High-Performance Regenerative Receivers

Discussion in 'Homebrew and Kit Projects' started by G3EDM, Sep 27, 2021.

ad: L-HROutlet
ad: l-rl
ad: Left-2
ad: L-MFJ
ad: L-Geochron
ad: Left-3
ad: abrind-2
ad: l-BCInc
  1. W9BRD

    W9BRD Ham Member QRZ Page

    Another aspect I'm thinking about is the method of coupling from the first AF amp to the push-pull 1AG4s. The primary impedance of the interstage transformer assumes that it's driven from a "10 k" plate, meaning a triode. Screen-grid amplifiers have a much higher output Z, and so in coupling your screen-grid first-AF stage to the push-pull output stage via that transformer, you're beginning with mismatch loss to the transformer from the 1AD4 screen-grid-AF plate. (In more-usual circuits, a triode driver would be used; or RC coupling would be used from the pentode-AF-amp plate to a single-ended AF power tube. In this case, though, rewiring your first-AF 1AD4 as a triode to better drive the transformer may give more gain, the same gain, or even less gain, "depending." And of course going to a single-ended output stage would halve the possible available output from that stage.)

    That said, the 1AG4, with its specified output of only a few tens of mW, was likely intended only to drive headphones. There are subminiature tube capable of hundreds of mW output, but headphones output is better for two-way work.
    G3EDM likes this.
  2. G3EDM

    G3EDM Ham Member QRZ Page

    Here's the latest schematic. Changes:
    • Removes the muting relay (this is now placed at the front of the 2nd AF).
    • Reflects that the output from the 1st AF is via a shielded wire (to try to reduce key clicks on sidetone).

    73 de Martin, G3EDM
    W7UUU likes this.
  3. G3EDM

    G3EDM Ham Member QRZ Page

    How's it going? :confused:

    73 de Martin, G3EDM
  4. W9BRD

    W9BRD Ham Member QRZ Page

    Getting parts together; have been traveling on weekends, so building to start soon.
    G3EDM likes this.
  5. G3EDM

    G3EDM Ham Member QRZ Page

    Interesting. I may try rewiring the stage as a triode, should be able to reverse back to a pentode if it doesn't improve output.

    Of course the reason it is not a triode is because originally, this was the final tube in the set, directly driving high-impedance headphones. When I added the push-pull final I should have paid more attention to the 1AD4's pentode operation and whether it was suitable to drive the final.

    Having said that, if I have discovered anything in building this receiver, it is that the detector and the first AF are inter-related and changing things in the AF can have strange results, especially as concerns parasitic oscillations.

    I never use the speaker except for one thing: "Showing off" when visitors ask about ham radio. (Usually just "polite interest" and without decent speaker volume, there's not much to demonstrate.)

    I've just been investigating the more powerful subminiature power pentodes. There are at least a couple that can give much higher output:
    • 5638: 0.6W plate dissipation with 100V on the plate, indirectly heated, filament is 6.3V at 150mA.
    • 5902 (or the similar 5640): 4.0W plate dissipation with 110V on the plate, indirectly heated, filament is 6.3V at 450mA.
    The "sweet spot" if going that route could be a pair of 5638's in push-pull. The B+ can come from a voltage doubler on the existing 52V B+ supply. However: yet another voltage to supply (for the filaments) and quite hungry. It could be five rechargeable 1.2V D-cells (the spec says 6.3V plus or minus 10% so it should function all the way down to about 5.7V). Or, probably better, just add another four D-cells in series with the existing filament supply (which is two D-cells in parallel).

    Another choice would be the 3V4 (or the almost identical 3Q4), which has a plate dissipation of 250mW. This is directly heated. The filament is center-tapped, and you could run the two filaments sections in parallel, in which case they are rated to function as low as 1.1V so could use the existing 1.2V heater supply and the filaments only take 100mA when connected in parallel. In typical applications the B+ is 90V, which could come from a voltage doubler. Putting two of these in push-pull should multiply the existing power output by a factor of at least 7: surely enough. Could pretty much drop these in to the existing circuit, fed with the higher B+.

    Of course the 3V4 is not a subminiature tube, but so what? Just need to add a couple of tube sockets.

    OTOH I came across this tube by accident, while searching for subminiature tubes. It was designed for use in AC/DC sets and there must be plenty of other 1.2-volt, directly heated candidates if you are willing to extend the spec to include standard miniature tubes.

    Edited to add: I've just realised that I don't know how to build a voltage doubler when the input voltage is already DC. Looks like it is time to learn about something called the "switched capacitor" circuit....

    73 de Martin, G3EDM
    Last edited: Oct 17, 2021
  6. G3EDM

    G3EDM Ham Member QRZ Page

    @W9BRD: After a week or two of testing the option of an instant A/B switch between a resistive (180K) detector load and an inductive (60H) one I have decided that the option is quite useful.

    Specifically, the resistive load can be a bit better at pulling in weak signals when there is a high level of QRN. I think that's because the 60H choke tends to strengthen the regenerative "whoosh" and that gets added to the QRN. When you switch in the resistive load instead, there is a noticeable drop in QRN/whoosh, and sometimes that is enough to make the weak signal pop out of the noise.

    However when QRN levels are low, the inductive load is noticeably better than the resistor: the resistor has a "cotton wool" effect.

    Other differences: There is noticeably more audio distortion/overload/pulling of the signal with the inductive load, requiring firmer control of RF gain, regeneration level, and regeneration throttle. It is a "wilder" environment but most of the time, more sensitive, unless there is high QRN.

    On another subject, as you know, I use a tunable RF stage in front of the detector. I get best results by tuning slightly below the monitored frequency although that does increase noise somewhat. With very strong signals, I tune way off the frequency, to add attenuation.

    73 de Martin, G3EDM
    Last edited: Oct 20, 2021
  7. N2DTS

    N2DTS Ham Member QRZ Page

    I have had much better luck building superhets that work then I did regens.
    My first regen was a 6SN7 in a design out of the encylopiedia we had 'the book of Knowledge' 1964 I think.
    I built it about 10 times and it worked once.
    That one time it worked quite well for the parts count.
  8. K6BSU

    K6BSU Ham Member QRZ Page

    I'd like to see what a 60 Henry inductor looks like. I can't imagine.
  9. KP4SX

    KP4SX Premium Subscriber QRZ Page

    Not what you are thinking :) They can be quite small. The Paraset used an inductor like that as did at least some of the LM freq meters. Only rated at a few ma, though.
  10. G3EDM

    G3EDM Ham Member QRZ Page

    It's a Hammond smoothing choke, similar in size to what you'd see in a B+ high-voltage power supply. Actually, this radio uses two of them. They are 60H and rated for 8 mA.

    Here's the detector load choke. Above it, the capacitor used as "regeneration throttle". Below it, you will see two resistors (220K and 1M) wired in parallel to give 180K ohms. There is a switch on the front panel that selects whether the choke, or the resistance, is used as the plate load. The black object at the right is the detector tube, a 1AD4 sub-miniature, enveloped with a makeshift shield (a ferrite clip normally used on power cords to reduce RFI) and fastened firmly with a plastic wire clip to a wooden support to increase stability. Sorry the photo is blurred, the lighting in the shack is not great.


    Here's the other choke, used as the plate load in the First AF stage and mounted almost 90 degrees rotated in reference to the detector load, in an effort to prevent mutual coupling and therefore parasitic oscillation (this seems to be effective, after earlier problems with "howling"). The detector choke is at the extreme back of the set, the AF choke is not far from the front panel (but not too close to it).


    Yes, it's a big mess visually. It happened because this was originally a "breadboard experiment" that got way out of hand. In theory I could take the whole thing apart and rebuild it to proper "RF construction" standards with shorter signal paths and short leads on the components, but at this point it may not be worth it given that I don't plan to use this set as my main receiver for very long.

    Physical arrangement of the Detector and First AF are crucial in a regenerative radio. I may in fact have made my set a little too small (not deep enough front to back), although with careful arrangement of the parts; shielding of the tubes; and attention to grounding of metal parts, the problem with oscillations has gone away. It's not just the oscillations. The detector will function better (and be immune to problems such as hand capacity effects) if it is built quite a long way behind the (grounded) front panel.

    Experiments with shielding (using a cardboard "cabinet" lined with grounded kitchen aluminium foil) demonstrated that shielding the detector (surrounding the whole set with a grounded shield) actually worsens the performance of the set considerably. So if I ever build a cabinet, it will probably be wooden, with grounded aluminium plates mounted on most of the inside but avoiding the detector area.

    There is quite a lot of stuff about physical arrangement of the stages in a regenerative radio in the articles by Charles Kitchin, N1TEV, from the 1990s. His regenerative projects were solid-state but the general principles apply to any regenerative set.

    73 de Martin, G3EDM
    Last edited: Oct 22, 2021

Share This Page