[sdiy] Basic core designs of commercial analogs

[Magnus Danielson]

From: Grant Richter <grichter at asapnet.net>
Subject: Re: [sdiy] Basic core designs of commercial analogs
Date: Wed, 25 Dec 2002 11:50:44 -0600

Dear Grant,

> Here is my thinking,
> 
> If you accept, that modulating the current to the transconductor is a
> form of multiplication.

You could either change the reference current to the expo-pair, or change
the offset into the balancing op-amp driving the common emitter of the
expo-pair.

The unscaled case would best be handled by adding to the current out of
the expo-pair and into the CCO.

> For polarity of multiplication: In the sawtooth ramp, modulation
> positive and negative slope can only be multiplied by a positive slope
> from the core.

It might be that you allow for a modulation of shifting polarity but this
is certainly not a thing all cores support. Usually they only allow for a
1-quadrant multiplication, i.e. that both (scaled) linear modulation and
the exponentiated pitch (which only can be positive).

> In a triangle core, modulation positive and negative slope can be
> multiplied by a positive or negative slope in the core.

Hang on here... you are refering to the polarity switcher of the input
current, right? That's part of the triangle CCO core.

For a triangle CCO core to handle deep negative modulations you would
have to modify the reset-curcuit. The triangle CCO core doesn't handle it
as good either since it's "state" needs to be changed.

> If we take a sine modulating a sawtooth core vs. a sine modulating a
> triangle core, the effect of both will be to bend the slopes of the
> waveform at the output.

Yes, and that results is the increase and lowering of frequency but the
average frequency will however be that of the unmodulated frequency.

> But if you take the modulated sawtooth and waveshape it to get a
> triangle, the negative modulation bend will point out the other side
> than if you modulate a triangle core directly?

No. Your getting confused by the sign reversals of the triangle core and
that which is motiviated by balanced modulation (like that of a sine).
The sign reversals which happend in the saw->tri converter and those
which the the triangle CCO core does should be equalent. However, these
are then reversed by the sign reversal of your modulating waveform, if
things work properly that is...

A triangle CCO which is able to really handle through-zero modulations
would then need to have a 4-quadrant multiplier for sign reversals as
indicated by the slope-control. The modulation core would need to have a
2-quadrant multiplier for the through-zero modulation and then naturally
the expo-pair to generate the 1-quadrant pitch signal. These could then
be combined by any of a number of ingenious methods. Most triangle
oscillators only handle 1-quadrant modulation sources and thus only
require a 2-quadrant multiplier in the CCO core.

It's all a bit confusing, I agree... ;O)

Cheers,
Magnus