Crystal Radios and Followups

[Swamp Thing]

Let me also dial back the hate on superregens. Yes, I called them "antisocial" and yes, they do have a tendency to radiate. But they don't have to. My grandparents had one from maybe the 20s or 30s. No interference at all.

I still would in general avoid it, as better alternatives likely exist, but it's not the radio equivalent of toxic waste.

"It's not a bad circuit. It's just misunderstood."

How about something like this? (See this post)

Edit: Purists might want to terminate the LNA output. Maybe even terminate the input and use a gimmick pigtail capacitor to the antenna.

 

[Baluncore]

How about something like this?

Broadband LNAs are class-A amplifiers, that typically require a supply of 20 mA. That is a lot of power, compared to a minimum radio receiver, that should run on a couple of milliamps.

Unless you have a good RF filter at the input, an LNA will amplify everything, and cross-modulate the signals you want. I believe they are still prohibited from being connected directly to an antenna in Germany, because they oscillate, and transmit that oscillation from the antenna.

 

[Swamp Thing]

I believe they are still prohibited from being connected directly to an antenna in Germany, because they oscillate, and transmit that oscillation from the antenna.

As I'm sure you know, superregenerative receivers at VHF are extremely sensitive considering their simplicity. Plus, the LNA adds 20 or 30 dB gain depending on the model.

Because of this, we can use very small valued gimmick capacitors at the input and output of the LNA and still get a usable overall sensitivity while avoiding intermodulation. Typically, the modules have a 1 dB compression point of around 20 dBm.


These LNA modules are MMIC based, using MMICs that are found in many professional and consumer devices. They are usually unconditionally stable (i.e. very immune to oscillating even with open circuited inputs and outputs. And if we terminate at least the output, the chances of them oscillating would be minuscule unless some horrible mistakes have been made in the layout and decoupling on the PCB.


Even if the LNA oscillates, the radiation would be at a level comparable to other electronic devices typically found in the home, since its port are very weakly coupled to the outside world. Also, MMIC based modules of this type are used by many, many people at the inputs of SDRs, and not necessarily with input filtering.

 

[Baluncore]

As I'm sure you know, superregenerative receivers at VHF are extremely sensitive considering their simplicity. Plus, the LNA adds 20 or 30 dB gain depending on the model.

Regenerative receivers have more than enough sensitivity. They can follow the noise floor without more gain from an LNA.

Because of this, we can use very small valued gimmick capacitors at the input and output of the LNA and still get a usable overall sensitivity while avoiding intermodulation. Typically, the modules have a 1 dB compression point of around 20 dBm.

Why burn 20 mA, just to waste it in poorly coupled terminals. Gimmick couplers are unpredictable, something a beginner does not need with an introductory project. The receiver can be isolated from the antenna with a 1 mA FET RF stage, without an MMIC and 20 mA.

They are usually unconditionally stable (i.e. very immune to oscillating even with open circuited inputs and outputs.

I disagree. For stability, they need to be terminated in resistive loads, usually with attenuator pads between stages. I have used MMICs in the first IF of broadband radar receivers, as the gain rises it takes more effort to tame the reactance and parasitic oscillations. I would never use such a wideband device if I did not need it for the signal. A regenerative receiver is inherently narrow band, and so is totally bandwidth mismatched to an MMIC. It is better to go with a tuned RF, FET front end.

Even if the LNA oscillates, the radiation would be at a level comparable to other electronic devices typically found in the home, since its port are very weakly coupled to the outside world.

Not so. When connected directly to an antenna, there is often a side frequency at which positive feedback encourages oscillation. That can be reradiated efficiently by the antenna. Much time has been invested, tracking down those nuisance oscillators.

 

[Vanadium 50]

How about something like this?

If I were launching on a project to build and understand radio receivers, maybe.

I would start with a crystal set, add an audio amplifier (which means a power supply) to drive an 8Ω set of headphones,then maybe a 2nd stage to drive a speaker, the maybe an RF amplifier, then maybe replace the crystal, and so on. This might be one piece.

To keep my sanity, I'd probably put these all on their own small breadboards (Radio Shack used to sell them in various sizes). Modularity is an important consideration in enginering (how this started) and while there is no reason that electrical modularity demands physical modularity, doing this makes evolving the design easier: this audio amp is no good - let's try again.

Because this discussion got me thinking harder about impedance, each part would likely have some sort of impedance matching up and down the chain. If not matching, at least defined specs.

And, so people can mock me, I'd probably go from board to board with shielded ribbon cable. "That's not the right stuff!" you say! No, but I have a lot of experience with it, and would do a good job - better than on alternatives where I don't know what I am doing.

 

[berkeman]

And, so people can mock me, I'd probably go from board to board with shielded ribbon cable.

At AM radio and audio frequencies, that should be fine. Just take care to choose your pinouts for the cables with crosstalk and loops in mind.

 

[Vanadium 50]

There are some very nice ribbon cables out there. You can get them with every 3d wire grounded, and you can get them as multiple twisted pairs in a common shield and jacket.

Getting them rated to 100 MHz is trivial, and 300 MHz if you spend some time with catalogs.

My point, is that sometimes the best choice is the one you will do the best with, and not the Theoretically Optimal Choice.

 

[tech99]

Here are some comments from my limited understanding. (I invite Baluncore and other experts to correct me if I go astray.)
1) Piezoelectric operation for the tan headphones seems right. See, e.g.
Amazon earphones
They have an impedance of a few thousand ohms. Why not 8 ohms, which is standard for HiFi/stereo speakers and headphones? Because of the need to match the load impedance to the crystal rectifier. The broadcast AM band is, for discussion purposes, pretty close to DC so we can look at the DC impedance of crystals, which is typically a few hundred ohms (see, for example, the center column in this table from Scaff and Ohl, Bell System Technical J., 1947).
View attachment 351422
These World War II microwave diodes literally used cat whiskers and semiconductor crystals inside of a cartridge.
Choosing the load impedance is a balance between best power transfer, which occurs with a matched load (500 ohms, say), and good selectivity (narrow bandwidth to capture just one station at a time) which requires high Q and therefore high impedance. I'm guessing that ~3000 ohms was empirically arrived at as a reasonable compromise.
EDIT: Another consideration is the need to forward bias the diode over at least part of the RF cycle in order to provide rectification. For a given power intercepted by the antenna, a higher circuit resistance will provide higher voltage across the diode. The forward drop sets the minimum voltage: 0.3V for germanium and 0.7V for silicon.

2) As a boy, I built a crystal set with enameled magnet wire wound on a toilet paper tube and held in place with airplane glue (toxic stuff, probably can't buy it anymore). The tube was nailed horizontally to a piece of wood and I used sandpaper to rub off the insulation along the top. A tin strip, nailed to the wood so it could pivot left to right and bent up to contact the coil, tuned the radio. I think this is easier than acquiring a ferrite antenna and variable capacitor, as you say.

3) I connected the RF input to a rain gutter downspout and the ground to a cold water pipe, and it worked! Gutters of old were steel with soldered connections, however, so this might not work so well with today's aluminum gutters and glued seams. Still, it's worth a try.

I think transistors are an obvious next step as the builder increases in ability to understand electronics (transistor action is not obvious...).

At the very small RF voltages being used the diode has high resistance. The highest possible headphone resistance is then the best. The piezo earpieces have infinite resistance so need a shunt resistor of say 100k in parallel. The traditional earphones for crystal sets were 4k; it is possible to use a transformer to allow lower impedance phones. In the UK, the AM broadcasting is being closed down now. Short wave broadcasting still seems to be used to some extent and crystal sets are easily made for short wave reception and need only a medium sized antenna.