PWM circuit to control 3-pin fan

[Eureka]

I found this simple circuit using a Darlington pair:
http://www.overclockers.com/forums/picture.php?albumid=395&pictureid=2336

As well as this one that uses a single mosfet and a cap (see page 2):
http://pdfserv.maximintegrated.com/en/an/AN3530.pdf

Which one would be a preferable circuit to take a 4pin signal from the motherboard to control 200mm fans (which are all 3-pin)?

What would be the value of R1 in the darlington pair design?

Also, to protect the fans, someone suggested using a Schottky diode. Would this be placed in parallel with the fan or in series?

Finally, does anyone know the connector names for 3-pin and 4-pin (PWM) connectors used on the motherboards/fans?

Thanks!

 

[Modelworks]

You need to smooth the PWM output with a low pass filter before you use it to control a 3 pin fan. Then all you need is a transistor that can handle the fan load. You will need to try different values of the capacitor and resistor to get the speed you want on the output, the voltage on the base of the transistor will vary based on the frequency of the PWM output and the RC .

 

[Eureka]

My filter theory is quite poor.. what happens when you use the LPF? I'm assuming the PWM signal is only one frequency, what does it output for various duty cycle lengths? And how does it affect the transistor?

Will a TIP31C BJT work with the filter?

Many thanks.

 

[esun]

The idea behind PWM is that instead of generating a constant voltage, you generate pulses whose average value turns out to be the constant voltage you want. Typically then you want to average a PWM waveform once it hits an analog device to get a useful signal out of it. However, if the frequency of the PWM is high enough (relative to the response time of the subsequent device---in this case the transistor), then the device itself will act as a filter and you won't need a separate on. So if your PWM is 10 MHz and your transistor's f_max is 1 MHz you're fine.

However, you may not need to filter it before the transistor either way if the PWM voltage isn't too high (i.e., beyond the maximum base/gate voltage of the transistor). The output will simply remain pulses (or pulse-like, anyway) and the fan itself will act like a filter. Naturally the fan doesn't have an instantaneous response so it will act in some capacity to average the rapid pulses of a PWM signal (assuming those pulses are at a high enough frequency).

The transistor must be able to handle the necessary voltage and current for your setup. See how much current the fans consume and choose transistors supporting at least 12 V with that current amount (times the number of fans if you have multiple fans hooked up to one transistor for PWM control).

The Schottky diode would go in parallel. The idea behind a Schottky is to provide a path for current going the wrong way to go around the device you want to protect. Naturally this means it must go in parallel with the fan (since it provides a parallel path for current).

As for the resistor size it is not a big deal, anything 1k to 10k will probably work just fine.