[sdiy] Re: [AH] Low slope VCF

[Andre Majorel]

[Moving the discussion to synth-diy]

On 2008-02-10 11:24 +0100, Andre Majorel wrote:
> On 2008-02-08 16:30 +0100, Lars ARNWALD wrote:
> 
> > Having reviewed EM discussions around a 1 dB/oct LP Spectral
> > Tilt Filter created by a Clavia Nord G2 I like to ask this:
> >
> > - Is it possible to create an all analog VCF (LP, HP, BP) with a
> > roll-off less than 6 dB/oct? Do the different filter types
> > (Butterworth, Bessel, Sallen & Key, etc) make a difference.  -
> > Do varying crossfadings between a dry signal and a 6 dB/oct make
> > something that I'm looking for?
> 
> Any pink noise generator contains an approximation of a 3 dB/oct
> low pass filter. Approximation because the shallowest filter
> available, a one-pole filter, has a 6 dB/oct slope. The 3 dB/oct
> slope has to be approximated by tricks. The Oakley module uses "an
> original design from Wireless World". Five caps and four
> resistors. It's not voltage controlled.

Several approaches outlined in this music-dsp thread :

  http://www.music.columbia.edu/pipermail/music-dsp/2007-January/065955.html

In it, James Chandler describes an analogue implementation, AFAICT
the one used in the Oakley module :

| In analog, various manufacturer application notes showed pinking
| filter circuits with several RC lowpass networks in series (with
| no electronic buffering between the simple stages). Three or
| four stages for audio-band pinking. [...] here is my naive
| 'hand-waving' assumed explanation--
|
| Usually analog RC networks only behave as you would expect a
| first-order filter to behave, if the destination impedance is
| much higher than the impedance of the RC node. A good rule of
| thumb is an impedance factor of 10X. So for instance, if the
| impedance of the RC network is 1 KOhm at center frequency, it
| will behave as you expect if the following buffer amp has an
| input impedance at least 10 KOhm.
|
| So if you wanted an kinda-Bessel' third-order RC lowpass and you
| were too cheap to use a lot of buffer amps (this is a really
| dumb way to do it, only as an illustration)-- If the first RC
| has impedance of 1 KOhm, the second RC has an impedance of 10
| KOhm, and the third RC has an impedance of 100 KOhm, and the
| final buffer amp has an input impedance of 1 MegOhm, it would
| work pretty close to an ideal three-cascaded first-order
| filters.
|
| If you 'load down' an RC with too-low destination impedance, it
| degrades the filter's performance (so you wouldn't get the
| expected 6 dB per octave behavior anymore).
|
| Those typical three-cascaded analog RC pinking filters-- The FC
| of each section is scaled, and the impedance of each following
| section is low enough to degrade the performance of the previous
| section.

-- 
André Majorel <URL:http://www.teaser.fr/~amajorel/>
Do not use this account for regular correspondence.
See the URL above for contact information.