[sdiy] Exponential Keyboard (KORG MS-20)

JH. jhaible at debitel.net
Sat Sep 1 12:50:04 CEST 2007

Try to locate US patent #3918342 about the Korg divider.


----- Original Message ----- 
From: "Tony K" <kalotony at videotron.ca>
To: <synth-diy at dropmix.xs4all.nl>
Sent: Saturday, September 01, 2007 11:42 AM
Subject: [sdiy] Exponential Keyboard (KORG MS-20)

Hi list,

  I'm building a keyboard controller for my MS-20 clone project using
a PAIA 37-note mechanism. It already has the diodes wired
into a 8 X 5 matrix and I'm tempted to use that with a uC and do a
digital scanning version with DAC (8051 based - I know I should
really graduate already, but those MMT8's make great development
systems - oops the secret is out ! ).

Analyzing the MS-20 keyboard circuit I'm a little confused about how
the scaling factor (427/403= 12th root of 2) is actually
'multiplying' the subsequent network. I tried this on paper by
starting off with 1.059 V, low C, and doing the math, but I don't get
the expected next value of 1.122 and so on down the string. I've
re-drawn the network and used the voltage divider rule it what seems seems
to be a simple network. Is the network 'on the left' affecting the
voltage as seen from the right (output) - it shouldn't as each key closure
is the 15 volt regulated supply. I don't remember but was the KORG
high note priority ? The legacy is last note priority but I'm not
sure it emulates
the keyboard faithfully (in single trigger mode).

Also the resistor values of 403 and 427 ohms are not available from
the usual sources -strange. The JH-720 use 430/402= 1.070 to four 
figures. Any other ratios that work as well as the original ?

I got excited about using a DAC-88 exponential DAC but those aren't
true log/exp DAC's only u-255 law and for audio use. I suppose the simplest
approach is to use a multiplying DAC and a look up table for those 37
keys (already calculated each keys voltage output).

All this leads me to scrap that part of the circuit altogether and do
a straight forward 1 V/OCT design by replacing the exponential input
scaling resistor R38 with a 24 turn MT trimmer - not ideal but I know
this works. Only things that concern me is the lower impedance seen
by the exponential converter and now FM wouldn't be Hz/V changing the
character of the machine (?).

I guess I'm thinking out loud at this point, but any opinions or
advice is welcome.

Tony K

P.S If anybody has a source for those precision resistors (403 and
427 ohms) I would love to know.

Synth-diy mailing list
Synth-diy at dropmix.xs4all.nl

More information about the Synth-diy mailing list