Philips Model 681A Repair

July 2014

 The owner of this receiver also owned the Marconiphone 219 refurbished recently (May-June 2014).

The set dates back to 1949 when it was sold for the princely sum of £48 which was about a months wages and something like four times the price of common radio sets.

 This set is rather special having a full set of bandspread short-wave bands as well as medium and long waves plus general short-wave coverage. Because of this the wavechange switching is especially complicated and there are plenty of adjustments to be made to bring the set back to its design performance. To complement the tuning ranges the loudspeaker is fed from push-pull output valves. An RF amplifier is provided and, to increase the selectivity and image rejection, the set is a dual conversion superhet. Plenty of scope for going wrong...



Here's the service data Philips 681A

 First the valve line-up.

EF39 RF amplifier: ECH35 Frequency changer: EBC33 Detector, AVC and Audio amplifier:

Two EL33 Audio output: ECH35 Bandspread frequency changer: EM34 Magic eye and AZ31 Rectifier

The 1st IF is 452KHz and the 2nd IF 3MHz

To add to the complexity the set is only operated as a dual conversion receiver on its 8 bandspread ranges, being otherwise single conversion.

The bandspread RF coils are fixed tuned to the range centre frequencies and the heterodyne oscillator is also fixed on all bandspead ranges at 3MHz.

Conversion from 2.75 to 3.25MHz is carried out by another heterodyne oscillator which operates from 2.298 to 2.798MHz resulting in the second IF of 452KHz.

The first task was to remove the chassis from the case. Not an easy job and reminiscent of fitting a silencer on a Saab 96. "There's no way this can be done", I told the Saab dealer. "It's tricky, you just have to keep jiggling it", was the response. Yes, with the car at jacked up to almost 45 degrees it could be done... Back to the Philips.. The first job was to remove the knobs which were well and truly rusted into place and had to be detached using a very large screwdriver as a lever on the inside edge of each knob after loosening the securing screw where fitted.

Later I worked out that the volume control and tone control knobs can be removed complete with extension shafts after removing screws in joining collars. This would make chassis removal easier.

 After removing the four chassis securing screws the fun began but, after lots of jiggling, twisting, levering and judicious bending the chassis came out. The case is actually fitted with an inspection panel but as there was an obvious problem with the tuning/wavechange mechanism the chassis had to come out for access.

 Once removed you can begin to see the intricate design of the wavechange fitted to the left side of the chassis, above. The electrical part consists of a pair of inordinately long yaxley switches. Each is operated by a brass gear running in a steel strip pierced with rectangular holes. The steel strip is moved by a centre brass gear wheel which selects one of the 12 wavebands. The design is such that only one of the yaxley switches is in operation at any time, the other only operated when the steel strip has selected the last range on the first. The whole mechansm was extremely stiff and was responsible for the problem with the various dial cords having come off the pulleys etc.

Below you can see the two yaxley switches. The left picture shows the ends connected to the gears. The right picture shows the remainder once its metal shielding cover had been removed.

Each switch has six sections and switch indents are carried at the back of the gear plate where the twin-pronged ends of the switches mate up with pairs of sockets in the indent mechanisms.



 The set like some other Philips models uses a number of very thin bowden cables, unlike the majority of receivers which rely on string. Hopefully all the wire and cables are intact otherwise the receiver will be too difficult to restore. The intial task was to dismantle, clean and lubricate all the wavechange parts, carefully scoring the parts for help with reassembly. The various metal parts are extremely well toleranced and I had to start again a few times having screwed the parts together upside down and back-to-front. Finally, I'd imagined orientation of the drive gear wasn't important, but then discovered there was so much force needed to turn the drive pulley it had be positioned with the securing screw exactly opposite a flat on the gear shaft (which I hadn't noticed). The drive shaft holds the pulley around which is wound very thin steel cable. I suppose I could have rewound the steel wire from scratch but it seemed too complicated and not at all like string.



 Above you can see the ends of the Bowden cable that wind around the gearchange pulley. The right picture shows part of the tuning mechanism, again using Bowden cables.

You can see the reason for having to remove the chassis evidenced by the loose cabling.

Pictures below show the wavechange switch operating gears removed from the chassis.

 A plastic pulley carrying the Bowden cable fits over the centre shaft. The essential flat which I missed when intially reassembling the mechanism is underneath the shaft.

 Here you can see hardened sticky grease which had virtually seized operation. Tension is maintained (or overdone due to stickiness) by a pair of springs.



 As the steel strip moves left the main wavebands are engaged by the LH gear whilst the right gear has been disengaged. As it moves right the left gear is disengaged and the bandspread ranges are selected by the RH gear.

 The pair of switch indent mechanisms, also gummed up with sticky dried gease.

Fortunately the gears and the steel plate looked OK. I once had a German radio using similar gearing and most of the teeth had broken off making the set scrap.

After cleaning and oiling the mechanism worked OK and I could set about restringing the wire. As you can see below the band indicator is set opposite 49m but as the wires connecting to it are loose this need not be correct. In fact logic can be brought into play here as the steel strip that mates with the gears is set to just about mesh wth the rear gear. The front gear selects the Long, Medium and two general coverage Shortwave bands. As the front gear is not engaged it seems that 49m may be OK, however we'll have to see where the pointer ends up once the drive wires are rethreaded. The first thing to do is to lubricate the pointer as dried gease is making it difficult to move, and probably responsible for the dial pointer drive wire being unwound from the drive pulley.

Once the pointer was oiled the Bowden cable could be pulled back and forth rather than jamming, and the main drive pulley and the guiding pulleys could be threaded. Once this was done I tackled the tuning whose main drive pulley was bereft of wires. I found dried grease was again the problem; the pointer guides were sticking on the upper and lower dial edges. After cleaning and oiling these the pointer could be moved freely and rethreading of the pulleys could commence. Fortuitously the wire hadn't stretched and was exactly the right length and once the pulley had been rethreaded the tuning mechanism and pointer worked properly.

I decided to turn on the radio and see if the wavebands were selectable. As luck would have it the radio came on and selecting the bands showed the waveband indicator was around two inches too low. Slackening the pointer securing screw and repositioning it was all that was required.

 Above is the most detailed dial on a consumer radio I've ever seen. You can see the pointer is slewed because its stuck with dried grease.

The circuit diagram and description from Radio & TV Servicing is short of alignment information, presumably because those attempting it should be informed enough to manage without?

Fortunately the radio seems to work reasonably well (at least on a long wire) so alignment should be easy, once the locations of all the adjusters has been identified.

The following shows the specific coils and trimmers to be tuned (I need to check the references)

C14/16/18/20/24/26/28/30/34/36/38/40 are all 3-30pF beehive trimmers

For reference:-

These are arranged in a diamond pattern of four for each stage. Nearest the dial is Long Wave, at the rear Medium Wave. The higher frequency short waveband (SW2) is nearest the output valves and the lower frequency short waveband (SW3) is nearest the tuning knob.

The oscillator group is nearest the output valves, the RF tuning in the centre and bandpass RF nearest the aerial socket.

The corresponding coil slugs are in a line at the tuning knob side of the beehive trimmers.

The front slug is Long Wave, rear is Medium wave, SW2 next to MW and SW3 nearest LW.

C6/7/8 (the 3-gang tuning control) are 10-490pF

LONG WAVE 900m-2000m: Osc S40.C40; RF: S14/C14, S30/C30

MEDIUM WAVE 192m-560m: Osc S38/C38; RF: S16/C16, S28/C28

SHORT WAVE 2 11.1m-34.2m: Osc S36/C36; RF S18/C18, S26/C26

SHORT WAVE 3 34.2m-110.5m: Osc S34/C34; RF: S20/C20, S24/C24

BANDSPREAD 11m : RF S110, S118, Osc S142

BANDSPREAD 13m: RF S111, S119, Osc S144

BANDSPREAD 16m: RF S112, S128, Osc S146

BANDSPREAD 19m: RF S113, S121, Osc S148

BANDSPREAD 25m: RF S114, S122, Osc S150

BANDSPREAD 31m: RF S115, S123, Osc S152

BANDSPREAD 41m: RF S116, S124, Osc S154

BANDSPREAD 49m: RF S117, S125, Osc S156

The bandspead trimmers are the group at the back of the chassis. For the RF amplifier stages aluminium slugs are used for the lower frequency bands and standard iron cored trimmers for the four highest frequency bands. The oscillator slugs are all of the latter type. Bandspread trimmers are located clockwise increasing in frequency from the top right viewed from the rear of the chassis with the oscillator in the right hand group.

452KHz IF S51, S52, S61, S62

3MHz IF S106

Bandspead heterodyne oscillator S40/C40 (tunes approx 2.3/2.8MHz)

The RF stage B1 is used for the non-bandspread ranges only and is capacitively coupled to the second frequency changer hexode, B2

The RF stage for the bandspread ranges is the hexode section of B7.

The picture below shows the speaker still in its original dust cover. This was left in place and a substitute used during testing. As it's a permanent magnet type and the output transformer is fitted on the chassis there's no problem using another.

 I checked each waveband in turn but soon realised that the 49m shortwave band was missing. Because of the way of arranging the gearing it turned out to be a wrong position left at the rear gear as the steel strip moved to the front gear (which operates the four standard bands). This was easy to fix. SW3 is selected, leaving the rear gear clear of the first hole in the steel strip. With a pair of long nosed pliers the flat operating lever running the length of the rear Yaxley switch was turned one notch clockwise to select the 49m band. All wavebands were now selectable and all were in reasonable shape as far as alignment was concerned.

Note that the position of the various adjusters is determined by their wiring to the selector switches so is not entirely logical. The front of the chassis carries three sets of four beehive trimmers and four dust core screws. The rear of the chassis carries three sets of eight dust core or aluminium cored screws. Nearest the front of the chassis are the LW beehive trimmers and at the rear the MW beehives. SW2 are nearest the output valves and SW3 to the chassis edge. Dust cores from the front are LW, SW3, SW2 and MW.

The set appeared to work well although the rear of the two output valves ran cold. Juggling the valve in its socket made the heater light up but it went out again later so this needs some investigation. Not surprisingly, due no doubt to tricky access for replacement, all four scale lamps were burned out. I'll also need to quickly check the main resistors and capacitors.

Another point of note was tuning backlash on the higher frequencies. This seemed to be caused by stiffness at the 3-gang tuning capacitor as the Bowden cable could be seen going very slack in the opposite direction to rotation, implying too much force was being drawn on the other cable. I applied oil to the cables and slackened the tuning capacitor end nut. Afrer applying switch cleaner to this to lubricate the ball bearing I gently tightened it to remove float. With some use the tuning should improve.

From a brief check it was clear that image rejection on the bandspread ranges was infinitely better than that of SW2 & 3 where at the top end of SW2 for example two responses 27.95MHz and 28.82MHz were received at much the same strength. The four adjustment apertures for the dust cores for the 1st IF were completely filled with sealing compound so I'm inclined to leave these alone.

I checked the IF response and centre frequency. The shape of the resonse curve was satisfactory. Rather than risk damaging the IF coils I left the centre frequency alone.

The display shows 10KHz per horizontal division so the 3dB points relate to something like 7KHz. Ignore the frequency reading as the peak marker wasn't lined up accurately.

 Above, the glass dial removed from the set. Over the years the inside surface had become covered with soot. Removing dust and soot on any radio dial should be carried out with extreme care and under no circumstances should a chemical solvent be used unless you want to end up with a clean, blank sheet of glass. Very slightly soapy water was tried and found to be OK used with cotton buds. The original paints were sound and not flaking off as you can get with some early dials. The method of illumination is interesting. Four dial lamps are positioned between a piece of perspex and simple convex reflectors. The perspex sheet is curved directing light onto the top of the glass dial which has a lower silvered edge. The technique is very similar to the methods used in modern flat screen displays which can use fluorescent tubes or more recently bright LEDs instead of filament lamps.

Strangely, at the rear of the dial, instead of being a pale reflective surface is a piece of woven material matching that covering the adjacent loudspeaker. Unfortunately the designers seem to have slipped up and the four lamps are so difficult to access that all four were open circuit and dial illumination probably completely ceased over twenty years ago.

During tests I'd noticed one of the output valves was cold because its heater was out. The sound seemed fine although I didn't try turning it up too much. In a logical moment I switched the two EL33s around and noticed that the cold one sometimes lit up. Something from the mists of time made me remove the valve and apply solder to the ends of the pins. This worked and the set now has a fully functioning push-pull output, although to be honest there was absolutely no change in audio level.

I'd like to check it with my spectrum analyser and see how the different wavebands are performing. In fact (see below) the only two scans I've included were made before alignment.

To monitor the RF response of the set I connected my spectrum analyser to the detector anode. The problem with checking a set like this is the action of AGC. In this particular set AGC action is pretty powerful and the amplitude of the monitored waveforms was confusing. The reason was twofold. On the lower frequency bands, because the set is very sensitive, it was picking up extraneous noise, in particular from my networked TV camera, or most probably radiation from the local area network cable plus radiation from the camera switch mode power supply. The noise level reduced the gain of the set dragging down the response to low level signals resulting in flat gain across a wide spectrum and enhancing image reception.

I found the only way of getting sensible results was to use a small capacitor to attenuate the tracking generator. In fact the 7pF capacitor selected at random was far too high in value and I ended up clipping the croc clip to the body rather than the capacitor lead. The end result, after turning off the camera as well, was quite remarkable. The wanted signal was now appreciably smaller than the image. The display showed the oscillator signal midway between the wanted/image signals and this was much greater in amplitude than the wanted IF signal. Whether some of the oscillator is being picked up on the short connection to the detector I'm not sure. I imagine it is because the 452KHz IF amplifier should not pass much of the oscillator signal so stray pickup although small will be large in proportion to that at the detector anode. This turned out to be true and was due to the unscreened aerial input picking up stray local oscillator radiation. Presumably the local oscillator pickup at the aerial would also acitivate the AGC, reducing its gain on weak signals.

I'll describe the way the picture below is produced as follows... the spectrum analyser has a tracking generator which is a swept oscillator. This oscillator is programmed to start and stop at specific frequencies and the purpose of the experiment was to view the difference between the desired frequency and its image. The output being monitored was the detector anode after the IF amplifier. To monitor the low end of one of the short wavebands (say 9 MHz) I set the sweep to cover 9 MHz plus or minus 5MHz. The local oscillator tuning for 9MHz is lower by 452KHz so two signals will be generated by the mixer valve viz. 9MHz and 8.096MHz. The third signal which will be present is that of the oscillator at 8.548MHz.

So, on the spectrum analyser display will be seen three frequencies side by side viz. 8.906MHz, 8.548MHz and 9.000MHz. These will be of different amplitudes. If the tuning and tracking of the RF amplifiers is optimum the wanted signal of 9MHz will be stronger than that of the image at 8.096MHz. By connecting the tracking generator directly to the aerial input, given that its amplitude is say 0dBm at 50ohms one is connecting a voltage of roughly 200,000microvolts, a level which will completely swamp the receiver, kicking in lots of AGC action. Of course one does need quite a high level of RF because the IF stages will attenuate the 9MHz signal by a significant amount. By connecting the tracking generator via a tiny capacitor to the aerial socket the swamping is reduced a lot and as a benefit the first RF amplifier coil will be much less damped. In fact the solution was to clip the input croc clip to the capacitor body rather than its lead. With this in place the wanted 9MHz signal was comfortably greater than the image.

The picture below shows the response of the receiver to an 8Mz input signal. Note the local oscillator in the centre and the image on the right. This was prior to adjusting the band for best tracking and in fact the signal on the right (the higher frequency) is the one that should be strongest. At this stage the dial is set to 8MHz and the strongest signal on which the marker stands is also 8MHz.

The span is indicated at 2MHz meaning each horizontal division is 200KHz.

 An alternative option would be to set the tracking generator to start at say 200KHz and run through to 10MHz. The display would then show, in addition to the three signals described above, the IF signal of 452KHz. The problem, insofar as what this last signal represents, is that it will be the sum of the wanted signal and its image.

I also tested the receiver to check the shape of its IF amplifier. To do this I set the tracking generator to run from 300KHz to 600KHz. The curve looked reasonably smooth with only a single peak. Using my signal generator and an audio output meter connected to the extension speaker sockets I set the level to avoid swamping and by shifting the frequency little by little found the centre frequency to be 453KHz. This is close enough to the design value of 452KHz and the only effect might be slight difficulty tracking the long waveband to correspond with dial readings. In practice this is of no consequence.

Just a note of warning, although anyone keen to check a similar receiver will already know about this... modern spectrum analysers are designed to be used with modern equipment. Old radio sets use valves and these require high voltages in order for them to work efficiently. As any modern test equipment will have its connections rated as something much less than the HT rail in an old radio you must use a suitable probe to make tests. Another problem must also be dealt with. That is the input and output impedances of test equipment. Most old domestic radios will have been designed with a high impedance aerial socket suitable for connection to a random length of wire and any point within the set will be sensitive to loading. This means that the test probe not only needs to have a rating of several hundred volts but it should also not have any significant capacitive loading effect. See a suitable home-made probe here.

I was interested in seeing exactly what the image rejection was like on the higher of the two standard short-wave bands. I have an accurately calibrated audio wattmeter so switched this to its most sensitive range and connected it to the extension speaker sockets. Tuning the radio to 10MHz I injected a 10MHz signal signal at the aerial socket via a small capacitor and set the wattmeter to read 10 milliwatts or so. First I tuned the radio to the peak of the signal (it only required a tiny adjustment as the dial seems to be very accurate) using the volume control to bring the wattmeter reading to a sensible level then set the signal generator attenator progressively higher, keeping the wattmeter in step by using the range switch and the receiver volume control. Eventually I was able to read exactly 50 microwatts of audio power and then recorded the generator output voltage as 10uV.

The next step was to tune the radio to the image frequency of 10MHz minus twice the IF. Then remove lots of attenuation until I could tune the image at about 9.096MHz. Then by adjusting the signal generator attenuator I set the wattmeter to again read 50 microwatts. The required signal level to generate this output was 1000uV. The ratio of the two signal generator output voltages represents the image rejection of the receiver. This is 1000/10 = 100 or 40dB. An alternative method would have been to leave the receiver set to 10MHz and change the generator frequency to 10.904MHz but it's easier to tune the receiver than my old digital signal generator.

I then checked image rejection at the top end of the range around 25MHz and recorded a ratio of 9/2.6 = 3.4 or around 10dB which highlights the benefit of having a dual conversion receiver for the short wavebands.

This picture shows the response at the top end of the standard short waveband, again prior to final tracking and alignment. In this condition the receiver has an image rejection of just about 1dB and will respond almost equally to stations either side of its local oscillator, effectively doubling the potential number of stations.

 Before finishing with this receiver a couple more points. It has a "magic eye" tuning indicator. These were excellent in their day, insofar as this example would have been nice and bright, displaying a green segment occupying about 270 degrees of a circular shape, closing as a strong station was tuned in. I must admit that it must be 30 years since I've seen one that had a visible display. The second point is the mains cable which like that used on nearly all old radios is a two wire cable. In respect of safety I suspect there is no real danger in practice, but in theory I guess things could go wrong. As the chassis is not connected to anything offering a return to ground for a stray mains connection the chassis could conceivably become live. This in itself is not a definite safety issue unless a user touches a metal part with a link to the set's chassis whilst simultaneously earthing another part of his body. Clearly for live chassis receivers (designed that way) there has always been a potential problem and in that type of set the designers were careful to avoid any exposed metal part, including the grub screws securing knobs which needed a sealing material to prevent a user from inadvertently touching them.

I guess that a solution for this receiver which overcomes two problems (ie. safety, plus finding a suitable mains lead) is to install an IEC mains socket and to supply a suitable moulded mains lead. I'll need to find a connector and a means of mounting it on the receiver, dispensing with the 2-pin arrangement. Like many sets the 2-pin connector needs to be disconnected before the rear cover can be removed. My usual source of IEC connectors is old computer power supplies, which also provide suitable mounting metalwork.

The job turned out to be rather straightforward after a little thought. The old two-pin connector is an integral part of the voltage selection panel. This panel is made of bakelite and fastened to a pair of metal brackets screwed to the chassis and these are located either side of the pair of pins. By cutting off the section of bakelite holding the two pins (using a small drill) I could substitute for it a piece of metal holding the new IEC socket. I cut off the section of metal from the old power supply on which was mounted the mains socket and after cutting it to shape drilled it to suit the pair of metal brackets. The RH bracket now mounts the voltage selector on the remaining section of bakelite plus one side of metal plate holding the new mains connector. The other bracket mounts the other side of the metal plate. This plate has an angle bent into it which I'd left intact from the original power supply box. This helps stiffen the plate and also enabled it to be screwed to the metal shield, at the side of the pair of EL33 valves, for extra rigidity.





 Wiring the new connector was easy. The two original wires go to the on/off switch and could be pulled slightly to line up exactly with the new connector pins. The green/yellow wire from the earth pin had a solder tag attached and this was clamped by the screw holding one of the mounting brackets. The original hole in the rear panel was enlarged slightly to accommodate the new mains plug. I found a 2-metre lead with a right-angle connector fitted with a 13-amp plug and I fitted a new 3-amp fuse in place of the original 13-amp.

In days gone by users were advised to use an aerial and a decent ground connection, the latter providing the path to blow a mains fuse if a fault occurred that placed live mains on the chassis, however with the design of modern houses an earth lead can be problematical. One can't easily drill through a nice new plastic double glazed window unit (and what's the point if you're at the top of a high rise) and with plastic water pipes a good earth connection is dubious. Admittedly, if all the green/yellow earth wires are in place a metal pipe might provide a return to earth, but equally it might provide a very noisy electrical path. The preferred method in 1949 was to dig a hole outside the nearest window, bury a piece of copper carrying the radio earth lead and fill it in with damp soil. Close to this advice you could also read about your aerial and the recommendation to fit a suitable switch to disconnect your long wire in the event of lightning. I can clearly remember listening to our old Murphy radio back in the 40s and 50s and hearing bursts of static, getting louder and louder, heralding lightning, but we hadn't got a switch so just ignored it.

The last job was to refit the chassis into its case whose owner has chosen to refurbish once the radio is working. Getting it out was very tricky and getting it back in was virtually impossible as I didn't want to damage anything. It almost fitted but not quite, however I'd noticed the volume and tone control spindles had 2-inch extensions which could be removed.

The extensions fit into sleeves and a small M3 screw holds each in place. Once the extensions had been detached the chassis slid into place very easily. I bolted it in place and fitted the two knobs to the extensions. The rear tone control extension was fitted and the screw inserted and tightened up.. no problem, however the front extension, for the volume control was impossible to fit unless you have three hands and X-ray vision. The solution was to drill a 13mm hole in the underside of the case, at the front corner as indicated below, through which the securing screw could be fitted into the extension and tightened up.

 The radio works very well and will not require much of an aerial unless the owner wants to explore short-waves. During testing Radio Peking (or whatever the official name is now) came in at good volume and the bandspreading feature makes tuning very easy. Strangely, I heard lots of jammers although I'd imagined these had long since disappeared.

Just before the radio was collected, and as I hadn't tried it in its case, I plugged it in and poked a long wire into its aerial socket. It warmed up and commenced to crackle and hiss so switching to medium waves I tuned down the band... fine.. so I tuned up to the high end and noticed the dial pointer stuck slightly before springing along to the end. A few twiddles and there was definitely something preventing the pointer moving cleanly. Suddenly there was a twanging sound and the pointer stopped moving... the dial cord had come adrift.

After removing the back I could see no alternative but to remove the chassis and sort out the problem. Having done this I managed to rethread the wire and then find out why it had unwound from the main drive pulley. The reason was two-fold. The clever curved plastic lightguide was fitted with mounting slots rather than screwholes for the sole purpose of positioning it away from the top of the dial pointer. To ensure it didn't move there were also two angled metal straps. I'd noticed these of course but hadn't thought about why they were needed. After positioning the mounting slots correctly I tightened up the nuts and tried the tuning again. This time the pointer touched something near the low end and I noticed one of the four plastic dial lamp holders wasn't fully rotated into place. A pair of pliers fixed this and now the pointer moves freely. I guess the motto is to consider carefully why the design of part of a radio looks a bit odd.. there maybe a good reason for it.

Return to Reception