1. Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

So you want PWM control of your 3-pin fan?

Discussion in 'Cases, Modding & Electronics' started by Lazzer408, Feb 21, 2010.

  1. Lazzer408

    Lazzer408

    Joined:
    Jan 6, 2007
    Messages:
    2,547 (0.88/day)
    Thanks Received:
    338
    Location:
    Illinois
    Ok just trying to get it straight. You don't need to "boost" the PWM signal. What is your reason for the 556?
     
  2. Velict New Member

    Joined:
    Jul 12, 2014
    Messages:
    24 (0.15/day)
    Thanks Received:
    1
    556 is to boost the PWM from the mobo to the 10 fans
     
  3. Lazzer408

    Lazzer408

    Joined:
    Jan 6, 2007
    Messages:
    2,547 (0.88/day)
    Thanks Received:
    338
    Location:
    Illinois
    Then either I'm misunderstanding what your doing or your misunderstanding how the circuit functions. You don't need to boost the PWM signal. Intel's PWM spec is more then enough to sink gate drive.
     
  4. Velict New Member

    Joined:
    Jul 12, 2014
    Messages:
    24 (0.15/day)
    Thanks Received:
    1
    Sorry, i'm not explaining it all too well. Okay, so i've heard of issues where the intel pwm spec is not enough to drive more than 5 pwm fans at one time. Not power, but the PWM itself. The fans are powered through molex or an 8 pin adapter or something custom. http://www.overclockers.com/forums/showthread.php?t=700653
     
  5. Lazzer408

    Lazzer408

    Joined:
    Jan 6, 2007
    Messages:
    2,547 (0.88/day)
    Thanks Received:
    338
    Location:
    Illinois
    It is true that the PWM signal has a limit but this circuit only uses the PWM signal to switch the mosfet's gate on and off. What's on the other side of the mosfet is invisible to the PWM signal from the board. You can connect as many fans as you want as long as the total current draw is less then 50% of the mosfet's rating -and- you keep the mosfet within it's thermal limit.

    The way they are buffering the PWM signal (via. 556) is overly complex. You could do it with 2 transistors and two resistors. I only result to IC solutions when a circuit is to complex to make a discrete solution fit the task -and- a production IC dedicated to the task is available.
     
    Last edited: Jul 18, 2014
  6. Velict New Member

    Joined:
    Jul 12, 2014
    Messages:
    24 (0.15/day)
    Thanks Received:
    1
    ...really. I had no idea. Is this because it's not multiple PWM fans? It's only going to drive one mosfet, not multiple?
     
  7. Lazzer408

    Lazzer408

    Joined:
    Jan 6, 2007
    Messages:
    2,547 (0.88/day)
    Thanks Received:
    338
    Location:
    Illinois
    Correct. Multiple PWM fans running off a single PWM signal can cause issues because the PWM spec calls for very little current on that pin. Spec calls for a PWM fan to provide no more then 5.25v at no more then 20ma of current (if I remember right). This is probably used for detecting the presence of the fan and limited to 5.25v in order to communicate with logic-level components on the board. The 20ma current limit is likely due to the device switching the signal being very small like a surface mount transistor or an IC. If you were to parallel, say, 5 PWM fans, the combined current could be too much for the device to switch and the device would be unable to sink (ground) the signal.

    A board simply grounds PWM pin (in short pulses) to slow the fan. The resistor in my mosfet circuit pulls the mosfet gate high (to turn it on) but only as high as the zener will allow. When the motherboards PWM pin ground the gate, the mosfet turns off. So all it's really doing is buffering the PWM signal to a level capable of driving fans directly.

    It's really a pain in the ass that Intel made it as complicated as they did. It uses much fewer components overall if they had just low side switched 3-pin fans. Some boards can control speed on 3-pin headers.
     
    Aquinus says thanks.
  8. Velict New Member

    Joined:
    Jul 12, 2014
    Messages:
    24 (0.15/day)
    Thanks Received:
    1
    That is a very good explanation, thank you. Also, yes I wish more motherboards had 3 pin control. This whole project is to prepare for socket 2011-3 and x99 chipset.
     
  9. Velict New Member

    Joined:
    Jul 12, 2014
    Messages:
    24 (0.15/day)
    Thanks Received:
    1

    Hey man! Parts came in. I built it today, tested it on one fan (still waiting on headers)
    So what should I be looking at running it without PWM attached at the moment, just running only power?

    What I saw was running stock, fan not attached to the circuit, NOTHING, i was able to start my fan at about 3volts, ran smooth and super low rpm.

    When fan is attached to circuit, pwm not attached to motherboard YET, my fan starts spinning at around 4.8 volts, and is very very smooth and slow. Is this what I should be looking for?

    Edit: Of course, at 12 volts, they were madmen... and i love it!
     
  10. Lazzer408

    Lazzer408

    Joined:
    Jan 6, 2007
    Messages:
    2,547 (0.88/day)
    Thanks Received:
    338
    Location:
    Illinois
    The fan should be running at full speed if the PWM signal isn't attached. Check for ~5v between gate and source.
     
    Last edited: Jul 27, 2014
  11. Aquinus

    Aquinus Resident Wat-man

    Joined:
    Jan 28, 2012
    Messages:
    6,848 (6.48/day)
    Thanks Received:
    2,456
    Location:
    Concord, NH
    In Intel's defense, running a low voltage and not a PWM + 12v makes it less likely that the motor will actually spin at low fan speeds. Applying short bursts of 12v versus constant low voltage will result in more consistent low speeds since the motor itself still realizing 12v, just less of it where a really low voltage like 3v or 2v might not even make the fan spin. So there are reason why a 4-pin design was made, but not if you don't want fine control of fan speeds and just rough control of it.
     
  12. Lazzer408

    Lazzer408

    Joined:
    Jan 6, 2007
    Messages:
    2,547 (0.88/day)
    Thanks Received:
    338
    Location:
    Illinois
    The reason they use PWM is because it's a switched-mode method of controlling analog devices like motors, lamps, LEDs, etc. and that's second nature to digital electronics. It's also more efficient and cost effective then a linear voltage regulator. The switching devices (carrying the current loads) are located in the fan so really all Intel is providing is a pulse train. They left the actual control circuit in the hands of the fan manufacture. This leaves them free and clear of any issues caused by the design variations of fan manufactures. As for start-up, that's on Intel's side. The PWM signal from the board ramps from 100% to X% on start-up.
     
  13. Velict New Member

    Joined:
    Jul 12, 2014
    Messages:
    24 (0.15/day)
    Thanks Received:
    1
    update: i'm at work now, I'm testing for 5v across gate and source. I'll be back in 20 minutes.
     
  14. Velict New Member

    Joined:
    Jul 12, 2014
    Messages:
    24 (0.15/day)
    Thanks Received:
    1
    Alright! I tested the gate and source, but when I did, my multimeter would short out the mosfet and turn the fan off... so i did a bit of troubleshooting and I didn't connect the source to the diode and resistor... oops. Now, i'm seeing about 4.7 volts. All is well :D Also, the previous issue of the fan starting at a higher voltage has been solved. Fans properly start at ~3v

    Edit: uploaded pictures....

    That grey stuff is dow corning 3145
     

    Attached Files:

    Last edited: Jul 28, 2014
  15. Lazzer408

    Lazzer408

    Joined:
    Jan 6, 2007
    Messages:
    2,547 (0.88/day)
    Thanks Received:
    338
    Location:
    Illinois
    Glad you figured it out. How is it handling the load of so many fans?
     
  16. Velict New Member

    Joined:
    Jul 12, 2014
    Messages:
    24 (0.15/day)
    Thanks Received:
    1
    Still waiting for the headers to test more than 1 fan.

    but I have a question about your second schematic. The 30% modulated one.... is the fan on this schematic also supposed to run at max RPM when PWM is not connected? The same issue is happening, too. I'm using the 3904s and tip3055

    Edit: R1 and r4 are receiving 12 volts, but r2 is only recieving 6v. Could I just replace the 10k resistor in this schematic with a variable 100k resistor?
     

    Attached Files:

    Last edited: Jul 30, 2014
  17. Lazzer408

    Lazzer408

    Joined:
    Jan 6, 2007
    Messages:
    2,547 (0.88/day)
    Thanks Received:
    338
    Location:
    Illinois
    None of the fans are modulated by that circuit itself. The circuit is just a switch being modulated by the motherboard's PWM pin.

    As for the PWM pin...

    PWM high (5v or not connected) = Fan at 100%
    PWM low (grounded by motherboard) = Fan at 0%
    PWM 50% = Fan at 50% (6v)

    The schematic you just posted is the mosfet version. A bipolar transistor version -can- maintain a minimal fan speed by keeping the transistor in a partially conducting state so when PWM pin is low, the fan doesn't completely turn off. You could also do this with a resistor but it will get hot since that is a linear method of idling the fans.

    I drew them some time ago. The transistor one was only an example of how the fan can be high-side switched so that the negative lead remains connected to the motherboard so the tach would function correctly.

    The RIGHT way to do any of this (if you want complete control over everything) is to make a separate circuit that creates a PWM signal and have that circuit controlled by the motherboards PWM signal. An RC network could convert the motherboards PWM to a voltage and have that voltage control your PWM generator.

    It's electronics. There's always 101 ways to skin a dalmatian. No wait...
     
  18. Velict New Member

    Joined:
    Jul 12, 2014
    Messages:
    24 (0.15/day)
    Thanks Received:
    1
    understandable, I have 556 timers to make my own pwm generator. But, i'll probably wait to do that. For now, what should I be seeing on the 30% circuit with the pwm not attached to the motherboard.
    My 50% circuit works amazingly. I'll be taking pictures of it tomorrow
     
  19. Lazzer408

    Lazzer408

    Joined:
    Jan 6, 2007
    Messages:
    2,547 (0.88/day)
    Thanks Received:
    338
    Location:
    Illinois
    Looking at the schematic in post 7, you can probably eliminate R1 and Q2 and place the zener where Q2 was. When PWM calls for full speed, you should see ~11.45v on the fan due to the C-E voltage drop across Q1. When PWM calls for 0, the minimum speed is set by the 100k variable resistor.

    Remember though, the post 7 circuit is linear when PWM calls for 0 fan speed and you have the 100k set to keep Q1 partially on. It needs a proper heatsink.

    The mosfet one shouldn't get hot at all and may work well without a heatsink at all.
     
  20. Vario

    Vario

    Joined:
    Oct 21, 2005
    Messages:
    3,080 (0.92/day)
    Thanks Received:
    1,015
    I am looking at running all 6 of my H1011's off of my motherboard's PWM CPU channel and doing away with my fan controller, could your circuit handle this? The fans are 12 volt .52amp 6.24 watts 3 pin.
     
  21. Velict New Member

    Joined:
    Jul 12, 2014
    Messages:
    24 (0.15/day)
    Thanks Received:
    1
  22. Vario

    Vario

    Joined:
    Oct 21, 2005
    Messages:
    3,080 (0.92/day)
    Thanks Received:
    1,015
  23. Lazzer408

    Lazzer408

    Joined:
    Jan 6, 2007
    Messages:
    2,547 (0.88/day)
    Thanks Received:
    338
    Location:
    Illinois
    I'm going to add another circuit to this thread that may help you guys.

    This circuit CREATES a PWM signal that can be used to manually control 4-pin fans. The potentiometer is labeled "throttle" because this was part of a DC motor controller I'm designing.

    Frequency is controlled by C1. R7 may not be required. By replacing the potentiometer with a thermistor, the circuit can monitor temperature and control the fans accordingly. R5 and R6 would then have to be adjusted to suit your thermistor and the temperature range.

    VCC is incorrectly labled 5v. The circuit runs on just about anything but was designed for 12v operation.

    [​IMG]
     

    Attached Files:

    Last edited: Sep 10, 2014
    Vario and Random Murderer say thanks.
  24. Vario

    Vario

    Joined:
    Oct 21, 2005
    Messages:
    3,080 (0.92/day)
    Thanks Received:
    1,015
    Lazzer, can you give me feedback on whether this circuit looks okay and won't blow up my motherboard?

    [​IMG]
    [​IMG]
    [​IMG]


    [​IMG]

    I used Velict's diagram
    [​IMG]

    Thank you.
     
  25. Lazzer408

    Lazzer408

    Joined:
    Jan 6, 2007
    Messages:
    2,547 (0.88/day)
    Thanks Received:
    338
    Location:
    Illinois
    Did you bench test it?
     
    Vario says thanks.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guest)

Share This Page