One valve FM receiver with loudspeaker.

From L to R: On/Off/Volume, headphones, regeneration, tuning. Speaker is 4".

This little receiver is a development of my original 12AT7 receiver, so if you haven't read that article you'll need to in order to familiarise yourself with this design.
With the success of that receiver, and also reasonable results when feeding my other 12AT7 FM tuner into a 6M5 pentode for loudspeaker listening, I had for some time contemplated a single valve FM receiver for loudspeaker use. It would be the simplest loudspeaking FM receiver imaginable. The initial idea was to use the reflex principle of the original, but with a power pentode to drive the speaker. As it is, the original 12AT7 receiver drives headphones with more than ample volume. However, it was found that with a speaker that the volume is rather lacking.

The 6DX8/ECL84.
I decided on using a 6DX8. This valve is one of the most popular Australian TV valves. It is otherwise known as ECL84 and is a triode pentode. In Australian TV sets,  the pentode is used as a video output valve and the triode for gated AGC or sync separator service. Many Australian sets also used it for audio amplification and output. Europeans would know this valve better as its 300mA series heater version; 15DQ8 or PCL84. There is also the 10DX8/LCL84 used for Japanese and U.S. designed 450mA series heater circuits. The reason for choosing this valve was that the triode has characteristics very similar to a 12AT7, and this was proven in previous experiments. Secondly, the pentode has high gain. This is important with no intermediate audio amplification stage.

The circuit.
The triode functions exactly as per all the 12AT7 receivers described on this site. It is a self quenched super regenerative detector with regeneration control performed by controlling the bias of the triode. The cathode RFC in conjunction with grid to cathode capacitance makes the circuit function as an ultraudion or electron coupled oscillator. Oscillation (and thus receiving) frequency is determined by the tuned circuit connected to the grid. The time constant of the 33pF and 270K causes the triode to go in and out of oscillation at a supersonic rate which is the key to a super-regen receiver. A 15 or 20pF tuning capacitor would be nice, but they're not as common as the 60pF units I have, so by inserting series capacitance of 22pF we can use a 60pF tuning condenser to cover the band. Audio appears at the triode plate along with the quench signal. The 150K and following 470pF is a simple quench filter. Too much quench signal will overload following audio stage so it is best to remove as much as possible.
Oscillation level, and thus super-regeneration, is controlled by a 10K cathode rheostat. It is, to a degree, self stabilising because of the DC feedback.
The filtered audio is fed into the output pentode's grid in the usual way, via the 1M volume control. A much amplified signal appears at the pentode plate and feeds the speaker via matching transformer. Cathode bias is used, obtained by the 180R and 47uF audio bypass.

Circuit of the receiver. It is hard to imagine anything simpler that can drive a loudspeaker from distant FM stations. Note the first filter capacitor is 330uF but can be much less than this.

No Reflexing.
Initial experiments trying to reflex the output stage as an RF amplifier, as per my original 12AT7 receiver, were a failure. Trying to use an audio power stage as a reflex amplifier has limitations. The problem is that due to the high voltage swings involved, the gain of any RF signal passed through the same stage will also vary at the audio rate. This results in instability at high volume. With a receiver like this, with only one valve, it is often necessary to have the volume at full. The original 12AT7 receiver can get away with it because its audio level is much less.
In order to keep to the "one valve" concept, it was clear that the RF amplifier would have to be omitted, and the 6DX8 used in a non reflexed circuit.
Now, the RF is fed straight into the triode grid via a 1.5pF condenser. This might seem to be a very low value, but in practice it does not detract from sensitivity to any significant degree. In fact, a 1uV signal can be heard. 20uV is quite clear. The low value prevents any serious aerial loading effects detracting from performance. The 100R resistor terminates the aerial in something like the correct impedance, and also provides a more definite load for the 1.5pF when fed by the oscillating tuned circuit. Essentially,  it reduces the effect of different aerials. Normally, I do not like connecting aerials directly to tuned circuits with no primary winding, but this circuit does seem to work very well. While an RF amplifier is ideal, remember this is a one valve receiver driving a speaker, so we can't have everything.
The tuned circuit is 4 turns of 18 gauge tinned copper wire, air cored with a diameter of 10mm. Easiest way to make this is by winding the wire over a 10mm drill bit.
The cathode choke is 75cm of 25 gauge wire wound on a plastic former of about 6.3mm diameter. 75cm is a quarter wavelength at 100MHz, and so represents maximum impedance in the middle of the FM band.
Note that if oscillation seems poor, the connections to the choke may need to be reversed. This is because of magnetic coupling to the tuned circuit. If the coupling is in the wrong direction, it will cause negative feedback, and thus a reluctance to oscillate.

Improved audio level.
Now that the RF performance was as it should be, the next thing was to see what could be done about increasing the audio level. Here, I found quite an improvement was gained by increasing the triode plate resistor, from the original value of 56K. Values as high as 500K worked very well with a substantial increase in audio output, but it was not possible to maintain oscillation across the entire band. 150K was found to be the best compromise. Although I did not try it, a 500K plate resistor might work with a much higher B+. As a result of increasing the plate resistor, the quench filtering was improved. This means less supersonic overload for the output stage. The grid resistor has been reduced to 270K to increase the quench frequency slightly, reducing the subcarrier beat.
It was necessary to include a 270R screen grid stopper to prevent oscillation in the output stage, but this is normal with high gain pentodes. There was some slight instability with the volume full on, but a 220pF from control grid to earth fixed that. It also provides some extra RF and quench filtering. It's possible that with different layouts the latter two components might not be required, but it's still good practice to include them.

I built the receiver in a diecast box, with the lid forming the bottom and the box itself the chassis. It is an ideal method of building small valve circuits if you don't have access to metal work equipment. Not only is it rugged, but it provides an excellent groundplane. The diecast alloy is very easy to work with.

Under the chassis. Speaker transformer is in centre with power supply to the left. The original aerial coil socket can be seen next to the tuning condenser.

My power supply uses a 140V transformer which feeds a half wave rectifier and 330uF 200V capacitor. Why such a high value? It's because that value is commonly used in computer power supplies and I have many of them. Normally, about 33uF would be used as the first filter. Needless to say, the output is hum free. The resulting 170V DC feeds the pentode plate. A 2.7K drops this to 157V for feeding the detector and 6DX8 screen grid. There is a degree of lamination buzz from the transformer due to the half wave rectification. However, it is not bothersome and could be eliminated simply by using a full wave bridge. There is a 5.8V winding for the 6DX8 heater which draws 720mA. As to the unusual 5.8V heater winding, the transformer was from a VTVM, where the lower heater voltage is not unusual. The lower voltage is used in VTVM's to reduce grid emission. In this receiver, the slight reduction of heater voltage is not important. As the entire B+ consumption is around 10mA, I could have used a 6AL5 for the rectifier instead of the 1N4007. Note that if this was done the first filter capacitor would need to be reduced to around 10uF to keep the 6AL5 peak current within limits, and a surge limiting resistor included if necessary. The .005uF across the diode was needed to remove modulation hum.

Power transformer before cover fitted. With exposed mains terminals it was essential to cover the transformer. Transformer has 140V 40mA and 5.8V 1A secondaries.

The 6DX8 pentode isn't being operated anywhere near its full ratings. It would be pointless to do so, as due to the low input voltage it cannot be fully driven.. For the output transformer I'm using my usual favourite which is a 100V line transformer. The impedance should be 11K, which equates to the 660mW tapping.
Initially, I tried a plug in arrangement for mounting the aerial coil, using a bakelite speaker plug and socket of the kind used in many Australian made radios and TV sets during the 1960's. Unfortunately, the plug in arrangement was a failure. The material the plug and socket is made of is lossy. This was clearly evident with a spectrum analyser showing the oscillation strength when the coil was connected directly and then via the plug and socket.
One useful experiment I did try was a one turn coil like the Heathkit FM tuner uses. This worked very well, and as I suspected, it eliminated the need for an external aerial on a great number of stations. Of course, like any loop aerial it was directional. If you want to experiment, wind one turn of 18 gauge tinned copper wire around a size C cell.
I subsequently removed the aerial coil plug and socket, and mounted the coil on a piece of fibreglass PCB instead. This fixed up the poor oscillation.

New aerial coil is mounted on a scrap of PCB. Don't use a plug and socket - it's too lossy.

Very good reception of the Sydney FM stations about 80km away was possible with an indoor VHF TV aerial. Central Coast and some Newcastle and Illawarra stations were also receivable over distances exceeding 100km. Like any good super-regenerative receiver, strong stations are receivable with no aerial at all. Just the tuned circuit will pick them up.
The audio is quite loud considering there is only one valve, and with headphones uncomfortably so with the volume turned up. It's certainly adequate for quiet room listening. For optimum performance, the regeneration control should be at the minimum that allows oscillation with freedom from subcarrier beat. At this setting, the set is at its most sensitive and will produce the greatest volume. Note that the regeneration control for a super-regenerative receiver is operated quite differently to that used with a conventional regenerative receiver. For a super-regenerative receiver, the detector is set so it is always oscillating. The most sensitive point, with the strongest audio, is just after oscillation has commenced. If the regeneration is increased further, sensitivity and volume is reduced. Sensitivity is into the single digit microvolt range.
Considering that this receiver has only one valve, receives VHF FM, and drives a loudspeaker, I think that's quite impressive. Anyone who built this receiver in its original reflexed form should upgrade it.