In some ways the RA17 receiver, introduced in the 50's, was ahead of its time but in an odd way it harks back to the twenties. The latter is due to the fact that to tune a particular station requires the use of four knobs! It was ahead of its time as it uses some of the techniques employed by modern digital receivers to overcome frequency drift, which in the wholly "analogue" RA17 is all but absent.
In the 30's when superhet receivers were being introduced to replace TRF, or straight designs, they were heralded by advertisements stating that a single knob was used for tuning. Gone were the days when first a tuning dial was positioned and then a reaction control was adjusted before one could sit back and listen to your favourite broadcasts. The latest TRF sets had been an advance on early models as the former had ganged tuning of several stages whereas the really old models needed a whole row of dials to be set and twiddled before a single station could be heard. The only way to cope with this was to write down the details of all the dial positions for each favourite station, and to this end each dial carried a plate calibrated 0-180 degrees.
Anyway back to the Racal. First you set the number of Megahertz (this is equivalent to a wavechange switch); then you set the Kilohertz knob to approximately where you expect the station to be. Then you set the aerial tuning bandswitch to the correct range and twiddle the associated tuning control. With luck you should hear the background noise peak. Then you return to the Megahertz knob and re-peak the background noise with that. Finally you return to the tuning dial and look for the station. Of course with practice these steps are second nature but it is certainly not as easy to use as an ordinary superhet. Is all this really necessary you might ask, and why was the Racal so successful in its field if it demanded all this fiddling?
Well it had two important features; first the dial was accurate to about 5Khz and if you knew where to look you could sit and wait with confidence for a signal to appear; second the receiver doesn't drift so once tuned the set could be operated solely by adjusting its subsidiary controls.. bandwidth, RF gain, AF gain etc. All this is now commonplace with the most elementary of modern sets because of the "digital revolution". Now that frequency synthesisers are used, the features that were once unique to the Racal are redundant. In a similar way to the Philips receiver, which was the ultimate in TRF design, described elsewhere on this site, the Racal reached the pinnacle of "analogue" superhet development.
The last great advance in analogue receiver design, adopted (for the first time in a commercial receiver) by Racal was the "Wadley Loop". This allowed for complete compensation of the the chief circuits that drifted. The design of the basic tuning VFO for which compensation was not possible was given special treatment and as its range is in the low HF region this circuit could be made extremely stable and to all intents and purposes, drift free.
In a nutshell the "triple conversion" Racal uses a tunable VHF oscillator (the MegaHertz knob), a tuneable low HF oscillator (the KHertz knob), several multi-section filters, numerous mixers, harmonic generators and the like, together with twenty-odd valves and of course lots of knobs to cover the range from "zero" to 30MHz. I say zero because signals can be tuned almost down to "no" KHz. Not quite... because of a huge heterodyne present at this frequency (or should I say "lack" of frequency?). Performance drops off below about 500KHz because the antenna variable tuning arrangement doesn't go that low and one has to rely on the untuned (or wide) position of the switch.
Above is a picture of the Racal now in my collection. It's a MkII which started life at GCHQ, probably in Cheltenham, possibly close to the office where I once used to work but maybe just down the road at a second site in the same town. Most of the GCHQ receivers were rack mounted and the Racal was probably the first type available in those days suitable for this mode of operation.
When I'd found a suitable military style "Plessey Plug", fitted it to a mains lead and plugged it in I was rewarded with a lot of background noise and absolutely no sensitivity at the aerial socket. I connected a long wire aerial (actually the wires to the intercom between the workshop and the house) and carefully tuned around 1.2MHz. With a little difficulty, using the BFO, I managed to resolve a faint heterodyne. By carefully tweaking all the knobs on the front of the set I eventually heard "Virgin Radio" with an awful signal to noise ratio.
The only real worry I had was that the main tuning control was very rough. At each revolution there was a very lumpy feel and generally the mechanism felt rough. I peered inside and saw some very inaccessible gearwheels. Was it damaged teeth or was it a damaged no-longer-obtainable slow motion component? The whole satisfaction of the receiver depends on a nice feel to the tuning as to go from one end to the other of the enormous tuning scale demands scores of revolutions of the knob! I turned the knob backwards and forwards lots of times looking for a corresponding telltale movement in the gears.. but there wasn't any juddering. Then the penny dropped. The knob had been bumped. I found an Allen key, slackened off screw in the centre of the knob and pulled it out a little. When I'd tightened the Allen screw, tuning was as smooth as silk. The cause was the back of the knob, which has a web design, catching on the little clips holding the transparent plastic rubbing plate in position. These rubbing plates prevent the paint around the knob from wearing out.
Back to the Racal's deafness. After removing the top aluminium cover I felt the temperature of the valves. All were lit, and closer examination under the black cans, revealed that most were either new or hardly used (the lettering was clear and fresh looking and hadn't burnt on). As I wobbled some of the more accessible valves the noise level changed and when I wobbled a particular, very hot E180F, Virgin Radio suddenly jumped to half scale on the signal strength meter. I removed two valves and applied switch cleaner to their sockets. The E180F had gold plated pins so I didn't bother scraping these! I could now hear the odd weak station on the medium waveband. I removed the aerial and tried a signal generator. Sensitivity was attrocious. I could just hear 10 millivolts of test signal across the whole range but nothing less, and 100millivolts was needed to give a workable S-meter reading.
Time for a bit of detective work. First check the odd valve on the valve tester or by substitution, then check all the resistors and then perhaps look at the alignment. All the important valves were OK, as I'd guessed. At one point I thought I'd found a discrepancy in the wiring to one of the mixers but hopes were dashed when I discovered the British and American versions were pinned differently... G2 and G3 were reversed.
A first study of the underside of the chassis revealed something interesting. Someone had been here before! I looked at the last IF strip which is at 100KHz and found lots of paralleled resistors. Now these old carbon "rod" resistors always go off. I think it may be corrosion or weakened grip of the brass end caps under the ceramic material. I don't think it's heating and cooling because I have an enormous number of brand new unused resistors of this type and very few are within spec. Most resistors climb in value, none get less. A simple way to compensate is to solder a second resistor across the old one. The Racal is not easy to work on so removing a resistor is not always easy and paralleling is a convenient option. Resistors for valve circuits are not too critical so I don't worry too much unless the measured value is say 700k instead of 470k. One can quickly place a second component across the offending one and if something worthwhile happens change the original, otherwise crack on. All the decoupling capacitors appeared to have already been changed so I didn't bother with these.
After removing all the lower screening panels, of which there are numerous, and changing a couple of resistors in one of the compartments, sensitivity improved slightly. I don't yet have a manual or a circuit diagram, and as the next step is quite critical I had to be careful. This step involves twiddling. There are literally dozens and dozens of trimmers and many of these are associated with filters. If the characteristics of the filters are altered the job may turn into a major exercise to get the receiver back in business so I first marked the position of every one of the trimmer capacitors in the filters. This was easy as air-spaced types are used and a fine pen gave a mark to which one can revert to in the event of a dead end.
I initially checked the 100KHz IF strip and using a signal generator set to 100KHz (without wobbulating it) and with the crystal filter set to its narrowest setting peaked the trimmers. Now Virgin Radio was clear and the meter read over half scale.
Next using the signal generator set to 1.2MHz, and connected to the aerial socket, I tackled one of the filters. There are two fitted with manual adjustments. Both have a row of trimmers going down the chassis then turning a corner and continuing across for a distance. I wobbled the trimmers for each of the sections of the first filter and got a corresponding wobble on the S-meter. Next I tackled the second filter. I gingerly "wobbled" the first trimmer. The S-meter wobbled in sympathy. The second responded, the third, the fourth and so on. The first one round the corner didn't have any effect no matter how much I rotated the rotor and neither did the next but one. The others responded with a sympathetic S-meter wobble. Now because all the sections of the filter basically look identical, it follows logically that if some sections tune nicely, then the others should as well. Across each trimmer was an immaculate-looking red 33pF siver mica capacitor. I removed the one across the first suspect section and fitted a replacement. The section now responded to a wobble with the trimmer set to half-mesh. I went to remove the second suspect capacitor and found that although the first end had a nice blob of solder on its lead it wasn't actually soldered to the trimmer. I removed it and fitted a replacement anyway. Now that section peaked nicely with the trimmer rotor about half mesh.
I backed off the signal generator and found to my relief that a strong signal was now tunable with the generator set to below 1microvolt. Sensitivity for half-scale indication had changed from 100 millivolts to better than 1 microvolt! I connected an aerial and found that my six inch crocodile clip test lead was now producing a better Virgin Radio signal than previously had the long wire. In fact, bringing my hand near to the lead caused the S-meter to press against its stop.
I refitted all the covers and de-upended the Racal back on its feet and connected a length of wire. On 21MHz I heard an OK1 talking to a W8. He said the W8 was S7 but I could only receive him at S3 but that's good enough for me on a short length of wire stuck in the back!
My father-in-law who was G3AQY, bought a Racal RA17 when they were first introduced. The price was astronomic. That receiver is still in the family. My brother-in-law has it now, complete with a manual for which he is now searching. Once I've borrowed it I'll be able to set up the filters properly. Thankfully there wasn't anything seriously wrong with my purchase. Looking back.. judging from the amount of resoldering, the last owner presumably spent fruitless hours looking for the elusive fault. Virtually every solder joint had been seen to... to the extent that one could be excused for believing that this example was the first prototype! However.....each of the metal screens covering the filters had been taped over with masking tape plus cellotape as if he'd steeled himself not to touch the infamous filters. If only he had, and resorted to "intelligent" twiddling, he may have been able to enjoy the delights of this wonderful receiver.