[sdiy] Re:freq shift allp filter ?

[Magnus Danielson]

From: "jhaible" <jhaible at debitel.net>
Subject: Re: [sdiy] Re:freq shift allp filter ?
Date: Tue, 14 Jan 2003 23:33:43 +0100

> > Thanks all for the help.
> > I will try to build JHs Dome filter. I hope I can manage calculations...
> BTW, anyone built it? At
> > least to have the values calculated correctly, I'm not a maths one...
> 
> If you can do simple multiplication / division, that will be fine. (;->)
> More seriously: You must measure capacitor values and resistor values
> accurately - otherwise all the effort is in vain.
> 
> 
> > Anyhow, it looks like not any quadrature oscillator will work well, but
> JHs looks too complicate for
> > me
> 
> Wait until you see the Moog / Bode oscillator ... *that* one is tricky.

Which you find on my webpage for it:
http://home.swipnet.se/cfmd/synths/companies/moog/

> The only thing about my FS-1 that is a bit complicated is the combined
> triangle/sine shaper and multiplier. It's an "economic" design (why
> linearize a gilbert cell if you want soft clipping for sine shaping
> anyway?),
> but this makes it a bit more complicated to understand. Much easier
> to build than a Moog / Bode, though.

Indeed.

> The best solution would be a digital implementation (counter, sin/cos
> lookup table, high resolution DAC).

A long phase-accumulator is a much better solution, behaves just like a linear
oscillator. Also, in digital, the cost isn't that big anyway. The sin/cos
lookup table stuff is however somewhat of a concern. You can by using mirror
and sign tricks reduce the table to be only for one quadrant, and by a bit of
multiplexing it will act for both the sine and cosine. However, for a 16 bit
system it will be quite large anyway. I'm sure someone out there have come up
with methods to compress them even further. I've only toyed around a little,
but not with any major knock-off results...

In the end, you run into a new set of troubles. Like, you also need a ADC for
the control voltage. You need a sine/cosine with good spectral purity etc. etc.
in order to reduce spurious mirror effects.

> Maybe you can replace the two multipliers with multiplying DACs?

You can.

> (But I don't know if there are 12 or even 16bit multiplying DACs available.)

Look at the AD569 for example, 16 bit multiplying (it doesn't say it on the
wrapper, but if you look at it closely, it is... ).

They are around...

Cheers,
Magnus