Bad Temp Sensor?

[Cluster]

I just finished installing a new Thermaltake V1 heatsink on my Phenom 9850. It was running on the stock heatsink doing 3ghz @ 1.4v for the last couple of days. Boot my system up with the new heatsink. Idles around 41c. Then i go to run Prime95 and my temps start sky rocketing north of 100c!? Topping out at around 116C according to 3 different pieces of software. Not only that but nothing was crashing. I have a monitoring unit with thermal sensors, i did the best i could to place it under my heatsink next to the heat spreader, and at the time my cpu was showing 116c it was showing low 50s.

Anyone got any ideas wtf is going on here?

Link to the heatsink:
http://www.hardwaresecrets.com/article/444

[demonkevy666]

I just finished installing a new Thermaltake V1 heatsink on my Phenom 9850. It was running on the stock heatsink doing 3ghz @ 1.4v for the last couple of days. Boot my system up with the new heatsink. Idles around 41c. Then i go to run Prime95 and my temps start sky rocketing north of 100c!? Topping out at around 116C according to 3 different pieces of software. Not only that but nothing was crashing. I have a monitoring unit with thermal sensors, i did the best i could to place it under my heatsink next to the heat spreader, and at the time my cpu was showing 116c it was showing low 50s.

Anyone got any ideas wtf is going on here?

Link to the heatsink:
http://www.hardwaresecrets.com/article/444

IHS on the cpu my not be flat. I used the stock heat sink once it seemed to force it flat.

or it could be a bugged senors.

[informal]

I would have to say bad Asus BIOS.You should visit the official product page and the forum attached to it.Like many other boards from M3A series,this one seems to suffer from bad temp. reports(reporting a lot higher temps. than actual).I have plain vanilla m3a board and after 806 bios update my temp. reading from my X2 athlon@3Ghz shot up from low 50s to high 70s degrees Celsius!(seventies!!!).I went to Asus forum for m3a and saw a bunch of people(phenom users included) who had the same issue.

You can contact the tech support and see what they tell you(i suppose this is a dead end since i know what they will say to you ,but you can try either way..)

Cheers and good luck!

edit : There is a chance as demonkevy666 said that you have non flat IHS or bugged CPU sensor(the later option is not likely).

PS My real temp. of both cores are around 55C under dual prime95 small fft test.As you can see,it's a far cry from near 80C my board reports after the bios update.

[Cluster]

Ugh, none of the above, PEBKAC... Bad mount, didnt think i was that rusty i guess. Ran Prime95 for 10 minutes, 43c, much better.

Which means my temps were right, that chip was actually doing 116c, havent seen that since the fan died on my 1.4ghz athlon tbird that was doing 2.2ghz lol

[informal]

Ugh, none of the above, PEBKAC... Bad mount, didnt think i was that rusty i guess. Ran Prime95 for 10 minutes, 43c, much better.

Which means my temps were right, that chip was actually doing 116c, havent seen that since the fan died on my 1.4ghz athlon tbird that was doing 2.2ghz lol

Actually that's the best news you could hope for .Glad you sorted it out

[daseto]

holy crap arent phenom's rated for like 65c max? wouldnt 116c like completely fry that chip?

[Cluster]

On further tinkering, it wasnt me at all, its this heatsink. Apparently it does not like to be vertical in a tower case... wonder who the genious was behind its design. The only way i can get the heatsink to work is to lay my tower on its side. If i stand my case upright, the temps skyrocket within 30 minutes. If i lay it on its side so the mobo is horizontal, 34c idle, 43c load at stock speeds. Too bad my mobo heatpipes are designed to be vertical... guess im shopping for a new heatsink.

And a note, dont buy a Thermaltake V1 heatsink unless you plan on keeping your mobo horizontal.

[soundood]

On further tinkering, it wasnt me at all, its this heatsink. Apparently it does not like to be vertical in a tower case... wonder who the genious was behind its design. The only way i can get the heatsink to work is to lay my tower on its side. If i stand my case upright, the temps skyrocket within 30 minutes. If i lay it on its side so the mobo is horizontal, 34c idle, 43c load at stock speeds. Too bad my mobo heatpipes are designed to be vertical... guess im shopping for a new heatsink.

And a note, dont buy a Thermaltake V1 heatsink unless you plan on keeping your mobo horizontal.

i have seen this before, not with this HS, but on others of the same flavour,

si the HS mounting bracket that is attached to the mobo tight? it might have a little play in it, that the stock HS and fan would be more forgiving about.

also it might be the v1 bracket that is a little loose, either way, when you sit the tower i the correct postion it is causing the V1 to lean off the chip slightly,

have another look, and when the mobo is flat to ground try and move the HS by hand to see if there is any play.

you might just need to pack out the mounts a bit

[Cluster]

i have seen this before, not with this HS, but on others of the same flavour,

si the HS mounting bracket that is attached to the mobo tight? it might have a little play in it, that the stock HS and fan would be more forgiving about.

also it might be the v1 bracket that is a little loose, either way, when you sit the tower i the correct postion it is causing the V1 to lean off the chip slightly,

have another look, and when the mobo is flat to ground try and move the HS by hand to see if there is any play.

you might just need to pack out the mounts a bit

Thats totally possible, i'll take a look into it.

But, i am leaning towards the orientation of the heatsink making the difference and here's why. If you look at the heat pipes on the heatsink, you'll see that each pair run in opposite directions from the block. Now when i mount this heatsink to my board, i have to mount it such that the heat pipes lie on a vertical plane. In relation to gravity, two pipes then have to go 'up' and two 'down'. Now, given that all the fluid in the up-pipes resides in the block from a cold start, and the fluid in the down-pipes is sitting in the elbow far away from the block one could theorize that i would end up with some cooling capacity from the up-pipes until the liquid vaporized enough that it made it past the elbow and condensed into the part running through the heatsink fins, and not back into the block.

The reason i went to this length to try and figure this out is because if a gap had simply formed when i mounted it vertically, would i not lose cooling almost instantly? I was able to run Prime95 x 4 for 25 minutes and maintain a temperature of < 50c. But once it did go above 50, it started to rocket up past 80c at which point i called an end to my observations. One thing i have not done yet is an active monitoring of the heat pipe temperatures themselves which im going to do next. I've got a monitoring unit with 4 thermal sensors, and im going to see if i can find the point where the up-pipes are pushed over the edge and simply lose their cooling effect.

[demonkevy666]

Thats totally possible, i'll take a look into it.

But, i am leaning towards the orientation of the heatsink making the difference and here's why. If you look at the heat pipes on the heatsink, you'll see that each pair run in opposite directions from the block. Now when i mount this heatsink to my board, i have to mount it such that the heat pipes lie on a vertical plane. In relation to gravity, two pipes then have to go 'up' and two 'down'. Now, given that all the fluid in the up-pipes resides in the block from a cold start, and the fluid in the down-pipes is sitting in the elbow far away from the block one could theorize that i would end up with some cooling capacity from the up-pipes until the liquid vaporized enough that it made it past the elbow and condensed into the part running through the heatsink fins, and not back into the block.

The reason i went to this length to try and figure this out is because if a gap had simply formed when i mounted it vertically, would i not lose cooling almost instantly? I was able to run Prime95 x 4 for 25 minutes and maintain a temperature of < 50c. But once it did go above 50, it started to rocket up past 80c at which point i called an end to my observations. One thing i have not done yet is an active monitoring of the heat pipe temperatures themselves which im going to do next. I've got a monitoring unit with 4 thermal sensors, and im going to see if i can find the point where the up-pipes are pushed over the edge and simply lose their cooling effect.

there isn't enough liquid for that to happen. also note that the amout is small insider there and is under vacuum, so it will boil at lower temperatures .

maybe it's a carrpy mount for Am2/3+ that doesn't apply enough pressure to the IHS of the cpu

[Cluster]

there isn't enough liquid for that to happen. also note that the amout is small insider there and is under vacuum, so it will boil at lower temperatures .

maybe it's a carrpy mount for Am2/3+ that doesn't apply enough pressure to the IHS of the cpu

Agreed. I just dont think this thing applies enough pressure. Lie it flat on its back and this thing is great, 34c idle and 46c prime95 load after 30 minutes with a 27c case temp. I setup the box in a different place so that i can lay it flat, all seems good now. Now, to hook up my cold air intake

[wesleys_dad]

Hard to believe that its the heatpipes.. most of the coolers with heatpipes are mounted horizontially and this dosent occur..if your problem was correct than all heatpipe coolers would have to take a turn upward to mount with the top of the pipes facing towards the top of the case...you have a wierd problem there...I tryed your theory on mine and temps are exactly the same with case upright or on its side..
ill be interested in seeing what you find...

[demonkevy666]

Makes you wonder about the max temperature for Phenom. If it can take 116C and not become unstable maybe AMD 66C is just trying to keep us from blow mosfets/VR's on their boards?

I just test my chip up to 1.7 volt (or as the board over volts form 1.650.... 1.696 volts 3.2ghz) the max temperature was 67C and nothing got unstable this on AIR too. My PS is getting hot and it's 1200 watts LOL

also why would they put the 9550 at 72C and the 9950 at 66C. shouldn't that added volts make it hotter and have a higher limit?

[Cluster]

Makes you wonder about the max temperature for Phenom. If it can take 116C and not become unstable maybe AMD 66C is just trying to keep us from blow mosfets/VR's on their boards?

I just test my chip up to 1.7 volt (or as the board over volts form 1.650.... 1.696 volts 3.2ghz) the max temperature was 67C and nothing got unstable this on AIR too. My PS is getting hot and it's 1200 watts LOL

also why would they put the 9550 at 72C and the 9950 at 66C. shouldn't that added volts make it hotter and have a higher limit?

My temps were reaching north of 100c with no overclock on the chip. Of course it was running Prime95 for a couple of minutes before it produced an error and i realized wtf was going on.

As for how things turned out... pretty much the same. The clip was a tad loose, so i put some packing in to make the clip tighter, no dice. Put some packing between the clip and the heatsink block itself, it helped marginally. I think my problem is that this heatsink just isnt fit for a phenom. Im running 2875@1.3v and it idles 34c, load is 44-45c.


But all i have to do is nudge the vcore to 1.35v in the bios and i get this:


It plateaus a couple of times and looks like its going to stabalize, then as soon as it breaks past the 50c plateau, the skys the limit!

Either way, im going water, enough of this air nonsense.