H60 Liquid Cooling Radiator not heating up?

TL:DR - the radiator attached to my H60 isn't heating up. One pipe is hot. Other is cold. Fan under radiator is spinning. If I put case on its side there is a noise coming from the unit attached to the CPU. The noise goes away when case is upwright. Not sure if noise is connected to radiator not heating up?

I have a i7-4770k and just installed a new 1070 Founder's edition. I have a minicase, so its a tight fit. Booted the system and was getting good graphics but horrible stutter / unstable fps. Checked temps and they were 90+ C and sometimes maxing out at 100 C. It was weird the system never shut off completely?

While installing I had to move the system fans to get to the old GPU & in the process must have loosened the cooling piece attached to the CPU (is this the pump?) because it was cold to the touch while CPU very hot. I removed the heatsink, cleaned off old thermal paste, applied new and booted up system to find a much improved 71 C BIOS reported temp which was more or less confirmed by other apps. I could also feel heat in the reattached heatsink. But still shouldn't be this high idle so I reapplied the thermal paste a second time checking to make sure the entire CPU was covered; booted up to a resting 60-65 C temp where I am still at. Get's up to 85 - 90 C gaming, but I am experiencing no instability or fps stuttering, which is weird.

I think the problem is the radiator at the end of the cooling cables is no longer getting hot. But I don't know why, so that's why I came here to ask the experts. The fan beneath the radiator is definitely spinning and so is the second attach to the side of the radiator. The power cable from the heatsink is plugged into CPU_fan on the MB and I've confirmed the rpm is 4600+ using speedfan. I tried positioning the case on different sides to see if that affected the temp and noticed the heatsink makes a rattling noise if the case is on its side and the cpu temp goes up about 5 C to a resting 70 C.

Any ideas how to get my radiator back involved in helping cool my system? Thanks for the help.

I was able to get the idle temps to around 40 C last night by removing the radiator from the case and physically manipulating it to get water flowing. This involves holding it at different angles and heights relative to the heat sink.

I reattached the radiator to the case with the system still running while on its side. Something about the angle or positioning threw off the water flow again and temps shot up. Removed the cooling unit a second time and got the water flowing but put it back in the case with the system turned off and upwright. Turned it on again and temps were back to 40.

So, physically manipulating the radiator got the water flowing and fixed my cooling problem. Does anyone know why this fixed the problem?

Most likely you have some sort of partial blockage inside the cooler and this is restricting the flow of water. Normally when this is the issue, your CPU temps will seem normal at cold boot, but then slowly and steadily rise. This is in contrast to most other issues where the temps rise and fall with voltage application.

You are going to need a replacement and this issue is likely to get worse with time. Contact customer support through the official site.

Most likely you have some sort of partial blockage inside the cooler and this is restricting the flow of water. Normally when this is the issue, your CPU temps will seem normal at cold boot, but then slowly and steadily rise. This is in contrast to most other issues where the temps rise and fall with voltage application.

 

You are going to need a replacement and this issue is likely to get worse with time. Contact customer support through the official site.

I thought that may be the problem, but I was able to get water circulating again and now when I boot the computer idle temps are around 35 C and even gets down to 30 C if the ambient temperature is cool. This indicates the cooling system is working as intended, no?

I think the problem may have been air accumulated in the pump area when I installed my graphics card so when I turned it on there was not enough liquid to begin the cooling cycle?

But now I am running into a weird problem that when I boot up a game the temperature climbs to a stable 75 C, which I understand is still running hot? Much better than the 95+ C I was getting, but seems like a large jump from idle 30 - 35 C to 75 - 80 C running World of Warcraft. Once I close the game the temp almost immediately drops into the 50 C range and from there slowly falls to the 35 - 45 C range over about 5 - 10 minutes.

Any ideas what could be causing the large temperature spikes just from booting up a game?

I have done no custom BIOS configs or overclocking.

Thanks for the input.

Air bubbles tend to cause noise and while I suppose it is theoretically possible for them to slow the flow en masse, the more likely culprit is something else. More commonly the sealant used in the construction gets into the flow and literally "gums" up the flow grate coming out of the pump. When you stop, take it out, shake it, etc., you dislodge the heavier material and it settles elsewhere. Eventually, it finds its way back to the drain.

Obviously I can't see inside the unit, but this is the most common reason and does fit your detailed description of events. The coolant levels inside the unit are higher than they should be. The coolant temp serves as the baseline minimum CPU temperature, thus even when the load is ceased, the idle temp remains high until the system is able to slowly remove the heat from the water. The blockage is not complete (or consistent) or the water temp would take hours to come down. Regardless of the exact nature of the blockage, your performance should not be so inconsistent and I recommend the RMA or replacement of another kind.

Air bubbles tend to cause noise and while I suppose it is theoretically possible for them to slow the flow en masse, the more likely culprit is something else. More commonly the sealant used in the construction gets into the flow and literally "gums" up the flow grate coming out of the pump. When you stop, take it out, shake it, etc., you dislodge the heavier material and it settles elsewhere. Eventually, it finds its way back to the drain.

 

Obviously I can't see inside the unit, but this is the most common reason and does fit your detailed description of events. The coolant levels inside the unit are higher than they should be. The coolant temp serves as the baseline minimum CPU temperature, thus even when the load is ceased, the idle temp remains high until the system is able to slowly remove the heat from the water. The blockage is not complete (or consistent) or the water temp would take hours to come down. Regardless of the exact nature of the blockage, your performance should not be so inconsistent and I recommend the RMA or replacement of another kind.

When the liquid was not cycling I would hear a loud whine from the heatsink when the computer case was placed on its side while the pump was running which would go away immediately when placed upright. Now that the liquid is cycling I can place the case on its side and there is no noise. Sometimes its very slight for a split second then goes away.

My theory was that the air in the closed system accumulated in the pump area and only air pushing against the heavier liquid above it & against gravity & twists in the tube was just too much resistence to get it going. Since I have a mini case the liquid tubes are compacted more than I'd like.

I really appreciate your insight into how these systems work. When you say:

"The coolant temp serves as the baseline minimum CPU temperature, thus even when the load is ceased, the idle temp remains high until the system is able to slowly remove the heat from the water."

How long should a system take to reach its idle temperate? Today I played WOW for about an hour & when I logged out temps were running stable around 75 - 80 C. Within 15 - 20 seconds temps had fallen into the low 50's. Probably another 5 minutes or so to reach 30 - 25 C. Should it be cooling off faster than this?

I read online about i7-4770k chipset running near my temps until a 'delid' procedure was performed that resulted in a ~20 C drop in load temps, which would put me right in the sweet spot compared to where I am. Is this worth trying before I replace the entire cooling unit? Would just hate to place the blame on a defective product if it isn't actually the problem since Corsair has been so good to me with every component I've purchased thus far.

Again, much thanks for sharing your knowledge!

How long should a system take to reach its idle temperate? Today I played WOW for about an hour & when I logged out temps were running stable around 75 - 80 C. Within 15 - 20 seconds temps had fallen into the low 50's. Probably another 5 minutes or so to reach 30 - 25 C. Should it be cooling off faster than this?

That first drop from 75-80C to 50C is the voltage stepping down from your steady load Vcore to the whatever your idle voltage supply is with C-State/EIST enganged (0.60-0.80v somewhere around there). That 25-30C difference between stepped down and load I usually refer to as delta over coolant. This value is affected by CPU design, materials, and mostly your voltage levels. Some 3.3GHz stock CPU will have a fairly small delta over coolant. When you overclock and crank up the voltage, this value will increase progressively as the Vcore goes up. If your coolant temp is 30C and your CPU temps are 85C, you know you are probably at your limit. This is where your de-lidding process comes in. You are improving the thermal transfer between CPU and cold plate and that is one of the factors listed above. The Haswell series is certainly the ideal candidate for delidding, but I have never done one and there is some risk. That can be tackled in more detail later, but it is not the origin of your current problem.

Your issue is the coolant temperature is too high and thus the entire CPU temperature curve has been elevated into a zone it would not normally reach. Heat transfer across the cold plate and CPU lid is bi-directional. When the CPU cores are 60-something degrees, the heat will travel through the metal, TIM, and cold plate into the water stream. However, the reverse is true as well. When you cut the voltage to the cores, they are no longer producing heat. However, if the water flowing on the other side of the cold plate is a steady 50C, that level of heat will also travel back to the other side. Effectively, your coolant temperature becomes the minimum possible CPU core temperature. Now the coolant is in constant motion and the whole thing is constantly looking for equilibrium, so this is not an unbreakable law of the universe. You will see core temps a few degrees below the coolant temp, particularly if left on long stepped down idle. However, it is a useful approximation of how the system should behave.

In your case, your WOW delta (and this is a good one to use - heavy CPU game) appears to be +25-30C. When the coolant is at 50C, your resulting temps are 75-80C. If the coolant was down at 30-35C where it should be, then your max temps would be +25-30 on top of that, or 55-65C peak value.

Is this a brand new installation? I am going to walk myself back from the RMA request for now, although you could start the support ticket anyway. It will take them some time to respond and it would be necessary if RMA is required. 5 minutes at idle is a good time for a H60 cooler to shed 15C or so from the coolant temperature. A small cooler like a H60 takes longer than a large surface area cooler like a 280mm H110i/H115i.

There are some other possible reasons for these circumstances:

1) Your CPU watt output is overwhelming the cooler. It cannot handle the load from a 4770K. What voltage and overclock (if any) are you running right now?

2) The case environment is also very warm and this heat is being transmitted into the radiator by proximity. This can be an issue with rear mounted radiators and heavy GPU use. The GPU below the radiator heats the thing up a pot of water on the stove. Once the load stops, the radiator is able to remove the heat in a normal fashion. I ran an H80 in the rear slot on a highly overclocked 930i for a few years. My MMO big multiplayer events would create higher CPU temps than Prime 95 because of the dual heat sources. You can check this by running a CPU only benchmark after a cold boot or long period on non-use. Don't use Prime. Try something like Intel XTU. It is very mild and the core temps graphs should lay out in a smooth wave pattern. If this comes back in the 50-60C range, we know where some of the heat is coming from. (set the 'wrench' settings you display each core temp individually on the graph. This can also help detect if the TIM job went sideways or you have a very quirky CPU).

3) The entire case is in some partially enclosed area where heat cannot escape. This results in the same problem as above. You can usually use the motherboard temp sensor reading from most boards to get an approximate case temperature. Also look at some drive temps since the motherboard sensor is often close to the PCI-E slots and can scale up with GPU heat more than other parts of the case. Big gigantic hot pocket around the whole thing.

I am wondering about #2, although I still have concerns about the noises and some of the other behaviors. However, I do think we should work through it. For some reason, if the H60 is not going to work for this set-up, it would be better to know now while you have maximum return options.

To supplement my understanding of #2 and 3, can you tell me what case you are working with (it doesn't matter if it's not Corsair). Also, describe what kind (size) fans you have in the case and which direction they flow (intake vs exhaust).

This is my last post for the night and I will be tied up all day tomorrow, but I will get back to it. Perhaps someone else will have some insights in the meantime.

Quick update - your initial thought was correct. I was mistaken that the temp would go back down to 30 C after 5 minutes or so. This only happened once or twice after I would boot up the computer, run a CPU intensive app for a few minutes then go back to idle. Running the CPU at load longer then waiting for temp to go down to 30ish would take a long time, as you suggest. It would fall to 50 very quickly (temp of coolant).

I bought an identical, new H60 and installed it. I can feel the water moving through the tubes much better than on my old H60. The radiator is heating up. Idle temps are in the 20's. I max out at 50C under load.

There must have been some blockage that developed and I could momentarily loosen it, but would always return as you predicted.

One last question - do you think the curve of the cooling pipes within the case can affect potential blockages? As stated previously I have a mini case. I bought it pre-assembled and the cooling pipes were quite twisted to make them fit. With my new installation I've limited the twisting by removing the side panel and leaving it off so the pipes can protrude. Or am I just being paranoid and as long as the pipes aren't pinched it doesn't matter how they twist / turn in terms of developing blockages?

Thanks again for your help working through this issue!