Low pass filter for high wattage 60 watts

Thread Starter

magudaman

Joined Feb 27, 2012
39
I am working on a micro controller to control two fans from a temperature probe. Basically I would like my heat sink remain at around 75C at all times but my fans would adjust in speed according to the load on the sink. These fans consume 36 watts each if not a little bit more and they really don't seem to like PWM power. In fact their data sheet says do not power them with PWM. Upon opening one up I see their is a decent sized capacitor inside that I am sure is taking the brunt of the PWM and would likely overheat.

So I know on low power signal line stuff I have done a resistor capacitor network to create a low pass filter but I am not exactly sure how to scale that up.

I can switch at around 50khz. My fans need 5.5v to 12V from a 12V source. They consume around 30.5w at full speed@12v.

From my lab power supply I made this little chart


Voltage Amps Watts
5.2 0.28 1.456
5.3 0.50 2.65
5.4 0.57 3.078
5.5 0.70 3.85
5.6 1.08 6.048
5.8 1.38 8.004
6.0 1.54 9.24
7.0 1.70 11.9
8.0 2.00 16
9.0 2.11 8.9
10. 2.28 22.8
11. 2.43 26.73
12. 2.53 30.36
 

JDT

Joined Feb 12, 2009
657
What happens when you power your fans from a variable DC supply? Does the fan speed control nicely when you vary the voltage? Some DC fans have switching electronics internally and don't like variable supply voltage.

If they do then I suggest you make a LP filter using a series inductor and a parallel capacitor. You will also need a diode to maintain the current in the inductor when the switch is off. See diagram.
 

Attachments

Thread Starter

magudaman

Joined Feb 27, 2012
39
What happens when you power your fans from a variable DC supply? Does the fan speed control nicely when you vary the voltage? Some DC fans have switching electronics internally and don't like variable supply voltage.

If they do then I suggest you make a LP filter using a series inductor and a parallel capacitor. You will also need a diode to maintain the current in the inductor when the switch is off. See diagram.
Thanks for the reply and schematic. They do seem to run just fine from the lower voltage. I believe the spec sheet say they can run on 5.5V to 13.5V. Any guess as to how to calculate the inductance value or the capacitance based on a frequency of 100khz?
 

crutschow

Joined Mar 14, 2008
34,282
The filter depends upon the ripple the fan can tolerate in the DC voltage. A simple LC filter rolls off at 12dB/octave so, for example, if you want the ripple to be 5% of the peak to peak PWM waveform you need 26dB of attenuation which is slightly more than two octaves above the corner frequency which would thus be below 25kHz for a 100kHz PWM frequency.

I would design for a low-pass corner of 10kHz or below which gives you a ripple reduction of at least 40dB or 1% ripple. Thus you want the inductive reactance to be equal to the capacitance reactance at 10kHz.

And remember that the inductor has to be capable of carrying the fan current at the maximum fan voltage.
 

JDT

Joined Feb 12, 2009
657
At 100kHz the inductor probably only needs to be about 30μH. The bigger the better actually. It must have a low DC resistance and a ferrite core and be able to handle the current without saturating. At 100kHz the capacitor needs to have a low ESR (equivalent series resistance) and should be by-passed with some 100nF ceramic capacitors with short leads. In fact the whole "loop" of diode, inductor and capacitor needs to be as small as possible.

Also at this frequency the diode needs to be a high speed type. If your supply voltage is quite low (under 20V say) then a schottky type should be used. The diode needs to be able to handle the full output current.
 

Thread Starter

magudaman

Joined Feb 27, 2012
39
Thanks for all the replies, thought I would get back on here and let you know what I decided to do. A guy on another forum introduced me to LT Pspice and I actually was able to play with the modeling and really get it tuned to be pretty darn clean! Very cool! Here is what I ended up with:

 
Last edited:
Top