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

FANS FANS AND MORE FANS! Some Nifty Fan Mods Including a PMW fan controller.

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

Skeith

Member
Joined
Jun 28, 2005
Location
Winnipeg Manitoba
FANS FANS AND MORE FANS! Some Nifty Fan Mods Including a PWM fan controller.

Here is a guide to a couple fan mods I have built and tested and can say that they work very well.

In this guide I will Cover The following:

PAGE 1: Pulse Width Modulated Fan Speed Controller YOU CAN BUILD YOURSELF!!

PAGE2: A safe 8.5V Fan mod that allows RPM monitoring.

PAGE3: Restoring failing/seized fans and abnormal rotation.

Visit my website for more of my projects.
 
Last edited by a moderator:
PAGE 1: THE PWM FAN CONTROLLER YOU CAN BUILD!


I noticed allot of people like to use the 7V fan mod, this is fine I suppose but it is very ineficient, and it doesnt allow for RPM monitoring.

I have Built a fan controller out of fairly cheap and readily available parts that is extreamly efficient, reliable and doesnt endanger your computers PSU. I feel very sure that anyone who knows how to solder can build this device, whether they understand the electronic workings behind it or not.

For those of you who dont know what Pulse Width Modulation is and why it is more efficient than voltage drop speed controll read the following paragraph, If you already understand the concept feel free to skip ahead to the construction stage.

Pulse Width Modulation (PWM) is the practice of rapidly turning a motor/device on and off to controll its speeds. This is a very eficient speed controll method because it runs the motor at its rated voltage, thus more torque, more accurate speed controlls and lower starting speeds. Motors can be slowed by loweing their voltage but this in turn reduces the motors torque and in computer fans this could mean less CFM. But how does it work?
A generated frequancy from an oscilator is passed into an OP amp. In the op amp is where the actual Pulse width modulation takes plase. The OP amps threshold is set. (the threshold is the voltage on the frequancy that will trigger a logic High (+5v) on the op amps output. By changeing the threshold you change the length of the pulses and thus change the speed of the motor. The output from the Op amp is then sent to a power management device. Usualy a power transistor, which operates like a normal transistor but can handle higher current. The result from the power output is the motors rated voltage very rapidly changeing from off and on stages.

Constructuion:
The parts you will need are as follows
(1) 390 Ohm resistor (orange , white , Brown)
(1) 330K Ohm resistor (Orange, Orange, Orange)
(1) 10K Ohm resistor (Brown, Black, Orange)
(1) 0.1uF (Microfarad) Capacitor. (commonly stamped with the number 104)
(1) 10K Ohm Potentiometer (Vriable Resistor)
(1) TIP31 or similar power transistor (I used a TIP122)
(1) NE555 timer.
(1) IC PC Board multi-purpose - 417 (by NeXXtech)
(x) 2 pin connector. (these are used to plug the fan into the controller. I salvaged mine from an old motherboard. The type that are used for jumpers.) The number of 2 pin connecters depends on the ammount of fans you wish to run on the channel. I used 2 sets. (to connect 2 fans) keep in mind that the Tip31 can only supply about 2 Amps without a heatsink.

Where to get the parts?
All of these parts are available from Radio Shack (or The Source as its now known in canada). The resistors and Capacitor can be ripped off from an old circuit board if you wish. All of these parts should cost you under $8 if you buy them all. Personaly im cheap so I scavanged for the parts in my junk bin, all I bought was the NE555, the prototype board.

Here is a pic of the schematic I used.

Now for construction. Here is a picture of my first channel that I have complete.
DCP01968.jpg
I built this one first so that I could determine if this actually worked. It did! So the information following is explaining how to build a replica of the device. (I built a second identical channel next to the completed one so I could take pictures of the step by step precedure. And belive me there are quite a few).
Follow the pictures. If you followed exacly as they show you should end up with a working fan controller.


So now the procedure starts. First take the NE555 IC and pop it into place. Its pins should end up on either side of the two middle strips in the board. Not in them but in the holes next to them on the smaller perpendicular strips. The middle strips will become your positive and negative rails.Make sure the holes right across from the pins are free as they will be needed to solder in components. Solder it in place.
DCP01969.jpg

Now that the IC is in place you need to give it power. Locate Pin one on the NE555. It is marked with a dot. Use a wire to connect this pin to the bottom rail. This will be your negative rail. Connect pin 8 (the pin directly across from pin 1) to the top rail. This is now your positive rail. (you may want to lable them with a felt pen so you remember as you will be using them in future steps.
Picture

Jump a wire across the IC to connect pins 2 and 6 together. Then connect pin 2 and 3 together as shown with the 330K Ohm Resistor.
Picture

Next take the 10K Ohm resister and use it to connect pins 7 and 8 together.
Use a wire to connect pin 6 to the 3rd free strip from the IC as shown in the picture. Connect the 0.1uF capacitor to that strip and then to ground. (remember its the bottom rail.)
Picture
And from a different angle

next comes the potentiometer. on the row of holes nearest the edge is where you will want to mount it. Align the center pin of the pot with pin 5 of the NE555. Push it in (it might require a bit of force as most pots have larger terminals). Solder it in place. run a wire from pin 5 of the NE555 to the center pin of the potentiometer. Left are the two end pins on the pot. Connect one to the positive rail and the other to the negative rail.
Picture

Now the next step might be a little hard to see, so look carefuly. First use a wire to connect pin 4 of the IC to the positive rail. Then use the 390 Ohm resistor to jump from pin 3 over pin 4 and conect it to the strip next to pin 4's strip.
Picture

Next comes the power transistor. Bend the leads right where they change size at a 90' angle backwards (toward the back metal plate of the tansistor). Pin one will be placed into the strip attached to the 390 Ohm resistor. Pins 2 and 3 will go into the single holes. Pin one is the base of the transistor and is what will be driving it, aka turning it on and off. The transistor simply acts as a high speed digital switch.
Picture

now comes a bit of a tricky part. Turn the board over. you have to conect pin 2 of the transistor to pin 4 of the NE555. Bend a wire in an L shape and solder it to each pin. Since pin 4 is being supplied with 12 volts this is the shortest way to get the supply voltage to the transistor.
Picture
Soldered connection.

Now comes the 2 pin connectors to supply the modulated power to the fans.
put pin 1 of the connector (not that it matters theres only 2 and they dont have special shape lol) right across from pin 3 of the transistor on the edge strip. I installed 2 sets for 2 fans. the pins on the side nearest the transistor will be the fans positive. Mark it so with a felt pen for future referance.
Picture

Now a bit more tricky soldering. Turn the board over and using a piece of bare wire solder pin 3 of the transistor to the positive strip of the connectors as seen in the image.
Picture

Now take a wire and connect it to the negative rail and bring it into the other side of the connectors. (the side away from the transistor). Its the blue wire in the center of the picture. Just poke it through dont solder it yet.
Picture

Now what you have to do is bend it over toward the connectors and solder it the strip. (not the positive on or you will get a direct short) This gives the fans their negative power.
Picture

Finnaly solder a molex connector to the corresponding negative and positive rails to supply power to the board. Check over your completed device to make sure that everything is right so that you dont end up blowing something on your test run.

Once your confident that its ready to run connect a fan to it (observe polarity) , conncet it to a power supply (I dont recomend your good one in case something is wrong use something cheap like an old school AT powe supply as I did.) once the unit is pluged in turn on the power supply and adjust the potentiometer. You should see your fan change speeds. If they do then you have succesfully built a home made PWM fan controller.
Sure beats spending like $40 on one, plus the building experiance is fun and when your at a lan pary be sure to tell and show your friends... and enemys :p that your fan controller is home made and built by none other than you!


Here is an image of my fan controller. 2 independent channels capable of controlling 4 fans (and more if i add more connectors.)
Good luck with your project.

DCP01993.jpg

If you have any questions please let me know, also I would like to hear from those of you that attempt this project.

Monitoring Fan RPMs while PWMing.

Here is a simple way to do this.
Take a couple connectors from on old PC case (the ones used for the pwr switch/ HDD LED etc.) Cut and strip the wires about a 1cm. Tin the ends with a soldering iron. Attach a 2 conector one to the +12 and ground of the fan. Use the coloured wire for positive and the white or black wire w/e it may be for ground. Now take a 1 connector wire. If your like me and you didnt have one available you can use a 2 connector. take only one of the wires strip it and tin it. It goes into the signal wire of the fan. put a dab of hot glue to keep the wires from poping out. The power conector will attach to the PWM circuit and the other one will attach to the signal pin on the motherboard. If you used a 2 conector wire just make sure the signal line is in the correct spot the others dont matter because those wires arent attached to anything.
sine all the circuits in a computer are common ground this will work with only 1 single connection to the motherboard!
Here are some pics to help you visualize.
Fan Connections
Motherboard connection
Complete setup
Peace out.

A PROFESIONAL LOOKING PCB THAT YOU CAN ETCH YOURSELF.

Alright A little late but like I promised Here is the working completed PCB for a 3 Channel PWM controller. The file is a .PCB and can be opened with Circuit Maker 2000. the DEMO can be downloaded

HERE

Im not sure if this link is legal or not all I know is that it was at one time legal to download the 30 day trial for free from microcodes page. However they dont seem to have it on their page anymore since a release of a new software package.

The PCB can be downloaded

here

in a ZIP archive. If you have trouble opening the link try right clicking and save as, the service I am using for free hosting doesnt like direct linking.

You will also need

this kit

to make the board. The kit expects you to use the Sharpie as a etchant mask but what you will do is instead use circuit make 2000 to print out the PCB on a piece of paper. Copy and paste them (i should have done it for you) so you get more than one copy before printing. This is key in case you mess up on the following steps as a begginer probably would, as I did the first time I did it.

Once you have the image printed (I think I fit 4 to a sheet.) get it photocopied to a transparancie. Or If you have a LASER printer you can print it directly on a transparancy. It must be a LASER printer because of the heat transfer method used with the toner.

Next mesure and cut your circuit board to your desired size. It should fit just within a 5" drive bay if cut properly. Now use a one of those green dish washing pads. (a clean one ) to clean off the copper it will become nice and shiny. This is key for the ink to stick.

Now get an Clothes iron, an news paper and a clean sheet of printer paper.
set the Iron to Mid High heat and let it heat up.

Cut out one of the transparancy PCB templates for the controller. Look at both sides. The image on one side is shiny and the other is dull. The dull side is the side with the Ink on it. Keep it in mind.

Place the cut and cleaned ciruit board on the newspaper copper side up and place a piece of printer paper over it.

Now Iron it LOL for a couple minutes.

Take the paper off and get ready to put the transparancy on the copper INK SIDE DOWN. Align it by peering through the transparancy with the copper before placeing it down because when it comes in contact with the hot copper the ink will stick.

Place the blank paper on top of the board once again and then Iron it. Iron it for a good 20-30 Sec. Carefull though you dont want the plastic to melt or it will smudge the ink. Now quickly put down the Iron and SLOWLY and CAREFULY peel the plastic off the copper. It should leave behind the INK on the copper. If you feel up to it you can try peeling it as I do by keeping heat applied to it by pulling up on the transparancy and slowly moving the Iron across the copper with the transparancy as it comes off the copper board.

If you messed up let the board cool and re clean it and cut out a new template and start again.

If it looks decent let it cool, it doesnt have to be perfect as you can now use the marker to touch up and copper showing through the lines. Once ready follow the instructions in the Kit to etch the board in the solution.

I find that using a 50/50 mix of the Fehric Chloride and water works beautifully. Move the board around in the liquid exposing it to oxygen perioduclay (ever few seconds) the faster you can move the board the faster it will etch. Try not to spill the chemical as is stains and is corosive. (not that splashing it on my skin ever did anything but dont take the chance) I sugges reading the MSDS sheet for Fehric Chloride before using it. Once your finnished you can bottle the solution and save it for the future.

Next drill the hole for the components on your board. I belive the kit come with a bit. It is very small so be carefull not to break it. If the bit looks too big for the pads i suggest going to a surplus store (like princess auto if your in canada) and pick up a smaller one. A drill press or a dremmel drill press converter is highly recomended because if the bit is pressed at and angle it will break verry easily. Also dont force the bit, If you see it starting to bend your pressing too hard.

Good luck everyone!!!!

PS. Please leave my name in the PCB if you make one as it is my work.
Although I dont mind if you edit it your your own needs.

Peace out.
 
Last edited:
Page 2: SIMPLE AND SAFE 8.5v fan mod.

This is very simple to do. You hook up a 5 volt regulator to the 12volt rail. but instead of connecting the common on the regulator to ground like normal, solder a 51K Ohm resistor in between it and ground. This creates a Bias on the common terminal causeing the common pin to "float" above 0v. This tricks the voltage regulator into thinking that the voltage its being supplied is less than it is and so it doesnt cut it down as much. Try different resistances for different voltages. Maybe even a 100Kohm pot would do?

SAVE0035.jpg
DCP01997.jpg
(yes that is an old 200Watt AT power suply modded with thum screw terminals as a bench power supply, I am so cheap :clap: )
DCP01998.jpg
Also im sure your setup will be a little neater than this :p . Twas just quick and dirty to show you it works.

NOTE: due to the fact that the Vreg is biased if you were to hook it up to a different source than a 12V source you will get a different output. Because the resistor only floats the regulator on a fixed value the vreg will output a higher voltage at a higher input and a output a lower voltage at a lower input. For example with the 51K Ohm resistor in place it but attached to a 5V input instead of 12V it outputs 3.4V So if you are to run it on a lower voltage/higher voltage test it with different resistances and a potentiometer. Im thinking a 100K ohm potentiometer would do best, then it can be adjusted exactly how you want it, then a dabb of glue placed on the knob so it doesnt get bumped or messed up. Im working on a formula that will determine the required resistance. might take a while though and when its ready dont take its word test the output voltage it to make sure!
happy modding :p
 
Last edited:
Have a fan thats beginning to seize, has trouble starting or is spinning abnomaly. Well here is a simple way to fix almos 90% of those problems.

First turn the fan to the side that houses the electronic circuit board. This is also the size that houses the berrings.
DCP01932.jpg

Now pull off the lable. If it wont come off as mine wouldent just cut through the center of it as shown.
DCP01934.jpg

Exposed is the berring. (some fans have a little rubber stopper to keep the berrings clean but most do not.) Now all you need to do is place a few drops of 3 in 1 oil in the berring.
DCP01935.jpg

Now connect it to a power supply and run it for about 5 minuted. You should imediately see the fan start to retore its origional state. While the fan is running the oil seeps down into the rest of the berrings and lubricates them. After 5 - 10 minutes shut off the fan. add some more oil and put some tape over the hole. (after puting in the stopper if yours had one). There you go one restored and happy fan.
DCP01938.jpg

I did this to my power suppy fan when it started to slow down and let me tell you its bee working ever since. Sometime lubing up brand new fans can give you a gain in a couple hundred RPMs. not much but the CFM is woth it.
 
Nice dude! This is my first time to say something deserves a sticky! A BIG PLUS is the fact that you have awesome pictures!
 
Captain Slug said:
If anyone is trying to find a good oil for fans I recommend sewing machine oil.
Yeah tis probably the best bet, but I didnt have any. 3 in 1 seems to work decently though.
 
this is quite a fantastic guide! I might find the motivation to try this out. i bet some pics from the scope on the output would be interesting and help to show how it works. im also curious as to how clean of a square wave the output would be.

note that a switching power supply is basically a PWM with some filtration thrown onthe end so the voltage smooths out and averages between the high and lows of the output. they just usually monitor output and change their pulse width to change voltage.

I;ve been pondering a method to make RPM monitoring work on a5V mod. the idea is that both the fan and the mobo are holding that pin at logic high. so if you were to put a resistorbetween ground and the RPM monitoring pin, it would pull the pin to logic low while the pulse from the fan wasnt going, and when the fan pulsed, it would be pulled to logic high. the pulses would work pretty much backwards from normal, but it should still time them accurately nonetheless. or at least thats my idea. problem is figuring out what resistance would be needed to let the fan still be able to pull it to high, but still keep it low enough to be counted as low. it might be possible to doby removing the resistor on the mobo thats to one of the rails, 3.3 or 5, that holds the pin high when the fan isnt grounding it sending it low. I'll probably play with it next time i have my case open, today i stuck one of the fans on the 7V and didnt think to unplug from mobo, and later realised it all after i had it all back together and in its place and runnin.

hope my rambling here makes sense, and was a little productive... haha.
 
four4875 said:
I;ve been pondering a method to make RPM monitoring work on a5V mod. the idea is that both the fan and the mobo are holding that pin at logic high. so if you were to put a resistorbetween ground and the RPM monitoring pin, it would pull the pin to logic low while the pulse from the fan wasnt going, and when the fan pulsed, it would be pulled to logic high. the pulses would work pretty much backwards from normal, but it should still time them accurately nonetheless. or at least thats my idea.

I read somewhere that every 2 rotations the fan pulls the signal line to low. So if this is true as long as the fan's ground is kept common it should work. Thats what my 8.5V Vreg mod is trying to overcome. I havent actually tested it on the motherboard to see if the RPMs are still readable though. In theory since all im doing is biasing the Vreg the fan should still allow the RPMs to be read because its ground line is still common.
 
Oh I just encountered a slight problem, but I found the fix. Becuase I have two PMW controllers on one input rail they were kinda interfering with each other because of the spikes and drops of the switching. Fixed it by putting a 1000uF cap on the Input power rail. Tis all good now.

EDIT:
Here is an updated picture of the controller with the 1000uF cap smack dab in the center on the power rails. The capacitor absorbs the voltage fluctuation on the rail caused by the rapid switching. I should have known that I would need this lol. Once again I liberated it from the PCB of an old computer monitor. (I just cant bring myself to throw anything away :shrug: )
DCP01999.jpg
NOTE: If you are only building one channel you shouldent need this capacitor, but if you are runnin into trouble with stable speeds or some other strange activities then perhaps try a cap on the power rail. If you are going to run 2 Channels then deffinately install a capacitor on the power rail.

Also I am working on a screen printable PCB template for the controller that will feature 3 Channels with one fan connector each. I will post the PCB template when I am finnished.
 
Last edited:
Skeith said:
I read somewhere that every 2 rotations the fan pulls the signal line to low. So if this is true as long as the fan's ground is kept common it should work. Thats what my 8.5V Vreg mod is trying to overcome. I havent actually tested it on the motherboard to see if the RPMs are still readable though. In theory since all im doing is biasing the Vreg the fan should still allow the RPMs to be read because its ground line is still common.


I too read that, and my thinking is that the mobo is normally pulling the pin high, and the fan drops it low on its pulses. but with 5V instead of ground, the fan is also pulling it high. so all we need to do is pout in something to pull it low. so a resistor to ground would be enough to over come the mobo holding it high while the fan isnt on its pulse, but when the fan pulses it would be enough to pull it high again.
 
four4875 said:
I too read that, and my thinking is that the mobo is normally pulling the pin high, and the fan drops it low on its pulses. but with 5V instead of ground, the fan is also pulling it high. so all we need to do is pout in something to pull it low. so a resistor to ground would be enough to over come the mobo holding it high while the fan isnt on its pulse, but when the fan pulses it would be enough to pull it high again.
Yeah sounds like it would work, only thing is that the resistor might put stress on the motherboard and sensor etc. I think the biased regulator would probably be the best bet, just personal preferance If I were to use a volt drop speed mod thats what I would do. But im sticking to PWM. Also some motherboards have a built in PWM on their fan ports but they dont take advantage of it. And as far as I know none of the software for bus monitoring supplied by the mobo manufactuers interfaces them either. speedfan will allow you to change the Pulse rates on a mobo with PWM onboard. I know that my NF7-S v2 has them onboard, I dont use them though. All my fans are on full power, I dont care about the noise. I think of my computer as a muscle car, the louder the better.
 
I have designed and etched a board for the fan controller. It has 3 Channels each controlls one fan. And it fits within a 5.5 inch drive bay. I will post the template when I know forsure that it works. I am currently waiting for the potentiometers as they are on order, so you can see 3 sets of 3 pins at the front with nothing in them. I also have to make a few modifications to the PCB because the heatsink got in the way of the filter cap. ( I had to drill in extra holes to make it fit) That way heatsinks can be added if they are wanted, no that they are needed. After the modifications are done and the testing is complete I will post the template and the PCB file.
Anyway here are some pictures at its current state.

DCP02012.jpg
DCP02015.jpg
 
great thread! I vote sticky! Info like this is very valuable for people around here.
 
Temporarily stuck, unless the mod of this section objects.
 
Nice, that PCB you got there its pretty cool, i'd do it myself, but Karma is kinda gettin back at me right now, so i dont feel safe messing with corrosive fluids
So i think i'll try it like the first PWM controller you have up there
I plan on making me a trip to Radioshack soon now ;P
Also, would there be anything bad to happen if i didnt put heatsinks on the transistors, will they overheat and mess with the speed?

awsome it is
-prince

*goes to his old taken apart keyboard to check resistors..hehe :p

edit: nvm about the heatsink part, i didnt read thoroughly, sry

I was wondering, could i get the input 12v rail and com from the motherboard fan header so i can stream the RPM detection wire to the mobo fan header and it still work and detect RPM?
 
Last edited:
Heres an update on the biased Vreg mod!
IF YOU ARE USING THIS MOD PLEASE READ THIS!!!!!!!

due to the fact that the Vreg is biased if you were to hook it up to a different source than a 12V source you will get a different output. Because the resistor only floats the regulator on a fixed value the vreg will output a higher voltage at a higher input and a output a lower voltage at a lower input. For example with the 51K Ohm resistor in place it but attached to a 5V input instead of 12V it outputs 3.4V So if you are to run it on a lower voltage/higher voltage test it with different resistances and a potentiometer. Im thinking a 100K ohm potentiometer would do best, then it can be adjusted exactly how you want it, then a dabb of glue placed on the knob so it doesnt get bumped or messed up. Im working on a formula that will determine the required resistance. might take a while though and when its ready dont take its word test the output voltage it to make sure!
happy modding :p
 
Back