I2500 stock fan/cooler suddenly ineffective

Int13

Posts: 7   +0
I had a problem with my PC turning off which occurred in Ubuntu and Windows. I turned to the bios hardware monitor and found that heat was building up to 78+ degrees and then turning off.

The case was full of fluff and dust which I cleared out. The (stock) fan/cooler was loaded with the stuff. No wonder it was overheating.

Now I cleaned off the old thermal grease with isopropol alcohol, qtips, and a glasses cleaning cloth. I reapply a small amount of thermal grease. I had a lot of trouble with one push pin on the cooler but got that running. Each time I fit the cooler/fan I use fresh grease after cleaning off the old gunk. I found that the grease seems to be about right as, applying it in a small blob in the middle of the CPU, I found it covered the whole area of the cooler when I later took off the cooler (as cooling isn't working). The grease isn't leaking outside of the area of the heatsink so I guess my loading isn't excessive. The grease is Arctic silver 5. It may be a couple of years old. Is that the problem?

The temperature still climbs up to about 80 before shutting off. I turn it off early now, once I see the cooling isn't working.

Taking the other side of the case off, I can see that all four push pins are correctly seated on the motherboard. I check the CPU fan and it is working OK. The other case fans are working, but even they being off couldn't account for overheating like this.

Any ideas?
 
What are you getting for a reading on "Vcore"?

If the CPU voltage is too high, that would explain the heat up.

That's all I got ATM.

Is this thing overclocked at all?
 
Your input is very welcome. The CPU isn't over clocked at all. I will have to check the Vcore when I get home. I'm at the office at the moment.

Could a PSU malfunction raise the Vcore and would this be reflected in what I see on the bios h/w monitoring screen?
 
Could a PSU malfunction raise the Vcore and would this be reflected in what I see on the bios h/w monitoring screen?
No, the motherboard does the regulation and monitoring.

It is my understanding that the motherboard checks PSU voltages and when they checkout, sends a power good signal back to the PSU which keeps it on. In other words if the PSU specs are not within tolerance, the motherboard will tell the PSU to shut back down. This is done within the first few seconds of bootup.

What is the RPM of the fan, when it appears to be over-heating?
Is the CPU heat-sink too hot to touch when this happens?
 
No, the motherboard does the regulation and monitoring.
You can change Vcore with the BIOS, jus' sayin.

I've had a BIOS decide to drop the CPU multiplier for no apparent reason.

Admittedly, that was a wild guess, and is predicated on the assumption that TS's assessment of the HSF and TIM is correct.
 
You can change Vcore with the BIOS, jus' sayin.
Sorry captain, I wasn't refuting your comment. Just saying it's the motherboard not the PSU that controls the voltage levels for the CPU. Yeah the BIOS does allow for changing these levels, if the PC manufacturer hasn't locked down (disabled these features) the System BIOS.
 
Yeah well, I was going to mention about maybe a couple of bad caps around the VRM, but it sounded too hysterical, even by my lack of standards....:D
 
I am at home and checked out the fan speed and vcore

Fan on cpu 2061 rpm
Vcore 1.280

There are other voltages listed around vcore: 12.091, 5.064, 3.296 - in case I am quoting the wrong thing.
 
No, "Vcore" is what I was looking for.

1.280 at idle seems high. I'm comparing that to my i3-530 Sandy Bridge (dual core w/ Hyper threading 73 watts TDP), which idles at about 1 volt.

That said, I'm not familiar with the quad cores CPUs at all.

So, wait for a second opinion on this.

You could also check CPU usage w/ task manager, to determine what percentage of CPU is being used. You could have a hanging app/ malware, that is pushing temp / CPU usage / Vcore up. (They're all interdependent).

All your other voltages are practically spot on. (+12.0 V, +5.0 V & +3.3 V are the nominals).
 
I am not going into the OS at the moment. When the PC boots I hit del and use the hardware monitor to see these values. The machine would overheat to shut down in a couple of minutes, so I don't want to bother to go in to the OS.

This is all really strange. The heatsink/fans push pins are engaged on the CPU as I can see them from the other side, coming through the Mobo. The thermal paste is sufficient but not spilling over. The fan is working OK and registering a high rpm. The Vcore is high, but is that any reason for the temperature to go towards 80 in a couple of minutes?

I wonder if all of this means that the heatsink/fan is actually broken in some way so that the push pins engage but actually the heatsink isn't sitting properly on the CPU. That would be wierd as it seems to be flat. But it would account for the overheating.

Otherwise, what could be broken to account for this?

Jeepers - it could be the temperature sensor. How could I test that?

Thanks for your suggestions and comments.
 
As far as checking the temp goes to rule out sensor problems, you're going to have to do as Clifford suggested, and touch the cooler right before the machine is tentatively going to shut down.
Or, for the less adventurous, a non contact infrared thermometer should work.

I found this HSF at Newegg: http://www.newegg.com/Product/Product.aspx?Item=N82E16835233081

It has a 92mm fan so you stand at least a decent chance, of being able to jack it into your case.
So, 25 bucks w/ free shipping, would get you on your way to a process of elimination showdown with your computer. Either Newegg or Zigmatek should have full size specs to guide your decision.

(I'm guessing your prebuilt has a fairly small case, (???)).

The temp sensor I can't offer any help with that. I you find that's bad, you may have to over ride the warnings in BIOS to keep the machine running, assuming it allows you to do so.

What, (I think), you failed to do is post the CPU temps from BIOS. Soooo.....?
 
Hi

I will touch the cooler before shut down and let you know the result.

My built-it-yourself has a big case with two fans sucking air in from the top, one from the front, one from the side and one at the back sucking air out.

That is a beautiful heatsink you linked to. I am in South Asia, and will have to see what they have here.

The temps - when I get into bios they are already above 50 on the CPU and keep going up a degree every second or two until they get to 78 when the machine shuts down. I don't let it go that far now. The MoBo temperature is 33.
 
The heat sink is not hot even when the CPU is showing mid 70s. Strange.

I don't think the sensor is off, as gradually increasing to 80 would be weird behaviour. I can't guess what the problem is.
 
The temps are rising to quickly. And if the heatsink is not hot when you touch it, that could mean only one thing. This leads me to believe, the heatsink is not thermally bonded to the CPU.

Pull the heatsink and check for anything that could keep it from seating properly. Check the heatsink for clearance all the way around the CPU socket.

I don't honestly know the answer to your question, about the age of thermal compound. I think the compound is still good as long as it is not dried up. This would be the case with practically any glue or sealant.
 
Each time I take off the cooler I can see the round shape of the thermal grease on top of the CPU where it has spread out. As I said, I change it each time. That says the cooler is pushing down on top of the blob of grease and spreading it out.

The Arctic silver 5 out of my tube is still silver coloured gunk. It should be working in some fashion, even if it is years old. But perhaps I've slipped up here and used one from 2009 (say) which has lost its properties. Something is going wrong.

I will also have to try using even less grease, in case that is the problem.

I have tried to see if the heat sink isn't seated properly but I can't judge it that finely. Perhaps it is sitting a fraction of a mm too high. But then how could it spread the AC5 out to the whole surface from a tiny blob?

I will squish out more of the AS5 to see if the stuff further in has better conductive properties. Almost certainly a waste of time. But there is nothing obvious at this stage.

You have been very helpful. Thanks for sharing your suggestions.
 
I have tried to see if the heat sink isn't seated properly but I can't judge it that finely. Perhaps it is sitting a fraction of a mm too high. But then how could it spread the AC5 out to the whole surface from a tiny blob?
Thats a good point.

What you are describing does not make sense to me. The heatsink alone should be enough to keep the CPU from over-heating that quickly. Without a fan it should still take a few minutes, before getting too hot.

I wish you had another CPU to try for a few minutes.
 
It's possibly that one of the mounting ears on the HSF is bent. It wouldn't take much to throw it off center. Even a small drop of bump could be enough to damage it.

I think you need another HSF. Even another stock unit to test the system with.

Ostensibly, Arctic Silver should know more about their thermal compound than me. But, I still won't use their method of applying it. I prefer to spread the compound with a razor blade, over the entire surface.

Vcore is still a bit high, (IMO), for a stock HSF.

Has "speed step" been disabled?

In any case, while a CPU failure is quite rare, (especially with a CPU running at stock frequently), it is possible. However, you do have to substitute the HSF, and work out any issues with your thermal paste application technique, before you can confirm that diagnosis.

 
Back