Question Corsair AIO showing CPU Fan Spd 0 in MSI BIOS

Oct 28, 2019
18
2
15
Hey all - just built a new PC with a MEG X570 ACE and Corsair H110 240 AIO. I have the single wire 3 pin fan connection coming off the AIO connected to the CPU-Fan header as said in the directions, but in the BIOS, the Fan shows CPU fan speed 0. Two questions - should it be plugged into the Pump Header on the MB, and also, why is it reading 0. The fans are both spinning on the AIO. Also, in the iCue software from MSI, it does show both fan speeds registering. Thanks in advance for the help.
 
What cooler do you have, specifically?
IIRC, H110 is a 280mm.... H110i V2 is 240mm.

I believe the H110i V2 connect the fans to the pump, then to the single header by default.
You should be able to connect the pump to the pump header on the board (whcih should have the pump run 100%) and fans to CPU_Fan and CPU_Opt headers (or off a splitter). If you do so, do you get accurate readings in BIOS?
 
What cooler do you have, specifically?
IIRC, H110 is a 280mm.... H110i V2 is 240mm.

I believe the H110i V2 connect the fans to the pump, then to the single header by default.
You should be able to connect the pump to the pump header on the board (whcih should have the pump run 100%) and fans to CPU_Fan and CPU_Opt headers (or off a splitter). If you do so, do you get accurate readings in BIOS?
Your right sorry. It's the H100i RGB. The two fans connect to the pump. There is a three pin fan connector that is supposed to be connected to "CPU_Fan." according to the instructions . That connector only has one wire though, and it's on pin 3. Power is supplied by SATA connection to PSU.

So should I connect that three pin connector to the CPU_Fan header or the Pump header on the MB?

Also, why would I not be able to see the speed in the BIOS? Even with that one wire, it should be he tach signal - correct?
 
  • Like
Reactions: Crosslhs82x2
You are correct. That 3-pin female connector on the end of a single wire from the PUMP component is solely to send the PUMP speed signal to the CPU_FAN header. This has two functions: it gets the pump speed into the mobo where the Corsair iCue software can access it, AND then that signal also is monitored by the mobo's own tools for possible FAILURE (zero speed) of the pump. The iCue software also will monitor for failure.

Since the rad FANS in this system are powered and controlled through the pump and the USB2 cable connection that iCue uses for communication, the speed of those fans is NOT available to the mobo fan headers at all. The ONLY place that can access the speeds of the rad fans is the iCue software tool, and it may only display one fan's speed. Although the CPU_FAN header may label the speed signal it can access as the "CPU Fan" speed, that really is the PUMP speed of the AIO system. Within iCue, though, that speed should be labelled properly as the Pump Speed.

This is NOT an item to ignore. Both the mobo and the iCue utility monitor that pump speed signal for failure, and failure of the pump is a BIG problem in this cooling system. Now the ONLY way for both of those tools to read that speed signal and check it is through the CPU_FAN header. Your post says NEITHER of those two tools shows you any pump speed, indicating that the CPU_FAN header either is not receiving the signal, or that the header is faulty. First, check the connection of plug to header. Is it loose? Is it plugged in the right way? The connector has ridges down one side that MUST fit around the plastic tongue sticking up beside the header. If necessary, unplug and re-connect this a couple of times to try to improve the connection. Is the header configured NOT to pay attention to this speed of the so-called "CPU Fan"? If it is told to Ignore this speed, it will show it as zero and iCue will NOT be able to access it.

Of all those items are correct, try a temporay change of connection as a test. Move that 3-pin Pump speed connector to another unused fan header, (e.g., SYS_FAN), boot up, and check whether the header can show you the speed of the pump. If it can, the CPU_FAN header may be faulty. OR, if the pump STILL cannot send its speed signal to the mobo this way, maybe the PUMP is the faulty item.
 
  • Like
Reactions: toro1966
Paperdoc - thank you for that great and very informative reply! I apologize if I misspoke in my initial post. The iCue software IS working. It shows both fans and the pump and labels them as such.

So based on your last post, does this mean that no matter what I do, becaue of the setup of the Corsair AIO, the CPU Fan Speed that shows in the MOBO BIOS will ALWAYS show as ZERO no matter what?

Assuming that is the case, should I have it plugged into the PUMP_Fan1 header on the motherboard instead of the CPU_Fan header? I am thinking the answer is yes - correct? Some of the the reading I have also done would seem to indicate that I need to ensure the header is providing full 12v in the BIOS to the header if it is connected to the CPU_Fan header, whereas the PUMP_Fan1 header should provide a consistent 12V.

Does that sound right? Again, the BIOS is MSI. Thanks so much for the help!
 
So I went back in and moved the three pin (one wire) from the CPU_Fan header to the Pump_Fan1 Header. After a short time, it started to show up so it does now read in the BIOS - though it seems to be jumping around a lot 2500rpm, 5000rpms, 2100rpm, etc. However, in the icue software, it shows the pump running pretty stable from around 2300-2400 .

Now I would say the biggest issue is temps. My temps seem very high, even at idle. My system:
Ryzen 9 3900X
MSI MEG ACE X570 MB
Corsair 680X Crystal Case
32GB DDR4 3200 Corsair
PCIe Gen 4 Sabrent 1TB NVME SSD
Corsair H100i AIO
Gigabyte RTX 2070 Super GPU
Fans: 3 120s in front as intake. 1x120 in rear as exhaust. 2x120s in bottom as intake, AIO at top of case with fans exhausting over the radiator out of the case.

Temps - 47-55ish at idle. Under load when doing the Heaven Benchmark temps went up to 71 (based on Core Temp 1.15.1 app)

AIO seems to be working and all fans running. Fans set to PWM in BIOS with MSI "Smart Fan" checked to regulate speed based on temp.

Any thoughts? Only thing I can think of is to set top radiator to intake and bottom 120s to exhaust.
 
Ok
Well you are running the 3900x you may have to listen to the fans on extreme as that cooler might not be able to handle the cpu.
It certainly sounds that you have ventilation.
What is ryzen master showing your temps @ ?
Have tried setting a more aggressive fan curve in ique?
What ghz are you showing at idle?
If you haven't adjusted the windows power plan then you could be at high clocks at idle which would keep your temps up at idle.
Check power plan
minimum process power state and set to
5-7%.
Then recheck
 
To be honest, I haven't installed Ryzen Master because I never intended to overclock this initially anyway. I'm just using the MSI BIOS mostly on default except for enabling XMP so I get full frequency on my RAM. I have been using Core Temp program and MSI Dragon Center to show CPU Temp. It's showing between 41 and 60 while not doing much but surfing the web. I did a CPU stress test with CPU-Z and after about 5 minutes or so the temp was at 77 so I stopped the test.

The one thing I have noticed based on what I have read, is that the CPU voltage seems to be at 1.4ish all the time. It goes down to 1.17 sometImes but mostly stays in the 1.4 range. Based on everything I've read, that's too high.
 
Thanks for that. It was a good read. I did confirm I was on Ryzen balanced power and that only XMP is enabled. Still have temps I'm not comfortable with. That said, there are lots of different opinions on normal/safe temps. Mine right now are 42-61 as of last night doing nothing but surfing the web and minor apps. Running cinebench and CPU-Z stress for a bit got me up to 77. I believe that's too high. But the opinions I have read vary.
 
That's a fascinating and detailed note on voltages and how to interpret.

Regarding your CPU temps, I do not know what they should be for that CPU. But I note than many high-perfornamce CPU's today are designed to oeperate aroung 70C for most loads, and do not start to throttle until the are well over 80C. So, OP, check carefully how YOUR particular CPU SHOULD operate NORMALLY, and at peak load. Your temps may be quite acceptable.

Your tales of pump speed readings still puzzle me. You say when you moved the 3-pin (with one wire only to Pin #3) connector on the wire from the Pump to the PUMP/FAN1 header it gave speed readings which it could not do on the CPU_FAN header. However, the readings were erratic, and 5000 rpm is NOT likely to be correct! That suggests to me that the signals being delivered to that header are "noisy" and inconsistent, causing bad readings. That would indicate a poor connection, so I suggest you get Corsair Tech support involved. THEN you say that, in that condition, the iCUE software utility DID show you quite reasonable and stable Pump speed readings. That is intriguing, for two reasons: (a) why stable when "read' this way, but unstable readings from the SAME signal by a different display; and (b) how did iCUE discover that the PUMP speed signal it was trying to display had MOVED to a different fan header? Moreover, if I read your post right, even before you switched the speed signal cable to that second header, iCue HAD been showing good pump speed signals even though the BIOS screens said they were zero. How could that be?

By the way, I think you have done this but correctly, but want to check. In that AIO system, the RAD FANS are supposed to be plugged into the two fan output connectors on a cable from the Pump. They do NOT connect to any mobo fan header. Done properly, those two fans are controlled solely by the iCue utility, and that is the ONLY place where their speeds can be "seen".
 
  • Like
Reactions: toro1966
You bring good points Paperdoc
On my h110i on my asus rog strix x470 f gaming it also when hooked to the cpu fan would not show in bios but when changed to the water pump header it does show now but it isn't erratic.
That single wire is the sense or tach wire more or less and
I really don't know why it even needs to be hooked up because all pump and fans readings and control is through a usb 3.1 header then connected to the side or bottom of the pump head.
That is the way both my 2x h110i's and my h100iv2 is.
When i first booted the ryzen upgrade from a fx-8350 for the 1st time I was meet with a cpu fan error knowing i had it hooked it to the cpu fan header per the instructions.

If it's erratic in bios but not in Ique
That would be certainly be a question for tech support.
 
  • Like
Reactions: toro1966
Thanks Paperdoc and Cross - I really appreciate the help.

Paperdoc - yes, everything you said is exactly how it was and how I have it hooked up. I expect you are correct that the erratic reading is due to a crappy connector on the AIO pump lead (3 pin with 1 wire populated) and so I may just exchange it (Amazon). In fact, it seems looser than all of the other connectors when connected to the MOBO.

That said, iCue shows all is well so I know it is operating as designed, even if it is erratic in the BIOS. Since the BIOS is not controlling it, I think that's a non-issue. What i still can't find a definitive info on - perhaps because of the newness of the CPUs - is the correct operating temp range. It's all what people think and those opinions vary. Is there any definitive source you know of for info on the 3900x?

I am also considering trying to see if I can put a 360 AIO on the front instead of the 240 on top. I can't fit a 240 on top but I might be able to fit a 360 in front. I know from reviews I have seen that this case will run hot because of the glass placement (bad design - placing form over function). I am also consider a push-pull on the 240 AIO I currently have to see if that might help pull some more heat out.

Any thoughts? Thanks again!
 
Your case specs is showing
Top 240mm and 280mm
Front 280mm and 360mm
Bottom 240mm and 280mm
Back 120mm and 140mm

As far as push pull maybe?????

That's exactly what i am thinking - see if I can do a push pull on the top (AIO). I haven't been able to run any of the stress tests yet. Will have to do them this evening hopefully. It would just be nice to know what the normal temps are to compare them to.
 
Push/pull used to be somewhat effective years ago when Noctua fans were king by a country mile and everything else was junk. Now Noctua is king by a hair, fans have improved significantly, so are more efficient and effective, which reduced the discrepancy vrs the rad. Basically push/pull now with any decent set of fans will net @ 2-3°C under heavy loads.

At @ 150w output at stock values, that @250w TDP 240mm AIO shouldn't be having temp issues. Just the 3x front fans should supply plenty of air. In bios I'd disable the smart fan control on the cpu_fan header. You don't want the motherboard trying to control the pump and not getting anywhere because there's no pwm fan signal, if possible, set that header to DC voltage mode instead.

You could try using Precision Boost Overdrive, not to increase but decrease. I believe stock settings are 142 95 140 for PPT, TDC, EDC and that can send voltages that are way high for stock clocks, resulting in overly high temps. Instead try lowering those to 130 95 125. Should drop temps and still be stable, you may need to play around with them a little.
 
  • Like
Reactions: toro1966
Ok - some progress - though I am not sure I am completely satisfied yet, but getting there. Still have some questions.

First I did some more research and it looks like max temp for the 3900x is 95deg C. Doesn't say where operating temp is but if we use the normal 10 degrees less than max method, than I should be good normally up to about 85 deg.

Secondly, I updated the Chipset Drivers to the latest from AMD, and third, I updated windows to the latest update - I think it was 1903. Finally, I installed Ryzen Master and OCCT.

After doing all of this the first thing I noticed was that my voltage was actually moving now instead of staying at 1.4ish. I also found that every piece of software I have that reads temps is 10 -15 degrees or so higher than Ryzen Master reports! Ryzen Master shows my idle temp at about 37-41 deg consistently. OCCT, Dragon Center, Core Temp, and iCue are all about 10 degrees higher on average than the reading in Ryzen Master. From the research I did I found that this is because all of the other software apps read any max on any of the cores instantaneously, while Ryzen Master reports an average over short time. This would seem to make sense to me. Just because a particular core gets to a particular temp for an instant does not really give you a good sense of how the entire CPU is doing - so I actually think this is good - thoughts?

I also ran the OCCT test that was recommended. I ran it for 5 min and the temps were almost always in the 67-69 range. There were indeed two or three times where it jumped into the low 70's and once where it went to 76, but that was it. Generally it stayed in the 68-69 degree range. Because the test stressed all cores and raised all temps, Ryzen Master and OCCT read almost the same temps the entire test! I am going to run a couple of the other tests to get a good idea.

How long is a good CPU stress test? I only did 5 minutes because I wanted to be sure. Thanks again!
 
  • Like
Reactions: Crosslhs82x2
For temp testing, Prime95 small fft torture test gives a 100% constant cpu load, no bouncing. Occt can and does bounce because it uses linpack, which adds in a value for the ram too, useful, but not ideal if looking for what the cpu is doing.

For an AIO or custom loop, half an hour. The liquid coolant at rest will be fighting to remain at as close to ambient temps as it can, so the test needs to be long enough for the coolant to reach saturation, the point where the cpu output is not enough to raise the coolant temp any higher. That's when temps will level off. With aircooling, that's about 5 minutes, metal heats up fast, dissipates heat fast, so it's just got to reach operating temps.

You might be on to something with the Ryzen temps. Ryzens don't work exactly like Intels have done for years, so most software is written around Intel and it's way of getting temp results, every 256 milliseconds. So software isn't going to be accurate until it is written to AMD's way of doing things. So far to date, there's only 2 I can think of that are accurate, Ryzen Master and coretemp, and even coretemp isn't quite there yet, the author is working hard with Amd to get better results. So Ryzen master will be the goto software for pretty much anything Ryzen related as is was written specifically for Ryzen.
 

TRENDING THREADS