Roberts R200 Overhaul (May 2015)

 This set was first introduced in February 1960, the year I started my electrical engineering course at Liverpool University. It sold for around £20 depending on the wholesale price and the exact amount of purchase tax. Now, put into context, when I started work as a fully qualified electronics engineer in 1965 my gross pay was about £15 per week, and that was considered a decent amount. Wages in those days were fairly stable and, in 1960, a radio like the R200 was considered as a luxury being affordable by only well-off families. About the same time a brand new Ford Popular motor car was about £275. Using a rule of thumb, the Roberts R200 at today's prices would be £500.

Below, a bunch of colourful examples of the R200 gathered from the Net.

 Sadly, Roberts Radios have long been "Roberts"-badged cheap Chinese versions since the original firm were bought out in the mid-1990s. Although many are adorned with Union Jack labels, read the small print to understand what that really means....

 I'm not a strong advocate of the original brand. Although electronic design was usually first class, mechanical design and construction was abysmal. Wiring between the circuit board usually employed, and a push-button assembly, often employed, was a rat's nest. A jumble of more or less identical wires made disassembly a tortuous exercise. Transistors prone to failure were often buried in tin boxes very difficult and time consuming to access. Despite this loads of Roberts portables seem to have survived, albeit in non-working condition for aeons.

This early example was in a very tatty condition when it was received; paint splattered, and battered, but nonetheless a treasured family possession.

Scrutinising the Trader Service Sheets shows there was more than one version of the R200, so I'll summarise these for any interested radio anoraks.

Serial numbers up to 12961 used the following line-up of transistors: OC44/OC45/OC45/OC78D/OC78/OC78 and from 12,962 dispensed with the OC78s and used newer audio transistors OC81D/OC81/OC81. There were other circuit changes also, but mainly the circuit and layout was similar up to serial number 70,000. Read on for other details relevant to this set which is serial number 52,698.

After 70,000 used: AF117/AF117/AF117/OC81D/OC81/OC81 (The AF117 types are horrible and most will have now developed nasty faults). The layout and some major parts were also changed.

The first step was to remove the chassis and speaker and clean up the case using warm soapy water and a brush. I then scratched off the paint splatters and glued loose rexine back in place.

The speaker grille was badly bent and had a large black oily deposit. After brushing with warm soapy water this cleaned up OK and after hammering between pieces of wood the grille regained most of its flatness. Being made from aluminium it's difficult to restore because the metal stretches easily.

Below is a picture of the case after treatment and with the grille temporarily refitted. Supergluing the grille to the inside of the case and temporarily fixing it in place with the speaker surround ensures that it will remain flatter. The colour is quite striking but had faded in patches and there were lots of scuff marks which had removed the coloured surface of the rexine. I struggled to find a method of sympathetically restoring the original colour and eventually settled on lipstick borrowed from my XYL. This was quite effective and finishing with wax polish should stabilise it.

 Above you can see the loudspeaker, the turntable removed from the base of the set and the pair of wooden wedges favoured by Roberts in most of their wooden cased portable radios to secure the chassis.

Below is a view of the chassis which is covered with a white oxidisation, but otherwise it looking sound.

The reverse view of the chassis. Note the missing nut at the tuning condenser. Strange as the nuts were loctited.

 The dial assembly after cleaning. You can see by the discolouration that two of the three knobs have been missing for a very long time.

The set uses early Mullard transistors: OC44 self oscillating mixer/RF stage, OC45 1st and 2nd IF amplifiers, OC81D audio amplifier and a pair of OC81s for driving the 3 ohm loudspeaker. A single OA70 diode is used for developing the audio and AGC.

 The makers' name plate and duty label. The R200 was manufactured to a modified design from S/No 36547 and to a much newer design from serial number 70,000. Looking at circuit changes it's possible that these were introduced for specific reasons, perhaps to counter instability and/or poor quality sound (maybe due to component tolerances or degradation of some parts). Read on for more details.

The duty label shows that Mr.Marconi's patent investments were still paying off even after 50 years when Godfrey Isaacs, the brother of the Earl of Reading, and managing director of the Marconi Company, was put in charge of extracting cash from makers of radio equipment.

 Above is the dial showing medium and long wave stations which were available in 1960. Vague Long Wave.

Next, I'll power up the set from a 9 volt power supply and see if it's going to be a problem. Hopefully not, because the early OC series transistors are generally reliable.

Before I powered the set I noticed there seemed to be a few broken wires at the largest coil on the ferrite rod. Closer inspection revealed that this was indeed the case and, unfortunately, the coil was wound with Litz wire. This uses seven strands of extremely thin enamelled wire twisted together. The reason for using Litz wire is that it has a lower resistance at radio frequencies for the amount of copper used compared with solid wire. Although direct current makes use of all the copper in a wire, RF currents use mainly the outer surface of the copper, making the core of the wire almost redundant. To maximise the use of copper as much surface area as possible must be available to RF currents. One solution in radio work is to use Litz wire. In this example the surfaces of seven thin wires are available for RF current. The data sheet tells me the resistance of L3 should be about 7 ohms but my meter tells me it's over 20 ohms, indicating that each of the seven wires in the coil should have a resistance of 49 ohms and the 20 odd ohms is indicative of only two of the seven wires being intact.

I noted the connections to the coils on the former holding L3. The former actually holds three coils: L3 plus coupling coils L5 and L1. L5 is underneath L3 so will be intact and L1 looks OK. Having removed the former and detached the connection to L3 I carefully unwound it. There were several areas of damage. One option is to find some new wire and another to attempt repair. I opted for repair. If the end result is too poor I'll consider fresh wire. I found the following: 1 break at 9 turns, 3 breaks at 20 turns, 2 breaks at 27 turns, 1 break at 33 turns and finally 2 breaks at 43 turns. Most of the damage had resulted from a dent in the coil caused by an impact with something.

Repairing Litz wire? Next to impossible? I tried anyway and managed to get the resistance of L3 down to about 10 ohms after resoldering 10 breaks in the wires (I probably missed one or two as 10 ohms works out at five good wires). If Long Waves work I'll leave it at that. I refitted the repaired coil and soldered back the wires. Next, I'll power it up from a bench power supply and see what happens.

The 9 volt input drew around 140mA. Not good as the set is supposed to draw 8mA with no signal! The good news is the set responded with a large increase in noise when the medium wave coil on the ferrite rod was squeezed, as did the set respond when the long wave coil was squeezed, but from the sound in the loudspeaker I can tell the set is oscillating at a supersonic frequency. Twiddling the tuning condenser produces squeals in several places indicating the set is basically OK. I notice the circuit board has been unscrewed from the chassis at some time and the IF dust cores are damaged. Clearly, I'm not the first to look at the set.

I looked at the circuit. Lots of potential problems. Electrolytic condensers and dark brown coloured condensers which from distant memories I recall are not always long-lived. At this point I noticed something odd. There are too many transistors. In the audio amplifier I can see two output transistors and two drivers instead of the one I was expecting. The extra transistor has a brown spot but it's number has been rubbed off. I think a brown spot might indicate an OC81D but there's one close by with a blue spot and that's marked OC81D. What remains of the code could be OC78D? If so that corresponds with the older circuit diagram, but the OC81D isn't shown. Very odd... In fact from S/No 12,962 OC78s were changed to OC81s.

Reading further explained the anomaly. An extra diode (X2) was fitted in versions after 36,547 across a new output transistor balancing control pot. Although this diode was intended to be a type OA81, an OC78 transistor with its collector and base wired together was used (no doubt to use up surplus supplies of the latter).

 

See the trader data sheet for versions up to serial number 36,546

See the sheet for serial numbers after 36,546

 Note that certain modifications to the early sets were only noted in the later Trader Sheet so it's important to read both when tackling an example of the early model.

Changes include transistor changes from 12,962, viz. audio types changed from OC78s to OC81s.

 

Circuit tests

 These Roberts printed circuits are very difficult to follow due to randomness of the layout of the tinned copper tracks. However I started by checking the resistors and found these to be in pretty poor shape. Examples are R6, R12 and R13 all marked 560 ohms measured 814, 674 and 617 ohms respectively. As these were measured in-situ their true resistances are likely to be even greater. R3 and R9 marked 3900 ohm measured in-situ 4600 and 4700 ohms and R7 marked 1000 ohms was 1300 ohms. Most other resistors were not worth checking in-situ.

I then tested the electrolytic condensers. C20 was 1uF and greater than 20 ohms (should be 2uF), C21, C22 plus a green condenser under the circuit board all measured around an ohm and not measurable capacitively due to resistive shunting. The electrolytic in place of C5 was open circuit (=100uF). It looks like I'll have to remove and test, then probably replace most of the these components.

I fitted a set of new electrolytics although only one measured open circuit. Refurbishing radio sets almost always demands realignment. This set has been meddled with so alignment is a must. I found the IF to be several Kc/s out and the stages not amplifying very well. One dust core was broken confirming past meddling. The trick is to carefully remove any cores with broken tops and turn them upside down. This generally gives you a new slot and, in this case, this was so. Because one's ears are not too good at accurate sound amplitude measurement it's essential to use an analogue meter for the job, connecting this across the loudspeaker. My meter indicated a half scale reading for no signal. This was due to the audio amplifier oscillating at a supersonic frequency.

Ignoring the oscillation, which could possibly be caused by extreme misalignment, I adjusted the IF transformers. These all peaked nicely at 470Kc/s, having shifted them from something like 480Kc/s. Next the medium wave alignment. A mark on the dial shows the HF alignment point at 1400Kc/s or 214m. The local oscillator and RF amplifier coils at this setting are adjusted using the tuning condenser trimmers.

It became obvious that alignment was so bad that realignment required several passes. The LF end is adjusted at 500m or 600Kc/s using coil L7. After many passes I had the alignment set correctly and numerous whistles could be heard. Switching to Long Waves the data sheet tells you to set the dial to 425m and inject a signal of 185Kc/s. This seems odd, but it's because there are no wavelength settings on the long wave dial. Having set up both medium and long waves I turned my attention to the supersonic oscillation, which was now modulated by motor-boating at some dial settings.

The power supply current registered 140mA and both OC81s were warm to the touch. A trick for finding the cause of the oscillation such as this is to use a condenser to short various circuit points to ground and after a few jabs the oscillation ceased, the output meter dropped to near zero and a quietish background hiss could be heard. The PSU output current dropped from 140mA to 8mA. What had I shunted to ground? It seems that shunting the live side of R11 to chassis via an 0.05uF condenser had done the trick. R11 is the resistor which with R10 defines the bias voltage for the base of TR4 (the OC81D audio amplifier). I twiddled the dial and was rewarded with lots of stations, but not at particularly loud volume. Continuing with investigations; tuning the set to Radio 4 Long Waves, showed that shunting one side of R21 to ground significantly increased the audio output whilst introducing distortion. The circuit layout shows R21 to be 47 ohms and is used in a feedback circuit with R19 (5.6 ohms) and C19 (0.02uF). R9 also has a bearing on the feedback and this 3.9kohm resistor had measured high (in-situ) at 4700 ohms. Clearly this area might determine the reason for both the oscillation and low audio. Looking at the circuit shows that shunting the loudspeaker with a low impedance meter (as mine is) will also affect feedback. The first step will be to remove the parts that measure wrongly and fit new parts in accordance with values shown in the circuit diagram.

After replacing the audio coupling condenser C21 and finding little had changed, I experimented further. The feedback circuit is a bit reminiscent of the reaction circuits from the 1920s and 30s and is an area of one of the major changes that took place during the life of the set and I reckon that drifting of component values has shown up weaknesses in the design. Shunting the base of the OC81D to ground with 0.05uF has countered the supersonic oscillation problem and revealed poor sensitivity. The first step was to re-check alignment. I reset the IF to 470KHz and this raised the gain slightly. I then checked the coils on the ferrite rod. The medium wave coil was OK and I could track it reasonably well across the band, but the long wave coil had too much inductance. This no doubt was a combination of damage and my scramble winding of the wire I'd removed. I cut off around 12 inches of the wire and realigned the coil to track over the long wave band. The set relies on a fixed capacitor, C3 of 40pF together with the tuning condenser trimmer (which is used for medium wave alignment). This is probably fine if you only use Radio 4 Long Wave, but as there is now an Irish station at the HF end of the band it's nice to be able to hear this as well. After some experimentation I found that reducing the inductance of L3 and adding an extra 20pF across C3 enabled good performance over the band. Read on.

 The picture above shows the pickup at the long wave coil L3 from the coupling coil L1 which is connected to my signal generator. I tackled lack of gain by using my spectrum analyser because my preferred method wasn't possible. I had in mind using a grid dip meter to check the resonance of L3 when the set was tuned to Radio 4 which broadcasts on 198Kc/s, but although I have three different GDMs, none had a long wave coil (probably because the US doesn't use long waves for broadcasting so far east designers didn't bother extending their GDO operation below around 500Kc/s). The method I used was ideal. As the R200 tunes high and low the above response moves right or left. The peak reading of the curve should correspond with the dial setting. Radio 4, as you can see above, gave me a reading of 198.750kHz. The Q of the coil, together with loading effects, defines the response shape. Changing the ratio of inductance to trimmer capacity, for example, will alter the shape, but tracking accuracy will suffer as Q increases. Setting the low end correctly by sliding L3 up or down the ferrite rod, then tuning to the high end, say to the Irish station RTE1 on 252Kc/s, then adjusting the trimmer for exact alignment, then repeating this a few times should result in perfect tracking, and therefore best long wave performance.

Next, I decided to investigate lack of overall gain. This is where I found the reason for the low gain was the component values in the audio feedback circuit. Over the manufacturing life of the R200, did the designer's change parts to give extra audio volume or better audio quality (or both)? I found that decreasing the value of R19 which is supposed to be 5.6ohms had quite dramatic effects. Shunting it with an 8.2ohm resistor improved overall sensitivity to the point where it was acceptable. Lowering the resistance further made the sound louder, but more shrill and, shunting R19 with 47uF made it loudest, but with complete loss of mellowness of sound. I decided to stick with the 8.2 ohm shunt which gave me the best compromise. It seems therefore that the overall performance of the set is too dependent on resistor tolerances. R19 might now be 3.3 ohms, but if the value of R19 had drifted upwards, as it certainly had, we're now looking at a new value for R19 which brings the performance back to nearer original.

 At this point I decided that the set was working well enough and likely to be quite acceptable. These old portable sets are useful because, being portable, they can be readily orientated to minimise interference, which gets worse and worse as more low energy lamps are fitted and mains data networks (pushed by BT for their TV boxes) are used. I have an old Roberts R309 which I use daily on Long Waves and, by rotating it slightly, performance is really good. Huge levels of interference can be completely removed by this method. You can also position a set so it's ferrite rod is close to a copper water pipe and improve weak station reception quite dramatically. Many years ago, it must have been circa 1963, I visited John O' Groats. There was a large grassed area for car parking and in the centre a telegraph pole with a metal sheath up the side. I placed a portable radio alongside the vertical metal sheath and found I could hear our local radio station, Radio Merseyside on something like 202 meters. Alas, I see that this station, along with all the other BBC local stations, faces closure on medium waves.... it's probably gone by now... cost savings combined with forcing listeners to use DAB.

Return to Reception