Dangerous Temps for OC'd Durons

[lostnthenet]

I will be getting my Duron 650 and ABIT KT7-RAID today and of course I will OC it ASAP. What is the temp I should worry about it hitting? Below is what I found in the AMD Tech notes. That seems pretty high but I assume they know what they are talking about. Gimme your opinions guys (and girls, if there are any around.)


"Thermal solutions must
monitor the processor temperature to prevent the processor
from exceeding its maximum die temperature. The maximum
die temperature is specified through characterization at 90C."

Thanks,
ESP

 

[Heifetz]

Anything below 50C is ok for stability. Anything above 50, you might begin to experience instability.

Heifetz

 

[lostnthenet]

OK, but 50 sounds very conservative compared to the 90 this says it can handle. Do you know why it says 90 or if the temps we are talking about are measured in different locations?

 

[PochiePooh]

The temperature AMD gives out is the limit beyond which your Duron will explode and die, I think. However, the ideal temperature within which your computer will stably work is much lower than that.

 

[rmblam]

Many guys are running their's over 50C without problems. Right now mine hits 54C max under full load in a hot room. Mostly it runs upper 40's under load. I' wouldn't worry unless you are over 60C consistently.

 

[johncar]

lostnthenet,
This is going to be a case of "apples and oranges" plus the effect of different locations of various temp sensors you find on different mobos.

First off, the 90C spec temp is specified at "top dead center" of the Duron chip, (verified by AMD application engineer), and is the max safe operating temp "at that location", (engineer and datasheet). To compare to that statistic, temp sensor "must be" in direct contact with that specific location on the chip.

Second, it is impossible to measure the temp of that location, even if you drill a tiny hole thru hs base and install tiny thermocouple bead flush with the hs surface...because you will measure a lower temp from the temp drop across any grease interface...impossible to make "direct/dry/no grease in between" contact with that specific location. Even with that difficult install/location you would have to estimate the temp drop across the grease interface and add it to the TC reading to compare to the spec 90C max limit...requires knowledge of power dissipation of the chip...LOL.

Temp sensors at any other locations can never be compared to that "absolute" 90C spec, because of unknown temp drops from that hottest exterior location. Unknown temp drops being the consequence of an aggressively cooled, non-isothermal environment...trying to keep the chip safely cool. But these "anywhere else" temp readings are not "completely" useless. They can be used as "relative" indicators of change in cpu temp. Build your system, oc it running most cpu intensive apps you have, but make sure system is stable over a reasonable period of time. Then record your "anywhere else" temp and worry only if/when it rises dramatically running same apps. That's all you can do with those temp readings vis a vis safe operation.

The increase to 90C from the typical 65C, 70C, of K6-2 chips reflects the change in the design of the chips. K6-2 chips 70C was specified at top dead center of the "case". So the temp of the K6-2 die top was "hotter" because of the temp drop from die top thru internal "goop", heat spreader, and case...all non-existant on Duron. We could speculate that 70C for K6-2 case top is equiv to 90C Duron die top...apples and oranges.

Hope this helps understanding of the issue. Build, oc/run stable, look for "change"...best you can do, and should be safe.
John C.

 

[lostnthenet]

Johncar,

Thanks. That was a superb answer to my question. Of course thanks go out to everyone who has replied. I really appreciate the feedback. I just want to make sure that when I go start overclocking tonight that I don't ruin my chip. I really like these boards and hope to become an active member.

Thanks all,
ESP

 

[Jaraxal]

I am not sure what the problem is. I have 2 Duron 650's on Asus A7Vs (rev 1.02).

I was using a Cool Master heatsink (AMD approved) and was averaging temps of about 50C for the CPU and 30C for the motherboard.

So I ordered some Alpha PAL6035 with Artic Silver thermal compound. What are my temps now? 50C and 30C respectively. There is no change.

I have noticed that the Alpha heatsinks don't seem to want to sit level and flush. If I try to hold the heatsinks level, the temp does drop, but not to the 30C level that I often hear people see. At the moment they are not overclocked.

Any ideas?

 

[johncar]

Jaraxal,
If temps do not change after improving cooling effectiveness with better hs and silver grease, then suspect the temp sensor/s are not close enough to the chip...certainly the one labeled mobo is not, so do not expect that one to change.

But you do get a change when you manipulate the Alpha, so that's an indication your mods may have had some effect.

But we would take steps to assure the hs lays flat against the chip,
like a crutch to the case floor, or a wire hanger support. Chance of frying the chip if the hs cocks off flat.
John C.

PS Read this thread here at same oc'g category for some ideas re temp measurement..."Why You Can't Measure Beneficial Effect Of Silver Greases"

 

[Jaraxal]

Well I found something that is just the right size to help stabilize the Alpha on my Duron. Qtip. Yes, that's right, if you cut off the ends of the Qtip (plastic tube kind, not cardboard/paper), and insert it between the top part of the socket (where it says socket) and the Alpha it will sit stable and temperatures drop by 10C.

On a related note, I was also attempting to overclock my Duron 650. Nothing seemed to work to keep it stable at anything over 650. I was using the bios to manipulate the bus speeds and voltages. I decided to try to use just the dip switches and jumpers. Bingo. I am running rock solid at 900@1.85v. My temps are between 50-52C with a full load (ie benchmarking quake, unreal tournament, or 3DMark 2000).