• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Dynamically controlling your water pump's voltage/noise level

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

dinofx

Registered
Joined
May 23, 2006
I have the Laing DDC pump installed in my Shuttle SFF, which has one of the later Intel chipsets on it. My motherboard has three PWM fan headers that can be controlled either via BIOS (useful when booting to Linux) or using software such as Speedfan.

After finishing my water cooling conversion, the water pump was the only device that did not throttle down when my PC was idle (all of the fans are PWM). At idle, it was the loudest component, even considering my dual 150GB Raptors. Many months later, I now have a custom circuit I designed that uses the motherboard's PWM signal to create a steady voltage that can be varried from 7 to 12 volts. The voltage can power two devices, such as a Pump and non-PWM fan.

The result is that the pump's tachometer reports speeds from 2800-4300. The pump is much quieter at idle time, and the wear and tear on the entire loop and pump is reduced, which should improve reliability.

pwmcircuit.jpg


Picture of the circuit installed, with wiring explanation:
usage.jpg


A picture of the boards and components used:
kitparts.jpg


Speedfan results (100% - 7% PWM in 1% decrements):
speedfan.gif


The pump's tach. shows that it slows by 1500, over 1/3rd of its ~12V speed.
 
Last edited:
or you could just use some sound dampener:D

looks pretty easy, nice find and job!!!

The only thing i see is a problem is the two pin power. Not sure how to go about that one since eveything I see is three on mine.
 
The 2-pin power is the power coming into the circuit, not going out to your pump/fan. In my case, at the other end is a standard MOLEX plug from my power supply. It's 12 volts, obviously.

The circuits output is two standard 3-pin fan headers.
 
Sweet. Do you plan on marketing these? I imagine that there are people that would buy these, especially if it can handle a D4/D5.
 
wow i was thinking of making somthing exactly like this after i had the same problem. if you decide to sell these as a kit or pcb send me a pm

No if he does plan to sell these, he won't PM anyone he will LIST THEM IN THE CLASSIFIEDS. If he or anyone for that matter, PM people with things to sell and we find out about it, you will be banned. There is a strict rule here, NO OFFERING ITEMS OR TRYING TO BUY ITEMS OUTSIDE THE CLASSIFIEDS. If you want to get banned and potentially get scammed and lose money, go for it.

nikhsub1
 
Last edited by a moderator:
Hum, nice... All it took was an opp amp, a power transistor, and some various capacitors and resistors? So what are you doing, feeing the PWM signal into that large cap through a resistor and using the op amp to make the output voltage on the power transistor match the average voltage of the capacitor?

Oh, and BTW, how much did it cost to get the PCB printed?

If you offer kits, I'm in for one. It's a lot easyer to buy a kit then design it your self on a crapy looking proto-board.
 
FYI there is no selling outside the classifieds, please don't do this again.

nikhsub1
Forums Moderator
 
Last edited by a moderator:
sir_pyro said:
Hum, nice... All it took was an opp amp, a power transistor, and some various capacitors and resistors? So what are you doing, feeing the PWM signal into that large cap through a resistor and using the op amp to make the output voltage on the power transistor match the average voltage of the capacitor?

Oh, and BTW, how much did it cost to get the PCB printed?

Not quite. The large capacitor is just to isolate the circuit from whatever the power supply is, so that it draws a steady amount of power. It is probably overkill and not necessary. The capacitor you are thinking of is there (C1), but it's a small surface mount one. It also performs double-duty in that it provides a feedback path from the opamp's output to prevent resonating (which would occur because the power MOSFET's gate has slight capacitance.)

I originally considered using a power transistor, but didn't. If I did, you couldn't approach 12.0 volts due to a transistors voltage drop. It is a P-channel MOSFET, which gets me about 11.7 volts on my water pump (1.5 amps). Note that a pump/fan's ground wire must be the true ground, or the tach signal wouldn't work. I've heard of this problem with other standalone circuits that use manually placed thermal probes.

The PWM voltage average varries between 0 and 5 volts, so you need to shift , invert, and scale it to control the P-FET.

While playing around with Prime95 and ATITool, I found that anything over 10.0 volts doesn't really improve the cooling ability of my setup. Of course, your mileage will varry.

PCBs cost me about $120.
 
dinofx said:
FYI there is no selling outside the classifieds, please don't do this again.

nikhsub1
Forums Moderator

Fine with me. It did not see that section. I assume we can still discuss the design of the circuit, its operating behavior in a water cooling system, and how it was constructed, can we not?

[edited]In fact, I still don't see that section, perhaps because it is blocked from my viewing it. It looks like I don't have access to the classified forum. If I could access it, I might encourage interested parties browsing that forum to perhaps PM me.
 
Last edited:
nikhsub1 said:

This product is not optimal (read "junk") for several reasons. First, it uses a manually placed temperature probe, and it requires manual tuning of POTs. Nowdays, your CPU, GPU, chipset, and even hard drives all report their temperatures to your motherboard, which has the ability to adjust its PWM signal to multiple fans/devices, without tuning/intervention from the user or even software. This product can not access any of those built-in thermal sensors, unless you upgrade and use some gigantic turd that contains a USB port, requires software setup, and costs lots of money. Secondly, I've heard that the tachometer signal does not pass through properly to your motherboard, so you can not monitor your pump or fans tach. Probably because they are using a slightly cheaper N-channel MOSFET. Finally, my solution is smaller/simpler at just 1"x1.5".
 
DrZaius said:
wow i was thinking of making somthing exactly like this after i had the same problem. if you decide to sell these as a kit or pcb send me a pm

No if he does plan to sell these, he won't PM anyone he will LIST THEM IN THE CLASSIFIEDS. If he or anyone for that matter, PM people with things to sell and we find out about it, you will be banned. There is a strict rule here, NO OFFERING ITEMS OR TRYING TO BUY ITEMS OUTSIDE THE CLASSIFIEDS. If you want to get banned and potentially get scammed and lose money, go for it.

nikhsub1
sorry about that wasnt really thinking
 
dinofx said:
Fine with me. It did not see that section. I assume we can still discuss the design of the circuit, its operating behavior in a water cooling system, and how it was constructed, can we not?

[edited]In fact, I still don't see that section, perhaps because it is blocked from my viewing it. It looks like I don't have access to the classified forum. If I could access it, I might encourage interested parties browsing that forum to perhaps PM me.
Yes you may discuss it to your hearts content. You can not see the classifieds yet because you need 100 posts to gain access to it. You will get there soon enough.
 
yeah you can't access the classified until you get 100 post and have a .edu or a ISP e-mail address.

About the two pin power, I knew it was feeding the circuit, I was just confused as connector for it. All I have are the molex connectors, unless there an adaptor to change a 4-pin to 2-pin.

or did you just take the wires out of the molex? <- which is probably what I would have done.

and one more noob question, I'm studying to be a computer engineer, so I have a little familarity with this stuff. In this circuit, did you try to use a npn instead of a pnp? (confused about what p-channel is)

edit - ^ nikhsub1 beat me to it
 
pcmaker401 said:
About the two pin power, I knew it was feeding the circuit, I was just confused as connector for it. All I have are the molex connectors, unless there an adaptor to change a 4-pin to 2-pin.

or did you just take the wires out of the molex? <- which is probably what I would have done.

Actually, "Molex" makes thousands of different types of connectors. These small connectors are called the K.K. "series". You can cut off the wires from one of your power supplies connectors (and you could always revert if you had the right crimp pins), or make an adapter from a spare "molex" y adapter, etc, which most people have lying around, and a 2-pin K. K. terminal housing. I was able to purchase 2, 3, and 4-pin housings and the crimp pins that go inside them online.

Only 87 more posts to go!!
 
There is a special price of $77 for 4 boards sized exactly 3"x4". But, because I needed each board V-scored so that I could break them into 8 circuits, it cost more. Technically, the order didn't qualify for the $77 special price, but they gave me a 1st-time customer price, otherwise it would have been $199.

www.pad2pad.com
 
Back