analog computers

[Grant Richter]

> 
> Maybe a mass/spring/damper system could be intersting if the spring
> get's some nonlinear characteristic (which it has in reality).

Dan's article covers a state variable implementation with a non-linear
element after the summer. One possible case is a cube function
formed from two multipliers. The sounds generated are quite intriguing.
With low Q, the non-linear function provides inharmonic partials in
addition to the filter actions. With higher Q, the circuit form a
chaotic oscillator (tending to oscillate in the mode of the
"attractor" with sudden breakovers to new modes as the energy
level is changed).

> 
> A large modular systems has four quadrant multipliers, summers,
> coefficents (mixer), so what's missing is the simple integrator with set
> and voltage controled time constant, as you pointed out. And maybe some
> nonlinear function networks.

Perhaps an arbitrary non-linear function generator, like an ADC feeding
a PROM and then a DAC. You can then map any arbitrary non-linear
function into the EPROM.