Well at least the base of Q2 operates at ultra low current, a few nA.
Q1 and Q2 act as a current mirror at DC. The current leading into the collector of Q1 is about 1uA. And that current is replicated at the collector of Q2. That means the current at the base of Q2 must be about 1uA/Hfe of Q2. A few nA.
Also the 1uA at the collector of Q2 is fed into the base of Q3 where it is multiplied by its Hfe, giving say a current of 300uA at the collector of Q3. You may be able to increase R2 or use an audio transformer in its place.
L1 and C2 form the resonant circuit and are coupled to an antenna some way of your choice.
The input impedance of Q2 under such low current is very high, however there is a lot of base emitter junction capacitance involved. That would cause a lot of phase noise if the circuit was used as a Regen detector but it is no problem for a TRF radio.
I built the circuit above, it works, but is not sensitive. The output volume is very weak. I'm driving a lm386 module at full volume. It receives powerfull stations in my area. I am using a litz wire ferrite antenna, which works great on a KM484 etc. if I connect a 2 foot wire to the coil, it works better, but not much. I am using 2N3904 and 2N3906. Tried various other transistors, all the sane result. The sound quality , while burried in noise, is good.
Here is the original
I tried this circuit as is above. It works, but the output is low, sensitivity is OK, with litz wire on a ferrite core antenna. I'll try an lm386 amp even though they are terrible, it has high gain. Seems like it needs a premap out or a buffer. Ill try the suggestions below.
I've only had a limited amount of time for testing but this version shows audio bandwidth limiting, you could use a IR LED for a slightly lower voltage into the 20meg resistor (say 2*10meg). The good thing is it is stable with an LM386 because virtually no RF gets to that chip.
I tried it at about 0.5 nA (500 pA) into the base of Q2 by connecting R1 (20 meg) to a 1.5 V voltage source (a lit red LED). I increased R2 to 220k and then I had to use an emitter follower to impedance match to the audio amplifer I used.
I noticed audio bandwidth restriction beginning to occur due to the very low current at the collector of Q2.
I'll try it at 100pA in maybe this evening. The lower the base current you use the better it seems up until the audio bandwidth is to low to be usable. However I haven't quite reached that limit yet.
I am testing at higher frequencies at the moment, around 13 MHz. Given the physical mechanism of operation I suggested I have to check it still will work good in the AM band.
The lower the current you go the more sensitive I find it gets. I got a good boost in detection sensitivity by increasing R1 to 20 megs I don't want to put 10 10meg resistors in series to try 100megs for even lower currents, I will maybe drop the voltage at the top of R1 to say 1 volt instead. At some stage audio bandwidth will become a problem and audio output impedance as well. However I will try. It is interesting to see a BJT transistor apparently operating in a different mode to normal at such low currents.
https://sites.google.com/view/analogelectronics/home/higher-am-detector
Yeh, absurdly sensitive even up a 10-15 MHz, higher I don't know. You may even have problems in the AM band with a ferrite rod antenna where even 20 V RF has sometimes been reported. That would cause base emitter reverse voltage breakdown of Q2 and possibly make Q2 permanently noisy. Not to mention you maybe would hear multiple stations at very high volume all at the same time due to the lack of selectivity of TRF radios. I live in a very weak AM band zone so I can't really say.
Of course people tend to be enthusiastic about their own projects and that distorts their viewpoint.
Probably there will be some indepenent testing at some stage.
One thing is the DC input impedance of Q2 is very high (20+ megaohms). That would mean it is very difficult to get current in and out of the base except by substantial voltage changes. And since BJT transistors are current controlled devices that should be a problem.
Or another way of saying that is there should be a substantial impedance mismatch between the transistor base impedance (20meg+) and the tuned circuit parallel resonance impedance (25-50k).
One offset is that with a regen the tuned circuit parallel resonance impedance is boosted by loss cancelling positive feedback to say 1 or 2meg.
I think the main saving grace is the base emitter capacitance of Q2. That allows much smaller signal voltages to push charge into the base and pull charge out, to charge and discharge the base emitter capacitance. While at the same time controlling emitter collector current flow.
What is normally a problem with transistors (base emitter capacitance) becomes a help in this case.
You might find using a very high ft transistor at very low frequencies is not the best idea then since the base emitter capacitance might be too low for maximum sensitivity.
Once you get use to using this board it does work as well as the old radio board, it is just a question of adaping to it. You just need to post a few times to get comfortable with it.
From the video it looks like it works quite well. The circuitry reminds me of an MK414.
Michael Dranfield built a 3 transistor TRF from a 1974 Practical Wireless article.
Somehow I don't think it can get the most gain possible at least from Tr1 and Tr2.
Cool, I want ti give this a try. So far, all the trf circuits I try oscilate a lot.