I have built a chopper-controlled high-pass Butterworth 8-pole filter for removing rumble and thump from cylinder recordings. The chopper ratio is controlled from a servo loop so that all the controlled resistors follow a master resistor in a D.C. bridge.
When the master resistor is O/C, the chopper switches open completely and the noise floor of the filter drops well below -60dBm, but with the chopper operating (at around 50:50 ratio) the noise floor comes up to
-50dBm. No particular device seems to be at fault, the noise seems well distributed throughout the filter.
There are some large switching transients apparing at the output of the audio amplifiers, so I would suspect that intermodulation could be at the root of of the problem - perhaps someone can explain the mechanism? Any ideas how to solve the problem without crippling the H.F. audio response?
The input signal level is around 0dBm and it contains some unavoidable
45 Kc/s crosstalk from a sensor in another part of the equipment, hence the 45 Kc/s filters on the input. The residual sensor signal in the remaining stages is undetectable (and the noise is still the same with the sensor disconnected).The bridge amplifiers are all in a TL074, the audio amplifiers are NE5532s and the choppers are CD4053s. All the chips in the filter run on well-smoothed and decoupled +7.5v and -7.5v rails, the other op amps have +15v and -15v rails.