> no more brass parts). however, each octave is about 20 cents flat -
> not really a huge problem with one octave, but with three octaves

i'm

> more than a quarter tone flat.
>
> i don't have the precision adder, so i can't try out that solution.

This is probably simply caused by the 'protection' resistor in the A-
170 output: at 1k it is higher than those in other modules handling
precise-CV-type signals, which are typically around 100ohms. A-100
inputs are typically either 100k if no pot (like A-110 CV1 in), or a
50k pot (like A-110 CV2 in); if feeding two inputs, it would be easy
to get the input impedance down to around 50k, and then a 1V input to
the A-170 would be seen as 50/51 of a volt at the output, or 0.98V,
so 20mV too low. 1V/octave is 1/12V = 83mV per semitone, so 20mV is
around 24 cents too low (which seems to tie in with your figures).

A simple solution would be to locate 'R2' on the PCB (above the 'A-
170' designation), and either replace it with a 100ohm, or simply
solder a 100ohm in parallel with it (i.e. across its legs), and this
will reduce the error by about a tenth, i.e. down to a few cents
instead. (And that will still leave you with *some* protection
against shorting the output, which is what it is there for I believe.)

Tim