My goal is to build an all-band communications receiver for AM, SSB and CW.
Therefore I have build a prototype IF-strip using 2 tubes EF183 (variable transconductance),
and did some measurements at it. The strip functions at 4.433 MHz using surplus colour-TV crystals,
but I intend to use a (commercial) 9 MHz crystal-filter. Here you find the schematics, and the prototype looks like this:

I have measured the gain of this strip, and how the AGC performs. The results are quite satisfactory:

Here the gain is shown as a function of the negative grid voltage on both tubes, with
two different inputlevels. It is clear that 100 mVtt is too much  since the gain is
clipping.

In this graph the level of the outputsignal (RF) is plotted against the inputlevel, with
3 types of autimatic gain-control: none (blue line), a single OA81 (purple) and two
1N4148 doubling the voltage giving the best result in the yellow line. The input
varies between 25uV and 400mV top-top.

Another graph showing the gain versus input with the AGC-types. The yellow line
fits very good; the gain varies between 66 dB and 9.5 dB, so the range is 56.5 dB.

Final graph shows the output audiolevel (purple) and AGC-voltage (blue), behind the
productdetector, against the RF inputlevel between 10 uV and 400 mV.