|The first task with the overhaul of any old set is to remove the chassis from the cabinet. Clearly this example had been tackled in the long distant past as in pencil on the inside of the case was "W/CHANGE KNOB PULLS OFF". Applying a large screwdriver proved the knob plus a metal blade came off together after a struggle. I feared the wavechange switch had disintegrated, but no, it was just the way it was made.|
|Also shown is the wavechange mechanism below the dial glass. Just to produce the front panel styling and layout demanded a complicated, and no doubt relatively costly, method of providing the wavechange switch. The metal plate slides left or right and its movement is coupled to the yaxley switch mounted on the front of the chassis. It might have been cheaper to just fit an ordinary knob and aesthetically balance this perhaps with a tone control mounted on the left.|
|Enigmatic message left for me on the inside of the case 50 years ago||
|Pulling off the single remaining knob (the volume control) was easy. The tuning knob was missing. Maybe this had been pulled off so that a potential purchaser wouldn't discover that the sets tuning mechanism was knackered? In fact, a length of ordinary parcel string had been used to rethread the pulleys and this was badly frayed, jammed in the metalwork, and generally useless. I made a careful drawing of the complicated threading to aid dial cord replacement (it had been threaded wrongly I discovered later)..|
Most dial cords don't work properly for long. The way this one works relies on the cord having just the right amount of friction. As the tuning knob is turned the loops of string wind forward or backwards along the spindle. At the same time the loops slip slightly so they don't hit the metalwork at either end of travel. Once the friction is no longer right the loops can pile up and jam, then they slip and tuning stalls. If the cord tension weakens tuning becomes hit and miss. The manufacturer states that new dial cord shall be "good quality fishing line, plaited and waxed", but I'll see what I've got as a reliable non-fishy alternative....
Under the chassis looks clean and I checked all the old carbon resistors which measured sensibly near to their markings. The circuit is pretty standard so I nipped out C22, the 0.02uF audio coupling condenser (see above) and fitted a modern equivalent (actually a pair of 0.015uF 400 volt condensers below the 47Kohm resistor on the left). The old schematic shows C22 as 0.01uF and has either been replaced previously or is a minor design change. The new condensers will keep the 6V6 output valve cool by ensuring its grid remains negative and not inflenced by a leak to HT. The Trader Sheet refers to the metal rectifier as a Westinghouse 15B. This is a half-wave rectifier with 8 fins and is shown below screwed to the chassis front. Being an AC set the rectifier is fed from the mains transformer.
The original output transformer (above.. centre chassis) had failed and has been replaced by an early Radiospares model marked "Output Transformer" together with the word "Choke". As you can see, new mounting holes have been drilled. In fact the transformer combines the functions of HT choke and transformer, not a very common design technique and I'd say impossible to source nowadays.
The HT is smoothed by C24 and C23, both 32uF. These are marked Mar and Feb 1948 and were therefore old stock when they were used or the radio predates the year of 1950 given in the Trader Sheet. Once C22 had been replaced I fitted a temporary mains lead and plugged it into my Variac. Slowly cranking up the voltage brought up one dial lamp and produced a slight hum from the speaker. Because this model uses a metal rectifier for HT the DC voltage will be present in line with the input AC voltage, hence there will be voltage present which is not dependent on warm-up of a rectifier valve. Note C25 which is a tone corrector. This is wired across the transformer primary. The alternative is connecting the tone corrector from anode to ground which is poor practice used by other designers (even the designer of the AR88). Why? Because if the condenser fails you'll lose the output transformer.
Some say that any old set should be turned on progressively to aid any reforming of electrolytic condensers. In fact I used a variac and slowly cranked up the mains voltage over 5 minutes. I did this before dismantling the set to determine if it was worth restoring as an open circuit transformer or broken wavechange switch, might make the exercise a daunting job.
With the chassis on the bench I again powered the set from my variac.
Increasing the input voltage to around normal mains value indicated nothing serious and after waggling the wavechange switch and plugging in an aerial I heard medium, then long wave stations. There was some intermittency which I suspect relates to the wavechange switch which is buried in muck and definitely needs cleaning. As stronger stations were tuned severe distortion crept in so I guess the AVC line is compromised by a leaky condenser or two?
Unplugging the aerial and inserting a few inches of wire proved alignment wasn't too bad as stations can still be received.
The HT voltages at C23 and C24 measured 148 volts and 132 volts respectively at an AC input of 234 volts, and are about 15 to 20% low. Either the smoothing condensers are a bit low or the rectifier has degraded, developing some excess resistance? Note how easy it is to measure these voltages as the condenser tags are sticking up and very accessible... but note they're also easy to touch and get a nasty shock..
Now that the set has proved to be in good order I can proceed with its refurbishment, starting with the dial cord, then as the wavechange switch is extremely mucky that will need cleaning before I check the alignment and tuning. All the valves are serviceable (but would look better after cleaning) but one dial lamp needs replacing.
For the record the resistors measured as follows:-
R1 33K (45K), R2 1M (1.3M), R3 10M (12.M), R4 47K (55K), R5 68 (63), R6 120 (128), R7 27K (32K), R8 18K (18K), R9 1.5M (1.4M), R10 ?, R11 Vol Control, R12 100K (123K), R13 150K (184K), R14 10M (12M), R15 500K (660K), R16 1.5K (?), R17 240 (253), R18 3.3K (?), R19 560 (?)
I couldn't find some of the resistors. Maybe a cost saving exercise eliminated these?
At first sight this model appears to be better than many because the three gang tuning condenser suggests an RF amplifier or preselector however, scrutinising the circuit diagram and the set's wiring shows the end tuning gang is not connected. Probably there were a large number of 3 gang tuning condensers in the Sobell stores or their buyers got a special deal. Another possible reason is that other Sobell models had an RF stage and it was cheaper to buy large numbers of 3 gang rather than a mixture of 2 and 3 gang condensers... we'll never know. Quite a few radios made after WW2 used parts familiar to collectors of WW2 equipments and I suspect the tuning condenser was used in something of that vintage.
The dial glass is removed by detaching the two lamps, clipped either side and sliding upwards together with packing materials. The bulbs are 6.3 volt, 0.3Amp and one was blackened and open circuit.
After preliminary cleaning you can see dating evidence in "B.B.C. LIGHT" programme which started in 1945 and "B.B.C. THIRD" programme which began in 1946.
A word of warning here for budding radio restorers. Any dial glass is usually irreplaceable and the lettering can often be in a delicate condition so cleaning has to be carried out with extreme caution. A water-moistened cotton bud is the best way to remove dirt, soot and probably nicotine, on the lettered side. Some lettering might be in a luminous substance which glows when dial lamps are on, but any radioactive traces will now be pretty weak so will be not much of a risk. Probably nicotine poisoning is the worst one can expect from a 1940s radio. Quite often the dial glass is operated as a light-guide. In this model the dial lamps illuminate the left and right edges so these should be clean.
During the rethreading I spotted tiny lettering on the front of the loudspeaker, "13 Jan 1950".
Dial threading; some advice
A piece of frayed parcel string was present and apparently threaded correctly, but was lumpy and slipped. Anyway, I made a sketch and removed it. There are three small pulleys, one at the left end of the dial and a pair mounted 90 degrees apart on a bracket at the other. There were two turns of string round the driving spindle. One end of the spring is anchored to a peg inside the tuning condenser pulley and the other via a spring to the same peg. Turning the spindle clockwise made the dial pointer go right to left which felt wrong. Also the string passed over the spindle of the wavechange switch, and from the amount of wear (and string fraying) has done for ages. My guess is the string should have been wound round the tuning spindle the other way. This would have reversed the turning direction for pointer movement and probably allowed the string to clear the wavechange spindle.
Lubrication of moving parts will be necessary if things are squeaky or pulleys are seized. Lubricate sparingly to avoid getting oil on the new dial cord.
I cut a new length of cord, making it a few inches longer than the original as I wanted to add an extra turn around the tuning spindle. With the tuning condenser at minimum mesh I knotted one end of the cord to the peg and threaded the new cord around the small pulley, along the back of the dial around the end pully and marked the position of the pointer in biro. I then removed the cord and fastened the pointer at the marked position. To do this properly the position of the cord needs to be flat and to be symetrical so the pointer doesn't tend to tilt. The cord was then re-threaded and the knot at the tuning pully set back on the peg. The cord from the pointer was then threaded under the wavechange switch spindle and wound three times anti-clockwise around the tuning spindle then onto the small pulley and around the periphery of the tuning pulley. At this point the spring must be attached in a position giving slight tension to the cord when anchored to the peg. This might need to be done a couple of times so don't make the knot at the spring too tight initially. With the threading change, the pointer moves to the right when tuning clockwise and is just clear of the wavechange switch. You can see most of the cord in the first chassis picture above or below where the new cord has been fitted.
The final check made on the cord is to check the exact positioning of the small pulleys in respect of the large tuning pulley. When received the cord had slipped off the large pulley. This was partly due to slackness, partly due to fraying but mainly the cord wasn't being directed to the groove in the centre of the large pulley. The latter is moveable along its shaft and wasn't exactly in-line with the pair of small pulleys. In this model the two small pulleys are fixed so that the cord from each is exactly in line and relies on the tuning pulley to be set correctly. Check also that the large pulley runs true and hasn't been bent from rough handling.
Just a tip about making a correction to the pointer position as it's unlikely you'll get it right first time. Once the new cord is in place the position of the pointer might be nearer one end of the scale than the other. This can be corrected very simply. Just slacken the screws securing the tuning pulley to the tuning condenser, set the tuning condenser to minimum mesh and move the pointer to the correct position at the left end of the dial. Temporarily secure the bush and tune to the right end of the dial. If the pointer isn't in the correct position slacken the bush and reset the pointer. When the pointer is equally spaced from the left and right scale ends secure the tuning pulley bush.
If you do this accurately, and no-one has fiddled with the coils in the past, the set might just line up with original factory adjustments. If you don't do this mechanical setting properly after a dial rethreading it's probable that none of the wavelengths marked on the dial will be right and lots of time will be wasted trying to align the coils. After I'd set up the dial edges with the tuning condenser meshing I checked the dial indications.
2000m = 150Kc/s, 1500m = 200Kc/s and 1000m = 300Kc/s were all spot on and noise level was pretty well constant over the band.
200m = 1500Kc/s tuned as 1440Kc/s, 300m = 1000Kc/s tuned as 975Kc/s and 400m = 750Kc/s tuned to 737Kc/s and 600m = 500Kc/s tuned to 500Kc/s.
The indication is that the medium wave oscillator padder is very slightly high in value.
50m = 6Mc/s, 40m = 7.5Mc/s, 30m = 10Mc/s, and 20m = 15Mc/s all tuned correctly.
I also set the signal generator to the IF frequency (which for this set is 470Kc/s) and found the set tuned roughly across 5Kc/s with the centre response about 2Kc/s low at 468Kc/s.
Connecting an aerial showed an even response of the set's range of tuning for each waveband.
As the dial almost corresponded to my Wavetek digital signal generator not much alignment was necessary. I connected the signal generator to the aerial input, set it to 470Kc/s and peaked the IF transformer coils. There are three, two in the IF can and one on the rear of the chassis. All had dust cores with slots and none was seized. I always use an audio power meter connected across the loudspeaker rather than relying on my ears. The meter must be an analogue type, not digital and an old AVO set to AC volts will work. Very slight adjustment brought the set's IF from 468 to 470Kc/s.
Below is one of my audio test meters. This is the Marconi AFN No1.
Medium waves were next. The manfacturer has marked the dial with a red dot to indicate the tuning point at 1.4Mc/s. The trimmer capacitors C30 and C28 were tweaked. As these are quite stiff I used a 4BA nut spinner. In the 511W chassis C30 is accessed through a hole in the chassis. Then, tuning to 500m or 600Kc/s I found the set was already peaked and I could just hear a 2microvolt signal at both ends of the band.
Long wave alignment requires only an adjustment at 170Kc/s (a red dot is provided). This setting was fine already as was the tuning at the LF end. Again I could just hear 2microvolts across the waveband.
Short wave alignment again was unecessary. I checked the band edges and found the set would tune from 5.9Mc/s up to 17.5Mc/s. Sensitivity was really good, however, placing a 50 ohm termination across the input (the Wavetech output) is miles from perfect for true sensitivity measurements of a domestic set, and I needed 50microvolts across the band to check dial accuracy. With a wire aerial having an impedance of hundreds or thousands of ohms, sensitivity should be a microvolt or two.
The last important task was to fit a proper mains cable. This turned out to be a very easy job. The old mains cable had only two conductors but I believe the new cable should have an earth. The old cable went through a grommet in the rear of the chassis and connected to the switch on the back of the volume control, but had been cut off by the last seller to avoid having to test for safety. This is a really barmy thing to do. I always recount a neighbours tale. She bought a TV from a street auction... no mains cable, just three inches of blue/brown wires. She found an old two-conductor mains cable with a 13-amp plug, cut off the "figure of eight" connector, bared the ends of the cables, bared the ends of the TV cables, twisted all four bare ends together, wrapped cellotape around the joint and plugged it in. You can see the short-sightedness of chopping off an old cable.
If you look at the rear of the chassis you'll see an oddly shaped hole. Goodness knows what this was for because pictures of other sets, as well as this example, have just an empty odd-shaped hole. As luck would have it the hole exactly fitted a computer power supply IEC mains connector. There are even M3 holes provided for its fixing, and a hole for anchoring the earth wire. I found a suitable connector, complete with protection varistors, a suppression capacitor and an earth lead with fixing tag and fitted this in a couple of minutes then wired the rear of the connector with short length of blue/brown cable to the volume control switch and plugged in a computer lead.
Before putting the chassis back into the case I removed and cleaned the valves, blew away dust deposits, and found a pair of matching knobs of roughly the right vintage.
Repairing the wavechange knob was necessary to complete the restoration as its design is unique.
The original material is bakelite with a metal blade set into the moulding, but what sort of material can be used to replace the missing part?
My XYL found some unwanted 10mm bamboo knitting needles and after cutting a short length from one of these I found it to be an excellent material for the repair.
It cuts cleanly, whittles without splitting and can be polished with emery cloth. The first step was to glue a piece of tinplate, with a rectangular hole for the blade, along the back of the bakelite. The broken edge was then filed flat and the bamboo part glued in place. To provide some strength I drilled into the bakelite and bamboo then inserted a short length of 22SWG wire into the hole in the bakelite before pushing the parts together and gluing into place. Final shaping will be done when the glue has set, then the bamboo can be stained to the correct colour.
The finished Rolland set with replacement matching knobs and repaired wavechange knob. The cabinet will need only a little remedial work to improve its looks