Fan'n'music (howto: measure rpm)

kickaha · Post by **kickaha** » Tue Jun 24, 2008 3:54 pm

I wanted to reuse an old psu fan (the psu burned 10 years ago) as case fan, but knew nothing of it. First of all I tried the fan with 3x1.5V batteries, just to avoid burning my new motherboard or psu whether something wasn't fine with it. It started easily at 4.5V and needed 0.40mA. Classic fan.

But what about rpm? It has just two pin for 12V supply!
Well, I happen to be a musician, and rpm is just a frequency, so this is the recipe:

- from a sheet of paper cut a little strip, something easy to handle (say, 1x10cm)
- with the fan turned off count the blades
- with the fan turned on put the strip near the fan, just to make it buzz, but without stopping or slowing the fan
- memorize the sound and choose from the possibilities:
1) you are gifted with absolute/perfect pitch, then at this point you know exactly which note you're listening at
2) you are not, but you have an instrument and you can search for the note
3) you have neither, then you google for any software that allow you to play notes on the keyboard or something and find the note
- use the reference tables below; I made them for 7 and 13 blades; scale 0 is the central one.

Theory

The frequency of the "buzz" is just how many times a blade hit the strip in a sec.
The A/La just above the central C/Do is 440Hz, every octave up the frequency doubles, every octave down the frequency halves.
Every semitone up/down, you have to multiply/divide the frequency by 2^(1/12).

Finding the note, you find the frequency. Once you have it, you divide for the number of blades and you have "revolution per second". To have rpm you just multiply by 60.

Assuming you can recognize the note, the error is 1 semitone, that is always 6%.

Example

My fan buzzed a F# just under the central C. Then starting from the reference A (440Hz) I have to go down 1 octave to the previous A (220Hz = 440/2) and then down again 3 semitones:

F# <-- G <-- G# <-- A <-- ... (central C) ... <-- reference A (440Hz)

My note has a frequency of 440 / 2 / 2^(3/12) = 185Hz
Then "rps" is 185 / 7 blades = 26.4
Then rpm is 26.4 * 60 = 1586
Error is 94 rpm: precise enough, I'd say.

In the reference table (7 blades) you see this values going back 1 octave from central and looking at the F# row.

Emyr · Post by **Emyr** » Wed Jul 16, 2008 2:03 am

I know it's just paper, but the buzz-strip might slow the fan down at lower voltage?

On a more general note, if every fan or vibration in your PC was turned to the same note or chord, would it sound quieter (sensitivity of human ear to dischord) or would harmonics and sympathetic vibration passing through the case make each vibrating object vibrate louder...?

jhhoffma · Post by **jhhoffma** » Wed Jul 16, 2008 6:06 am

This is just funny. I love this community because of the wide range of ideas people get to deal with all the different aspects of quiet computing based on their backgrounds.

Brilliant writeup!

kickaha · Post by **kickaha** » Wed Jul 16, 2008 1:56 pm

Thanks!

Emyr wrote:I know it's just paper, but the buzz-strip might slow the fan down at lower voltage?

Sure. You have to make the note just audible, but without slowing the fan.
I checked the method with motherboard rpm and it seems to stay within the error range.
I've measured a 80mm fan downvolted at 5V (about 700 rpm) without problem.

Emyr wrote:On a more general note, if every fan or vibration in your PC was turned to the same note or chord, would it sound quieter (sensitivity of human ear to dischord) or would harmonics and sympathetic vibration passing through the case make each vibrating object vibrate louder...?

I think you can hardly do that

You'd have little differences in frequency and wave shape (higher harmonics) and every fan/component wouldn't keep exactly the frequency over time. I believe they would roughly sum up.