Now this is a bare-bones circuit! For example, the base to collector leakage provides the working DC bias and the headphones double as a radio-frequency (RF) choke.
I had the circuit working on the breadboard in a matter of minutes using a 455kHz shielded IF "can" in place of their "loopstick." My 2N107 transistor is date-coded February of 1957 and my "4k Ohm" headphones date from the same time-period.
As is typical with these early, low-gain bipolar transistor circuits, the regeneration adjusts very smoothly and without hysteresis. While one expects a fairly narrow peaked response (especially with a ~350kHz operating frequency), I noticed what appears to be a positive feedback loop for audio-frequency (AF). Increasing the overall feedback past the threshold of RF oscillation progressively narrows the audio bandwidth until the point of sustained AF oscillations is reached. While a small dose of AF peaking is not unwelcome for CW reception, too much results in an unpleasant "sea shell" effect, along with unnecessarily critical beat-frequency tuning. Fortunately, as it stands the AF bandwidth isn't excessively narrow at the optimum RF feedback setting.
With the aim of producing an ultra-simple 40m superheterodyne receiver, I next threw together a self-excited mixer, or converter, to precede the intermediate-frequency (IF) regenerative amplifier/detector. The circuit I chose was well-known in the late 50's and early 60's. For example, it appeared in the "Hints and Kinks" column of QST Magazine in 1960.
Here it is again in the March 1963 edition of DL-QTC; "Kristallgesteuerter Transistor-Konverter"
I chose to use a 2N78 in my 40m converter stage. This rate-grown, NPN, germanium transistor was introduced by GE in 1955.
“The General Electric NPN transistor type 2N78 is designed for RF and IF amplification in broadcast receivers. It employs a new and revolutionary manufacturing technique – the exclusive GE rate-growing process." General Electric History
Although the "alpha cutoff frequency" for the 2N78 is only 5 or 6MHz, it functions well enough as a self-oscillating mixer on the 40m band. The device I used is date-coded October of 1956. The snapshot below was taken from the 1956 Radio Shack Catalog. The 2N78 would cost $77.32 in 2010 dollars.
Here's the circuit that I'm using.
C3, C4: 10nf
R1: 52k Ohms (optimum value depends upon transistor used)
R2: 100 Ohms
T1: 3.5 to 6uH slug-tuned; 16 turns with 3 turn coupling
T2 455kHz shielded "yellow-slug IF Can"
X1: 7.3764MHz quartz crystal (salvaged...like everything else ;o)
L1: 5.6mH molded plastic inductor
L2: 3.3mH "gum drop" inductor
CV1: 15 to 360pF, 6992 to 7070kHz main tuning capacitor
CV2: 15 to 360pF, regeneration control
HP: high-impedance magnetic phones
Q1: 2N78, NPN germanium transistor
Q2: 2N107, PNP germanium transistor
Frequency stability is excellent although grasping the headphone cable originally produced a slight shift in the receive beat-note. The RFC and capacitor low-pass network connected between the 2N107 collector and the headphone (shown in the above schematic) eliminates the problem. I can clearly copy my weak-signal test generator/step attenuator at 0.2uVrms on this little receiver.
The first station I copied was VE3BYR/QRP, operating north of Toronto with an excellent, S-7 signal. On the next evening I copied TY1KS in Benin at 559 among a dozen other DX stations. I easily copied OE5KC with the headphones lying on the workbench. Obviously he was 599! When N4NQ mentioned he was running 75w to a Windom antenna, I connected my step-attenuator between the antenna and the receiver with the 33dB pad switched in. He was a 439 copy; the same as he would have been if he were running 37.5mW with no added attenuation on my end. I was equally pleased at what I did not hear this evening...no SWBCI whatsoever!
I swapped several 1950's vintage 2N169 and 2N170 transistors in place of the 2N78 with roughly the same results.
Here's an old German transistorized regenerative detector circuit that's nearly as simple as the one shown above. This one uses a Siemens AC-151 junction transistor (ft ~ 1.5MHz) in the common-base configuration. I've lost track of where this circuit came from, although I suspect it was from one of Herbert G. Mende's books; late 50's to early 60's. (Yes, DL3PB confirmed that it's from H.G. Mende's, Leitfaden der Transistortechnik. Thanks Peter!)