Water Pump Plugged Into CPU_FAN Header

[ProbablyRaging]

Hi,

Just recently installed a Corsair H105 CPU cooler and while installing I noticed the manual said to plug the cooler pump into any 3 pin fan header but didn't clarify if it had to be a CPU or CASE fan header so I went a head an plugged it into 1 of my 2 CPU fan headers.

From what I can tell it is running as normal and I haven't had a single issue, but after doing some reading online a lot of people say to use a CASE fan header and not the CPU fan header.

May I ask why this is and what the harm of using the CPU fan header is?

Thanks.

 

[alexandergc]

If you plug in the pump to your CPU FAN header, you should set the fan curve to maintain a constant flow, preferably above 50% at all times.

Normally, the CPU fan will swing RPMs from low to high as needed when it detects a temperature change in the chip.
Fans are built to take this kind of changing load, and they provide less resistance anyway.
Pumps on the other hand, should always keep circulation going. The liquid you're pushing doesn't make it any easier either.

The main worry is if your fan curve has a portion that goes too low for your pump to run, causing it to shut down.
When the temps inevitably go higher, the fan header will push more voltage to try and get your pump back on, but it MIGHT (rare, but it happens) not work and instead fry the pins.

3-pin CASE fan headers are non-PWM, so will run at a constant voltage all the time and are the safest bet for connecting a pump.
You could also go a little bit ghetto and hook up the pump to a 3-pin-to-Molex connector direct to your PSU. It does the same thing as a 3-pin fan header.

 

[ProbablyRaging]

If you plug in the pump to your CPU FAN header, you should set the fan curve to maintain a constant flow, preferably above 50% at all times.

Normally, the CPU fan will swing RPMs from low to high as needed when it detects a temperature change in the chip.
Fans are built to take this kind of changing load, and they provide less resistance anyway.
Pumps on the other hand, should always keep circulation going. The liquid you're pushing doesn't make it any easier either.

The main worry is if your fan curve has a portion that goes too low for your pump to run, causing it to shut down.
When the temps inevitably go higher, the fan header will push more voltage to try and get your pump back on, but it MIGHT (rare, but it happens) not work and instead fry the pins.

3-pin CASE fan headers are non-PWM, so will run at a constant voltage all the time and are the safest bet for connecting a pump.
You could also go a little bit ghetto and hook up the pump to a 3-pin-to-Molex connector direct to your PSU. It does the same thing as a 3-pin fan header.

I see. That is basically what I figured would be the problem and I actually had a little look at a program from MSI called Command Center which shows my fan info, including the pump which it thinks is 'fan 1'.

From what I can tell in MSI Command Center is that the 'fan 1' (pump) is running at a constant 1800RPM all the time which is normal correct? Even if I try to change the fan curve it wont actually change the RPM of the pump, from what I can tell anyways.

So with that information would I be safe to assume that I am OK to keep it plugged into one of the CPU fan headers or should I still go a head and swap it to a CASE fan header?

Thanks for the response.

 

[alexandergc]

PWM fan speed control requires that the pins AND the plug have 4-pins each.
If your pump runs on constant anyway, the motherboard should be smart enough (and in your case, it is) to detect that and disable the fan curve.

Seems like you should have no problem leaving it on your CPU fan header.
It's an advantage that you now have a free CASE FAN header for any extra fans you'd like to put in the case.

 

[JackNaylorPE]

CPU fan headers are just about always PWM .... CHA fan headers were almost always DCV (even when 4 pin) up thru Z87. Now they are almost all 4 pin (at least on gaming / enthusiast boards) and do both. Use PWM header for PWM devices. Using DCV to control PWM is not a good thing.

http://www.swiftech.com/pwmcontrollers.aspx

[Using] DC operating voltage to modify fan speed [edit: in PWM devices] is not recommended. Transients generated by that approach can irreversibly damage motor commutation and control electronics and dramatically shorten the life of a [PWM] fan [or pump].

Which you use would be based upon the type of pump..... if it's PWM you'd use a PWM header. I don't know what type is used in CLCs as we don't recommend them. Speed control is CLC pumps is kinda useless.... and only minimally effective on custom loops. In a custom loop, the idea is to maintain a flow of 1 gpm or better thru CPUs, you can get away with less on a GPU because if the massive block. I have my dual pumps in my loop and they are capable of delivering up to 2.5 gpm. I have them set up to deliver from about 0.75 gpm minimum, and 1.5 gpm maximum under normal usage.

Martin from martinsliquidlab tested the Corsair H100i and found the pump flow was only 0.11 gpm. At that flow rate, I certainly would not adjust pump speed any lower.

 

[alexandergc]

I'm guessing the H105's recommendation of 3-pin fan header is for DCV only, since the 4th pin is usually the PWM signal.
In any case, it would be safer to just plug it into the recommended pins since it's really just a matter of changing the wire location, though it doesn't seem to be having any problems in the CPUFAN header right now.

 

[Karadjgne]

Corsair recommends plugging the pump into cpu_fan for 2 reasons. 1st is cpu alarm, if the pump fails, it'll read as 0 rpm and the pc will immediately shutdown or refuse to start, not having a viable cooling solution. It's a safety mechanism. The second reason is corsair doesn't want you to have any control over pump speeds. The cpu_fan header is, as previously stated, pwm controlled header, if you have a 3-pin connector, the header runs 12v constant that software cannot override. If you plug the pump into a case fan header, there is a good possibility of it being analog controlled, and people don't set it at 100% in fan curves, so complain or RMA due to pump failure, which wasn't the case.

While I do prefer clc's and do recommend them, if budget isn't an issue or especially if mobo/ram/case restraints force the issue, I don't follow Corsair's recommended policy on this as I'm adept enough to keep tabs on temps and much prefer my rad fan monitored by cpu_fan header, so it doesn't run 100% fan at all times, instead it's running the same fan curves as any aircooler, slow at idle and fast under high temps.

The pump/fan can be used on any header or direct to psu with no worries. There is only 1 stipulation. There must be something connected to cpu_fan header, or you'll not be able to run the pc.

 

[JackNaylorPE]

The reason that today's MoBos have a CPU and CPU_OPT header is precisely this reason. So that you can control both pump and fans from the CPU. The CPU_OPT **on most MoBos** is a dummy header. Control of both CPU and CPU_OPT is based upon the feedback signal from the device connected to CPU header, the wire from the device on CPU_OPT is for all intents and purposes, "clipped". So having a 3 pin connection to CPU_OPT is exactly what is expected as 1) it is considered redundant and 2) it eliminates the issue of the 4500 rpm pump and 2700 rpm fans sending two different signals on the same channel. If your fan curve is set to run at 40% at 30C, 60% at 55C and 100% at 75C, then at 55C:

Pump will run at 60% of 4500 and fans will run at 60% of 2700

BOTH your pump and rad fans will in fact not run at 100%.... both will speed up and slow down in response to t he fan curve you set up in the MoBo utility

This is not true on all MoBos however, Gigabyte Force Boards for example allow independent control of each header.... this is kinda nice as:

1. You can shut fans off whereas you never want the pumps off
2. You can essentially set a max pump speed.

This is of less value with a CLC as the pumps are comparatively weak (H100i is only 0.11 gpm)..... I see no value in controlling pump speed at all in this instance. My pump ranges from about 1400 - 4500 rpm (2.5 gpm max), tho the MoBo limits it to 1450 - 3000 (about 1.5 gpm max)

 

[Karadjgne]

Kinda wierd to have control over cpu_opt. My Asus p8z77-v LE runs h55 pump on cpu_opt at @1480 +/- 10rpm, and nf-f12 is on cpu_fan and ranges from 254rpm to 1055rpm, yet doesn't budge the pump one iota. I actually prefer it that way, since the pc is basically my wife's, have Asus suite alarm set to 1400rpm on the pump, so she'll freak out about it long before any damage can be done.

 

[JackNaylorPE]

I couldn't tell ya how CLCs work as we don't install them other than the Swiftech H240-X / H220-X and soon the Predator from EK. If someone wants one, we'll do the build and install the Stock cooler and they can take it from there.

As I had indicated, I would not vary the pump speed on a pump that maxes out at 0.11 gpm... and that is for the H100i .... dunno what the single 120/140mm ones do. I set min pump flow at 0.75 gpm on SLI builds ... sometimes 0.5 w/ single card builds. My problem is that users wanna take the things home and start gaming as soon as it's done so I never get a chance to try "this and that" for very long. All I have here for every day experimentation is a custom loop and a half dozen air cooled desktops and half dozen lappies.

 

[Sacco_Belmonte]

Both my AIOs had been in the "chassis 1" fan header for the last 4ish years. The reason is that I have a sound studio and I need the PC as quiet as possible.

I used my last AIO (Cooler Master Seidon 120mm, crappy loud pump) at 1600rpm for 2 years. Still working I swapped it for a Tt Water extreme 3 250mm.

The pump on the Seidon stopped at 1500rpm. Literally stopped at that voltage. Still the pump held fine and worked reliably at 1600rpm.

The pump in the Water extreme can run at 1500 or lower (as low as 1200rpm), so I've set it to 1450rpm fixed.

In your case you might wanna see if your pump can work all the way down to 1200rpm or if it stops, you need to listen carefully and put it back up if it does and find the point it stops and stay 100rpm over it, don't set the pump at the bare limit, mine stopped all of the sudden while it was set at 1520rpm or so.

Seems like those with the rounded low profile pumps (like the ones from Corsair and Tt) are better in that regard, they can run real slow and are also more silent.

I would rather recommend against ramping up and down. When I change the speed I can hear how the pump slowly ramps up and stays on the new voltage, I can hear how the pump takes some small time to adjust to the new speed so I assume that can harm the motor and actually compromise water flow due to turbulences.

I rather have EVERYTHING at fixed speeds....fans howling up and down are a disgrace. 750rpm to 800rpm is enough. Even here in summer.

1400rpm on the AIO pump, if you pump can support it and you don't hear any rattling while doing so go for it. No need to have it at 1800rpm drilling your ear. I can hear it at 1500 but at 1450 it almost "disappears" (on my room fully closed)

At those speeds you can do everything you normally do (working on your audio/video projects, gaming, browsing, and only while rendering or encoding, or anything that can take 100% CPU usage for a long period you just load the Full Speed profile.

That said this all applies to both my overclocked machines (3770k at 4.6Ghz and Ryzen 1800x at 4Ghz) in a good case, nothing extra fancy but well ventilated.

For (dust free) systems running at stock speed you don't even need to ramp up the fans ever.

BTW: If you have a 3 pin pump you need to set the fan controller to DC in the BIOS. At least that's how it is with my X370 PRO (Ryzen). It worked right away with my old P8Z77-V LX MOBO but I never had to check anything in the BIOS.

 

[Karadjgne]

Agreed. My nzxt Kraken X61 has a software control for the pump, it's designed to work that way, slow speeds at idle liquid temps and raises as the liquid temp does, not the cpu temp. The fans are the same, so small jumps in temp (very common on Intel cpus) don't raise fan speeds as they are temporary. Under sustained loads, the liquid temp rises, so do the fans. Currently the fans are set on silent mode which keeps them @450rpm but will go upto @700rpm if I push it, only going to my user set cap of 900rpm if I stress test. Which is still pretty much whisper quiet. This on a 3770k currently at 4.6GHz (funny that) which still has speedstep enabled so idles at 1.6GHz. Dead silent.
But that entails actually following nzxts recommendations as far as installing the clc goes.
My wife's h55 is on cpu_aux, the fan on cpu_fan. This gives the cpu direct control over the fan speeds and the pump remains at a very quiet/silent (round Asetek pump) 1480rpm ±10rpm, while the fan varies from @450rpm to 900rpm. The stock fan at 2200rpm, hooked up as Corsair recommended, was constantly, abysmally loud, it had to go.