An interesting circuit appeared in the April 1961 edition of CQ Magazine; "The RXT-2," by W6MMC. This was a 40m regenerodyne receiver using three Philco transistors and one Raytheon CK-768. A crystal-controlled converter was followed by a regenerative IF amplifier/detector; tunable from 360 to 650kHz. This, in turn, drove a two-transistor audio amplifier.
This circuit came to mind when I discovered several 2N107
and 2N123 Germanium transistors in a box of miscellaneous
semiconductors purchased at a hamfest. The 2N107
was introduced in 1955, as a way to generate revenue from the
large quantity of rejects coming from the 2N43, 44, and 45
assembly line. These devices were sorted at the end of what
was essentially the same manufacturing process. The best devices
were given the 2N43 designation. The next lower specification was
the 2N44. Those devices not meeting the specification for
the lowly 2N45 were dumped into bins destined
for disposal, until an engineer by the name of Carl David
Todd convinced GE to sell a subset of these rejects
to hobbyists. Here is a page concerning the 2N107
from the online Semiconductor Museum.
Todd relates a story about an engineer working for a large toy company that stumbled upon a batch of 2N107's rejected because their specifications were too far above that of a 2N43. It seems the poor fellow designed a talking doll using these devices (you can read the passage here). It's true the engineer allowed himself to get snookered, but the thought of making QSO's with a receiver built from transistors rejected for use in a 1950's talking doll sounded like too much fun to pass up!
All three of my 2N107 specimens function well as regenerative detectors at 400kHz. I also noticed right away the regeneration control is quite smooth for a bipolar transistor. This may be a result of the unusually low current gain of these devices. A tunable (296 to 366kHz) IF regenerative detector drives a second 2N107 AF amplifier. A crystal-controlled (7376kHz) converter was fashioned from a 2N123.
L2 and L3 are ferrite-cored junkbox finds, each having a measured unloaded-Q of 44. The narrow band impedance matching circuit comprised of L3, C11 and C12, present Q3's collector with an 8,500 +j0 Ohm load at 650Hz. Q3's collector current measures 665uA, while the impedance looking into the base of Q3 is 1,550 +j0 Ohms at 650Hz. L2, C8 and C9 transform this impedance to upwards of 14,000 Ohms - again, all resistive, at 650Hz - at Q2's collector. The loaded-Q of the entire audio section is 3.8. The measured power gain from Q3's base to the headphones is 34.3dB. It's interesting to note that the low-frequency current gains of the 2N107 transistors at Q2 and Q3, are each a mere 20.
The total current draw for this receiver (at a supply potential of 6Vdc) is 960uA. The minimum detected signal is -119dBm (0.25uVrms), while a comfortable copy requires a signal of -107dBm (1uVrms). The receiver noise figure has not been measured, but neither has it been an issue. The frequency stability appears to be up to the task. Of course, the front-end is not exactly "bulletproof;" an extremely strong signal in the vicinity of the operating frequency will cause it to fold-up.
Other than that, it's a pleasure to cruise 40m with a receiver made from three hobbyist's Germanium transistors that were manufactured while the Everly Brother's Bye Bye Love was at the top of the charts. All three transistors that I use were manufactured in the Spring of 1957. As it happens, the 2N123 at Q1 is date-coded the very week that I was born!
Here is a short audio recording of VE3OU's signal as heard on this receiver.
above image to enlarge
The above scan, taken from my design notebook, shows an experiment in which a tunnel diode converter was used in place of the Talking Doll's transistor converter. A number of QSO's were made using this set-up.