user 37518 said:
I read in a book that DC servos can be as bad as caps in low frequency phase shift, I believe Deane Jensen did some work on this.
Beware when the term 'phase shift' is brought up as a source of bad sound. Most of the time, it means that the person blaming phase shift really doesn't know why things sound bad, but it's simple enough to blame 'phase shift' as a mysterious cause because it sounds somehow plausible and it's been used as a culprit many times before.
In a minimum phase filter (basically a filter with no extra delays added in), the phase shift and magnitude response are tied together inseparably - if you want a certain magnitude response (frequency response) then you have to have a certain phase shift, no more and no less.
So, if you obtain the same highpass from a servo that you obtain from a coupling cap, then their phase responses must be exactly the same. Something called the Hilbert transform (and a few assumptions about causality) will let you transform magnitude into phase and vice versa with mathematical exactness.
This doesn't mean that the two filters have to sound the same - the servo can have all manner of other flaws unrelated to magnitude or phase, and so can the coupling cap. But, beware when 'phase' is trotted out as a universal source of ephemeral bad sound. It rarely has much to do with anything related to 'bad sound', it's just the necessary result of a filter with a specific frequency response.
All that said, I find DC servos to be quite effective in terms of size, cost, and performance compared to coupling caps for low impedance audio circuits. They also have the benefit of correcting the offset of the amplifier that they're wrapped around, and not just highpassing the signal passing through, so you get a well defined DC offset from the entire system, and can thus use signal handling amplifiers that have slightly sloppy DC performance because they have great audio performance. These days, for about $5 in parts, you can make a balanced DC servo with a net residual offset of about 5-10µV that takes up a footprint of 10 mm x 12 mm and is no more than 1.75mm thick. Sure, you can screw them up if you're careless, but you can also get them right too, just like coupling caps.