This project was the result of needing an FM receiver that could be run all day, every day, in my workshop, off the 12V solar supply. Until now I had used the Mains operated 12AT7 Receiver with the Rohde & Schwarz vibrator inverter. However, the current consumption of this arrangement was higher than I'd like, especially when there are several cloudy days in a row. With the recent development and success of the 6BL8 super-regenerative receiver, I had ideas of developing a new receiver based on this design, but operating from the 12V house supply.
This project was also going to be the test for another concept I had been thinking about for quite some time - how to use 6 volt vibrators and transformers efficiently on a 12V supply. Broadly speaking, the design would incorporate the new 6BL8 circuit, a low power one or two valve audio stage, and a synchronous vibrator. Total current consumption would be around 1A at 12V, this being an insignificant load on the house batteries.

Receiver Module.
From my design work for the Model T Ford car radio back in 2003, I had left over an alloy diecast box and chassis which had once contained a 12AT7 receiver used for experiments. The idea was that the box would be clamped to the steering column of the Model T with wires run to the amplifier and power supply under the seat. The final outcome used a different method of construction, but I still had the receiver box and chassis left over. It seemed ideal to use the box and chassis for the new receiver, since all the metal work was already done, and the same layout suited the new 6BL8 design. Using a modular concept such as this, it meant that the receiver could be optimised independently of whatever design I chose for the power supply and audio stages. Furthermore, with all the VHF circuitry contained in a shielded box, the layout of the rest of the receiver would be non critical.

The 6BL8 circuit was constructed on the small chassis, and worked exactly like the prototype, with good reception of 2NUR-FM, a station about 135km away. This has proved that the circuit could be duplicated successfully. The B+ and heater connections are fed in via a 4 pin plug and socket, and the audio output is via an RCA socket. Aerial input is via a BNC socket - this being a legacy of the original Model T radio experiment.

Receiver module being tested prior to fitting into diecast box.

The chassis is attached to the box lid by means of the two potentiometers. As with some of my other super-regenerative receivers, an MSP metal tuning capacitor is used, originally intended for MW superhet receivers. The aerial section is not used, but the lower capacitance local oscillator section works quite well with a series capacitor to restrict the tuning range. The drawback to this scheme is that the stations at the high frequency end of the band tend to be cramped together, while those at the low end of the band are spread out. Nevertheless, the whole FM band is covered, along with part of the aircraft band. For those wanting a receiver specifically for the aircraft band, the circuit merely requires a small modification of the tuning circuit.

Top of receiver module chassis. At left is the audio output, followed by the 4 pin power socket, and at right, the aerial socket.

The diecast box was sprayed in green hammertone paint. This vastly improved its appearance, and once the vernier dial and knobs were fitted, it had a very professional look about it.

Under chassis shows compact construction.

Main Chassis.
This would support the speaker, vibrator, transformer, audio amplifier, and of course, the receiver module.
A suitable looking 4" speaker was taken from my collection, but as is typical with many speakers of small size, there were no mounting holes. To support it on the chassis, I made a U-shaped bracket which clamps around the magnet. This worked perfectly. I have some reservation in mounting a bare speaker like this with no baffling, because it means bass response is less than it should be. However, a super-regenerative receiver is not hi-fi, and this receiver would not be operated with particularly high volume anyway.
Once the speaker type had been determined, the next thing was to confirm the vibrator transformer I wanted to use would be suitable. The details of this will be described later.

Vibrator transformer being tested.

With the larger parts now known, the chassis plan was drawn up, and the aluminium cut and bent. The hole for the vibrator socket was punched out, and the socket itself mounted on rubber grommets to reduce mechanical noise. The vibrator, a Mallory 245, requires a UY-5 socket. It so happens that the plane of the reed inside the vibrator is in line with the socket mounting. This method of mounting works well because the vibrator can swing in the direction of reed travel.

Starting to resemble the final product with the major parts mounted.

Some thought had to be given to the audio amplifier because of the B+ voltage being around 150V, and as will become apparent later, the heater current needed be around 700 to 800mA. Various combinations and types of valves were considered, and ultimately a 6Y9 was chosen. This was tested on a solderless breadboard to confirm suitability for audio use with this receiver. Once this was done, the speaker transformer and a 10 pin valve socket were mounted.

Speaker transformer and 6Y9 socket mounted, along with fuse holder and supply cable grommet.

The new design uses a novel method for operating a 6 volt vibrator. Voltages are marked in red.

1. Receiver.
This has been discussed in detail here. A 6BL8/ECF80 triode pentode valve is used, with the pentode section performing as an untuned RF amplifier. This is required to prevent aerial loading affecting the detector performance. It also provides a small amount of gain and reduces detector radiation.
The triode section operates as a self quenched super regenerative detector. The triode is wired as a Colpitts oscillator, with a quarter wavelength cathode choke and the grid to cathode capacitance causing oscillation. It oscillates at the received frequency (88-108Mc/s) by virtue of the tuned circuit. This consists of a four turn coil and a 100pF tuning capacitor with a series 33pF capacitor. The latter is used to restrict the tuning range. Ideally, a 15pF variable capacitor would have been used, but as I have a good supply of N.O.S. 100pF types, one of these was used instead. The tuning capacitor is actually a type intended for superhet MW receivers, and in this circuit I am using only the local oscillator section.
Quenching is obtained by the grid circuit time constant; i.e. the 330K and 33pF. The long time constant causes the circuit to go in and out of oscillation at a super-sonic rate. The amount of oscillation needs to be adjustable for optimum reception, and this is achieved by the 20K cathode rheostat. It is bypassed for VHF by the 1000pF, and for audio by the 1.5uF.
At the triode plate, the detected audio is present, along with the fairly high amplitude quench waveform. This is filtered to a sufficient degree by the 150K and two 1000pF capacitors.
Because both ends of the 6BL8 heater are above earth in this receiver, both heater pins are bypassed with .0022uF ceramic capacitors. This is important because the cathode for the detector triode is live with RF, and some energy will be capacitively coupled to the heater.

2. Audio Amplifier.
This part of the set is based around a 6Y9/EFL200 dual pentode. The 6Y9 is a frame grid valve introduced in the 1960's for television use. It is unusual in that it has a 10pin base which looks very similar to the more common B9A type. In Europe, the series heater type PFL200 is more common, but in Australia, the 6Y9 was very popular in sets using European valve types. The two pentodes are not the same. One is a low power signal pentode typically used as a sound IF amplifier, and the other is a high gain video amplifier.

Video output valves  work well as medium power audio output valves, with the higher gain being useful in many instances, and in Australia this was commonly done. For instance, there were several mid 1950's STC radios using the 6CH6 in the audio output. Most commonly, the 6DX8 was used in TV sets, and a few radios and record players. Philips in Australia actually gave the 6DX8 official audio ratings for both single ended and push-pull use.

In one article of the Philips Miniwatt Digest, some thought was given to using the 6Y9 as an audio valve when this valve was introduced. However, if they did give it ratings for audio use, it would have been in a later issue which I do not have. I have seen the circuit of a Calstan portable radiogram using a 6Y9, so the valve has been used in at least one commercially made product this way. A particularly interesting aspect of the 6Y9 is the incredibly high gain, at 21mA per volt for the power pentode. One could envisage the power pentode being fully driven from a crystal pickup, which is done with the Calstan circuit, or a one valve regenerative MW receiver able to drive a speaker at full volume. However, it must be remembered that these high figures of gain can actually be problematic for audio use, and careful attention to design is important.

Part of the Australian Philips data for the 6Y9.

As I have a lot of 6Y9's in my collection, I thought that this new receiver would be ideal for testing this valve for audio use. Furthermore, I was looking for a valve which would pass about 25mA at 150V (the reasons for this will be described in the power supply section later).
A test circuit was wired up for the power pentode, and there didn't seem to be any difficulty with getting it to operate. Because of the high gain, screen and grid stopper resistors are necessary. A cathode bypass capacitor was not required in view of the high gain, and eliminating it also provides a convenient method of obtaining negative feedback.

With a B+ of 136V, maximum output was 285mW, and input sensitivity was 1.2V. If a cathode bypass was included, sensitivity was 540mV. B+ current was 26mA. The ideal load impedance is therefore about 5.2K, which suits a 5K P.A. line transformer as I have used with other projects.

The super-regenerative detector can't quite drive the 6Y9 power pentode to full output, so the signal (RF) pentode is also required. Like the power pentode, this too has a very high gain. In view of this, gain was kept low by means of a low value plate resistor (47K as opposed to the usual 100K to 220K), and  again by not bypassing the cathode. Without cathode bypassing, the stage gain is 12, and with the capacitor it increases to 52.
It could be imagined that with both pentodes of such high gain, using them together as an audio amplifier is likely to produce instability, and not surprisingly this did occur. A 330pF capacitor at the signal pentode grid cured the problem, along with a 3900pF bypass at the power pentode plate.

A super-regenerative receiver is quite noisy on a weak signal, and as this receiver was intended to listen to a station which is often weak, it was thought a good idea to include a treble cut option. By reducing the high frequency response, the noise is reduced and the receiver is more pleasant to listen to. In this set, a .0015uF capacitor can be switched in between the 6Y9 signal pentode plate and earth.

Finally, as noted elsewhere on this site, some P.A. line transformers can be problematic as valve output transformers. The resultant audio often sounds very shrill and unpleasant to listen to. This can be improved by operating the output in ultra-linear or triode mode. This was the case with the new receiver, and the 6Y9 screen was fed from the 2.5K tapping on the transformer primary. Further details of the problem are described here, and in fact the same type of transformer was used.

3. Power Supply.
This set was going to use a vibrator to enable operation off the 12V house supply. While there are several vibrator powered sets described elsewhere on this site, this set takes a very unusual approach and most of this section will be spent describing it.
I have a lot of 6V vibrators in my collection and it seems a shame not to use them in new projects. The problem of course is that 12V supplies are much more common than 6V in the modern day. The series drive types can easily be used on 12V with a resistor in series with the drive coil.
But, a large proportion I have are shunt drive types and I wanted to use one of these. However, except for a very few types, it is not possible to isolate the drive coil to include an external dropping resistor. Thus, the vibrator can only work on 6V.
Furthermore, I wanted to use a synchronous vibrator with synchronous rectifier circuit. This of course requires a proper vibrator transformer with centre tapped secondary. And, as it happens, most examples I have of this type are also 6V.

So, how can one use a 6V vibrator and transformer on a 12V supply? One method which allows efficient use of a 6V transformer is described here, but is only suitable when the vibrator is 12V, or is a series drive 6V type with coil resistor in series. The vibrator must be a split-reed type, and a valve or solid state rectifier must be used.

At first thought, a dropping resistor would be a simple way to use a 6V vibrator and transformer from a 12V supply. However, there is a problem in that until the valves warm up, the load on the vibrator is much less than normal, and because the resistor is chosen for dropping 6V at normal operating current, the vibrator and transformer will be exposed to higher than normal voltage.. Apart from not being good for the vibrator drive coil, the transformer may suffer damaged insulation, along with the buffer capacitor being overly stressed. In the case of circuits using a synchronous or solid state rectifier, the filter and other capacitors may also be exposed to excessive voltage. Also, it needs to be remembered that unless there is a large amount of capacitance connected to the transformer centre tap, the supply actually rises back up to 12V in between the time when the contacts make. Keeping in mind the drive coil is in circuit during this time, it can be seen that it will be exposed to excess voltage.

This is NOT acceptable because the vibrator, transformer, and other components, are exposed to excessive voltage when the supply is not fully loaded.

If, however, the supply is always loaded, the idea becomes viable. In this power supply, this technique was used successfully. The B+ output of the power supply is clamped  by zener diodes, and thus the vibrator input current is always the same, regardless of if the valves are drawing current or not. There is a large electrolytic capacitor connected across the 6V supply so the driving coil is not exposed to excess voltage.
Alternatively, the 6V supply can be clamped by a zener diode instead. During warm up, the zener conducts, clamping the input to the vibrator to a safe level, and once the valves have warmed up, the zener ceases to conduct.
The drawback of course with a dropping resistor is efficiency loss, because half the input power is wasted across the resistor, but in this case the output power required was very low to start with, and the power wastage was acceptable. In practice, the zener diode of course needs to be a high power type.

Zener diode prevents the 6V supply rising during periods of no load, such as valve warm up. Drive coil is also protected when reed is between contacts.

A further extension of this concept is quite ingenious and can result in efficiency as good as if a 12V vibrator and transformer were used to start with. Until now, we have considered only the vibrator and its transformer, but in most instances the vibrator will be powering something with valves; e.g. a radio receiver. Most indirectly heated valves, such as the ones used in this receiver, are designed for 6.3V operation. So, what if the heater circuit happened to have the same current consumption as that of the vibrator? It can immediately be seen that the valve heaters could take place of the dropping resistor. Effectively, the vibrator would act as the dropping resistor for the valve heaters, and conversely, the valve heaters would act as the dropping resistor for the vibrator. The result is that both the valve heaters and vibrator get their required 6 volts without any other resistors and full efficiency is obtained.

The choice of zener voltage requires some thought. It should not merely be 6V (6.2V is the closest preferred value). This is because a 12V battery supply can vary from about 11V to around 15V depending on charge, and if the battery is being charged when the receiver is in use. If the supply was 14V as when the battery is being charged, and a 6V zener was used, then 8V would appear across the valve heaters. Also, the zener would be conducting, getting warm, and wasting power. By having a zener of 7.5V, the voltages between vibrator and heater would divide equally up to a supply input of 15V. Thus, if the supply was 14V, the heaters would receive 7V which is much more acceptable, and no power would be lost in the zener diode. The warm up voltage of the vibrator would be a maximum of 7.5V, but this is within ratings as vibrators are designed to work over normal battery voltages. Essentially, the zener diode operates only during warm up. It should be obvious that when conducting, the zener diode current will be the normal full load current (at 6V) of the vibrator supply minus its no load current. If the vibrator is removed, the zener has to carry the full heater current.
Under certain fault conditions that cause the vibrator to draw excess current, such as a short circuit on the B+ supply, the valve heaters would receive excess voltage. However, with modern capacitors used the chances of this are so remote as not to be worth worrying about. Several methods could be used to protect the heaters if this was of concern.

In practice, to design this setup to work efficiently requires a bit of a juggling act so that the vibrator current works out the same as the heater current. This means the total B+ current has to be taken into consideration, as well as the valve types and their heater currents.
Of course, a balancing resistor can take care of any discrepancy between the two currents. For example, if the vibrator draws more current than the heaters, a shunt resistor can be connected across the heater circuit, and vice versa.

A disadvantage of connecting a zener diode across the vibrator input is that the supply must be polarised. If the supply is reversed, the zener conducts like a normal diode with a drop of about 0.7 to 1V. In this case, the 6.3V heater string would receive virtually the full 12V.
The zener diode has to be made bi-directional therefore, so it clamps with both polarities. This can be done by including a bridge rectifier before the zener, or more simply with two zeners of the same voltage back to back.
In the case of the bridge rectifier arrangement, 1.4V needs to be added to the zener voltage to obtain the clamp voltage. With back to back zeners, only 700mV is added.

It is true that for synchronous vibrator circuits, the input is already polarity conscious, but accidents do happen, and polarity may be reversed until it's realised the radio is not working after the warm up time. While the B+ electrolytics will be exposed to reverse voltage in this situation, they will stand it for the short period of time. The vibrator may be overloaded since the B+ current will be somewhat higher, but no lasting damage should be done if the set is switched off quickly.
Simply connecting a diode in series with the 12V supply to protect against reverse polarity wastes voltage, which is important particularly as the batteries approach discharge. A diode shunt connected so it blows the supply fuse is commonly used with 12V circuits, but it is inconvenient as one must have a supply of fuses to hand, "just in case". Alternatively, a relay can be used; the normally open contacts in series with the supply, and the coil fed via a diode. Thus, the relay only connects the supply if the polarity is correct. A simpler and better scheme is to connect a diode in series with the first filter electrolytic on the B+ side. Here, the small voltage drop is trivial. Of course, where a non synchronous vibrator is used, input polarity is not important, provided any capacitors on the 12V side are non polarised.

Now to the design of the power supply used in the new receiver. The transformer chosen was a type made by Radio Corporation and used in Astor receivers amongst others. It was tested and found to provide about 150V. Clearly, it was intended for operating battery type valves, and would have been used in a domestic vibrator set. Typical current rating for such a transformer is about 20 to 40mA. Another transformer I had was of different manufacture, and with a larger core and 250V output would have been intended for car radio use. I chose the 150V transformer in view of wanting to keep the input current around 1A, which it would be with around 20mA current draw at 150V.
The next question was what vibrator to use. This time I wanted to draw on my supply of U.S. made types. Most of my synchronous types happened to be made by Mallory and I decided on type 245. This fits a UX-5 base which is one reason I chose it; I have a good stock of these. Being of Mallory quality, it didn't take long to get the vibrator started and the contacts clean after many years of not being used.

Timing Capacitance.
One of the most important aspects of vibrator circuit design is of course the timing, or buffer, capacitor. The value is important, and if not chosen correctly, vibrator life will be short. Much discussion has already been provided in articles elsewhere on the site with regards to this.
For the Radio Corporation transformer, a look through all the circuits in the Australian Official Radio Service Manual showed that all sets using it had a .004uF timing capacitor across the whole secondary, and a .05uF across the primary. The latter value has only a very small effect on overall timing capacitance and exists primarily for interference suppression. The vibrators used with this transformer were all Ferrocart synchronous types. Looking through types that would have been used, all were 100 or 115 cycle. Therefore, the nominated timing capacitance should suit the Mallory types operating at 115 cycles. Some time was spent determining a suitable value, as it seemed that .004uF had been chosen as an "ideal" value which does not allow for vibrator aging.

Selecting the vibrator timing capacitor with a decade capacitance box.

Mallory advises that the slope of the waveform should be 65%. In other words, the amount of timing capacitance is made slightly more than the ideal. This is because as a vibrator ages, the contact spacing may increase slightly. If this happens, and the circuit has been set up with the ideal value of capacitance, there will now be insufficient capacitance, with all the problems that entails. It is more acceptable to increase the value of timing capacitance than decrease it, since contact damage as well as transformer insulation failure can occur with insufficient capacitance. Increased capacitance is not ideal from a timing point of view, but the transformer will be protected, and contact wear will be minimal provided the capacitance is not excessive.

Left waveform shows .004uF; at right is with .0068uF.

Experiments found that for a 65% slope, the capacitance should be .0068uF. As can be seen in the right hand side waveform, the slope is measured at 15.4V from the top of the waveform, which is 23.4V. A simple equation of 15.4 / 23.4 gives 66%.
A 2000V polypropylene type was used.

It is good practice to include primary damping resistors, so 100R 1W types were connected across the primary contacts.

6 Volt Vibrator Supply.
Once the audio section had been constructed, it was found that the input current to the vibrator was 1.25A at 6.3V. Conveniently, this happened to be exactly the same as the heater current; 450mA (6BL8) and 800mA (6Y9). Thus, no balancing resistor was required.
Two 6.8V 10W zener diodes were connected back to back across the 6V vibrator supply. During warm up, the voltage rises to about 7.5V. This is within limits of the vibrator circuit. As the zener diodes dissipate power for only a short time, minimal heatsinking is required. A strip of aluminium on which the diodes are attached to each other is adequate.

12 Volt Input.
A two pin polarised plug connects the receiver to the house supply. A fuse is included to protect the wiring, particularly in case of switch break down. The gauge of wire connecting the set would burn up before the house breaker tripped otherwise.
The power switch is a two pole three position rotary switch of Oak (MSP) manufacture. As well as turning the receiver on or off, the third position also switches in the treble cut capacitor.

Under chassis wiring completed. The two 10W zener diodes are visible at the top towards the middle.

Once the wiring was completed and the receiver was operating correctly, the front and rear panels were labelled. While it is true that only I would operate the receiver and therefore know what the controls are, it does add something of a professional appearance to label everything.

Front panel labelling.

Rear panel labelling.

Performance.
In terms of sensitivity, the receiver works just the same as the other valve super-regenerative receivers described on this site. Sound is slightly lacking in bass because of the speaker not being baffled. The tone control works effectively, and makes weak signal listening easier.
The  low current consumption is very pleasing, and the new technique of using 6V shunt drive vibrators with a 12V supply works perfectly. This method will be used in future projects.

Home