XD3 Pump - Slow or No Startup

[ifly2high]

Hello, looks like I might be having an issue with my XD3 pump either being slow to start, or not starting at all upon resuming from an off power state. Has anyone else been having any issues?

 

I came home one day to find a warning from HWINFO that said my cpu temp was at 85 degrees and I was a bit confused. When I opened up the program and looked at my temps I noticed my GPU was also hot and then when I looked at my water pump rpm's it was at zero. So I shut down the system opened up the covers and sure enough the system was hot! I pulled the spare power supply from my cabinet and plugged the XD3 pump into it to test and see if it died and it fired right up. I let the pump run for a bit and let the system cool down. I then hooked up the pump back to the power supply in the system (EVGA 850 p2) and fired the system up and well, everything is working. I setup HWINFO to monitor the pumps rpm as the tach lead is plugged into my cpu fan header (it's set at a 40% constant speed in the bios) so that in the case it's at zero for so many read cycles to give me an alert and if's it's unattended for so long with reading zero rpms I have HWINFO shut my system down so I don't have another run away hot system. I've contacted EVGA thinking it could be a power issue and they suggested to plug the molex harness into an empty SATA connector on the power supply and see if the issues goes away. Well this morning when I woke my system up, I noticed the the pump rpms were sitting at zero when my desktop loaded (I have rain meter setup with a skin to monitor system information). After about a minute I then noticed the rpm's start to come up before my panic started to set in that it was not working again. I'm now wondering if it's the pump and not the power supply. My next step is to wire in a 12v led to the pump's power harness so I can make sure that the power supply is in factg supplying voltage to the pump to rule out if either the power supply is causing the issue, or if the pump is just not starting up all the time.

 

Are other people having or seeing any issues with these pumps?

[c-attack]

This is the first issue of this type I have seen. Most people with troubles have lost PWM control or have obvious motor issues. This appears to be something else.

 

Make sure the control header is set to PWM mode in the bios or “4 pin” in the Commander Pro settings. Most cpu fan headers are automatically PWM but an auto-detect feature could misidentify the DDC motor signal. It’s possible it sends the full 12v needed for start up on normal power on, but fails to do so on wake. As a PWM motor it needs 12v all the time.

 

You need to be careful if using a molex to SATA adapter. 12v is usually less of an issue than 3.3 or 5v, but sometimes there is a problem on the rails. If the lighting is off too, that would be an additional indicator this is a problem. However, if this problem only occurs when waking from sleep vs normal power on or reboot, I think we are looking at PSU cause and this would seem to be the area.

 

I would contact Corsair Support now and if it gets resolved — great. But better to get things in motion if the issue is more serious.

[ifly2high]

I woke my system up this morning and when I arrived at the desktop my monitoring display showed me a zero rpm for about maybe 30 seconds or so. Be reminded on me having to wake the system, wait for the login screen, enter my password and wait the desktop to load. The monitoring software was functioning because I was watching all the other things such as temps and my other fan plugged into the aux port on the mobo was being updated but the pump rpms was sitting at zero. The pump did spin up before it triggered to shut the system down for no pump speed, and I could watch it on the monitoring software, it was late to the startup party.

 

I checked my bios settings (MSI Z270I Pro Carbon mobo) and it is set to pwm, locked at 45% and the system reports back 2400 rpm speed, they stay consistent. I did however change the step up/down time to .1 from .7, maybe that will help?

 

EVGA had me plug their molex harness from the "perf" port to the "sata" port stating that the harness would work with out any changes because the sata and perf ports are wired internally the same inside the power supply, so by swapping their molex harness to the other port no damage would occur and still receive the 12v needed. I guess that is their easy way to test and see if it's the port or not.

 

I don't have the leds on the pump even plugged in, so they are not in use at all.

Edited April 24, 2021 by ifly2high

[ifly2high]

I got parts in today and made a 12v molex to led adapter that I have plugged into the next connecter after the pump, and ran the led to the front of the case so I can see it's glow near my radiator. Now we wait and see, if I see a green light and no pump rpm running then it's the pump or if not, the power supply is not suppling voltage. Crossing fingers this issue just goes away....

[ifly2high]

Looks like it's the pump that is slow to startup. Everytime the system wakes up the power led I hooked into the harness turns instantly on, so it's not a late power delivery issues it seems. Only other thing I can think of would be the motherboard not sending a pwm signal to the pump right away, and I can swap the pumps pwm plug over to the motherboard aux port and see if the pump starts right away or see if it acts the same. It has started regardless (knock on wood) but sometimes right away like this morning, and others I'm sitting there wondering if it will start but does eventually come on.

[LeDoyen]

if the pump doesn't receive a PWM signal, it will default to full speed. and if it receives a ridiculously low signal like 1 or 2%, it also defaults to its minimum which is like 30%.

if power is good, the only thing i see faulty would be the pump controller itself > RMA.

That's why i hope you took C's advice to create a ticket already. If it's indeed the pump and it dies you'll be without computer until you get the replacement shipped :/

[ifly2high]

Well after many weeks with no issues, today, just now actually, the pump failed to startup, and my monitoring software shut down my system for me. Upon powering the system back up, the pump is functioning again. The led I have tied into the 12volt plug it's connected into was lite, so power delivery was there. The only other thing I can do is to swap the pwm connection to the aux connector and the fan I have in it to the cpu connector on the motherboard and see if the pump behaves the same way. I can swap the monitoring software to look at that port and let it behave the same way so my system would still shut itself down in the case of a prolonged zero rpm reading.

[vidkid]

My XD5 pump has been plagued with similar symptoms.

 

I have tried everything you have, with no positive results. Where are you at with getting this resolved?

[if_you_build_it]

Im having the same issue with my new XD3 after only about 2 months of light use.

When PC goes to sleep or powers off and comes back on, the pump struggles to start.  Makes like a clicking noise like its trying to spin up or something but you can see its not spinning when looking at it.  After maybe about 1-3 mins it sometimes engages and start running.  It happens sometimes from cold boot as well.

When this happens, the pump info is missing from iCue.  Just started happening the other day after a couple months of use.  Am I dealing with a bad unit or is this something else?  

[c-attack]

1 hour ago, if_you_build_it said:

Makes like a clicking noise like its trying to spin up or something but you can see its not spinning when looking at it.

Contact Support.  A pump should not click -- at all.  

[LeDoyen]

And while waiting for the replacement pump, see if you can give it some airflow, and when you'll receive it, try to run it slow-ish. Not too slow that it causes your waterblocks temps to rise, but not full speed either 🙂 

DDC pumps are air cooled, so if you can reduce pump power draw, you'll lengthen its lifespan (if that was what killed it.. )

[if_you_build_it]

7 minutes ago, LeDoyen said:

And while waiting for the replacement pump, see if you can give it some airflow, and when you'll receive it, try to run it slow-ish. Not too slow that it causes your waterblocks temps to rise, but not full speed either 🙂 

DDC pumps are air cooled, so if you can reduce pump power draw, you'll lengthen its lifespan (if that was what killed it.. )

Its mounted to my RAD which has 2x140mm fans blowing on it.  Pump only runs about 1300 rpm set in iCUE, almost silent.  Keeps my radeon 6800xt and ryzen 5900x temps around 48C and 43C respectively at idle.

1 hour ago, c-attack said:

Contact Support.  A pump should not click -- at all.  

Its like the motor tries to rev up but stalls right away... and repeats a dozen or more times until it catches and starts running.  maybe 'clicking' isnt the best way to describe.

[c-attack]

You might call it "slipping", but it does not matter either way.  These are the early symptoms of a mechanical issue and one day it simply won't be functional.  This is not the expected behavior at 2 months or 2 years.  Follow up with Support.

[CableNS]

Background. 

I have two Xd3 pumps in my system.  My CPU pump has never failed.  My GPU pump is on the Third replacement Xd3 pump. When the pump fails while in the Operating system , I no longer "see it in ICue commander".   When it fails in the OS , if I reboot and sit in the bios it runs full tilt.  My Pwm controller is plugged into one off my Commander Pro XT's This  would possibly explain the fan running wide open in the bios.  When I come back to the OS the fan stops running or is running so slow I can not tell but the GPU temp climbs.  After several reboots it will again show up in ICue commander and life is good.  This pump is maybe 6 months old.  I have even replaced the commander core Xt's .  I can not make the CPU XT fail, only seems to be the GPU one that the OS stops seeing it.  

 

Any ideas ?  I am about to open a support ticket but I suspect I am going to have to do something different.  

[c-attack]

It depends on why the XD3 is failing.  Impeller failures are pretty obvious.  You hear clicks.  The pump is trying to start and can't.  No flow, rapid temp increase.  

 

The other commonly reported issue is a tachometer or PWM sensor fail.  The pump is running but is no longer responsive to controls.  Since the Commander XT runs an auto-detect check on each fan header at boot, it shuts down the PWM header with the XD3 if the pump doesn't send a signal back at boot.  It's hard to say where the fault lies from this side.  If you encounter this, do a full shutdown, flip the PSU off for 15 seconds, then back on and start up.  Obviously if this is happening on a regular basis this is not going to be an acceptable work around.  

 

Not directly related to this, but do you need two DDC pumps in the system?  Presumably this is for separate loops on the GPU and CPU, but that is not always beneficial or necessary except in very large systems or those that run near maximum for long periods of time.  

[CableNS]

 "  Since the Commander XT runs an auto-detect check on each fan header at boot, it shuts down the PWM header with the XD3 if the pump doesn't send a signal back at boot"

That's a very key piece of the puzzle.  Thanks for that info.  

The PSU shut down does work i think most of the time. I have to pull the power and wait 30 sec or so for the board to drain power.   I have not had the problem on this newer pump. It just started :(. 

What is interesting is the Pump comes right up if I boot into the bios.  It runs full rpm and I suspect it comes up full RPM because the PWM is not connected to the board directly but through  the Commander core XT.  

Does the Commander Core do anything intelligent if the OS is not up yet ?  I am assuming it does not.  My next step may be to bypass the commander Core and go straight to the mother boards PWM.  

Also for clarity  I am not getting any noise from the pump out of the ordinary. 

As additional background and not super relevant but for complete picture.  

I have two commander core Xts   one commander pro and one lighting pro.  The PSU is 850 Watts.   The GPU is actually sitting off one of the Commander Core Xt's and the CPU is sitting of the Commander Pro. 

Un- related question answer. 

C- Attack I run a loop for the GPU and CPU - Its a Lian Lia Dk05F Desk and it is as much for aesthetics as well as being able to run the fans super quiet.  I have a 3080 ti and at 1950 mhz and it runs a cool 39c. 

Thanks for the response! 

[c-attack]

OK, two XD3s it is. 

 

2 hours ago, CableNS said:

Also for clarity  I am not getting any noise from the pump out of the ordinary. 

I think this is the PWM/tach issue.  Bad impeller is painfully obvious.

 

2 hours ago, CableNS said:

Does the Commander Core do anything intelligent if the OS is not up yet ?  I am assuming it does not.  My next step may be to bypass the commander Core and go straight to the mother boards PWM.  

It should do the fan check at power on and then decide whether or not activate the PWM header.  So that poses an interesting question if the XD3 is slow to respond.  Some pumps have a deliberate delay built into the power up to take some stress off the unit.  Possible that causes it to miss the Com Core or XT fan check?  I don't know.  Those two controllers did not exist when I was using the XD3.  Regardless, I agree and moving the XD3 tach/PWM connector to the motherboard is the next logical step.  That may help differentiate between a XD3 and Com Core/XT issue.  

[CableNS]

Next time it fails I will take the glass of this beast and wire it differently.  I will keep you posted.  Today was the first fail so may take a few days to fail again.  Thanks for your responses.  Really appreciate the insights into how the thing works.  Very helpful.