R206 MkI Electrical Refurbishment

 I was finally happy with the mechanical overhaul. There was almost no wear but all the bearings were seized and took a lot of persuading to free up. The next step was to apply power and see what happened.

For those that are interested the circuit diagram, is given below in three parts, also the complete user manual:

The RF front end

The IF amplifier and the remainder of the circuit

Plus the power supply which is a seperate box weighing in at 40 pounds.

See pictures of the R206 Power Supply

And the complete manual

The first thing to do was to find a new ARTH2 which had a broken top and to see if I could repair the adjacent ARP34 whose top cap was detached and wedged in the top cap clip. I found an old ARTH2 which seemed to be OK then tackled the ARP34, or to be precise the CV1053, which had been substituted sometime in the past and looked pristine apart from the damage. This was straightforward.

See how to repair a radio valve

 Well, with a full complement of valves inserted, powering the set proved something dire was wrong. What it was I couldn't tell, but there was very little to be heard in the loudspeaker. Removing the top cap from the output valve and finger testing proved the output transformer was not open circuit and removing the top cap from the LF amplifier proved that was working, but in neither case was much heard, just a mediocre hum. Next, I applied a 465Kc/s signal at the aerial input but, despite the input being a whole volt I heard absolutely nothing. Moving the signal into the two IF amplifiers showed nothing was getting through.. in fact I could hear only a weak tone from the loudspeaker when the 465Kc/s signal was connected to the detector.

Maybe there's an open circuit resistor or a short-circuit condenser? I measured all the resistors in sight. All measured with some resistance and most weren't too bad. There were well over a dozen rough-looking 0.1uF 350vw condensers (some associated with wax blobs stuck to the chassis which was slightly odd as this can only have happened with the set upside down). As a trial I removed all the components from the 1st IF amplifier. The cathode resistor of 390 ohms was 405 ohms, the 10kohm screen resistor measured 12kohms and the bias resistor of 100kohm was 110kohm. All serviceable but I fitted new ones. There were three decoupling condensers and all measured just over a megohm and around 0.3 to 0.8uF.. pretty strange. I fitted three new 0,22uF 400vw condensers. This made absolutely no difference, so it's something else. Maybe one of the IF transformers is open circuit, although I'm pretty certain the LF amplifier is deaf also?

In line with most WW2 radio equipment the components are coded with a common number for their value and a letter for the specific part, so R19B is the second example of a 10Kohm 0.5 watt resistor.

See the condenser/resistor listing for the R206 MkI

   
   These views show the original parts for the first IF amplifier (top left), the panel with the parts removed and left, the new parts fitted. The original 0.1uF condensers (C21E, C21B, C21D) are replaced with 0.22uF 400vw condensers because I have a large box of these. There are three resistors: the 390 ohm cathode bias resistor (R5B) is visible, the 100Kohm grid bias resistor (R6B) between the top two condensers and the 10Kohm screen feed resistor (R13A) above the bottom condenser.

 The next day, as there was a lull in customer's repairs, I tackled the R206 again. I started by repeating the previous days test which was to input a 1 volt signal at 465Kc/s to various points in the circuit, but this time using an oscilloscope to see the amplitude of the demodulated tone. I soon found that I could reduce the input to around 50mV and still see some recovered audio. I was able at this point to check the tuning of the IF transformers and filters. Tuning was a bit woolly and I found a strange intermittent at IFT3. This wasn't significant but varied the output a little. I also tried changing various valves, but to no avail.

During further testing I discovered the volume control seemed to be open circuit. I removed it and opened it up and found the track wasn't connected to the brass rivets to which the solder tags were fastened. I found a suitable Radiospares pot of 1Mohm and fitted this. Now I can actually hear a tone in the loudspeaker and finger tests produced much louder results, but the receiver was still very deaf.

Below is the faulty 1Mohm volume control lacking any continuity between the carbon track and brass studs.
   

 I found the second IF valve anode and screen were low, sitting around 60 volts or so with the valve unplugged, whilst the HT at IF1 was around 200 volts. I decided to change the same parts as I'd done at IF1. The only really bad component was the centre condenser. I fitted new parts for IF2 just like those for IF1 and tried the tests again. The output signal was much better but still very weak, however I could now hear a tone in the loudspeaker which could beat with the BFO. This is a big improvement and to verify this I reduced the input down to around 20mV.

Below is the panel carrying the new second IF stage components, similar to those fitted for the first IF stage. These are R5C, C21I, R6D, C21H, R13B, and C21F.

 
  I removed the side panels from IF transformers 2 and 3 and was able to check continuity of the four coils. All were OK, but it was during further testing that I found one or two of the trimmer adjusting screws were sitting at around 100 volts or so. I didn't pay much attention to this until later when, on checking the circuit diagram, found that this shouldn't have been possible.

 

Above is a view of the inside of the second IF transformer. At this point I hadn't checked the third transformer on its right. The design is a little odd in that earth returns are carried to a single point under the chassis to eliminate earth currents which might result in instability. Top right is the fixed trimming condenser for the anode circuit of the first IF amplifier valve. This part in the adjacent transformer was leaky resulting in HT being present at the rotor adjusting screw.

Above, I accidentally connected IF3 preset trimmer rotor to ground and the set burst into life. At least the loudspeaker suddenly resounded with interference noises. Up until that point the speaker had only emitted a quietish tone from the incoming 465Kc/s very large test signal. I connected a lead from the rotor to chassis.

Turning to Range 5, I connected a long wire aerial and I found several stations at listenable volume. I also noticed that someone had been adjusting the coils in the drum as RF3 had been broken and a saw cut now replaced the original slots. Probably the original owner back in the late 50s? I'd also found the volume control had been resoldered and if you look at the original pictures, the knob was missing. Was this fault the last straw resulting in the set having been retired from service?

What now? Well, the trimmer rotors should defintely not be carrying HT and the wire from IF3 secondary rotor should be carrying something close to earth. In fact it measured open circuit. Time to study the circuit diagram and ponder the reason for the anomalies, which hopefully will also explain why the set receives only signals of tens of millivolts instead of tens of microvolts!

 

 My first thought was to study the circuitry around the IF amplifiers in some detail. The R206 uses a pair of crystal filters (SO1A and SO2A above) to assist in digging out weak signals. One has a nominal bandwidth of 2.5Kc/s and the second around 700c/s. Both these and the wide setting, described as having a bandwidth of 8Kc/s, can be used with the switchable 900c/s audio filter.

Ideally, the IF amplifier should be adjusted to its centre frequency of 465Kc/s with the narrowest filter in position. In practice, unless the narrow filter has been adjusted, the centre frequency of the IF amplifier should be tuned to the actual centre frequency of the 700c/s filter which should be pretty close to 465Kc/s. You'll see from the circuit diagram, that the primary and secondary of each IF transformer is tuned independently, however there is a filter switch setting which varies the primary tuning of the first two IF transformers. Clearly the tuning of these circuits will interact with the filter switch setting and it's pretty obvious that the designer's intention was to broaden the settings to provide the 8Kc/s pass band.

Once the R206 is working properly a spectrum analyser can be used to adjust the crystal filters and then the IF tuning can be tweaked to exactly 465Kc/s. Unfortunately, this R206 is not working properly so that exercise will have to wait.

What will happen to the set's performance if there's a component that's drifted away from its design value, for example, say one of the low value condensers associated with tuning? As long as the secondaries tune sharply then the primaries should be OK unless their padding condensers have drifted in value. These are C33A and C33C. The effect of drift in value of all the other trimmers and padders should be essentially dealt with by the tuning process. So, all the trimmers should be able to be peaked. This is reminiscent of RA17 filter problems when a silver mica condenser had drifted miles out of spec (click to read this).

What I'm seeing in my testing is the rotor of one or more of the variable trimmers sitting at a high voltage. Assuming this is a DC voltage it can only happen if a fixed trimmer condenser is leaky. C18A, C20A or C19C are the only potential culprits unless of course there is a short between the stator and rotor of C5G, C5I or C5K.

Another fault that could occur is a leaky decoupling condenser. These are C21A, C21C and C21G which can be tested quite easily by removing the associated valve and checking the voltage across the respective feed resistors R2G, R2I and R2H. If there's a voltage across any of these with the valve removed the decoupling condenser has developed a leak.

One more potential problem is the feed condenser to the AGC amplifier, C22A. This condenser if leaky would also draw current through the HT feed resistor R2I. It would also put a positive bias voltage on V2D compromising the effectiveness of the amplified AGC.

Other possibilities are an open circuit padder C33B or C33D or a missing earth connection. The latter is the most common fault in cars and, in order to reduce stray earth currents, the designer commoned all the earth connections in a particular stage to a single chassis connection. If that connection is missing it would nicely explain the R206 deafness.

 

 You can see here the wiring of the IF coil trimmers (all three stages are similar). Unusually, neither of the rotors is at ground potential so that any HT leakage through C19C will make the rotor of C5K sit at a high voltage determined by any leakage through C33E and partly C21G.

In other words C19C and C33E+C21G will act as a resistive potentiometer.

If the leakage through C19C is really bad (as in my case) L26E was so badly detuned that the IF amplifier was acting as an attenuator.

 After studying the circuit diagram I was keen to check out the theories. The first step was to measure the resistance between the two trimmers. Open-circuit was the answer. Then I switched on the set and measured the voltage at the trimmer rotor connected to the anode circuit of the second IF amplifier. It slowly rose to 176 volts. Not so good as the only connection to the rotor is C19C, a 400pF fixed condenser, the other side of which connects to HT.

Why the open circuit was the first thing to determine? Now the R206 weighs not far short of 100 pounds (or 40 kilos) so turning it over is not easy. However, over it went and I could see a wire connecting the two centre pins of the IF transformer. Each of these connects to a trimmer rotor, So I measured the resistance between the two tags connected by the black covered wire. Open circuit... so I poked at the black wire and one end moved... the solder tag was cleanly broken off from the IF transformer pin. I soldered this back in place and heaved the receiver back on its base. Next was C19C, which I unscrewed from the can (it's secured with two 6BA nuts and bolts) and measured it's resistance. It was certainly leaky so I hunted around for a new 400pF condenser. Not that easy as most nowadays are rated at low voltages, however I fished out my box of ancient condensers purchased many moons ago from Greenweld in Southampton... their entire stock of condensers which I bought on Ebay for a few pence (there's a lot... the box weighs around 25 kilos and there are also three drawers full weighing about the same). After 15 minutes I'd turned up two NATO coded 350vw condensers.. a 300pF and a 100pF. Soldering these in place of the old leaky condenser should do the trick.

Could the wire link have been cut by a previous repairer to identify the reason for receiver deafness? Probably not as the solution would have been there to see once the rotors of C5K and C5L had been electrically separated.

As an aside; having worked in the defence industry during most of the cold war period , and seeing how equipments were described in terms of reliability, I was forever mystified why reconciliation of how the reliability of stuff was calculated didn't compare with what was found in practice. It's all very well counting parts and using reliability numbers for these parts to come up with equipment reliability but, in practice, faults were commonly due to bad design which resulted in component failure, or bad wiring such as poor connections from solder or wire-wrap troubles (and of course engineer error). Including a figure for the reliability of joints would have been a good idea, but certainly in Plessey this was never considered.

I turned on the R206 and instead of a healthy background mush... silence.

After puzzling a minute or two I decided to measure the voltages at the anode feed resistors conveniently set in a line at the front of the set for the serviceman. The 100volt stabilised line to the local oscillator was missing... I unplugged the EF50 oscillator and the neon in the PSU lit and the 100 volts had returned. A replacement EF50 worked a treat and the set came back to life.. the R206 was back in business. Next, I'll align the IF strip..

   As the aerial plate is removable and the original was missing I decided to add a PL239 connector and an S-Meter.... at least temporarily.

 The IF strip tuned up nicely at 464.5KHz which seemed to be the centre frequency of the narrowest crystal filter, with all the trimmers at neither fully-out or fully-in position. With sensitivity reasonably good, I plugged in a long wire aerial and tuned to various signals but noting a kind of "rubbery" effect on strong examples. As a strong station was tuned in the strength would suddenly change. I suspect the oscillator is shifting so that its frequency overshoots. This happens when tuning from either direction and prevents one from accurately tuning a signal, unless done extremely slowly.

I guess the best way to figure out what's happening is to set the receiver to a particular frequency, plug in a signal generator, and increase its output level whilst monitoring various voltages. Possibly, a faulty component is resulting in changes to oscillator frequency and working in reverse to an AFC circuit? The R206 uses amplified AGC which is pretty unusual. This means that a strong signal will produce a powerful feedback voltage. Surely the designers would not have been that foolhardy to produce circuitry such that an AGC voltage would alter the oscillator frequency?

Assuming then that the frequency pulling is due to one or more components becoming degraded it remains to determine exactly which...

Well, I carried out the tests and found that the receiver oscillator definitely wasn't pulling due to any electrical reason. This was easy to diagnose because there's a fine tuner which can move the local oscillator a small amount up or down. Although the main tuning knob had the weird tuning affect the fine tuner was OK. I could zero beat a strong station without any trouble using the fine tuner but not the main tuning knob. The problem must be that the tuning condenser mechanism is partly seizing and backlash in the setup is responsible for amplifying the behaviour of the stickiness.

The main tuning has three parts. The slow motion drive, the gear train which connects the slow motion shaft through 90 degrees to the tuning condenser, and the four gang tuning condenser itself. I suspect the problem is in the tuning condenser bearings. Do I remove the whole thing, dismantle it and lubricate the bearings or just try to fix it in-situ?

Problem solved! After removing the two tuning knobs, locking catch and slow motion assembly and using a temporary tuning spindle, and having disabled the drive to the drum, I was able to test the mechanism. Tuning backwards and forwards resulted in the strange tuning characteristic so I applied pressure to the tuning condenser rotor and found the tuning was nearly normal. This was so when applying pressure clockwise or anti-clockwise. I removed the cover to expose the anti-backlash gearing and applied pressure to the pair of gears. The tuning condenser rotor not only rotated in sympathy with the tuning knob but moved ever so slightly along its axis. Because of slight stiffness in the condenser bearings the rotor was moving against its end stops and because of stiffness it was staying put instead of springing back. The end bearing float is adjustable so I was able to slacken the locknut at one end, tighten up the screw and remove the end float.

What was happening was this: Tuning a signal using the BFO and overshooting zero-beat was OK. At this point I turned the tuning knob back towards zero-beat but, instead of rotating the tuning condenser, the rotor moved along its length decreasing the oscillator frequency and instead of moving nearer to zero-beat it moved further from zero-beat. Stopping and tuning in the other direction moved the tuning condenser rotor back to its centre position, increasing the oscillator frequency, then continuing to tune the set would zero-beat the signal. If one overshot again, even slightly, the whole process would be repeated.

Once the end float had been removed tuning a station with the BFO on was perfect. Next, I'll try to unseize whichever of the moving parts is still reluctant to free-up so the tuning flywheel effect is restored.

 
Above is the underside view of the main tuning condenser with its gearbox cover detached so you can see the anti-backlash gears. There are several bearings involved, any of which if seized will affect tuning. Underneath is the worm gear which is adjustable for end float via a long screw fitted with a locknut at the rear of the screening case. There's a plain bearing at either side of the gearbox and an end float adjusting screw at each end of the shaft. I'm slightly surprised that no ball bearings are used anywhere in the construction other than in the slow motion drive, but as probably no-one imagined the war would last long enough for these to be necessary and ball bearings were in short supply, plain bearings were deemed to be OK. The tiny set-screws on the left of the gears are used for adjusting the end stops so that the tuning condenser isn't stressed at range edges.
 
 Above is the RF module detached from the receiver, probably the first time since it was manufactured? I removed it for two reasons. Firstly to check on the tuning condenser gearbox adjustors and secondly to replace any bad capacitors. Sometime in the past a couple of the springs used for the turret have been damaged. One has lost its contact so I soldered a short piece of phosphor bronze relay spring in its place (6th from the right)
 

 Surprisingly the designers did not use any wax condensers, instead sticking to mica types. Possibly this was done to minimise the need to remove the module? If you look carefully you'll see that I've fitted several new resistors.

Above, left to right are.. V1A (EF50), V2A (EF36), V3A (ECH35) and V1B (EF50)

Resistors: R1=4K, R2=1K, R3=470, R4=330, R5=390, R6=220K, R7=100, R8=75K, R9=4.7K, R10=30K, R11=47K, R12=30

Condensers: C1=4 gang tuner, C2=10pF trimmer, C6=0.5pF fine tuner, C14=200pF, C15=0.01uF, C16=50pF, C17=50pF

The handbook contains some errors viz.R12A/B and mostly standard resistor values are used in my set
 
 

 I replaced these resistors which had all gone high in value in varying degrees. None were high enough to prevent the receiver from working but, as changing them was straightforward with the unit detached, it seemed a good idea.

The worst was the 33K (V3A screen) which measured 52K and one of the 470ohm (V1A anode) which read 700ohms. The pair of 4K (a strange value) had each risen to over 5K and were wrapped in plastic tape then a woven sleeve, presumably to insulate them from possible shorts. I used a modern 8.2K and didn't bother refitting the insulation.

The other resistors measured over 20% higher than their marked values.

 Below is a view of the area revealed once the RF module was removed. The long wire connections to the tuning condenser are not ideal but clearly work OK as the receiver's performance is pretty good. Central is the adjuster for the tuning condenser worm gear. I tightened this to the point where free movement was restricted, then backed it off half a turn.

There are two small trimmers connecting to the oscillator and aerial tuning sections of the tuning condenser. These are left in place and did not need detaching.

 

 The main exercise was to investigate stiffness in the tuning system.... so I returned to the various bearings armed with penetrating oil for the nth time. Eventually the stiff bearing freed up. I think it was the one at the front of the tuning condenser gearbox which isn't visible without serious dismantling. I was then able to remove the shim in the slow motion drive (added to increase the output torque) and refit this. After refitting the RF module plus the coil turret and re-engaging the dial drum and putting back the two tuning knobs the tuning assumed a silky feel much like my original set.

I also made a special tool with which to carry out alignment. After hunting around the workshop looking for a non-conductive material half an inch diameter I found an old tooth brush with a handle about 10mm diameter which I cut into shape so it fitted the slot in the coils which is about 5mm wide.

I now have a choice of continuing to discover any more leaky condensers and out of tolerance resistors, or align the front end... Both jobs need to be done.

I thought I'd carry out some alignment to see if anything needed to be done before finishing alignment completely.

Each of the RF coils is housed in a wedge shaped copper box in the turret assembly which is essentially the same as those used in multi-channel TV sets of the 60s. This makes for a superbly stable and precise arrangement but has the disadvantage of making for awkward access for servicing. To give an example. In my first R206 the oscillator coil for Range 6 needed a copper tuning slug to replace the iron dust slug to enable that range to be aligned with the dial markings. The same problem has occurred in this second example in Range 5.

The problem may be the padding condenser. This is the component required to modify the main tuning condenser so that the oscillator tracks the RF stages. In the case of the higher frequency bands the oscillator frequency is pretty close to the incoming signal, but on Range 6 the oscillator tunes 1575Mc/s to 2.685Mc/s to tune 1.11Mc/s to 2.22Mc/s. The difference in frequencies means the padder C12A has to be 360pF and this could have drifted higher in value such that it's beyond the range of adjustment offered by the iron dust core. This means the box has to be removed so a new condenser can be fitted.

 

 RANGE

 1

 2

 3

 4

 5

 6

 OSCILLATOR PADDER

 150pF

 3000pF

 2580pF

 1150pF

 610pF

 360pF
 After removing the ten 6BA screws necessary to detach the metal box carrying the Range 6 oscillator coil, there was the 74 year old 360pF condenser looking pristine. I have a very accurate auto-capacitor tester and this said the 360pF condenser measured 362pF, so back to the drawing board... in fact I aligned all six ranges and all had the same half-inch dial error which meant every range was reading low in frequency. Correcting this meant that two or three of the oscillator tuning cores (and some RF coils) were too far out of their coils. The thought passed my mind that the main tuning condenser had a film of oil covering its plates. This would have occurred when I was trying to unseize its bearings. I may try and introduce something to remove this oil film...

 

 After carrying out a preliminary alignment I air-tested the receiver and was surprised to find it lacked overall gain. Of course it was infinitely better than it was when it first arrived here, but background noise level was almost absent and a check with a signal generator indicated that sensitivity was such that reception fizzled out at around 20 microvolts when I expected half a microvolt. A bonus met during alignment was that several ranges end with exactly double their starting frequency. This means that coils can be adjusted at say 1.1Mc/s and trimmers at 2.2Mc/s using the second harmonic of the signal generator, thus saving time having to reset the signal generator.

Next I tested all the valves. Apart from one EF39 which was 60% the remainder were all excellent. As there are still lots of wax condensers lurking under the chassis plus lots of resistors there's a chance that one or more is responsible for the poor performance...

 Another day passed and my repair of lift boards finished about 2pm so I returned to the R206. I've noticed that although it works really well using a long wire aerial, unplugging this results in complete silence from the loudspeaker even when the volume control is turned up. I'd prefer a background hiss so I proceeded to test and replace parts.

There are several groups of components under the chassis, mostly mounted on tag boards. At the back there's a tag board carrying resistors and condensers for the first IF amplifier so I cut out the condensers and checked them. All three showed serious leakage so I fitted new parts, two 47nF and a new 0.22uF. I use the latter in place of old 0.1uF because I have a large box full of new old stock RS orange things. One resistor was burnt so I fitted a new 4.7K. Sensitivity improved slightly after this, but with valved equipment the old resistors are rarely critical and often leaky condensers are not that important. I'm keeping the old wax-covered condensers because I intend to replace their innards with new high voltage chip capacitors.

Next, I looked at two pillar shaped assemblies carrying resistors and, in the centre of each, a vertically mounted electrolytic condenser. I checked the first and surprisingly found it was perfect, with good capacity and a very low ESR of 0.2ohms. The other seemed equally good. Sitting at the side of the assembly nearest the turret was an ancient Sprague condenser that must have been fitted in the early 60s. As it had a low voltage sitting on its end I unsoldered it with the set turned on. This produced an unexpected result as I hadn't checked to see its function. It was part of a top cut circuit and the audio of the local radio station to which the set was tuned took on a pleasant crispness so I left the condenser unsoldered. Clearly the designer intended to introduce the mellowness found in old domestic radios... The resistors around these pillar assemblies all seemed respectably within tolerance, apart from one 47K which read in the high 50s. There are several more wax covered condensers left so I'll tackle these later.

 

 Another day and, after finishing a lift repair, I changed the remaining wax condensers in the R206. I'm going to investigate all these because they behave very strangely on my capacitor testers. Checking capacity makes the tester sort of flutter in auto mode and display a stupid reading on specific ranges. Leakage measured with an ohm-meter is also odd. See the table below.

I followed a logical sequence in setting the IF amplifier centre frequency. First I set the filter to narrow, then determined the filter centre frequency of 464.65Kc/s then peaked the six trimmers for IFT1, 2 and 3. This resulted in the narrow and wide settings working OK, but the middle 2.5Kc/s setting was rubbish. I guess the internal trimmers or fixed condensers must have drifted making the centre frequency 2Kc/s or so away from 465Kc/s?

Aligning the six wavebands also ended badly. I couldn't reconcile the dial readings with the trimming range offered by the dust cores and trimmers for ranges 4, 5 and 6. In each case I had to remove the dust core from the oscillator coil so either I've missed something or something electrical is wrong. One possibility is the inductance of the coils has increased over a 70 year period. This may be more serious with the multi-layer coils rather than the simple single layer types. In my first R206 I had to use a brass insert to reduce the inductance of the Range 5 oscillator coil in order to get it to track correctly.

 

 The simple solution I used to resolve alignment problems was to substitute for the iron dust core a piece of brass.

The dust core was glued to the bakelite adjusting screw and I found that by gently twisting it I could pull it off. I found a brass fitting about the right diameter and cut this into short lengths and drilled this so it fitted tightly over the bakelite stud.

This worked well and allowed alignment of two of the rogue coils. One, the oscillator for Range 6 and the other an RF coil for Range 4. The oscillator coil for Range 5 aligned perfectly without its adjuster fitted.

I have a suspicion that one or two RF trimmer condensers are at either maximum or minimum settings which suggests reduced receiver performance. This is probably due to drift in the value of fixed condensers. The way to check this is to confirm that each trimmer (none of which has an end stop) has two different settings.

 

 Below the picture of the rogue condensers is a table giving test results for 13 condensers removed from the R206. I measured them with an auto-ranging capacitor tester and leakage was measured using a high voltage power supply and a high impedance digital voltmeter. Leakage 1 (Note 1) is the voltage across the capacitor when fed via a 1 Megohm resistor connected to the HT voltage. Leakage 2 (Note 2) is the voltage between one leg of the capacitor and HT negative with the other leg of the capacitor fed directly from HT positive. Because the voltmeter is a very high impedance digital type you can just about ignore the internal resistance of the old condensers.

For comparison purposes, Sample 14 is a tiny 500V surface-mount chip capacitor and Sample 15 a modern 400 volt RS capacitor.

 

 Sample

 Marked value

 Type

Measured value

 DC resistance ohms

 Leakage 1 (note 1)

 Leakage 2 (note 2)

 1

 0.1uF

 ZA24256

 271nF

 1M

 6.6V

 226V

 2

  0.1uF

 ZA24256

 381nF

 1.5M

 5.4V

 228V

 3

  0.1uF

 ZA24256

 177nF

 800K

 13V

 227V

 4

  0.1uF

 ZA24256

 161nF

 1.7M

 18V

 227V

 5

  0.1uF

 ZA24256

 299nF

 1.1M

 6.6V

 227V

 6

  0.1uF

 ZA24256

 183nF

 730K

 17V

 227V

 7

  0.1uF

 ZA24256

 159nF

 140K

 24V

 228V

 8

  0.1uF

 ZA24256

 96nF to 0

 cycles

 95V

 226V

 9

 0.1uF

 345

 213nF

 2.4M

 5V

 226V

 10

 0.05uF

 ZA24257

 141nF

 4M

 14V

 224V

 11

 0.05uF

 10173

 150nF

 4M

 10V

 222V

 12

 0.05uF

 10173

 124nF

 2.4M

 5V

 225V

 13

 0.05uF

 10173

 125nF

 3M

 25V

 224V

 14

 0.1uF

 chip

 100nF

 greater than 500M

 0V

 0.4V

 15

 0.22uF

 RS 115-281

 220nF

 greater than 500M

 0V

 0.25V

 From the results you can see that none of the old condensers would be suitable for coupling the anode of an LF amplifier and the grid of an output valve although half the samples would be adequate for decoupling purposes. Most had melted wax indicating that they had got pretty warm at some time. Sample 8 presumably has some sort of short circuit.

In all cases the leakage value seems to have no simple relationship to the measured DC resistance.

Read about stuffing...

 
 Above is the receiver now working tolerably well. Look carefully and you may notice the crystal filter modules normally fitted at the back, behind the IF transformers, are missing (see below). I decided to examine these and check alignment. Also, I've fitted a new knob for the volume control. I didn't have a proper replacement so I used a knob from a 19 set and made a skirt from a piece of black plastic (a lid from a jar) using a circular cutter. I'll need to etch and fill some markings later to complete the job.

 

Two plug-in filters are provided. One for CW reception and the other, which provides a 2.5kc/s response, for R/T reception. 

This is the narrow filter with its lid detached. There are are a number of crystals fitted inside the box and these are arranged to provide a CW filter of about 700c/s, meaning a bandpass at the 3dB points of 700c/s. There are three trimmers, as you can see, which are used to define the shape of the filter and these are extremely critical in their adjustment.

I was surprised to find this filter was still well adjusted after 74 years.

I carried out some tweaking and testing using my spectrum analyser but made the mistake later of further twiddling whilst listening to a signal and found it impossible to get it back to its original shape without recourse to the spectrum analyser as changes of a fraction of a pF made a huge difference resulting in multiple responses to the received signal..

Because the filter uses crystals the centre frequency of the filter is determined by their natural characteristics. This means that the IF needs to be set, not to exactly 465kc/s, but to the centre frequency of the narrower filter. In my case about 465.6kc/s. This setting, and filter adjustment, are best checked by disabling the local oscillator which is easily done by inserting a piece of paper between the appropriate stud for the oscillator coil on the turret and its mating spring.

 

 Lurking at the back of the front panel I found the coupling condenser (C21S=0.1uF) that connects the LF amplifier to the output valve. I must admit I hadn't bothered to change this because it doesn't connect directly to the grid of the output valve but nevertheless it was leaky. One end is fed by 95 volts, being the anode voltage of the EBC33, and the other end measured 9 volts. The condenser connects to a small intervalve coupling transformer via a 10K resistor (R14C,which measured 12K). Of course the fact that the condenser was leaky was important because it meant that some direct current was passing through the transformer primary and this reduced the overall audio gain. Disconnecting one end of the condenser and resistor and adding new components hugely increased the audio output. The leakage must have been around 90Kohm and resulted in about 1mA flowing through the transformer primary.

I've noticed also that the filter that can be switched in to improve CW reception wasn't working. I must investigate the reason for this. It uses a number of condensers and a pair of coils to provide a narrow bandpass at around 900c/s. These old condensers not only leak but can be vastly different in capacity to their markings. You'll also notice W1A which works as a limiter by clipping the loudspeaker audio when it's switched in. The switch appears to have no effect.

 

 Another odd thing is the overall gain of the receiver is less when the AVC is switched off and in this setting stations can only be heard over maybe 10 degrees from the maximum setting of the RF gain control. You can see above that V2D is an amplifier which provides a separate path for the AVC setting. V4A diodes provide a detection function as well as dealing with BFO injection and not AVC. Instead V5A is the AVC voltage detector. An obvious candidate for more gain in the AVC setting must be C23B which, if leaky, would provide a positive bias to the AVC line thus counteracting the negative voltage output from the rectified IF signal at the anode of V2D. C22A may also be important but I think this has only a very low capacitance and is not the type of condenser that generally goes leaky. C21Q which is the smoothing condenser for the AVC voltage is a prime candidate for leakyness but would reduce rather than enhance the gain, which is the opposite condition to the anomaly I've noticed.

I think I'm delving into marginal things now that the receiver is working tolerably well. I investigated the shortage of background noise with no aerial using headphones and a digital voltmeter. C23B seems to blameless, One end carries over a hundred volts and the other about minus 0.2 volts with no signal. Poking around with the voltmeter I found something slightly odd. C33E had about plus 0.16 volts present which must mean that C19C is slightly leaky. I'd already changed one of these condensers when I'd discovered HT present on an IF trimmer rotor. I checked and of course Sod's Law dictated that I'd already changed C19C for so-called "new-old-stock". I left this for now because when I checked the receiver's sensitivity wearing headphones and the volume control turned full up I could just hear a 1.2Mc/s test signal of 0.1uV.

What else is there yet remaining to be done on the electrical front?

The audio filter isn't working. I hadn't realised but this filter (L32A, L32B plus five condensers), is contained in a can similar to the three IF transformers. I'll unscrew the front cover and see what's gone wrong. I'd always imagined this can held the BFO coil.

 

 

 The audio filter is a twin T type. Basically the resonant frequency of L32A and C21J is designed to create a high impedance at 900c/s and a low impedance at higher and lower frequencies, then L32B and C21K, which have the same values creates a second high impedance path for 900c/s and further attenuation either side of this frequency. The two sections are linked by C27A which is designed to be a low impedance path for 900c/s. C28A and C28B are also low impedances to 900c/s and provide a measure of matching to the source and output circuits.

Differences in actual values of the components will result in a slightly broader filter in practice (ie. the parts have manufacturing tolerances) than theory predicts.

All a bit hit and miss due to the available components back in 1940, but it works. The actual audio frequency passed by the filter and its tuning width is not too important as long as it provided the operator with a method of listening to CW in noisy conditions. The BFO setting and exact tuning can be set to match the filter output. The frequency might have been partly chosen to suit the natural resonant frequency of the type of headphones in use.

 

Five condensers (three are fitted on the vertical tagboard) and two chokes. Note the method used by the manufacturer to secure the wires carrying input and output signals.

 

The faulty filter switch disconnected for testing

The replacement DPDT toggle switch. Recognise this?

Below the switch was C21S connected to R14C, except this is now a surface mount capacitor inside the old case!

 Sorting out the audio filter turned out to be time consuming but interesting. Detaching the case was simple, just two 4BA screws holding it to the chassis. Firstly, I found the filter switch S3A was faulty and after searching around the workshop I found a very similar WW2 switch which I'd used to modify an ex-MoD power supply. In the interests of authenticity I used this for the R206 and fitted a newer version in the PSU. As I'd removed the filter I decided to check the components. The results are shown below and are dramatically bad considering the worst ones are supposed to be tuning condensers. Roughly speaking, if the filter was designed to be resonant at 900c/s it would be nearer 220c/s. Less of a whistle and more of a groan.

 CONDENSER

VALUE 

 CAPACITANCE
 RESISTANCE

 ERROR

 C28A

 0.02uF 5%

 50nF (0.05uF)

 4M

 250%

 C21J

 0.1uF 5%

 413nF (0.413uF)

 8M

 413%

 C27A

 0.01uF 5%

 179nF (0.179uF)

 500K

 1790%

 C21K

 0.1uF 5%

 337nF (0.337uF)

 1.9M

 337%

 C28B

 0.02uF 5%

 57nF (0.057uF)

 1.1M

 285%

 Finding suitable replacements for the old condensers was tricky as some modern types have wide tolerancing and can be miles different to their marked value. The manufacturers must have had some trouble also because some of the old parts were rated at 750 volts, no doubt these were the only types available with a tight tolerance. Using a capacitor tester I found suitable replacements within the original 5% tolerances and soldered them in place. Next, I tested the filter. My spectrum analyser unfortunately only goes down to 9KHz so is not suitable so I had to use a signal generator and oscilloscope. Initially I found a strong resonance at 1080Hz with a smaller response at around 900Hz. The bandwidth at the 6dB points was 70Hz peaking at 1080Hz (1040 to 1110Hz). By trial and error I found that adding extra capacitance across L32A lowered the stronger response and I finally got a very well defined peak, with no noticeable secondary response at 951Hz with a bandwidth of 106Hz (873 to 979Hz). A quick arithmetic check seems to suggest L32 should have been about 400mH. Maybe it had changed over 75 years?

Looking at the EMER the filter should be resonant at 900c/s plus/minus 80c/s with a bandwidth of between 180 and 300c/s so my end result of 951Hz with a bandwidth of 106Hz is pretty good. Testing the audio filter by listening to morse on 20m proved it performed very well and the receiver stability was excellent. As with most WW2 equipment: Weight=Stability.

 

 The refurbished audio filter with "selected" capacitors shown before refitting to the chassis

 I'm finally near the end of the electrical refurbishment. A third or fourth realignment shows several coils seem to have moved in resonance at the LF end of their ranges, possibly due to drifting fixed trimmer condensers?

As an afterthought I measured the voltages at the rotors of the IF trimmers. The IFT I'd fixed previously stood at around 100mV but would you believe it.. the first IF trimmers were sitting at close to 100 volts and those of the second at around 5 volts. I suppose this will reduce the gain of the IF amplifiers slightly by damping their tuning? Clearly C19C and C20 are both leaky.

pending

Return to Reception