I have a fairly new build on my hands, built it in early january. Up to this point, it has been working fine with two exceptions: a message has popped into view, saying that "Warning! Your CPU-fan is not working properly. You must shut down immediately" or something to that effect. This has happened twice now during the past month. Alarmed, I always immediately checked the temp reading of my core (always visible on desktop) and in neither case has it risen above 50 degrees C (considering I was doing something fairly core-intensive calculations at the time, it was nothing to be alarmed of; my CPU might go up to 60C on prolonged high-intensity use with idling on windows desktop being around 33C).
So I simply clicked the message away and as the core temperature was within limits, I shut down the program instead and my core temp immediately began to cool off as the CPU returned to idle state. I ignored the messages (besides running full virus sweeps etc that came up empty) as my core temp didn't spike and my system remained stable and without any problems. I simply remained vigilant for any subsequent problems which never occured.
That was during the past month, as I said.
Just a few minutes ago, I received the message again, only briefly. It was followed by an immediate blue-screen, first-line saying that I've experienced a critical hardware malfunction and that windows was shut down to protect my computer. That's all I could read before I reached the powerswitch and turned it all off.
I opened my casing to see if there were any visible malfunctions, problems, indications of short-circuits etc. I found none. Carefully keeping an eye on the hardware, I booted up.
Normal start-up without a hassle, though the CPUID Hardware Monitor reported CPU temp at 45C at startup (had been roughly 48C before alert/shutdown), climbing to 51C as windows continued to boot up, then dropped down to normal idle temperature of 33-34C within minutes.
I've tested my cpu-fan connections while in shutdown, the powercable is in mint condition and the connectors both on the cable and the MOBO are like new as well. The screws connecting the fan to the CPU radiator were tight and secure. There was no significant amounts dust anywhere; I like to keep both the inside and outside of the case clean. Visually, the fan works as it's supposed to and there are no alarming noices, only simple quiet hum.
I'd like to know your opinion on this, and any possible measures you'd recommend for monitoring/fixing this issue.
Software-wise I usually have CPUID Hardware Monitor up, and speedfan 4.40 on the background to control fan speeds (I have two intake case fans, 80mm and 90mm)
Specs on my build:
OS: Win7 64-bit Ultimate
CPU: AMD Phenom II x4 925 (overclocked to ~2.95 GHz per core), stock fan and radiator
RAM: 4Gb DDR3 1333 dual-channel (manufactured by Kingston)
MOBO: Asus M4A785T-M (based on AMD 785G -chipset)
DVD-drive: Standard cheap SATA manufactured by LG, DVDRW 22X BLACK BULK
PSU: CORSAIR PSU 550W 12CM ATX12V2.2 80+ (Manufacturer's product code: CMPSU-550VXEU)
HDD: Basic SAMSUNG 500GB SATAII
GPU: (CLUB 3D) HD 4670 PCIE 1GB
As a curiosity, could this be caused by a 2-wire case-fan getting somehow mixed up for the CPU fan by the OS or software? Because as far as I can see, that is the most probable source.
In this case, would it help to remove the case-fan from the MOBO-connector and just plug it straight in the PSU-cable? This would mean the fan would be running at 100% the whole time, but I can bear the increased noise if it meant secure knowledge that there really is nothing wrong with my CPU-cooling.
If this was a BIOS problem, shouldn't it crop up more often? I admit I don't know for sure, but that would sound more likely if it really was a misconfigured BIOS giving me problems.
That message really sounds like it comes from the BIOS loop that monitors the CPU cooler fan speed. To be sure, do you have your CPU cooler fan plugged into the CPU_FAN pinout on the mobo? Because I bet this message is telling you that the speed signal from the fan (on a 3-pin fan system , it is the yellow wire lead) is NOT getting to the mobo all the time. In fact, the history you tell suggests that, on several occasions in the past, the signal was not detected (causing the message) and then came back. BUT more recently the signal suddenly quit and did not come back quickly, so the message caused Windows to shut down to prevent the CPU from overheating. This is a fast-acting protection circuit designed to prevent overheating, rather than just wait for the CPU temperature to go high when a fan really fails.
The likely cause is a faulty connection somewhere in fan leads. On a 3-wire fan, the red is +12VDC (varies less than 12 for slower fan), black is Ground, and yellow is the fan pulse signal going back to the mobo. A brief bad connection in any of them could cause this. Examine all leads again closely. Maybe take the side off and wait for the malfunction, then look VERY quickly inside to see if the CPU fan really is stalled - you may or may not see that. You could try replacing the fan. It's also possible the mobo circuit has a subtle weak connection somewhere.
Okay, thanks for the help - I checked out my mobo on the BIOS side, everything was set up a-okay and version is up to date. The fan is connected to the pins clearly marked both in the manual and on the mobo as cpu-fan, the connector is there securely and there are no other cables bending the connector or the cables. Visual inspection with magnifying glass also showed no problems.
Based on your posts that makes me thinks that its likely the cables between the fan and mobo; perhaps a poor connection to either the plastic connector at the mobo-end of the wires (which is pretty flimsy now that I think of it) or at the connection between the cables and the fan.
Looks like I'll have to run tests with this, and take out a spare mobo+fan to test the parts individually...
It seems that the yellow fan signal wire has a poor lead to the plastic connector, which is then connected to the mobo pins. At certain positions of the wire relative to the connector, no signal comes trough (tested with a multipurpose electronics meter) and apparently a strong draft inside the case is enough to move the flimsy wire a bit, consequently most likely sometimes briefly hitting a position of no-connectivity.
So, problem solved, case closed and I'm off to find a replacement fan without faulty leads. Thank you.
I solved my issue with this by unchecking the option in AMD overdrive to warn you if CPU fan is broken and it was telling me it was and it wasn't, so unchecked it, clicked apply and re-booted. AMD overdrive is running in the back ground always I guess if installed even if you think it's not.