Which capacitor for DC blocking in audio applications?

Question

Which capacitor (electrolytic, film, ceramic) do you use for DC blocking in audio applications? From what I have seen most people use electrolytic capacitors. But, also from power decoupling theory we know that we need to add a 10uF electrolytic capacitor with a 0.1uF ceramic capacitor for removing the high frequency components.

But, isn't it the same idea for DC blocking applications? Shouldn't we put electrolytic and ceramic capacitor in parallel?

Answer 1

For audio applications (as opposed to decoupling) there really aren't any high frequency transients to deal with, so a capacitor with a self-resonant frequency well above the audio range should be sufficient.

• Electrolytic capacitor: it is of course polarized, and only useful if the DC+signal is always unipolar. Also, an electrolytic capacitor has a shorter lifetime than many other types.

• Ceramic capacitor: in the high capacitance dielectric types it will have significant change in capacitance with applied voltage. This can add distortion in audio applications.

• Film capacitor: it isn't very space efficient, but it is reliable, non-polarized and has very good performance in audio applications.

Answer 2

You would prefer that a coupling capacitor have very low voltage across it at all frequencies of interest. A cheap electrolytic capacitor might have an ESR of 1-2 ohms which might vary by less than one ohm *(over the audio bandwidth**. The input impedance of a line in might be 10K. There's thus negligible voltage change (-0.002dB) which doesn't change much over frequency.

Other than leakage and noise/microphonics the characteristics of a coupling capacitor in an single-ended system are not very important- it is supposed to act as a 'short' so if it's a bit better or worse or slightly nonlinear short it's still part of a fairly negligible voltage change. Most of the signal appears across the input and almost nothing across the coupling capacitor. The capacitor inductance may start to have some small effect at >10,000Hz, but again that's not very important. Probably few, if any, human beings can tell the difference between a 10kHz sine wave and a 10kHz square wave. The overall impedance looks like the below from here. The minimum series impedance is at the resonance point (when it is just ESR). The audio frequency range will be in the area where Xc and ESR dominate and Xl just becomes important. Aside from the roll-off at the low end, it's all pretty negligible.

Filters are another matter- the capacitor characteristics such as voltage coefficient, temperature coefficient, dielectric absorption and so on matter a lot.