[Asus P5Q] Very low CPU temps, resulting in fan stuck at low rpm

[toyo]

Hello,

I'm having this problem lately with the CPU temp (not the cores, these temps are reported correctly and are into the low 40s on idle), it's most of the times detected as low as 1-5 degrees (Celsius). I did not made any changes I can remember other than replacing a faulty HDD with new SSD/HDD (but I cannot guarantee this was the start of the problem). It just started to happen (I have AIDA64 OSD running).
Important mention: the CPU sensor is not stuck when it indicates low temps, the temps go up on load, but they're 15C or so lower than normal.

This is bothering me because I use a Freezer PWM cooler set on Turbo on BIOS. The fan needs to go as high as 2500-2800 rpm to keep the CPU cool when it needed to work on the hot ambient here, but now, because the low CPU temp is detected, the fan sits lazily at 1100 rpm and the cores go up to 65C (tested with IntelBurner). What's even more annoying, is that sometimes the correct CPU was reported after a reboot (this is not happening any more - the temps are wrong for good it seems) and the fan immediately starts working correctly again, but it never lasts. Everything on the motherboard is correctly seated with Arctic MX-3. I did a fresh reseating for the Arctic cooler, but it did not help.

Additional problems I have noticed: the CHA fan is no longer detected, albeit it's working OK. A wrong DRAM reading appeared out of nowhere during my attempts to repair this issue, detecting a 2.02 V - incorrect, as the RAM sticks are working at 1.8 V.

I have also tried reflashing the BIOS to older versions, resetting CMOS, overclocking the CPU up to 3.5 GHz, everything in an attempt to get the sensor IC unstuck, but none worked.
Speedfan is not a possible resolution - I cannot control the fan on my MB with it.

Is there any way I can "reset" the sensor to its former, correct behaviour? Please help, I'm getting desperate here, as the only way I can tackle more demanding tasks other than browsing is to disable the Q-FAN BIOS option and let the fan always work at 2700 rpm - it's like having a blowdryer in your ears all the time.

Thank you!

E7200 2.53 GHz, Asus P5Q (2209 BIOS), Kingston 2x2GB DDR2-800 5-5-5-18 1.8v, Asus EAH4850 512MB 625/993, 23" Fujitsu-Siemens LSL 3230T, Corsair Extreme SSD x64/Samsung Spinpoint F3EG 1TB SATA-II, Windows 7 SP1 Ultimate x64

2.53 GHz, stock:

3.16 GHz, OC:

DRAM wrong entry:

 

[FreedomEclipse]

Sounds like the board could be borked if ram sensors are off too. maybe a little power spike shorted something on your board.

 

[toyo]

The system is stable, I put it through some torture testing hoping to get the sensor unstuck. That DRAM reading might be a HWMonitor flaw, I never saw it before (I have Kingston value RAM, nothing fancy, not sure the voltage can be really read).

 

[FreedomEclipse]

have you tried using other programs to display your temps???

 

[toyo]

Of course... I usually use AIDA64's OSD, posted HWMonitor screenshots for the sake of having more info on the same image. But the problem comes from the sensor. In BIOS, the same low temps are read. The Intel diode readings (the cores) are working just fine, only the readings from the Winbond IC went crazy. It even reached 0C at one time.

The PWM CPU fan depends on this CPU TEMP read, and this is my primary concern: getting the fan to work properly depending on system load.

I've sent an email to Asus support too, but I might be out of luck as this MB is practically 3 yrs old.

 

[FreedomEclipse]

well. I just put it down as the board is on its way out... CPU sensor is stuck. Ram sensor is borked. and your board wont detect the case fan thats plugged into the cha_fan.

if it was just the CPU sensor that was stuck that would have been alright. but all these issues are just too much of a coincidence,

 

[toyo]

I hope there's an alternative way of using the fan's PWM (like making it depend on the readings from the Intel diodes), as there's no way I can get another MB now, and I don't see better times on the horizon.

 

[FreedomEclipse]

I hope there's an alternative way of using the fan's PWM (like making it depend on the readings from the Intel diodes), as there's no way I can get another MB now, and I don't see better times on the horizon.

If youre looking for a board. try ebay, they will most likely do second hand and brand new ones.

Failing that. you might need to get a fan controller and hook everything up so you can control the fan speeds.

not totally ideal but when a ship starts taking on water, do you get a bucket and start bailing water or do you jump ship.

 

[toyo]

Sometimes it's just plain impossible to jump ship. Might sound cheap to a person that can upgrade every 6 months or so, but not all countries are made equal, and I happen to live in one that's cannibalised by the banks and its own gov. for long years. Investing in non-essential things could prove a totally fatal move for me right now, even if it's just 10$.

 

[FreedomEclipse]

I ran with my Q9550 for at least 3 years before i ditched it for a i5 setup. Im far from fitting into the 'upgrade every 6 months' stereotype.

Looks like a fan controller is the ideal option for you right now. but i do recommend replacing that board as you dont know if the problem can short out and effect any of the boards other functions.

 

[toyo]

Took the board out, resetted CMOS again and gave every chip on the MB some light finger shocks. CPU temp seem back to normal now, although still too low I think (equivalent to ambient on idle), but the fans kicks in properly on load now. DRAM entry is gone.
I'm still open to hear if there are other more permanent solutions, as I don't trust this lucky event to last for long.

 

[Mussels]

all those readings tend to come from the one sensor chip, and it sounds like its going borked, giving out weird readings on various sensors.


you might have to give up on automatic fan speed, and just manually set it via a fan controller or hard wired voltage.