[SOLVED] How to design specific filter ?

[Audioguru]

I made a pink noise generator with an MM5837, it was not a 5937. I designed its white to pink filter with trial and error so that its frequency response perfectly matches an expensive professional pink noise generator.
Sometimes I can hear it repeating is pseudo-random sounds. Its sequence is different each time it is turned on.

 

[schmitt trigger]

You are correct, MM5837...it was a typo error.
Darn phone keypad, it is too small.

 

[Ibtissam Aziz]

Hi,
Thank you for your help !
I have found several circuits wich do the job on the internet.
But if you had to design from scratch how do you proceed to compute the values for each R and C ?
Let's choose this design for example : (FIGA1)
**broken link removed**
As you can see tranfert fonction will take ages to hand compute and after you got it how to determine the right values to aproximate the -10dB/dec desired ?
Compute the log magnitude curve then use a non linear solver and make a curve fitting as suggested by FvM ?

 

[FvM]

Compute the log magnitude curve then use a non linear solver and make a curve fitting.

That's clearly a working method. With a moderate number of RC ladder elements, also manual fitting is an option.

 

[D.A.(Tony)Stewart]

sunnyskyguy:

Indeed, in the late 70s I built a pink noise generator using some of the schematics you show in your link.
The MM5937 white noise generator was a wonderful little IC, although its sequence was not very long would repeat every two or three seconds. One could listen to a "bump" when this happened.

I did the same... and found it sounded too much like a choo-choo-train, so I designed my own with better results. We wanted to make a room equalizer using this with a Yamaha 1/2 octave graphic equalizer and a microphone. The tough part is getting the mic to match your hearing perception. The easy part was locating the speakers for best amplitude response. Balanced Phase response required your ears with a mono sound in the middle of your head.

To realize the filter from scratch, use
H(s)=1/(1+s/(2PI*fo))^0.5

Intuitively, the square root response matches the Cap impedance ratio of cascaded filters where sq.rt. of 10 is 3.3 used approx. for successive stages

Rigorously, I leave for students, but a 1st order RC filter has a -3dB breakpoint and the root of 10 is ~3.3