The CBCSRK-1T Circuit Description For The Electronics Engineers

Why do we need yet another single transistor AM radio design?  "I'm sick to death of them," I hear you cry!  They're all rubbish anyway, are highly variable, dependant on weird properties of this and that, use components that were last seen when dinosaurs still roamed the Earth, and are all highly annoying in some way or other.

Bear with me on this one, and don't write into Barry Took to complain until you've actually tried it.

PSPICE simulations do not qualify as "trying it," as PSPICE would probably explode in a hissy fit of recursive matrices if you fed this into it, or you'd spend more time making the models sufficiently accurate than it actually took to build it.

The "simple" explanation provided on the main page is pretty much it. The aerial winding part of L1 pins 2 to 1 is straightforward. Depending on how capacitive the aerial system is, moving the coil away from the main resonant tank coil can improve selectivity, as can reducing the number of turns. The arrangement shown is a good compromise most of the time.  L1 pins 6 to 5 and VC1A comprise the resonant tuning circuit. Both sections of the variable capacitor are used with the capacitor trimmers set to minimum. You can tune most of MW this way, and get the rest by moving the coil up and down on the rod.  It's a, "Mr. Slidey." The resonant tank is RF grounded at the bottom end through C2 and C3. Then it gets kinda interesting.

Q1 is wired as an emitter follower. The idea is that the current gain in emitter follower mode allows direct connection to the top of the tank without damping it too much, so you get the maximum possible voltage gain from the resonance of L1, despite getting non at all from the emitter follower...  Yet.  A small amount of initial base bias is fed through the bottom of the tank from R1, avoiding using a DC blocking capacitor or damping the tank with the biassing resistor. You'll see this a lot in old transistor radios. In emitter follower mode, even a BC549C running at low voltage can provide RF current gain at MW frequencies, so amplified RF current comes out of the emitter. It goes through the regen coil L1 pins 4 and 3 and is induced into the L1 system. From this we have some RF positive feedback gain for regeneration, without actually making any voltage gain with the transistor. Collector-base Miller effect is non-existant.

At the top of L2, the positive current flow on RF peaks builds up a voltage. Q1 can only flow in the positive direction, so it's acting as an active demodulation rectifier after the regen coil.  You see that voltage across L2 primary and the parallel capacitor. L2 is wired as a step-up transformer, and the signal voltage gain is about 10dB. We have used the transistor in current amplification mode and used the existing L2 transformer to get some audio signal voltage gain.

C3 eh?  In some other simple designs, you'll see a single transistor used as a voltage amplifier, and a horrible old Clacton Coils / Mullard transistor type radio frequency choke last seen in 1978 used to block the RF, and feed audio back to the base, where it is indeed voltage amplified again by the same transistor doing the RF amplification. You can see the RF riding on top of the audio in such designs. Those particular RFCs are unobtainable, even if you substitute a not very effective 4.7mH SMCC ferrite inductor. A more complicated reflex feedback arrangement would use more parts than another proper transistor stage, so would be pointless. Here, C3 couples the audio signal from the top of L2 primary back through L1 pins 5 and 6 in a reflex bootstrap fashion. It effectively provides more base bias for the RF peaks and the audio signal.  Q1 acting in this bootstrap mode also effectively multiplies the values of the capacitors. 100nF is more than sufficient. You can't see much of the RF voltage riding the audio in this design, because it's not there. It's a current.

You may initially feel aggrieved about transistor Hfe and voltage supply variability with the very basic bias arrangement.  Fine.  Because the bias is provided mostly by the bootstrap, once Q1 is tickled into conduction by R1, adjustment of the amount of regeneration effectively removes the effect of Hfe variability. The regeneration adjustment is surprisingly stable even when tuning across the band. The loudness limit obtainable ends up being more dependent on the point where the circuit breaks into oscillation than the Hfe of a particular transistor. Supply voltage variability effecting the gain of the receiver is reduced in a similar way. The initial quiescent current does vary with supply voltage, but the Re of the transistor and the small DC resistance of L2 helps to make this less critical for supply voltages between 1 and 3 Volts.  Anyway, you wouldn't connect a 12V car radio to a 24V lorry battery and expect it to survive the experience.

Oscilloscope Pictures

The RF voltage on Q1 base is on the upper trace and the demodulated audio at Q1 emitter on the lower. You can just see a bit of RF voltage on the lower trace. The upper upper trace is on a X1 probe, the lower trace on a X10 probe, with the attenuator settings as shown in the photograph. The signal generator is an ancient Maplin "Precision Gold" unit set to internal 1kHz modulation.




Here are the top of L2 primary and top of L2 secondary with both probes set to X10. The voltage gain across the transformer is a factor of 3.5, or 10.9dB. It may be possible to do better with a higher ratio transformer, but this has worked well here. The audio fidelity is OK, though it gets poorer with increasing regeneration.




Current Consumption Versus Supply Voltage Table

Under the signal conditions shown above, with moderate volume of 1kHz tone coming out of the earphone, and moderate regeneration in action, the current consumption varies according to the following table. You would expect to see this vary with different transistors within the "C" gain banding, but the low supply voltage, intrinsic emitter resistance, and L2 primary DC resistance will always keep it reasonable at the data sheet open-ended high end.

Supply (V)   Current (mA)

1.01         0.10
1.11         0.12
1.21         0.13
1.30         0.16
1.40         0.19
1.50         0.21
1.60         0.22
1.70         0.24
1.80         0.27
1.90         0.29
2.00         0.31
2.20         0.35
2.40         0.39
2.60         0.44
2.80         0.48
3.00         0.52
3.40         0.61
3.80         0.71

Not Good Enough For Yah?

You can stabilize the quiescent current consumption quite well using a more conventional bias arrangement, with some DC emitter degeneration and a bypass capacitor.



Supply (V)   Current (mA)

0.95         0.04
1.10         0.06
1.26         0.09
1.47         0.14
1.69         0.19
1.90         0.24
2.10         0.27
2.30         0.31
2.61         0.38
2.96         0.45
3.33         0.53
3.52         0.57


You can also stabilize the quiescent current consumption very well up to stupidly high voltages using the nice variable Vbe scheme seen in Horowitz and Hill.



Supply (V)   Current (mA)

1.05         0.07
1.16         0.08
1.26         0.11
1.36         0.12
1.45         0.14
1.55         0.15
1.78         0.17
1.99         0.18
2.38         0.19
2.99         0.21

Woah, Buttercup. Nice regulation, but that's looking way too complicated, and what's the point? I had four spare holes in the terminal strip, so there's no point getting too fancy if it doesn't actually give the end user any more sensitivity or give louder output. Within the boundaries of any project, it can be better just to do the job, and to provide fewer opportunities for failure. This applies equally as well to the marketing departments of global high-tech mobile device manufacturers, as it does to kids making radios.

As a final note, if you have one in your component racks, you might want to use a TO-92 NPN Darlington transistor such as a BC517, go up to a 3V supply, and increase R1 to about 10M accordingly. I tried this. It works really well, even on 1.5V. I considered using one in the kit, but I've been caught out with obsolescence before. Obsolete parts; The electronic engineer's nightmare. Using a BC549C or anything else like that keeps it simple, and gives the customer an entry into the exciting world of the transistor, with just five additional parts added to the original kit.

Navigate Up

Have you seen a mistake? This is a new page, so feel free to contact me at the email address below. This address has been the same since 1997, and unless I'm on holiday, it is checked daily including the spam folder.
 


Recent edit history

24-JUN-2025: Page created

© Henry J. Walmsley 2025.