[sdiy] DC blocking caps on inputs - or not?

[Neil Johnson]

Input coupling capacitors prevent unwanted DC biases from upsetting
your circuits.
Output coupling capacitors prevent your DC biases (wanted or
otherwise) from upsetting external circuits.

If you don't know the polarity of the external bias then use bipolar
electrolytics, which oddly enough seem to exhibit less distortion than
their polarised versions.  And talking of distortion Self has
experimental evidence to suggest that you want to keep less than 80mV
of signal across electrolytics, so think about the impedance network
and the lowest frequency of interest.

There are definitely places where you do need DC-blocking capacitors:
around pots to stop the scratchy sounds due to DC through the
track-wiper interface for example.  If you want to preserve anything
of the bottom end then you really should be aiming for around 2-3Hz at
the -3dB point of the coupling network.  By the time the signal has
been through 10-20 of those the phase distortions soon start to add up
and the -3dB point creeps up.

As in all things engineering, there is no one-solution-for-all.
Sometimes adding components makes things better - balanced outs and
ins for example.  Also, the "100k input is fine" mantra needs to be
carefully weighed up against the objectives of the system and how the
input is structured: a non-inverting buffer with a 100k bias resistor
is OK (the thermal noise will be dominated by the source impedance),
whereas an inverting buffer with 100k input resistor is probably not
OK as that is adding quite a lot of thermal noise into the audio path.

Neil