You can see that I didn't really need another receiver ;-) It was just fun!
The schematic diagram is here (regen_sch.jpg):
A version in better quality is here (regen.pdf):
It uses two 12AH7 double triode tubes. They just happened to be available in my junk box. The first two triodes are connected in series, so that they act as a synthetic tetrode. This "cascode" configuration was discussed in detail by David Newkirk on his excellent website.
Regeneration is controlled by adjusting the voltage of the second triode's grid. When the regeneration control is at its minimum position, that means we have 0 Volts at pin 5, there is also no anode voltage at pin 3. The voltage at pin 6 is at its maximum then. When the regeneration control is engaged (voltage at pin 5 increases) the voltage at pin 3 also rises up. At the same time the voltage at pin 6 decreases, because the tube is drawing current and there is a drop at the 100k resistor.
The second 12AH7 works as a two stage audio amplifier. I tried to reduce the number of components to a minimum so, there is no cathode RC-combination in the audio stages. The necessary negative bias voltages are produced by the small grid currents flowing through the 10 MOhm grid resistors (grid leak bias). With my 10 MOhm VTVM I can measure abt -0.65 Volts at the grids so, the actual voltage should be abt -1.3 Volts. There is no noticeable distortion of the audio signal.
The coupling capacitors in front of the grids are quite small. This helps to reduce the hum. I tried an output transformer in an earlier version (you can still see the slots on the chassis where it was mounted). But this was a disappointment. So, it was replaced with the RC-combination (25k/1µ) shown in the diagram. Now I can use high and low impedance headphones. The low impedance ones sound much more "bassy".
The detector circuit was used in a receiver that was lent to me by a fellow ham in 1968 when I just started my amateur radio career. It was a kit sold by a German company (Technik-Versand Bremen) and worked perfectly well in those days even on the 10 meter band. I always regretted having given it back ;-) so, I wanted to build a replica now. This works even better than the original.
The sensitivity is extremely high. Using a two feet test lead as an antenna the regen hears everything on 40m that the R-388 can hear (which uses a 20 feet antenna). My shack is located under the roof in the second floor. But, the regen needs a good ground! In my installation this is a connection to the heating pipe. Of course, the selectivity is not comparable. But in most cases this is no problem (the selectivity is between the ears of the operator ;-)).
Regeneration control is very smooth without any noticeable frequency pulling. The big center knob is for band-spread tuning. The cap is a very simple device which I also found in my scrap box. I had to remove some of the rotor plates to get a tuning range from 7.0 to 7.1 MHz. This was very easy because of the simple construction. The advantage of this cap is that it rotates very smoothly. I don't use any reduction drive but tuning in an SSB station is no problem at all. In the lower left you see the control for the band-set capacitor. The knob in the lower right is for regeneration control.
The PS for my regen is mounted inside the box of a butchered PC supply. I ripped off everything and just kept the case, the AC input socket, the on-off switch and two 330µF capacitors. It delivers about 70 Volts B+. In spite of the simple circuit, the voltage is quite stable and clean because the receiver draws very little current (abt 4 mA). Stabilization of the heater voltage was necessary to reduce the hum. The plug-in coil is tailored for the 40 meter band where we find a lot of active CW stations here in Germany. It has 17 turns (abt 9 µH) with the tap at 4-1/2 turns. The next one will have a lower tap so that oscillation starts at a slightly higher plate voltage.