On the old Radioboard the circuit which I posted which had the largest response was a regenerative medium wave receiver using only an LM386 audio amplifier chip.
The front end of the LM386 was configured as a Colpitts oscillator using the intrinsic Base-Emitter capacitance of the input transistors and an external 220pF capacitor. The tuning inductor used is a standard MW ferrite rod inductor and the tuning capacitor is a PVC MW 2 gang variable.
Because the gain of the LM386 used in these configurations is so large care should be taken to use short lead lengths and a ground plane would be preferred. Using a 9v supply a speaker may be driven to a comfortable volume with the MW version.
Regeneration could be controlled in a number of ways. In the simplest version reducing the value of the 1K regen variable resistor increases the audio gain, but reduces the regeneration. Increasing the value the opposite occurs.
The three other methods of controlling the regeneration include a 1 mH choke which isolates the RF gain from the audio gain. Audio gain may be varied by adding a 100 Ohm variable resistor in series with the choke. Changing the DC operating point of the LM386 with the regen control inhibits the oscillation. These alternate methods can be tried if the technique used in the simplest versions of both the MW and SW receivers proves to be inadequate.
The shortwave version is basically the same as the medium wave version except for the values in the tank circuit and the omission of the 220 pF capacitor. The intrinsic capacitances in the front end of the LM386 provide the proper ratios for Colpitts oscillation at SW frequencies. Using a 9V supply and chips from National or Samsung the SW version functioned well to over 8 MHz which includes the 80 and 40 meter amateur bands. Using a ferrite rod with 20 turns no external antenna is needed to receive a considerable amount of activity on the SW bands.
The links below are videos posted by a radio amateur who built an earlier version of the MW receiver.
youtube.com/watch?v=-UCACxbQS1U
youtube.com/watch?v=PTWd82KI53s
I got one of those TRF AM780 kits several years ago to play with, and it was absolutely dreadful as a MW receiver in my location - zero selectivity, which I think the TA7642 is somewhat infamous for.
I have used the TA7642 in other applications, and it is not too bad when used as the IF amp in a superhet following the selectivity, but it is a just awful in the AM 780 TRF configuration.
I have been sort of looking for a ZN414 to put in the AM780 to see if that worked better; but I may try to recreate one of these regenerative receivers with it when I get a little farther through my project backlog.
Thanks for the tip.
Win W5JAG
For any interested in trying this citcuit. Elenco makes a MW kit with all parts necessary which uses a TA7642 as the front end.
On one of the receivers which I constructed I removed the TA7642 and with a few modifications reconfigured the wiring to create the receiver in this thread.
I came upon these circuits a while ago and they started me back on my hobby in electronics. Most of my attempts at these worked but the oscillation was so strong and so unpredictable they really were quite squealy. I didn't know what I was doing but they did work. It's really cool that you posted here, I appreciate it. Lets see where qrp-gaijin can take it
I've been thinking about building a shortwave regen again (starting to get bored with my well-working but complex 1.2-volt regenerative superhet), and this circuit is interesting, though I wonder if it's possible to make it work above 8 MHz.
For example, could you use an off-chip oscillator, perhaps the well-known cross-coupled/differential pair oscillator, as the Q-multiplier -- which would operate all the way up to 30 MHz -- and then use the LM386 as a high-gain detector and AF amp? Implementation should be trivial -- just add a link winding on top of the ferrite rod antenna, and connect the link winding to the two-transistor cross-coupled oscillator (whose regeneration is controlled via emitter resistance). It should be possible to use a couple of diodes to make a voltage regulator for stable operation of the Q-multiplier.
The assumption is that using the LM386 as the detector for the Q-multipled RF signal is more sensitive than attempting to detect the RF with a separate transistor, then feeding the AF into the LM386. Do you think the assumption is correct? I seem to vaguely recall that your early experiments may have involved these kinds of circuits (an off-chip Q-multipler, plus using the LM386 as envelope detector), so any comments you have would be appreciated.