6 Transistor Super Regenerative Receiver.

The following is a design for a separately quenched super regen receiver I first tried in early 1992.  It is based on a circuit which appeared in Practical Wireless for July 1981. It worked far better than any other solid state design, so I built a portable version for use during my daily commuting on the train. Running off 10x AA nicads, this gave me a week's listening before recharging.
My alterations to the PW circuit were to include varicap tuning, an RF amplifier, and an audio amplifier.
I submitted my modified circuit to Silicon Chip, whereupon it was published in the April 2003 issue.
My portable version differs in that it uses varicap tuning, the output transformer has a 1K primary, and the output transistor is a BC108, with bias components to suit. It also uses the headphone lead for the aerial. There is no doubt that the separate quench feature makes all the difference. Simple one transistor self quenched circuits cannot match this design for sound quality or sensitivity.

Circuit of the prototype using an external aerial and a conventional tuning capacitor.

Circuit description
This super-regenerative receiver is essentially a VHF AM receiver, with slope detection used for FM. By tuning to one side of the carrier, the receiver's tuned circuit converts FM to AM. The bandwidth is about 200kHz so wideband FM stations can be demodulated by tuning the receiver to the most linear point of the response curve, rather than the top of the curve as one would for AM. In practice, this simply means tuning for clearest sound.
The heart of the receiver is Q2 which is a Hartley oscillator, with its tuned circuit in the base circuit. It determines the frequency of oscillation and hence the receiving frequency.
RF amplifier Q1 is a self biased, untuned common emitter amplifier, included to prevent aerial loading from affecting the detector's oscillation frequency and amplitude. It also reduces any RF radiated from the aerial. RF is coupled into the oscillator coil by C2. The aerial can be a piece of wire cut to 75cm. A 75cm telescopic rod aerial is better, but a proper outdoor FM aerial is preferred for non portable use.
Most simple super-regenerative detectors are self-quenched, however this makes it difficult to obtain the optimum quench waveform. Particularly for wideband FM, the quench waveform has considerable effect on sound quality.

Quench waveform at the UJT emitter. The 74.9KHz quench frequency shown here provides a good sound quality, but should be reduced if more sensitivity is required.

In this receiver, the quenching of the detector is achieved by Q6, a unijunction transistor (UJT) relaxation oscillator. The emitter of the UJT provides an approximate sawtooth waveform, which as it also provides the bias supply for Q2 takes the detector in and out of oscillation at about 50kHz.
It is necessary to be able to set the optimum quenching voltage and this is done by adjusting Q6's supply by pot VR2. This effectively functions as the regeneration control.
Present at the collector of Q2 is the demodulated AM or FM signal as well as the supersonic quench. This is of sufficient amplitude to overload the following audio stages, so C6, R7,C7 and C9 provide simple low-pass filtering.
Transistors Q4 and Q5 form a class A amplifier, which can provide about 80mW output. Bias stabilisation is automatic using current feedback. If the current in Q5 rises then Q4 turns on harder, reducing the bias for Q5. Negative feedback is obtained from the secondary of the speaker transformer and fed into Q4 via R18. The windings of the transformer must be phased correctly, otherwise the amplifier will oscillate. The transformer is a standard transistor 500 ohm to 8 ohm output type.

The prototype receiver uses the local oscillator section of a plastic AM radio tuning capacitor, which has a maximum capacitance of about 60pF.  To restrict the tuning range, a 39pF capacitor is connected in series. (The aerial section of this variable capacitor tunes a ZN414 AM receiver in the same enclosure, sharing the same audio amp).
The air-cored coil (L1) consists of four turns of 18 gauge B&S tinned copper wire with 3/8" ID and tapped at one turn. With this coil, frequency coverage is about 60-150MHz depending on tuning capacitance.
As with all VHF circuitry, some care needs to be taken with construction. The protoytpe was assembled on a piece of blank PCB with the copper cut into small squares, to form isolated pads. The portable version was built on a small piece of Veroboard.

The portable receiver runs from 10x 500mAh NiCd cells to provide 12V. Not shown is the charging circuit which is simply a 330R 1W resistor to provide a relatively constant charge current from a 30V power supply.

When using this, or any other super-regenerative receiver, it may be found that an audible tone is heard in the background when listening to a station transmitting stereo or SCA programs. This is a result of subcarriers beating with the quench frequency. Adjustment of the quench frequency will usually minimise the problem.
With this receiver, if adjusting VR2 doesn't get rid of it, then it's worth experimenting with C11. It's important to note that raising the quench frequency too high will reduce receiver sensitivity. Decreasing the quench frequency will improve sensitivity but the subcarrier beat will be more evident. As SCA is now largely extinct, it is possible these days to use a lower quench frequency around 35KHz.
Further decreasing quench frequency will make the quench audible at all times. For non FM stereo/SCA applications, C11 can be increased until just before the quench becomes audible.
Optimum sensitivity occurs with VR2 adjusted to the point where the receiver has just gone into oscillation.
At this point, a "rushing" noise becomes evident and stations can be tuned in. With very weak signals, it will become obvious that the settings of VR2 and C4 interact slightly. I tested this receiver with an HP8654B signal generator and could receive a 3uV signal, albeit with some noise.

Portable version with headphone lead as the aerial.

Note the three pots. The portable version uses varicap tuning. The 2.5mm socket at the rear is the 30V input for charging.

The prototype shown alongside the portable version. The small PCB on the left is an MK484 receiver for Medium Waves, while the PCB on the right is a two transistor audio amp. At rear is the HP8654B signal generator used for testing sensitivity.

Taken from my original notes, this is how to use varicap tuning and the headphone lead as an aerial. The 1N914 in series with the zener diode provides temperature compensation.

Fellow FM enthusiast, Andy Mitz, who authored the now defunct "somerset FM" site, decided to have a go at building this receiver with a few minor changes.
The front end was kept as is, but an audio amp IC replaced my two transistor circuit, and an 18V supply was used. This is what Andy had to say: "I have attached some photos of a regen build using much of your design. This version uses a Motorola varicap diode and a Philips audio amp chip. The unit is sensitive (does not need the whip antenna), selective, and has enough audio to overdrive the speaker."

Inside and outside views of Andy's receiver. The 18V supply would provide better
stability for the varicap tuner's zener diode stabiliser as well as providing high audio output.
Here's the circuit in .pdf.

And, finally a disclaimer. While super-regenerative receivers can provide very good sound quality, this is dependent on a number of things. Don't build them if you are specifically wanting hi-fi performance.