Erratic QL Fan light pulsing

[Salsiccia]

Hi guys,

I have a very weird scenario going on here !

I have a newly purchased H150i Elite Capellix in my new build.
I swapped the ML-120 fans that came with it for some QL120s I had as I wanted the extra LEDs that the QLs have over the 8 LEDs in the MLs.

I have just noticed that one of the QL fans pulses erratically whenever I have a changing light pattern on the pump head.
If I have no pattern on the pump head then all 3 fans are perfect. Either in a Static colour or a changing pattern, doesn't matter. For now I have them in static white for debugging purposes.

As soon as I enable some kind of changing light pattern (Watercolour spectrum, Light Shift, any of the Out of the box patterns) on the pump head then one of the fans (always the same one, in Port 1 of the included controller) emits an occasional pulse of random coloured light. The pulse is less than half a second in duration, probably only a couple of milliseconds, but enough to be noticeable and this occurs roughly 6-8 times per minute.

I will do more testing in the next few days and change the port that the fan is connected to to see if I can narrow it down to a fan issue or a port issue.

I am assuming that changing the fans from the 8 LED MLs to the more numerous, (I think it might be 16 or 32 LEDs) on the QLs is legitimate, but maybe it isn't.

In case it matters, I have a further 3 QLs on the same controller, but not on a rad, just mounted underneath the GPU on the base of the case. It only seems to be this one errant fan on the rad that is the issue.

Does this sound familiar? Have I done something stupidly wrong or just been unlucky with a faulty QL? (I bought all 6 in 2 x 3 packs last week, so they are brand new)

I also have a spare QL I can swap in for testing purposes, but that's not a long term solution as the ones in play now are white to match the build, and my spares are black, but it would be a useful data point.

As always, any and all advice appreciated and welcomed !!

Sal

 

[c-attack]

Unlike the older LNPro, Commander Pro, and Lighting Node Core controllers, the Commander Core/XT has independent RGB headers.  If there is a fan problem, it should be local to one header and not ripple down the line like the others.  Swap fans 1<->2 on the RGB side and see if the issue moves with the fan.  My guess is physical fan problem, but we'll see.  

[Salsiccia]

Thank's c-attack. That is my plan to do a simple fan swap which should shed some light on it. I will get to that later today (just started work here) and will report back in a few hours with an update.

 

[Salsiccia]

Hmmmmmm, I swapped fans 1 & 2 on the RGB side as per c-attack's suggestion and the problem was not evident.

I then swapped them back again and the problem is still not evident.

Maybe the connection wasn't seated properly yesterday, or I did only notice the problem after I had dome some stress/stability testing in AIDA64 so I will repeat that stability testing now and see if the problem resurfaces.

 

[Salsiccia]

It now appears this is happening under CPU load.

I ran the Stress CPU test under System Stability in AIDA 64 and the problem surfaced, but on fans 1 and 2 this time.

CPU Package was at 72 degrees C, AIO coolant was steady at 38 degrees C.
My motherboard LED display shows CPU temp after it has finished posting and it reported 72, and iCue's CPU Package sensor reported the same so I am comfortable with the accuracy of that temp.
CPU is a new 7950X so I think the temps are OK - CPU Load was at 100% for minutes so it was definitely working hard.

At idle CPU sits at a steady 45 degrees C.

Ambient room temp is around 22 degrees.

I have stopped the CPU stress test and with just a very dim static colour on the Pump and CPU at 45 the problem is not evident.

Adding a shifting pattern to the pump (Watercolour Spectrum in this instance), CPU temp remains at 45 and fan #1 is now pulsing erratically.

Next I might try swapping fan 1 for a spare QL I am not using and repeat the stress scenario above, but after I have let the machine cool down for a few hours.

Any other thoughts or tests I can try ?

[Salsiccia]

It is now starting to look like a QL fan problem.

I disconnected the fan that was behaving erratically and connected a spre QL 120 I have and repeated the stress test with a changing pattern on the pump and no issue at all so far.

I am cycling through the different colour changing patterns and watching for any glitches.

If all goes well then I guess I am opening an RMA ticket with Corsair as I can't use my spare fan as it is black and all the others are white so I need a replacement white fan, but I can live with the black until the RMA delivers I guess - I may just replace 3 whites with 3 blacks so its not as bad as 5 and 1.

[Salsiccia]

I have now added GPU Stress from AIDA64 into the mix to push it even further:
- CPU temp has gone up to 85 degrees
- GPU Temp, Hotspot and Junction temp are all elevated as well.
- GPU Load via AIDA64 is 100%
- GPU Load via iCue is 0% with occasional spikes up to about 20%, but basically flatlining at 0%

iCue correctly recognises the card as an RTX 3070Ti, but it seems confused about the load. 

 

[c-attack]

4 hours ago, Salsiccia said:

iCue correctly recognises the card as an RTX 3070Ti, but it seems confused about the load. 

That one is a known issue with both AMD and Nvidia cards. The origin was with Windows, but other monitoring programs have made adjustments to compensate. 
 

I’d start a support ticket and swap the white QL. No reason to wait until it gets worse and we do see this at times with the QL series. However, you might try to replicate the original fan issue using a different stress test like OCCT, cpu-z bench, y-Cruncher, or even Prime set to mild-moderate settings. AIDA has a tendency to get entangled with cue. This usually shows as bad data, like impossible highs or lows but sometimes results in loss of control or internal device errors. It may or may not do the flickering, but be prepared for someone to cite the Aida conflict as a cause and address it first. 

Edited June 19, 2023 by c-attack

[Salsiccia]

11 hours ago, c-attack said:

That one is a known issue with both AMD and Nvidia cards. The origin was with Windows, but other monitoring programs have made adjustments to compensate. 

I’d start a support ticket and swap the white QL. No reason to wait until it gets worse and we do see this at times with the QL series. However, you might try to replicate the original fan issue using a different stress test like OCCT, cpu-z bench, y-Cruncher, or even Prime set to mild-moderate settings. AIDA has a tendency to get entangled with cue. This usually shows as bad data, like impossible highs or lows but sometimes results in loss of control or internal device errors. It may or may not do the flickering, but be prepared for someone to cite the Aida conflict as a cause and address it first. 

Thanks for confirming the known issue with Load. Its not a big deal for me, I will just ignore the iCue reading for the time being. I'm not really relying on the iCue dashboard at this stage because I have built an external sensor panel using AIDA64 (which is why I have been using its stress testing capability). The external panel sits on my desk right in front of me rather than being built into the case as others have done. This way I don't have to turn my head away from the screen - it works for now !

Good thought re the other stability/stress tools too. Again I only went with AIDA because it was already installed for the sensor panel work, but maybe it is better to go with a product more dedicated to that testing space rather than AIDA's overall capability.

I will kick off the support process as you suggested, although ironically when I swapped the white fan back in yesterday and repeated the testing there was no problem, so there may be an element of randomness in play. I will raise the ticket and see what Corsair come back with.

Thanks again c-attack, appreciate your input.

Sal.

[Salsiccia]

Fun fact: I installed OCCT yesterday and tested with it all day and the fan behaviour was flawless. 

So either the connecting/disconnecting process over the last few days has resulted in a cleaner connection, or it was due to an anomaly in AIDA64s stress testing.

I have raised a ticket with Corsair anyway so will see what they say, but as of right now - my problem has "gone away" and that is good. Now I am thinking about "what lurks beneath"!!!

 

[c-attack]

It's not a lock guarantee that AIDA will interfere with RGB lighting.  I think it is possible since it does have issues with the polling of the Commander Core for fan/pump speed and that in turn often causes CUE to behave strangely.  So while its something to look at, I am going to be a bit surprised if that is the genuine cause.  The real reason to pursue it so someone else doesn't slow you down later by suggesting the same and having a ready response might be helpful.  I don't typically see problems with lighting patterns when testing at 100% CPU.  However, using 100% of RAM may cause CUE to get sluggish and lighting patterns to do strange things.  I don't know which AIDA stress test combo you were running, but a quick check on my older version of AIDA gobbled up all 31.7GB of my available RAM in a few seconds.  OCCT does not use 100% of RAM unless you deliberately set it to that and there are lots of reasons not to when testing in the Windows environment.  

[Salsiccia]

Good points c-attack.

On the current version of AIDA64 that I have you can select which tests you want to run. In my scenario I only selected CPU. In past tests I selected RAM to watch the RAM, GPU to watch the GPU, but in the tests I reported above I had only selected CPU. I was trying to keep the testing as "clean" as possible.

I did note that OCCT does seem a lot more flexible in terms of what options you can give it so it is probably more reliable from that standpoint.

 

[Salsiccia]

Update on this thread:

Kudos to Corsair Support.

I raised the ticket addressing my problem and Corsair responded with requests for more data re serial numbers, proof of purchase etc and they also asked for a video of the problem - all totally reasonable requests in my book.

I replied with the proof of ownership details and serial numbers etc and promised to reinstall the problematic fan and get a video for them.

I was delayed getting the video to them, but when I checked just now (I have not yet sent the video of the problem) I saw a message saying a replacement is en route via Fedex to me.

That is what I call exemplary customer service.

I have seen some negative comments on here of other folks' experience with Corsair customer support and I cannot comment on their circumstances, but what I can comment on is that when I reached out, I found Corsair Support responsive and supportive and willing to help.

Good job Corsair, thank you for your approach.

Best,

Sal.