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

FEATURED Building PWM Controller for 4 wires PWM fan

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
Glad to see you now understand 3 vs 4 wires fan, this is the best choice when it comes to controlling & monitoring the fan performance. 4 wires fan is new & better design and more advanced.

About your PCB routes, a minor corrections , deleted connection at pin 11 and added route to pin 4.

View attachment 82359

Also, I couldn't find a 680 micro Farad, so I will use the 470 for now and order a 680.

Its PICO Farad (pF), not (uF)micro Farad, the one you want to order is 470 pF or 680 pF, not 470 uF or 680 uF.

Watch the suffix between p (pico) vs u (micro), that pico Farad size is 1 million times smaller micro Farad.


.
 
Last edited:
Hrm.... I kinda thought it was pF, but I didn't see any pFs so if it is 1,000,000 times smaller then that would be .00047 uF?

Can you recheck my circuit bing? I edited it before you corrected it.

PWM%20Circuit.PNG
 
Last edited:
Interesting...


I was actually thinking about doing something similar last year, except with switches/motorized pots and normal 3-pin fans. I used that type of 555 circuit in an fuel injector cleaner/tester I made quite a while ago, which pulses them at ~50Hz from 0-100% duty cycle. Then I realized you could probably do the same to fans. At the time I was fed-up with how fast fans were still loud at 7 volts, and that quiet fans [at 7 votls] flowed poorly.

The main reason I came here was to see if anyone was doing work with PWM stuff. Since the PWM line won't require amps like the power wires, couldn't you possibly just wire the fan's control wire to the mobo's header control pin? I was thinking about hooking 2-3 fans to the signal fin, but feeding them "external" power via molex. They make y-cables that let you hook 2 pwm fans to the cpu fan header, and the only issue is that it can report weird rpm's if both tach lines are hooked up. ATM I can't pull any info up on that fan, so all I could guess is that it's incompatable with anything but a 16kHz signal.
 
I have the Asus Maximus III Formula, which actually has 7 - 4 Pin PWM Headers. The problem is, 5 of them only allow 50% duty cycle. The CPU and PWR Fan Headers allow 0-100% Duty Cycle, but only on Manual setting. Using the pre-configured settings, 20% is the lowest. In addition, it raises the PWM in steps. 10-20-30- etc. up to 100%. By doing this it is not very fluid. By building this circuit, it will allow me to actually set it as needed. If Speedfan every picks up support for the Asus MIIIF, then maybe there will be a viable software alternative, otherwise I think that manual control will be the way to go.
 
Miah, I think you should consider to invest on good and beast like PWM fans like this one, equipped with controller, imo they're really handy at benching session where you can rig up multiple of these to blow off the condensation.
When in use for condensation control, the fans need to be on 100% all the time anyhow. :-/

This idea only really makes sense to me for air/water cooling ;)
 
Hrm.... I kinda thought it was pF, but I didn't see any pFs so if it is 1,000,000 times smaller then that would be .00047 uF?

Can you recheck my circuit bing? I edited it before you corrected it.

Yes, you're correct, 470pF equal to 0.00047uF or 0.47nF, any type of cap will be ok since this circuit doesn't need a very stable frequency, a slight drift at the freq won't hurt the performance.

They're really cheap at digikey for example -> HERE or HERE , again, if you're going to purchase it, buy both 470pF and 680pF.

I'll try my best to check that pcb's route, though that will take a while.

Are you sure you don't want to breadboard the circuit 1st ? Since its quite simple, it should be quick, prolly 30 minutes to 1 hour if everything is correctly routed, it will be up & running.


Since the PWM line won't require amps like the power wires, couldn't you possibly just wire the fan's control wire to the mobo's header control pin? I was thinking about hooking 2-3 fans to the signal fin, but feeding them "external" power via molex. They make y-cables that let you hook 2 pwm fans to the cpu fan header, and the only issue is that it can report weird rpm's if both tach lines are hooked up.

Yes, that is possible by grouping/connecting all fans PWM wire together, and then connect to the PWM source like from the mobo or from the controller. That is the other advantage of PWM fan compared to legacy 3 wires fan.

About the weird rpm, of course you should split it and connect each fan's rpm wire to each own mobo's fan header at the RPM pin, so it will be read separately. Cause connecting RPM wires together will just give you garbage signal since the physical fan rotations and the signals are not synchronized together, anyway its almost impossible to have the same speed even from the same type of fan with exact same PWM dutycyle.


ATM I can't pull any info up on that fan, so all I could guess is that it's incompatable with anything but a 16kHz signal.

Looking at Brutal's result using that 16Khz fan at his mobo's PWM signal that is at 25Khz, I think it is working fine.
 
Last edited:
I took your recommendations and picked up a breadboard. I can see the logic, but I guess I was just looking to cut costs. The breadboard was $10 and the Wire for making the circuits was another $5, but hey... if I have it, I will use it. I found a good electronic store with good prices, so the stuff from Radio Shack can go back. As far as the timer goes, he didn't have any TLC556 in stock, but he did have TLC555. You said I could use both, but I will need an idea in the differences from the schematic you built for me. I will post a screen of my breadboard when I am done.

Question: does breadboard circuit link with the spot next to it? I have never used one before, so I guess I will be doing some more reading.
 
DSCN0565.JPG


This is what I have so far. I am a little confused, because the Schematic with the TLC555 does not have a 10K Resistor and the TLC556 Circuit does.

I am using a TLC555, so if you could help me out, I would appreciate it. :)

I hope you like my progress. Its my first usable circuit, other than preschool light circuits and such.

My first integrated circuit.
DSCN0568.JPG


DSCN0569.JPG


This is based off of the first Circuit, with the TLC555 and only a 1K resistor. Please let me know if I need to change anything.
 
Last edited:
DSCN0570.JPG


DSCN0571.JPG


DSCN0572.JPG


Updated Circuit. Remember, according to the picture with the TLC555. Input please.
 
Great ! :thup:

About the 10K resistor, that is for the double 555 configuration, so for this single 555, you don't need it !

A quick spot on, that power conection from molex is wrong, the positive is connected to the +12 volt line, it should be on the +5 volt line, becareful. From the last pic, that positive should came from the right most hole at the molex.

Btw, that brown 680pF (labeled 681J3) is a real good cap for a stable frequency ! :thup:

I'm currently working, give me few hours, I will try to check it out & post if there is any correction !
 
Great ! :thup:

About the 10K resistor, that is for the double 555 configuration, so for this single 555, you don't need it !

A quick spot on, that power conection from molex is wrong, the positive is connected to the +12 volt line, it should be on the +5 volt line, becareful. From the last pic, that positive should came from the right most hole at the molex.

Btw, that brown 680pF (labeled 681J3) is a real good cap for a stable frequency ! :thup:

I'm currently working, give me few hours, I will try to check it out & post if there is any correction !

DSCN0578.JPG


DSCN0579.JPG


DSCN0580.JPG


Corrected Molex connector. That other one was one i had laying around, I thought it was wrong, but was going to see if you corrected it.

Added GND connection. Will await your input. :) I am getting excited. I hope putting this on the PCB goes real smooth. My soldering skills are not that great. But I think I can make do.
 
Well, at my best looking at those routing, I think you're set.

About the 3 pins fan header you're using, I assume you're aware that on 4 pins connector, the PWM is the right most (blue cable) like below pic.

14403d1117682408-psu-configuration-4pin_connector_002.jpg

Before powering it up, again, suggest you to re-check one last time, and if you have any stock Intel HSF fan which is also PWM fan, use this 1st, instead of hooking up to that expensive Sanyo Denki fan.

Fyi, Intel stock HSF fan even at 0% duty cycle, will never go down to 0 rpm, it will turn only at slower rpm, prolly below 1000 rpm if I'm not mistaken.
 
Ok, here are the results so far.

I had to hook up a bridge for the ground side because on the breadboard there is no ground connections between each set of 5 holes.

Next, I hooked up a PWM Fan, The fan runs without the circuit hooked up on 12V.

When I hook the circuit into it, I can control the fan, however, I measured voltages from the output signal side and I am getting from 0.65V up to 3.56V as I turn the potentiometer. This isn't right correct? I should have around 5V the entire time, pulsing On/Off. Now I don't have my good multimeter here, so I know the PWM signal can be hard to read, but Its not right correct? Also with the fan, the speed does not go to zero, and although the fan speed changes, its like it only has two speeds. Granted its only a 1500 RPM Fan, but should that make a difference?

DSCN0581.JPG
 
Last edited:
The 3.56 is a average of the voltage I'm sure. To see the peak voltage you'll need an O-scope and view the waveform. But you should see close to 12VDC if the PWM is working right. Whre is the meter connected? Should be to the red/black.
 
Brutal,

The PWM out can not be measured with plain DMM since its switching between 0 and 5 volt real fast, so what ever your measurement will not be accurate.

Now, before trouble shooting the circuit it self, try this to verify if the fan's PWM signal is working right, looking at the last pic, that Blue wire is the fan's PWM right ?

Connect the power (red) and negative (black) wire as usual to the 12 volt to power it up, and then connect that blue wire to the negative, this is equal to 0% duty cycle, see how fast the fan rotates, once noted, then move that blue wire to +5 Volt (becareful, do not accidentally connect to the +12 volt cause it will toast the fan's internal circuit instantly). Once connected at +5 volt, this is equal to 100% duty cycle alias full speed, take note on the speed as well.

Back to the circuit, see if at least the same as the min and max speed when connected on the controller, you mentioned its like two speeds only, this circuit is not supposed to work like this, it should give you a smooth transition when you dial the pot.

Now on the circuit, using themultimeter, on capacitors at those blue, brown or red one, they should be read by the multimeter like they're not connected between two legs aka not conducting at all, while resistor just measure it, if it within 10% from 1K ohm, it should be fine. Also, do you have any other 555 IC around ? Try to replace it if you have any spare.

I did recheck again all your wires outes & connections at the breadboard and still can't find any mis-wired.
 
Last edited:
Yes the Blue wire is the PWM signal wire. I was using a crappy Radio shack handheld meter. I will be bringing my Fluke 87V (True RMS) home tonight to recheck, I know it does PWM for sure. I am pretty sure the circuit is working right, but I am using a xigmatek fan that might be working like the way you said an Intel fan works. Barring anything else I can bring it to work tommorrow. The radio shop has a DSO, so I can hook it up there. Also gonna try it on my San Ace, perhaps the smaller spectrum of the Xigmatek doesn't show up. I did get two TLC555s, so I will try that out.

[Q]Now on the circuit, using themultimeter, on capacitors at those blue, brown or red one, they should be read by the multimeter like they're not connected between two legs aka not conducting at all, while resistor just measure it, if it within 10% from 10K ohm, it should be fine. Also, do you have any other 555 IC around ? Try to replace it if you have any spare.[/Q]

10K ohm? The first schematic shows 1K. Maybe thats the problem?
 
I feel like this should be right, but w/o the Ocilloscope, its hard to tell visually. I am pretty sure my Fluke will figure this out. ;) I should have brought it home to begin with, but I really use it a lot at work. Thats why I have a cheapie at home. I have actually looked at a digital graph meter. I also have a auto scan tool that shows voltages in a graph, but I can only use it on cars and it doesn't have a place to hook up probes. :(
 
Back