After the 1R5 one valve superhet mostly worked, I added a 1S5 diode-pentode as a detector and audio stage.
The two valve circuit is also taken from an RCA tube manual, using the mixer and detector/1st audio stages.
The 1R5 stage is the same as the One Valve Superhet, with R1 removed from the grid bias, and an AGC line connected from the detector circuit. (Yes, the AGC should connect to the L1 - C1 junction, but the layout made that awkward. I didn't think that far ahead when setting out the parts.)
Notes on initial build:
Output is currently taken by just putting the headphones across R6. I need to figure out a proper output, that doesn't involve putting B+ through the headphones.
I moved the 1R5 plate connection to a tap on the first IF transformer. I wish I'd tested that on the single valve version, it made the audio sound noticeably clearer. Improved Q on the first tuned circuit? Some web pages say that transistor circuits use that tap for the collector connection for that reason.
AGC has very little effect. I know it's connected to the wrong place, but it seems to work the same either way, and that was a lot more convenient. (I made the layout up as I went along.) The end of the T3 winding only seems to vary from about -0.4V for no signal to -0.9V for the strongest signals. The data page for the 1R5 seems to suggest that the usual range for that line should be 0 to -4V, though that is with an IF amp stage. After the 1Meg resistor, there's probably not much of an effect, although I don't have a meter sensitive enough to see what's really going on at G3.
IF transformer experiments
Tried a real valve IF transformer, "sharp response 455kHz" from eBay. It has a 1:3 step up ratio, which seemed to improve the volume a bit, and also AGC action moved up to -0.4V to -1.4V. Tuning was sharper, otherwise seemed about the same. Had to do some adjusting of the oscillator coil to get it to oscillate again.
Went back to the transistor IF, but connected the same way as the valve ones, with plate and diode connection at one end of the coil, and B+ and AGC/Gnd connections at the other. AGC action is now -0.4V to -1.4V, same as with valve IF, although tuning is back to previous sharpness. Oscillator coil worked at same settings as for valve IF, did I have it a bit off before?