Racal RA17 Mk2

 I picked up this Racal locally. It was said to have no audio but I think that's just dirty valve pins as it came to life after plugging it in and waiting ten minutes. After this it seemed lively but very deaf and intermittent. Hopefully just bad valve connections, but it will need an overhaul to deal with lack of sensitivity, I'm sure. The Megacycle dial tunes nicely and the responses are definite but perhaps a bit lopsided?

The main problems other than misalignment are not too serious, lack of plastic or glass behind the escutcheon and the labelling on the front panel is pretty worn. From the wear on the switch toggles it's had a useful life. Somewhere I have a decent escutcheon if this one doesn't scrub up well. Cleaning the panel will be awkward as most of the labelling is well worn and not all appears to be engraved.

Above before, and below after cleaning

 The picture above shows the set after cleaning up the front panel which I'd guess was coated in nicotine. I used sugar soap to remove the discolouration.

Some of the lettering is engraved but some seems to be transfers. For the time being I'll leave further cleaning etc until after electrical work's been completed.

I checked a really nice example of this receiver which I bought over 10 years ago and most of the lettering was signwritten rather than engraved. I'll consider using lettraset but I may just signwrite new labels with black ink in the future. Just below the escutcheon should be a Racal badge. Oddly my original RA17 has been marked "MEGAHERTZ" and "KILOHERTZ", something I hadn't noticed before but was presumably written by GCHQ service engineers during a regular service?

 

 

 The balun came with the receiver as did a mains lead and aerial lead, both usefully fitted with correct plugs.

 

Refurbishment

 The first task was to get the receiver into a half decent state. Panel staining and discolouration was sorted out then a couple of fairly important jobs were undertaken.

The dial glass was missing so I removed the escutcheon. This is impossible to remove without breaking off one or more mounting pegs as these are secured by push on fasteners which are very tight.

The escutcheon had been removed in the past when the pegs weren't too brittle and had been glued back to the panel, less its upper "glass" window. I used an old CD case to make a replacement window.

By cutting a piece diagonally the size is just right.

 

 

 I cleaned the escutcheon first. Not easy but I finally got it looking black instead of greenish yellow. Sugar soap was best, finishing with a spray sold for restoring car bumpers.

I was unsure how superglue would work but it was perfect for the job. I kept clear of the aperture to avoid leaking onto the plastic window. The glue reacted with the black plastic producing red and green colouration but stuck really well.

Below are views of the panel before cleaning. You can easily see the nicotine staining now the escutcheon's detached.

 

 

 

 

 The two tuning knobs were both binding on the panel. The centre Allen screws were slackened off and the knobs pulled off and cleaned.

The kilocycle scale looked worn and made a scraping sound when the left knob was turned. This turned out to be probably a common fault with old Racal receivers.

Just to the right of the scale in the fourth view above, the scale passes over a roller. The roller was seized solid. It was so bad I initially thought it must be fixed, but I could see a ball race at the top of the roller so guessed it should rotate freely. I think the lower end of the roller probably rests on a single ball and dried grease is the reason for seizure. I was able to drip a light oil onto the bottom end and eventually the roller freed itself. I managed to clean its surface which had become coated with ink from the markings. Fortunately the scale is serviceable. Now the scraping noises have gone and tuning is relatively silent.

I refitted the dial after cleaning the front panel. I'll have to check on the pointer position as the rear of the scale where it was produces a very vague image. I'm pretty certain the last owner bent the needle so it fitted behind the scale instead of in front.

 

 

 Above and below are views of the chassis showing what happens when a receiver is stored in a garage.

 I cleaned most of the dirt away by using a toothbrush dipped in light oil. The only rusty parts are the 6BA screw heads.

When I first tested the set it was pretty certain that valve pins need cleaning. For the moment I just wobbled them all in their holders.

Further testing showed the intermittents had gone and connecting a long wire resulted in lots of noise from the loudspeaker. I had to turn off my network camera as this was produced too much noise to hear any stations on the medium waveband. Although the set tunes stations fairly well I can detect a poor IF response as there are two or three tuning points for strong signals (not that anything is really strong). The megacycle knob tunes well with each resulting megacycle tuning point quite distinct and close to the mark. Having read a receiver test report the other day (it was in a 1932 magazine), I logged all the signals in the medium waveband. Note that this covers two settings for the megacycle knob viz. 0 and 1.

 There were 23 stations audible. In 1932 there were 19 stations logged by the tester. The test was made around 4pm.
Frequency Station Transmission
696 Radio 5 live 693
716 French 720
832 Classic Gold 828
885 Radio Wales 882
898 unidentified 901
912 Radio 5 live 909
1000 unidentified 999
1027 Radio Jersey 1026
1055 Talk Sport 1053
1090 Talk Sport 1089
1108 Talk Sport 1107
1117 Radio Guernsey 1116
1172 unidentified 1170
1198 Virgin AM 1197
1216 Virgin AM 1215
1260 Virgin AM 1260
1335 unidentified 1332
1361 Radio Solent 1359
1380 French 1377
1407 French 1404
1463 unidentified 1458
1553 unidentified 1548
1560 unidentified 1557

 You can see an error in recorded frequencies of about 3KHz or more, partly because the pointer was wrongly fitted.

I could find no long wave stations and the signal strength of the medium wave stations was not good.

Next I'll attempt to carry out alignment of the receiver. No easy task.

I started the next day having lugged the set into my workshop. Initially I thought I'd try various things to get the feel of what exactly needs to be done. I injected signals into the aerial and found a bad connection between the coax braiding and plug in the short length of coax that came with the set. Having corrected this I found sensitivity was such that around a 2 microvolt signal was audible at the low end of the band 1000 to 2000KHz but this needed boosting to 50uvolts at the high end to hear the same level of signal. Checking the IF transformers revealed very little change to the S-meter with more effect when tuning the first. Tuning the second transformer had only a just discernible effect on signals which seems wrong, maybe a low emission valve or resistors gone high?

As I'd had trouble previously with the lower frequency VFO, I checked this and found it ran from 3.150MHz to 2.063MHz with 3.100MHz able to indicate the zero scale marking, but the end frequency at the other end of the dial read around 3KHz high. In fact this error was noticeable at 2.5MHz when the dial read 603KHz and didn't change much by 2.1MHz when the dial still read 3KHz high. The amplitude of the VFO on a test receiver showed a variation from S8 to S6 from 3.1 to 2.1MHz.

Connecting a long wire brought in long wave stations but there was a strong heterodyne on the 198KHz station and the Irish station on an indicated 252KHz was solid but not as strong as I'd have expected, neither was France Inter on 162KHz (indicated). Testing the band 1000 to 2000KHz showed plenty of broadcast stations but these repeated between 1,7 and 1.9 MHz suggesting a filter problem. I also tested the shape of the sets response using the S-meter. I found 4 or 5 distinct responses for each strong signal, which is not right. Finally I tried test signals for higher frequencies. 26.02MHz was not present.

Next I'll check the two VHF filters and see if they produce an acceptable pass band or otherwise. In the past I've found bad silver mica capacitors in these.

Checking the 40MHz filter showed it was serviceable but not quite right. The results of realigning it will be shown later. The 37.5MHz filter is easier to set up. All the trimmers could be peaked and only one needed a little retuning.

Further tests revealed the receiver was basically working, in that it received signals, but needed all its gain to produce good results. Something's not right and gain variations kept showing themselves up. I looked again at the 100KHz IF amplifiers. There are two EF93 pentodes in the main signal path with a third tapping off the first to produce a pair of 100KHz outputs on the rear panel. As I'd earlier discovered the first stage tuned fairly sharply but the second did not, being very flat and little change in output as the transformer was tuned. After lots of testing using the S-meter as a guide I switched off and checked the components around V14 and V16 the two IF amplifiers. I was concerned about the screen voltage of V14 as this was 120 volts instead of 60 volts. I found the reason. A previous owner had soldered resistors across the 15Kohm screen ground resistor and the 47kohm screen HT resistor. I removed these and fitted a new 15kohm resistor as the original read 33kohm. Checking the results now showed the correct value for the screen voltage but reduced overall gain. I wonder if the last owner was suffering from the deafness fault and increased the screen voltage to improve matters? I connected up a scope to the output of the second IF amplifier, V16 and was somewhat surprised to see a voltage on the diode of the detector, V21 of 38 volts RMS. This was the sort of level that resulted in a comfortable audio output. At this output level the input voltage was a mere 5mVolts, representing a gain of 77dB. I suspect the IF amplifier is OK and the real fault lies elsewhere.

After more tests I'd managed to reduce sensitivity even more by swapping valves. I was now struggling to hear 200uV at 1.8MHz when previously I could hear 50uVolts. Is there a rogue valve in the set? Maybe I should fire up the valve tester and do the job properly? There are some clues. Gain drops off as one tunes from low to high on each band. It could be a component in one of the inaccessible boxes, like that carrying the RF amplifier, or maybe one of the three mixers or one of the miscellany of amplifier valves has a bad anode, cathode or screen resistor?

Connecting my long wire aerial resulted in lots of medium wave and long wave signals, but it's a really long wire and signal strengths are high. A crystal set would probably work just as well as the Racal.. Radio 4 on long waves is accompanied by a heterodyne possibly indicating the set is working much harder than it should and picking up an internal oscillator?

The next step was to check the valves. As I have a working RA17 I opened its top and just changed valves one by one. Setting the dial to 1.8MHz and the level set to give an indication of 30 on the S-meter I changed each valve in turn. V8, an EF91 caused the meter to move to 35 and V4 gave a further increase to 40. Others made no difference. I then started tweaking trimmers, leaving the 40MHz filter alone. The trimmers for the 37.5MHz filter improved things slightly.

I then compared results with the two sets side by side. The newer set was still slightly deaf but not as bad as before. At 1.2MHz signals of 20uV on the new set was equivalent to 2uV on the older one.. not too bad all things considered, but probably indicating a high value resistor in one of the more-difficult-to-test modules.

Finally an air test on 20 meters. The set was now quite lively and resolved weak SSB stations quite well. Time to call it a day until I choose to continue alignment using the spectrum analyser to squeeze an extra bit of performance.

See the filter alignment...

... keep watching...

return to Reception