE Temp.


Mike

 

[Krueger81]

Hello,

I am using the ABIT KT-7 and I have the thermistor bent up far enough so it touches the bottom side of the CPU. It was at 51 before i put the RADIO Shack stuff on. then I threw the Arctic Silver Stuff on there and it jumped to 60+ took that back off and put the Radio Shack back on ... running stable at 49C now and at 900MhZ.


Krueger81

 

[Mikewarrior2]

Do you really think the thermistor underneath the socket is accurate? What does it touch besides IDLE substrate?

Anyways, try a little bit thicker of a layer. Artic Silver is very easy to thin out when clipping the heatsink onto the cpu.


Mike

 

[toph99]

are you sure the heatsink is in good contact with the core? it's always good to double check

 

[Krueger81]

Hello,

I am sure its in contact with just the core as I can see a little blue mark on the CPU itself. ANd the Heatsink sits on there tight yes becasue it works fine with the Radio Shack goop.

Krueger81

 

[H4rm]

you obviously did something wrong. have the KT7-Raid, 1 ghz OC'd to 1.145 ghz with an Alpha Pal6035 and Artic Silver and my temps dropped initially 10 deg F when I applied the Artic Silver onto my old stock heatsink and fan.

 

[DaddyG]

MikeW,

The temps may not be accurate but the thermistor tends to 'lie' consistently. Maybe the hsf is not flat and the thin layer of AS isn't enough. If the hsf base has 'ridges' the thickness needs to be increased a little.

 

[Mikewarrior2]

Hence why I recommended trying a thicker layer of AS.

 

[spidey07]

I had a great deal of difficulty properly applying Artic Silver. It is so slippery that when mounting my Heatsink it would slide around on the Silver (I have a socket mobo). I tried three times and could never get it right. Here are my full load 1 hour temps:

Artic Silver - IMPROPERLY APPLIED!!!
58c, 56c, 61c

Goop from 3dfxcool.com
35c

Moral of the story is you may not have put enough on or rubbed it off when you mounted your HSF as I did.

 

[Rudee]

The key to effective cooling is applying it very thin. i.e you should be able to see through it. If it pushes out and over the sides of the slug then the effective cooling is diminished.

I managed to squeeze an extra 50mhz out of a friends CPU overclock by re-applying the thermal grease (he had it on too thick). Like I said, you should be able to see through it - it has to be a very very very thin film or it doesn't cool properly.

 

[Mikewarrior2]

No, the layer of Artic Silver should not be see through. THe stuff is very thin as is. if you Read their isntructions for application, you need to put at least enough to fill the gaps between heatsink and cpu.

In the end, a little extra is far better than having too little. Any extra will squeeze off the side and wont' hinder performance of the grease in any way.

Spidey07,

What heatsink are you using?

Mike

 

[resinboy]

I tried it last night for the first time on my rig. I was using the Chemtronics silver. Load was about 115' during heavy UT, idle was usually around 85. With the initial application, using a credit card for a smooth edge, I was unable to boot up at my overclocked speed of 1008( 144 fsb). Had to keep dropping the bus back to 133 for it to boot. Tore everything back down, added a really thin layer, wouldn't even boot at 133. This time I went into Hardware Monitor, and saw my cpu temp was 168* !!! Tore it back down again, added a layer to the slug, and a layer to the sink ( Alpha P3125)booted back up- everything back to normal. My idle temp is still the same, but I did notice a 5* drop under load.

 

[Krueger81]

Hello,

well guys I have it figured out. I run at 43C Stable at 900 MhZ and at 1.75 volts that's with a full load of RC5 . Had to re apply AS. I didn't have enough on there. I was able to lower the voltage and the temp dropped also.

Before AS with Radio Shack Stuff: 49C at 900 and 1.8 volts
After AS: 43C at 900 and 1.75 volts

Krueger81

 

[johncar]

DaddyG et al,
Re the repeatability of temp sensors...yes, agree most if not all of today's temp sensors are very repeatable, esp over small range we use them. But they can still "mislead" when used to compare effectiveness of interface materials....here's how...using estimated numbers just for illustrative purposes.

Let's say arctic silver lowers temp drop of interface 10C over the RS zinc-oxide stuff...then the chip's internal temp should also drop about 10C. But because of the lower thermal res of the "hs path", more of the "constant total heat flow" goes thru the hs path and "less" goes thru the ceramic bottom. So the thermistor touching the bottom should not show a 10C drop in temp, because that would say the branch heat flow stays the same because the temp drop from internal chip to thermistor is the same...ie both changed 10C. Remember heat flow = temp drop/thermal res of bottom path...which does NOT change. So if we say bottom temp drop doesn't change we have to say that neither does heat flow...but we explained above that heat flow "does change" path to path because of cpu/hs interface path improvement.

So let's say the "diverted heat flow" was 33% of former "bottom flow"...(which might have been only 10% of total). Then temp drop from internal chip to thermistor would only be 2/3 of former temp drop, which means that the thermistor "cannot" drop 10C, "the same amount the internal temp changed". That's the "mis-leading" aspect of temp sensors not measuring "directly" what you really want to measure.
(And we don't even know what the original temp drop from internal to thermistor was...2/3 of what???...so all we might get is a "system specific indication&quot.
John C.

 

[DaddyG]

Good post Jon C, as you know I'm a fan of PCTC but I will NEVER believe that AS, when applied correctly, has higher 'relative' temps than Radio Smack.

Hadn't really considered that by lowering the Thermal Resistance of one pathway, you could infact, effect the ratios of the multiple pathways.

 

[johncar]

DaddyG, you wrote....
>Good post Jon C, as you know I'm a fan of PCTC but I will NEVER >believe that AS, when applied correctly, has higher 'relative' temps >than Radio Smack.

Thanks for kind comment, and hope we understand higher 'relative' temp comment. Anyway...we didn't say bottom thermistor temp would "go up" with AS replacing RS stuff, just that it should not go down "as much" as internal temp goes down. It should go down, but not as much because the "temp drop" from internal to thermistor becomes less with the lesser bottom heat flow.

Also, additionally interesting, if you had a sensor "on the hs side of the interface", that temp should "go up", (proportionately to hs heat flow), for AS because of the "slightly increased heat flow thru that path" with no change in hs thermal res. Another instance of getting a technically correct result, but which most people would interpret as "AS is no good"...misleading if they don't understand "the heat transfer process".

>Hadn't really considered that by lowering the Thermal Resistance of >one pathway, you could infact, effect the ratios of the multiple >pathways.

Yes, most people don't consider that, and as above, can cause some temp location results to be misinterpreted.
Good dialogue DaddyG, John C.

 

[PliotronX]

In my own experience with AS, a PEP66, and a 700E, the best way to apply the AS was to tack a small bit of AS in the center of the core, and apply the PEP66 as flat as possible. When I tried applying the AS before attaching the heatsink, I made it paperthin, opaque consistency. This resulted in no warm air coming from the heatsink (delta 60 fan), and the back of the slocket was very warm to the touch. After 6 retries with the razorblade, I decided to try the "slop some in the middle and watch the excess be squished out" trick. Sure enough, the back of the slocket is room temp, and the air extracted by the fan is warm, as it should be. So, there are different ways of applying the AS.. there's a LOT of goop in that little syringe.. I still have 3/4 of the syringe left

 

[Mikewarrior2]

Artic silver is too smooth/thin in consistently to apply "paper thin". IT needs a thicker layer.

AND YES, a heatsink clip will squeeze out excess AS.


Mike

 

[jebus]

So what is the final word on applying Artic Silver? I just purchased it for my Duron 600 with Alpha 6035 overclocked. Should I apply thin, almost transparent? or should I apply thick, spread everywhere? Or apply to center, then let the HS squish it to the side? Thanks.

 

[resinboy]

personally, I would say a reasonably thick, even coat. I swapped motherboards this weekend. All I did was pull out the slot-1 cartridge, and re- install it. Ran the new board for a couple of hours to discover the voltage select didn't go high enough . Reinstalled my Asus, and the temp reading was now 20* Fahrenheit higher!! Pulled apart the HSF, added more silver, and back down to cool temps again . I never had that happen with the old stuff

 

[DaddyG]

Jebus,

The amount of AS to apply really depends on how flat and how smooth your heatsink is. A 'lapped' heatsink will need less than one out the box. Do not apply a blob in the middle and hope that it spreads. (Nevin should really comment here).