Chaos LFO : who to manage 1/X ?

[Grant Richter]

The National LH0094 is a monolithic module designed to do complex 
multiplication, division and exponentiation It is just a group of log/
anti-log circuits with access to the current summing points. Unfortunately
they are like $100 if you can find one. Burr-Brown and Analog Devices also
had equivalent units back in the day.

BUT you can do the same thing with a CA3046 and some quad op-amps. That is
build a discrete LH0094. You must us a group of matched transistors for the
log/antilog circuits, but no tempco is needed since the tempco of the log is
cancelled by the tempco of the antilog. Electronotes devoted a whole issue
to doing math with log/antilog summation and ratios.

Among other things, you can make a circuit that does x=y^m where m is
variable from 1 to 4. The log circuits are only a single quadrant so you
have to either bias the chip or use a full wave rectifier on the input and
switchable inverter on the output.

The Fairchild/JRC RC4200 uses the sum of logs technique to make a vary good
and cheap multiplier (0.1% linearity compared to 1% for AD633) they are also
less than a buck in volume, but require much more support circuitry. If you
study the datasheet for the RC4200, it shows a bias scheme to get +/- 10
volt ranges from a single quadrant device.

Try looking up the LH0094 datasheet, and the RC4200 datasheet, they are good
tutorials on doing math with log/antilog techniques.

----------
>From: Martin Czech <czech at Micronas.Com>
>To: harrybissell at prodigy.net
>Cc: synth-diy at node12b53.a2000.nl
>Subject: Re: Chaos LFO : who to manage 1/X ?
>Date: Thu, Nov 9, 2000, 2:30 AM
>

>
> :::MHO:
> :::
> :::Dividers are a real weak point in electronics, especially analog.  They
> :::usually are
> :::very limited, especially when the denominator approaches zero...
>
> Well, I'd say a lot of algorithms on digital computers suffer from that, too!
> The AD534 data sheet shows some divider appl.
> Perhaps it is possible to avoid the division by changing the structure?
>
>
> m.c.
>
>