September 13, 2009
Nearing completion, tuning cap tacked in place.
The Final Phase
Builders of tube-type regenerative receivers favor low-amplification tubes for better regeneration control. This receiver started out using J310 FETs for the rf amplifier, regenerative stage and detector. I wondered if a lower-gain FET, such as one designed for audio applications, might work better. The 2N5639 switching FET provided the best results in Q102. I had only J201 audio FETs, and they wouldn't oscillate at the low end of either tuning range. J202s might work, and I'm pretty sure J203s would work. But the switcher works very well. Having obtained some J201s I used one for Q103, though any N-channel FET should work here. The circuit wanted a little more source resistance for Q103. You want to set R102 as low as possible for minimum noise. Adjust it at the low end of the lower band segment, so the receiver just breaks into oscillation with the Regeneration pot in mid-range.
Having to add source resistance to Q103 makes me believe a hotter FET would work better there, and might even be quieter. A J310 may work better in that position, or a hotter audio FET, but I haven't tried it yet.
Regeneration is much smoother with the 2N5639 than it was with a J310, and probably smooth enough to remove R101 from the high-band tank circuit. Selectivity and gain are not noticeably affected with R101 in place, so I left it there. On any frequency on either band, I can separate the sidebands of my signal generator modulated at 400 Hz, hearing two distinct peaks. CW operation requires very tight selectivity, AM-carrier reception is more forgiving.
Skirt selectivity is still very good. I have a strong local station at 1450 kHz, only a couple of miles away, but I can listen on 1440 and 1460 kHz without desensing or splatter.
Every time I was almost ready to publish this page I got an idea for an improvement. One came from comparing apparent sensitivity on the low and high bands. Could the rf chokes be too lossy? It was easy enough to wind a new coil on an FT50-61 core. I made this one 63 uH because that's where it happened to end up, and that changed the frequency coverage slightly. The low band now tunes about 494-900 kHz. Maybe I'll hear one of the experimental 500-kHz stations. The new coil has higher Q. Where the chokes gave smooth regeneration entry and exit, it was now a bit jumpy above 800 kHz. It was usable, but another 150k-ohm resistor gave just enough swamping. The Regeneration control had to be advanced slightly, but the transition is much smoother. So R101 moved over to swamp both inductors.
There is a 140-kHz overlap between the two bands. I may eventually take a few turns off the high-band coil, which should extend that band into the 160-Meter Amateur band.
T101 on the rf board started out as a step-up transformer to match the low-impedance output of my loop preamplifier to the receiver's high-impedance input. Right now it has the same number of turns on primary and secondary, 20 turns wound bifilar on an FT50-61 core. If you are using a random-length wire (not a "long wire," which would be several hundred feet long on the broadcast band) you can leave it out and connect the antenna directly to the 300-pF (value not very critical) cap. I didn't include an rf-gain control and haven't missed it.
Test equipment will tell you a great deal about how a circuit performs, but there is no test better than actually using it. The last 10 days or so of development on this project took so long because I didn't want to introduce too many changes between on-air sessions. And anyone who DXes the AM band knows the night time is the right time. Every night it was almost like sitting down with a new radio.
Most articles about regenerative receivers point proudly to the extra manipulation required to use them. Since when did lack of convenience become a feature? Yes, the Regeneration control adds to the operation, and may even add some fun as well. But there is no reason you should have to readjust regeneration every time you change frequency. If you do, then when you tune lower in frequency, gain drops precipitously. Just when you are homing in on a signal, it disappears. You have to readjust regeneration for a heterodyne and come at it again. Meanwhile, the signal is fading up and down. About the time you catch up, it fades into inaudibility. This receiver can be tuned more than 100 kHz without having to readjust regeneration except for very weak signals. Then, it will let you zero in without detuning. It still requires a bit more skill than a superhet, but most of the time you won't know you're using a regen.
As of September 13, 2009, in three weeks of use, the receiver has logged 15 states, 5 countries and 2 continents, with an indoor amplified ferrite-loop antenna. I've since switched to a large air-core loop.
Update September 25, 2009: The receiver has now logged more than 130 stations in 20 states. Best DX is KSL, Salt Lake City, Utah, 2030 miles (3330 km). The receiver shows no anomalies or annoying quirks; no "fringe howl," microphonics or other issues. This past week I switched to a 14-foot (4.25-m)-circumference untuned air-core loop, and dispensed with the dual-2N5109 preamplifier.
Final Thoughts
Aside from the tuning capacitor and vernier drive, which came from hamfest flea markets, every part used in this receiver is available off the shelf from major online retailers.
There is a school of radio building that believes in using as few parts as possible. You shouldn't use parts you don't need, but you shouldn't leave out parts you need, either. I'm not nostalgic about regens; I never had one as a youngster. So I'm not drawn to circuits that duplicate the radios of my youth. I'm drawn to radios that work well. I like regens because, although easy to build, they work surprisingly well. With a little work they can be made even better. The extra rf chokes and bypass capacitors shown at various places on the schematic may not be required; I put them in during construction to be sure the rf and audio signals kept to their respective sides. Maybe adding an audio filter is a step away from the regenerative-receiver spirit, but it sure helps me hear the weaker signals. If your listening interests run to local stations only, you probably don't need the filter.
Update October 12, 2009: Used the receiver in the Fall Crystal Set contest. Fall 2009 Crystal Set Contest Log.
Update November 9, 2009: Log of all stations logged through November 8, 2009.
Notes
1. H.A. Robinson, "Regenerative Detectors," QST, February 1933, p. 26.
2. James Millen and Robert S. Kruse, "An Analysis of A.C. Operated Short-Wave Receiver Design," 1930 (James Millen Society).