Homebrew thermally controlled CPU fan

Soldato
Joined
15 Jan 2006
Posts
7,811
Location
Derbyshire
edit - *LESSON TO ALL: DON'T DO THIS*

Just thought I'd run this by someone with a little more electrial knowledge than me. My motherboard has no speed control for the CPU and System fan headers: it just runs them at 100% all the time. Until now I've been using a zalman fan mate II to achieve the desired balance between noise and coling on my CPU fan. Running at low speeds renders my CPU overclock unstable. I'm using an arctic cooling freezer pro64 cooler and fan at the moment on an X2 5000+ BE.

I had a wee idea. Left over from my old socket A system is a Thermaltake volcano (11 or 12 - I'm not sure) with a smartfan fitted. That ran at 100% for anything over about 55 degrees. Now what if I were to take the thermistor from the Volcano and add it in series with my Freezer Pro fan? I have a spare 3 pin fan extension cable and can jerry rig the thermister along the 12v line on it. I'd rather run off the motherboard connector rather than molex, as then I'll still be able to view the speed in Speedfan.

Any reason why this wouldn't work or can people foresee any potential pitfalls (besides my clumsy soldering skills...)? Any tips on thermistor placement for AM2 CPUs? I'm thinking somewhere near the bottom of the heatsink that doesn't get much air flow.
 
Last edited:
Don't see a problem with that TBH. Placement will just need to be somewhere where the thermistor can do it's job ie near or in contact with an area that will get hot so that it will in turn increase the volts to the fan.

I've actually scavenged some temp sensitive circuitry from some old PSUs at work and use them to control the fan speed of my case fans though the way I have the circuit mounted means the fans never speed up so I need to look into that at some stage.
 
In short: DON'T DO THIS. Typing this while backing up important stuff onto a USB pen with one eye on Speedfan. I've messed up my motherboard's (or hopefully daughter board's) voltage regulation or something. It didn't work. The fan didn't move and something made a whiney sound... :(

Now, having removed the mod it the fan works but something's gone horribly wrong: the CPU just gets hotter and hotter but doesn't cool down. VCORE climbs gradually. Heatsink's contact is still good. It gets hot to the touch.

Fortunately I just won a motherboard and CPU on the bay yesterday intended for a 2nd PC. Should be arriving soon :)
 
Without knowing the precise resistance range of the thermistor it'd be hard to predict the behaviour when in series with a fan.

I made an auto fan-controller a couple of years ago using a thermistor. The difference was it was part of a potential divider and used a transistor to give enough current drive the fan.:)
 
Hang on - simplest solution first, I think. Maybe I knocked the heatsink when swapping fan connectors? If I'd affected the contact could that produce the above behaviour (CPU heats under load and fails to cool at idle). The VCOREB results in Speedfan might be a red herring. Apparrently they generally read too high on this board. VCOREA is as I set it in BIOS.

I'll try reseating it with new thermal compound this evening and see if that helps.
 
If the circuit worked in line with the previous fan then I can't see why it wouldn't if used in the same fashion as the old fan. The only proviso might be the current draw of the respective fans, so if it controlled a 60mm fan and you've added it to a 120mm fan then the current draw from the 120mm may well cause a problem.

As I said I've taken a temp control circuit board from an old PSU and use that to run 2 80mm case fans in parallel and the fans work fine.
 
Well, I effectively put a high enough resistance on the CPU fan header to stop the cpu fan turning. The fan works without the thermistor in line - the heatsink just gets rather hot and the CPU doesn't cool down at idle.

As I said -simplest turn of events is I've reduced the contact between the heatsink and CPU. Heat transfer is still taking place but if there's air in there it could be acting as an insulator.

The more worrying thing is if I've somehow caused problems with something else. Maybe the CPU voltage regulation on the board or even the PSU?
 
Mama mia figaro magnifico!...

Reseated heatsink using a half grain of rice sized amount of AS5 (as per instructions) rather than spread on with a credit card like last time. Currently running S&M, just passed first test.

Now running 45°C under load @ 3GHz. We'll see how FPU test does. I reckon reseating the heatsink with less thermal compound has dropped temps 8-10°C from before the problem. A blessing in disguise no less!

With the improved cooling I might even be able to get a higher overclock or run silent.

Now the rubber hits the road - FPU test - Just hitting 50°C -

And staying there! If it drops I'll be extatic...

:D:):cool:

It's coming back down baby - yeah. All tests passed

We're back in business!

Edit: Further to the above. Turns out it was a Zalman Fanmate 2 causing the problems all along. When I added the mod, I put the Zalman controlling an antec Tricool 120mm fan that I didn't turn up to full. Looks like the fanmate didn't like it. Fanmate's removed and all is well.
 
Last edited:
Back
Top Bottom