PWM Fan controller [Archive] - jonnyGURU Forums

shinji2k

07-23-2009, 07:00 PM

I've been playing around with an analog PWM Lamptron fan controller. These don't play nice with a lot of fans and I was hoping to smooth the output out with some caps. Hopefully I can reduce some of the motor noise. They've got PCB space for output caps but I guess they decided they weren't needed. At lower voltages the frequency is in the 100-250Hz range. What kind of caps should I look for that would work best for that frequency?

shinji2k

07-23-2009, 09:33 PM

Well I did some experimenting with some scavenged caps. I figured it would mess with the circuit a little and I was right. Using a 16V 330uF cap limited voltage adjustment to 7-12V but all motor noise was eliminated. I tried a 50V 47uF cap and it give me adjustment down to 5V, and motor noise was still suppressed. I tried a 50V 10uF cap and it gave me adjustment down to ~3V but then it did not help with motor buzzing. I'd like to be able to get down to 4V so maybe I should find a cap in the 30-40uF range to test with before I order some.

Is there any danger to the cap or fan controller running it like this? Is there a certain brand/model of cap that would hold up better to these low frequency situations?

KeriJane

07-24-2009, 07:47 PM

Hi shinji

I've noticed this effect also.
A lot of PWM fan controllers on motherboards end up creating a lot of fan motor noise. Some boards appear to have a spot for a capacitor right next to the PWM connector, others don't.

Could one wire a cap into the motors leads? Which would be the best ones?
i'd love to get these noisy little chuggers quieted down.

Thanks for the insight so far,
Keri

shinji2k

07-24-2009, 10:27 PM

Well, I've found that every fan I have behaves differently. Most normal fans will be silenced with a 47uF cap in parallel and still give decent voltage adjustment. Most of the time I could get the fan controller down to between 5 and 7V. If you need lower voltages, you've got to decrease the capacitance. But then you run the risk of the capacitor not being very effective. What I did to test was to take a fan extender and just splice the cap into that. Some of the more powerful fans like the Kaze 3k needs a 100uF to tame them. But those have some bearing noise at low voltages anyways so I'm fighting a losing battle there. 47uF and 100uF 16V caps should be about all you need.

There is also the issue of running more than one fan in parallel on the same fan controller channel. For example, putting two Kaze 3k on one channel forced me to put a 330uF and 100uF cap in parallel (effective 430uF) to get the desired effect. With three YL 14cm medium speed fans, I needed 2200uF! I'm not sure why this is, even three YL only draws as much current as one Kaze 3k so it's not a current issue.

But for fixing motherboard controllers there shouldn't be too much issue with 47uF or 100uF not working. You could just splice one inline and put some heatshrink over it and be done.

370forlife

07-26-2009, 11:49 AM

These lamptron pwm controllers, how high of an amperage can they run?

I have 3 nidec va350dc 92mm 110cfm fans in my computer that are pwm capable, but use 1.1A each, too much for the motherboard.

As you can imagine they are quite loud :wall:

shinji2k

07-26-2009, 08:47 PM

They've got six channels and are supposed to be good for 45W each channel.

CBRworm

07-31-2009, 07:58 PM

The only concern I know of with adding a capacitor is that the capacitor will draw much more instantaneous current when it is charging during each pulse than a fan motor normally would. Whether this is enough current to damage the controller would depend on the transistor used. It seems like they should be somewhat tolerant.

That being said, I have tried adding caps to PWM controllers and have not yet damaged one, but also haven't successfully gotten rid of the noise and kept a reasonable range of adjustment.

shinji2k

08-01-2009, 01:13 AM

They've got the transistor labels sanded off so I've no clue what they are rated for. If it dies eventually, so be it. The way it is now it is pretty much useless and PWM compatible fans aren't exactly in abundance. I've had some luck getting most of my fans to where they aren't resonating and I still get decent adjustability, so for now I'm just happy I found a quick fix.

Ok, just ran across this (http://ww1.microchip.com/downloads/en/AppNotes/00771b.pdf). It looks like they are recommending a low-pass filter with an RC between 1 and 4.7ms. So an RC of 1ms would mean a cut-off at 160Hz and 4.7ms would be ~34Hz. Please stop me if I am wrong, I am pretty new to this. I'm gonna try this out next since I will be able to use much smaller capacitors.

cypherpunks

08-02-2009, 08:38 PM

Is there any danger to the cap or fan controller running it like this? Is there a certain brand/model of cap that would hold up better to these low frequency situations?

The concern I have is that a capacitor is lower impedance than the fan by itself. If you connect a capacitor charged to 7V to a power supply at 12V via the switch in the PWM controller, you're going to get a lot of current flowing all of a sudden. In fact, you're creating a dead short; if all components were ideal, it would be infinite current. Now, in practice there will be some resistance in the wiring which limits current, but I don't know if the PWM controller is designed for that.

Thinking, it's very possible that it is, as fans have input filtering caps. The PWM controller could have some kind of overcurrent clamp.

To make this work in real life, you want a filter inductor and a catch diode, and congratulations! You just invented the buck switching regulator. (Real fans are on long wires, which have—guess what?—inductance, and that might be assumed.)

I can't tell you off the top of my head how much would be needed, but you can try it. Just add a schottky catch diode near the PWM controller (it has a small one on-chip, but I don't know if it's enough) and some inductance in the power lead.

shinji2k

08-03-2009, 12:08 AM

cypherpunks

08-03-2009, 12:25 AM

Thanks for the info. Making these into buck regulators definitely sounds like something I can handle. I'll do some research and gather up some parts. Would converting it to a buck regulator be better than the low-pass filter I suggested in my previous post? I assume it would be. I may just build both for the fun of it.

It would be more efficient, but it actually is a low-pass filter. Just a lossless LC one (no resistors). The downside is, it's harder to design, and using a resistor to solve the infinite-current problem is easier.

I'd go with the RC unless you're trying to learn more about circuits.

shinji2k

08-03-2009, 12:58 AM

cypherpunks

08-03-2009, 01:36 AM

OK. I think I'll try the RC circuit first since it is easier. Also, I was trying to calculate the value of the inductor needed and I think the switching frequency of the PWM controller may be a limiting factor here. Assuming 200Hz, a rough calculation gives me a fairly large value of 0.364H. I would still hope to have each channel handle a couple amps. I don't know much but it doesn't seem possible to buy inductors able to do that. I can see why most buck regulators use 100kHz+ frequencies.

Oh and thanks again for the help.

Yes, the lower the frequency, the larger the components. Corner frequency for an LC filter is 1/π√(LC), so it depends on both. But plugging 200 Hz and 0.364 H into that shows that you've got 7 μF of capacitance. (Maybe a bit more, as you want the corner frequency below the switching frequency, but I don't know how much fudge factor you used.) Surely you can get more than that? A 1000 uF capacitor and 200 Hz corner frequency would call for a 2.5 mH inductor. That's not hard to find with a 1A current rating. You could make that without a core using 85 turns of thin wire in a 1x1 inch cylinder. (inductance calculator (http://www.daycounter.com/Calculators/Air-Core-Inductor-Calculator.phtml))

shinji2k

08-03-2009, 02:22 AM

I was using step 2 on this page (http://www.gbint.com/Files/Apps/General%20Apps/GB-4400-001.htm) to calculate the inductance needed. The figures I used were Vin = 12V, Vout = 5V, Iout_min - 0.2A, freq = 200Hz. It doesn't seem like that was the right formula to use though.

The RC circuit may be simpler, but from what I have read it may not completely solve my problems. This thread (http://www.hardforum.com/showthread.php?t=797390) has some relevant discussion. It looks like the RC circuit would involve a lot of trial and error and it would be different for each channel (each channel has a different type and number of fans). It may also not correct any problems from back EMF from the fan and since I don't have an O-scope I can't really trouble shoot the cause of all the noise. There were two components. There's the electrical noise which I assume is interference between the PWM frequency and any resonant EMF from the fan. Then there was the pulsing from the fan motor. The RC should help smooth out the pulses causing the motor noise but it may not help with any resonance. Converting this to a buck regulator by adding an LC circuit would smooth out the square wave and approximate a DC voltage, so long as I design it correctly. That would seem like the ideal solution. Am I just talking out my ass?

CBRworm

08-11-2009, 10:14 PM

Another option is to try different fans that don't make as much noise as the power is being PWM'd.

simontay1984

04-02-2010, 02:37 AM

Not sure if this would help, but I bookmarked this post I found ages ago while looking for a circuit to convert PWM to DC: http://forum.allaboutcircuits.com/showthread.php?t=8998

I also emailed PaQt and someone called Peter was kind enough to send me a schematic of the circuit in http://www.paqt.co.uk/docs/PaQ_PWM.pdf
I can send it you if you want, it's just an MS word file.