XK Falcon Car Radio (AWA 966-A)

This set came to me for restoration from another early Falcon enthusiast. From the manufacturing date on the radio being 1960, I soon worked out it belonged to an XK, which was the first Australian model of Ford Falcon.
This is the first hybrid car radio to be described on this site. It was manufactured by AWA, and given the model number 966-A.
Hybrid sets were popular in Australia from around 1959 to 1962. They were the 'interim' technology between all-valve and all-transistor sets.
When transistors first became available for consumer use in the late 1950's, they were not entirely suited for an all transistor car radio. The summer temperatures inside a car could be problematic for the heat sensitive germanium transistors used in the RF and IF circuitry.
As a compromise, valves were retained for the front end, and more rugged transistors used for the audio section. Since the front end valves will actually work quite well with only 12V B+, there is now no need for the vibrator power supply. It is impossible to get more than milliwatts of audio power from a valve audio amplifier operating at 12V, hence the need for 200 odd volts and the vibrator power supply previously used.
For a more detailed description of low voltage valve operation, this article describes a full superhet using ordinary valves at 12V.

The AWA 966-A.
This is one of AWA's wedge shaped designs which had its origins around the mid 1950's with the 930-A / 931-A. These were six valve vibrator sets sold under the "Cruiser" or "Cruisermatic" names. The last version was the 945-A / 946-A produced in 1965.
However, a few hybrid models used the same case and chassis. The rear section of the set where the vibrator, transformer, and rectifier once resided were now replaced by a heatsink and transistor output stage.


Rear section of set originally contained a vibrator power supply.

The front end of the set is laid out exactly the same as the all-valve predecessor, with the IF transformers, valve sockets, and tuning unit, in the same position. Essentially, all that has changed is the 12V valve types, and the circuit being fed with only 12V.

The Circuit.


The set described is the Model C version.

Incoming signal is tuned by the stray capacitance of the aerial (around 100pF), C1 and C2, with L3 being the variable element of the circuit.
The set is of course permeability tuned, which was by now standard for car radios. L1 is the ignition filter. It presents a high impedance around 40Mc/s where ignition interference is at its greatest intensity.
The RF amplifier is a 12BL6. This is essentially a carefully manufactured 12BA6. Its plate load is tuned by L4, C6 and C7, and the amplified signal then fed into the signal grid of the 12AD6. This is a converter equivalent to 12BE6. The local oscillator is tuned by L5, L6, C10, C11, and C29.
IF from the 12AD6 plate proceeds to the IF amplifier via T1, a double tuned IF transformer. The IF amplifier is a 12BL6. Plate load is the second IF transformer, T2. The secondary of this feeds one diode of the 12FK6, equivalent to 12AV6, for detection. The other diode generates AVC, and is shunt fed from the 12BL6 plate via C21. AVC feeds the RF amplifier control grid via the filter consisting of R13 and C4. The converter and IF amplifier have a separate AVC filter consisting of R12 and C9. It will be noted that the AVC for these two stages is reduced by the presence of R4. This in conjunction with R12 acts as a voltage divider. With such a low plate voltage, it would be easy to cut off these stages entirely, much sooner than when operated at a more 'normal' B+.
The detected IF is filtered in the normal way by C17, C18, and R8, leaving just the audio component. This feeds a loudness compensated volume control, and then the grid of the 12FK6 triode. R10 is a tone control. At the low setting, the high frequency component of the audio signal is shunted to earth by C23. For the high setting, the loudness circuit is shunted to earth via R9, and C23 is isolated by 1M of resistance. S1 is not included in this version of the 966-A (Model C). It is for the push button tuned models, and shorts out the audio when stations are tuned by the push buttons. The purpose of this is to silence the perhaps undesirable sound, as the receiver is suddenly tuned across several stations when a button is pushed.

Transistor Amplifier.
This consists of a 2N591 driver and 2N301 output transistor. The 12FK6 plate is directly coupled to the 2N591 base, which provides some of the required bias, along with R17. Thus, the conduction of the 12FK6 determines the bias for the 2N591.
Note that as the transistors are PNP types, the collectors are on the earthy side of the circuit (this being because of the negative earth requirement). The 2N591 bias is stabilised by the emitter resistor R19, with some AC feedback via R18. Collector load is driver transformer T3. Since R17 is connected to the collector, it creates some local feedback to help reduce distortion.
The output stage is biassed to 400mA, thus operating in class A. Base bias comes from the voltage divider consisting of R21, R22, and R23. It is stabilised by emitter resistor/choke, R24. This component is shown schematically as a resistor, but is in actual fact a choke, carefully made to be of 1 ohm resistance. Using a choke here provides AC stability. Previous experience shows that a wire wound resistor can be used as a replacement, should this choke be burned up due to a shorted output transistor. Collector load is the choke L9, which shunt feeds the 15R speaker. The choke has a high impedance at audio frequency, and is included so that minimal DC flows through the speaker voice coil. Additionally, if it were not for the choke, there would be loss of collector voltage across the voice coil resistance.
The Model A version uses a more powerful push-pull class B amplifier.

Restoration Begins.
Just out of curiosity I powered the set up as I received it. It was, as the owner said, dead. That is, until I connected the aerial input to my long wire outdoor aerial. Some stations could be heard weakly, but the dial calibration was also way off. I was hoping this would all clear up after the resistor and capacitor replacement.
Three lots of components needed replacing. Electrolytic capacitors, particularly low value types, of the age of those in this set, were now doubtful. There were numerous paper capacitors, of the AEE metalised paper type, which first appeared around 1956. Being paper means they're now leaky to a lesser or greater degree. In the case of transistor circuitry, they could in theory be left in situ since the impedances are low, but in the case of valve circuits, performance is likely to be affected. In any case, what leakage there is will only get worse as time goes on. As this was an expensive repair, where reliability was to be made as good as sensibly possible, the obvious thing is to replace them all. Finally, there are the carbon resistors. Again, these deteriorate over time, especially the high values. Thus, anything over 100k was replaced. Last but not least, was the mica capacitor C33 in the audio feedback circuit. These are now unreliable enough to warrant replacement in audio circuits. They are often the source of intermittent faults, particularly scratching sounds heard from time to time.
And so, after about five hours work, all the relevant parts had been replaced. This is the time consuming tedious part of the job, which I doubt few non-technical people appreciate. Unfortunately, AWA was one of many manufacturers who insist on wrapping component leads right around the tagstrips and valve socket pins. It makes for difficult removal, as the lead has to be unwound while the solder is kept in a melted state. In all my years of electronic construction I've never found it necessary to do this. The solder alone is more than adequate to make a secure mechanical connection. The reason for the leads being twisted around tags is actually to do with the manufacturing. The components were installed first on the assembly line, and then the soldering was done further down by another operator. Thus, it was necessary to make sure the parts were secure mechanically first.

The Challenges Begin.
During the resistor replacement, by chance I measured R5, the 1M 12AD6 signal grid resistor, before replacing it. I was somewhat surprised to see it reading 105k. It is extremely rare for a resistor to reduce value like that, so I started disconnecting other components to find where the low resistance was. In the end, it was found to be internal to the 12AD6. The resistance was between pin 5 (plate) and 7 (signal grid). This is strange because there is a suppressor grid between these two electrodes, which did not show any resistance reading. I simply burned it out using a picture tube reactivator. After that, there was an open circuit between plate and grid, as there should be.
With such a high leakage between pin 5 and 7, the signal grid would have been biassed positive which would explain the extreme lack of sensitivity. What caused the leakage will never be known. Quite possibly it was a fragment of material loose inside.
Anyway, the set could receive a few signals now on the car aerial, but the dial calibration still was way out. There's a station around 1629kHz in my area which was appearing almost half way down the dial. This was a lot more than just a bit of alignment drift.

Trimmer Capacitors.
The sensitivity seemed a lot better when the aerial was connected directly to the converter, bypassing the RF amplifier. This is the opposite of what should be happening. So why was the RF amplifier acting as an attenuator instead of an amplifier?
While pondering this, I thought at least I could get on with one obvious thing that needed repair; the aerial trimmer, C1. Someone had been a bit careless in the past, and the adjustment shaft had a force applied to it which was never intended. The bakelite trimmer casing was broken in several sections. It looked like it could be super glued back together easily enough, so it was extracted and dismantled. However, I still thought the glued join might not survive the rough treatment it was likely to get in the future. Then I had an idea...I noticed that the RF amplifier trimmer, C6, was of the same construction. I could swap the trimmers over, since the more delicate one would be quite OK inside the set. To do this entailed swapping over the control shaft of the aerial trimmer with the screw adjustment of the RF trimmer. It all went to plan and the trimmers were duly replaced. Surprisingly, sensitivity was considerably improved, and the RF amplifier was doing its job.
It would appear there was some foreign matter lodged in between the plates of C6 which fell out when I dismantled it. This would explain why the trimmer wouldn't peak.

Local Oscillator.
Aside from the local oscillator, the only thing that could cause the stations to appear at the wrong dial locations would be if the IF channel was way off frequency. That was worth a check first. It was close to the expected 455Kc/s, although some improvement in gain was had by tweaking all the IF transformers. This is easy to do with a weak 455Kc/s signal fed into pin 7 of the 12AD6, and each IF transformer adjusted for maximum  audio output. Note that IF transformer cores should be adjusted for peaking towards the outer ends of the coil formers, to get the intended coupling characteristic.
Obviously, the local oscillator was still way off. In fact, the lowest frequency the receiver could be tuned to was 600kc/s. It should be 530kc/s.
I checked all the capacitors associated with the local oscillator and found no fault. There was only one thing left to try and that was adjusting L6, which AWA calls the "padder" adjustment. Normally a padder is a series capacitor in the tuned circuit, but because this set uses permeability tuning, a coil is used instead. L6 is in parallel with the local oscillator coil (L5), which is tuned by the user. Adjusting L6 effectively controls the inductance range of L5.
And so, careful adjustment did bring forth the ability to tune down to 530Kc/s. As expected, now that the tracking was better, the sensitivity was improved. As well as L6, C10 has to be adjusted to set the upper limit of reception. In this set it's a trimmer which I believe is of Philips origin. This trimmer capacitor consists of a ceramic tube with one conductor running up the centre. The other connection is to a fine wire wound around the tube. As more wire is wound on, capacitance can be increased, and vice versa. The only problem is adjustment is not easy. The wire can be heated up with a soldering iron and turns removed to reduce capacitance, but to increase it, it's necessary to add turns of wire. In its favour,  however, one can at least be sure there's no screw adjustment to be got at by non technical people, or vibrate loose.
In this case, I had to remove turns to get the upper band limit to 1630Kc/s.


Trimmer is centre of photo. Note the turns of wire removed from ceramic tube.

At last we had correct band coverage, but I had noticed the sensitivity dropped towards the low frequency end of the band.
A variable capacitor connected across the existing aerial trimmer as an experiment could peak up sensitivity. Trouble was, the value of capacitance required was not consistent across the band.
At this point I'd pretty much had enough and needed a break. There was was an obvious tracking problem, and the aerial seemed not to have enough capacitance.
The amount of obscure problems which can occur have made me question the economics of doing this kind of work. To illustrate what I mean, five hours has already gone into just replacing components. Add another 2hrs dealing with the trimmers, and then another couple of hours trying to find out why the local oscillator was out. And now this. Add up the hours plus the cost of parts, and you can work it out.
Truth be known, one could get away with leaving the set as is, if it was used in a strong signal area, but such is not the case with the owner of this one. In view of his location being far from AM transmitters, it had to work down to the noise level.


New resistors and capacitors installed.

Tuner Adjustment.
There was nothing left but to adjust the RF and aerial cores in the tuning unit. AWA gives instructions on how to do this; referring to a 0.56" gap in a slot in the side of the tuning unit, as to being the reference point for the initial setting of the tuner. With the vague alignment instructions, it took a while to work out what they were referring to (another hour in time here), but eventually I found it. Suitably set with vernier calipers, the RF and aerial cores were set to peak up with a 1000Kc/s signal. Fortunately, the oscillator core was spot on - and so it should be!
This at last did the trick. The aerial and RF trimmers peaked as they should, and the gain was consistent across the band. It brought in 2ZB from Wellington, NZ, as it should in the evening. 2GN from Goulburn was receivable late afternoon which is also a good indication.
How the set got so far out of alignment will remain a mystery, and I can't recall anything like it that I've dealt with before.

Audio Output Power.
The owner had queried using 4 or 8 ohm speakers, since 15 ohm types are obsolete. 16 ohm is a good modern equivalent, but sizes are limited. I assured him it wouldn't damage the set to use 8 ohms, but just that the power output would not be as much as it should be. Out of curiosity, I thought I'd measure it and make a comparison between 8 and 15 ohm loads.
And here was another mystery. I don't have the specs for this set, but it does use the same 2N301 output stage as a number of other sets AWA made. One quoted 3W output. The 966A could only just manage almost 1W driving 15 ohms. Quite a reduction in power! But, when I thought about it, there was no way 3W could ever be produced. With a supply of around 11V across the output transistor, the peak to peak voltage can never be higher than this. As there is no transformer step up to the speaker, the speaker will therefore receive this peak to peak voltage. This is what I measured; 10.8Vp-p, or 3.8Vrms. That equates to 972mW into 15 ohms.


Peak to peak voltage into 15 and 8 ohm loads.

I repeated the test with an 8 ohm load and the reduction in output power was severe. Now it was 6.29Vp-p, or 2.22Vrms. That results in the much lower power of 278mW! AWA point out the load impedance is critical, and they're right.
It's an interesting thought to modify the output stage to operate an 8 ohm speaker efficiently. The output choke could be tapped down its winding. Alternatively, the bias could be increased for the output transistor, provided it could still be driven adequately. It might be necessary to change the 2N301 for a more powerful silicon transistor.
If the owner cannot get a 15 (or 16) ohm speaker, I would recommend two 8 ohm speakers in series, correctly phased of course. I doubt a single 8 ohm speaker will give enough volume with the car in motion.

Aerial Socket.

Prior to the Motorola plug being standardised, other types of aerial connector were also in use. Bayonet types were popular, as were microphone connectors. This set uses what looks like an RCA socket. As requested by the owner, an adaptor was made up with an RCA plug connected to a Motorola line socket. I could have simply installed a Motorola socket in place of the original RCA type, but would have detracted from originality. As I discovered, the socket wasn't actually an RCA type at all. It looks like it, and the centre pin fits perfectly, but the diameter of the outer (earth) connection is a fraction smaller. To get a tight connection I soldered a thin piece of brass shim to the inside of the RCA plug.



How Does it Perform?
This being the first hybrid car radio I've done extensive work on, it was interesting to compare to an all valve set.
Having used my 12V superhet for almost 10 years, I can say that low voltage operation does not detract from sensitivity. And so was the case with the 966-A. It's as sensitive as a 'normal' valve car radio.
The sound quality is good. Probably because it's operating in class A, it doesn't have any kind of harsh transistor tone. It's actually quite nice to listen to, and bass response is good - probably helped by the loudness circuit associated with the volume control.

Just for fun, I tried a 12BA6 and 12BE6 in place of the 12BL6 RF amplifier and the 12AD6. No surprises at all when the set 'worked' with these common 250V type valves. The 12BA6 worked quite well, but the 12BE6 less so. Although it oscillated, RF gain was down somewhat. Some optimisation of the bias would be in order, or alternatively trying a quantity of 12BE6's to find one with the right characteristics. The point here is the 12V valves are made so that their curve is uniform at low voltages, but otherwise they are the same as their 250V counterparts. For the 12V superhet, the bias had to be optimised for the particular valves used, and it could well require alteration if any of the valves were to be replaced.
The exception for 12V valves being the 'same' as their 250V counterparts are a few space charge types like the 12K5. These provide sufficient drive to the output transistor. Essentially, by biassing what is normally the control grid positive, this forces a greater current flow to occur. The signal is fed into what would normally be the screen grid. The 966-A does not use this kind of valve; instead another transistor is used.

From the outside, the 966-A works just like any other valve set, warming up in the same amount of time. What is strange is no vibrator buzz - that's probably the most peculiar thing to get used to.
I have to say I didn't have much enthusiasm for hybrid sets previously, but having got this one working well, it's a nice set and works well. The only thing to be wary of is to ensure correct supply polarity - something that isn't important with a non-synchronous vibrator set.




Home