input attenuation

[Martin Czech]

Quite a while ago Juergen Haible proposed that balanced audio wiring of
large modular systems is not such expensive, because most "real" inputs
have allready + and - pins.  E.g. otas, ladder differential stages etc.
We simply don't use the other input.
But some means of input attenuation is needed, this would mean a dual,
good matching pot, thus this proposal was more or less negated.

My design goal is "clean signal", I have spend money on real 16bit ADC
balanced recording. I would accept noise in "vintage circuits" (maybe
part of the original sound), but no other artefacts, like hum -70dB down,
or radio. I do not assume that the input signal of some module is
allways "on", but that there may be silent passages, too.

Is it really necessary to have a dual pot ?  I don't think that most
modules need -80dB, but only -20dB or so at the input. We simply want to
control distortion, and S/N. This could be done with a linear or audio
"crossbar" pot from the - to + line.  Ie. + input , 10k resistor, pot,
10k resistor, - input.  The voltage across the pot is then fed to the
usuall ota input voltage devider.

The input impedance is thus variable, but this won't hurt.  The CMRR will
not be better then -40dB for 1% components, but I think this should be
enough if combined with twisted pair cables and shielding. The CMRR will
of course get worse for higher frequencys, a crossbar cap and maybe two
additional caps to local input ground would cut common and differential
mode hf dirt down.

The only real expense would be to have a balanced output driver, that
is one op amp more, and stereo jacks...

And a 2P2T switch would give input inversion, without extra cost...


m.c.