My fan-voltage-contoller-strip

Control: management of fans, temp/rpm monitoring via soft/hardware

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
cmcquistion
Posts: 278
Joined: Sun Aug 10, 2003 6:05 am
Location: Tennessee

My fan-voltage-contoller-strip

Post by cmcquistion » Fri Jul 02, 2004 3:48 pm

A couple years ago, I read this article on building a "Diobus", which is basically a rheobus for voltage control, built using diodes, to drop voltage. I built a couple of these and have used them in various computers.

My home computer, today, is an Asus P4P800-E Deluxe motherboard and P4 3.0C (overclocked to 3.6 GHz.) I wanted to use the motherboard's Q-Fan control to lower the speed of the Zalman CNPS7000AlCu heatsink and the Evercool AL12025 exhaust fan, lower than 5V. The Fan-Mate voltage controller, provided with this heatsink, only allows voltage down to 5V and it, somehow, defeats Q-Fan control. Speedfan (software control over fan speeds) is also defeated by the Zalman Fan-Mate. It keeps them from lowering the fan speed any more, so you are stuck at 5V.

I knew that the Zalman heatsink and the Evercool fan could both operate at lower than 5V and still do an excellent job, so I started considering some other options. I remembered the Diobus and it intrigued me, because it works off a different principle than the Zalman Fan-Mate and I could hook it up so the RPM wire passed through. The idea was to lower the voltage with a Diobus and then let Q-Fan lower it, even more, dependent on temperature.

I wanted the machine to be very quiet, but I didn't want any switches or knobs jutting out of the case and I didn't want there to be an easy way to accidentally turn off the fans, or anything. I decided I would use the Diobus technique, but without a rotary switch.

I built a fan-voltage-controller-strip (FVCS), and this is how:

I went to Radio Shack and bought a Terminal Strip and a diode assortment, which included several IN4001 diodes (along with some higher-rated diodes). I bent the diodes into "U" shapes and clipped their legs a little shorter. I then jammed them into the holes of the terminal strip, such that they were in a daisy-chain arrangement, and then tightened down the screws.

I cut up a couple fan extension cables to supply voltage to the fan-voltage-strip and have some headers to plug my fans into (so I wouldn't have to clip the fan wires of my Zalman heatsink or Evercool exhaust fan.) Supply voltage goes in the end, before the diodes. In this arrangement, each diode drops the supply voltage by about 0.75V. Since they are daisy-chained, you can hook your fans into the strip at any point and get voltage from that point. For instance, if the supply voltage is 12V and you hook a fan in after the first diode, then you will get about 11.25V. If you hook a fan in, after 8 diodes, you will get about 6V.

I built my voltage controller strip with 9 diodes and then started experimenting with where to plug in the fans, so they had adequate startup voltage, but didn't completely stop when Q-Fan turned them down to their lowest point. In my system, with this heatsink, 8 diodes seemed to be the sweet spot for the heatsink and 7 diodes seemed to be the sweet spot for the Evercool fan. The RPM wire of the Zalman heatsink is passed through the FVCS, but the RPM wire of the Evercool is not. Therefore, the BIOS gets RPM information from the Zalman, which I used because it is the faster (higher RPMS) of the two fans.

After I was finished and the system was running, I measured the voltage being supplied to the Evercool 120mm aluminum exhaust fan and the Zalman CNPS700AlCu heatsink, with my FVCS and Q-Fan. At full speed (startup), the supplied voltage is 11.63 volts, going in to the FVCS. After 7 diodes, the Evercool fan is getting 5.95V (it needs around 4.5-5V to start) and the Zalman heatsink (8 diodes) is getting 5.17V (I think it needs about 4V to start.) Since Q-Fan is turned to 12/16 in the BIOS, it slowly turns down the supplied voltage to the CPU fan, depending on CPU temperature (which is generally very low in my system.) After a few minutes, it has dropped the supplied voltage to 9.29V. Now, the Evercool is getting 3.98V and the Zalman is getting 3.24V. When I had the Evercool hooked up to the same location in my diode chain as the Zalman (8 diodes), it was stopping, when it reached the lowest power. This is why I moved it up the chain one notch (7 diodes) and I have Q-Fan set to 12/16. With Q-Fan at 11/16, the Evercool would occasionally stop at the lowest voltage. Not always, but often enough that I decided to just bump Q-Fan up to 12/16, and fix it, that way (instead of moving it up the diode chain, again.)

When the system is under a high load, Q-Fan bumps up the fan speed, just enough to keep that CPU temp cool (around 50-55C). This is transparent to me and I generally don't even notice. If you don't have an Asus motherboard with Q-Fan control, you can get the same results, using Speedfan if it supports your motherboard. A few other motherboard manufacturers provide similar CPU fan control, but I'm not familiar with any, but Asus' Q-Fan.

I put some velcro on the back of my FVCS and velcro on the back of my power supply. It is kept it there, out of the way, and I can easily get it down, if I need to fiddle with it.

Here are some pictures, which can probably explain this contraption better than my instructions:)

This is an overhead picture of the FVCS. It shows the terminal strip from above. From this view, the diodes are on the bottom and the power wires on coming out of the top. The far left is the power-in (this is hooked up to a cable going to the CPU fan header.) The next one over is the voltage going to the Evercool fan (after 7 diodes.) The next is the voltage going to the Zalman CNPS7000AlCu heatsink (8 diodes.) The far right (next to last terminal) is hard to see, but I have all of the common ground wires plugged in, there. There is NOT a diode going to that terminal, of course. I just needed a place to keep the grounds, together. The big advantage of using a terminal strip is that everything is removable and flexible. There is no soldering required, but it is all held securely and if you need to add, remove, or change something, it is a piece of cake.
Image

This is a side view of the FVCS, showing the diodes in their "daisy-chain". This side should probably be covered with some tape, or otherwise isolated, so none of the metal leads from the diodes can accidentally touch any metal part of the chassis. This could start a short. In my case, I have the FVCS attached and isolated with Velcro, which you'll see, below.
Image

This is an angled view. It is very fuzzy, but it shows the diode chain a little closer and from a different angle.
Image

This is a view, showing the opposite side, where all the voltage wires and the ground wires are screwed into the opposite side of the terminal strip, from the diodes. (When I took the picture, I chopped off the left side of the terminal strip, a little bit, so you can't see the first red wire (power-in).)
Image

This is a picture of the case, showing the FVCS velcroed to the power supply. You can see the heatsink, exhaust fan, and PSU. These, and the Arctic Cooling VGA Silencer, are the only fans in the system. The front intake is open and allows air to be sucked in the front, across the hard drive (cooling it) and exhausted. I've built several computers with this case, in this arrangement, and it has worked GREAT!
Image

This is the final picture, showing my whole case.
Image

The case is modded for soundproofing. I build a hard drive suspension rack and the power supply (Fortron FSP-300) is modded for 5V, which makes it incredibly quiet. You can find the instructions for the hard drive suspension rack here, the instructions for the case mods here, and the instructions for the FSP-300 mods here.

Hope you enjoy this and can take advantage of this incredibly cheap, flexible, and effective fan-voltage-controller-strip. I've been meaning to write this up and take some pictures for over a year, since the first one I built for a noise-concious customer. I've built several since then, and this one is the most recent.

(P.S. I can get away with hooking two mid/high power fans up to one motherboard fan header, because their voltage is reduced, which reduces their current draw, so their combined current draw is relatively low.)
Last edited by cmcquistion on Wed Feb 01, 2006 6:11 pm, edited 2 times in total.

Ralf Hutter
SPCR Reviewer
Posts: 8636
Joined: Sat Nov 23, 2002 6:33 am
Location: Sunny SoCal

Post by Ralf Hutter » Sat Jul 03, 2004 6:02 am

Very nice! Three thumbs up from Ralfie!.

markjia
Posts: 171
Joined: Sat Apr 24, 2004 4:31 pm
Location: Vancouver BC

Post by markjia » Sat Jul 03, 2004 3:26 pm

So are all the diodes IN4001? What about using resister in place of the diodes?

lenny
Patron of SPCR
Posts: 1642
Joined: Wed May 28, 2003 10:50 am
Location: Somewhere out there

Post by lenny » Sat Jul 03, 2004 10:52 pm

markjia wrote:So are all the diodes IN4001? What about using resister in place of the diodes?
With diodes you know you're getting a fixed 0.7V drop for every diode. If you use resistors, the behavior is different for different fans.

cmcquistion
Posts: 278
Joined: Sun Aug 10, 2003 6:05 am
Location: Tennessee

Post by cmcquistion » Sun Jul 04, 2004 5:56 am

markjia wrote:So are all the diodes IN4001? What about using resister in place of the diodes?
They are all IN4001 (or higher) diodes. Using higher rated diodes provides the same voltage drop, they are just rated for higher voltages (overkill).

There are no resistors. Using resistors is a totally different principle.

silvervarg
Posts: 1283
Joined: Wed Sep 03, 2003 1:35 am
Location: Sweden, Linkoping

Post by silvervarg » Mon Jul 05, 2004 2:21 am

Neat and simple solution that does the job. Great work man!

Just in case you want to try this yourself I would like to correct some minor facts.
The diods are not named IN4001, they are named 1N4001. (The first "1" is to tell you that there is one junction=diod, where a "2" would make it a transistor, e.g. 2N2222).

Not all diods has a voltage drop of 0.7V. The 1N4001 is rated with 1.0V drop at maximum current (1A). The voltage drop is close to constant for other currents.
For instance a 1N4004 is rated for 1.1V drop, so using different diods is an option for fine tuning. Common diods usually have 0.7 - 1.3 V drop. Most of them have 1.0-1.1V drop.

cmcquistion
Posts: 278
Joined: Sun Aug 10, 2003 6:05 am
Location: Tennessee

Post by cmcquistion » Mon Jul 05, 2004 5:03 am

Thanks for the corrections. I've only used diodes from 1N4001-1N4003 and they all appeared to have the same voltage drop, but it is good to know that the higher rated diodes will probably have slightly higher voltage drops

Also, I would like to point out something that I didn't really cover in my instructions, which is that diodes must be oriented properly, because they are designed to allow current to pass in only one direction. If you turn one around wrong, you won't get any current through it, at all. The side of the diode with the silver band around it is the "arrow" that tells you which way current can flow. Make sure that they are all facing the same direction, in the direction of lower voltage. My pictures, above, probably aren't clear enough to see, but the diodes are all pointing to the right, which is the side where the fans are hooked up. The power-in comes from the left side.

Just thought I should mention that, for anyone not familiar with working with diodes.

cpemma
Posts: 351
Joined: Sun May 11, 2003 2:31 pm
Location: S Yorks, OK
Contact:

Post by cpemma » Wed Jul 07, 2004 2:52 pm

silvervarg wrote:Not all diods has a voltage drop of 0.7V. The 1N4001 is rated with 1.0V drop at maximum current (1A). The voltage drop is close to constant for other currents.
For instance a 1N4004 is rated for 1.1V drop, so using different diods is an option for fine tuning. Common diods usually have 0.7 - 1.3 V drop. Most of them have 1.0-1.1V drop.
Diode voltage drop follows a nice smooth loggish curve with current, shown on the Fairchild 1N4001 datasheet. For silicon types from around 0.6V at uA levels up to about 1V at max continuous rating, so a very steep curve, unlike resistors where the volts dropped is directly proportional to fan current so will change a lot with a change of fan loading.

For a typical 150-200mA fan load the drop is around 0.75-0.8V with 1N400x. For finer jumps the 1A schottky diodes (eg 1N5817/8/9) are ideal, around 0.35V drop at 200mA.

Post Reply