Regenerative Receiver -- Advice Sought

(1/6) > >>

MARTIN MARRIS:
I'm posting this as a separate topic rather than rolling it into the receiver "build" thread. If forum denizens feel the quantity of my posts is becoming excessive, speak up!

I have spent the past several days doing further assessment of the "Three-Transistor Radio for the Beginner" (ARRL 1968). The receiver is really "hot" on 40m, where I am hearing hundreds of signals, with surprisingly good selectivity and stability. This is a lot of fun and very good for improving CW copying skills. On 80m, the performance is so-so but perhaps this has something to do with current band conditions??

The real problem though is on 20m, which is a band I very much want to monitor. Sensitivity on that band is fine: I hear plenty of signals, both CW and phone. The problem is inability to set the regeneration for proper reception. For instance on CW, all I hear is bursts of carrier "white noise." If I advance the regeneration control further, it breaks into howling oscillation.

The receiver is otherwise working fine in that HF neighborhood. For instance if I shift the bandset capacitor slightly to go down a few hundred KHz below the 20m band, I hear plenty of international broadcasters in AM, with good audio quality.

The author of the original ARRL article warns about this issue, says that it is caused by overloading, and suggests solving it by adjusting the antenna trimmer capacitor to reduce antenna coupling. So that's the first thing I tried; I opened up the trimmer all the way to minimum capacitance but it made no difference. The next experiment was to reduce the capacitance even further by adding other capacitors in series. This made no difference. Finally, I tried additionally attenuating the signal by adding a 10 ohm resistor across the antenna terminals. This made a difference, but a very small one. I managed to get one or two CW signals to sing a tone, but it was extremely rough: it basically sounded as if each pulse of RF was triggering the "overoscillation" howl, rather than being any kind of controlled reaction. As for SSB phone signals, they are distorted to the point of unintelligibility, regardless of any of the above configurations.

My next step was to do a lot of reading about regenerative sets, culminating in an article in QEX (Nov/Dec 1998) by Charles Kitchin, N1TEV, who seems to be the 21st Century guru on regenerative sets. But this still didn't advance my knowledge as to what might be happening with my particular issue.

So, ladies and gentlement, what do you think I should try next? The schematic is here:

http://tinyurl.com/obskp2z

As Kitchin says, "Although a regenerative circuit may look very simple, how it actually functions is not simple. Its operation is both complex and quite fascinating."

73 de Martin, KB1WSY

:
You are embarking on a road that many have traveled: attempting to optimize a regenerative receiver. Be warned: it is an addictive road!

The circuit you posted is interesting. Though the article claims it is a Colpitts oscillator, I would say it is actually a Vackar oscillator, because neither end of the inductor is grounded (the inductor "floats" above two capacitances). This is a somewhat unusual topology for a regenerative receiver.

There are a couple of things which may be problematic with the existing design. First, the regeneration is controlled by altering the oscillator operating point via the source bias. Second, the audio is taken off of the drain, which causes drain voltage to vary with the demodulated audio. Both of these can lead to an unstable operating point and a touchy regeneration control and frequency drift.

Your overload problem could normally be solved by reducing the RF gain, but the simple circuit you have does not include an adjustable RF stage, so it appears that even with your gain reducing techniques you still need to reduce the gain further to prevent oscillator pulling/blocking. A grounded-gate JFET input stage with adjustable antenna attenuation would help here.

To reduce frequency pulling it is advisable to use a transistor with low internal capacitances and also to make the tank C as high as possible so that any drift in the parasitic capacitances will be swamped out by the high tank C.

As for what to do next, I would suggest trying to build one of Charles Kitchin's circuits, like this one: http://www.electronics-tutorials.com/receivers/regen-radio-receiver.htm . I've built that one (3 times, in fact) and it works well, if you closely pay attention to all of the advice that Kitchin gives. His circuits generally include optimizations of the small details that are important for a serious regenerative receiver - details like coil placement, lead length, operating point stabilization, and RF attenuation.

If you want to continue working with your existing circuit, then adding a grounded-gate or grounded-base RF stage would be a good start. Also - and this is just an idea for experimentation - I would try removing C9 from the circuit. That might conceivably help with the howling oscillation (audio-frequency squegging) that you are experiencing.

MARTIN MARRIS:
Thank you, I will try your suggestion regarding C9 and report back.

I'd already thought of adding an RF stage, which is often recommended anyway in vintage homebrew projects for bands starting at 20m and above. How about the one-tube preselector from the 1963 ARRL Handbook -- which would have the advantage of being separate from the receiver so I can use it on other receivers later? I already seem to have most of the parts for that. However, the preselector itself is regenerative and perhaps having two regen units coupled together is asking from trouble (?). The ARRL instructions for the preselector state that if the receiver is overloaded by strong signals, this can be dealt with by turning down the preselector's regeneration control. Come to think of it, if you turn it down enough -- so that there is no regeneration at all -- that might lower the gain dramatically enough to serve as a workable attenuator. Is this a crazy idea?

Your suggestion of building one of Kitchin's (solid-state?) designs is good too but for the time being I'm a tube man. This 3-transistor set was really a one-off departure, mainly for nostalgia's sake, using discrete components with no ICs or PCBs, and partly also because it was rather easier to construct than a tube project (no power supply to build). I'd be happy to tinker with the existing set though, now that it's been built!!

In general I am really impressed with the receiver's performance -- except when it doesn't work! But that's the joy of homebrewing I guess. I initially had low expectations of ever being able to use it for QSOs but I am now confident that it will perform OK under the right band conditions, especially on 40m.

73 de Martin, KB1WSY

:
Quote from: KB1WSY on June 20, 2013, 07:39:12 AM

However, the preselector itself is regenerative and perhaps having two regen units coupled together is asking from trouble (?).

It certainly will be tricky to get 2 cascaded regenerative stages working; you have two high-Q stages tuned to the same frequency and fairly tightly coupled. This could lead to permanent oscillation of both stages and/or an over-critically coupled dual-hump frequency response.

Another easy thing you could try is to reduce the number of turns on L1, the antenna coupling link. You can also move the link farther away from the oscillator coil to further reduce the antenna coupling. That should allow you to reduce the antenna input as much as needed. If you really wanted to get fancy, you could consider building a variocoupler to allow you to physically swing the antenna link coupling closer to or farther from the oscillator coil.

Peter Chadwick:
Firstly, the Vackar is only a variant of the Colpitts - just as the Clapp is. But what are the values of  C2 and C3 on each band? It sounds from your description as if it is 'squegging' when the regeneration is advanced too far: in 'tickler' regenerative detectors, I remember that the advice is to use very tight coupling and a small tickler. You might find reducing the 1 Megohm gate to ground resistor helps  try 470k, 220k, 100k.

Navigation

[0] Message Index

[#] Next page