Improving DC-servos

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
unclejed613 said:
and it's on the verge of becoming an oscillator (hence your strange behavior at 1khz). the feedback is contensing with your input stage for control of the overall system, and making it unstable
Yeah; there are too many poles in the loop, making the system marginally stable. Each pole can be compensated with a zero, but this requires accurate matching of the components, and it adds to the complexity.
I think a similar result could be reached by a clever rearrangement of the topology.
But until now, I've had no illumination.
This is why I call luminaries to my help.
 
the example i posted works quite well. it controls DC well, and rolls off the amplifier response below 5hz (important if you still listen to vinyl). with the Jfet switch across the integrating cap (not shown because of picture size constraints), the amp does not get "confused" when the power is cycled repeatedly (it doesn't power up with the servo or a feedback capacitor in an undefined state). the 100uf electrolytic shown was replaced in production by two 220uf caps in opposition (in series, + to + to avoid use of a nonpolarized cap). since the isolation was through a 1Meg, even a latched rail at the output would not damage the cap (IIRC the caps were 25V, even though the rails were +/-65V) or the op amp. using a servo keeps the AC feedback loop free of reactive components, and negates the need for a feedback cap rated at or above the rail voltage (which would also preferably be a nonpolar cap).
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.