Recovery Advice for a stupid Pump Error?

Associate
Joined
10 Dec 2009
Posts
21
I managed to "get myself volunteered" to refurbish a PC for a friend of mine. Part of this work has involved replacing 5 radiator fans, 2 pumps, 2 loops worth of soft tubing and 1 water block [GPU] and giving a very thorough [Mayhems Blitz] clean to the rest.

But I've hit a weird problem.

I couldn't tell you what brand the original pump/reservoir units were, but they looked like this. The new owner of the machine gave me a pair of these and asked me to fit them instead.

At first everything went brilliantly well... The new tubing runs went smoothly... I made up some unorthodox but effective mounts for the pumps, got everything plumbed in, added coolant [Mayhems Blue, pre-mixed] and all looked good.

For the initial coolant loading, rather than use the system PSU, I used my bench power supply and a cannibalised molex cable. Again, this worked brilliantly... I was able to prime the loops and then leave them running on the Eventek unit for a couple of hours to get all the micro-bubbles out of each circuit.

When running on my bench PSU, the new pumps were *completely* silent. Well, I had to get my ear to within 3-4 inches of the pump to hear it. I did all my testing with the pumps "final position" fitted to the case. Happy, I re-connected the pump molex connectors to the system's Zalmann PSU and fired the machine up. The good news is that the system booted first time, no issues.

The bad news is that these pumps sound like a couple of jet engines going through a startup cycle. From what I can tell, they appear to be running flat out. I don't understand [at this point] why the pumps were silent and ran at a self-moderated speed from my bench PSU, but run flat out under the PC's Zalmann PSU. (On the bench unit, they were taking 12.0v and 1.25A).

After scratching my head in disbelief, I looked everything over and came to the conclusion that perhaps the reason is that the new pumps are "PWM" units - each pump has 3 cables emerging - power, LED and a 4-pin PWM, only the first of which is currently connected.

So here are my questions...

1. Is the fact that these are PWM pumps likely the reason that they are now running flat out [i.e. is this just because the +12v is running 100% of the time with a good bit of current behind it?

2. Is there a neat, simple way that I can implement something to allow me to control pump speed without having to spend a fortune or remove/replace the pumps? I suspect the first place to look would be the motherboard - which is a now-elderly MSI Big Bang X-Power II hosting a 6-Core i7 4960XE CPU and 32Gb RAM. This board has 6 4-pin fan connectors [1 x CPU, 5 x System] and the manual shows them to support PWM fans... I'm a bit reluctant to try without seeking advice, so: would it be possible for me to connect the PWM cables from the pumps to a couple of these PWM fan headers and then try to control the pump speed through the BIOS?

3. Secondly, should I instead consider some form of manually set, in-line controller? Specifically, I was thinking of something like this, although TBH I'm not sure if this specific one is a step forward, since it looks like it requires a 4-pin PWM-like feed from somewhere (motherboard)? [Motherboard Manual PDF.]

4. If all that fails, should I consider maybe re-purposing one of the currently external USB ports and instead put something like an AquaComputer Quadro inside the case and then use AquaSuite to give the two pumps some sort of profile?

Can you tell that I'm clutching at straws here?

Any advice would be very gratefully received.

Thanks in advance
 
OK... I think this might be the final update. I followed MickyFlinn's advice as above, but found that this did not actually solve the problem. It let me view the pump speed [which was about 4,500rpm for both pumps] but not alter it. So I bought a couple of these:-

https://www.amazon.co.uk/Noctua-NA-FC1-4-Pin-PWM-Controller/dp/B072M2HKSN/

Which are Noctua NA-FC1 in-line PWM fan controllers. The documentation provided suggests that they can work in "passive" mode - i.e. just plug the controller in to the 4-pin socket from the pump - or "in line" - i.e. connect the controller between the motherboard and the device. I tried the first approach but that didn't work. Added the second cable for an in-line connection and got a perfect result.

This is a little disappointing, since I'm going to need to run some careful testing with Prime95 and a GPU stress test to get the machine up to a heavy load and then adjust the "volume dial" on the controller to suit. But the positive news is that this little widget looks to have resolved the problem.

I did try to figure out why MickyFlynn's suggestion wouldn't work. I'm not saying that this *is* the reason, but I suspect that it is because the cable that comes from the pump is only 2-pin, not a full 4-pin PWM cable. But that's just a guess. In case anyone is wondering the pumps are EKWB Quantum Kinetic 160s... (DDC).
 
I was able to see the pump speed in 2 different places… First, it was visible in my BIOS. The motherboard in this case is an odd-sounding but very stable MSI Big Bang Power II… It runs AMI Bios v2.8 and that has the ability to set fan speeds and the utility tells me current fan speed.

I was altering it through the use of the external NA-FC1 units… To work, I had to connect these between a couple of motherboard PWM headers and the PWM tail provided by the pumps.

That seems to work just fine - I am able to put the pumps up to maximum and then slowly and carefully dial back until the pumps become inaudible, whilst keeping an eye on temperatures courtesy of CPUID.
 
I absolutely agree with you in terms of what I would have expected to work and what I would not expect to work. All I can tell you is that when I connected the pumps directly to the motherboard, I could not find any means of positively controlling their speed. When I connected via the adapters, I could. I would have preferred a direct-to-motherboard solution as it would be cheaper and neater… but it seems to be working ok.
 
Back
Top Bottom