One way to get a mirror-in-the-lid set, without paying the horrendous prices for a pre-war model, is to get a Sparton 4940TV. Fashion changes (and picture tube technology) saw the mirror-in-the-lid design as obsolete by the 1940's, but nevertheless there were a few models available post war, which were also produced in greater quantity. The best known of these is the Sparton 4940TV, produced in 1949. Despite the antiquated styling, the set internally is just like any other post-war U.S. made television set, and even uses the ubiquitous 10BP4 picture tube. This in itself is a good reason to get a post war model. Apart from sets using type 1802 CRT's, pre war sets have the worry of obscure and rare picture tubes, which because of their shape are not easily substituted.
The 4940TV was also available in a larger cabinet containing a radio and gramophone; the model 4900TV.
See my set operating here https://youtu.be/2rX8uyODfwg

The set at the Eli Buk collection auction.

Getting one to Australia.
I had known about the 4940TV for a few years and learned that they were not uncommon, and not hugely expensive. As someone who wanted a mirror-in-the-lid set, and with a particular fondness for American styling, it was the obvious choice. In February 2016, I happened to see one on the U.S. ebay for sale in Florida. Two things caught my attention. Firstly, the set was in immaculate cosmetic condition, and secondly, but more importantly, the seller was quite happy to ship internationally. The seller was a dealer in vintage TV's and had apparently sent sets all over the world. Just out of curiosity, I enquired about shipping to Australia, not really thinking anything would come of it.
He came back and said US$650 for shipping by air, and $50 for packing. The set was $800. So, US$1500 and I'd have a round CRT mirror-in-the-lid set at last. I couldn't pass it up. The dealer gave me references, one from a TV museum in Norway, which I followed up and all seemed good.
I did an internet bank transfer and all went OK. Pity about the exchange rate, because all up it cost about $2300 in Aussie dollars. However, what modern people pay for their "home theatre" stuff is a lot more than this, and I'm getting something a lot better!
The payment was received by the seller on the Monday afternoon, and the set was immediately sent out via a private courier in Florida. It then came via Emirates in Dubai and much to my amazement, I get a call on Friday afternoon saying it's in Sydney. To get a console TV set from the U.S. to Australia in less than a week is pretty amazing. You can tell you live in Australia when the payments don't end there...at this end I had to pay a customs broker for clearance, and then when I picked up the set at the Dnata freight terminal at Mascot, had to pay a "handling fee"; about $150 for a bloke to put it on a forklift, drive it 20m and put in the back of my ute. But at last it was here and I couldn't wait to get out of Sydney and back home to unload it.

It was really well packed. The set was wrapped in plastic and then layers of styrofoam insulation and finally cardboard outer covering. It was all attached to a wooden pallet. Inside the set, foam cushion had been packed in to protect the CRT and other parts. Even the decorative cabinet handles had been covered in tape. I can't imagine a more professional job. The set had arrived in perfect condition.
One thing I had noticed is that most of the valves were unusually clean. When I enquired about the history of the set, I was told it was part of the Eli Buk collection, and had been auctioned off after his death. Looking at it in the auction photos then, the valves are quite dusty, so it would seem that the seller had cleaned them.
You can see the set at the auction here.

Inside the cabinet. The valves without shields had been cleaned. CRT is a Hytron 10BP4. The B+ and EHT power supply chassis is separate to the main TV chassis.

Restoration commences.
It was not until the end of 2016 that I actually started work on the set. It was interesting to see that apart from just four capacitors, everything was original. Not bad for a set made 68 years ago. True, we don't know when it was taken out of use, but for an old valve set that's still very good. Curious, I decided to see if the electrolytic condensers would reform. With a variac feeding the 240-115V step-down transformer, I tried those on the power supply chassis first. With the voltmeter connected across the B+, I knew it was time to raise the input voltage a bit further once the B+ stopped rising. Over couple of hours the capacitors seemed quite OK and could even light a 15W bulb for a couple of seconds. This indicated they had a decent amount of capacitance. There was no warmth from them which indicated the leakage was also low. Chances were the electros in the main chassis would also be good.

Power Supply Chassis.
The power supply chassis is simple with a conventional transformer and 5U4 rectifier. Of course, being American and post war, the supply is designed for 117V operation. Hence the need to use a step-down transformer for Australian 240V mains. Also on this chassis is the RF EHT power supply. Interestingly, of all valves they could have chosen, a 6SN7 with the triodes paralleled is the oscillator. Normally, valves like 6V6, 6Y6, or 6L6 are typical for this application. The RF transformer itself is quite conventional and similar to those used in smaller electrostatic sets. The rectifier is a 1B3 (or 8016 if you prefer the pre RMA type number). Output is 10kV.

Inside the RF EHT power supply.

Main Chassis.
Turning now to the main chassis, the first thing to do was replace most of the paper condensers. A quick check showed none of the resistors had drifted. I wasn't surprised in view of a similarly aged 7" Admiral I once restored about ten years previously. It seems the Americans could make resistors that didn't drift. It's a pity we did not have the same manufacturer in Australia, because high resistors in local sets are just as much of a problem as leaky capacitors. Only having to replace the paper condensers would speed up the restoration. I had it all done over about three days. It was nice to see component leads not wrapped around the tags multiple times, and this made replacement easy.

Chassis in original condition. The gold coloured electrolytic at the top right and three other smaller capacitors in the line oscillator were the only components changed in the set's prior history.

To reform the electrolytics on the main chassis, I used a separate power supply; my BWD 215. I did this because I didn't want to run the valve heaters which would cause extra B+ current draw, and thus be misleading in terms of leakage current. Again, it was a case of raising the voltage in steps each time the leakage current dropped off.

Paper caps have been replaced, except where leakage is not of concern. Astute readers will notice the neon lamp near the centre is illuminated.

Now the set was ready to be powered up. Initially there was no light on the screen because the ion trap was wrongly set, and also because the 15uF electrolytic (C6) associated with the brightness control needed further reforming. Soon, there was quite a good raster, and most importantly, brightness and focus were good, indicating an excellent 10BP4.

U.S. vs. Australian Television Standards.
At this point it's worth briefly mentioning the use of U.S. made TV sets in Australia. The difference in mains voltage is easily dealt with using a commonly available transformer. However, the transmission standards are different, but still close enough. Both systems use negative modulation, FM sound and operate in the VHF band. For 525 line operation in the U.S., the line oscillator operates at 15,750c/s (15,734 c/s for colour transmissions), and the frame oscillator operates at 60c/s (59.94c/s for colour). For 625 line operation in Australia, the line oscillator operates at 15,625c/s. This is so close that usually not even the line hold control needs to be adjusted. However, the frame oscillator needs to operate at 50c/s. This simply is a matter of adjustment of the frame hold control. To compensate for the lower field frequency, the height control then needs to be readjusted. Because the height actually needs to be reduced at the lower frequency, there is no problem in it not having enough adjustment.
So far, all that is needed to produce a 625 line picture on a 525 line set is simply to reset the frame hold and readjust the height (and possibly the linearity).
To receive an off air picture, the differences in channel frequencies also have to be accommodated. It so happened that prior to the analog switch off (Dec 3rd 2013), the Australian capital city channels were receivable by an American tuner with some adjustment of the local oscillator.
The stumbling block with using American sets in Australia comes down to the difference in the video-sound spacing. In the U.S., it's 4.5Mc/s, but in Australia with the wider bandwidth video signal, it's 5.5Mc/s.
In order to receive the sound it is necessary to retune the sound IF to the higher frequency. Sometimes there is enough adjustment, other times the tuning capacitors associated with the IF and detector coils have to be reduced to get the frequency high enough. Once tuned, it was quite easy to use an American set in Australia with the off air channels. It is true that the video IF should be realigned in view of adjacent channel traps and so on, but in practice it usually worked quite well left as is.

However, that is all now history as there are no longer off air transmissions receivable by old sets. Now, the only analog signals come from devices that produce base band audio and video, such as digital tuners, VCR's and DVD players. It therefore makes sense to simply modulate these sources using a system M (U.S. standard) modulator. This will provide the correct video-sound spacing, as well as be compatible with the set's tuner. Invariably, the output of these modulators is on the U.S. channels 3 or 4.
Thus, no RF or IF realignment needs to be done. System M modulators are easily obtainable from the U.S. ebay, and inexpensive. In fact, the postage will cost more!

This is the approach I took with the Sparton. Effectively, the set is now being fed with a system N signal. That is, 625 line 50 field, but with 4.5Mc/s video-sound spacing. This system is actually used in parts of South America.

First Pictures - No Sound.
My first attempts in feeding a picture into the set were disappointing. There seemed to be a lot of ringing in the picture and there was no sound. At one end of fine tuning adjustment a reasonable picture could be obtained, but still no sound. At the other extreme, a weak sound could be heard, but the picture was well out of tune.

One of the first pictures. A non technical viewer might find it acceptable, but there was an obvious problem with the alignment.

First thing I decided to check was the sound IF alignment, seeing that a reasonable picture could be obtained. This was only slightly out and was easy to touch up. However, the problem still remained.
Given this set uses a split-sound IF, the absence of sound could now only mean the tuner's local oscillator was off frequency. Indeed I was able to confirm that all channels had drifted up by about 2Mc/s.
Trying another 6C4 did not correct this, so I can only assume one of the capacitors in the 6C4 circuit has drifted slightly. There was not enough fine tuning adjustment. Chances are if I had started replacing capacitors in the tuner, I'd have to realign it anyway, so I simply readjusted the local oscillator slugs for channels 3 and 4 (the only channels I'll ever use as these are the standard U.S. modulator channels).
Now, we had sound. Problem is the picture was now so off tune as to be unwatchable. There was some buzz in the sound, and I suspected the 5uF electrolytic across the ratio detector output was largely responsible. If this capacitor fails, the limiting feature stops working, and there is no AM rejection. Replacing the capacitor improved things somewhat.

IF Re-alignment.
Obviously, the video IF alignment is what was really out of adjustment to start with. At first I readjusted the IF coil cores to the spot frequencies listed in the service manual. Alas, the picture was even worse. I had also noticed one of the coils (the 22.4Mc/s mixer plate load) didn't seem to peak properly. I went over this procedure a couple of times but without improvement. It was clear I'd have to see what was going on by using a sweep generator. Not surprisingly, the response curve looked pretty awful.
The coil that wouldn't peak had me curious. By placing a ferrite slug into the coil former, I could make the curve change so the coil must have been doing something. Each stage of IF amplification, from the mixer plate onwards is electrically much the same. All the other coils had a definite peak.
To cut a long story short, I could peak the coil at 23.5Mc/s but no lower. It needed to be 22.4Mc/s. It seemed that there was too not enough capacitance. The IF coils are not resonated with an actual capacitor as such, but by stray capacitance present in the chassis wiring and between the valve electrodes. Because it was easy to try, I replaced the 6AG5 mixer, but this didn't do anything. There was the 270pF in series with the next stage so tried that. Still no joy. One thing I had overlooked was the 27.75Mc/s trap coil in series with the mixer plate, but as it had peaked OK and was on a much different frequency I had discounted it as having any effect. As I was running out of options, I bridged it out with a clip lead and tried again. This time the 22.4Mc/s coil peaked! So what was going on? Well, the 27.75Mc/s coil peaked too, so nothing was making sense. That is, until I thought about where it peaked... tuned circuits have two peaks; an inner and an outer. I'd tuned it for the inner peak. When I realised this, (and the manual does point it out) and peaked it on the outer, both it and the 22.4Mc/s coil were correctly tuned.

The picture was still unwatchable and the response curve totally incorrect. As peaking each of the coils to their supposed correct resonant frequency had not achieved anything, I decided to ignore that method of alignment, and simply adjust for the correct curve. For one thing, the bandwidth was too narrow, and I sensed the last stage was actually oscillating. The curve would vary wildly just by putting my hand near the last IF coil without actually touching it.
Sure enough, tuning for a broader bandwidth brought the gain down and everything was stable. It didn't take long to peak things up with the markers in the correct places.

From left to right, markers at 22.7Mc/s, 25.5Mc/s, and 26.5Mc/s (barely visible half way up the left side of the waveform).

Much joy at last when I fed in the RF from the modulator and got a good picture, and even the sound was correctly tuned!
Out of curiosity, I decided to see what the "correct" spot frequencies were for the IF coils, and did indeed find them slightly different to those specified, particularly for the last two coils.

Retrace Blanking.
One thing that had been evident all along when looking at the picture was the presence of retrace lines. It seems that early sets like this rely (or try to rely) on the video signal alone to blank the screen during the field blanking interval, because they do not contain any blanking circuitry. In theory, it should work with a DC coupled or DC restored video stage; the latter being present in this set. In reality it is not completely effective, and blanking effectiveness depends on brightness, contrast, and picture content.
It is however, an easy matter to arrange for proper blanking. All that's needed is a pulse of sufficient amplitude to cut the CRT off during the field blanking interval. Either the grid needs to go negative, or the cathode needs to go positive. So that the blanking circuit does not interfere with the video signal, it needs to be injected into the electrode that does not modulate the CRT with video signal.
In the case of the Sparton, the video is fed into the grid. This means our blanking circuit needs to inject a positive pulse into the cathode.
As it is, the cathode is fed via the brightness voltage and decoupled by a 15uF capacitor. The cathode needs to be decoupled from this capacitor by a resistor; no more than about 100K. Then we need a suitably timed positive pulse. Conveniently, this is available at the 6V6 frame output valve plate. All that now needs to be done is to connect this pulse via a DC isolating capacitor and voltage dividing resistor to the 10BP4 cathode.

Blanking modification for the Sparton.

The circuit worked first go, and with an optimisation of components, the decoupling resistor was 33K, the voltage dividing resistor 1M, and the capacitor .01uF. In view of the high peak voltage, the 1M needs to be 1W, and the capacitor at least 1000V.  I used a 3kV ceramic. Too much blanking voltage will blank out the top part of the picture, and too little, will of course, leave visible retrace lines. A three lug tagstrip was mounted just next to the 6V6 for the additional parts.
At last, the set was ready to go back into the cabinet.

Lid Switch.
The set is turned on and off by raising or lowering the cabinet lid. I had noticed that the mechanism was either out of alignment or missing something because the switch itself did not move with the lid.
A closer look seemed to indicate a pivot might be missing. I had an aluminium spacer which looked like it would make a good substitute, and suitably drilled, it fitted perfectly. The switch now reliably changed positions as the lid was raised or lowered.

Centering Adjustments.
Because of the Earth's magnetic field, it was pointless to check the centering adjustments with the chassis on the bench while the picture tube was horizontal. The adjustment works on the basis of pivoting the focus coil around its axis by means of spring loaded screws. That is to say, the focus coil electromagnet also functions as the centering magnet.
For the test pattern I used my Philips PM5544 .jpg file played in the DVX digital box. I'm not sure if something around the picture tube has residual magnetism, but I could barely get enough adjustment.
Nevertheless, the end result is good. The deflection coils also needed a slight adjustment to the tilt.

Tuner.

This uses a 6BH6 pentode as an RF amplifier. Input is the usual balanced 300 ohm impedance. Pentodes had become obsolete for this purpose by the mid 1950's when the lower noise cascode triode circuit had taken over. The mixer is a 6AG5 pentode, and the local oscillator is a 6C4. By the mid 1950's separate mixer and oscillator valves had been replaced by a single triode pentode specifically designed for the purpose. Well known types include the 6U8 and 6BL8.

Top of tuner shows RF, oscillator, and mixer valves, from left to right. The pulley at the front is for fine tuning via a dial cord mechanism. Underneath, the rack and pinion can be seen. From left to right, the coils are RF, oscillator, and mixer in their separate rows.

The mechanical construction of the tuner is quite interesting. Each set of coils for each channel is independent from the others, so electrically it works like a turret tuner, but instead of separate biscuits, all the coils are mounted on insulating strips which move back and forth over the contacts by means of a rack and pinion. The 4940TV is fitted with U.S. channels 2 to 13, but there is space where the channel 1 coils were fitted when this tuner was used with an earlier model. The IF is 26.25Mc/s for video and 21.75 for sound.

Video IF amplifier.
Four stages using 6BH6 valves are used. Each stage is tuned to a slightly different frequency so that the overall bandwidth is adequate for the video signal. In other words, stagger tuning. Traps are provided for sound and adjacent channel rejection. The sound IF is taken from the cathode of the third 6BH6 and this coil also functions as a sound trap.

The Split Sound and Intercarrier systems.
This is a split sound set, so unlike modern intercarrier designs, the sound IF is that which comes out of the tuner; in this case 21.75Mc/s. Modern sets use the difference between the video and sound frequencies as the IF. This is known as the intercarrier system, and was first implemented around 1949. In effect, the video detector acts as a mixer. The video IF signal performs as the local oscillator, and therefore the sound IF output is simply the difference between the vision and sound carriers. In the case of the U.S. system, this is 4.5Mc/s. It can be seen for this to work that there must be a video signal present, otherwise the sound carrier has nothing to mix with. Therefore, one stipulation with the transmitters (or RF modulators) is that peak white does not cut off the video carrier. If this happens, a prominent buzz at the frame scanning rate is heard in the sound. Of course, the intercarrier system can only work where the sound is FM. Television systems that use AM sound are unsuitable because the video signal is AM also and would modulate the detected audio.
The advantage of the intercarrier system is that it renders set tuning non critical for the user. Furthermore it makes local oscillator construction less critical in terms of frequency drift. The sound IF is always perfectly in tune because it is set by the vision and sound carrier oscillators at the transmitter, which are at the least crystal based. With split sound sets, the set is tuned for best sound, and hopefully this coincides with best picture, if the set alignment is in order. However, as the FM detector must be accurately tuned to avoid distorted sound, it can be seen that any drift will require use of the fine tuning control. This becomes more problematic with higher frequency channels of course.
For the intercarrier system to be successful, good attention must be paid to set design, and anything that distorts the video signal such as overloaded video IF or detector stages can lead to annoying buzz in the sound. Poor AM rejection in the sound IF stages and detector can also lead to problems.

The Sparton sound IF and Audio stages.
There are two sound IF amplifiers using 6AU6 valves. Apart from the power supply which will be discussed later, the circuit is conventional. However, something seldom seen in an FM sound IF amplifier is the application of AGC. The amplified 21.75Mc/s signal feeds a ratio detector using the ubiquitous 6AL5 double diode valve. Ratio detectors have inherent AM rejection which is why there is no separate limiting stage. The negative going output from the 6AL5 provides the AGC voltage which controls the first 6AU6. The AGC is delayed by means of a 2.2M resistor which is effectively connected to +150V.
The 6AT6 diodes prevent the AGC line going positive when there is no signal and thus causing the first 6AU6 to draw excessive current. Delayed AGC circuits are used to prevent loss of gain with weak signals. Here, the signal strength has to be strong enough to overcome the current flowing through R87 (2.2M) before the 6AU6 (V13)  gain is throttled back. If the AGC is not delayed, there will be a negative voltage produced from atmospheric and circuit induced noise. Thus the 6AU6 would be throttled back already with weak signals which is just when it needs to operate at full gain.

From the ratio detector, the audio signal passes through the volume control potentiometer and into the triode of a 6AT6, and thence a 6V6 beam tetrode in the usual way. There is a top cut tone control in the 6AT6 plate circuit using a 1M pot and 5100pF capacitor. The volume control has a loudness circuit by means of a tapping on the volume control and a .03uF capacitor. This enhances the bass response at low volume. The loudness control is taken out of circuit when the tone is at the maximum treble setting. Here, the .03uF (C67) is shorted out.
What is surprising is that despite this extra attention to the tone control, the 6V6 is operating without negative feedback. While I have not attempted to measure the performance of the audio stage, it could simply be that the tone control has been designed to accommodate the frequency response of the 6V6 operating with no feedback. That still doesn't reduce distortion, however.

Video Detectors and Amplifiers.
Another unusual aspect of this set is that there are two separate video detectors which feed two independent video amplifier stages. Starting with the more conventional stage, there is a 1N34 diode for the detector, which then feeds the demodulated video signal into a 6AU6 operating as the first video amplifier. This then feeds a 6AK6 as the video output. Two video stages are required when fed from a negative going signal to retain the correct video polarity when the signal is fed into the picture tube grid. This circuit is completely conventional, and has the usual compensating chokes in the plate circuits of each valve. The contrast control is a 3K cathode rheostat for the 6AU6 and provides variable degeneration in the usual way. It is frequency compensated with a choke and capacitor to retain picture sharpness as the gain is reduced. With the exception of the video detector to 6AU6 grid, the video amplifier is all AC coupled, and this includes the connection to the picture tube grid. However, there is DC restoration applied at this point.

The second video detector is half of a 6AL5, and interestingly provides an inverted video signal. This feeds an "Auxiliary Video" amplifier consisting of a 6AU6. Its purpose is solely to drive the AGC and sync circuits. It is curious why the other half of the 6AL5 was not used instead of the crystal diode for the main video amplifier. As the 6AU6 is concerned with sync information only, it does not need the frequency compensation required as when driving a picture tube.

Video AGC.
This operates using a peak level detector using one half of a 6AL5 diode, fed from the auxiliary video amplifier. Here, C53 (100mmF) charges to a negative voltage representative of the sync pulse tips. The sync pulse level, of course, remains constant regardless of the picture information, so is an ideal reference for AGC circuits. R63 (2.2M) and C52 (.1uF) filter out the video signal, creating a clean DC voltage to control the RF amplifier, and first and second video IF amplifier valves.

Sync Separator & Amplifier.
Here's another interesting piece of design; a diode sync separator! Often seen in pre-war sets, it soon faded into obscurity, to be replaced with triodes or pentodes operating with no bias and low plate and screen voltages. Yet, the technique is perfectly logical and viable. In fact I used a diode sync separator for my 2" TV set.
The video signal is fed into another 6AL5 diode much like the AGC peak detector. The coupling capacitor, C54 (.25uF) charges to the peak video voltage through the diode and R67 (10K). Thus, the 10K has a voltage across it proportional to the .25uF's charge current. Because of the time constant of the .25uF and R66 (220K), capacitor charge current only flows at the peak voltage, in other words, where the sync pulses are. Thus, at the 6AL5 plate we have a clipped video signal with only the sync pulses present.
The recovered sync pulses are then fed into a two stage amplifier based around a 6J6 twin triode. It's an interesting choice of valve given that it was designed for VHF work, but there's no reason not to use it. Both triodes operate with no bias and therefore provide further clipping of the sync pulses to ensure no picture signal remains. The amplified and clean pulses emerge from pin 1 of the 6J6.
There is a connection between the sync amplifier and the DC restorer which is not entirely clear as to its purpose. It is suggestive of a blanking circuit in which the CRT grid is driven more negatively during the sync pulses.

DC Restorer.
Common in U.S. designed sets until the 1950's was the diode DC restorer. If the video signal remains AC coupled all the way to the picture tube, the average brightness remains the same regardless of scene content. It is not possible for the viewed picture to show a true black level. It is certainly cheaper and technically more convenient to make television sets with AC coupled video amplifiers, and many sets have been made this way. The typical non technical viewer is oblivious to the loss of true black level, being distracted by the program content.
One can DC couple the whole system from the video detector right through to the picture tube, but any changes in operating characteristics of the video amplifier will alter the black level, as will signal strength changes, unless the AGC circuit is particularly good (i.e. complicated/more costly). Alternatively, the video circuit can remain AC coupled, but the DC component restored at the picture tube.
Given that the sync pulses remain fixed to the black level, the operating point of the picture tube can be referenced to the sync pulse tips. In a typical diode DC restorer circuit, including the Sparton, the minimum picture tube grid voltage is always that of the peak sync pulse voltage, as obtained by C51 (.1uF) charging through V11A.

Frame Timebase.
This part of the set is largely conventional, and would not look out of place in sets made 10 years later.

A blocking oscillator is used based around a 6SN7 triode which drives a triode connected 6V6 in the usual manner. Sync pulses are fed in via the low pass filter (integrator) consisting of R115-R117 and their associated capacitors. This filters out the higher frequency line sync pulses.
Triode frame output stages are much easier to obtain good linearity from than those using pentodes, which require complex feedback networks. The trade-off is the triode stage is less efficient, but this is not important with low angle deflection picture tubes like the 10BP4. In the 1950's purpose designed triodes were developed for frame output stages, such as the 12BH7, 6S4, and 6CM7.
There are two interesting departures from 'modern' practice in this circuit. Firstly, the oscillator plate voltage comes from the ordinary B+ that drives the output stage rather than a higher 'boost' voltage.
Secondly, the method of height control is simply a 10K wirewound rheostat in series with the B+ to both the oscillator and output stages.

Line Timebase.
Of all the sections of the set, this is undoubtedly the most unusual. Like the diode sync separator, the line output stage in the Sparton is a throw-back to pre-war methods.

The AFC circuit and line oscillator are actually quite conventional and not out of place in a modern set. The only exception is there is no tuned circuit in the line oscillator plate circuit. Such a coil/capacitor combination is tuned to the line frequency and termed a "stability" or "sine wave" coil, and basically improves the frequency stability of the stage. Evidently, with other prewar aspects evident in the set's design, they thought a direct sync line oscillator might be taking things too far, and so AFC using a 6AL5 is included. This compares the frequency and phase of the line oscillator to the incoming sync pulses. A positive or negative going correction voltage is developed across C90 (.1uF) which lowers or raises the oscillation frequency until it is the same as the sync pulses. A rather strange inclusion is a neon lamp (M1), which supposedly functions as a regulator for the line oscillator. It is unclear how it can regulate effectively when there is a 330K resistor in series with it.

Turning now to the line output stage, we have an ordinary audio output valve, a 6L6, functioning in this position. By 1949, a 'proper' line output valve, the 6BG6, had been developed and was widespread.
Valves like 6L6 and 807 were certainly not uncommon in prewar electromagnetically deflected sets in this position. The use of a 5V4 rectifier valve as a damper diode is not however unusual in post-war sets. The 5V4 was popular for this application because it has a suitably low plate resistance. However, it does require a well insulated 5V heater winding either on the normal power transformer, or as here, a separate 6.3V to 5V isolating transformer with low inter winding capacity. By the mid 1950's, valve technology had produced a diode with low plate resistance but with a well insulated cathode that could withstand the necessary high voltage peaks and run off the normal 6.3V heater circuit. This was the beginning of the 6AX4 and related types.

One obvious omission is a line output transformer. There isn't one. Instead, the line deflection coils are choke coupled to the 6L6 plate by T6. The width control is simply a 5K rheostat in series with this.
Because of the negative and positive B+ power supply distribution in this set, not all the 6L6 plate current comes via T6 as first thought. Most actually comes via the 5V4, so the 5K rheostat is not as stressed as it may first appear.

Power Supplies.
Not uncommon for the time is an RF EHT supply. The normal B+ power supply is conventional with a transformer and 5U4 rectifier.

The set operates with a split power supply. That is, both negative and positive rails are used. Parts which require the most voltage are connected between the positive and negative rails, whereas sections like the IF stages are connected between either the positive rail and earth, or the negative rail and earth. Going back to the sound IF stage, it will be noticed that the plates and screens are connected to earth. This is because the cathodes are fed from the B- rail. The B- is -140V and the B+ is 180V. This technique allows for efficient voltage distribution. For example, it is wasteful to operate a 180V IF amplifier off a 320V supply via a resistor when other sections can be used as a 'dropper resistor' instead.
The entire set's current also flows through the focus coil. This has a variable shunt across it to set the exact coil current for optimum focus.
I am not keen on the absence of a fuse or circuit breaker and may add one.

While the use of an RF EHT supply was not unusual in 1949, the use of a 6SN7 as its oscillator does tend to stand out. Both triodes are paralleled. The circuit is conventional otherwise, and I discuss RF EHT supplies in more detail here. Output is 10kV.

Of course, for Australian use, it is necessary to use a 240V to 120V step-down transformer. I use one of my 300VA toroidals for this set.