A Kenwood L-07TII I once owned had very high sensitivity on the bench. But with an outdoor antenna connected, it showed a fatal flaw: background noise on strong signals, and no hint of weak signals that other tuners received clearly. These symptoms suggested front-end overload.
This is the mixer output spectrum with the tuner driven by two 79 dBf test signals 200 kHz apart. Very strong intermodulation products result.
An RF gain control can solve this problem. With it you can dial back the front-end gain until a signal clears. But there was no rotary control available on the tuner's front panel, and I didn't want to drill a hole. I decided to see if automatic gain control was feasible.
The tuner used cascaded dual-gate MOSFET RF stages. Nonsignal gate G2 was biased at 7 V in the amplifiers, which had identical circuits. Lowering the G2 voltage greatly reduced amplifier gain.
A separate IF strip drove the signal-strength meter, with the first amplifier stage feeding a pair of ceramic filters. I wanted the AGC to respond to the wideband mixer signal so I fed an AGC detector from this amplifier. The ceramic filter input impedance varied somewhat across the passband, but it didn't affect the signal enough to warrant building a separate AGC amplifier.
This is the AGC detector circuit. The 2N3904 collector connects to G2 in both RF stages, pulling the gates toward ground as the signal level increases. Gain reduction begins when the input signal reaches 52 dBf. The gain reduction provided by controlling both RF stages was easily enough to handle any signal on the air. Each amplifier bypasses G2 and I saw no instability when paralleling their gates.
This shows the mixer output with AGC, with the spectrum analyzer sensitivity readjusted to yield the same peak level. All intermodulation products are gone except for two tiny pips about 57 dB down. This was the worst case I could find. The AGC dramatically improved reception in my high-RF location.
The downside of AGC is that a strong untuned signal in the RF passband not causing intermod may activate the AGC. This can raise the tuner's noise figure and make a weak tuned signal noisier. In retrospect, an AGC threshold of 52 dBf seems too low, inviting desensing even for moderately strong signals. Replacing the germanium 1N277 with two silicon 1N4148s in series should raise the threshold to 58 dBf. A more complex AGC that first reduces the second-stage gain and at higher signal levels reduces first-stage gain might help.
A Kenwood KT-880D exhibited front-end overload with my antenna pointed toward a 14.5 kW station 2.4 miles away. This orientation yielded 123 dBf (about 400 mV) at the tuner—a tremendous signal level. Several stations within a couple MHz had clear signs of desensitization or IMD. The KT-880D used 3SK122 dual-gate MOSFETs in the RF amplifier and mixer. G2 was biased at 3 V in the RF amp. Grounding G2 dropped the front-end gain 22 dB. This slightly lowered IMD on one or two signals, but much more effective was inserting a variable RF attenuator in the feedline. The attenuator cleaned up every signal if I dialed in enough loss (nearly 20 dB was required in some cases).
Many FETs become less linear when biased for lower gain. Although I saw no sign of this problem in the L-07TII, the KT-880D was a different story. I've seen the 3SK122 used in several tuners, but never with RF AGC. Perhaps now I know why.