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

[Mattias Rickardsson]

On 6 February 2016 at 14:11, P Maddox <yo at vacoloco.net> wrote:
> Matthew:
>>> Another capacitor question popped up:
>>> What's your view on having electrolytic caps right at the input of an
>>> audio
>>> device in order to block DC and/or protect the circuit?
>>
>> I think it's not so much a question of "protection," as of allowing you to
>> choose the DC bias point of the first stage.  If you don't block DC, then
>> anything you connect to that input is going to affect the bias in a
>> hard-to-predict way.

Yes, I figure it's more of a bias thing than a protection thing. It
could also be a noise modulation thing - some noise sources increase
with signal level, and a DC offset means a higher level even if the
signal is silent.

> I'd agree... I'd go a step further also.
> If you don't *HAVE* to use caps in the audio path, don't.

In many other audio fields you are more lucky than in the synthesizer
field. An amp or a CD player can have DC more or less right through
the audio path without problems, but every VCA or signal switch in a
synth needs to have a DC-free signal in order to avoid thumping or
clicking like a Minilogue. ;-)

> Everything you put in the path affects the signal, so the less you can put
> in the path, the less the signal is messed with.

That's true. Even the phase, as mentioned before. I generally prefer
staying under 1 Hz as -3 dB point if I can afford it.

>> I'd avoid electrolytic caps in this application because they don't
>> tolerate reverse voltage, and so it'd become necessary to protect the cap
>> from that, one way or another.  Not so much an issue of signal quality.
>>
> yeeee, no....
>
> every cap does have a reverse voltage limit, all be it much smaller than the
> forward voltage limit, but you can get caps that work 'bipolar' for not much
> more than unipolar caps.

There isn't any voltage over a DC-blocking cap unless you are in the
low frequency range where it actually blocks DC. So if you don't
expect a DC-offset situation, and keep the cap big enough to pass all
the expected signal, then it doesn't matter much that the cap is
polar, and its marked voltage doesn't matter much either - it could
even be less than your expected signal swing.

/mr