Emu SSMs (was Re: [sdiy] A potential problem with a voice-assignable organ)

[Antti Huovilainen]

On Sat, 6 Jun 2009, thx1138 wrote:

> Yes, I modeled this on MatLab using the DSP56300 Tool Box and Simulink.

Is the result something like the Moog model I published back in 2004?

In that model you have for each stage
y = y + g*(tanh(x) - tanh(y))

the tanhs come from the transistor pairs.

The SSM2044 core would then be
y1 += g*(tanh(x) - 2*tanh(y1/2))
y2 += g*(2*tanh(y1/2) - 2*tanh(y2/2))
y3 += g*(2*tanh(y2/2) - 2*tanh(y3/2))
y4 += g*(2*tanh(y3/2) - tanh(y4))

The middle stages are equivalent to darlington differential pairs and 
therefore tanh(yn) is changed to 2*tanh(yn/2) to take the Vt change into 
account while keeping cutoff the same (for y1, y2 & y3).

The last stage uses feedback to essentially make it equivalent to regular 
differential pair as far as nonlinearity goes.

> The Patents on the SSM2044 probably ran out quite a while back and it could
> be nice to re-build this chip with a few improvements on Thermal stability
> For tracking a CV in oscillation mode when the Q is turned up.
>
> Still Dave Rossum did a good job on this design with Ron Dow.

Do you happen know the reason why SSM2044 was designed?
The folded Moog ladder is a clever design (and a way around the Moog 
patent?), but IIRC, SSM already had SSM2040 at the time.

Antti

"No boom today. Boom tomorrow. There's always a boom tomorrow"
   -- Lt. Cmdr. Ivanova