[rfeenstra]

The sensitivity on my restored 20 varies dramatically across the dial, being least sensitive at the low end.  Since I had to rewind the primaries on both coils, I thought maybe it was a problem there.  However, that rewound portion of the coils is in the plate circuit and not really tuned.  Hooking up the spectrums analyzer to check the antenna coil and the 2 intermediate coils yielded a flat response across the dial. This was with the radio off, the sweep hooked to the primary and the analyzer input hooked to the grid wire with isolation resistance.

Turning the radio on yields a very different result.  With the volume on full, sweep hooked to the antenna, and the analyzer input hooked to the grid of the plate detector, the signal level drops about 5db from 1500 kc to 550 kc. With the volume turned quite a way down, signal drops 14db from 1500 to 550.  Quite dramatic. 
Much of the volume control on these radios is via a change in bias on the 2  24 tubes (the volume control changes the cathode voltage while the grid is at 0 volts.) Grid bias is at -2.9 volts at full volume and -11 volts at minimum volume.

So the question to you tube gurus is:  Are the #24 tubes less sensitive to lower frequencies when approaching cut off bias?  If so, then there is potentially no solution to the lower sensitivity without restructuring the entire volume control system.  I would rather keep it original.  I may try changing the 2 tubes to see if there is any difference.

[mikethedruid]

I have noticed this same drop in sensitivity as one goes down in frequency of the AM BCB on several radios. I don't think it is restricted to the 20, or to particular tubes. It may be an interaction between the tuned RF circuit and the antenna, as I have noticed the need to adjust the antenna trimmer on my communications receivers between the low and high ends of the AM BCB range.. Have you tried using an antenna tuner between the antenna and the radio?

[rfeenstra]

No, Mike, I have not. this shows up dramatically with the spectrum analyzer. Interestingly, I changed the inductance of the second intermediate coil (the one that feeds the plate detector) by inserting brass in the coil tube. I was able to increase the 550khz signal by about 6 or 7db without affecting the sensitivity at the high end of the dial. I'll glue the brass in rather than unwind some of the secondary to reduce inductance. At least I gained a little!