Looking for controller build hints..

[Byron G. Jacquot]

>Any recommendations on how to do this multiplexing?
>Any tips for a CPU that can be *easily* interfaced into such a set-up, with
>say onboard ram, flash rom, serial ports, and perhaps a simple way to
>interface to an LCD?

I did something very similar with a Mot 68HC11 a while back, but only 48 knobs.

The ADC in the '11 can do 4 channels multiplexed in parallel, into 4
internal registers.
The E9 part has 8 input pins to the ADC, and a register for selecting which
set of 4 it reads (hi nybble or lo).  It can also be set to a continuous
conversion ascheme, so that the value in the converion register is always
fairly recent.  So you have the basis of an analog scan matrix right there.

I used a 74LS154 4:16 decoder/demux driven by an output port as a select
line, which (through a buffer/inverter transistor)selected a group of 3 pots
together.  The pots were wired as voltage dividers, with the top connection
coming from the demux chip.  The selecteg group went to Vref, the others
stayed low.

The outputs from the first second and third pots from each group went to an
opamp summer, and then back to the ADC for conversion.

The algorithm for reading went like this: 
Select a group of pots.  
Wait for the conversion to happen.  
Read the ADC result registers.
Select the next group and repeat...

The 11 itself is reasonably easy to interface, with onboard EEPROM, RAM,
EPROM, and serial ports.  Toss in a few more decoders, and you'll be
alright...a 4*32 matrix would address 128 pots.

You might want to maintain the value of every pot in a memory buffer, and
only send MIDI messages when a pot has changed, to keep MIDI bandwith low.

Byron