Using a base-collector strapped transistor as AM detector, sharper knee for better weak signal demodulation
How to test a regular or SMD bipolar NPN or PNP transistor with a DVM in diode mode, to tell it apart from a dual common anode or cathode diodes inside a 3 legged SMD devide.
Another sensitive AM detector circuit that can do 100% modulation levels, high impedance input, low impedance output.
Using JFETs to assist AVC in transistorized AM radios, also widens the IF passband on strong stations, allowing higher audio fidelity.
A simple radio receiver signal strength indicator not using the AVC line.
Using a small signal silicon transistor with its collector and base tied together makes the collector turn on once the base is biased on, making a sharper knee. Making for a good weak signal AM radio station demodulator. Collector saturation doesn't seem to be an issue.
I kept the geranium (yes, this radio does use geranium, er, germanium signal diodes, I tried a silicon diode but strong stations distorted some) diode (D3) that came with the radio for the AVC circuit (R29 and a new 0.01uF cap to ground), and used a transistor with its base and collector tied together for the audio signal detector function. I split off the audio signal path (R30, C38) off the now AVC diode and put it on this transistor. Transistors like a 2N2222 or other silicon small signal types should work well. A sharper "diode" (base-emitter of the base-collector strapped transistor) knee will make for better reception of weak stations, if you add bias to overcome this "diode" (base-emitter) drop voltage. This is done with some forward bias current to improve weak signal detection, as this bias will help overcome the "diode" (base-emitter) drop loss of signal. A base-collector strapped transistor with a higher drop but sharper knee would be the better choice. Too many variables to calculate the resistor value to use (what current on the "diode" (base-emitter) knee will make it detect best, what's your supply voltage, what's the voltage of the "diode" (base-emitter) drop voltage, what's the resistive load on the "diode" (base-emitter) circuit such as the volume control, and so on), so trial and error would be quickest.

The emitter on such an NPN transistor would be the cathode, and the base tied to the collector is the anode Here I used a 2SC2785 salvaged from a VCR or such. If the transistor is a PNP, its emitter would be the anode, and its base tied to its collector the cathode. Try various transistors. I found that 680K seems to hit the sweet spot in this radio, but it's not a critical adjustment on a trimpot (which you can use to quickly find the value you'd want in a fixed resistor). I had thought that using a transistor this way was the mark of a sloppy company (did they run out of diodes?), but I got the best weak station detection performance with this. Oh, check strong stations too, which worked well too.

Turns out I needed a bit more sensitivity in the AVC detector diode, so I used a germanium transistor with its base and collector strapped together. I found that physically larger transistors have lower base-emitter drops (the 2N1304 has 0.14V with the base and collector tied together) than smaller germanium transistors. But be aware that most larger germanium transistors may not be up to switching fast enough to act as a detector. I used the 2N1304, its data sheet says it can do a few MHz. I now get less or no distortion on very strong stations, as the AVC now kicks in with less "delay" ("AVC delay" means the AVC needs a certain voltage amplitude before the AVC starts reducing the stage gain); it's not time delay.
The big transistor with heat shrink over it is the large germanium transistor. Its case is tied to its collector, and the heat shrink is to prevent shorts.

One another copy of this radio, I found that I really don't need a separate AVC diode. The AVC circuit produces enough bias on the new transistor-as-diode to get good demodulation on weak stations. Measure the diode drop voltage with a DVM with this feature, and find a transistor-as-diode with a similar drop to replace the diode with.

On yet another version of this radio, it turned out I needed a separate AVC detector, now "D3". The audio detecting transistor (this set used silicon) strapped as a diode "Dau" needed less bias current than the AVC circuit imposed on this diode. If the AVC circuit imposed less of the current the "diode" needs for its sweet spot for best audio detection, one could add a resistor to B+ to get the current on the diode into that sweet spot (the immediately above radio had this AVC current already on the sweet spot). But it's not so easy to reduce the AVC current, so I added a 2nd diode "D3" to provide the AVC action. Oh, you need bias current on the audio detector diode "Dau", else you may not hear anything. 390K to +5V put the new diode in its sweet spot. The weak AM reception is now better, and this set lives up to its badge as a "Musaphonic" radio.

Some simulations of a silicon diode, a base-collector strapped silicon transistor,

and such a transistor with a bias current.

I'm not sure how much, if any, knee can be below 0V here. Conduction on the other side of 0V would probably mess up the detection. IIRC, the FCC requires that 10% be the minimium amplitude of an AM station's carrier, which should permit some knee curvature of stations that aren't too weak (else signal to noise would make them unlistenable anyway).

Above is an AM radio, a GE C1432, that was designed and built using a base-collector strapped transistor as the detector. The AVC circuit provides the bias to overcome the diode drop. Stronger stations will produce AVC voltage which will change the bias voltage, but this change won't much matter as the stringer station will overcome any mismatch of the bias vs the diode drop voltage.

AM detector that does 100% AM modulation.

Source web page. "AM detector more sensitive than simple diode". EDN September 28, 2016 by Lyle Williams
I modified this circuit a little, mostly changing the diode to a diode strapped transistor, for a sharper detection knee. I also kept the old detector in place in my radio to take care of AVC (but moved the volume control to the output of this new circuit (and used a resistor to substitute for the volume control resistance in the AVC circuit)).

Below left my modified schematic. Q1 is essentially a current amplifier, turning on at the peaks of the AM modulated signal, to charge up C2 and R2 (filter, time constant circuit). This briefly turns Q1 on at the AM signal peaks, charging the filter cap C2. And R2 bleeds this charge off between times. Q1 also keeps a high impedance load on the detector, like an emitter follower at the AM peaks. I found that putting C2 right at Q1 avoids regenerative heterodyne whistles, as it keeps the RF circulating currents in a small area, reducing coupling into the AM antenna. The RC filter time constant essentially includes the B+ supply bypass cap to get to ground. This detector operates on the positive peaks, while the AVC circuit operates on the negative peaks.

The AM signal is a 455KHz carrier with 1kHz modulation at 100%. The input impedance of the detector is around 300kO, which doesnít appreciably load the source. The transistor Q2 strapped as a diode is operated with a slight forward bias so a regular silicon transistor can be used. The output impedance of the circuit is low and the load here has little effect on the distortion level.

The simulation input and output patterns are shown above right . The output audio waveform is shown with the correct DC level with respect to the center line which is 0 volts. Note that the output voltage does not go to zero even with 100% modulation. There is no visible distortion in the output and that is unusual with 100% modulation.

Using a JFET as a voltage controlled resistor for variable IF bandwidth, and to act like an AVC circuit in a transistor AM radio. This circuit widens the bandwidth on strong stations (adjacent stations are so weak you wont hear them), for more audio high frequencies, better fidelity. Weak stations will have narrower passbands, avoiding "monkey chatter" from immediately adjacent weak stations. I live 6 miles north of a 50,000 watt flamethrouer, WABC 770KHz. They briefly played music to celebrate their past. This mod helped make it sound good, almost hifi.

Here the usual AVC circuit in the AM section of this radio gets an assist from a JFET operating as a voltage controlled resistor. This is in the triode or ohmic region (at and near 0V and 0mA current, the blue region in the figure below) where the FET can operate as a voltage-controlled resistor. The triode or ohmic region in an FET is sometimes known as the linear region. The JFET operating as a voltage-controlled resistor works in this region. If we extend the VDS voltage range to include slightly negative voltages for a particular gate-t-source voltage, we see that the there is still a resistive effect. Preferably, there is no DC voltage across the JFETís drain and source terminals in the voltage controlled resistance mode. This JFET's source and drain is connected across the entire LC circuit of the first IF transformer, which has no significant DC across it. The source and the drain both have IF frequency energy on them, out of phase with each other. The LC circuit has a tap near the center of the winding, tied to an RF ground (which is why the drain and source has out of phase IF energy) The gate is roughly at the midpoint of the resistance, which should reduce the effect of the gate capacitance on the IF.

When the gate voltage VGS here is dropped below around -4V, the JFET (a J211) becomes near infinite resistance. At around -2.7V, the JFET becomes a resistance around 1500 ohms. This variable resistance across the IF transformer LC circuit attenuates the signal strength, and also as a side benefit, will make that LC circuit have wider bandwidth. Which will allow higher audio frequencies from the strong station to be heard, higher fidelity (also better fidelity because the attenuation keeps the strong station from overdriving and distorting in the rest of the IF strip). Weaker stations will have narrower passbands, avoiding "monkey chatter" from immediately adjacent weak stations.

The controlling voltage is acquired from the AVC line via an inverting transistor (a 2N2222). This transistor level shifts and amplifies the AVC voltage range. Weak signals create a higher AVC voltage, and the transistor goes into more conductance, drawing, via the collector resistor, its collector voltage more negative. Strong signals will produce lower AVC voltages, and the transistor goes nearly into cutoff, and the collector approaches the positive supply rail. Be sure to use a bypass cap between this collector and the B+ line, to avoid ripple and such from "modulating" the voltage controlled resistances of the JFETs. Which would distort the detected audio. You need to size this cap to smooth out ripple, but still allow the circuit to respond to the AVC changing due to tuning in stations. I used a trimpot to get a sweet spot of an emitter resistor value, setting the gain of this transistor (a value that has the JFET do nothing on weak stations, but attenuates strong stations to minimize distortion), then I used a fixed resistor of this value in the finished circuit. Not a great way to do circuit design, as variations in the transistors and JFETs could throw this off in production. The IF transformer's LC circuit is also at the positive supply rail. From the JFET's point of view, (its gate is connected to the transistor collector) its gate going more negative makes the JFET resistance go higher. The gate getting less negative makes it lower resistance, attenuating the strong station's signal. As the AVC circuit is a feedback loop, the JFET's resistance will settle to a value after this loop does its action.

A second JFET on the IF transformer between the first and second IF stages is also installed. You don't want to put it at the last IF transformer, as all you'd do is attenuate distorted signal. A third JFET is also used on the antenna LC circuit (I removed a 1.5pF cap there, as the JFET has some internal capacitance, a few pF's, inside it). Then I retweaked the antenna trimmer cap.

Above right shows an exaggerated relation of passbands (how wide some value of X dB down) to radio station signal strngths. The passbands really vary by a factor of two at most. Attenuation is related to AVC action, in that weak stations are amplified more than strong stations in regular AVC circuits, and here there is actual attenuation being done as well (which via the voltage controlled resistive JFETs changes the passband widths of the various IF transformers and the antenna LC circuit).

Another radio, this one AM:

Here is a single JFET connected across the 2nd IF transformer T2. The IF amp transistor TR2 also acts like the 2N2222 in the above AM/FM radios (as TR2's base gets biased by the AVC line), and the JFET control signal is picked off a bypassed resistor R10, was 1K, now 1.5K (to increase the control voltage range). The full LC circuit of T2 is bypassed by C18 and biased by voltage divider R6 and R7. There was a diode between T2 and R10, but I removed it before I added the JFET. The voltage divider was reworked (by changing R6 to 180 ohms) to have the JFET go almost into infinite resistance when the radio is tuned to an empty spot on the dial. And to drop resistance value on strong signals. This then widens the IF bandwidth and reduces the IF strip's gain.

Another AM radio, an RCA RZA202B, with the JFET mod. One JFET across the antenna LC circuit L1, and another JFET across the first IF LC circuit T1. Driven by a similar buffer transistor that uses the AVC line voltage and converts to to the JFETs' gate control.

If you are okay with infecting a valve radio with solid state parts, then:

The above mods on solid state transistor radios, using JFETs as variable resistors across IF transformers. Idea is to use the AVC voltage to increase or decrease the JFET resistance to assist the AVC function, and to make received audio bandwidth on strong local stations wider. And perceived audio signal levels are more even while tuning across the AM band. Someone asked about doing this on tube sets. I put the JFET on the secondary of the 1st transformer. This secondary is riding on the AVC voltage line. For the JFET I used (J211), to have infinite resistance, its gate should be around -4V compared to its source and drain. About -2.7V gets about 1.5K resistance. Here, the source and drain (connected to the IF transformer secondary) are riding the AVC voltage, and I'd need a fixed negative voltage to connect the gate to. I used a half megohm trimpot on 12VAC (tapped the heater string) then rectified that and filtered it to get an adjustable negative bias for the gate. RF Bypass this bias to ground. I and added another JFET to the antenna circuit. JFETs add about 2pF of capacitance so you'll have to tweak the trimmers.

This radio is the classic All American 5ive hot chassis radio. This will work for transformer powered sets too.

Simple radio receiver signal strength indicator, not using the AVC line

This circuit samples the voltage of the IF signal on a collector of a transistor in the IF amp. And is independent of the AVC circuit. The diode D2 "detects" the amplitude of the IF signal (the diode pointed so that more IF amplitude makes the bias on Q1 increase, thus making the LED light brighter), either 455KHz or 10.7MHz in an AM/FM set I put this into. Resistors R2 and R3 were a single 470K trimpot I used to adjust the bias on Q1 (the wiper connects to the base) to get the LED on Q1's collector to light or dim as I tune across the AM or the FM dial. Measured the trimpot, and then used 39K for R3 and 560K for R2. Your mileage will vary, depending on the diode you use for IF level detector D2 and transistor for Q1. R1's resistance was chosen to be a negligible load on the signal getting amplified by the IF stages of the radio. That then drives the selection of R2 and R3's resistance. The LED is a modern high efficiency one that doesn't require much current to light up bright. It came from a recently made Xmas light string. Thus Q1 doesn't need much biasing from the "detector" circuit to delver enough current to light the LED.

I wanted to avoid loading the AVC circuit itself. This above circuit shows what the AVC has already done to the signal. Loading the AVC circuit directly would tend to make the AVC circuit less effective in controlling the IF gain.

The FM section of the radio I put this in doesn't really have AVC, as you want the FM signal to get clipped before the discriminator detector anyway. I can see a station with full quieting still be intermediate signal strength, and stronger stations are of course full quieting. This above circuit shows the varying levels of station strength beyond full quieting (useful for antenna aiming). I tapped the 2nd to last IF stage (TR7's collector in the schematic in the JFET section of this page), where clipping hasn't fully kicked in yet, but full clipping does happen in the final IF stage.

How to test a bipolar regular or SMD transistor with a DVM in diode mode, to tell it apart from a dual common anode or cathode diodes inside a 3 legged SMD devide.

You know the old loose bipolar NPN or PNP transistor test method of checking the PN junctions (with your DVM in diode mode), like from the base to the emitter, and then base to collector. You can figure out if it's an NPN or a PNP transistor (compare to a diode). And additionally, you can see some transistor activity by doing this following test: First, having one test lead on the emitter and the other on the base (but not the collector), in the direction you see a diode drop voltage, typically you'd see about 0.68V on a silicon transistor. Now while keeping the same test lead on the emitter, touch at the same time with the other lead both the base and the collector. You should see a lower voltage like 0.59V (around 85% of just the base-emitter only value). This test can also tell you, with surface mount devices, if you really have a transistor, or a device with two diodes, one with a common anode, or a common cathode on what would look like the base pin.