I can totally relate to that and I was a bit harsh in my comments. Sorry about that. I just had a crap day yesterday.
Experimenting for the sake of learning and having fun is something I do a lot of myself.
Anything that moves with signal and causes a variance in the transmitted signal can be considered in the signal path. However this truth doesn't negate the effectiveness of a DC servo as seems being suggested.
Capacitors in the DC servo are not passing full magnitude high frequency signal currents through it to a subsequent network. Rather high frequencies are being post filtered from the low impedance side relative to a virtual ground in a servo, becoming then turned into a DC current fed back through some resistance to the input. The manner and extent of the feedback, and the capacitors used, determines the extent of passthrough signal currents left over, not the entirety of those currents.
In a recent high gain application the nature of the amplification was such that the offset could be trimmed to near zero output, however this could not account for thermal drift. Hence the output was trimmed for zero output out of the DC servo, that when left alone the servo adjusted for the thermal variances within its confined range of operation.
In an application that eliminates the capacitor in the feedback, the input error voltages are magnified by this gain at the output. If the output is connected to a transformer, such as in electrostats for example, the current can become significant.
