12 Volt Superheterodyne Receiver

This Medium Wave AM Superheterodyne receiver works entirely from 12V with no high voltage supply.

A valve radio that operates from 12V? Nothing unusual you may think; after all millions of valve car radios were powered by 6 or 12V accumulators. However, these still required the usual 200-250V supply which was normally provided by a vibrator power supply, or in some sets, a genemotor. The set described here does not require the high voltage supply.
Now, you're probably thinking of the valves released in the late 1950's used in hybrid car radios. These did indeed run off 12V for the plates and screens. The audio output stage in these sets used one or more transistors as even this series of valves was still incapable of much power output at low voltages.
No, this set uses ordinary 250V valves of the kind used in mains operated radios and runs entirely from 12V DC.

Low voltage operation of valves.
My attention was drawn to the possibility of low voltage operation in the mid 1980's when I was doing extensive work with my two valve regenerative receiver. In designing an automatic regeneration control I became aware that the detector valve normally operated at around 20V with the optimum regeneration setting. Later, I built a car radio based on this circuit, and thought it would be interesting to see what happened if I tried 12V B+. So, I unplugged the vibrator and connected a clip lead from the 12V supply to the second B+ filter capacitor which normally had 150V across it. Not only did sounds issue forth, but I was surprised at the performance given the low voltage. It wasn't very loud, but all the usual stations could be tuned in. The valves used were 6BL8 for the detector and 6BM8 for the audio; these being ordinary TV valves.
Over subsequent years I tried experimental regenerative detector circuits with 12V high tension with good results. It was clear that they had the same sensitivity as their mains operated counterparts, but with lower output. Eventually, along came the Kosmos Radiomann with its 12V high tension. It was my previous experience with low voltage operation that made me aware that all was not well with the design of this set, which I discuss in detail in that article.
Low voltage one or two valve regenerative sets are actually not unusual and circuits have occasionally appeared in various publications over the years. Commonly, an audio output valve is used, such as 1Q5 with something like 9 or 13.5V B+. An audio output valve is used as it passes somewhat greater current at the lower voltage. In fact, such a receiver will provide similar volume into its headphones as a conventional set operating at 45 or 90V.
Another method was to use space charge techniques. The type 49 valve is an example of such a valve used here. It is classified as a "dual grid" valve. The problem of course in using mains valves on something like 9V is weak electron flow. But, by applying a positive bias to the grid closest to the cathode, and increased electron flow can be forced to occur. The second grid is used as a control grid in the usual way. The 49 was especially suited for this service, but other circuits did appear using pentodes such as 6C6 or 6SJ7 in a kind of faux space charge mode operating at 6V on the plate. Apparently, the heaters must be run at a reduced (and critical) voltage. I've not had success with these latter designs yet.

Hybrid Car Radio Valves.
With the advent of power transistors in the late 1950's, a new car radio design appeared where valves were operated at 12V high tension but because of the low output, a power transistor (typically a 2N301 or OC26) was used to drive the speaker at the usual 2 or 3W. The advantage of this new design was elimination of the vibrator and power transformer, making the set more compact, and also reducing power consumption. The valves in the front end (RF, converter, IF, detector) were essentially their 250V mains counterparts, carefully constructed or selected to provide consistent predictable results, and given new type numbers. Even though the plate current is a milliamp or less for each stage, that's still enough to function. Gain is not reduced to the extent that one may think.
In many of these sets, a special valve of the space charge type is used to develop sufficient power to drive the output transistor on its own. Best known of this type is the 12K5. It is unlike any mains valve. Other sets continue on with the conventionally constructed valves but use two or more transistors preceding the output transistor to obtain sufficient drive.
I discuss the 12V hybrid valves here.

It started as a joke.
With all the misinformation on the internet, usually coming from those who have "just discovered" valves, I was becoming somewhat annoyed with comments that you need to use space charge valves if you want to use a low voltage supply, or that all the valves in a hybrid car radio are of the space charge type. Knowing full well that mains valves work with 12V high tension and that the supposed "hybrid valves" were in fact quite normal in construction (with the exception of a few space charge audio driver types), I thought I'd prove the naysayers wrong. Something I must say I enjoy doing...people who stick to the same old circuits and won't think outside the square. As regenerative receivers operating at 12V aren't that unusual, the project I decided on would be a normal Medium Wave superhet receiver. It would use the same circuit as a standard mains design of the kind that was made in the millions during the 1950's and 60's, with normal mains valves. Nothing weird that the naysayers could claim was cheating. As it turned out, not only did the receiver work, the performance exceeded by far even my own expectations. Instead of just proving the idea works, this receiver has better sensitivity than many other superhets.

A typical superhet.
In Australia during the 1950's -60's a typical valve radio running off the mains used one of two valve lineups, or a combination of both. Those using Philips valves would use 6AN7 as the triode hexode frequency converter, 6N8 as the IF amplifier and detector,  6M5 as the audio output, and 6V4 as rectifier. Because the sensitivity of the 6M5 is almost twice that of a 6V6, some sets did not bother with a preceding audio stage. Other variations of this lineup used a 6BH5 as the IF amplifier, with a 6BD7 for the detector and first audio stage.
From the opposition came the AWA choice of valves; 6BE6 frequency converter, 6BA6 IF amplifier, 6AV6 detector and first audio, 6AQ5 as audio output and 6X4 as rectifier. As AWA was the Australian associate of RCA, their designs tended to follow American practice. Note that all these valves have 6.3V heaters. Radios with series heaters and B+ derived directly from the mains, as is common in other countries, were no longer made in Australia once DC mains ceased to exist, and even then they were a rarity. They were seen as a dangerous shock hazard by manufacturers and technicians in this country. Thus, most Australian valve radios are fitted with a power transformer.
Until the 1950's, mains operated radios simply used a long wire aerial. Ferrite loopsticks started to appear in some sets at this time. Loop aerials of the type attached to the back of the cabinet as is done with U.S mantel radios were generally only ever used here with portables.

The design of my set.
At the start it was decided to use completely conventional valves in a completely conventional circuit to prove there is nothing unusual being done to obtain the end result.
I had the remains of an AWA 568MA chassis which would provide the oscillator coil,  IF transformers, and tuning condenser. The aerial coil was missing so I simply wound one using the data I used for this set, ignoring the regeneration tap of course. The converter would be a 6BE6 as the oscillator coil I had was designed for this valve. For the IF, I chose 6BA6 as I have a good quantity, likewise 6AV6 was chosen for the 2nd detector and first audio stage. Previous experimentation with audio output valves had shown 6AQ5 to be a poor performer. 6CM5 was the best I'd tried but the high heater current was undesirable for this set (I wanted to run it off my home lighting plant for long periods). A good compromise was found with the 6CW5/EL86. This valve was originally designed by Philips to drive their 800 ohm speakers in an unusual push pull circuit without a speaker transformer. However, a more common use in Australia was as a TV frame output valve. All the valves chosen for use in this set were taken at random from my stock. All were second hand and untested.

Circuit of the 12V Superheterodyne Receiver. Note the circuit uses ordinary mains type valves in the conventional circuit.

The design follows any other typical MW superhet receiver. The 6BE6 pentagrid accepts the incoming 550-1600Kc/s signal and modulates it at the local oscillator frequency. The local oscillator is a conventional Hartley design in the cathode circuit. As the first grid is fed with the local oscillator signal, it modulates the electron stream passing through the 3rd grid which is where the incoming RF is fed in. The  difference frequency is fed via a 455Kc/s double tuned transformer to the IF amplifier stage. This is a 6BA6 pentode which is the variable mu version of the 6AU6. A variable mu pentode should be used where the IF amplifier requires its gain to be adjustable, either by an AGC circuit or a manually adjustable bias used to control the receiver's volume.
The amplified 455Kc/s signal is then fed via another double tuned transformer to a standard diode detector using the diodes of a 6AV6. After filtering and passing through a 1M volume control, the detected audio is fed into the grid of the 6AV6 triode for amplification, prior to being fed into the 6CW5 power pentode. Again, the circuit is conventional with a 10K to 8 ohm transformer feeding low impedance headphones or a loudspeaker.
The DC component of the rectified IF signal is used to provide AGC in the usual way. Up to now, this description would apply to any mains operated valve superhet.
Now, let's look at using this circuit with only 12V.

How to use valves on 12V.
Remembering we're using normal 250V valves, the plate current is going to be very low with only 12V. So first thing is to forget any idea of screen dropping resistors. We aren't going to need them, and they would be a hindrance. Any voltage drop to the plates and screens is to be avoided. What about bias? Again, the plate current is so low on 12V we actually don't need to negatively bias the valves for the purpose of limiting plate current. However, for the purpose of linear signal amplification they still require negative grid voltage. Without bias, the positive going signal would cause grid current to flow resulting in distortion. At 12V the bias voltages are much less than when the valves are run at 250V. In fact, typically a few hundred millivolts. And this voltage is somewhat more critical for correct operation. A volt either way mightn't make a huge difference with a 250V supply, but here with 12V it is the difference between the receiver working well or not at all.
How to obtain grid bias? The logical method is of course to use cathode resistors. But in this circuit such resistors would rob the valves of precious plate and screen voltage, so it is not preferred. We could use a bias battery which would overcome this. Such a battery would last until it basically rots away as no current is drawn from it.
There is an even better way to get our negative grid voltage with no battery or cathode resistors. Because the required voltage is so low, we can use contact bias.
When a valve cathode is heated it emits electrons which are of course negatively charged. The grid being close to the cathode accumulates some of these electrons and thus acquires a negative charge. How great this charge is depends how fast the electrons leak back to the cathode. This is simply determined by a suitable resistor between grid and cathode. The lower the value the faster the electrons leak away, and the lower the negative grid voltage.
The circuit is commonly used with high mu triodes like 6AV6 or 12AX7 in low level audio circuits where a resistor of typically 10M is connected from grid to earth. No other bias components are used.
Contact bias turned out to work perfectly for all valves in this receiver. It is important to use only a high input resistance meter (e.g. a DMM has a typical input resistance of 10M) when attempting to measure the voltage developed from contact bias.

A closer look at the design:
Frequency converter.
The aerial coil is home made, not having a suitable commercially made example to use. Because of aerial loading problems I included a series capacitor for the aerial input. This has been usual procedure with my regenerative sets, as towards the middle of the broadcast band it is difficult to achieve oscillation. As it turned out, with the superhet it isn't required. However it has been found useful for attenuation where the AGC is insufficient. This illustrates one of the advantages of using a superheterodyne circuit; the input tuned circuit does not have a critical effect on receiver performance. Receiver selectivity and gain are largely determined by the IF stage.
The local oscillator circuit is exactly the same as used with a 250V powered receiver. Cathode feedback causes the oscillation. The 47uuF and 22K are the grid leak components and bias the first grid of the 6BE6. In series with the tuning condenser is a 420uuF padder condenser. The value of this is critical to ensure the local oscillator always runs at 455Kc/s above the aerial tuned circuit, from one end of the band to the other. In view of my aerial coil being non adjustable, the padder should be made adjustable instead, but in practice this wasn't necessary. The circuit oscillates strongly from one end of the band to the other with 12V. It was not necessary to increase the feedback or resort to unconventional circuits.
AGC is fed into the 3rd grid of the 6BE6 via the aerial coil in the usual way via the 100K decoupling resistor and .068uF RF bypass.

Under the receiver. The home made aerial coil is at the top right.

IF amplifier.
This is the simplest stage of the receiver with the 6BA6 having no cathode or screen resistors. AGC is fed in via the 470K decoupling resistor. The .22uF functions as an RF bypass for the IF transformer grid winding, as well as a time constant to prevent audio signal decreasing receiver gain.

Detector& 1st audio.
Both diode plates of the 6AV6 are paralleled in view of the simple AGC circuit used. Detected audio appears at the earthy end of the 2nd IF transformer's grid winding. RF is filtered out using the conventional circuit consisting of two 100uuF condensers and a 47K resistor. The filtered audio proceeds to a 1M volume control whereupon the required level comes back to the grid of the 6AV6. The plate resistor is the same as one would use on 250V, being 220K. Any RF that has got through is bypassed via a 330uuF condenser. Initially, I used a 10M grid resistor as one does for 250V, but found the bias was way too high, In fact the plate voltage was up around 11V, resulting in distortion. Reducing the grid resistor to 470K brought the plate voltage down to around 6 for proper class A operation, thus clearing up the sound.

Negative DC also appears at the earthy end of the 2nd IF transformer secondary (because of the diode polarity) which is dependent on signal strength. It is thus used for AGC. Again this is completely conventional. Filtering is achieved by the 470K and .22uF time constant. However, a slight problem arose in that the level of contact bias developed across the 470K resistor is a little too high. The problem was that even with weak signals the 6BA6 and 6BE6 were not operating at full gain. I simply used a delay circuit as used in TV circuits to fix this one. By offsetting the negative voltage by means of a 3.3M resistor to B+, the problem was fixed. The value of this resistor is critical and had to be selected for maximum gain. Alternatively, a 2.2M resistor connected to the wiper of a pot across the 12V supply would provide adjustable delay.
As can be expected, the AGC voltage developed with such low plate current in the IF amplifier is fairly low and control is limited. A long aerial used near strong transmitters could be problematic, although easily overcome with an attenuator. The other option would be to to forego the AGC altogether and have a manual control of the 6BA6 and 6BE6 bias, to function as a volume control, as was done in the 1930's.

Audio output.
As can be imagined, this section had the most thought and experiment put into it. While the RF and low level audio stages are happy with 12V B+, audio from a power output stage is severely limited. Ordinary power valves only pass a few milliamps at 12V. This means power output can never be very high. However, with careful choice of valves quite reasonable results can be obtained. In selecting a suitable type, we need to look for ones that have relatively high current at low plate voltages, such as TV deflection pentodes. For example, 6CM5/EL36 passes 100mA at 100V. It can be seen that 6V6 or 6AQ5 would give poor results as these only pass 45mA at 250V. Television audio valves of the type used in stacked audio/IF circuits such as 12CA5 or 6BF5 have likely looking possibilities also, but these are not common in Australia. The other thing to look for is high heater current. The higher the heater current, the hotter the cathode and the more electrons emitted.
What I decided on using was the 6CW5/EL86. I was going through a box of valves for an unrelated reason and upon spotting it I remembered its characteristics, and thought it worthy of investigation for 12V use. While the 6CM5 is the best "normal" valve that I've tried for 12V audio output work, it does have a high heater current at 1.2A, which is a lot for only 11mW audio power! In view of the receiver being operated for long periods off my 12V home lighting plant, I wanted to keep current consumption under 1A.
The 6CW5 is an excellent compromise with its 760mA heater.  This valve passes 70mA at 170V so looked promising. Indeed it proved to be, providing good sound level in a quiet room with just a 4" speaker.  Optimum load turned out to be 10K and the grid resistor for contact bias, 1M. Output power is only 3.3mW before distortion. While that sounds horrendously low, it is actually enough to cause a definite vibration to be felt on the speaker cone. Naturally, a well baffled 12" speaker would give a somewhat louder sound. Sound through headphones is of course much louder, and can be made uncomfortably so, quite easily.

Top view. Speaker transformer is at the left rear. Filter choke is immediately behind the speaker.

Power supply.
The incoming 12V is fed via a 4A fuse. Although the current consumption is only around 900mA, a 4A fuse allows for the switch on surge when the heaters are cold. Because all the valve heaters are 6.3V they are connected in series parallel. The combined heater currents of the 6BE6, 6BA6 and 6AV6 is 900mA. Thus, 140mA has to be shunted across the 6CW5 when its 760mA heater is connected in series. One could use 12BE6, 12BA6 and 12AV6 where the output valve also had a 12.6V heater and use a conventional parallel circuit.
Entire B+ consumption is only 4mA! Because of noise on my 12V supply coming from various loads, it was necessary to filter the B+. Here, I obtained excellent results using a small iron cored filter choke of the type used in a modern car radio. It measured 5mH, so perhaps an ordinary RFC could be used. A choke was used instead of a resistor to avoid voltage drop. The 220uF provides further filtering and bypassing. If the radio is not used on a noisy supply, these components are not needed. RF bypassing for the front end B+ supply is by a 1uF polyester condenser. Electrolytics are unsuitable for this application. As the valves are all indirectly heated, it is not necessary to filter the heater supply.
No reverse polarity protection is provided. One advantage of valves is that they aren't damaged by reverse polarity. The only components in a valve radio that might be damaged are the electrolytic condensers connected the B+ supply. There is only one in this set, and a non polarised type could have been used if there was a chance of reverse polarity connection. As it is, the receiver is connected by a polarised plug, and even if it was connected incorrectly, the 220uF would not be immediately damaged.

At the rear are the aerial and earth terminals. Note the 12V polarised plug.

The receiver worked immediately upon switch on and brought in all the Sydney stations (about 80km away). This was before I'd even done any alignment! Adjusting the IF transformers and then the trimmer capacitors brought up performance considerably. Finally, adding the 3.3M AGC delay resistor got the receiver sensitive down to the noise level. 2LT in Lithgow and 2BS in Bathurst came in at entertainment quality. After dark, all the usual interstate stations started coming in just like any 240V operated valve superhet. Of course, 2ZB from Wellington (NZ) was also receivable without problems from Sydney's 2KY being only a few channels away.
Initially I had being testing the receiver from a 12V regulated power supply, but upon trying it on the 12V house supply the noise was unbearable. This is a result of several switchmode type power supplies being fed from this source. Fortunately, it wasn't too difficult to filter this out of the supply as previously described.
Because of the aerial coil being non adjustable, gain does fall off towards the low frequency end of the band. The tracking alignment could be improved by using an adjustable padder, but this is not a priority in view of the ample sensitivity.
Weak stations require the volume to be turned full up for speaker listening, while for local stations the audio power output capabilities are exceeded at much lower settings.
AGC is limited, and for strong signals better performance is obtained by connecting the aerial in series with the 390uuF condenser. However, my aerial is about 40m long and provides a very strong signal.

Further options.
As a car receiver the design has possibilities. However, as it is, it would only be suitable for headphone reception and therefore should only be used by passengers. The aerial coupling would also need to be made tighter in view of the shorter aerial used on a car. Several possibilities would exist here, such as feeding the aerial into a tapping on the secondary winding, or even to the entire winding via a small capacitor. The capacitance of the aerial lead in would have to be taken into account when doing this. (This is why AM car radios have a user adjustable trimmer capacitor).
As far as increasing audio power output is concerned, one could parallel another 6CW5, adjusting load impedance and perhaps bias to suit. However, a single 6CM5 will give nearly twice the power output as this arrangement with slightly less heater current.
An interesting option in view of the very low B+ current drawn by the output stage would be to connect one or more 9V batteries in series with the 12V supply to the plate and screen of the 6CW5. A considerable increase in audio power output would be obtainable in this way. One could have the extra batteries able to be switched in as required, and thus they would last a very long time. Of course, bias and load impedance would need to selected for increased voltage operation.
If several watts are desired, then an audio IC such as a TD2002/TDA2003/LM383 could be driven from a cathode follower after the detector. For example, one could replace the 6AV6 with a 12AU7, having one triode connected as a diode for the detector, and the second triode as a cathode follower to drive the IC.
It should also be possible to simply drive a one or two  transistor amplifier from the existing speaker transformer, either from the secondary winding or a tapping on the primary. In this case it might be possible to replace the 6CW5 with a lesser power valve. Note that I have not tried these circuits and they would require some experimentation.