^^ That's pretty unusual (and also explains why a PVC works in your system)!
The problem you often run into with direct-coupling is the circuit can have bandwidth that can go lower than the actual power supplies (which always have some sort of time constant, unless powered by batteries).
If this is not handled correctly IMD will rise and there can be low frequency instability.
On top of that servos can be a bit tricky (we use servos as well in our direct-coupled outputs) and in general will have some sonic artifact unless there are at least two poles in the servo's bandwidth.
If the servo isn't right, IMD can rise and there will be low frequency instability.
If you can change the servo IC or other component in the servo without changing their values and that results in a change to the sound of the system you know you have a problem.
This is why 99% of everything out there uses coupling caps.
The problem you often run into with direct-coupling is the circuit can have bandwidth that can go lower than the actual power supplies (which always have some sort of time constant, unless powered by batteries).
If this is not handled correctly IMD will rise and there can be low frequency instability.
On top of that servos can be a bit tricky (we use servos as well in our direct-coupled outputs) and in general will have some sonic artifact unless there are at least two poles in the servo's bandwidth.
If the servo isn't right, IMD can rise and there will be low frequency instability.
If you can change the servo IC or other component in the servo without changing their values and that results in a change to the sound of the system you know you have a problem.
This is why 99% of everything out there uses coupling caps.