Jump to content
Corsair Community

Do I have a faulty coolant temp sensor? (H100i RGB Platinum SE)


subn3t
 Share

Recommended Posts

I've got a confusing situation with my new H100i RGB Platinum SE where the coolant seems to run hotter than the CPU itself.

 

I zip-tied a temperature sensor to the inside of the case just as a sanity check, and here are some numbers I'm seeing:

 

Case temp: 26.5

Coolant temp: 41.5

CPU Avg: 38

 

hVpq723.png

 

After 10 minutes of AIDA64 stress test:

Case temp: 28

Coolant temp: 46

CPU Avg: 75

 

BvRXcHP.png

 

From what I can see, it looks like my case temp is about normal, and it looks like my CPU is about normal, if a little hot (though it's a 9900K, and they run hot).

 

The coolant temp for the load readings strikes me as high, but the thing that really gets me is the sensor giving me 41.5 at idle. This causes the built-in fan profiles to be useless. I know I can make my own custom profiles, but given that this is a brand new unit, the coolant being 15C hotter than the ambient temperature makes me suspect that something is "off".

 

Any ideas?

 

Pddcwn3h.jpg

Edited by subn3t
Link to comment
Share on other sites

You're right, it's not normal and it is a bit perplexing.

 

The CPU core temp will occasionally dip below the coolant temp at idle. This is the standard power saving features cutting a lot (or all) voltage to that core, although usually only momentarily. While we say coolant temp is the minimum possible CPU temp, that really on exists on the surface of the CPU. As you move away from the contact point, temperatures shift. Down when no voltage. Up when the pins apply it from the other side. You typically see this more frequently after load when the coolant is warm, but I think in this instance the high idle coolant temp is making it more obvious. This is not a problem per se.

 

41C coolant temperature is too warm for both your case and room temp. That is about where I would expect your max load stress test coolant levels or high combo loads. This could suggest a physical problem with the cooler or an environmental problem. We can look at both.

 

1) Physical cooler problem - the flow is restricted somehow. We can see the positive pump speed, so it's not that. Some type of obstruction is preventing coolant from making the expected number of round trips. We have not seen this on any of the Platinum or Pro coolers as of yet. A more probably explanation would be a rather pesky air bubble causing the same flow issue at the CPU (and coolant sensor). You can try tipping the case back and forth (while pump is running) or cycling the pump from low to high to low to high (2000 to 3000 rpm) every 10 seconds to create a disturbance that might push it out.

 

2) Environmental or heat management issue - While the above is possible, I would expect your load coolant temp to be horrific with an abnormal coolant delta. Instead, +5C is actually good for 10 min of combo or CPU stress AIDA. My H115i Pro would also be about +5C. So this raises some questions. Blockage or other flow obstruction would cause a large coolant delta, likely +10-20C. Instead, it is actually good. So why is the cooler retaining so much heat at idle? Based on your room temp, my predicted idle coolant temp for you would be 32-34C. The difference between the two situations is further complicated by the 2400 rpm fan speed at idle. If there was an airflow restriction like a dust filter or most other things, the 2400 rpm fan speed should power through it. That's a not a livable situation, but it should get cooler when you do that.

 

a) I am curious if the coolant temp gets worse when you drop it to a normal 900-1000 rpm idle level.

 

b) What happens if you take the glass off the top? I have not seen anyone else on a 280X be so crippled, but perhaps you are the first to notice. That cooler won't fit on the front rail with the rad + fans + GPU, will it?

 

c) Is the case temp accurate? It looks like the motherboard temp sensor is measuring 37C (although there is a lot of junk readings in the iCUE motherboard section). It's possible the room temp probe strapped to the back is reading from the rear and getting more an outside the case temp. Not sure if you have another way to verify or quit iCUE and all its services in the task manager, then try HWMonitor or HWInfo to get a clearer set of readings. Don't start taking panels off until you get an accurate baseline case temp reading for comparison. Once you verify the 37 is or is not the motherboard sensor reading, then you can use iCUE and make changes and watch coolant temps. iCUE does not get along with other monitoring programs and running them simultaneously usually results in gibberish.

 

d) In the interim, continue to use your manually created fan curves.. The 2400 rpm is too much to handle. I would probably set fixed speeds I could tolerate while working through this.

Edited by c-attack
  • Confused 1
Link to comment
Share on other sites

Also, put a temp sensor in the exhaust flow of the cooler. It should be close to the liquid temperature so that'd be a good confirmation. I don't know how accurate that device in the picture would be; ideally, you'd have a thermistor that you can put right above the radiator itself.

Keep in mind, too, that temperatures inside the case can actually vary pretty widely. Getting the sensor right in the middle of the intake airflow for the cooler would likely be more telling.

Link to comment
Share on other sites

First off, I can't thank c-attack enough. Without their thoughtful and detailed reply, I'm not sure I would have undertaken all this testing.

 

tl;dr H100i RGB Platinum SE LEDs add 6.3C to the coolant temp. IMHO, this is probably due to the proximity of the temp sensor and the LEDs, and not representative of actual increase in coolant temp.

 

* * * * * * * * * *

 

Well, I'm back after several hours of testing. The results have been very interesting.

 

I went through several stages of troubleshooting and made detailed notes at each stage. In summary, I went from 41.5C to 21.4 on the coolant temp by doing the following things:

 

  • Fixing AIO fans and pump to their maximum speeds
  • Removed top glass and dust filter
  • Turned off all RGB in the system

 

Even though I have no intention of running my PC like this, I was able to determine that there was likely nothing fundamentally wrong with the AIO, particularly the coolant temperature sensor.

 

The lowest coolant temp I achieved tonight was 21.4, and the highest was 38.9C. The breakdown is as follows:

 

  • Difference between 600 RPM and MAX on the fans: 7.2C
  • Difference between "Balanced" and "Extreme" on the pump: 5.1C
  • H100i's RGB: 6.3C

 

YLw9r2y.jpg

 

Yep, the RGB on the AIO added a very fast 6.3C to the coolant temperature, and the temperature dropped just just as rapidly when the RGB was again switched off. The rest of the system's RGB appears to have no significant effect on temperatures.

 

Given the speed with which the coolant temp rose and fell, I suspect that the LED heat didn't actually get transferred to the coolant, but that the coolant temp sensor picks up heat from the LEDs, leading to this 6.3C discrepancy.

 

With the case reassembled, I turned off all the RGB and set the AIO fans to 1,300 RPM. (This seemed to be the sweet spot to my ears.)

 

Coolant: 28.8C, CPU Core 32C

 

I cranked the fans up to maximum and re-ran my stress tests. (Based on my data, I'll make my own fan curves. For now, I think it makes sense to stress test with max fans.)

 

Coolant: 39.4C, CPU Core 71C

 

IMHO, the 6.3C added by the AIO RGB is very significant and should be noted and/or fixed and/or accounted for in built-in fan curves.

Edited by subn3t
Link to comment
Share on other sites

 

Yep, the RGB on the AIO added a very fast 6.3C to the coolant temperature, and the temperature dropped just just as rapidly when the RGB was again switched off. The rest of the system's RGB appears to have no significant effect on temperatures.

 

 

 

That is one of weirder things I have seen. I am not sure how to account for it at this point and I don't have a Platinum cooler to compare.

 

A lot of this stuff was for diagnostic purposes and not intended to be a standard operating procedure. However, I would permanently remove the top dust filter. Radiator + dust filter is never a good combination. When you add in a glass panel too it's going to put a strain on airflow from idle to load. Most of your dust will be outside on the exterior anyway.

Link to comment
Share on other sites

That is one of weirder things I have seen. I am not sure how to account for it at this point and I don't have a Platinum cooler to compare.

 

Well I just tested turning off my LEDs and well, the coolant temp instantly dropped by 2°C, from 36.5°C to 34.2°C, in less than 2 seconds.

Turning em back on gave an instant 1.5°C increase. My CPU is currently under 100% load, all cores sitting around 52 to 60°C.

 

After some more toying around with the instant lighting colors:

Red: 36.4°C

Blue: 35.0°C

Green: 34.8°C

Yellow/Orange: 37.6°C

Light Blue: 35.7°C

White: 38.1°C(peak)

Off: 34.1°C-34.6°C

 

Edit: Adjusting brightness also affects these temps, lowest setting is a minimal increase

 

These are instant temp jumps & drops happening in when changing the colors.

 

At first glance this doesn't seem to affect my CPU temps under load. But still if this cooler is setup in a default config, where the fans react to coolant temp, then yea this is not ideal.

 

 

H100i Plat firmware: 1.00.33

iCUE version: 3.13.94

 

Off

xR9oIDN.png

 

On

j0IXMhO.png

Edited by Frankz
Link to comment
Share on other sites

This came up on the beta team - a couple of users (Zotty, for one) confirmed that their coolers did this as well. Personally, mine does not. I don't know why mine doesn't ... but it doesn't. So it would appear to affect some coolers and not others.
Link to comment
Share on other sites

Let us know if you need sys specs to check what is happening there.

 

I am running Same CPU as you but H115i Platinum...... I have found its the Center 4 LED's that give the biggest jump in temp with the middle top one causing the biggest jump in temps.

 

I am 100% sure there is no actual increase in Liquid Temps... I personally think its the LED's warming up the area where the temperature probe is......

Link to comment
Share on other sites

I am 100% sure there is no actual increase in Liquid Temps... I personally think its the LED's warming up the area where the temperature probe is......

 

 

I agree. It just isn't possible, unless you've opted for the 150W D-RGB edition. The most likely explanation is an exposed sensor too close to a particular section of lighting. With any temp sensor, it you expose the sensor tip end against something, you get strong reactions.

Link to comment
Share on other sites

  • 2 weeks later...

Have a problem with my h100i rgb platinum and found this thread. Sorry for my english.

 

So I bought this in dec 2018, and til about one month ago, my idle water temp when starting the PC, was about 24c and when playing fortnite about 43-47c.

 

My room temp is the same and nothing else has change. But one month ago I started to have idle temp of 30c+ when starting the pc. And 53c+, often 56-57c when playing fortnite.

 

Tested now to turn of all lightning on h100i rgb platinum and now my watertemps is the same as before the ”problem” started. Now I have 24c idle temp when starting computer. If I turn on the lights, booom 30c+. Turns off the light and it shows 24c again.

 

I have always had default lightning. So nothing changed there. Only thing i have Done is that I updated iCUE one month ago. Have not moved the pc either. So this problem is very strange. Only wanted to ad my info to this.

Link to comment
Share on other sites

just remember guys.. this is not your liquid heating up... if you watch your CPU package there will be no change.. most likely this is an LED heating the area around its self which is likely to be next to the temp probe in the pump....

 

my H115i Platinum also does it by the way... my H150i Pro does not

Link to comment
Share on other sites

Since the temperature change is not actual, there is no need to be alarmed. However, it does make using the three preset curves a bit pointless. Even without this issue, I encourage everyone to make there own anyway. You will need to shift your scale up a bit. Figure out the idle resting coolant temp with your preferred LED effect. Load rise is going to typically be +4-8C on top of that, depending on CPU model and any additional heat contributors (GPU, case layout, etc.). Key the fans in check. Nobody needs to be running 2000 rpm on a 240mm or 280mm cooler. Set it to your noise comfort level. If you think the CPU temp is too warm, you can bump fan speed up another 100-200, but keep in mind that may only bring a 1C reduction. Usually not worth the trade in noise.
Link to comment
Share on other sites

Ok. That's what I'll do. Work on setting a custom fan curve. I think the big question then is what is too warm for the liquid temp to be? I know I was told when I was running a custom loop that most pumps and such couldn't take temps over 60 C.
Link to comment
Share on other sites

You will hit the cpu limit long before you reach a physical temperature limit on the rubber and metal. Most manufacturers list 55-70C as a coolant limit. Anytime you reach 50C you should ask questions, although I would not consider it a safety issue yet.
Link to comment
Share on other sites

So after playing a game for about an hour with my custom fan curve my cpu temps were in the mid 50’s on my CPU. According to icue my water temp was 48. But that’s with leds on. So I’m figuring it was more like 43. Not sure what to think about this.
Link to comment
Share on other sites

So after playing a game for about an hour with my custom fan curve my cpu temps were in the mid 50’s on my CPU. According to icue my water temp was 48. But that’s with leds on. So I’m figuring it was more like 43. Not sure what to think about this.

 

43 is a bit warm but not dangerously so.

 

Question: do you have the radiator in the top and configured as exhaust?

Link to comment
Share on other sites

 Share

×
×
  • Create New...