MOTM

> I'm still struggling with the concept. Is it a fixed architecture
> voice? If so, how do you control VCO volumes into the filter, poly
> mod amounts, resonance, etc - all those things in a P5 voice which
> required a separate VCA?

Just an ordinary Modular, but 4 parallel inputs and 4 parallel outputs
instead of single ones.
For instance a VCF module will be not one VCF circuit, but 4 VCF
circuits, with individual inputs, outputs and modulation signals,
but with shared manual controls. So, one knob for cutoff frequency
will set the initial cutoff for all 4 filters, but each filter will get it's
own envelope, keyboard tracking and so on.
Where you usually would have a pot to scale an input or output,
I don't use a quad pot, but 4 VCAs and a single pot.
I'm putting these VCAs at the output of the modules, so every
modulation depth, filter overdrive etc is controlled by the level
of the *sending* module; inputs have fixed sensitivity. (This
concept works nicely in the VCS3 for instance.)
As I have a VCA at the output of every module anyway, I won't just
use it to emulate a quad pot, but I can sum individual CVs to individual
VCAs as well, so I get a complete VCA with every module for free.
Take 2 VCOs routed into a VCF as an example: The built-in VCA at
the VCO output can just be used to manually set the mix of both VCOs
and the filter overdrive. But it can also be used for dynamically mixing
the VCOs with an extra envelope. Like: 1st VCO goes to VCF with
manually set amount, second VCO has manual output level set to zero
and is controlled by a short AD envelope to be active for a shor time
only.
So, as long as there is only manual level control, the 4 VCAs per module
are surely an overhead, but as soon as I patch dynamic mixing etc. the
concept starts to become interesting.

> Or is it something wilder, with physical patch cords representing
> somehow identical patching in 4 sets of electronics? How in the heck
> would you pull that off without an elaborate electronic switching
> matrix?

I'm using 5-pin "Midi" cables. Not the worlds best connectors, but
available and affordable. It will be clear when you'll see a photo;
on the front panel layout you just see the 16mm holes and two tiny 3mm
fixing holes for each 5-pin jack.

> I'm so curious because I designed many poly synths in the old days of
> SSM chips, but still never built any because of the sheer amount of
> hardware required.

Yes, the wire harness inside the first module is quite impressive. I've
made a detailed plan of a mechanical concept before I started, and
apparently ot works. For the interface module I have 2 strong metal
braces attached to the front panel (19" 3U) which hold a rectangular
wooden frame. The frame is made of 15mm x 15mm wood, and it spans
an area that holds 4 eurocards (160 x 100 mm) such that the bottom of
the pcbs (veroboard, actually) is acessible from the bottom for repair
and mods, even when the whole thing is assembled. Wiring to front
panel comonents is done from the upper side of the pcbs, with ribbon
cable for unsensitive signals, and 4-wire shielded cable for sensitive
signals. The wires are soldered at the front panel components, and there
are tiny pcb connectors on the veroboard end.
Every module will probably get its own tiny pcb mount toroidal
transformer, so I don't expect grounding problems and everything is
self contained.

JH.