[Guide] Build home made attenuator

mikeraach · Post by **mikeraach** » Wed May 11, 2005 9:56 pm

With the help of Webfire (he supplied some pictures of the mcubed attenuator so that I could indeed verify that my design would work) I have successfully made 2 attenuators for $10, and can easily make 2 more for only $2. Best yet is the fact that I believe that my attenuators are more effective than mcubes (based on what I could find the values of the resistors and capacitors provided by Webfire's pictures). This device does infact reduce the slight clicking noise that my 120mm yate loons produce!

What you need:
3-pin housing connectors
The prongs that go in those housings
The male end of the 3-pin housing connectors
Some caps (1000uf)
Some resistors (5ohm would be good, but Radio shack only carries 10ohm so I wired them in parallel to make 5ohm)
Blank circuit board.

With 5ohm resistors (as opposed to the 12ohm resistors mcubed used) I can get a higher voltage to the fan when there is 100% given to the fan (10.9v vs 9.6). You HAVE to build this circuit with resistors, so dont think you can just take them out to get 12v at 100%. This is basic dc stuff and if you want to know more I suggest getting a book and reading as I took a whole semester of this stuff in college and it'd take too long to explain.

Also using a 1000uf cap results in a much smoother voltage given to the fans compared to 100uf (as will be shown below). This would hopefully further reduce fan noise).

How about some pics?

The basic wiring diagram:

Front:

Back:

Wiring diagram (basically shows you where the positive and negative parts of the wire are):

Oscilloscope (software) showing the difference between 5ohm with 1000uf and 12ohm with 100uf cap:

5ohm,1000uf:

12ohm,100uf (notice how this looks more rough--the voltage isn't as 'clean')

Aleksi · Post by **Aleksi** » Wed May 11, 2005 10:07 pm

Hi,

nice looking project! Probably smoothing the signal too much will result in loss of the tachosignal, but that's not very important.

Also, if there is no fan connected, then you would be driving (12V, 5ohms) 2.4A of current through the resistor, correct? you could add a resistor in parallel with the fan, that would help in this situation.

mikeraach · Post by **mikeraach** » Wed May 11, 2005 10:36 pm

Edit: I was really tired when I posted this last night and I wasn't thinking too well. Now that I look at it, NOTHING will happen when there is no fan.

The only thing connected in the circuit would then be the capacitor, and when it is fully charged it acts as an open, thus preventing any current going through the resistors.

/Edit

Good point as to with my current design there is no safe guard for if there is no fan plugged in. I guess its a good idea not to juice it up without a fan

sthayashi · Post by **sthayashi** » Thu May 12, 2005 6:29 am

Maybe I didn't follow the solder connections very well, but how would NOT having a fan hurt the resistor?

BTW, a very good job with the attenutator. How well does it work with low PWM percentages?

One last thing, and this is more for everyone. Does anyone remember how to calculate the frequency of Low Pass Filtering when a capacitor and/or an inductor is used?

Aleksi · Post by **Aleksi** » Thu May 12, 2005 7:54 am

using a regular resistor (1/4W) in that scenario would result in the resistor frying. If there is no fan connected, you would be driving in the worst case 2,4A (5ohm) or 1A (12ohm) through the resistor, meaning wattage of 28,8W or 12W.

Calculating the cutoff freq for the lowpass filter using a resistor and a capacitor is f=1/(2*pi*RC), is this what you were looking for?

mikeraach · Post by **mikeraach** » Thu May 12, 2005 8:51 am

Aleksi: Check out my edit above to see why there will be no problem with not having a fan connected.

sthayashi:Yes low pass filtering is f=1/(2*pi*RC).

I dunno if you were thinking this, but my attenuator is not a low pass filter (for it to be, I would have to make the fan voltage be taken across the cap which I am making the fan voltage go in parallel with the cap).

Aleksi · Post by **Aleksi** » Thu May 12, 2005 9:02 am

Hi Mike,

just noticed you had edited your post.

you're absolutely right, it doesn't matter if the fan is not connected. For some reason I was thinking the resistor stayed parallel to the resistor

. Ignore my post

cpemma · Post by **cpemma** » Sat May 14, 2005 1:19 pm

The idea looks interesting, but why 850Hz? That's well in the audio range and will be pretty noisy IMHO.

Looking at a less powerful fan (100mA = Pan 80L1

) and lower frequency (85Hz) it's removed the sharp attack that seems to be a major factor in PWM noise creation. Must test...

95% duty cycle

50% duty cycle

mikeraach · Post by **mikeraach** » Sun May 15, 2005 12:34 pm

The higher the frequency, the smoother the output voltage of the circuit will be. The fans make clicking noises when using PWM because of the constant on off. A smooth dc input makes no noises and thus would be the most desirable to feed to a fan.

Yes something at 850hz is within the audible range of humans, however it is the fan that makes noises, not the hz. As stated above, the 850hz makes a smoother output which is much closer to a dc signal than 85hz. It is the irregular voltage (provided by PWM or by the diagram you posted) that produces the noise that we hear from the fan.

For the most part, as long as the signal is smoothed and have less drop offs and spikes(as what you have done with 85hz) the fans should create very little noise compared to solid PWM, however if you are able to make pure DC (closer to what I have done with 850hz) then the fan will sound nearly identical to if the fan was getting fed DC.

Btw, what program did you use?

cpemma · Post by **cpemma** » Sun May 15, 2005 1:51 pm

mikeraach wrote:The higher the frequency, the smoother the output voltage of the circuit will be.

Btw, what program did you use?

True, but I was looking at this as an add-on to a commercial PWM controller, where you don't have much choice on the frequency. Low frequency (under 100Hz) seems to be common, with Microchip and others recommending around 30Hz, with frequencies above the audio band (>20kHz) reserved for 4-wire fans.

The simulator is Circuitmaker.

trpltongue · Post by **trpltongue** » Fri May 27, 2005 4:40 am

Mike,

Just to clear something up, this is a 2-channel version right? I'm assuming the wires going out of picture are connected to the PWM signal source (MB header). Is my thinking correct here?

Russell