AR77 Alignment

 Alignment of this complicated receiver is not going to be easy.

The main reason for this is the fact that the main tuning dial is finely calibrated and interacts with the bandspread dial, at least over the top four frequency ranges.

The operating manual is very unhelpful, as it assumes the receiver will never need full alignment and is not very clear about a couple of important things. First the bandspread dial. Should this be set to the vertical line at the HF end for proper calibration of the main dial or to the highest marked frequency of the band in use? If the receiver is accurately aligned then this doesn't matter as trial and error will prove it one way or another. Secondly, is the local oscillator designed to be high or low of the incoming signal? The AR77 manual is silent on this point and I had to read the AR88 manual to see if this mentioned anything. The AR88 oscillator is set to be high of the incoming signal so hopefully that was the principle adopted with the AR77?

Pause for thought

Let me explain a little. A superheterodyne receiver is capable of simultaneously receiving several signals of different frequencies at the same dial setting. Of course it is designed to receive only one, that corresponding to the dial marking. Unfortunately a specific dial setting is not a guarantee that you will not hear anything else.

In the case of a very sensitive receiver designed for weak signal reception, for example a distant amateur station, you will be able to receive strong signals that have different transmission frequencies: The reason for this is the way a superhet receiver works. The AR77 is a relatively poor performer as it uses an intermediate frequency (or IF) of 455KHz. Basically, the higher the IF the better the so-called image rejection.

Take for example the 20 meter band. Let's say the AR77 is tuned to 14.050MHz (call it f1), so its local oscillator will be set at f1+0.455, or 14.505MHz (call it f2). The mixer which does the work of adding these two frequencies (f1 & f2) is a bit dumb and unfortunately cannot distinguish between adding and subtracting so will simultaneously provide two outputs which are 14.050MHz and what is called the "image" frequency of 14.505MHz plus 0.455MHz, or 14.960MHz. Both mixer outputs will appear as a 455KHz signal.

Resorting to equations..

The arithmetic for the desired result is f2-f1=0.455MHz which is the same thing as f1=f2-0.455MHz

Because a mixer cannot distinguish between a plus and a minus the equation can also be written f1=f2+0.455MHz. Here, f1 is called the image or the unwanted signal.

Thus f1 can be f2 plus or minus the IF frequency.

This means that whilst one is digging out a weak CW signal at 14.050MHz a second signal at 14.960MHz might be apparent, ostensibly as far as the operator is concerned, on the same frequency. Of course you may be lucky and no-one is transmitting on 14.960MHz, but if the image were to fall in a short-wave broadcast band the unwanted signal might be tens of thousands of times stronger than the weak CW station.

This problem isn't just confined to domestic or amateur band superhet radio receivers. Because many TV sets and communications satellites also employ superhet principles large numbers of people are constantly engaged in frequency planning. In days of old, ones TV reception on the fringe of a major transmitter might be plagued by interference, usually patterning, from another transmitter broadcasting on an entirely different channel but whose image fell into the frequencies of the channel to which you were tuned.

Satellites too are not immune and it would be upsetting if a satellite started re-broadcasting an image signal. To help avoid such problems one can polarise ground transmissions to be either horizontal, vertical or right or left circular, but that's another story.

The designer of a communications receiver will understand the practical problem and will have arranged for some rejection of the unwanted signals. In the main this rejection is carried out by using multiple circuits tuned to the desired reception frequency. The AR77 uses a single RF amplifier between the aerial and the mixer and the quality of the components in this amplifier will determine the amount of image rejection. The 1939 advertisement for the AR77 boasts about its superior image rejection controlled by a user operated control on the front panel. This is really an aerial trimmer but it's better than an internal preset trimmer as it can be precisely adjusted by the operator.

In practical terms the peakiness of the tuning will determine rejection, however on the highest frequency range which is 18 to 31 MHz it is difficult to derive a high image rejection because of the proximity of the desired and undesired frequencies. The performance of the AR77 is given as 60 at 18MHz and only 25 at 31MHz. This compares with 50,000 at the lowest frequency of 540KHz. The latter is because at 540KHz the image will be 1.450MHz, a long way off in practical terms because a tuned circuit at a low frequency will have a much superior performance than one at 31MHz.

Returning to the 20m band we can expect an image rejection of maybe 200. Our one microvolt CW signal will be equivalent to a 200 microvolt broadcast signal image, and that is not a very powerful signal as broadcast stations go.

So in order to use the AR77 on the amateur bands, particularly above 40 meters, it must be accurately aligned. This means that all the tuned circuits must reflect the dial markings or at any setting of the dial any tuned circuit needs to be peaking and this is a real problem. Not only must the oscillator tune to 455KHz above the dial marking, the tuned circuits must all be working at peak performance at any dial position. Because of the physics involved this means that the inductance of the coils must be precisely set to correspond with whatever capacity exists across the coil at any setting of the dial. This is what alignment means, it's the overall adjustment not just that at one specific frequency. One speaks about oscillator tracking which is ensuring that the local oscillator is always exactly 455KHz higher than the dial markings.

If one were to connect a signal generator to the aerial socket of an AR77 set to say 20 meters and crank up its output to say 100millivolts then tune the generator from say 400KHz to 30MHz you will hear not just the main and image responses but countless others also. For example you will hear 455KHz and, depending on the quality of the signal generator, and the physical condition of the AR77, you will hear loads of signals corresponding to harmonics. You may also hear broadcast stations, particularly if the generator is set to CW because the generator will act as a local oscillator. The only way to make things tolerable is to ensure any receiver is properly aligned.

This article is chiefly intended to describe my alignment of my AR77 so I will proceed.

A general look at the AR77 front end

The first problem I encountered was adjusting the trimmer capacitors. I believe these were set up in the factory with the RF front end on a test bench before being incorporated inside the chassis. There are adjusting holes for trimmer access but these are now inaccessible. The trimmers are "special air-spaced low loss" ceramic-bodied contraptions with a long brass spindle carrying a hole at right angles. The spindles are secured by concentric nuts which require slackening before adjustment can be done.

A very unhelpful suggestion in the operators manual is the advice that the lower screening panel should be in place when adjusting trimmers. OK, you can twiddle coil cores through holes in the lower panel, but not the trimmers, which are the most sensitive parts in relation to the screening panel. I guess by trial and error one could waft the panel around whilst trimming, because you really need to use the locknuts?

 Trimming, lets say, is therefore very awkward.

The layout of the tuned circuits looks very methodical. It is and it isn't because the layout is dependent on the wiring from the wavechange switch.

From left to right the ranges are 2-1-6-5-4-3, a bit reminiscent of the firing order of an engine.

This mainly reflects the aim of keeping wiring as short as possible at the higher frequencies.

After spending a few hours twiddling I was unable to align the higher frequency bands so withdrew to discover the reason. Maybe one or more "padder" capacitor has drifted, maybe the 6K8 mixer is bad or a resistor or decoupling capacitor is bad?

As the AVC switch has been disconnected with parts left dangling, maybe a previous enthusiast has done something wrong?

Below is a chart showing the six wavebands and the likely local oscillator tuning requirements in MHz (the IF is 455KHz)

I've assumed the oscillator is always on the high side of the signal

RANGE RF LOW RF HIGH OSC LOW OSC HIGH IMAGE IMAGE
1 0.540 1.340 0.995 1.795 1.450 2.250
2 1.340 3.300 1.795 3.755 2.250 4.210
3 3.300 5.800 3.755 6.255 4.210 6.710
4 5.800 10.200 6.255 10.655 6.710 11.110
5 10.200 18.000 10.655 18.455 11.110 18.910
6 18.000 31.000 18.455 31.455 18.910 31.910

 Each band will produce two main responses from a signal generator.

For example at a dial setting of 10.2MHz you will hear 10.200MHz AND 11.110MHz.

Setting the dial to 11.110MHz and tuning the signal generator to 11.110MHz at a level of say 1mVolt will produce a huge response.

Retuning the dial to 10.2MHz will produce another huge response.

Switching to Range 6 and tuning to 22.220MHz you'll find another sizeable signal and another at 23.130MHz

In fact, because the IF is 455KHz the images are always roughly a meghertz higher, or more precisely 0.91MHz higher.

Finding an unknown IF or coil values and padding capacitor

As an aside; many years ago I was aligning an old Hallicrafters set whose dial markings on its highest range had no connection with received signals because a long wave range had been substituted for it and after hours of twiddling decided to use a spreadsheet to help sort out the problem of restoring things to their original state. At that time, before Excel, I used Lotus 123 which, like Excel, lets you write equations into cells and produce tables. I investigated the short waveband which had been removed. By measuring the maximum and minimum values of the tuning capacitor a listing of frequencies can be produced for various tuning capacitor settings in a column given a particular coil value. For example at 100pF you would tune a signal of 18MHz with the short-wave coil. These frequencies can be compared with the dial markings until a perfect match is obtained.

Next a similar calculation is performed to determine the value of the new oscillator coil. Its inductance is estimated by trial and error and after measuring the tuning capacitor max/min values and using the known IF comparisons are made with marked dial readings. This time an important component must be taken into account. This is the padding capacitor which effectively modifies the action of the tuning capacitor. Depending on the precise circuit position of the padder the oscillator frequency equation needs to be modified. The oscillator and RF columns in the spreadsheet can then be compared.

As the exercise had been primarily to determine the new coil values and the padder value these items were kept outside the table and used as presettable constants. Keyboard "F4" was used in Lotus and it was also adopted in Excel for this purpose. The padder was appended with the label F4 and by changing the value from say 100pF to 200pF one could see its overall effect on the main tuning and the oscillator tuning at the same time.

At this point I'd got things working to the level where some fine tuning on the spreadsheet was feasible. I added in trimmer capacitors, again as F4 values, and set these as reasonable values, say 10pF. By typing in numbers, progessively using more significant figures, things resolved themselves and I was rewarded with new coil values and a padder capacitor value.

Click to see the original set on which I did this

Work proceeds 5th March 2014

I decided to sort out some of the really bad resistors and also swap some of the old capacitors.

I changed a 470kohm resistor reading 675kohm for a new part and then replaced an open circuit 100kohm resistor. The latter I discovered later resulted in the BFO starting to sort-of work.

I suspect this 100kohm resistor failure stopped the last "restorer" in his tracks as a new switch dealing with BFO selection was wired in place but no solder had been used and the connections were very loose.

I also fitted about a half dozen new capacitors in places where high value resistors were associated with them. This would remove any significant current leaks as old capacitors tend to develop resistance in parallel with, and very often also in series with, their capacitance making the efficiency of the capacitor pretty poor.

At this point I decided to retest operation and turned to the lowest range, setting the dial to exactly 0.54MHz where I was rewarded by a signal from my generator slightly off tune.

The correct method for tracking is to adjust the oscillator coil at the low end of the band, then retune the receiver to the top end, 1.3MHz and trim out any frequency error with the preset capacitor.

Having set the peak exactly on 0.54MHz, I then tuned the set and generator to 1.3MHz. By increasing the generator output level I could just hear a signal off tune so I twiddled the oscillator preset capacitor and peaked the output. Turning back to the low end the error was now very small, and by twiddling the coil and trimmer a few times the dial settings corresponded perfectly.

Next was to repeat the alignment but this time on the two RF amplifier coils. Both responded well to tuning and tracking and I found I could reduce the generator output level to 0.2 microvolts at both ends of the band and still just hear a signal.

During this process I had to unplug my network camera which generates a tremendous broadband mush and found I could hear broadcast signals even with the generator plugged in because the underside of the AR77 was unscreened. I plugged in a long wire aerial and there were masses of powerful stations right across the band, although mostly distorted.

So far so good. Presumably there are no serious problems with the set that will affect alignment of the other five wavebands.

Before I proceed further I must fix a couple of glaring faults. First there is no automatic volume control operation and to hear decent audio I have to reduce the RF gain control. I would have expected at least one setting of the various knobs to provide AVC. Second, and this may be related to the lack of AVC is to solder the new switch connections and find out if all the wiring is in place.

Fixing the AVC switch

The first job was to remove the "new" old AVC switch which had six positions and find a replacement with only three positions which is correct. The junk box turned up a four pole three way switch and after sawing off half an inch of spindle I fitted this. I removed a capacitor and three wires from the old switch and then decided to draw a circuit for the switch and trace the wires to see how they corresponded. During this process I found that the circuit diagram I have and the actual AVC circuitry didn't quite match, but the two versions were close enough for me to re-establish the wiring to the new switch.

There are three sets of wires.

The AVC circuitry which operates at position 1 and 2 and connects the AVC detector diode to the RF amplifier and first IF stage grid bias feed. My set uses only one of the two diodes in the 6SQ7 valve and incorporates a 1 megohm resistor whereas the circuit diagram shows both diodes in parallel and a 220kohm series resistor.

Negative feedback selection at position 1 links the audio output via a 5.6kohm resistor to the earthy end of the volume control. A new wire had been added grounding the earthy end of the volume control so I cut this off, letting the new switch place the ground connection in positions 2 and 3.

The third set of wires are for position 3 switching HT to the anode of the BFO and simultaneously grounding the AVC line.

Once the new wiring was in position I'd discovered the previous restorer had disconnected the negative feedback feature, failed to discover an open circuit anode feed resistor to the BFO and disconnected the AVC line. This sort of supports my theory about why some apparently good receivers are put on one side for ever, until sold or scrapped. When I turned on the set to check it was still working it had been totally transformed. Tuning across the medium waveband revealed it was full of intelligible stations. Before, the RF gain control needed to be carefully adjusted in order to resolve anything, but now with the gain at maximum everything was really clear and as a bonus the S-meter was working properly.

Alignment again

I returned to alignment and found range 1 was almost perfect and after a few adjustments I was able to see a small S-meter change when tuning across a 2 microvolt signal. I switched to range 2 and tweaked this and found it was also perfect. Range 3 proved to be a problem. No amount of twiddling could line up the dial markings with the signal generator so back to thinking about potential problems and wondering whether my spectrum analyser could help as it had with aligning my Racal RA17. A spectrum analyser enables one to see the local oscillator moving up and down as one tunes a receiver both in terms of frequency and amplitude. In the case of the RA17 the VFO signal had suddenly vanished mid-scale whilst tuning the receiver because of a bad EF91.

7th March 2014

Still tackling the problem of dial alignment on the higher frequency bands, I traced the circuit to discover exactly how the various ranges are configured.

The table below lists the various coils and capacitors

COIL TUNING TRIM MIN FREQ MAX FREQ
RF STAGE MHZ MHZ
L1 C2+C16+C1 NONE 0.54 1.34
L2 C2+C16+C1 NONE 1.34 3.3
L3 C2+C16+C9 NONE 3.3 5.8
L4 C2+C16+C7 NONE 5.8 10.2
L5 C2+C16+C8 NONE 10.2 18
L6 C2+C16 NONE 18 31
OSCILLATOR PAD
L13 C6+C5 C24 C42=330PF 0.995 1.795
L14 C6+C5 C23 C40=1000PF 1.795 3.755
L15 C6+C15 C28 C44=1000PF 3.755 6.255
L16 C6+C13 C27 C43=2700PF 6.255 10.655
L17 C6+C14 C26 NONE 10.655 18.455
L18 C6 C25 NONE 18.455 31.455
MIXER
L7 C4+C3 C18 0.54 1.34
L8 C4+C3 C17 1.34 3.3
L9 C4+C12 C22 3.3 5.8
L10 C4+C10 C21 5.8 10.2
L11 C4+C11 C20 10.2 18
L12 C4 C19 18 31
IF 0.455

 The main tuning capacitor has three sections and each section two sets of rotors and stators (making 6 sets in all)

These are labelled C1 through C6 and are in pairs each approx 300pF+150pF

The bandspead tuning capacitor also has three sections but has three sets of rotors and stators (making 9 sets in all)

These are labelled C7 through C15 and are in sets of three approx 25pF+15p+10pF

The antenna trimmer C16 is operational on all bands and is separately tuned and looks about 30pF.

Each coil, apart from the coils in the RF amplifier, has an independent trimmer, C17 through C28.

There are 4 oscillator coils fitted with padding capacitors. These are the 4 lowest ranges.

After puzzling over the results I've been getting and comparing these with the above table I was somewhat confused.

Basically I managed to align ranges 1, 2 and almost range 6 but failed dismally to align ranges 3, 4 and 5 where, in each case, the oscillator failed to have enough range to tune the selected band. Today I found the reason and it's nothing to do with the padding capacitors which I'd begun to suspect.

Covering the tuning capacitors is a large metal screening plate and of course this prevents one from seeing what's happening. I suddenly had an idea whose proof I tested using my signal generator. Setting the bandspead dial to 7.2MHz and the main tuning dial to a point fairly close to 7MHz I found a signal. Turning the bandspead dial to 7.3MHz should have enabled me to hear the generator set to approximately 7.3MHz but instead I found I had to tune to 7.1MHz. The penny finally dropped and after detaching the metal screen I saw the bandspead capacitor was turning the wrong way !

Either the dial cord is wound back-to-front or the tuning mechanism is wrongly assembled.

Because restringing the dial cord is a real pain I needed to find another AR77 so I emailed the 19 set forum and was quickly informed by Donovan, VE1BDC that his military version of the AR77 which has the designation GR10 has a bandspread knob which turns clockwise to increase frequency. Thankfully I hadn't restrung the dial.

Why then should my bandspread work backwards? I figured it out.

There are several mechanical parts in the bandspead tuning set-up.

(1) The dial which can be locked in place by screws.

(2) The slow motion drum carrying a few turns of cord and which can also be locked in place by screws

(3) The bandspread capacitor which can be turned through 360 degrees because it has no end stops.

(4) An endstop device which allows the capacitor to rotate through only 180 degrees and can be locked in place by a grubscrew.

Because some ranges use bandspreading in parallel with the main tuning the AR77 needs to be operated in a specific manner in order to get reliable dial readings. The manual says to adjust the bandspread dial to its maximum frequency setting at which position the main dial reads accurately. This applies to ranges 3, 4 and 5 only, otherwise for ranges 1, 2 and 6 the bandspreading is not electrically connected as can be seen in the above table. During my alignment exercise I duly set the dial to a marking at its high end and, as the tuning capacitors are inside a screened box, I'd taken it for granted that all was well. By now however after twiddling trimmers for hours I was somewhat expecting something odd. What I found was at the high end setting the capacitors were at two thirds mesh and tuning through the 180 degrees to the low end of the band progressively put the capacitors less in mesh until the vanes were fully disengaged at 60 degrees then continuing to tune downwards they started to mesh again to 30 degrees. Over the last 30 degrees the tuning knob was correctly sending the oscillator down in frequency. The results as you can imagine were very odd indeed.

After puzzling over the mechanics and trying various new settings without success I eventually realised the problem lay in the endstop device. Somehow this had become out of phase resulting in two thirds mesh at the HF end of the dial. By loosening the grubscrew and pushing the capacitor vanes I could begin to put right the problem, but only so far because the grubscrew securing the endstop collar hit the body of the capacitor. The solution was to remove the grubscrew and rotate the collar another 90 degrees, refit the grubscrew and lock the collar into place so that the endstops corresponded to full mesh at the LF end of the dial and unmeshed at the HF end of the dial.

How had this problem happened?

 I suspect it's connected with a weakness at the endstop collar (left centre above). There's a pin on the shaft which hits a peg and this peg is slightly bent. I think the dial or tuning system was forced beyond its end position and the pin managed to force itself over the peg, bending it a little. Once over the peg it was impossible for the pin to get back and essentially from then on the bandspread mechanism was working back-to-front. Because of the screening cover obscuring the operation of the tuning capacitors the previous owner probably never did figure out why his set suddenly went deaf on the popular amateur bands, not to mention tuning backwards and picking up some stations twice!

Having set the bandspread tuning capacitor to its correct position I set the corresponding dial to the end marker line (at the HF end).

Magically, the ranges where the bandspread is in operation became really easy to align.

After a few coil and trimmer adjustments each of the ranges fell into place and dial markings were fairly close to input signals.

Air testing proved that each amateur band was tunable and SSB stations relatively easy to resolve. Everything isn't perfect because various parts are well worn but stability is reasonable. Medium wave reception is exceptional using a long wire.

Return to the AR77 overhaul

or

Return to Reception