R208 Refurbishment 

 I needed to make a replacement test probe for my spectrum analyser because I'd mislaid my usual one when recently rearranging the workshop. I wanted to check the alignment of an HQ170 receiver on which I'd been working and whilst waiting for delivery of the parts I looked at my R208. I removed the receiver from its case but before plugging it into the mains supply checked the mains input panel which gives the connections for 100 to 250 volt input. Surprisingly this was set to 210 volts which I suspect was the voltage tapping used by its last military user. Although the set has been in the hands of at least one radio enthusiast since then, pound to a penny no-one ever looked under the cover to check the mains setting. I would imagine that at least one valve is low in emission from being over-run, but nevertheless I plugged it in and checked for life.

At this point I had no inkling of the history of the R208 and assumed it was just an ordinary government surplus receiver designed by experts and fully meeting its design criteria but, after over 70 years, just needing a little tweaking to restore perfect operation. As you read on you'll get a feeling that all was not well back in the early 1940s. It would be easy to redesign the old set but that would ruin its originality so I'll bash on and make the best of it...

See the user manual


 Although the receiver responded with crackles when the wavechange switch was operated there was very little background noise except when the BFO was turned on. I fitted a 20dB attenuator to protect my signal generator and connected it to the aerial terminals. The set was clearly not only a bit deaf, but the local oscillator didn't appear to work fully across each waveband. On the 10-20MHz band I needed 100mV to produce a response at something like 29MHz and at other frequencies in that band I suspect the oscillator wasn't functioning. Maybe a bad 6K8 mixer/oscillator, or a resistor gone high/condenser leaking... or both.

In the band 20-40MHz I could hear rough responses (at least on the lowest band the signal, although weak, was clean), both main and image signals with 10mV input. One of these, I can't say which until I've read the technical spec was stronger but not much stronger than the other. On the 40-60MHz band I could hear a signal at 58MHz with maybe 5mV input, but this needed more input voltage as the frequency was decreased and soon disappeared at below 50MHz.

The frequency coverage (10MHz to 60MHz) of the R208 is a bit peculiar, but could be explained by the fact that many WW2 communications receivers were limited to 20MHz, and a few reached 30MHz, but apart from specialized aircraft receivers, the low VHF band was inaccessible.

Dealing with the wayward local oscillator.... a simple trick to test the local oscillator is to place a wire aerial from another communications receiver near the chassis and so I did this to look for the presence of the R208 local oscillator. By using an S-meter you can quickly see whether the signal is good or bad, (for example a constant level or fading out). I once used this method to diagnose my RA17 extreme deafness. A low emission EF91 VFO was cutting out before the tuning condenser reached full mesh. A replacement valve fixed the RA17. If your monitor receiver doesn't exactly match the band being checked you can tune to a harmonic of the oscillator.

According to the R208 handbook the intermediate frequency is 2MHz and the local oscillator is on the high side of the incoming signal on all three wavebands so, for example, an input of 25MHz requires the local oscillator to be 27MHz putting the image at 29MHz. Both the 25MHz and the 29MHz signals will be heard when the set is tuned to 25MHz and their difference in level will be a function of the tuning and tracking of the RF amplifier coils. So if the mixer is wrongly tuned to 29MHz the image will be received strongly. Equally, one can hear and therefore alter the tuning of the receiver to 25MHz by inadvertently setting the local oscillator to 23MHz. This will result in tracking difficulties.

Before testing the valves I'll first check the voltages at the test panel. These should be close to the values shown in the handbook.


 Here are a couple of sets of measurements. The second test was with a better HT voltage.


 V1A Anode

 V2A Anode

 V2A Cathode

 V1B Cathode

 V1C Cathode

 V1D Cathode

 V2B Anode

 Ideal values








 First Test








 Second Test








 I checked the readings at the test panel (see First Test) and Anode V2B' reading was about 142 volts (= the HT voltage), and varying. No doubt the hum from the speaker was connected with this variation so I wired a 22uF 400v capacitor across the metal-cased condenser C1B and the hum dropped considerably and the HT rose to 175 volts (see Second Test) Clearly the old condenser is past its best. Looking at the circuit diagram there are two similar condensers C1A and C1B acting as reservoir and smoothing respectively for the HT. C1B was bad so where was C1A? This is located underneath the power supply chassis (below) which is fastened to the main chassis by four 2BA screws. To detach the power supply it's convenient to remove first the horizontal bar between the front panel and the rear frame. This done, the power supply can be unscrewed and turned upside down to access the reservoir condenser. I used the trick of connecting the condenser in series with a resistor across a variable HT supply whilst monitoring the current by measuring the voltage across the resistor (in this case 20Kohm). With 200 volts applied the current drawn was initially pretty low but as I watched it rose steadily to about 1mA. Increasing the voltage to 400 increased the current to 4mA. It then very slowly dropped and stuck at 3.7mA. I then tested the smoothing condenser. This gave roughly the same results so leaving them both disconnected, I wired a couple of new 22uF 400v electrolytics as reservoir and smoothing. The HT was then 175 volts, still a bit low.

Here's a view under the power supply chassis.. Many of the parts are for 6 volt battery option using a vibrator. The previous owner had removed the Mains/Battery switch and some of the wiring and the vibrator.


 Under the main chassis there are a further two metal-cased condensers. These are 2uF and rated at 350 volts. I tested these and to my surprise, both were in excellent shape with only tens of microamps leakage at 300 volts. Both measured very close to their marked values of 2uF with an ESR reading a small fraction of an ohm so I left these in place.

The decoupling condensers are pretty substantial 70+ year-old components and quite unusual but swapping these might help reduce the demands on the HT rectifier. I tested the electrolytic condenser at the bottom of the picture (clearly not original 25uF 12vw component) and found it was marked 50uF and measured 51.87uF with an ESR of 0.39 ohms which is remarkable as it will date from 1958.

While I had the chassis upside down I checked all the resistors. All were high by varying amounts, but apart from a couple were not too bad but as none were open circuit or dramatically high I left all intact.


The next test was to investigate what appeared to be the local oscillator dropping out, and sure enough whilst listening on another receiver, the oscillator was giving S9 over about a third of the two higher ranges and for only a quarter of the lowest range. I'd already swapped the two 6K8 valves around so my initial thoughts were that the problem may be low HT or a bad component in the 6K8 oscillator. It's quite possible the previous owner had changed the mains transformer tapping to deal with this?

 I shelved the local oscillator problem for the time being because I couldn't easily read the dial markings because of tarnishing and I needed to read the markings in order to align the set. The Muirhead dial was best removed for this job. It's held in place by a screw securing the drive output to the tuning condenser shaft. I had to use a torch to see the screw head which needs to be turned to a convenient point for slackening. You'll need a long thin screwdriver to access the screw. The dial mechanism has a tab which locates in a metal block but before it can be pulled off the small plastic bandmarker needs to be slackened (one screw is sufficient, allowing it to be turned out of the way of the dial). Once the assembly is off four screws are removed to free the metal dial which can then be cleaned with Brasso. Once cleaned it was easy to see the numbers and gradations.



 The dial, partly cleaned before deciding to remove the metal plate and get rid of all the tarnishing. This meant removing the salmon-coloured finish to produce a shiny blemish-free surface.

At some time it might be a good idea to apply a matt finish to help with readability.


 Whilst sorting out the dial I noticed that the RF gain pot wasn't connected. I swapped over the wire from the centre where it was soldered to the ground wire and connected it to the end of the wiper. This made the receiver totally deaf because someone had not only disconnected the pot but they'd completely removed the resistance wire so the wiper was rubbing on plastic.

It soon became obvious what had happened. The pot must have seized and broken its winding. This may have been shorting to the chassis so had been removed. Why not replace the pot? Well the pot is directly in front of the coils for the RF amplifier so there wasn't room to get it out. Two of the coils had been broken from their mounts in an effort to remove the defunct pot but still there was half an inch too little space so they'd given up.

The solution was to break off the spindle. This is brass rather than steel so was easily broken off with a large pair of pliers. I was then able to extract the remains and fit a replacement. I had a small WW2 US pot small enough to fit in place. The new pot is 5Kohm rather than 2Kohm so I fitted a small resistor to shunt the resistance down to the correct value.





 I'm probably revisiting something that may have happened back in 1958 when Sputnik 1 was in orbit?

The RF gain potentiometer had failed but there wasn't enough room to remove and swap it even though the two coils directly behind it had been forced from their mounts. The securing screws for the coils can only be accessed once the whole RF front end chassis has been unbolted from the main chassis and all of its cable harnesses unsoldered. A daunting task...

Maybe the resistance wire had become jammed in the wiper and shorted to surrounding metalwork so the wire had been painstakingly removed?

The solution was to break off the brass spindle, giving half an inch extra clearance, which enabled the pot to be detached.

Below, the new potentiometer installed and wired up.

It took 60 years to fix.. but what about other problems?


Above is the coil pack. Left to right RF amplifier, Mixer input and local oscillator and top to bottom, 40-60MHz, 20-40MHz and 10-20MHz.

The coilpack above was typical of receivers designed in the 1930s. It's a utilitarian design also found in things like the AR77 but not in newer receivers aiming for better performance and easier servicing such as the R206. As is the case for all superhet receivers it's vital to accurately track the oscillator and RF amplifier stages. The problem is the oscillator is set at 2MHz away from the RF stages but the tuning condenser must peak the coils right across its whole range. To track the tuning beehive trimmers are adjusted at the HF end and the coil slugs at the LF end of each range. This process must be repeated until the receiver's maximum RF amplification is always achieved at any setting of the dial.

I connected a long wire aerial and initially checked the lowest band 10 to 20MHz because this was completely dead, whereas the two highest bands were active over a small dial rotation. I suspected the wavechange switch and after spraying it with switch cleaner and rotating it backwards and forwards a few times the lowest band sprung into life and loads of broadcast stations appeared. The type of switch used in the R208 is "self-cleaning" so any slight tarnishing is scraped off as the switch is turned, but after 70 years a little help is needed hence the use of switch cleaner.

I turned on my signal generator and after 20 minutes twiddling the lowest band was nicely aligned, as was the IF strip at roughly 2MHz. The first IF transformer had an intermittent but as I looked for a means of removing it I noticed its two fixing nuts were very loose. I tightened these with a 4BA nut spinner and found the intermittent had magically gone away.

Now that the dial was clean I could read the dial markings and I was able to see that the oscillator was cutting out at 33MHz and 53MHz on the two higher bands. This might have been a bad 6K8 or a low HT voltage although the latter isn't too bad now that the HT is cleaned up. I removed the RF amplifier and the mixer/oscillator valves and checked the valveholder pins. Anodes and screen voltages were all present, reading 195 volts. As the HT is 175 volts with all valves in place, I suspect the HT rectifier (a full wave bridge) maybe has more resistance than ideal and is adversely affecting the HT rail. This tends to support the reason for the mains selection tapping of 210 volts. I might change the setting to something in-between... say 230 volts rather than 240 volts. Before that I'll swap any leaky decoupling condensers because I imagine the oscillator problem could well be due to poor decoupling which will reduce the gain.

Most of the decoupling condensers are very large black things marked "H.C" and "O.F." with either "41" or "42" (=dates 1941/1942 ?) which, surprisingly, all tested as perfect. I changed a 75Kohm resistor that measured 92Kohms but this didn't fix the local oscillator. I may use an external HT power supply to see if this works... it didn't.

Below, I've indicated the various measurements made (with the HT measured at 182 volts). Not easy because the underside of the valveholder is buried under the coilpack and it was easier to stick a short length of wire into the socket with the 6K8. I tried various examples of the valve and ended up with one that tested 100% on my AVO valve tester. The voltages are what might be expected and suggest insufficient feedback to maintain oscillation. The lowest band works over the whole dial but you can see the local oscillator amplitude is dropping at the LF end (ref. the grid voltage). The cathode bias increases as the oscillator amplitude drops because the triode is drawing more current (going from circa 2.6 volts to 3.1 volts). Also, the triode anode voltage drops as the oscillator amplitude drops ( from 125 volts to about 114 volts and then 81 volts when it's stopped oscillating)


 Waveband & dial setting







 Local oscillator



 Stops at 33MHz


 Stops at 53MHz


 6K8 cathode







 6K8 triode anode







 6K8 grid, grid1Triode







 6K8 anode







When the local oscillator has maximum amplitude the grid voltage is minimum (ie. -2.5 volts) and when it stops the grid voltage increases (ie to +2.7 volts).I disconnected the R208 HT feed at the HT choke and measured the current at 66mA then connected an external HT supply. This was adjusted to provide 250 volts and the receiver drew about 70mA. The internal HT supply provides about 180 volts. The local oscillator did manage to get down a little further with the increased HT but only improving to see 46MHz. At 200 volts this went to 47MHz, but I need to get the oscillator down to at least 42MHz so a further study of the circuit diagram is needed to try and identify a rogue component.  I checked all the resistors around the frequency changer and the condensers (decoupling, grid coupling and feedback) for value plus any leakage and all were fine. Shorting out the cathode resistor slightly increased background noise level but only affected the local oscillator cut off point by a few KHz and made the cut off more abrupt. I suppose the next step is to change the feedback condenser C12A from 150pF to say 330pF and see what effect this has...


 Changing parts in the R208 local oscillator area is tricky because of the position of the valveholder. Its buried beneath wiring and wedged between screening plates and the wavechange switch. As I studied access I noticed someone had been here previously because there was an earth solder tag with solder but no component or wire and the condensers had been cut off and resoldered in place. Also, although the grid blocking condenser and the feedback condenser are both supposed to be identical, one was correct and the other (the grid coupler) was 200pF instead of 150pF. Oddly that was going to be my next step... Instead I fitted a 125pF and put a 330pF in place of the 150pF feedback condenser. Checking the 20-40MHz range proved that this had absolutely no effect on curing the oscillator cut-off problem. Looking at the oscillator on my spectrum analyser however showed something rather odd. The total tuning range of the oscillator was correct (22MHz to 42MHz) although cutting out with the tuning condenser two thirds closed and of course not matching the dial. Maybe the padder is too high? This was accessible so I checked it and found C8A, the 0.002uF condenser measured 1800pF. Thinking about this I reckon even at 1800pF it must be too high (because the full tuning range was achieved with two thirds of the available rotation) so I fitted a handy condenser marked 830pF. To my surprise the oscillator amplitude remained steady right across the band and tuned 21 to 40MHz. After twiddling the slug and trimmer this ended up at 22-42MHz... perfect but very puzzling. Almost as if the wrong tuning condenser is fitted, a wrong coil or it's very degraded in some way. Anyway a new oscillator padder fixed the problem so two ranges are now correct.

Looking at the dial markings the tuning range for the highest band marked 40-60MHz only covers two thirds of the dial. Because of the findings on the 20-40MHz range I should really check the oscillator padder so that the frequency coverage matches the dial markings. The lowest frequency of full dial rotation is indeterminate bcause the dial markings stop at 40MHz. Also, at this stage, although I aligned the 10-20MHz band, I'll recheck this with the spectrum analyser. Because of the padding condenser puzzle on Range 2, I thought I may have confused the main RF input to the image.

Just a recap...that oscillator padder change from 2000pF to 830pF for the 20-40MHz band is really odd. The implication is the tuning condenser has too great a range. Maybe I should check this? Physically, the tuner has three identical sections and the user manual confirms 3 sections each providing 110pF swing. Maybe some calculations are necessary to indicate what's happening? I cleaned the tuning condenser contacts and checked the tightness of all the securing screws in case one was loose and extra inductance was creeping into the equation. Some screws were loose but tightening them didn't have any effect.

After some rough calculations I worked out that the 830pF padder capacitor would work perfectly given around 0.47uH for the oscillator coil and, for the RF amplifier, a coil of about the same value would tune 2MHz lower. The answer must be in the oscillator coil inductance. If this is different to the original, perhaps from ageing (ie. a change in characteristics of its former or the iron dust core) it might explain what's going on. Another explanation, a rather slim possibility, is that the wrong coil was fitted to the oscillator circuit in the factory. But now, I'm happy with Range 2 (albeit puzzled) I'll look next at the 40-60MHz band...


 I must have spent a few hours working on the 40-60MHz band. Using my DSA815 I could see the local oscillator take a dive into the noise at much the same dial position each time I tried something new. I tried a different grid coupling condenser, a different feedback condenser and added extra ground wires but each time, although the amplitude and maximum frequency would alter somewhat, the thing cut out as the dial was rotated lower than around 50MHz. Then I noticed that twiddling the RF amplifier coil and trimmer dramatically affected the amplitude and exact frequency of the loss of local oscillator output. It seemed that at a specific point on the dial the oscillator output was being sucked away.

I looked again at the wiring. Because of the high frequencies involved and the mechanical layout dictated by the physical sizes of the parts, the earth wiring is quite long and, practically speaking, must be part of the tuned circuits. Certainly, in the highest frequency range this extra earth wiring must be a significant part of the tuned circuits. Then again.. perhaps the 6K8 and the immediate electrical circuitry is being asked to do a job beyond its capability. Could this explain the reason for the dial markings being incomplete in this range? Also, it's quite possible that a previous owner has made modifications to the local oscillator wiring?

Maybe the adjacent 20-40MHz oscillator coil is resonant and as the 40-60MHz coil is tuned it passes through a resonance point of the 20-40MHz coil which is disconnected from the tuning condenser? Maybe I should short out the coil and see if this has an effect? It didn't.

Shorting the adjacent coil didn't have any effect on the local oscillator problem because scrutiny of the wavechange switch revealed it was probably a shorting type anyway. This type of switch uses a grounding wiper to deselect the desired coil rather than select the one required. There is however a slim chance that because of the relatively long RF connections in the coilpack some sort of tuned circuit remains even when the unused coils are grounded.

Why all these problems you may wonder? Well the answer is that the 6K8 wasn't designed for use at VHF and to get it to run at 60MHz must have been a challenge for the designers. The coilpack and its connection to the valves is not ideal. For frequencies under 30MHz it's a classical design but, because the coil sizes are relatively small at 60MHz, the wiring has assumed a significant portion of the tuning. Not only that, but the condensers used in the circuit intrinsically include some inductance so swapping these for different types affects the way the oscillator behaves and there's evidence that at least one previous owner has struggled to get the set working so any further work involves undoing previous remedial work that didn't achieve its object.

During testing I noticed the response of the IF strip was strange. I could see the local oscillator signal flanked by main and image responses but the RF responses were doubled up giving the impression of five signals rather than three. Perhaps I should fix this before proceeding...

I checked the IF strip and found that having tuned it by ear it wasn't too bad, but extra adjustments removed the double humping and resulted in the shape below.

I used a sweep of 1.8 to 2.2MHz with my home made probe connected to the final IF transformer output. The results are qualitative rather than quantitative because the probe is very lossy. The curve shows a bandwidth of 80KHz at 30dB down and at the image frequency around 2MHz the IF gain is 40dB down.


 Note that the wavechange switch has extra contacts which are used to modify the IF transformers for the 10-20MHz band.(see circuit diagram). I suppose this was done to give adequate gain in the two higher ranges whilst not too much gain in the lowest range.I guess, with a little extra work the response below for the 10-20MHz band could be improved.

I then carried on with trying to improve the lowest local oscillator frequency in the 40-60MHz band. By experimentation I'd got this down to 45MHz = reception at 43MHz. I'd noticed that that as the dial was tuned lower the oscillator amplitude would drop very suddenly giving the impression that an absorption effect was present although continuing to tune down didn't resurrect the oscillator output. When the cut off point was reached I could back off the tuning slightly then changing the tuning of the mixer would increase the oscillator amplitude back to its normal level. Lowering the oscillator frequency then showed a slight improvement with cut off taking place at a lower frequency than before. The overall impression was the two tuned circuits... oscillator and mixer were crossing over at a point on the dial so that their resonance was coupled through the 6K8. As I'd had trouble getting the tracking right for the middle band 20-40MHz, I wonder if bad tracking between the oscillator and mixer is the reason for the oscillator cutting out. It may be that the problem is being aggravated because the 6K8 characteristics result in significant unwanted coupling at VHF, whereas at lower frequencies the coupling is not important.

For perfect tracking the oscillator coil should track the mixer coil such that it is always exactly 2MHz higher. Given the wrong coil inductance in either the oscillator or the mixer there may be a point on the dial where the two are at the same frequency. At this point the oscillator output is absorbed by the mixer coil. That would certainly account for the abruptness of the oscillator output cutting out and the shift in this changing as the mixer coil is trimmed. A way to prove this might be to lower the Q of the mixer coil? If this works it will be a lot easier to adjust the tracking.

To support this idea I'd noticed that shorting out the 6K8 cathode resistor would increase the oscillator amplitude by maybe 30% but the oscillator cut off effect was even more pronounced.

Above is the final layout of the parts for the two higher local oscillator ranges. The oblong green capacitor measured about 830pF and replaces the 0.002uF original padder. In the centre wired together are two capacitors in parallel, being 100pF and 270pF, because 370pF was the best value for the padder for the highest range. This is in place of the original 1000pF. I had to put back the 3-30pF trimmer when I found the 2-9pF trimmer didn't allow the dial to match the lowest frequency. I had to fit a different iron dust core when I found neither the one fitted nor the correct one which I found in another coil worked well enough to maintain oscillation. I also had to raise the frequency of the oscillator by a couple of MHz by adding the extra copper wire. The tuned circuit was formed, not only by the coil but by connecting wires and, to some extent the surrounding metalwork. The final result was that the dial markings of 50MHz and 60MHz now correspond exactly to local oscillator frequencies of 52MHz and 62MHz. The oscillator will run down to about 45MHz before cutting out, not quite achieving the 42MHz in order to tune the receiver to the lowest dial marking of 40MHz. Maybe a selected 6K8 would manage the extra few MHz?

I'm a little suspicious of the 40-60MHz coil. This seems to be wound over paper wrapped around the former and its turns are spread wide apart unlike the mixer and RF amplifier coils which are relatively close-wound. Maybe the coil has been replaced at some time?

Looking at the picture of the dial above, the lowest band has a 1000pF padder and this works fine but I don't think a 0.002uF padder sounds right for the 20-40MHz band and as the highest band is spread much more than the two lower bands its 1000pF padder sounds too big as well. Although an 0.002uF part was fitted I wonder if this was a documentation error and the correct part should have been 1000pf? Also a 1000pF padder for highest band might be useable with the 2-9pF coil trimmer and dust core. In fact the trimmer was specified as 2-9pF in the schematic and parts lists in bold print but a 3-30pF was factory fitted. It does strike me as a failure to embody manufacturing changes ie. the 0.002uF should not have been fitted and the 2-9pF trimmer was probably later re-specified as a 3-30pF trimmer and duly fitted? A mystery we'll never solve....

Finally I'll check the front end amplifier and mixer to see if their tuning range can be adjusted to match the dial readings? Here we are below after slight tweaking of trimmers and coil slugs...with a word of explanation in case you haven't seen this type of scan before..

The tracking generator was input to the R208 aerial post and a high impedance probe placed on the 6K8 mixer grid. The tuning control was set to 10MHz then 20MHz for Range 1, 20MHz then 40MHz for Range 2 and 40MHz then 60MHz for Range 3. You can see the peaks of the RF tuning for low and high settings for each range together with a spike which is the local oscillator running 2MHz above the tuned frequency (except for the 40MHz setting for Range 3 which is missing the oscillator). The final pictures are around the mid-band position of Range 3, with normal tuning then slightly lower just before the local oscillator disappears into the noise.


Of note is the fact that the amplification was pretty well flat whilst tuning across each range from low to high. Interestingly you can see the response curve broadening and overall gain dropping slightly as the frequency increases because the ratio of capacitance to inductance gets less.

When I air-tested the receiver the two lower bands worked exceedingly well.

It was just about at this point in proceedings I learnt of something interesting. The R208 was designed at a time when valves good for frequencies over 30MHz were in short supply. Developments in radar and other things had priority and newer WW2 valves capable of running at VHF were as scarce as hen's teeth. Although the requirement for the R208 was important it had to take a back seat and this meant that the designers were stuck with the 6K8 mixer. By careful selection and probably a lot of effort the R208 managed to get through its manufacturing stage but with the proviso that the performance of the highest frequency band might be a bit hit and miss. I understand a second radio company was brought in to help but they gave up and as I'm finding out over 75 years later the 40-60Mz band is well nigh impossible to fully commission and back in 1942 a decision was made to make do until a later completely new receiver became available... this being the R308, looking very R206 MkII'ish in appearance.


 See more R208 pictures

 Return to Reception