H100i v2 early failure signs or thermal paste?

Hi folks,

I've noticed over the past couple of months, the effectiveness of my H100i v2 has started to slightly diminish. I have always run a quiet fan / quiet pump profile with a i5-6600K (stock voltage) without any issues over the past 2.5 years since I setup my build.

Now over the past couple of months when my CPU is put under load (call it 70%+), quiet pump mode is no longer sufficient to dissipate heat and temps on the cooler will climb to 40-42 degrees, triggering full fan speed when cooler temps hit 40 degrees. The good news is when the CPU load reduces itself, the temps fall rather quickly back to around 30-32 degrees idle so the cooler is still doing its job, just at about 80% of what it used to when under load.

Switching the pump to extreme mode provides sufficient cooling to maintain a pump temperature of 35-37 degrees (for now) under heavier CPU load, however I'm not sure how long this will last.

I will also mention ambient room temperature is typically 18-20 degrees and my CPU temps remain around 65 degrees under load (no concerns there) even when the H100i v2 is at 40, so this is very much at the margin of not being handle load and only in quiet pump mode. When the system is idle, the cooler temp hovers around 30 degrees.

Given the above, anyone experienced this before? Is this a sign my cooler is starting to lose its effectiveness and cooling properties after only 2.5 years, or could this be a simple case of thermal paste wear that needs re-applying?

I've also attached a sample load chart showing the cooler at 40.5 degrees when the CPU was under heavier load for a few minutes (cooler pump in quiet profile). Appreciate your thoughts.

Thanks

The fact that the temps drop quickly is actually a good sign. And the cooler is still absorbing the heat from the CPU.

So ... any other changes in your system lately? Perhaps ambient is a bit warmer? New GPU? Does this happen with a CPU-only load (e.g. from Prime95) or do you see this in a mix CPU/GPU load (e.g. gaming)?

I don't really like that ambient to coolant temp differential. I can also see a rather sharp increase in the coolant temp mini-graph. iCUE leaves something to be desired with the accuracy of the tiny graphs, but that is a steep change. The noticeable difference in pump settings is also a warning. Most people might see 1C of difference, maybe 2C if you set up extreme conditions. On your CPU, it should be pretty much zero difference. This suggests there likely is some kind of flow restriction building, as you have already theorized.

Unfortunately, this seems to be the most common complaint on that specific series of coolers. You can contact Corsair Tech support through the ticket system and see if they will RMA the cooler. I don't know if they will readily replace in anticipation of a more serious failing, but you can certainly approach it that way. In your ticket, focus on the ambient temp to coolant temp differential, both idle and load. +10-12C above at idle and +20C at load are definitely out of bounds. You may need to dissuade them off contributing factors like GPU heat or general case management. A quick and easy CPU only stress test like Intel XTU or the tool in CPU-Z that confirms the results should be convincing enough.

The other option is to move on. They will give you a brand new H1100i v2, but this fate seems to be the most common end. You will need to pay shipping to Corsair from wherever you are and you might want to factor that against the cost of a new model. I really like the new Pro series. It addressed the faults with your current line and has proved to be extremely reliable thus far. No one has reported anything like your current condition on the replacement series and there have been only a handful of other reported failures in the last 16 months since it launched. You could also do both, if you had a need for a backup cooler or had another purpose in mind.

The fact that the temps drop quickly is actually a good sign. And the cooler is still absorbing the heat from the CPU.

So ... any other changes in your system lately? Perhaps ambient is a bit warmer? New GPU? Does this happen with a CPU-only load (e.g. from Prime95) or do you see this in a mix CPU/GPU load (e.g. gaming)?

Yes and no with respect to temp fluctuations. Good that it drops quickly when CPU is not under load, bad that it increases rapidly when CPU is under load.

Absolutely no changes to the system or its environment since it was built 2.5 years ago. Ambient is up about 2 degrees from about a month ago, call it 20-21 degrees, but nowhere near 30+. GPU temps are not impacted by CPU load. Case is a very free-flowing Fractal R5 Mid-tower.

Yes this is a CPU-isolated problem. Have a look at the attached, I just performed a quick test run of CPU load ONLY (no GPU activation).

1) Pre-CPU temp test (idle conditions) - Cooler 30 degrees / Core 1 CPU mid-30's / GPU 54 degrees

2) Put CPU under load at the 70-80% load level and hold at these levels for approximately 10 minutes. After the ~10 minutes, Cooler is now 39 degrees (quiet fans / quiet pump modes) / CPU 50 degrees / GPU holds 53-54

3) Return computer to idle conditions and after 3 minutes of idle, Cooler is 36 degrees / CPU 43 degrees / GPU holds 54-ish

4) Idle conditions continue and after 6 minutes since stopping CPU load, Cooler is 32 degrees / CPU back to mid-30's / GPU remains 54-ish

What do you think DevBiker, is this consistent with early pump failure symptoms and potentially RMA-eligible?

Thanks for your comments c-attack, I think the above quick test is ample support this is CPU-isolated and indicates a problem with the cooling abilities of the cooler at this point in time. Good to hear you and others have had success with another product line.

The tricky bit is +9C is not completely out of bounds for a CPU stress test on a 240mm radiator for general discussion. I do think it is high for your CPU and specifications, but we are talking about fine margins and a few degrees - typical of early stages of a blockage or flow problem. The two parts that bother me are the high coolant temp at idle compared to the case ambient and the different coolant temp between 2000 and ~2900 on the pump.

iCUE is terrible at reading Asus' native board values, but most likely the 21C is your motherboard temp sensor. See if you can confirm that with another program. Most things like HWiNFO, HWMon, AIDA, etc. try to poll the same sources as iCUE and give back bad readings. Check the Asustek values in iCUE (most will be gibberish), then quit iCUE and its services. Load up another monitor. Any of the above will give you the Asus motherboard temp sensor. I think this 21C temp sensor vs the 30C coolant temp at idle is a strong piece of evidence. While it is typical to idle +4-7C above your room temp, at idle with standard Intel power saving features (EIST, C-states, etc) that coolant temp should be within 1-2C of the motherboard temp. +9C is way off and suggests the cooler cannot get rid of its heat efficiently even when load is minimal. This is a classic sign of flow restriction.

The other thing is the coolant temp differential between the two pump modes. On a 95 TDP CPU, this should never matter. A 3-4C difference in coolant temp at the flick of the switch suggests the coolant is having trouble making the rounds. It is these two pieces I would focus on in the ticket. Your heat up and cool down are still close enough to normal that not everyone will see the issue right away. The elevation of the entire temperature is more telling of the current state of affairs.

The other thing that I'd like to eliminate is environmental - we've seen (multiple) cases where the case builds up heat at the top. This will also increase the coolant temperature. What case are you running? Do you have an accurate, clear way to measure the internal case temperature? What happens if you remove the side of the case and any possible obstructions to the radiator exhaust? Finally, have you cleaned the radiator anytime recently?

I don't think it's a case where repasting will do anything. That doesn't seem to be the issue as the radiator is warming up under load. While I'm leaning towards this being early signs of failure, I'd like to make sure that we eliminate all other possible scenarios before we come to that conclusion.

The other thing that I'd like to eliminate is environmental - we've seen (multiple) cases where the case builds up heat at the top. This will also increase the coolant temperature. What case are you running? Do you have an accurate, clear way to measure the internal case temperature? What happens if you remove the side of the case and any possible obstructions to the radiator exhaust? Finally, have you cleaned the radiator anytime recently?

First and foremost, my case: Fractal R5 mid-tower, 1 stock intake mounted upper front, 1 stock exhaust mounted upper back, cooler mounted to top, case is filled with vents all around (front, top, bottom, rear), so air restriction is not a big issue.

Case was cleaned, including vents, radiator, and dust filters, about a week ago (few days before I started this thread) to see if that would help. It did not.

This took some time to try and get this as accurate as possible, but I ran three different load tests by using a thermometer that was placed both inside and outside of the case. My only caveat to this is in test #1 below, I had to rest it on the GPU plate which was warm and transferred heat merely by contact (whereas CPU + cooler are not touching GPU plate, and PSU is more neutral temp), but the results were in-line with the other tests. Here's what was done:

Scenario #1: Evening / Case Closed

Ambient Room Temp: 18.0C

Idle Status:

Internal Case Temp (thermometer resting on PSU): 21.4C

Internal Case Temp (thermometer resting on GPU): 24.0C

Cooler Temp: 27.7C

15 min CPU Stress Test Result:

Internal Case Temp (thermometer resting on PSU): 22.1C

Internal Case Temp (thermometer resting on GPU): 24.0C (no change)

Cooler Temp: 39.0C

Scenario #2: Morning / Case Closed

Ambient Room Temp: 17.2C

Idle:

Internal Case Temp (thermometer resting on PSU): 20.3C

Cooler Temp: 26.9C

25 min CPU Stress Test Result:

Internal Case Temp (thermometer resting on PSU): 21.0C

Cooler Temp: 40.2C

Scenario #3: Afternoon / Case Open (side panel removed)

Ambient Room Temp: 17.8C

Idle:

Internal Case Temp (thermometer resting on PSU): 19.6C

Cooler Temp: 26.6C

25 min CPU Stress Test Result:

Internal Case Temp (thermometer resting on PSU): 19.3C

Cooler Temp: 39.1C

Takeaways:

1) In all tests, the internal case temp did not fluctuate much (no more than half a degree) while the Cooler temps ran up to 39-40C each time.

2) The move from 27C or 30C up to 39C was quick (~8-10 minutes), however the move from 39C to 40C took an additional 10-15 minutes to achieve.

2) GPU temp did not change in the 53C +/- 2C degrees territory, although this would contribute to additional internal case heating when activated during gaming. GPU was purposely left out of the equation to isolate problem to CPU cooling for this testing.

3) Cooler did remain about 1C lower with the side panel removed during the stress test, however was still considered high at 39C.

The tricky bit is +9C is not completely out of bounds for a CPU stress test on a 240mm radiator for general discussion. I do think it is high for your CPU and specifications, but we are talking about fine margins and a few degrees - typical of early stages of a blockage or flow problem.

I think it's safe to say the Cooler is working and still doing it's job, just not to the same degree as out of the box. It's like one of those used car parts that is showing signs of wear and needs replacing soon, but you could probably still get by for 6-12 months easy in current condition.