Would a receiver of this sort offer the 
same degree of sensitivity and selectivity 
as a leaky-grid-with-reaction valve receiver? THE TRANSISTOR REFLEX SETS
  Details for this fully portable receiver were published seven times from 1962 to 1981; it clearly became very popular.   Gilbert Davey believed that this popularity amongst simple receivers was because radio frequency (RF) transistors became available at reasonable cost, and there was no need for aerial or earth.   Having built the BBC Focus two-transistor radio, with its unselective crystal detector front end, I was keen to try out a Davey transistor reflex design.   Would a receiver of this sort offer the same degree of sensitivity and selectivity as a leaky-grid-with-reaction valve receiver?   My chance to find out came during the 2020 Covid-19 lockdown.

Welcome and introduction,
contact details, disclaimer,
rights notice, BVWS link,
site map.
The man who introduced radio
construction to several generations
of boys, many of whom became radio
or communications professionals.
The history of the famous
one-valve circuit, 1948-78.
The Studio 'E' 1-valver and
the Focus Transistor radio:
downloads for these famous
designs, and lots more!
(This page)
A growing resource for those who
built Davey designs years ago, or who
simply wish to know more about them.
Davey sets built by contributors -
including designs that could have been
lost but for some prolonged detective work.
Two home-made coils to use in place of the all-too-scarce
commercial types, suppliers for other components, and
other sources of useful information.
Latest news, links to other sites of interest,
and news archive.
Hover your mouse over the
navigation buttons above for
brief page contents.
Pages marked  >  give access to
one or more subsidiary pages -
see also Site Map at bottom
of WELCOME page.
I decided to construct a  
unit that was fully testable 
before housing it in a case.
    My receiver, ready to fit into its case.

Publication history

This is both a regenerative and a reflex receiver.   The RF signal from the ferrite antenna is amplified by the first transistor, OC44 or equivalent, then part of this amplified signal is fed inductively back into the tuning coil (regeneration).   Having been rectified by a diode detector, the audio signal is fed once again (reflex) to the same transistor for amplification.   Further amplification is provided by a subsequent stage or stages.

The design appeared in two versions.   The first version was published twice; the later version appeared five times.

In the first version (3-transistor plus earpiece), the volume control takes the form of a rheostat in the collector circuit of the first stage (OC44 or OC45).   Two stages of amplification (OC71, OC72) follow, with the earpiece forming the OC72’s collector load.   In the October 1962 edition of Boy’s Own Paper, it appeared as part of a centre-spread “Amateur Radio” Gift Supplement.   Further notes were promised for the November 1962 edition, but nothing appeared.

This version appeared again in the Boy’s Own Companion No 5 (1963).   (Somewhat confusingly, details of Davey’s well-known homebrew LW-MW coil, used in versions of the Beginner’s One-Valver, were included within the text.)

In the later version (2-transistor plus earpiece), the audio output from the first transistor (OC44) goes to the slider of the volume control, whose track serves as one of the bias resistors for the second transistor (OC71).   The earpiece is wired across the collector load resistor.   This version first appeared in another centre-spread Gift Supplement in the Boy's Own Paper December 1963 edition.   Apart from recommendations for alternative transistors, it remained essentially unchanged for the following subsequent appearances:

Fun with Transistors, 1st edition, 1964 (Chapter 4).   The following chapter (5) carries two suggestions for powering a speaker.   The first describes (with text but no circuit diagram) the addition of a third transistor (OC72).   The second suggestion is to use the first stage plus diode detector to feed Davey’s version of the Mullard four-transistor push-pull amplifier;

Fun with Transistors, 2nd edition, 1971 (Chapter 4).   Speaker addition options were not included in this edition;

Fun with Radio, 6th edition, 1978 (Chapter 6).   Alternative (more modern) PNP transistors were suggested.   Here a note of caution is required.   Of the alternatives suggested, the AF117 is one of a series known to give trouble with the growth of “tin whiskers” which can short out the internal connections.   Unfortunately, too, Davey’s suggestion of the AC127 appears to be an error: although it is a germanium device, it is of NPN polarity.   A three-transistor circuit for silicon NPN transistors (BC108, BC109) was given at the end of this chapter, but it is configured somewhat differently from the above versions, and cannot really be considered as their stable-mate.   Although it employs regeneration, it does not appear to be a reflex design;

Fun with Silicon Chips in Modern Radio, 1981 (Chapter 4).   Robustly (p15), Davey excused inclusion of the second reflex version (and a crystal receiver) in a book about silicon chips by pointing out that germanium devices were the ancestors of the chip.   He clearly felt that the design still had life in it.   Once again, alternatives to the original transistors were given (AF117 or AF127, AC128).   The same cautionary note as above applies to the AF117.

My version

I decided to build the later version, as given in Fun with Transistors, 1st edition, Chapter 4, with the addition of the third (OC72) transistor to power a speaker (Chapter 5).   I bought new-old-stock transistors, and used as many “period” components as I could, but all the electrolytic capacitors are modern.   Davey does not give a value for the RF choke, but I had one from a 1960s radio kit which I built in to start with.

Above: Basic structure, with longer    
ferrite in trial assembly.    

Right: Ferrite and coil,    
with colour-coded leads.    

The front controls are united 
with the circuit board by a folded
Z-section bracket.
Ferrite aerial with
colour-coded leads
I decided to construct a unit that was fully testable outside its case.   Incorporation of the speaker required that, for minimum case size, my circuit board should be set back behind the speaker, with a hole to clear its magnet.   I therefore evolved my own component layout, rather than using Davey’s layout, which was based on a plastic case available when the second version of the design was first published.

Holes were pre-drilled for all small components, and these were placed from the rear, with most of the interconnections on the forward-facing side of the board.   The front controls are united with the circuit board by a folded Z-section bracket.   There is a hardboard packing piece, to stiffen the bracket, receive the countersunk heads of the three capacitor mounting screws, and minimise the projection of the potentiometer boss.   An extension of the bracket, beside the tuning capacitor, forms a spring clip to hold the PP3 battery against the side of the case.

I was concerned at the lack of stabilising components (resistor and parallel capacitor) in the emitter circuits of both the OC44 and the OC71, but decided to give Davey the benefit of the doubt to start with.   However, I allowed space (and drilled holes) for these components should they prove necessary.   Also, I had a choice of four output transformers to try, so I cut a "catch-all" rectangular hole (bottom right in the picture) that would accept the pinouts of any of them.

For the ferrite rod aerial, I had in stock two 3/8” diameter rods, one 4” long, the other 6”.   I made the sleeve from thin card, and wound the tuning and regeneration coils as Davey directs, but using 26swg enamelled wire instead of the double cotton covered that he specified.   I added extra turns and tapping points in case of need, but first trials have taken place with the number of turns that Davey specifies.   The complete coil is just loose enough to allow removal of one rod and insertion of the other.

The connections are:
1 (brown lead): start of main coil;
2 (red): tapping at 10 turns, feed to base of Tr1;
3 (orange): end of main coil, 50 turns total;
4 and 5 (yellow and green): regeneration coil, 7 turns, starting near tapping on main coil, and wound in same direction.

The mounting arrangement allows either rod to be adjusted for position within the coil.   There is ample scope for experiment with this aerial.

First trial

I first built up the radio with only the first two stages operational, exactly according to the published circuit, powering a crystal earpiece.   First trials were promising in terms of selectivity and volume, but there was some instability, with some inter-action between the volume control setting and the trimmer capacitor setting.   Then the OC44 gave up the ghost.   The OC44 spec gives a minimum hFE of 100; my example tested at 198 before failure, and around 230 afterwards.   I sent for some more OC44s, did some reading, and sent the published circuit to a knowledgeable friend, Andy, for his opinion.


I have re-drawn the circuit diagram entirely, in order to bring out a couple of points relating to the design, to provide a diagram of the output stage for which no diagram was published, and to record the changes I have implemented.

Circuit diagram (second published    
version), with my changes to the    
first two stages, and the output    
stage as described by Davey.    

Circuit diagram,
re-drawn entirely.

The 2-stage circuit (second version), as published, is in white.   The OC72 third stage is shown in orange, as described by Davey as text only in Fun with Transistors, 1st edition, and as I built it.   In red are shown the changes agreed for the first two stages after discussion with Andy, as follows:

A:   Stabilisation components were added in the emitter circuit of Tr1.   Davey, along with many young constructors, must have been lucky not to experience runaway, but it may be that the Mullard OC44s available at that time generally exceeded the 100 hFE minimum by less than later clones do (mine are made by Texas Instruments).   When I received my replacement OC44s, I tested them all for hFE, and chose one that tested at about 145.   I did notice that my metal-canned OC44s are very heat-sensitive: touching a finger to the case causes the hFE reading to rise rapidly!   By the way, the original glass Mullard OC44s now command very high prices, as the fuzz-box fraternity value them for their “Mullard germanium sound”.

Left: First choke fitted (top), and    
coil former used for the    
replacement choke.    

Right: Replacement choke in situ.    

Discarded choke, and the vintage
former as basis for its replacement. New choke in postion.
B:   The original choke I had used came from a medium-wave radio kit from the 1960s.   It tested at 360μH / 10Ω DC resistance.   Davey does not specify an inductance for the choke, but my reading had suggested that a figure of 1 – 2mH (millihenries) was needed.   I found an old (1930s?) coil former, in which the former itself is varnished paper, about 3/8” bore, with a brass ferrule, mounting bush and nut, and a dust-iron core with brass adjusting screw.   To this I fitted 1-inch diameter paxolin cheeks, about 3/8" apart, positioned so that the adjustable core could be centred within the coil or withdrawn entirely.   I pile-wound on as much 36swg enamelled wire as this would hold.   This tested at 5.1mH with the core centred within the coil – ample!   So I unwound wire until I could achieve 1.1mH with the core withdrawn and 2mH with the core centred.   Installing this in place of the original required a small mounting bracket and a cut-out in the circuit board to minimise the choke’s projection rearwards.   For the moment, I have left the core centred within the coil.

C:   Andy suggested that C2 should be increased from 0.1μF to 1μF for better audio and impedance match to the following stage.

D:   The volume control was re-wired conventionally, with its slider connected to Tr2’s base via a 10μF capacitor.   Thus the load upon the first stage should not vary much, if at all, with volume setting.   I used screened leads for the connections to the potentiometer.

E:   We reviewed bias resistor values for Tr1 and Tr2.   Because of the way the circuit was drawn as published, it took me a while to realise that R5 and R6 are TR1’s bias resistors – connected to the base by 10 turns of the tuning coil.   My re-drawn diagram hopefully makes this clearer.   Bias resistor values for Tr1 were not changed, but for Tr2, R3 (220kΩ) was replaced with a 62kΩ resistor.

F:   A 4.7kΩ resistor was fitted as Tr2’s other bias resistor, in place of the volume control’s track which had served this role.

G:   Stabilisation components were fitted to Tr2’s emitter circuit.

Front view of assembly, with    
shorter ferrite in position.    

Front view of assembly.
Making changes like these can be tricky, with the need to find space for new components, and the risk of damaging components and connections unaffected by the changes.   As mentioned, I had anticipated the possible need to add stabilisation components.   Squeezing in the larger home-made choke entailed some careful work, but was helped by the change of C2 from a large 0.1μF component to a small 1μF electrolytic.   The workmanship on the front face of the board is now a little less neat than I would wish, but I will live with that.

After these changes, the radio worked well, with ample earpiece volume and good selectivity, comparing well with what one expects from a leaky-grid-with-reaction valve set.   With no external aerial, it cannot be expected to pull in distant stations, though.   Davey recommends that the trimmer capacitor should be adjusted for good performance over the band, then left alone.   I find that (as with a valve set with reaction) there is not really one sweet spot that always serves.   If I were to build the set again, I would be tempted to incorporate a knob-operated variable component instead of a tool-operated trimmer.   If building the set into a case, one could compromise by providing an access hole for a trimmer tool.

I got the third (OC72) stage working without trouble, with more than adequate speaker volume on the most powerful stations.   Davey does not give a spec for the transformer, but I had four midget output transformers to try, with a small 8-ohm elliptical speaker.   All four worked satisfactorily, none causing excess current to be drawn.   I chose one that gives a turns ratio of about 30:1, as recommended by Andy.   As mentioned, I had pre-cut a hole in the circuit board that allowed the terminals of any of them to drop through, and then made a paxolin adaptor plate with holes to suit the pinouts of the selected transformer.   Soldering on over-long connecting wires made the plate captive on the transformer.   This assembly was bolted in place, and final connections made.   The speaker was connected by way of a chocolate block and generously long cable, to ease testing and assembly into the case.

The only problem I have noticed is that the set does not quite cover the whole of the medium-wave band; a little is missing at each end.   For proper coverage of any band, the tuning capacitor must have a capacitance range between fully meshed and fully unmeshed (expressed as a ratio) that is the square of the ratio of the highest and lowest frequencies of the band in question.

So for medium wave:
1600kHz : 530kHz = 3.02 : 1.
Capacitance range = 3.02 squared = 9.11 : 1.

My tuning capacitor, a high quality component, has a maximum capacitance of 399.5pF, and a minimum capacitance of 30.2pF, giving a capacitance ratio of 13.2 : 1.   From the circuit diagram as re-drawn, it will be seen that the tuning capacitor is not simply connected in parallel with the tuning coil, but that C3 (0.01μF) is in series with it.   By calculation, this very slightly reduces the capacitance range to 384 - 29.9pF, i.e. 12.84 : 1.   This should still be ample, so I can only conclude that my wiring layout has introduced some stray capacitance that adds to the minimum value, thus limiting the coverage.   I will look out for a 500pF capacitor of similar construction.   Alternatively, I could alter the inductance of the coil so as to bring within range the busier HF end of the band, while losing a little more of the quieter LF end.

I might also try adding a switch to put extra capacitance in parallel with the tuning capacitor, sufficient to bring in Radio 4 on 198kHz long wave.


For the case, I had in mind a simple plywood cabinet that any home constructor could have made, but that could be face-lifted as a 1960s-style appearance.   As with my BBC Focus radio, I cheated a little here and there, but again I will confess to these liberties as I go.

Above: Front view of case.    
Note captive nuts in front panel.    

Right: Rear view of case, and    
rear cover.    

Basic case, front view.
Basic case, rear view, with
rear cover removed.
The case is built up from 4mm plywood (back of an old wardrobe), glued together.   Overall dimensions (excluding controls) are 221mm long x 120mm high x 61mm deep – about the size of a house-brick.   The top, sides and floor are in two layers of ply, imparting extra strength to the corner joints, and as an easy way of forming rebates at front and rear.   The front panel is recessed by 4mm to allow addition of a grille and fascias.

At the rear, it can be seen that the inner ply layer on the floor of the case is narrower (front to back) than the other three.   Together with a thin fillet of wood (planed down from the stick of a rocket that landed in my garden), this leaves a groove that receives the two aluminium locating strips screwed to the rear panel (a gap is left to clear the home-made choke).   This arrangement leaves the floor of the case unobstructed, allowing easy insertion or removal of the chassis.

The chassis is fastened into the case entirely from the rear.   Near the controls, the front panel was drilled with three 6BA clearance holes, then carefully counterbored 5mm diameter from the front.   6BA nuts were pressed into the counterbores, thus forming captive threaded fasteners (I cheated and superglued them in place, but a 1960s constructor could have used one of several suitable adhesives available then).   At the other end of the circuit board, screws are driven into wooden blocks glued in place.


Many constructors will have been quite satisfied with a case in this form, perhaps with a coat of paint, but I decided to “face-lift” it somewhat, as might have been done by “Mr 1963”, a slightly more resourceful constructor.

Before gluing the front panel in place, I had cut narrow slots at its top and bottom edges to receive the folded-back top and bottom edges of an expanded-metal grille.   I covered the case sides, top and bottom, and back panel, with faked pale leathergrain-look Fablon, made in the same way as I did for my BBC Focus receiver: a suitable pattern printed off on the PC, covered with matt plastic film, and spray adhesive on the back.

"Face-lift kit".    
Inset detail: corner of fascia piece,    
made in two layers to produce rebate.    

Facelift kit.
The “face-lift kit” consists of:

* An expanded metal speaker grille, with top and bottom edges folded back;

* A black paper mask, to ensure that the speaker aperture is invisible through the grille (the edge of the front panel aperture was blackened for the same reason);

* Two decorative fascia pieces, each made in two layers: 1.6mm Perspex and 0.5mm acetate sheet, glued together to form rebates that master over the raw side edges of the grille (see inset detail).   These were covered with faked Fablon with a blue woven texture.   The right-hand fascia has two holes for the controls;

* Two “period” grey control knobs with brass brights, found among my spares.   Their plastic bodies were smartened up and polished with Brasso.   For the tuning control, I made a 60mm diameter disc from 2mm clear Perspex.   On the rear face of this I scored a 1mm wide vee groove, filled it with red acrylic paint and squeegeed off the excess.   This produced a dense red pointer line.   This disc was glued to one of the knobs to form the tuning control.   It projects a little from the fascia, so can be used instead of the central knob for fine tuning;

* Finally, a paper scale card, visible through the Perspex disc, carries index marks indentifying the most powerful stations.   (You guessed: I printed this on the PC, but “Mr 1963” could have produced something similar using text from a magazine, coloured paper and careful draughtsmanship.)   If I alter the tuning characteristics (see above) I can make a new dial to suit.

The finished case,    
front and rear views.    

The finished 60s-style case.
Rear views of finished set.

This is a lively little set with potential for experiment!   Suitable components are generally not hard to find.   It’s a pity that there is now so little of interest on medium wave, but listening to sports commentaries on Radio 5 Live conjures up the pleasure that many young constructors in the 1960s must have had with these receivers, at camp, while relaxing at home, or in the “dorm”!