other H80 and H100 pump noise solution

There's a lot of unhappy people out there (like me) whos H100 (or H80 because of the same pump) is making a 1990's hard-drive noise.

For a long time this was believed to be caused be caused by air bubbles in the system. The only recommended solutions were to get it replaced by Corsair, or to shake the cooler while it's working, until the noise goes away.

Recently someone discovered that ts actually caused because some PSUs give a little more than 12 V. his causes the pump to work faster than it should. Like at 2300 RPM instead of 2000.

The new recommended solutions include:

• Use something like a fan controller that has an extra molex for the pump or something.
• Build a small adapter that has a 1N4001 diode.
• Build a large chain of random adapters and cables between the PSU and your pump.

Sources:

• http://www.overclock.net/t/1209387/corsair-h100-pump-noise/10
• http://forum.corsair.com/v3/showthread.php?t=97733&page=22

I have yet to try any of this, as I just found out about it. I'll try to get some fan speed measuring program first to see how fast my pump is.

EDIT 1:

Maybe I could just use a molex to 3 pin adapter (which I already have) and use that to plug the pumps molex right into one of the PWMs on my motherboard? The case fans are meant to work at up to 12V, right? Or maybe this would be bad, because the motherboard measures its temperatures and only turns those up to full when it thinks its hot? Either way, I have some experimenting to do.

I could get a fan controller, but I'm trying not to have to do that, because 2 of the case fans use molexes and I don't want my whole case to become full of adapters and extension cords.

EDIT 2:

According to "SpeedFan", the pump is working at 2165 to 2181 rotations per minute. According to the people at the Corsair support forum, most of their pumps became quiet at about 1700 RPM.

EDIT 3:

According to the "CPU Thermometer" (by Microsoft), the cores are between 24 °C and 30 °C at idle. The stock cooler kept them at about 32 to 37 °C at idle.

EDIT 4:

Turns out that the molex to 3 pin adapter I have has the wrong (male) 3 pin part. (Duh! ) So all actual tests will have to wait.

Edited again for spelling.

The H100 does not have those resistors, because it has a button on the pump that can set the whole thing to 3 different speeds. The fans are plugged into the pump. As far as I know, the pump speed remains the same at all settings.

quote Some dude at http://forum.corsair.com/v3/showthread.php?t=97733&page=22

quote Some other dude

I can confirm this. Now when my pump reports ~2040rpm instead of ~2190 i haven't seen any change in temps at all. The difference is just that the grinding noise is gone due to the lower voltage

I second that! The pump speed does not affect the temps at all (well if it stops or is really slow it does). The flow-rate on the H100 is very slow and the difference in flow-rate at 2000RPM or 2200RPM is negligible. I've modded my H100 and have played around with adding a second pump in the loop to up the flow rate, which also had no effect. Winding the pump down to about 1800RPM is starting to have a noticable effect on temps..

I've taken one of my unused PWM extension cords apart and put the female end onto the molex to 3 pin adapter.

It looks like this now:


I'll plug that into an empty PWM in my motherboard and see what the pump speed and temperatures are like.

EDIT 1:

Good thing I remembered something first. That PWM on the motherboard is meant for a single fan, but I need power for not only the pump, but 3 fans as well, because they're plugged into the pump.

I guess I'll need that fan controller after all, unless I'll get more molex to 3 pin adapters, but... There has got to be an easier way around this. Maybe I can find some diode or resistor.

EDIT 2:

I'll test each PSU rail, just in case.

no, not the diode. that just stops back current...though to be honest, i dont know how it helps if all computer hardware uses DC, not AC. but then there must be something thats reducing the RPM somehow. when you say theres a voltage drop, where is it being measured from? the PSU rails?

my confusion with the diode was about HOW it drops the voltage, if it isnt by resistance. from what i know in electronics, the voltage from the power source (in this case, the PSU), should be pretty much constant. therefore, the only way (that im aware of) to drop the voltage would be with a potentiometer. otherwise, its just increasing the resistance to decrease the current.

quote

I'm not even sure that PWM of a simple DC motor is a good idea

there are no AC components in a computer, other than the input power from the outlet to the AC side of the PSU. all fans are DC. PWM is perfectly fine.

EDIT: wait...im not sure if the clockgens/crystals are AC or DC...ahh doesnt matter, its got nothing to do with fans

quote

How it drops the voltage is to be honest not something I can easily explain but it goes beyond the simple realms of V=IR. A diode is a semiconductor device with a PN junction and it doesn't matter how much current you stick through it (providing you don't exceed its maximum rating) it WILL drop 0.6V, fact. This is also true for base-emitter or base-collector on a transistor, same thing.

i guess part of my confusion stemmed from there. when i think of PN junctions, i think of transistors, and when i think of transistors, i think of three possible connections to one, whereas the diode only has two.

quote

When did I say there were any AC components in a computer? All I am saying is it is perfectly fine for fans to be run from a pulse width modulated supply because there's is very little resistance to continued rotation. Try running a simple DC motor that's pushing water and has a fair amount of resistance and you will reach a stall point were the motor can't rotate and reach the next set of windings.

well when you said 'a simple DC motor', it sounded like you were comparing AC to DC or something. never mind, then.

as for the PWM and stalling, im just speculating here, but perhaps PWM only kicks in after a minimum RPM?

Well I've decided to do this right and get a Zalman ZM-MFC1 plus. Seems like a cheap and effective thing.

My soldering skill is just shit and I don't want to have any suspicious loose ends. And I definitely don't want to burn out any part of the motherboard.

quote biomed

Is 7W enough to control the water pump?

Good question. There's no wattage listed anywhere, but I just found someone at the Corsair support forum, who had measured that the pump gets 1.1 to 1.2 A and he claims that it works "fine" on his 1A 12W motherboard fan connector.

If the 7 W is not good enough, I'll just use the controller for fans and connect the pump to the motherboard. My motherboard fan connectors are good for up to 24 W and 2 A.

So either way, this will become super quiet, as I'll also be getting some of Corsairs new quiet edition fans.

EDIT:

Strange coincidence that this video should now show up, but I'll still use my first plan.

They'll replace it for free now:

quote http://forum.corsair.com/forums/showthread.php?t=107252

Released from Cooling Product Manager:

"Corsair has now tracked down and fixed a rare and hard to reproduce issue on some of our Hydro series coolers, specifically the H60, H80, and H100 when used in combination with certain power supplies in certain circumstances. The failure will present itself as a buzz or hum, similar to a hard drive seek noise, and may disappear and reappear at random times. This issue affects a very small number of units, and typically is more prevalent on power supplies that provide higher than +12.2V on the +12V rail, which explains why some users have experienced multiple units with the same issue – they’re plugging them into the same power supply.

Regardless, this has since been corrected and we are now shipping units that do not experience the issue. If you’ve got a noisy H100, H80, or H60 unit, and the noise is coming from the pump (not the fans), we’d like to offer you a RMA. Please Contact us by using the link on the left and request an RMA to start a replacement with a known good unit.

Thank you, and we appreciate your patience as we corrected this issue."