Jump to content
Corsair Community

High temps w/ H100 and i7-2600k; radiator not distributing heat?


mattw

Recommended Posts

I have an i7-2600k, testing with small-fft torture prime95 at *stock* speeds. H100 mounted in a 500R corsair case; mobo is a P8Z68-v pro 3 from asus.

 

The asus mobo sensor tends to be ~10 degrees south of RealTemp. When I fire up prime95 from a purely cold boot, I almost immediately get ~65c temps as read by realtemp. If I leave it running 20 minutes, temps will climb to ~80-82c in realtemp on the hottest core (78-79 on the coolest).

 

Friends who recommended the H100 to me are running overclocked CPUs at 100% load at 60c. Ideally, I'd like to get the i7-2600k up to 4.4ghz (which is where the asus autotuner puts it if I allow it to OC from), and run p95 torture and stay at 80c or less.

 

So, my broad question is:

 

These numbers are obviously stupidly high based on reports from friends and tests I've seen that have benchmarked temps on the H100.

 

What could be wrong?

 

When this first happened, I decided to reseat the backplate, and clean out the stock paste and apply arctic silver 5, so I carefully and thoroughly cleaned the sink surface on the H100 and the cpu, removed and reseated the backplate, and reapplied the H100 being very careful to tighten evenly.

 

When I retested, results are very similar. I fared perhaps 1-2c better than before my reseat and reapplication of thermal compound.

 

This concerns me: the radiator isn't getting hot, or even warm, anywhere except the very edge where the hoses connect. The hoses themselves get hot (quite hot as the temp reaches 70-80c), as does the very edge of the radiator. The radiator obviously has those 2 smooth blocks of metal with a large radiator "grill" in the middle. Only a tiny bit of warmth is making it onto the grill, and only then just barely, right where it connects to the edge (the solid block part of the radiator).

 

So aside from the general "What could be wrong?" question: can someone who is using an H100 and pushing their system for an extended benchmark tell me: do you notice heat spreading along the radiator? I have the feeling maybe there's a manufacturing defect in the actual radiator part, and so while the pump is working fine, it's not pulling enough heat into the main grill body of the radiator, so it ends up recirculating water that is too hot and continues to let the temp climb. (Remember, my temps slowly climb throughout a 30-minute run of Prime95)

 

Any help is appreciated. I was going to call support, but they're M-F. :/

Link to comment
Share on other sites

I suggest you verify the pumps speed, do you have the three pin fan connector with one wire from the pump connected to a fan header? The pump speed should be ~2000 RPM.

 

The radiator never really gets warm in my experience, even with a H60, whose radiator is half the size. But, only the inlet hose should be warm, if both hoses are close to the same temperature, say in the middle of each hose, there could be a problem with the radiator. The fluid returning to the CPU should be noticeably cooler than fluid coming into the radiator.

 

You should also check if the pump is not loose due to a back plate issue seen with some boards. If you can turn or lift the pump off the CPU even a bit (please lift the pump from the bottom, not the top portion with the fan controller!) it is not firmly attached. Did the TIM imprint look good when you re-applied TIM?

 

When running Prime95, an i7-2600K will get hot very quickly with any of the H-series coolers, particularly if OC'd. They are not in the same league as a full blown water cooling system with a high volume pump and 1/2 liter or more of fluid, which can draw heat away from a CPU quicker than virtually all other coolers. Once Prime95 is stopped, the CPU temp should drop almost to idle temps in a short time, and then diminish to idle temps in a minute or so. BTW, what do the radiators owned by your friends feel like when the CPU is stress tested?

 

A 4.4GHz OC is good but not so extreme it should cause excessive CPU temps with a good cooler. My i7-2600K does not reach the low 70's C running Prime95 until the OC is 4.6 - 4.8GHz. When you OC, I recommend not setting manual CPU and other voltages, or enabling other extra voltage settings, such as PLL OV. I've had mine at 5.0GHz with everything on Auto and only small tweaks to the voltage regulators. OCing Sandy Bridge CPUs is very different than the i7-900 series CPUs, and much easier.

 

Yes, the ASUS temp software is off (low) by ~10C, as I have ranted about here many times. :rant2:

Link to comment
Share on other sites

I have the feeling maybe there's a manufacturing defect in the actual radiator part, and so while the pump is working fine, it's not pulling enough heat into the main grill body of the radiator, so it ends up recirculating water that is too hot and continues to let the temp climb. (Remember, my temps slowly climb throughout a 30-minute run of Prime95)

Back on topic...This would be impossible due to the way they are manufactured. All the tubes in the Radiator would have to be closed shut. If the pump is running it is definitely circulating through the rad.

What are the pumps RPM's being reported in your BIOS? I'm wondering if it is just not running at full speed...

Link to comment
Share on other sites

  • 2 weeks later...
I suggest you verify the pumps speed, do you have the three pin fan connector with one wire from the pump connected to a fan header? The pump speed should be ~2000 RPM.

 

It is. I do have the 3-pin connected, it's ~2k.

 

You should also check if the pump is not loose due to a back plate issue seen with some boards. If you can turn or lift the pump off the CPU even a bit (please lift the pump from the bottom, not the top portion with the fan controller!) it is not firmly attached. Did the TIM imprint look good when you re-applied TIM?

 

It seemed to have an extremely tight fit. The TIM looked "feathered", but it may have just been from pulling it off.

 

When running Prime95, an i7-2600K will get hot very quickly with any of the H-series coolers, particularly if OC'd. They are not in the same league as a full blown water cooling system with a high volume pump and 1/2 liter or more of fluid, which can draw heat away from a CPU quicker than virtually all other coolers. Once Prime95 is stopped, the CPU temp should drop almost to idle temps in a short time, and then diminish to idle temps in a minute or so. BTW, what do the radiators owned by your friends feel like when the CPU is stress tested?

 

I asked them after posting here, and they confirmed for me that their radiators also remain cool.

 

A 4.4GHz OC is good but not so extreme it should cause excessive CPU temps with a good cooler. My i7-2600K does not reach the low 70's C running Prime95 until the OC is 4.6 - 4.8GHz. When you OC, I recommend not setting manual CPU and other voltages, or enabling other extra voltage settings, such as PLL OV. I've had mine at 5.0GHz with everything on Auto and only small tweaks to the voltage regulators. OCing Sandy Bridge CPUs is very different than the i7-900 series CPUs, and much easier.

 

And this is what friends report. They describe getting ~60degree temps with H100s at significantly higher clocks. (4.4-5.0; I think the guy with the 5.0ghz was on an i970)

 

But, only the inlet hose should be warm, if both hoses are close to the same temperature, say in the middle of each hose, there could be a problem with the radiator. The fluid returning to the CPU should be noticeably cooler than fluid coming into the radiator.

 

Interesting. I'm on the road, but my recollection is that both hoses were close to the same temp in the middle. One may have been hotter, but both were quite hot under load.

Link to comment
Share on other sites

mattw - did you manage to solve this?

 

I'm having a very similar problem. Same motherboard, basically the same chip and a similar case with the H100. I overclocked to 4.8GHz just changing the multiplier and fixing the voltage. At 1.35v it was almost stable at 100%, but the temp was hitting 81 degrees, which is hotter than I'm comfortable with so I didn't test any higher and dropped back th 4.6GHz. From what I've read in other places, this is much hotter than most are getting with a similar overclock and the H100.

 

I did find when I first seated the backplate that it was a bit loose and I wonder whether that is part of the problem for me. I'm planning to reseat the backplate with some washers and then apply some new thermal compound and reseat the cooler.

Link to comment
Share on other sites

  • 1 month later...

I did find when I first seated the backplate that it was a bit loose and I wonder whether that is part of the problem for me. I'm planning to reseat the backplate with some washers and then apply some new thermal compound and reseat the cooler.

 

I have a Corsair H100, cant say it hasn't been a problem. I wont get into it now I'm sure every one knows the issues. I was getting hight temps with my cpu/board (i7 2600k / P8Z68-V LX)as well. I upgraded from a P6T i7 920 D0. Didnt want to but that speaks to the issues I mentioned earlier. wont get into it now.

 

Ok so how I fixed it was simple and I hope Corsair starts to include something like this in the h100 packaging soon. You need a spacer behind the cpu back plate and the cooler mounting bracket. Plastic is ideal but I used Aluminum tape folded over till it was the right thickness. I cut it to the size of the cpu back-plate and was sure no contact to any solder points was made. Once the spacer was in place I am getting around 52c under full load @ 4.4GHZ. Find any spacer you like and install it. Problem solved.

 

JaS

Link to comment
Share on other sites

mattw - did you manage to solve this?

 

I'm having a very similar problem. Same motherboard, basically the same chip and a similar case with the H100. I overclocked to 4.8GHz just changing the multiplier and fixing the voltage. At 1.35v it was almost stable at 100%, but the temp was hitting 81 degrees, which is hotter than I'm comfortable with so I didn't test any higher and dropped back th 4.6GHz. From what I've read in other places, this is much hotter than most are getting with a similar overclock and the H100.

 

I did find when I first seated the backplate that it was a bit loose and I wonder whether that is part of the problem for me. I'm planning to reseat the backplate with some washers and then apply some new thermal compound and reseat the cooler.

 

The backplate is meant to be loose, don't put washers in there! The back plate gets tightened when you attach the coolblock. If you put washers in there, you'll put to much strain on your cpu\socket and can possibly break your motherboard!

 

The problems you guys are describing could be caused by an air bubble that is 'trapped' close to the inlet of the fin-part of the radiator. This can prevent coolant from entering the radiator. The inlet-chamber and the outlet-chamber of the rad have a 'shortcut' inbetween them. With a bubble 'blocking' the inlet of the fin-area of the rad that could mean that hot coolant from the cpu gets transported to the inlet-chamber, via the shortcut to the outlet-chamber and back to the cpu again, without only a very small amount of coolant flowing through the whole radiator. This could very well explain the high temperatures. Try shaking the rad while holding it vertically (upright) with the hoses pointing downwards, so all the air can travel up into the fin-area of the rad, and see if that helps

Link to comment
Share on other sites

The backplate is meant to be loose, don't put washers in there! The back plate gets tightened when you attach the coolblock. If you put washers in there, you'll put to much strain on your cpu\socket and can possibly break your motherboard!

 

It was flush on my P6t when I had it installed. Moving it over to the P8Z68-V LX it was not. The CPU backplate on the P8Z68-V LX is thinner then on the P6T so when the mounting screws were installed it was loose. I guess the metal CPU backplate of a socket B is thicker then the one on the H2.Or maybe its on Asus that it's loose. IDK but adding the spacer fixed the temp issues.

 

JaS

Link to comment
Share on other sites

The backplate is meant to be loose, don't put washers in there! The back plate gets tightened when you attach the coolblock. If you put washers in there, you'll put to much strain on your cpu\socket and can possibly break your motherboard!

This is correct to a point. We have seen several MB that seem to be made from a thinner PCB ,or a difference in the socket thickness, than most and does indeed have some slack that needs to be shimmed to get the pumpblock to tighten down correctly. Most use plastic or rubber washers .

Link to comment
Share on other sites

Oooh! Oooh! Can we talk about religion next? :p:

 

OMGosh, how did I miss this one!! :naughty:

 

Now at first I was thinking that Corsair was mixing up the shipping up their H-series coolers after they were blessed for the local religion. For example, the ones blessed by a Catholic bishop were sent to Japan (Shinto, Buddhist), and those blessed by a Muslim Cleric were sent to the US and Europe (mostly Christian and Jewish), etc.

 

Then I realized that the US Congress mandated that Corsair must make their products strictly secular (as in CPU-agnostic), which angered the Gods (particularly Neptune), which resulted in some overall bad mojo. :evil:

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...