Patchable modulars

[Magnus Danielson]

> Ray Peck writes:
> 
> 
> 
> This brings up something I've been thinking about for a while....
> 
> 2. computer control of all patches and parameters.
> 
> 
> 
> Check out the Harris matrix switcher IC's. The most dense unit I think they
> offer has 128 analog switches that go from DC to 50 mhz with very little
> non linearities over that range. 40 pin dip style or PLCC.
> 16X8 patchable, any or all switches on simultaneously and has a clear all
> pin and built in latches.
> Very low noise, offset and crosstalk too.
> 
> 4 of them could be a 32 x 16 patcher or a 64 x 8. There are lots of
> possibilities.
> 
> 
> P/N is CD22M3494E
> 
> They cost less than 8 dollars each.

Oh, only 8 dollars a piece.... then I will use them!

I have been looking at them some time now but now they are sure closer to show
up in my modular.... whenever that one gets completed (never by definition... 
).

There is BTW other sources for the same IC but with a sligthly diffrent pinout
(I think it is 4 pins or so being interchanged) so it may be wise to make
jumpers so that replacements can be used if necessary.

> In fact, Harris was kind enough to send me 10 of them as samples for no
> charge whatsoever.

I'd better call my Harris agent rigth away :)

> Mitel makes a duplicate of this chip as well.

Packages: 40 Pin Ceramic DIP

MITEL	MT8816AC

Packages: 40 Pin Plastic DIP

HARRIS	CD22M3494E
MITEL	MT8816AE

Packages: 44 Pin PLCC (Mitel lead compatible)

HARRIS	CD22M3494MQ
MITEL	MT8816AP

Packages: 44 Pin PLCC (SGS Lead compatible)

HARRIS	CD22M3494SQ
SGS	M3494 (sorry, lack details on SGS)

The two PLCC pinouts differs in only 4 pins which could be good to make
interchangeable on the card so that there is support for both pinouts.

These switches are really easy to control and is dead easy to expand (if you
don't care about contact resistance and fan-out problems).

The datablades of both the MITEL and HARRIS chips claims a 0.01% THD @ 1 kHz
and 2 Vpp. I guess that would suffice for most synth's.

>From a control point of view you can see them as a 128 x 1 write only memory.
I guess that maybe the best way of haveing them CPU controled it to have them
connected to a buffered databuss and let both the address lines and the data
bit (they are 8 in total) connected to the data bus. The write pulse is 
connected to the STROBE inputs and then a simple chipselect circuit to select
which chip to create the individual CS signals.

Cheers,
Magnus