After installing narrower ceramic IF filters in my Sony ST-S555ES tuner, adjacent-channel selectivity was 32 dB at the standard 45 dBf test level. But it fell to just 22.5 dB at 75 dBf. The problem was evident for signals on the air as well as with instruments.
This is the IF strip. Bipolar differential amplifiers separate the ceramic filters. But the amplifier outputs are single-ended, so each stage functions as an emitter follower driving a common-base amplifier, not as a true differential circuit. With no local feedback these amplifiers are not particularly linear, even well below clipping. Clipping commences at the last stage and occurs at common signal levels.
Nonlinearities prior to the final filter in any IF strip degrade selectivity by generating in-band distortion products. With a spectrum analyzer I could see a spur appear at 10.9 MHz when injecting high-level signals at 10.7 and 10.5 MHz.
I tried adding emitter degeneration to linearize the IF amplifiers. This worked, but it lowered the IF gain enough to noticeably reduce sensitivity.
This AGC circuit improves IF linearity without compromising sensitivity. It picks up the unused meter-drive signal from the LA1235 chip fed by the IF strip. The signal rises to 5 V as the input level increases. Q1 thresholds and inverts the signal and applies it to current source Q2, which with R5 replaces the original emitter resistor in the first IF amplifier. The current source sets the stage gain. The AGC keeps all IF levels well below clipping. 75 dBf selectivity falls only 4 dB to 28 dB. Audio S/N does not degrade at any input signal level.
The ST-S555ES develops its meter-drive signal from a separate IF strip. Originally the metering circuit sampled the IF signal at the collector of Q101. Because the gain at this point was no longer constant, I added a JFET signal sampler at the base of Q101.