High Chipset Temperature

[ej24]

Any input would be appreciated!

 

The problem PC

mobo: GA-Z87n-wifi (BIOS rev. F6)

cpu: i3-4130t (35W)

ram: crucial ballistix tactical ddr3 at 2000mhz (1.35v)

gpu: XFX RX460 2gb

ssd: 2.5in Crucial m500 240gb

OS: Windows 10 Pro

 

The Problem

Recently moved my home office PC from a small restrictive case, Lian-Li PC-Q12, to a new, larger case with more expandability, airflow and options, Lian-Li PC-Q01. Now that everything is in the new case I'm having really high chipset temps, 52C at idle, (TMPIN0 according to HWMonitor) I don't recall it being an issue until I assembled everything in the new case. I have other gigabyte Z87 and Z97 boards with TMPIN0 temps around 38C at all times. This case is larger with more airflow so that's not the issue. When I touch the chipset heatsink its almost too hot to touch so I'm almost positive TMPIN0 is the chipset temp. It's causing really weird case fan behavior too which is the main problem. The TMPIN0 reading starts out at room temp when I first boot, 22C. Leave it idle (even in power saver mode) and it goes up by 1C every 5 minutes. The moment it hits 50C my 140mm Noctua NF-A14pwm case fan instantly goes from 700rpm to 1600rpm and won't go back down until I reboot. CPU core temps are around 30-35C. GPU is around 38-40C. Chipset temp (according to TMPIN0 in HWMonitor) stays above 50C even when the case fan is going nuts. I'm not really that bothered by the chipset temps (it'll tolerate 50C no problem) so much as the fan going insane and being loud as hell. Everything seemed fine until I put it in the new case with the RX460.

 

Process of elimination... Things I've tried,

• I've tried clearing CMOS, loading optimized defaults. Still behaved the same (hot chipset and loud fan).
• The RX460 is also new, could a GPU be causing a chipset to overheat? Maybe some sort of pcie bus clock problem?
• Before the transition to the new case, I delidded my Haswell i3 (mostly to see how difficult it is before I consider doing it to my devils canyon i7, much cheaper than ruining a $300 cpu, used IC Diamond between CPU and IHS). i3 was fine after delid, also did it to my G3258 and its fine too.
• I've also tried 3 different PSU's and none make a difference.
• I've used three different CPU coolers, just in case the TMPIN0 is actually CPU temp related. There is no difference in TMPIN0 between stock intel, NH-l9i or NH-l9x65.

 

Some things I think I'll try next,

• check the thermal pad between the chipset and it's little heatsink, even though I really doubt thats the issue as the heatsink is hot, ie it's clearly conducting heat
• swap CPU's to the G3258 or an i7 I have lying around (if the issue only occurs with the delidded i3 I'll check the paste under the IHS)
• swap the GPU for the old HD7750
• somehow position a tiny 40mm fan over the chipset to see what happens?

[★ Coups ★]

Try a different temperature tool. I had this problem a few days ago, it was just one of the software I was using wasn't working correctly, much like how you explained

 

Can you also tell us the temperature under load please? 

 

Perhaps stress test and measure temps in Aida64, that is what I used for my temperature reading problem.

 

[Nord]

The new case you bought is just bad in terms of airflow.

 

I allready had a hard time finding proper pictures of that case online but if it looks somewhat like this:

(ignore the germanness, clicky on the pictures) https://www.caseking.de/lian-li-pc-q01b-mini-itx-gehaeuse-schwarz-geli-639.html

its indeed the problem.

 

From what I can see it has only intakes at the bottom of the case itself and a slim intake hole at the very bottom of the sidepanel (or is it on both panels?).

I can see no exhaust options whatsoever, no way to even mount a exhaust fan. The CPU seems to sit pretty much directly below the PSU, which is the next problem.

The CPU coolers you are using (yes all 3 of them) pull air from above them and exhaust it over the heatsink - the PSU pulls air from below it (from inside the case) and exhausts it out of the PC, they sit directly near each other, in fact the PSU sits on top of the CPU fan, so basically they fight each other for airflow.

And to further complicate the matter of airflow, before the air, which is hopefully taken in by the bottom case fan, even reaches your CPU or PSU, it needs to pass your RX480.

 

So what happens is:

Air is pulled from below by your noctua case fan. I'd say a little below half will go into the GPU, what is left goes up and both the CPU & PSU fight each other for it, literally.

Once your system is on for a while, as heat rises, your GPU (which sits below the CPU/PSU) gives a fair share of its heat upwards towards the CPU & PSU while hot air gets re-used and re-used by your CPU fan as there is neither a real way for the air to escape the case nor a proper one to take new air in, as the entire rest of the case appears to be solid sealed (apart of the back where the IO is obviously)

Like honestly if it approximately works as I described, that case is a nightmare and if you did pay more than 20$ for it, its overpriced on top of it.

 

Btw case fans are usualy "bound" to the CPU TCase temperature in BIOS, which means the noctua fan gets that loud because your CPU's heat is rising and if even with 1600RPM intake of a proper aftermarket system fan you cant seem to hold it below 50°C in idle, something is very wrong, not even budget laptops run that hot. I mean they do but with extremly silent operation, like quiet as in "is that thing even running?".

 

[ej24]

RX460 fan doesn't run until 55C and even when it does it exhausts almost completely out the rear I/O and is idling at 38C so it's producing essentially no heat. Its only blocking about half the Case Fan and even then there's several inches between them and on all sides of the GPU, the air can easily just go around the GPU. PSU is a Corsair SF450 SFX psu so it blocks nothing and isn't fighting the CPU cooler for air because it's fan is never running. CPU temps, as I said above are about 35C as its a 35W TDP cpu so it simply never even makes hot air. My CPU temps in the previous case were around 40C and the chipset was around 40C. You don't need massive 40 liter cases and 17 fans to maintain low temps. I had a 65W i7 in the PC Q12 before as well and it never went above 60C under full load.

 

As for my current set up, under load the gpu hits 60C and CPU hits 48C using Skydiver stability test (20 cycles) and in a CPU only test the CPU hit 44C using intel processor diagnostic tool. Again, I used to have an i7-4770s on this board in a smaller case and the chipset never hit these temps. The CPU cooler is positioned such that it exhausts directly across the PCH heatsink, so in this case with the i3 its blowing typically 40C air over it. That doesn't explain why at idle the PCH temp keeps rising when the CPU is only at 29-30C.

[Nord]

Intel processor diagnostic tool is not a programm that will make your CPU hit its actuall limitations in every way and hence will not make it produce its maximum amount of heat, alone the fact that Skydiver gives you higher CPU temps than a programm that is supposed to stress your CPU should tell you that. Use Aida64 or prime95 for real CPU stresstesting - though this is not relevant to your issue, just an FYI.

 

 

However I probably missed to point out the issue properly.

I explained to you what happens with the airflow from bottom to top and that heat rises (which I assume you know), so the longer your system is turned on the more heat goes up to the top of the case and has no means of escaping or being moved out, so it just stays there. This means that hot air will just stack up at the top of your case, sometimes the CPU fan will recycle some of the air, most of the time it wont. The top of the case is also where your motherboards passive cooling heatsink is located (aka the thing you are worried about). Since it is not passively cooled really as there is no airflow going on at the top of your case, it runs hotter. Further as the air is essentially "still" in its area and also hot its easily explained why it runs 10°C hotter than in the previous case. I also belive your RAM sticks will run quite hot aswell.

To sum it up, Its the location of the heatsink and the fact that you dont have a exhaust fan or any other "working" way for the air to espace the top of your case.

 

You could in fact easily test this even, remove the sidepanel and check the temp again, at the very least it should climb way slower than before.

[ej24]

The top of the case (which is aluminum) is cold to the touch so there isn't that much heat accumulating up there, otherwise the cpu would begin to overheat by sucking in its own exhaust too. It's been sitting idle all day, CPU is 37C and chipset is 50C, GPU is 39C. The case is under insane high pressure from the 140mm fan. The vents by the rear IO are gushing with exhausted air that is warm-ish. The case is ~13 liters and the fan is pushing 115m^3/hr, aka ~2000 liters per minute, so 147 case volumes per minute. That's a ton of turn over. In the interest of being thorough I'll take the panel off for a while.

19 hours ago, Nord said:

 the fact that Skydiver gives you higher CPU temps than a programm that is supposed to stress your CPU should tell you that.

Skydiver warmed the CPU up more because there was a 60C GPU right next to the CPU intake. The Intel diagnostic test (which is all I had installed at the time, I'll run aida later) leaves the GPU idle so no extra heat from below.

 

Also none of this explains the case fan behavior. It doesn't slowly ramp up to full RPM, it's instantaneous. Its 750rpm all day, the moment that chipset hits 51C its 1600rpm. GPU and CPU temps plummet to 30C when that happens and the reading for the chipset stays the same.

 

[ej24]

Update: with and without the side panel the PCH temp changed 2C. Aida cpu temp was 49C. So nothing crazy. It's weird in an open case situation that the chipset would be 52c at idle. 

 

Pulled heatsink off pch, there's no thermal paste or sticky pad but a firm black spacer that looks like graphite maybe? I put the tiniest squeeze of thermal paste in there and it made no difference. Case fan still freaks out at pch temp of 52c. A sleep and wake cycle or reboot calms it down. But it'll eventually reach 52c again... and then its blast off for the case fan

 

Also my SSD is giving weird temp readings that clearly errors (175C? I think not. Voltage spike? Perhaps), but it may be a sign of an issue with the PCH or SSD causing the PCH to act up. I say this because SATA ports and I/O like USB are pretty much the PCH's entire job. If something is wrong with the SSD or the PCH it may explain weird readings causing the PCH to overheat. It is genuinely too hot to touch even with the case open.

[Nord]

Thats interestingly odd.

 

Try to jury-rig a fan near the heatsink to directly blow on it , or just use a standard table ventilator to blow on the heatsink somewhat and see if that makes a difference. I mean it obviously should, question is just on how much so you can probably think of some solutions depending on the outcome.

There also is the possibility that the heatsink is not a heatsink but just a heatspreader and the thing is designed to stay at around 50°.

 

About what you can do with the casefan, I really havent used a GB motherboard since DDR2 times, so no idea what they can or cant do really. But looking at my own motherboards from asus, they always did offer software options at the very least since the first I3/I5/I7 series CPU mobos came out, to set custom fan curves for case & CPU fans inside windows, or to some degree, even in the bios itself.

So I suppose there should be something simular from gigabyte aswell, where you can set your casefan to tolerate "higher temps" before it ramps up, but then again I dont really see that a casefan should be connected with the heat of anything other than either TCase or core temp.. but who knows how that specific motherboard bios calculates the TCase, maybe it takes CPU IHS temp + other temps and averages them...

I mean you could also, just to rule it out, look at the TCase temp of your CPU when the fan starts to ramp up, just to rule out thats not it.

[ej24]

2 hours ago, Nord said:

Thats interestingly odd.

 

Try to jury-rig a fan near the heatsink to directly blow on it , or just use a standard table ventilator to blow on the heatsink somewhat and see if that makes a difference. I mean it obviously should, question is just on how much so you can probably think of some solutions depending on the outcome.

There also is the possibility that the heatsink is not a heatsink but just a heatspreader and the thing is designed to stay at around 50°.

 

About what you can do with the casefan, I really havent used a GB motherboard since DDR2 times, so no idea what they can or cant do really. But looking at my own motherboards from asus, they always did offer software options at the very least since the first I3/I5/I7 series CPU mobos came out, to set custom fan curves for case & CPU fans inside windows, or to some degree, even in the bios itself.

So I suppose there should be something simular from gigabyte aswell, where you can set your casefan to tolerate "higher temps" before it ramps up, but then again I dont really see that a casefan should be connected with the heat of anything other than either TCase or core temp.. but who knows how that specific motherboard bios calculates the TCase, maybe it takes CPU IHS temp + other temps and averages them...

I mean you could also, just to rule it out, look at the TCase temp of your CPU when the fan starts to ramp up, just to rule out thats not it.

I have 3 of this motherboard and a Q87 varient of this board too. The others have chipset temps of 35-38C. As for this questionable board, it doesn't matter what bios settings for the case fan are used, even set to "silent" aka 0.75x PWM it still goes full speed, instantly, when the PCH (TMPIN0) hits 52C. According to Aida64, my motherboard diode is reading a constant 50C on this particular board (even with the case completely open). My other boards are more like 38-40C even during stress testing. Again, it's very interesting that my Crucial M500 SSD, about once per hour, gives a reading of 170C. I've had probably a dozen Crucial SSD's and none have ever exhibited this behavior on other gigabyte z87 or z97 mini-itx boards. It may be the SSD is problematic and messing with the PCH? I could swap out the SSD for something else, maybe thats the issue seeing as SATA and USB are the bulk of the PCH's duties.

[Nord]

If you do suspect hardware incompatability and/or causes, you pretty much have to look at the entire PC in this case. My knowledge here is a bit rusty but from what I recall the chipset pretty much is responsible for all peripheral devices connected to the PC via the mobo IO,  direct communication with the CPU, all devices that transfer data (DVD Drives, SSD, HDD, Thumbdrive, SDCard reader ...) aswell as RAM / Chip RAM (I'm not 100% sure if its either both or just chip ram). So basically you are looking almost at the entire system.

 

At this point, your options pretty much boil down to "exchange parts until the problem goes away". So you might aswell try a different SSD, especially since you suspect it due to the incorrect temperature reads, with a clean windows install first. I would also not rule out a software "problem". For example if you would have a virtual DVD drive installed, with something mounted in it, that could potentionally cause additional data travel, hence more work for the chipset...  in my theory at least.

 

Bottom line is, if you belive the cooling is sufficient and you have other mobos, in the same or very simular scenario, that get lower temps on the chipset, your options are like I allready said, swap parts untill it goes away or deem the mobo broken and RMA it if you still can or just put up with it.

Thats pretty much all the remaining advice I can offer you.