Antique Radio Restoration Handbook en
Antique Radio Restoration Handbook en
Antique Radio Restoration Handbook en
2. Scope
3. Abbreviations
4. Objectives
6. Supporting Information
11.Passive Components
15.1 Emissions
15.2 Immunity
16. Photographs
17. Glossary
1. Foreword and Acknowledgements
Telephone: +44-(0)2392-596836
E-mail: [email protected]
2. Scope
3. Abbreviations
4. Objectives
The restoration materials indicated here are types available in the UK.
Recognising the international readership of this book, a glossary is
included in an attempt to define the materials so that locally
available substitutes can be identified if necessary.
6. Supporting Information
The error rate in original documentation will be found far higher than
would be considered acceptable nowadays. Problems and uncertainties
due to this cause are best resolved by reference to a “gold standard”
example of the model in question if you can find one near where you
live, and if the owner is sympathetic enough to let you examine it in
detail.
Because each radio design had its own set of stock faults when in
service, a lot of time can be saved by getting wise to these sooner
rather than later. Magazine articles can be invaluable as a source of
informed material, as can old hands who used to work on these sets
prior to retirement. Sometimes the list of stock faults can be
surprisingly specific. For instance, the mains transformer on the B40
is unreliable… but only on the original production variant. Some B40s
mistrack… but this problem will usually be found only on those examples
of the later “D” variant that somehow escaped one particular authorised
service modification package.
If faced with this problem, make a pair of handling frames right now at
the start of the job, one fitted to the left skirt of the chassis apron
and one to the right. Chipboard usually works fine for this job.
Wooden chocks can be screwed into position, to tilt the chassis
slightly backwards at a convenient angle for working on the bench. By
making the chipboard sideplates a little higher than the tallest
component on the chassis, this technique would allow full inversion of
the radio on the bench without damage.
Silver plating was used by Marconi and Eddystone amongst others, and
tarnishes with age unless it was lacquered when new. Unlike other
metals, the oxide of Silver is nearly as conductive as the metal
itself, so there is little to be gained by polishing. Quite the
reverse in fact. All polishing will achieve is a reduction in the
depth of the highly conductive plating. If it is desired to shine-up
an oxidised Silver surface, a quick wipe over with a cotton wool pad
carrying a very little Silver Nitrate (AgNO3) will be found more
beneficial. This material is hazardous to health, however.
7.2 Earthing
WARNING: AC/DC sets must not have their chasses earthed, and not all
radios fitted with mains transformers have an isolated chassis.
Examples exist of radios with transformers which nevertheless have the
chassis connected directly to one side of the mains. (This design type
is most commonly found on broadcast receivers.)
Many radios were assembled with paxolin valve sockets. All too often,
the securing screws were also used as grounding points. The paxolin
tends to creep with age, leading to the securing screws becoming
somewhat loose. The repair technique used by this author is to
introduce externally serrated washers between the soldertag and the
chassis. These washers have considerable in-built springiness and the
sharp teeth do a good job of providing continuity.
7.4 Grommets
Very often rubber grommets are used to support variable capacitors, RF
subchassis or anti-microphonic valve sockets. These must all be
replaced at this stage with brand new items. Don't bother trying to
make the originals last. They probably won't go on for very much
longer, and the development of related faults can be very insidious.
Tuning backlash, creeping paralysis of the dialdrive mechanism and
slowly developing crackles due to intermittent waveband connections,
are all phenomena that this writer has positively tracked down to
failing rubber bushes at various times over the years.
Gearboxes need checking for bent gearwheels, correct split gear spring
preloading, tooth wear, and bearing maladjustment. Start by verifying
that the gearbox sideplate securing screws are good and tight, and that
the gearbox frame is firmly attached to the RF unit subchassis or
whatever else it is bolted to in your radio.
The correct endfloat for all gearbox shafts is zero, with very slight
endthrust. Note that on the average gearbox, over-tightening any one
shaft will spring the sideplates apart slightly, reducing the endthrust
on adjacent bearings. Any adjustment should therefore be made only a
little at a time.
All split gears should run true and have a full compliment of
tensioning springs. B40 owners should note this particularly. The
amount of tension should ideally be equal to about 1½ teeth. Check
that the two halves of each sprung gearwheel are free to move with
respect to each other within the limits imposed by the mesh of the
gears, and that the two split faces have not become corroded together,
or separated too far because of wear in the hub. With the gearbox dry
but uncoupled, the spring loading should be sufficient to accommodate
all effects of friction and stiction in the geartrain.
Some split gears are very large in diameter, even as big as 3". With
these sizes it is particularly important to check that the wheel
rotates concentrically, and that the two halves are properly supported
by the hub.
The plastic gears found on Drakes are sloppy at best, and can impart a
very lumpy feel to the tuning, due to wear, grit or long term
underlubrication.
Limited application of the special grease sold for use on the nylon
gearboxes of model racing cars is a good first step. It will be best
to avoid using any of the usual automotive greases.
7.7 String and Wire Dialdrives
Most pointer dialdrive systems used an open fabric or nylon cord, and
these rarely present insurmountable problems. If replacing a cord on a
radio originally fitted with waxed string, be sure to clean the pulley
grooves carefully as otherwise lumps of old wax residue may cause
uneven travel of the new cord. Cord knots should be sealed with a bead
of polystyrene cement. Replace rusted or stretched tension springs as
necessary.
Some radios, such as the BRT400 and those of the Swiss manufacturer
Paillard, used an open stranded steel wire which is terminated at each
end by soldering, without the use of a separate tensioning spring. The
writer is unaware of any source of this material, though modelmaking
shops do stock plastic covered stranded aircraft locking wire that can
be used instead. However, it is noticeably less flexible than the
original. Care is needed on later models of BRT400 (suffix B and
onwards) to grease the wire where it passes over the bronze guides.
Early BRT400s, and most other sensible radios, used pulleys instead of
guide sleeves and hence do not suffer this problem.
Collins used an open plastic insulated stranded steel wire from the
beginning on their model range. In general there are no problems, but
inside the KWM-2 PA cage it is necessary to make sure the wire ends do
not touch anything to prevent any possibility of unwanted RF currents
flowing along the steel core of the wire. The ability to control this
problem was purportedly the reason why the factory selected the plastic
covered wire in preference to the (then) more common uninsulated stuff.
Lubricant should be kept well away from cord or nylon dial strings,
because it can rot the fabric in time. One useful modern lubricant
that can find an application in old radios is the type sold for
lubricating the carriages of computer printers. It contains PTFE in
colloidal suspension, and has proved excellent for lubricating pointer
slides.
This technique was used by Murphy in their B40 design for the British
Admiralty and in an HRO variant (a German copy?) which the writer saw
23 years ago at Aston University, and which he now wishes he had
inspected rather more closely than he did. It was also used by Peto
Scott in the Trophy-8.
A worn chain drive will always cause backlash, the only variable being
the exact magnitude of the problem. Tightening the inevitable jockey
wheel reduces the backlash, but introduces excessive friction due to
bearing sidethrust. The solution used by this writer is lots of medium
grade lubricant, with the tensioner set to only just take up all the
slack.
Fitting a new chain will usually stop the backlash, but in the writer’s
experience, the use of a new chain running on old sprockets is a recipe
for stiffness and notchiness in the drive.
Some highly ingenious friction drives have been designed. The key
aspect for a restoration programme, is to ensure an absence of uneven
wear on the friction surfaces. It is also important to make sure that
the spring tensioners are functioning correctly, and that the
lubrication conditions are correct.
Even the smallest trace of ordinary mineral oil can quickly ruin a
friction drive. Do not therefore, lubricate friction drive mechanisms
other than as indicated above or better still, exactly as recommended
in the manufacturer’s documentation.
Make sure the LF and HF endstops are correctly adjusted. This can
prove extremely difficult, for some designs seem not to have been right
even at the start. The National NC100 comes instantly to mind. The
writer has had to file the endstop wedges on his example, and can see
no way that it could ever have worked properly without this rather
bespoke operation being performed.
Other designs were far less competent, with the result that severe
damage may be caused by mishandling. The NC100 and HRO are examples.
The first does have stops, but has no means of slowly unwinding the
flywheel. The second has no endstops at all, relying on the torque
absorption capabilities of the (admittedly beefy) gearbox and variable
capacitor. The friction drive Eddystones and the SP600 can wear flats
on the rim of their driven pieceparts if spun repeatedly against the
endstop, causing lumpiness in normal operation.
Set Oldham couplers to have specified endfloat, and make sure the anti-
backlash spring is present, and effective against the rotational
stiction of the output shaft.
On most good quality receivers the tuning gang has a cover, to keep out
dust and draughts. Notable exceptions include the KW equipments such
as KW202, certain Eddystones such as the 940, and certain Hallicrafters
receivers such as the SX62. For these equipments, consideration should
be given to fitting a plastic cover, ideally of polystyrene or ABS
material. A good place to start is the plastic potting boxes available
from RS and Maplin. These things come in a range of sizes, and are
devoid of the corner fixings found on boxes designed to accept lids.
They can be held in position with Evostik or silicone rubber.
Some receivers, such as the Murphy B40 and the Hammarlund SP600JX use
variable capacitors which have ceramic shafts. These are very fragile,
and can sometimes be found fractured, especially if the radio has been
dropped. In the event of unexpected tuning backlash in a newly
acquired receiver, this point is worth checking. It is usually
possible to effect repairs by using Araldite.
The projection may be of images marked onto a glass drum or disc, and
is usually presented on a ground glass strip or plate. Some very
ingenious arrangements were made, many of which gave a quasi-linear
presentation of markings originally made in a circular format. The
technique is efficient in the use of space and very reliable in
service.
7.15 Cooling
In the case of radios such as the Murphy B40, which are fitted with
metal rollers or skids underneath, proper feet will have to be fitted
anyway. This is to avoid severe damage to the surface upon which the
radio is to rest. When fitting feet to these radios, consideration
needs to be given to the height of the feet, and to the weight of the
radio. When in service, the B40 had lots of air available due to the
design of its anti-vibration mounting cradle. So a standoff height of
at least 1/2” would seem reasonable.
Damaged Allen and Bristol pattern grubscrews are self centring for the
drill bit, but they can sometimes be removed more easily with a reverse
threaded stud remover. American grubscrews always seem to be made of
much harder material than the usual flat headed British ones.
Occasionally the knob will be found stuck fast, even after removal of
its captivating screws. The obvious solution is to remove the shaft as
well, so that repairs can be performed away from the radio. Very often
though, this is impossible. On these occasions it is necessary to
decide up front, the extent to which brute force is to be used.
Various techniques are available. It may be best to drive the knob
further onto the shaft prior to any attempt at pulling it off, but be
extremely careful in the case of collet knobs. It is generally
inadvisable to lever against the panel, because of the near certainty
of causing cosmetic damage. In desperation, this writer once used a
miniature butane torch on the knob, with a fibreglass mat being used to
protect the panel. The knob was quickly ruined, but at least it came
off easily. The panel was unscathed, and the control shaft quickly
cleaned up with some emery paper, ready for the fitment of a new knob
which had been obtained for this purpose beforehand.
On radios with badly designed over-travel stops, the tuning knob may
have slipped round in service, especially if it has unusually large
inertia, as on the Marconi Atalanta. Sometimes the wavechange detent
is so strong that the selector knob suffers the same fate. The result
can be that the shaft gets so severely scored that the knob cannot be
removed after removal of its securing screws, even though it can spin
freely on its shaft. This is because either the shaft or the bore of
the knob bush, or most likely both have become severely burred.
Because of the size of the tuning knob, a hefty puller can sometimes be
used. If mishandled, this can damage the over-travel clutch and the
mechanism, causing even more work. If the knob can be replaced or
repaired, it may be worth drilling the centre of the knob to allow the
jackshaft of the puller to bear directly on the end of the shaft. When
the set is rebuilt, make sure the shaft has no burrs, and tighten the
grubscrew(s) hard. Then take care in the future.
Knobs which are taper seated, or which tighten down onto a wedge, are
best assembled onto the shaft with a trace of grease so that they will
come off more easily next time. Do not over-tighten the securing
screw.
Black bakelite knobs often have engraved arrows, letters or lines which
were colour filled, often in white. The original colouring often wears
off over the years, leaving the engraving unfilled. The original
appearance may be restored quickly and easily by the use of a Kwikfyl
crayon pencil. The various available colours can be useful for
restoring the engraved panel legends, labels and warning signs.
Completely remove all old colouring material from the engraved
character with a pin. Then refill the engraving with new Kwikfyl,
applied with a candle. Remove surplus material immediately, with a
clean rag.
Towards the end of the valve era, these came into widespread use.
Unfortunately, the substrate materials then available were very poor as
regards mechanical strength, water absorption, dielectric strength, RF
loss, dimensional stability, thermal survivability and track/substrate
adhesion. Taken all in all, this is a rather damning list, and
indicative of the problems nowadays found in restoration.
Many PCB etch patterns were lacquered on the print side, after
soldering. This coating can, and usually does partially wash off
during the cleaning process. After repairs have been completed you may
wish to consider brushing (do not spray) a thin coat of clear gloss
exterior grade polyurethane varnish over the print. The good points of
this technique include preservation of the PCB appearance, the
avoidance of Copper oxidation, and the extra support - albeit slight
- given to the Copper print. Bad points include difficulty in the
event that rework is found necessary, or in the event that repairs have
to be made in the future.
At first sight it may seem that the first generation radial ended
components used in these equipments would cause few problems, but
unfortunately this tends not to be the case. Heavy components like the
bigger polyester capacitors in moulded cases, often relied on being
physically supported by the PCB, rather than being captivated by a P-
clip as on earlier equipments. Modern components tend to be much
smaller, with a reduced pitch spacing of the leadout wires. When using
today’s components as replacements for period types, the inevitable
wire doglegs look bad, and the body of the components are usually left
at some distance from the PCB surface, standing proud by perhaps 5mm or
so. This can place too much strain on the print doughnuts in the
longer term, often resulting in the appearance of an intermittent
fault. This writer would advise a small bead of Araldite beneath each
incorrectly sized replacement component, to save the fragile print from
having to take all the strain.
Nowadays it is recognised that for PCBs having print on only one side,
it is good practice to cross-hatch all large solid areas. This avoids
undue stress on the track/substrate bond, which prevents buckling of
the PCB at temperature extremes. In the days of valve radios this
problem was not recognised or if it was, then it was not addressed.
Many intermittent faults stem from long term thermal cycling of these
early PCB structures, resulting in cracked Copper tracks. Beware in
particular, of large PCBs with output stages or rectifiers in the
middle. Be even more suspicious if the PCB is horizontally mounted
directly above the valves.
The new PCB could be a straightforward mimic of the original, but this
writer would advise that the print be roller tinned, that all large
groundplane areas be cross-hatched, and that pitch spacings should suit
currently available components. Put a component silk screen onto the
board, even if there was not one on the original. All resistors,
capacitors and valve sockets would normally be renewed as a matter of
course. Ceramic or PTFE sockets would be used for the RF, oscillator
and hot-running valves. All wound components including any RF chokes,
would be the originals transferred from the existing board.
The lower RF losses of FR4 laminate & modern valve sockets at VHF,
could lead to parasitic oscillation not found with the original PCB
assembly. Be on the lookout for this phenomenon, and have 10Ω
resistors or ferrite beads at the ready.
On the chassis, check all valve sockets for contacts that do not grip
the valve pin snugly. Be particularly wary of the side contact family
(CT8, type E). These suffer a variety of problems. A slightly tight
fitting socket often causes the Copper rivets on the valvebase to
loosen, causing intermittent contact. On the other hand, a slightly
loose fitting socket can also be a source of actual or latent
intermittent faults. In the middle there may be a happy medium… but
not usually, with this awful family of valves. Each individual contact
comprises a hairpin spring leaf, the ends of which are supposed to be
free to move with respect to each other, not firmly connected together
by a lump of solder. This is a point worth checking carefully.
Because little of the CT8 valvebase stands proud of the socket when the
valve is seated, it can be difficult to remove the valve except by
pulling on its glass envelope. This is an unsafe practice, and of
course there is every possibility of the bulb becoming loose in its
base. Perhaps more commonly on removing CT8 valves from their sockets,
the bottom edge of the metalizing becomes detached from the bakelite
base, leaving the valve looking ragged edged. It certainly pays to
make sure there are no bits of debris or loose metalizing, lying in the
well of the bakelite moulding. The sockets for this family usually
seem to have a marking pip at pin 7, which seems unusual considering
the valve carries eight pins. Another peculiarity of CT8 valves is
that the pin numbering advances anticlockwise as viewed from underneath
the valve, which is the reverse of every other popular valve family.
Check also for contacts that have become a loose fit inside the overall
socket insulation moulding. This problem tends to be specific to
certain families, especially those B7G, B8A and B9A moulded composition
sockets where each individual female contact comprises a piecepart
resembling a miniature tuning fork. If your radio has these items
fitted, it is definitely not a good idea to power the set without its
valves because sometimes the individual contacts can be so free to
move, that adjacent connections touch each other. When reworking one
of these sockets, be sure not to rigidify the wiring more than
originally. This is to ensure the valve pins have the same amount of
freedom in the valve socket as previously. Failure to do this can
result in microphony, especially in sets with an integral loudspeaker.
Or broken valve pins, of course.
Some sets are fitted with special sockets which positively locate the
valve. This may be done to get the heat away, or to guarantee local
oscillator performance under operational conditions. The Eddystone
770R for example, uses standard McMurdo valve sockets in its RF unit
but they are modified to suit the application. Certain contacts for
the EF95s are replaced by special turned pieceparts.
Inspect the sockets of hot running valves and all barretters and output
valves for signs of decomposition or dry joints. Rectifier sockets
should additionally be inspected for evidence of flashovers or tracking
paths leading from the anode pin to chassis, or more often between the
anodes in the case of full wave rectifiers.
The acorn family (954, 955, 956) of valves are usually seated in large
ceramic bases which have strongly-sprung gripper contacts. Because of
the fragility of the radial seals on these valves, it is important that
the contacts mate correctly. For this reason, it is advisable to
lightly oil each radial pin before insertion of the valve into its
socket, then remove all traces with methylated spirit to avoid problems
later on.
This writer believes that slightly bent castings are best left well
alone unless of brass, which can stand a certain amount of resetting
without much likelihood of cracking. In the event of a casting
becoming cracked as a result of being dropped, or because of a failed
attempt at straightening it, there are several optional ways ahead.
You are advised not to attempt to close up the inevitable small gaps
when the major components are offered up to the repaired casting.
Tinplate or brass shims should be constructed to fit the major
components perfectly into position without appreciable distortion.
Given the availability of the right tool, new thread inserts are easy
enough to fit, though most are metric types nowadays. If you wish to
retain the original BA or UNC securing screws, a matching hexagonal nut
may be secured in position on the bulkhead using Araldite.
Unfortunately, tapping a BA/UNC thread into a metric insert is
impractical because the material is far too hard.
Some receivers were fully sealed against water and dust ingress. At
least, they were when they were new. Many of these designs were
intended for tactical use by the armed forces. Good examples include
the British R209 and the American Collins R-392 HF receivers. Because
most such sets were fitted to tanks, Landrovers, halftracks or Jeeps,
the construction had to be rugged in the extreme.
Army CNR sets were capable of operation over wide temperature ranges.
For operational reasons, they were designed for use by an operator
wearing arctic mittens and so chunky knobs were generally fitted. Many
radios were installed in their host vehicles in positions which
deliberately encouraged their use as toe holds. Therefore the front
panels, knobs and windows had to be “boot proof”. Nevertheless many
surviving CNR sets are found in badly damaged condition. Be on the
lookout in particular, for CNR sets which rattle or which have huge
dents or broken castings. Such sets have probably been dropped. The
internal damage will probably be far worse than you at first imagine.
Other waterproof sets were intended for marine use, and these include
commercial types as well as Naval. Here, the standard of construction
is better than normally found in other classes of equipment, but less
robust than the military CNR types. The marine sets tended to be well
looked after by the users, and by the maintainers as well.
In general, sealed radios with valves in run very hot because of the
obvious lack of ventilation louvers, and most have led a very hard life
indeed. At least no dust or grit will have got in! The extent of the
required overhaul will depend on whether the seal has failed, allowing
internal corrosion to develop. If it has, then most of the work will
usually stem from this cause. Especially on marine equipments.
The intricate mechanical design of sealed CNR sets will provide many
challenges caused by interlocking structures and high component
densities. Some of the better sets were modular. This makes them
easier to repair if all the necessary connectors and umbilical patch
cables are to hand. The unfortunate corollary is that sets designed
this way often prove harder to repair than a straight chassis layout if
no service harnesses are available. One thing to be on the look out
for with all modular CNR sets is intermittent connectors caused by all
the maintenance that had to be done over the years, to keep these sets
fully operational when in service.
Very much on the plus side for sealed equipments is the joy of opening
one which has had intact sealing throughout its working life. The
interior gleams like new and the chassis requires no cleaning at all.
It must be said though, that the smell which escapes when the seal is
first broken is often very unpleasant, especially if MFP doping has
ever been sprayed on!
The seal itself is often in the form of one or two O-rings. These are
commercially available in different sizes of round and a variety of
rectangular cross-sections. A ring made from Viton material of about
60 Shore “A” durometer hardness would be a good starting point for most
radios. Note however, that this material produces Hydrofluoric acid
which is extremely corrosive, if grossly overheated. Therefore do not
touch a Viton seal with a soldering iron. If necessary, sections of O-
rings can be joined together, and proprietary kits are available to
make this job reasonably easy. Clean the groove in the casting, and
then lightly coat the new ring with MS4 silicone grease before fitting
it.
Sets with silica gel desiccator cartridges should have this component
present and thoroughly dried-out before resealing the case. Unless all
the control shaft O-rings are also renewed, it may be expected that the
overall sealing of the radio will not be perfect. Therefore, drying of
the desiccator will need to be an occasional preventative maintenance
chore.
Rotational Bowden cables often work in push-pull mode, which avoids the
need for a strong tension spring. Philips used this type of
arrangement to operate the tuning gangs on some of their radios, for
example model 643A. In practice it all worked better than may be
expected, with minimal backlash and no need for the tuning gang to be
strongly located on its chassis mountings.
8.3 Turrets
Turret tuning mechanisms are to be seen in the B40, SP600, DST100 and
51S-1. Although generally trouble free, it is important to get the
rotor contacts to mate centrally with the stator contacts. This means
that the axial position of the drum needs to be correct, with zero
endfloat. Usually, one end of the drum is supported by an adjustable
bush bearing, while the other is spring loaded by a wavy washer.
Wherever turret grounding springs are supposed to be fitted, they
should all be present, correctly tensioned and properly lubricated.
Some radios were designed to have their turret turned in one direction
only. If working on one of these, do not remove the interlock/detent
mechanism without being very careful.
Quite often the sprung contact leaves are found to be broken. New
contacts can be fabricated from thin phosphor bronze sheet, and held in
position with a brass collar prior to soldering. After the solder has
cooled fully, the contact can be bent gently into exactly the right
shape, or at least near enough to do the job. On most turret designs,
access to the mating contacts is quite good, the SP600 being a
noteworthy exception to this rule.
Eddystones seem to suffer from dirty contacts. HROs suffer from broken
rivets securing the spring finger contacts to the insulator blocks
under the tuning gangs. This fault can be quite hard to repair in
situ, although it is definitely possible without removal of the tuning
gang.
Slide switches have changed footprint quite a bit over the years.
Sometimes, exact replacements for early types are impossible to find.
Whilst this is rarely a problem for panel mounted slide switches, there
can be severe headaches in the case of PCB mounted ones. Often, the
best way ahead is to salvage the baseplate from the original switch,
and use it as a header for a new component mounted piggyback, with its
legs trimmed as short as possible or splayed apart, then soldered to
the header. Deft work is required. If the new switch is slightly
smaller than the old one, the overall dimensions of the new hybrid will
be approximately the same as the original component.
In the event that the original switch has become loose on the board, or
stiff in its operation, the connecting print doughnuts may have ripped.
In this case, strengthening is necessary. Start by gluing the switch
baseplate to the top surface of the PCB substrate, prior to soldering.
This task is often made difficult by the residue of lubricant which
will have been applied over the years.
Switches carrying signals at low voltage and zero current need gold
contacts to avoid developing high contact resistances. On wartime
radios where this is a recurring problem because the switch contacts
were of cheap material, it may be possible to introduce a bit of bleed
current through the original switch (say 2mA or so) to keep it clean.
On one SX28, the switch on the noise limiter would click convincingly
but not actually make or break reliably, even after careful
lubrication. This same fault was also seen on an original replacement
switch which was then fitted as a swap for the first one! In the end,
a modern replacement had to be used, a great shame since this component
is at the top of the chassis and extremely visible from above.
Iron cores come in two basic types with a range of grades to cover
different applications. They have a wide range of relative
permeability.
True ferrites are shiny, and coloured either charcoal or black. They
have a high to exceptionally high value of relative permeability, and
are best suited to LF and MF work.
Powdered Iron cores are usually matt grey. These have a lower but
still positive value of relative permeability, and are well suited to
HF and VHF applications. For both ferrite and powdered Iron types, the
coil resonant frequency will drop as the core is inserted.
It is important to ensure that each coil has the correct type of core,
and that the length of the core is as originally specified by the
manufacturer. Usually, coil assemblies that look untouched need not be
investigated further. Coils that look to have been the source of past
problems, should be surveyed to see if the right type of core is
fitted. This can be difficult without reference to a gold standard
radio.
Cores that do not move as they should, can cause major headaches.
Before attempting anything potentially hazardous to the health of the
former or its windings, some basic measurements need to be made so that
if the need arises, the coil assembly can be replicated. The complete
coil assembly should be removed from the chassis. Disconnect any
capacitors, and measure each winding in turn with all other windings
disconnected, to determine their individual inductances. Note the wire
gauge, wire type and the relative sense of all the windings. Then if
possible, measure the mutual inductance. Note whether the windings are
lap wound, wave wound, pile wound. Or indeed, a mixture of these.
Finally, be sure to measure the length and diameter of each bobbin, and
record the physical distance between the windings.
The most troublesome cores are those which have a hollow hexagonal
bore. These cores look very robust but actually, they are very prone
to shearing longitudinally. Removal techniques for jammed or broken
cores include eating away at the material with a small drill, or
creating a new screwdriver slot by scraping the surface with a scriber.
In the case of hex cores it is sometimes possible to drift one of the
broken flutes down and out of the coilformer, after which the others
can usually be removed very easily by careful use of a hatpin.
Routinely check the integrity of all mains wiring and insulation using
a proper Megger or other high voltage insulation tester. Do not try
this test using a DMM, as the results may indicate the equipment to be
safe when in fact it is not. If in doubt, take the radio to somewhere
equipped with a calibrated PAT machine. You could try your local tool
hire shop.
American radios are often supported by handbooks written only for the
home market, where the transformer had a single 117V primary. The
export sets were often referred to as the “universal” models.
(Warning: in the UK this would often mean an AC/DC variant). When
operating American universal sets from 234V, the mains fuse should
normally be halved in value compared with the figure in the book.
Change cable grommets where they have hardened, even if this is highly
inconvenient to do.
Many awful types of mains chassis connectors have been used over the
years. Strip out anything suspicious and fit modern components, even if
this further worsens originality. What's at stake is worth far more
than mere cosmetics.
Make sure the main chassis of the radio is earthed, even if originally
it was not designed to be. For AC/DC sets, or other live chassis
designs referred to in the UK as “universal” radios, it is prudent to
use an outboard isolation transformer to achieve this objective.
This writer advises that in all cases, the mains tap be set to 10V
above the prevailing mains line voltage.
When the set is first running, keep an eye on the mains transformer
temperature. These things can run warm, or on some American radios,
hot. They should never smell or smoke, or have core temperatures
exceeding +90ºC. The large thermal mass means that the transformer
core temperature may take several hours to stabilise.
In the case of so-called “line cord” radios, which were mostly cheap
American midget domestic or enthusiast sets, measure the mains cord
resistance carefully, examine the condition of the insulation and take
the right decision about its future. The original length of line cord
cannot be shortened, of course. Do not simply replace the line cord
with ordinary mains cable, and be especially careful with the mains
plug connections of these sets. Some had two wires to be connected to
the Live pin, one to the Neutral and none to the Earth. All too often,
the chassis of the radio went live as the set was switched off, a
thoroughly nasty characteristic. This writer is inclined to condemn
line cord radios on sight, but fully acknowledges their collectability.
In the writer’s own collection, all pre-war mains radios are routinely
modified to have an over-voltage varistor wired directly across the
power input. Most such radios then have a Brimar CZ3 or similar
negative temperature coefficient thermistor fitted in series from there
on into the radio. This arrangement will hopefully protect the old and
fragile mains transformer from grossly excessive mains transients, and
allow a slightly easier warm-up for the bulbs and filaments every time
the radio is switched on from cold.
9.4 HT Insulation
9.5 Re-impregnation
This technique can also be used to mend coaxes which have developed
short circuits between the inner and outer conductors. This is
especially likely to happen at the connectors to valve top caps.
Plastic wiring sometimes goes brittle with age, especially where it has
been heavily doped with an MFP compound. Sometimes shorts develop
between different PVC insulated conductors in the loom, if it had been
laced too tightly when new. This is because PVC is a plastic material
which creeps out of the way in time, leaving the conductors exposed.
PTFE will only occasionally be found on radios of the valve era, but
the comments regarding creep are even more relevant to this material.
PTFE has the added problem of being very easily nicked by the knife
whilst being prepared for tinning, or by sharp pieceparts whilst the
loom is being positioned in the chassis.
Varnished canvas and fibreglass insulated wire rarely if ever gives any
trouble at all.
Still on the bench, bundle the new loom initially with tywraps, then
lace it properly with new braided nylon cord or string. Start at one
end, and remove each tywrap as you go along. This work is not
difficult, but will be time consuming if you have never done it before.
Be patient, and work carefully. Aim for perfection. This is one job
where you can achieve it, given time. Each loop should be
independently knotted. Do not use tywraps or spiralwrap anywhere on
the final loom. These things look awful inside an old radio, and are
very eyecatching.
Remove all solder and old wire ends from the various connecting lugs on
the chassis. Then strip and tin the cable ends of the new loom, and
offer it up into position on the chassis. If you have done your work
properly, the new loom will fit perfectly. Reuse the original P-clips
if you can.
If only partial rewiring is needed, snip each loop of the lacing cord
back as far as you need to go, and carefully feed out the old rotten or
burnt wiring complete with its lacing. It is best to remove complete
wires if you can. Then fit new wires into the loom. Where joins are
essential in mid-loom, use a Hellermann sleeve to cover each solder
joint. Everything has to be done in situ, which is why the use of
heatshrink sleeving cannot be recommended. End by re-lacing the loom
using the original positions for each loop of the cord. Partial repair
of a loom like this very often takes longer than fabrication and
installation of an entire new loom assembly, and never looks half as
good. Except in the case of very localised damage, this writer
recommends complete replacement of damaged looms, every time.
Suitable wire for most radios will be PVC covered 16/0.2mm for signal
and HT circuits, 24/0.2mm for runs feeding up to 3 filaments, and
63/0.2mm for high current runs to the filament transformer of a big
receiver. Umbilical power cables for the 358X, HRO etc should use
63/0.2mm for the filament, and 24/0.2mm for HT+ and HT return. The
colour brown looks unobtrusive in most radios.
One curious feature of Collins and Marconi radios is that they used
screened cables that had no insulating sleeve over the braid. It is
necessary to take particular care not to let the braiding fray, or lie
against any live metalwork.
9.7 Motors
Most problems occur not with the motor itself, but with the driven
mechanism. As a rule of thumb, a worn motor will indicate a much more
severely worn geartrain. Indeed, before purchase of a motor tuned
radio it is well worth a good search for metal filings near the driven
gears. You can form your own conclusions, if any are found.
9.8 IFTs
The primary of the first IFT may have a higher impedance than the other
IF windings in the set. This is especially likely with radios using a
6K8 or Mazda/Ediswan 6C9 mixer, and was considered necessary to give
the best conversion gain with these types.
In contrast, the windings of the final IFT may have lower than usual
impedances, in order to efficiently drive the usual AGC and signal
detector diodes, which present low impedance loads.
In battery sets, QPP output stages would normally use a step-up driver
transformer to feed drive voltage into the high impedance control grids
of the output pentodes. In contrast, class B sets would use a step-
down type to drive positive grid current into the zero bias triodes.
Both of the transformers here will ordinarily be of the unbalanced to
balanced type.
In general, the American military suffix-W valves are better than the
average commercial grade, though they are not necessarily any more
reliable. In this respect they are similar to the CV100 series used in
the UK.
The WA, WB suffix American military valves will each safely replace
lesser suffices, but the converse is not true. For instance, a 6AU6WC
will replace a 6AU6, 6AU6WA or 6AU6WB in all applications, but a 6AU6WA
will only replace a 6AU6 and should never be used to replace a 6AU6WB
or 6AU6WC. Some of the suffix upgrades affected secondary parameters
such as warmup time, heater/cathode insulation withstanding voltage, or
the mechanical resonant frequency of the electrode structure.
Frame grid valves have an extremely small spacing between the cathode
and the innermost grid, which is usually the control grid. This was
necessary to obtain higher mutual conductance than was otherwise
obtainable. An unwanted side effect of this construction is that these
valves are far more sensitive to mishandling or electrical damage than
non frame grid types, and they seem noticeably more prone to develop
intermittent faults. Valves of the frame grid type and certain others
such as the 6SG7 and 7H7 are rather prone to developing control grid
emission, especially if the heater voltage is too high. This problem
can cause very poor AGC action if the affected valve is somewhere in
the RF or IF amplifier chain.
With side contact valves (CT8, type E), it is important to make sure
the valvebase has no loose contact rivets.
Bearing in mind that so many of the older valve types were basically
common bulb/electrode assemblies mounted onto a family of alternative
bases, it may be possible to “make” a new valve, if you have a source
of donor pieceparts. This writer has successfully grafted the
bulb/electrode assembly from a new spare octal Brimar 6F6G onto the UX
base from a defunct RCA 43, to create a 42; and converted an octal
AZ31 into a side contact AZ1 using the base from a dud ECH3. The work
is rather fiddly, and you must perform a comprehensive set of tests to
verify that there are no inter-electrode shorts before you put the
“new” valve into service. The best way to remove the bulb from an
octal base is to start by loosening it rotationally, wearing gloves and
goggles and with all parts of the valve held inside a big rag in case
of breakage. Then one by one, gently bend each pin in the bakelite
moulding back and forth until it snaps at the root, making sure each
wire remains attached to its pin. Then melt the solder at the end of
each snapped pin and pull it off the wire. The next step is to paint
the end of each wire in a distinctive way so that it can be uniquely
identified later on. After doing this 8 times or as necessary, the old
bakelite base should pull straight off. Then the bottom of the bulb
can be cleaned up, and all wires extended and sleeved as necessary
ready for insertion into the new base. Araldite works fine for the
bond, but do be careful in the case of magic eyes to align the bulb
onto the base in the correct angular position. A quick check on a
valve tester will confirm that you have connected the wires to the new
valvebase pins in the right order.
In the case of those valves that are no longer obtainable from anywhere
at any price, or where cathode rejuvenation is inappropriate, repair of
the original valve is the only logical alternative to the manufacture
of a new one, unless rewiring of the socket for a slightly different
type is considered acceptable. Many faults, such as a broken cathode
connecting strap, or a heater wire that has snapped at its weld to the
pinch leadout wire, would appear to be very simple to repair given the
necessary trepanning, welding, pumping and gettering equipment. The
writer would very much welcome contact with anyone with experience of
this type of work. It is noted that unlike large transmitting tubes,
the valves used in receivers were never designed to be rebuilt after a
period of service. Hence the envelope glass may well be a type
fundamentally unsuited to repair.
Loose top caps are best unsoldered and removed. Clean the cement
surface, but do not attempt to dig it out unless it is loose, in which
case be sure to remove all unattached pieces. Then apply a bead of
Araldite to the pip without disturbing the lie of the wire, which is
quite fragile. Immediately refit the cap, push it home and quickly
solder it. Do not disturb the assembly for a day or so.
The term “valvebase” is easily the least confusing of the various names
used for the bakelite base forming the bottom part of many older style
valves, and which carries the male connecting pins or protruding side
contact lugs. They are found on standard glass octal, tubular octal,
UX, British, CT8 etc valves.
The best way to fix a loose base is to invert the valve, squirt some
solvent based contact adhesive through one of the spare holes in the
bakelite, then try to swill it round a bit. Finally, store the valve
upside down with an elastic band squeezing the bulb into its base for a
couple of days. Any adhesive that dribbles through the gap between
envelope and valvebase, can be trimmed off with a scalpel after it has
set.
A slightly more unsightly repair, for use mainly when the valve has no
spare pin holes, is to run a fillet of Evostik between the bulb and the
valvebase. Then apply the elastic band and store the valve the right
way up for a day or so. If using this technique, you are recommended
not to attempt trimming of surplus adhesive, because it will leave a
mess and weaken the joint.
In general, the early grey coatings are very prone to peeling off. To
make things worse, this type of coating can get so dirty as to
completely obscure the markings of a VP41, AC/VP1 etc. The later gold,
black and red coatings were much better in both respects.
The metalized coating should be well grounded. Even a few ohms can
cause incipient or actual instability in RF and IF stages.
Because the metalizing is very brittle and because the earth contact
wire or clip is captive to the valvebase, it is obviously necessary to
make sure the bulb is not at all loose in its base. If it is, fix this
problem before attempting to restore the connection from the valvebase
shield pin to the metalized coating.
On most old radios, valve cans are usually fitted somewhere or other.
They are all there for a purpose, therefore they should all be fitted
- except in some American equipments, where certain specified cans
should be removed for static rack mounted service. Check which valves
should have cans fitted, for your particular type of installation.
Do not assume that cans should always be fitted wherever a skirted
valve socket is to be found. This is not always so. The 12BY7A
transmitter driver in the Yaesu FT101 family is a good example.
Red cans are often of heavy µ-metal material. They were sometimes
fitted to valves operating at high audio sensitivity such as the EF86
to reduce hum, or valves which are susceptible to AC magnetic hum
fields such as the 7360 mixer, where the objective was to reduce hum
modulation of the signal. Another application was on battery-powered
radios fitted with integral loudspeakers. The static magnetic leakage
field from the magnet could damage certain of the more fragile B7G 1.4V
filament types.
The Hammarlund SP600 uses a special can for its 6C4 first local
oscillator. This has a special lug which must be screwed down onto a
support pillar, to provide proper mechanical support and to guarantee
efficient grounding of the can. The objective was to improve frequency
stability and possibly also, to reduce local oscillator radiation.
The best cans for B7G and B9A valves are the contact-cooled types made
by IERC such as the TRNC6000 family, which was licence manufactured by
Garrard in the UK. These can sometimes be fitted as an upgrade, in
place of the standard type. Avoid fitting contact-cooled cans to
barretters or neons, though. Marconi were quite fond of using an
elaborate fully screened can which contained wire wool to get the heat
away. These cans screwed down onto a special threaded ring which
surrounded the valve socket. These things cannot be used as
replacements for ordinary valve cans. One side effect of all these
cans is to feed heat from the bulb to the skirt of the valve socket,
and hence down onto the chassis. This is generally not a problem.
However, care is needed with VFOs, which may drift more with a heatsink
can than they do without because of increased component heating near
the valve socket. Another relevant fact is that for valves without an
internal shield around the anode cylinder, heatsink cans increase the
anode capacitance to ground more than ordinary types.
Certain American radios used spring fingerstock inside B7G and B9A
valve socket skirts, presumably to support the valve and promote
cooling. If originally fitted, all such hardware should be present and
correct. Make sure none of the spring fingers has become detached and
fallen on to the top of the valve socket insulator or got jammed
alongside one of the socket contacts. This problem is frequently
encountered.
Some radios by Philips and Heathkit and perhaps others, used push-on
cans which connected to a grounded metal finger running up alongside
the valve envelope. Sometimes the finger was intended to be trapped
between the valve and its can, and sometimes it pressed onto the can
from the outside. In either case, it is important to make sure the can
is properly grounded, so that it acts as an effective RF shield.
The Marconi B8G ‘80 series (like loctals) and B9B (EF50) valves used
all-metal valve retainers which often need new side springs, but little
else ever seems to go wrong with this pattern.
10.6 RF Amplifiers
The worst option of all would be the use of a straight frame grid valve
such as an EF95 to replace a big old variable slope standard type such
as a KTW61. The difference in capacitances and mutual conductances
between these two families is so great that correct tuning would be
difficult to achieve on the HF bands, at least with adequate stability.
Even if this condition were found to be obtainable, the AGC
performance would now be poor because the front end would be trying to
perform almost the whole of the variable gain function on its own
whilst lacking the intrinsic linearity to cope properly under strong
signal conditions.
Having said all of this, some radios certainly do benefit from a modest
upgrade. For example, the SP600 gives 10dB NF on 6m with a 6DC6,
compared with 12dB with the original 6BA6. No circuit changes are
needed, the stage is totally stable under any conditions of source
impedance, and the two-signal performance is only slightly worsened.
In designs that use high gain or frame grid construction front ends
such as 6CB6, 6DC6 and 6BZ6, the valve will often be found distinctly
flat, giving poor HF sensitivity and worsened intermodulation
performance. These high performance valves also seem more prone to
inter-electrode shorts than lower gain types with larger electrode
separations, especially when the valve envelope is mounted horizontally
in a way which allows sagging of the electrode structure.
If any nuvistors are missing, and direct replacements are not traceable
or affordable, you may wish to consider the use of an available
miniature wire ended glass type such as a 5840/EF732 with legs bent to
fit into the original socket. The new valve will be taller than the
original, so support it at the top of the envelope using a grommet.
Keep the wires very short, but do not bend them within 1.5mm of the
glass seal. Be particularly careful not to damage the valve socket,
otherwise you may have problems fitting a new nuvistor, if you ever
manage to get one later. It may prove possible to open the old broken
nuvistor and use its bottom end to anchor the new component.
At the risk of being branded a heretic, this writer would point out
that it may be possible to use something like a TO92 packaged N-channel
JFET to replace acorns or nuvistors in certain situations, especially
where the operating potentials are low. For a little while, Fairchild
and possibly other manufacturers too, produced pin compatible non-
thermionic replacements for certain nuvistors. These devices were
called “Fetrons”. The list of available types was small, but did
include the ubiquitous 6CW4.
10.8 Compactrons
These were used in some American radios but not for long, and their
production life seems to have been quite short.
10.9 Mixers
The key to healthy mixer operation is to use the right valve - not a
near equivalent - and feed it with the right amount of local
oscillator injection. Too much causes whistles. Too little causes a
lack of conversion gain, and hence low sensitivity. Although
multielectrode mixers have rather complicated surrounding circuitry,
surprisingly little usually goes wrong with them. most types remain
widely available. In general, it is acknowledged that the octal 6A8,
6K8, 6J8, ECH35, 6TH8G, 6E8, X61M, X65 & X66 types are all broadly
interchangeable despite various differences in electrode constructions.
Care is needed with the 6TH8G as it is physically huge, and draws a
large heater current. The 6SA7 and 6L7 types are deliberately not
included in this list, since the mechanism of their operation is very
different from the valves on the list - and from each other. In
addition, these two valves are not pin compatible.
In the loctal family, the 7S7, 7J7 & X81 are all broadly
interchangeable.
The Mazda/Ediswan 6C9 and the far more common ECH42 are also broadly
interchangeable, though these types have very different filament
current ratings. Some adjustment may need to be made to the triode
grid resistor value if this substitution is made. Note that the
American 6C9 is an entirely different valve from the British
Mazda/Ediswan, the latter being constructed as a sort of miniaturised
6K8.
The 6E8, ECH35, X61M & X81 types of mixer are unusual in that the mixer
section is a hexode rather than a heptode. In this respect, these
devices are similar to the 6K8, though of more conventional
construction. All of these valves rely on a large anode-screen spacing
to prevent secondary emission distorting the Vg1/Ia transfer
characteristic.
The 6JH8 and 6ME8 beam deflection have an extra grid compared with the
7360 and work rather better as receiver mixers. They are cheaper, too.
With all types of beam deflection mixer, the local oscillator
injection needs to be at a very high level, perhaps 10V RMS or even
more according to design. Stage gain is proportional to excitation
over an extremely wide range. The local oscillator injection may be
applied to the deflection plates differentially, or in single ended
mode. Balance of the DC current paths within the valve itself is
necessary, and this is invariably achieved by slight differential
adjustment of the deflection plate DC voltages. The IF output from the
mixer stage may be taken from the anodes differentially, or in single
ended mode.
VFO valves sometimes had a different part number from other valves of
the same type used elsewhere in the set. As an example, the 6BA6s used
in the VFO of the 75A-4 were allocated Collins part numbers different
from the 6BA6s used in the IF strip. This may indicate that these
components were pre-aged, had unusually low heater/cathode leakage, or
were in some way selected to be especially suitable for the job.
Perhaps they were sourced from a different manufacturer.
By way of contrast, a weak oscillator valve will cause low drive to the
mixer leading to insensitivity of the receiver. Alternatively, the
oscillator may drop out or fail to start correctly, somewhere in the
coverage range. Weak LO valves also tend to cause seemingly endless
frequency drift.
Reactance valves are used for small frequency changes. These have
variable DC voltage drive, and are typically strapped across a local
oscillator tuned circuit.
10.12 Calibrators
10.13 IF Stages
IF strips often use several examples of the same type of valve. The
final IF amplifier has to develop the most signal power. It often gets
little AGC, in order to suppress unwanted modulation rise. Therefore
it is wise to fit the best specimen you have in this position. A weak
valve here may cause an unacceptable amount of distortion on strong AM
signals.
A curious feature of 6BA6s and 6SG7s is the frequency with which they
are found to be very flat indeed, even though the radio itself may have
been working fine with them fitted.
Beware the Mullard EBC33. For unknown reasons, this type is very prone
to problems with its diodes. Inter-diode shorts are known, as are
anode-to-cathode shorts. The problem seems not to be found on EBC33s
from other manufacturers. For instance, the wartime military black
metalized EBC33s from Rogers in Canada seem almost indestructible.
AM detector valves, and product detector valves may need to have low
heater/cathode leakage to avoid hum problems. This is particularly
true in the case of receivers with series-connected heater chains.
Some pre-war battery sets used indirectly heated double diodes such as
types 2D2 or 220DD. These are now believed to be completely
unobtainable. The best answer in event of problems would appear to be
to use a semiconductor diode or two, leaving the original valve in
position for the sake of appearances.
In some cases, µ-metal cans were fitted to the detector and/or the
noise limiter. These were often painted red for easy identification.
These cans are heavier than the usual type and have the necessary soft
magnetic characteristic.
Sets using delayed AGC may suffer from “differential distortion” of the
audio as the AGC diode drops into and out of conduction on a weak
signal subject to fading. Quite a lot was made of this phenomenon
years ago, and some very clever circuits were devised to compensate for
the problem. The cleverest approach was to provide a level of forward
bias for the detector diode which was proportional to the strength of
the received signal. In the opinion of this writer these complicated
circuits are of esoteric interest only, since most detectors work
perfectly well enough for communications purposes when correctly
restored to original specification.
Radios having double AGC systems obviously have about twice as much AGC
generation hardware to restore as the usual variety. There are no
special pitfalls to avoid during restoration. The Hallicrafters SX28
Super Skyrider has a complicated dual AGC system, operating at
different bandwidths. Despite the grand claims in the promotional
material relating to this receiver, there seems little practical
benefit to the user. Each AGC system works so well that the writer’s
example works fine with only one (either will do!) AGC loop being
connected - a useful test of performance.
The main design problem with magic eyes is the loading they put on the
signal detector, in the many circuits where the magic eye is fed from
this stage to avoid the effect of AGC voltage delay. Little can be
done about the resulting AM distortion, except to use a large grid
series resistor (in comparison with the diode load), together with a
low leakage decoupling capacitor on the control grid of the eye
amplifier triode section. To avoid the AC loading problem, some of the
more expensive sets used an entirely separate detector to feed the
magic eye without any voltage delay. Some of these radios put the main
magic eye on the signal detector to help tuning on weak signals and had
a second, reduced sensitivity, magic eye running on AGC for use on very
strong signals.
Special mention needs to be made of the pre-war Philips magic eye type
EFM1, which was a combined AF pentode and magic eye used in sets like
their 735A. The EFM1 was also made by Tungsram, and by French
manufacturers. This valve required its pentode section to be connected
to the signal detector for DC as well as AC purposes, which meant it
automatically introduced feed-forward AGC. This in turn, meant that
the audio amplifier needed heavy feedback to reduce the curvature
distortion to acceptable levels. Some good news with the EFM1 is that
there were no problems with AC loading of the signal detector caused by
the magic eye section. Nowadays, the main problem with type EFM1 is
one of finding supplies. Though produced by Philips in large numbers,
this type is now very rare and correspondingly expensive. One
perfectly good reason for this scarcity is the lack of light output
resulting from heavy inking inside the viewing end of the valve, at
least on early production examples. Frequent replacement would have
been necessary. Post-war EFM1 production had an ordinary clear viewing
window, and these later devices are a lot brighter. In the event of
non-availability, the best approach would be to change the socket to
suit an ordinary octal magic eye such as the 6U5G, and then to mount a
wire-ended variable slope pentode such as a 5899 directly on the socket
tags. Alternatively, it should be possible to rewire the base to suit
an EF9, and completely do away with the magic eye functionality
altogether. The EFM1 was also made on the German octal base. This
valve was called type EFM11. It is just as rare as the EFM1.
To obtain the correct shadow pattern at zero signal input, the cathode
of the magic eye was often tied slightly above chassis potential,
typically to a point part of the way up the cathode string feeding the
receiver audio amplifier or output stage. In the event that an
incorrect shadow pattern is found, look at this area of circuitry
first, and only then move onto the anode and grid supplies.
The main restoration problem with magic eyes of all types, is lack of
light output. This has led to a distinct shortage of certain types due
to the need for frequent replacement in the past, combined with modern-
day collectability. The problem usually seems to be caused by phosphor
burnout rather than lack of cathode emission. This is shown by the
fact that many weak magic eyes still glow brightly from the part of the
phosphor that is only illuminated under extraordinarily high levels of
input signal. Interestingly, many older magic eyes took a lot of
target current - for example the Mullard type TV4. Sometimes the
current drain was very variable between samples. This can lead to
significant differences in light output even between new and unused
valves of the same type.
One clever receiver once owned by this writer was made by the Swiss
manufacturer Paillard, being their model 59. This only energised the
target electrode of its UX6-based 6E5 magic eye to full HT when the
maximum selectivity position of the IF strip was selected. In the
other positions, the valve received rather less than 200V and glowed
only dimly. The original magic eye had remained in service for 57
years, and was still in very good condition. This intelligent piece of
design has not been seen on any other receiver known to this writer.
Much confusion surrounds tubular octal types EM34 and 6AF7. Both are
European magic eyes of dual sensitivity, the EM34 being directly
equivalent to American type 6CD7. The 6AF7 is a French type. Despite
its number, this is not an American valve. The military CV394 is
supposed to be equivalent to commercial type EM34. Now consider this:
the writer has three new & boxed Amperex (a USA brand owned by
Philips) valves which are actually type 6AF7 even though they are
marked EM34/6CD7. The Soviet bloc never seems to have made copies of
the genuine EM34. So: what are the differences? They can be
summarised as follows:
How much these differences mean in your radio, may well depend on
whether the heaters are wired in series or parallel, and whether any
provision exists for rotating the magic eye relative to the cabinet.
Some sets that were designed for a genuine EM34 cannot mount a 6AF7 due
to their inability to accommodate the larger bakelite base.
Noise limiters are very often troublesome. Most benefit from selection
of the limiter valve for minimum hum. Exceptions are found in some
Collins equipments (eg. 75A-4) and some AR88Ds, which operate the noise
limiter valve under current-starved heater conditions, and some
Eddystones (eg. S880/2), which apply positive bias to the centre tap of
a separate noise limiter heater supply rail.
One problem with the Eddystone approach is that because the limiter has
a dedicated filament winding, it gets full voltage at the moment of
applying mains power. This can cause the cold filament to flash
brightly, with the result that open circuit filaments are commonly
found.
Audio output valves run hot in normal operation. They get even hotter
if the grid coupling capacitor goes a bit leaky. To check a grid
coupling capacitor, briefly short the anode of the audio driver valve
to earth with a screwdriver, whilst measuring the cathode voltage of
the final stage. There should be no change in reading greater than
100mV. This test only applies to output valves having automatic bias.
Gas problems are cumulative, and must never be ignored. Excessive heat
causes gas that was previously trapped in the metalwork of the
electrode structure, to be released into the vacuum inside the envelope
of the valve. This gas then ionises with a positive charge and the gas
ions wander towards the most negative electrode, which is the control
grid - pulled there by electrostatic attraction. On collision, this
causes grid current to flow. If the grid leak is much over 470kΩ, the
negative grid bias can reduce appreciably because of ion bombardment.
The anode current then rises, so that even more heat is generated in
the plate and screen grid. This process can happen cumulatively during
the first few minutes of operation, causing a steadily rising HT
current. Even more gas then gets liberated. The valve soon goes
“soft”. The situation escalates until the HT fuse, valve, output
transformer, HT choke, rectifier or mains transformer fails. Some of
the gas ions may collide with the cathode instead. If severe, this can
lead to the mechanical fragmentation of the oxide coating.
The vicious spiral can appear in American sets with fixed grid bias
lines, but is seen at its worst in British sets using high gain output
stages operating with little bias. The problem is relatively common in
AR88Ds using the 6K6GT output valve, and is endemic in AC/DC Eddystones
using the high gain UL41.
Gas ionisation can be seen as a pale mauve glow inside the metalwork of
the electrode structure. It must not be confused with fluorescence,
which is a sign of unusually good vacuum. This is seen as a dark blue
glow just inside the glass wall of the envelope. Sometimes the inside
of the glass is sprayed with graphite to stop this secondary emission
from the bulb - or at least, to prevent it from being visible! Osram
and Brimar were very inclined to use a graphite coating. Mullard and
Mazda were not.
The DC voltage drop across each half of the output transformer primary
should be within 10% at no signal. Note that because of the way it is
constructed, the DC resistances of each winding will often not be
identical, even though they have the same number of turns.
Check that in actual service, both output valves get equally hot and if
not, investigate. One of the valves may be flat, or oscillating at
VHF. Be particularly suspicious of any crackling distortion from the
loudspeaker, sounding for all the world as if the speaker voice coil is
rubbing on the polepiece. This is often an audible symptom of unwanted
RF oscillation in the output stage.
One problem unique to push-pull, and the reason this writer dislikes
the use of this technique in communications receivers, is the variation
in HT current with signal at medium and high volume levels that occurs
unless the stage is operated in true class A mode. As the valves start
to draw extra current on peaks, he HT voltage then becomes modulated by
the bass content of the audio. On the higher RF frequencies
particularly, this can cause the local oscillator to suffer severe
pulling or fluttering. This is especially true if no neon regulator is
fitted, and when listening to SSB transmissions. Bias problems can
cause HT current variations greater than intended, and a careful check
of grid and cathode voltages is an essential early step in the overhaul
of any push-pull output stage.
Battery sets often had intricate QPP or class B output stages. The QPP
type used a pair of pentodes biased almost to cutoff, with grid
excitation usually coming from a step-up voltage transformer. Class B
designs used a pair of zero bias triodes which were driven into
conduction by feeding their grids from a step-down voltage transformer
which was capable of providing significant AC drive current.
10.18 HT Regulators
Neon regulator tubes are usually trouble free, but not invariably so.
Quite often, especially with S130 and 85A2 types, the operating voltage
will be found too high. There is no alternative to replacement of the
valve. Fortunately, most types of neon are still cheap and plentiful.
AF oscillation of the regulated line is sometimes seen, typically
causing unwanted sidebands on the local oscillator. This is usually
caused by excessive capacitive decoupling of the regulated rail. Make
sure a previous owner has not inserted an electrolytic across the neon
in a misguided attempt to improve performance.
A final caution for neon regulators concerns the suffix WA (US military
ruggedised) variants of OA2 and OB2. These contain a radioactive
isotope, which was intended to promote ionisation, and hence lower the
striking voltage. The activity level and half life of this isotope is
not known to the writer.
Rather than use inert gas voltage stabilisers, some German designs used
a solid state rare earth regulator. These are normally reliable, which
is just as well since these devices are nowadays very difficult to
obtain.
Some valve radios used a Zener HT regulator diode for the VFO. The
RA117E is one example. These early devices were unreliable and should
be checked very carefully. Replacements can be difficult to find,
especially as the stud needs to be the anode (negative) pole of the
device. A quick catalogue search revealed that a 150V 5W Zener, with
its anode connected to the stud is not available nowadays. An OA2 may
have to be fitted instead.
Rectifier valves should always be checked for the presence of any bits
of oxide cathode coating floating around loose inside the envelope.
This is usually due to a past flashover caused by an HT short or an
extraordinary mains transient. Some loose oxide is OK but a lot of it,
or a big bald patch on the cathode, would condemn the valve.
Some rectifiers take ages to warm up, especially Mullard type CY1.
This is not necessarily indicative of a fault. Thickwall cathode
tubing may have been used, to make sure the rest of the valves in the
radio all get a chance to warm up properly before the HT appears.
One fact about rectifiers that puzzles this writer, is the great
difference in HT voltage developed by different samples of the same
type. Some radios such as the Hallicrafters S36, are generally short
of HT and would benefit from a really low loss rectifier. Others like
the Eddystone S940C, produce excessive transformer secondary voltages
and are at their best with a low output rectifier. In the case of
these radios, special care to select an appropriate rectifier is a very
good idea.
In general, Silicon devices have stood the test of time well but
Germanium types have not. Modern hazards include intermittency
developed as a result of metal migration on the surface of the chip,
noisiness (usually with pronounced leakiness as well) due to
inadequately clean conditions during manufacture, and short circuits
developing between the electrode structure and the outer casing. In
particular, members of the Mullard AF10x and AF11x series of RF/IF
Germanium alloy transistor nowadays often suffer from low gain, and/or
shorts from collector to case.
Most early devices are still available very cheaply in 2005. For low
level audio applications, you may need to try a selection to get an
acceptable noise level. Unlike valves, just because these early
semiconductors have not been used does not stop long term problems from
developing. Remember to use a heat shunt when soldering Germanium
devices, because these things are thermally fragile.
Use HMP alloy solder to connect to the wires and lugs of hot running
dissipative components like wirewound resistors and heater thermistors.
The iron should be a medium/large type with a tip temperature of
+450ºC or thereabouts.
Use LMP alloy solder to attach the leadout wires of new components to
their termination soldertags, using a medium/small iron running at
about +350ºC. This will avoid excessive reflowing of the original
soldering material.
Most radios were originally constructed with each joint being wrapped
before soldering. However some receivers, especially Eddystones, used
laid-on joints that were not wrapped. They seem none the worse for it.
This really does open up the question of whether it is worth the
bother of unwrapping and rewrapping the solder joints during repair of
all the other makes. It must be accepted that ideally however, the
repair should use the same technique as was employed in the original
manufacturing process.
Wick seems to work better than a plunger sucker for removing old
solder, but there is great variability in the quality of the wicks on
offer. Don't keep too much in stock for long periods, as this stuff
loses efficiency quickly when the braid starts to oxidise.
11.2 Resistors
Check in particular, the values of all cathode & screen resistors, and
all carbon HT dropper resistors. If more than 20% away from the
nominal value, replace the component. Where sufficient space exists,
use the next power rating up from that originally specified, to stop
the problem happening again in the future.
Many of the better quality radios have low wattage series resistors
fitted in the HT feed to each stage, typically 2.2kΩ/¼W. These all
need to be checked carefully. If darkened, burned out, or incorrect in
value, first find out why. Then replace the component with the same
value and wattage as originally. These resistors are intended to act
as telltale fuses to identify problems downstream in the circuit
diagram. They will not do this job properly, if you uprate their power
dissipation capability.
Dubilier grey and brown power resistors with fibrous bodies should be
condemned on sight. These components are extremely prone to going high
in value, and they can get very noisy in old age, too.
Check that the AGC circuit resistance exceeds 100MΩ when all shunt
resistors are disconnected, but with all of the AGC capacitors left in
circuit. This is a good opportunity to measure each of the shunt
resistors individually, while they are temporarily disconnected from
the rest of the circuit.
This writer recommends that you change all electrolytic capacitors more
than 25 years old. Be especially diligent with red and yellow plastic
cased wire-ended Plessey types, and all others that show signs of
bulging or cracking of the rubber endcap or where there are signs of
electrolyte leakage. American radios were inclined to use octal plug-
in reservoir and smoothing capacitors. These components are usually of
excellent quality, and need not routinely be replaced. The only
problem associated with these things occurs when the base or socket
contacts get a bit tarnished. This can cause mysterious hum and/or low
rectified HT voltage.
This writer recommends that you avoid the use of NOS electrolytics. If
you really must use one of these things, you are advised to reform the
internal insulation of the component first by a slow, progressive
buildup of voltage, fed through a high value current limiting resistor
to prevent excessive gas pressure buildup.
German and Russian equipment running from AC/DC mains supplies tended
for some years to be fitted with an Urdox (Uranium Dioxide) filament
current regulator. This device comprised a conventional Iron/Hydrogen
filament barretter section, in series with which and inside the same
glass envelope lived an NTC thermistor made out of Urdox material.
This should in principle have prevented the usual switch-on flare of
the iron filament, thus prolonging the life of the barretter section.
A clever detail is that the thermistor element was located immediately
above the Iron filament, acting to ensure a very low running resistance
for the solid state element.
11.6 Bulbs
Bulbs require a lot more care than might at first be expected. Part of
the problem is finding the things on the right base, in the right
voltage/current combination, with the right envelope style, in a
quantity sufficient for your chassis - plus a few spares.
For the sake of appearance and longevity, it is often best to fit bulbs
rated at 12V, to radios having filament systems running at 6.3V, in
order to avoid excessive illumination. For series connected bulb
chains, this is very good practice, otherwise manufacturing tolerances
may mean that one or more of the bulbs could fail prematurely due to
voltage hogging. Keep an eye on the voltage distribution if you use
this trick on AC/DC sets.
11.8 Potentiometers
Use Cermet or wirewound replacements for any components that carry DC,
as in all BFO and VFO fine tune arrangements and some squelch and
volume controls. Suitable replacements can often be quite difficult to
find, especially potentiometers having reverse log (antilog) tapers, or
taps along the track for tone compensation.
For all Collins types, both end coils should have the same DC
resistance, often about 50Ω. The end windings should be insulated
from each other, and also from the outer casing. Collins mechanical
filters can go slightly lossy and develop passband ripple, but this can
be difficult to diagnose without a gold standard available for use as a
comparison. These components are not repairable, and can be extremely
difficult and expensive to replace. Make sure that any capacitors
responsible for keeping HT off the filter coils are the very best types
you can find.
11.11 Thermistors
Do not leave in place Silicon rectifiers, where these are found bridged
across thermionic types by a previous owner. The voltage developed
during the first ten seconds can be high enough to damage the
insulation of wound components. The subsequent high running voltage
can over-stress the output stage and output transformer.
The 0Z4 fullwave gas filled rectifier had a period of vogue in radios
powered by asychronous vibrators. This type of component is
electrically fragile. It is intolerant of incorrect reservoir
capacitance (it should be 8uF or thereabouts), and this rectifier will
not survive open circuits or short circuits for long. The 0Z4 requires
a minimum of 330V peak starting voltage per anode, and must deliver
between 30mA and 110mA to the load, otherwise service life will be very
short. Thus if your radio has been running with a defective vibrator,
or is a low current design that has operated for some time with
slightly flat valves, you may expect to find the rectifier in poor
condition. Despite not having a filament or heater, these devices do
take some time to get going from cold. This is not in any way
indicative of a faulty device. In normal use, the cathode becomes
incandescent as a result of ionic bombardment, and then emits electrons
in the normal way. Type 0Z4 is considered to have ended its service
life when its forward voltage drop exceeds 25V at +70ºC ambient
temperature.
11.13 HT Vibrators
Both synchronous and asynchronous types were used. Vibrators were very
popular for low powered portable receivers running from batteries,
where the current drain did not justify the use of a dynamotor, or
where space or weight was limited.
Many types were sealed in cans where the guts were supported by sorbo
rubber. Years of disuse will probably have caused severe tarnishing of
the contacts because of the sulphurous emissions from the foam as it
decomposed. Sometimes it is possible to revive an old vibrator by
spinning the top off using a lathe, burnishing the contacts and then
resealing by using the plastic top from an old aerosol canister.
Metal rectifiers were used for AGC and even signal detection in some
valved sets, the Marconi Atalanta for instance. In general, these
components are unreliable and should be replaced. This writer has
always used small Silicon diodes such as the 1N914 or 1N4148 to replace
these small metal signal rectifiers, and has had no problems.
The most troublesome relays by far, are open types that have no form of
protective cover. The next most troublesome, are plug-in types with
unsealed plastic covers which clip into position. Unfortunately, most
receiver relays fall into one or other of these categories.
Relay coils give far less trouble than contacts, but they can
occasionally go open circuit. This is firmly in the “bad luck”
category of faults. Rewinding is generally possible. It is worthwhile
pointing out that most of the 110VDC coil types of relay found in some
Collins equipments are still manufactured, and available from component
suppliers in the US.
These are octal or B9A glass devices found mainly in American radios,
intended to delay the production of HT for typically 30s, 45s or 60s
depending on the time delay value of the device. These components are
little seen in British and European equipment because of the more
widespread use of indirectly heated rectifiers, which did the same job
a different way.
Sealed types give less trouble than unsealed ones, which is only to be
expected. Old sealed meters can be very difficult to enter though, in
the event of problems. All too often the bakelite case can break, or
bits of mastic become loose and jam the movement.
Loose glass is best scraped with a razorblade, and then bedded back
down into a thin fillet of Evostik around the edge.
Broken glass is best replaced with new glass, rather than plastic,
which causes too many static problems. Cutting new glass to the right
shape seems not to be too difficult for skilled glaziers. The best
source of glass is from a junked meter rather larger than the one
fitted to your radio, though it has to be annealed before cutting.
On some radios such as the Hallicrafters SX28, the meter case was drawn
from sheet metal. They are often found cracked into sections splayed
apart rather like flower petals. Presumably this was caused by
stresses created during the manufacturing process. To bind the whole
lot together and keep dirt out after overhaul, this writer advises that
the casing be bound with masking tape and then secured with an
automotive hose clip.
11.19 Loudspeakers
WARNING: AC/DC sets must not have their chasses earthed, and not all
radios fitted with mains transformers have an isolated chassis.
Examples exist of radios with transformers which nevertheless have the
chassis connected directly to one side of the mains. (This design type
is most commonly found on broadcast receivers.)
When fully powered-up, check all cathode and screen grid voltages under
zero input signal conditions. Then check all anode voltages.
Check carefully for high temperatures and burning smells. Listen for
crackling, frying or hissing noises.
Alignment data is one specific area where handbooks are often found to
be incorrect. If your radio is known to be totally out of alignment,
refer to the gold standard radio, and set all cores and trimmers to the
same physical positions as on that one, to start with. Never
arbitrarily twiddle anything. Always be systematic. Keep careful
notes of what you are doing.
Marine radios which use 110VDC supplies such as the RCA AR8516L and
also battery sets, will not give as much gain per stage as radios
running on higher HT voltages. One piece of good news that results
from this, is that these low voltage sets are very immune to RF and IF
instability problems.
On unusual sets for which you have no information, and which refuse to
tune up, be on the alert for possible use of harmonic mixing. The
Atalanta is one radio that used this technique on the HF range of
coverage, because the RF valves are mounted outside the coilbox to
improve accessibility for servicing. Stray capacitances were such that
the top range uses the second harmonic of the LO to track the signal
circuits. Their earlier Marconi CR100 used the same mechanical design
but did not suffer this problem, possibly due to the use of gridcap
valves or because of the smaller number of ranges covered.
a) When the coil is tapped near one end of its winding, the
effective tap position depends on which end of the coil gets the
extra inductance due to the core.
RF/LO cores on brass leadscrews: use the peak which places the
Iron core furthest from the leadscrew support nut, to give
correct inductance temperature compensation (least positive
value of dL/dt).
The scale tracking of the local oscillator may be found incorrect, even
with a completely clean variable capacitor. If equally bad on all
ranges and the errors are all in the same direction, you may consider
bending the local oscillator plate vanes. This is an iterative
process, which must be performed very carefully, especially if there is
HT across the plates. It can take ages.
If the problem occurs equally on all ranges, first check the BFO by
listening to it on another radio. If the phenomenon occurs only on the
higher frequency ranges of coverage of dual conversion radios, check
the 2nd local oscillator the same way. Then check the following, for
both the 1st LO and the 1st mixer.
Some mixer circuits are partially neutralised, even where the circuit
diagram does not show it. The AR88D is one radio that has invisible
neutralisation. This technique gives an improvement in the symmetry of
the RF tuning response if the circuitry is correctly designed, and if
the hardware and layout has not been butchered over the years.
All of the foregoing means that it is necessary to take much more care
over the alignment of the mixer signal grid circuitry than would at
first appear necessary. The best way is to align for a noise (not a
signal) peak after first setting up the RF amplifiers, provided you can
be sure to avoid selecting the image. Otherwise, keep rocking the
tuning back and forth slightly during alignment, whilst listening to a
weak wanted signal. Tune the mixer grid circuit for the peak response.
When it is believed that the general alignment is correct and the set
is operating properly, it is time to look hard at AGC operation.
Meaningful voltage measurements require a measuring instrument with a
very high input impedance. The usual DMM or scope x10 probe is not
good enough for this job because they give only 10MΩ input resistance,
which is inadequate for measuring an AGC system running at, say, 2MΩ
impedance. If at all possible, you should use a proper VTVM such as a
Marconi TF1041C instead. The 100MΩ impedance given by this class of
measuring instrument gives negligible loading on even the highest
impedance AGC systems, enabling accurate measurements to be taken.
Look first at the AGC delay voltages, noting that delayed AGC was not
always used - especially in American sets. When present, the delay
voltages are usually produced from a valve cathode or from a potential
divider fed from a high voltage point. Examine the way the set
develops and distributes its AGC as the input drive signal voltage is
progressively increased from zero. Make sure that every stage that
should receive AGC gets it, in the right proportion, and in the right
sequence of initiation as the input signal rises through the first few
micro-Volts of aerial PD.
In the case of radios which distort at high signal levels with large
modulation on AM, the cause may be modulation rise in the final IF
amplifier. Check that the right type of valve is fitted, and that the
correct fraction of AGC is applied. This valve may be a straight type
in comparison with the variable-µ specimens fitted elsewhere in the IF
strip, and the correct amount of applied AGC may be zero. A good
specimen is always needed in this position.
Second detector problems generally stem from the high impedance nature
of the circuitry. This can cause difficulty in the event of even
slight leakage in the circuit.
As with AGC systems, beware detectors fed via a small capacitor from a
source of high combined AC+DC voltage such as the final IF anode. Any
leakage in the coupling capacitor would cause detector distortion, or
other more subtle problems such as unexpected squelch, as is sometimes
found, especially in the RAF R1155.
Because most noise limiter circuits operate at very high impedance, all
the usual problems with high value resistors and leaky capacitors
exist.
Crystal oscillators give far more trouble than most people suppose.
Eddystone was one manufacturer clever enough to spot this problem, and
address it head-on. Their S880/2 model provides a very small cathode
bleed current when the calibrator switch is in the OFF condition. This
is sufficient to prevent poisoning, but not sufficient to cause
oscillation or even noise output. All in all, this is a very elegant
solution to the problem, and indicative of a receiver design that was
better than average in several other respects too.
12.16 Ovens
In all cases known to this writer, VFO ovens can be switched off. This
is recommended, unless the radio is to be operated continuously in
areas having high diurnal temperature excursions.
12.17 BFO Injection
Injection locking of the BFO is sometimes seen. Here, the BFO appears
to suck-in towards the zero beat position under strong signal
conditions. Having acquired lock, the BFO will not release back to its
natural frequency unless the input signal is greatly reduced in
amplitude, or moved to a frequency well outside the capture range. It
is particularly annoying on SSB, where the effect is to impress severe
syllabic rate FM on the demodulated audio.
The Eddystone S940C suffers badly from this problem, despite (or
because of?) having a self oscillating 6BE6 product detector. A
partial solution can be achieved by improving the decoupling
arrangements. No total solution has yet been identified by this
writer. The good news is that this phenomenon does have a practical
application. Given some operator skill, the S940C can behave as a
useful synchronous AM receiver giving better results on selectively
fading 41m band signals than available from its envelope detector.
This type of fault is not at all uncommon in old radios, and can be
extremely hard to trace. Usually, the cure is easy and straightforward
once the cause of the fault has been positively identified. The trick
is to avoid inducing secondary faults along the way.
In sets that used a soldered chassis frame which has cracked apart, it
is often a good idea to resolder whole areas of metalwork en masse, in
case the fault is due to a dry joint in an important RF ground path
somewhere.
The final solution is for the desperate, and should only be used for
thermal intermittent faults where all other avenues of analysis have
already been explored. Be warned that this technique can cause
secondary damage, but it usually does positively identify the original
fault. Cover the radio with a fire blanket to block off all
ventilation. Fit a thermometer or thermocouple probe in the airspace
under the blanket at the top of the chassis. Run the set until the air
temperature reaches about +75ºC (use +65ºC if your chassis uses waxed
capacitors or coils, or was originally built with second grade
components). Open the blanket slightly at the bottom, and try to
stabilise the temperature at this figure. Stay with the radio, monitor
it very carefully and leave it to run until something fails. Usually,
what fails is the cause of your original intermittent fault, but
normally the failure will now be permanent, and therefore traceable.
You really must be present all throughout this test, and have a
suitable fire extinguisher readily to hand.
If you are convinced one of the stages in your radio is unstable, there
are three alternative ways of proving the existence of parasitic
oscillations when a spectrum analyser is unavailable, as follows.
Technique No 2: Add a small 47Ω resistor in the grid, and a small 10Ω
resistor in the anode of each stage in turn, working from the 1st RF
valve back through to the final mixer. Of course, this will identify
the problem by curing it. The need to put the resistor directly on the
valve socket contact does make this technique very fiddly to implement.
On the American equipments, the screws that hold the modules onto the
chassis are usually 6/32UNC with Phillips round heads, painted green.
Do not be tempted to remove any red-painted screws without a thorough
search through the handbook first. Red screws are rarely found, but
normally indicate fastenings which ought not to be disturbed. In
extreme cases, red screws could cause danger if removed, as a result of
suddenly released turret springs or a dropped mains transformer.
The screws found elsewhere in American radios are mostly 3/48UNC or the
larger 4/40UNC and 6/32UNC sizes. It is worth pointing out that
Phillips heads are similar to Pozidriv, but without the extra guide
flutes of the European system. It is therefore important to use the
right screwdriver, especially on the tighter 6/32UNCs. Most Phillips
drivers have red handles, whereas the Pozidriv ones tend to be blue.
Unfortunately, UNC hardware is rather hard to come by in Britain, and
most of the smaller stuff that is available has Pozidriv heads.
You will need, at the very least, Xcellite tool number 99-66. If you
or a previous owner has attempted to use an Allen key, it may be
necessary to drill out the grubscrew. This is difficult since the
material is very hard.
Beware cores that have become detached from their leadout wires.
Repairs are very simple once the loose core has been pulled free of its
former.
Set the calibration clutch to the middle of its travel. Make sure this
humble mechanism works properly. Repairs are often needed, and may as
well be performed now, rather than later. Collins-derived radios seem
to have fragile calibration clutches.
Check that all spring loaded split gears are properly preloaded. Then
make sure the Geneva mechanism that drives the octave coil selection
wafer switches is correctly adjusted, and properly ganged to the MHz
readout cylinders.
The mechanical counter display needs to have fully functional "-" and
"+" rollover display cylinders. These should appear when the kHz shaft
is set to below 000kHz, or above 1000kHz on any MHz range, heading
towards the adjacent mechanical endstops. Typical limits of travel
will be indicated at "-966" and "+034" kHz as the endstops engage.
Note the symmetry of the wrapround numbers.
Next, check that the MHz crystal oscillators, and any associated
frequency multiplier coils, are properly selected by the drive shaft
feeding from the bandchange mechanism. Dial up the “magic number”
given in the handbook (this varies from model to model), and set all
the cams to exactly the right positions. This should result in a
properly aligned geartrain, onto which the VFO can now be tracked.
Finally, set the VFO to track the kHz mechanism, and then lock up the
grubscrew. Be absolutely sure at the end of your work, that the main
mechanism dialdrive endstops activate before the follower nut inside
the VFO canister runs out of thread on its leadscrew.
The VFO Oldham coupler, and all others in the chassis, should have
about 0.75mm endfloat. The anti-backlash spring should be present and
functional.
VFO endpoint adjustments vary from model to model. They do all have an
ability to be set to exactly the right frequency at the 000 and 1000kHz
points or at least, they did when they were new. Typically the
endpoint adjuster is hard to get at, sealed by a screw plug, and
delicate. Not a nice combination. Frequency tracking across the range
is set by internal adjustment, but should be within 1kHz (500Hz when
new) anywhere between the 000 and 1000kHz endpoints. The error is
allowed to worsen between either endpoint and its adjacent endstop.
The VFO tuning ranges of the R-390 and the R-390A can in fact be
adjusted in the chassis, though the handbooks say otherwise. Target
2.455MHz to 3.455MHz from the VFO, over exactly ten turns of mechanical
rotation. If this cannot be achieved within the adjustment range of
the endpoint adjuster, the VFO will need internal surgery to remove
some inductance from the small coil in series with the large main coil.
Messing with the tracking is probably best avoided, unless your pocket
is deep enough to afford another VFO if things go wrong. The only
American VFO type that seems to give incurable problems with tracking
is the Cosmos, a late species of R-390A VFO. Most other types,
including the Collins original, are generally there or thereabouts
after half an hour of careful fiddling.
Use heavy mineral grease on the corrector stack mechanism and its
follower. Apply EP90 hypoid oil to the endthrust bearings. Sealing O-
rings should be lubricated by MS4.
Inspect most carefully, the follower nut that runs on the leadscrew.
On some designs, this nut is carried on a sprung cage, the fingers of
which can break. This is one identified cause of backlash. Repairs
generally prove possible, but it is important to retain some sort of
spring loaded flexible mounting arrangement for the nut.
After refitment to the chassis, always switch the VFO heater oven OFF
if you have a choice on your radio.
All permeability tuned BFOs seen by this writer, have input shafts in
the form of a very fine leadscrew. Typically, this is connected to the
operator's shaft by a bellows coupler. Make sure the coupler is at its
natural, ie unstretched/uncompressed length, with the BFO set to the
centre of the IF passband. Comments relating to relubrication of the
internals of the BFO canister are generally as for VFOs, though the job
is easier due to the absence of the tracking corrector stack.
Any shaft grounding springs must be making contact properly, and well
lubricated to avoid galling of the mating surfaces.
Drake permeability tuned VFOs never seem to give problems, and rarely
need adjustment despite their relative simplicity. They track within
1kHz, even though the ferrite arrangement looks really crude. Overall,
the Drake VFOs surely represent a more elegant engineering compromise
than the Collins approach.
British precision VFOs found, for instance in the S880/2 and RA17, tend
to use high quality variable capacitors from the likes of Wingrove &
Rogers, rather than permeability tuning. They generally benefit from
replacement of doubtful looking ordinary Rs and Cs, but never seem to
require retracking.
The PBT arrangement used by Collins in the 75A-4 relies on rocking the
VFO frame in synchronism with movement of the BFO capacitor, using a
spring loaded bronze belt & pulley arrangement. It all works very well
in practice, though there are two special points worthy of mention.
The right valvecans must be used, these being of slightly shorter than
standard height. This is to stop the VFO cans hitting the support
frame at either end of the PBT travel.
Radios having two RF, and two or more IF stages usually start
developing AGC at about 2µVpd input. Thereafter, a +100dB RF step
should generally result in <+9dB rise in AF output. For radios having
one RF or IF stage fewer than the above, the AGC action typically
starts at about 4µV, above which a +80dB RF step gives <+9dB audio
rise.
Maximum AGC voltage varies from about -10V on radios having simple AGC
circuits, to -45V in the case of radios having complex AGC distribution
arrangements and AGC amplification.
The star AGC performer in my collection is the GEC BRT400K, for which a
+100dB step above 3µV gives +2.5dB AF. This set has separate RF and IF
AGC voltage delays, a complex anode bend AGC detector, and feeds only a
part of the available AGC to the final IF (6BA6) stage. Further
evidence of excellent AGC management in the BRT400K is the way the
recovered audio SNR rises rapidly and progressively with the RF input
signal level. Certain other radios are able to match this receiver for
audio levelling, but none yet measured by this writer does as well on
SNR recovery in the difficult 1µV to 10µV signal region when receiving
an AM signal.
The BRT400 family uses an S130P neon HT regulator for its oscillators
and for the RF/IF screened grids. This type shows unusually large
sample-to-sample variability as regards actual regulated voltage.
Samples varying between 110V and 147V are held at this address. In
general, GEC shouldered-glass types are at the top end, and Cossor
balloon-top types are at the bottom of the range. Care is needed to
select an S130P sample which gives adequate voltage stability over a
reasonable range of mains input voltage.
The truth is that noise figure on the HF bands need not actually be
better than about 11dB for normal terrestrial shortwave reception. Of
the vintage manufacturers, only Collins, Drake and Racal seem to have
properly understood this. Other manufacturers generally produced
receivers that had excessive gain, and hence suboptimum dynamic range.
Note that for proper image performance, the radio should be fed from
the source impedance specified in its handbook. This applies to
alignment, testing and actual service. If the radio has a balanced
input, especially one that has to be adjusted for correct balance as
well as for resonance, be sure to feed it from a balanced source of
correct differential impedance. For alignment purposes, this may often
be achieved by a toroid and a network of impedance stabilising
resistors.
This writer normally measures third order intercept point on 10m, since
in the case of multiple conversion receivers, by choosing different
tone separations the performance of the first and second mixer chain
can be analysed separately. In the case of tunable first IF designs
such as the 75A-4, tones spaced at 20kHz will pass straight through the
RF and first IF systems, so that the performance limitation will
usually stem from the second mixer. With 100kHz spacing, both
fundamental tones and the third order intermodulation products will
still pass through the nose of the RF tuning, but the most remote tone
will generally now be sufficiently far down the skirt of the first IF
response for the second mixer not to be subjected to these two strong
tones at the same time. The measured performance will then be
determined by the first mixer. Or by the RF stage(s), if the first
mixer is unusually robust as may indeed be the case with a 6BA7, 6K8 or
Mazda/Ediswan 6C9.
When taking measurements with close tone spacings, the signal sources
used must have very low phase noise, otherwise performance will be
masked by reciprocal mixing mush. Also, the attenuators must be truly
linear, and must not interact with each other or with the combiner
network. This writer uses an identical pair of Marconi TF144H/4S
signal generators, each of which uses a high level unsynthesized
5763/QV03-12 oscillator without any subsequent amplification, followed
by entirely resistive output padding. To combine the two generator
outputs into a single 50Ω feed to the receiver under test, an Olektron
B-HJ-302U dual directional coupler is used. This provides good inter-
generator isolation without excess transmission loss in either leg.
Tubular 10dB fixed BNC pads are fitted directly to each port of the
coupler to provide a further 20dB of inter-generator isolation, and to
provide a reasonably accurate 50Ω source impedance for the receiver
under test. The path from each individual generator thus has 10 + 6 +
10 = 26dB loss in the line. On the end of this chain, is an 8-way
Raven 50Ω multicoupler unit. This uses a pair of FET devices to
provide sufficient overall gain, together with a very high IP and low
noise figure. This test system provides accurate and repeatable
results down to 5kHz tone spacings. Intercept points are measurable to
beyond +20dBm as and when (rarely) necessary.
An even better way ahead, especially when measuring with very close
tone spacings indeed, is to use high level crystal oscillators to
provide the tones. The crystals need to be tightly specified, of
exemplary quality, and designed especially for the job. You should
expect to pay very good money for a suitable crystal. The precision
oscillator circuits are usually bipolar designs containing a special
amplitude feedback mechanism which is quite separate from the
oscillator RF circuit. This approach reduces the output problems due
to limiting noise and waveform distortion. The output voltage is
usually derived from the current flowing through the crystal slice.
A supporting rationale for the choice of 10m as the basis for critical
third order intercept point measurements of valved HF receivers, is as
follows.
The lemon giving the -30dBm result is the HRO-MX, the main problem
appearing to be its inadequate mixer, even though the level of
injection from the local oscillator is quite high. This radio is
noteworthy for having a very transparent mixer and front end. This
causes a very high level of local oscillator breakout from the antenna
port on the higher frequencies. This was a well known problem with
this model in the days of Band I television broadcasting. To make
things worse from an intermodulation viewpoint, the HRO maintains high
levels of gain right down to the LF bands. Perhaps in the spectrally
uncontaminated days when it was designed, the smaller number of signals
made outright sensitivity a necessary design target even at low
frequencies.
As the strong signal is then detuned so that it falls down the skirt of
the first IF response but stays within the RF bandwidth, the first
mixer will become the stage that crushes first. This will occur at an
input level lower than the adjacent channel measurement by a factor
equal to the conversion gain of the second mixer plus the skirt
rejection of the first IF. Sometimes, the second mixer will not ever
compress under these conditions because the first mixer will be quite
unable to deliver enough power through the first IF selectivity skirt
to achieve it, however much input drive is applied to the antenna
socket.
15.1 Emissions
Radios with valves often had very high levels of local oscillator
injection to the first mixer, causing high levels of unwanted conducted
emission into the antenna feeder. Often, this conducted emission will
be 455kHz above the operating frequency. On the higher frequency
bands, the front end valves will provide little reverse isolation of
the fundamental or harmonic frequencies of the local oscillator, and
the RF tuned circuits will do little to stop the local oscillator
getting into the anode of the preceding stage. The HRO is especially
bad in this regard. The antenna will be reasonably efficient at the LO
frequency, since its resonance is only some 1½% removed from the
radio’s operating frequency at 30MHz. This obviously assumes that the
antenna is correctly adjusted. The net result can be a high level
radiated emission from the antenna structure occurring at the local
oscillator fundamental frequency, or at harmonics of it. The
fundamental can cause problems to other users of the frequency band you
yourself are operating on. The third harmonic can fall into ITU-R
broadcasting Band II, in Europe at least. The situation can be
worsened appreciably if any RF screening plates are loose, or if any
valve cans are missing.
15.2 Immunity
16. Photographs
Despite being simpler and cheaper, the R-390A was not really inferior
to its predecessor. This represents a significant engineering
achievement. Many manufacturers made the R-390A. The complete list
together with the estimated delivery quantities (originating from
N5OFF to whom go the author’s thanks) is: Collins 6185, Motorola
14202, Stewart-Warner 6587, EAC 11686, Capehart 4237, Amelco/TSC
3642, Capehart o/b/o Adler 5, Imperial Electronics/TSC 3976,
EAC/Hammarlund 118, Dittmore-Freimuth 215, Fowler 2. This last order
was as late as 1984. Total production of complete R-390A radios was
thus 50855. Very many modules were also manufactured as spares. The
complete radio photographed here shows Collins serial number 24 of
order 14214-PH-51. This order was satisfied by the very last R-390
deliveries as well as by the very first R-390A deliveries. This set
is thus the twenty-fourth production R-390A model ever made. It is
in excellent condition following a very detailed restoration. Every
one of the modules is original to this particular chassis. The front
panel is a late-production engraved example. The original silk
screened front panel is in fair condition, and safely stored for
posterity. This set has one non-standard mechanical filter, open-
frame line and audio output transformers, and a 6BE6 product
detector. It is otherwise exactly as built by Collins.
16.7 An Unusual Chassis Design: the RME model 99.
The RME99 is a very rare radio - at least in the UK. This example
is unfortunately rather corroded, but now once again fully functional
after many years of damp storage in a shed. Note the cast Aluminium
frame, onto which is screwed a flat tinplate chassis bed. There is
no separate front panel, because the radio chassis assembly slides
into the case from the rear. All the markings, the meter and the
glass are directly carried on the front of the box. This unusual
construction must have caused production of the RME99 to be rather
expensive. There are lots of loctal signal valves, plus an octal
neon and a UX4 rectifier. The Iron components are far larger than
normal for a set of this class. The main tuning dial is the tiny
window on the right, which has a superb hand-marked vernier scale.
The big dial is for amateur radio bandspread. This set works very
well indeed, and is a credit to its designers all those years ago.
This set used octal, B8G (similar to loctal), B4 and B7G valves.
Later variants such as the BRT400K, used B9A and B7G valve types
whilst keeping the B4 neon and a couple of octals. What a strange
mixture of valve bases! This particular radio has the BRT403
calibrator fitted. The BRT400D was broadly equivalent to the AR88LF
& AR88D combined, but GEC fitted a calibrator as well. (In
comparison, none of the AR88 family was ever given a calibrator.)
The underside photo shows the inaccessibility of the RF unit which
cannot be dismantled, since all its partitioning bulkheads are not
only riveted, but soldered into position. Only the LO gets the
benefit of ceramic switch wafers, though all signal valve sockets are
also made of this material. These sets run very hot indeed because
of the unusual active smoothing arrangements. Two beefy KT81 beam
tetrodes are used - one for the audio output, the other for the HT
smoothing. The HT current drawn by a healthy BRT400D specimen is no
less than 167mA. This is almost double the requirement of most
similar sets. (The later BRT400K used EL84 pentodes, and got by with
a more modest 140mA of HT current drain.) The BRT400 has an anode
bend AGC detector which works well, and there is a particularly
praiseworthy signal detector and audio system. These features make
BRT400s very suitable for SW broadcast reception. The pointer
travels on a stranded steel wire. Each range dial strip is
individually lit. The dial register is backlit by a bulb which also
illuminates the interior of the set. Yet more bulbs are fitted to
the dial vernier and to the meter - making ten bulbs in all, plus
two spares! This is surely a record number for an HF receiver. In
general, BRT400s are found well cooked and in very poor condition.
But service lives of 50 years are known. This indicates their
toughness, and the excellence of the components. Moral: the Brits
could make decent receivers when they really tried.
Radios like this one should be treated with deep suspicion unless all
of the original Sprague Black Tubular capacitors have been replaced,
and unless the winding/chassis insulation of the HT choke has been
checked carefully. Removal of the RF unit on this radio is very
difficult, and the turret stator contacts are impossible even to
inspect without prior removal of all biscuits from the drum. The
ceramic biscuits are not individually screened, but are extremely
well made. The variable capacitor has a ceramic shaft. The HF
coverage of this set exceeds 50MHz. A 5R4G is used as the rectifier,
which is a highly unusual (unique?) choice in this class of
equipment. The SP600s did not stay in active service long because of
their incendiary tendencies. Many of them were quickly made
available to American amateurs for use on MARS USA-Vietnam
communication links. The on-air performance is generally good, other
than suffering overload problems due the excessive RF/IF gain caused
by the many 6BA6s and 6BE6s. There were many production variants,
including at least one VLF type. This example is the relatively
common US Navy SP600JX-6 type R-274B/FRR, which was built to JAN
specifications.
This ISB adaptor was initially intended for use with the the R-390,
but it is equally usable with the R-390A. It is very large and
heavy. This equipment contains 44 valves, almost half of which are
double-section types. There is motorized AFC, and the sideband
filters are physically huge. In the days when this was made and for
many years afterwards, it was conventional to operate the radio with
a much wider IF bandwidth than would be considered acceptable today.
This was necessary for the AFC to work properly. Each sideband
filter in the CV-157 is 6kHz wide. It is important to understand
that this unit was primarily designed for teleprinter sideband
traffic, not voice.
This novel design of radio has 1MHz ranges and extremely elaborate
screening for the various oscillators, in order to reduce oscillator
breakout to extremely low levels. The construction is modular.
Unfortunately, the tuning is very stiff due to the complexity of its
permeability tuning mechanism. Rather than using cams and followers,
this set uses screw jacks to raise and lower the slug platform, which
carries very many cores. The RF coils warrant very close inspection
to discern their design and manufacturing cleverness, which easily
out-did Collins in terms of design elegance. The VFO uses a
precision Wingrove & Rogers variable capacitor, and is extremely
stable and linear. The IF strip relies on quartz 500kHz twin-slice
B7G filters which were inadequate compared with competing radios
fitted with Collins mechanical filters, or those other radios using a
100kHz final IF. This set does have a proper product detector
(albeit of suspect design using a 6BE6 heptode); AGC tailored for
SSB; 50kHz/revolution main tuning; a separate electrical fine
tuning vernier; and a switchable fixed or variable BFO. These are
all features which make the S880/2 feel fairly modern to operate.
The front panel is a casting, which can be unplugged from its
cableforms to aid servicing. This set is extremely heavy due to its
extensive use of brass plate, much of which is Silver plated.
This was really a super broadcast receiver rather than an out & out
communications receiver. It uses a conventional twin pentode 6AG5 RF
strip and eight-gang variable capacitor all the way up to broadcast
Band II, though the mixer is a loctal triode type 7F8. A 455kHz IF
is used on the lower frequencies, and 10.7MHz at VHF. This receiver
is single conversion on all ranges of coverage. It runs hot because
of thirsty push-pull 6V6GT audio. The lack of any cover over the
tuning gang or any screening under the coilpack means that drift on
Band II is always a problem. There is a neon HT regulator, but no
AFC, tuning indicator or signal strength indicator. So retuning on
FM broadcast stations is a frequent event with this set. The
reproduction quality is truly excellent. The user handbook supplied
when new, shows 7H7 and 7A4 loctals in the IF strip, but this example
actually has 6SG7 and 6J5 octals fitted instead. There is a non-
original 240V/110V autotransformer mounted piggyback on its mains
transformer. This was fitted by the dealer when the set was sold new
in Switzerland. Cost-cutting in the design is evidenced by the lack
of any fuses whatsoever. To make things worse, the mains filter
capacitors are of the notorious Sprague BT type. These are very
prone to crack, and develop serious leakage.
This design is a hybrid which uses valves for the signal stages and
transistors for the ancillary circuitry. The performance of these
sets is excellent but depends on expensive multipole plug-in crystal
filters. As always with Drake the ergonomics were unusual, and the
tuning feels rough due to the use of cheap plastic gears. The very
last few R4Cs used brass gears, which worked much better. This is a
mid-late production variant, with 6EJ7 mixers. Earlier R4Cs had 6HS6
mixers, and the ability to select only one CW crystal filter. This
situation is typical of Drake’s seemingly endless variations in
production build standard. The design quality of the R4C is very
variable. The signal stages are superbly engineered. In contrast,
the power supply is very badly designed and prone to burning its
printed circuit board. Class A audio is used to avoid DC line
fluctuations on speech peaks, a design nicety evidently not
understood by those who would have you fit a modern class-B audio
chip instead. However, it has to be said that the standard R4C audio
output transistor is mounted directly below the VFO, and this does
cause slow thermal frequency drift. This problem could so easily
have been avoided by a using better physical layout. There was a hum
problem on nearly all R4Cs caused by the layout of the chassis, and
its wiring. Part of the reservoir AC current was allowed to flow
through the audio board. On the very last few R4Cs, some additional
thick wiring was used to route this AC hum current past the audio
board without having to flow through the Copper print on its way back
to the rectifier anodes. The R4C does not share its mixing
configuration with any Drake transceiver except the rare, and much
later TR5.
This radio was extremely expensive when new, being roughly the same
price in the UK as a contemporary Jaguar car. Plug-in B9A mechanical
F455J-series filters provided the IF selectivity. The angular
position of the PTO subchassis was ganged to the angular position of
the BFO capacitor in such a way that front panel control of passband
tuning was provided. The coupling mechanism was a flexible bronze
drivebelt. This same technology was also used to raise or lower the
slug rack. A regenerative adjustable notch filter was fitted, and
the set had an early twin-triode product detector. This particular
example has been modified to incorporate an E88CC frame-grid twin
triode first mixer and a 7360 beam deflection second mixer. An IF
tailing filter has been added, clamped to the inside of one of the
chassis apron sidewalls. The result is a considerable improvement on
the original circuitry, which used two 6BA7 mixers and was very prone
to premature overload. Although impossible to quantify, the 75A-4
has a very nice “feel” when in operation, and it is an unfatiguing
set to operate for long periods. Collins lost these nice ergonomic
touches with their later S-line equipments.
This example has the 136B-2 four valve TRF noise blanker option,
which was fed from an external VHF whip antenna. It also has the
rare B&W 340-A notch filter option. Early winged emblem examples
such as this one used hard-wired unencapsulated relays on the
underside of the chassis. They gave regular trouble in service.
This set has been fitted by the RAF with the enclosed plug-in relays
found on later round emblem radios. The receiver section of this
radio works very well indeed, despite most of the valves also
performing transmitter functions. The piggyback speaker and mains
power supply unit is the model PM-2, which was intended for light
duty work only. The mains transformers were rather fragile. This
example has had a Gardners unit fitted by the RAF. It is of far
higher quality than the original Chicago Standard Transformer Co
item. The only operational irritation when using SSB mode, is the
lack of an RIT control. In the days when Collins were the only SSB
manufacturer, this feature would not have been needed because
everyone else was also using Collins gear. None of it drifted much,
and all of it drifted in the same direction in any case. The CW mode
of operation is rather poor, as this radio gives a 1.5kHz receive
frequency offset and has no narrow filter. The radio was in scrap
condition when acquired from the RAF, and required a lengthy rebuild
using many mechanical parts taken from another junked set. This
particular example is kept in a model CC-2 fitted suitcase. By the
time the KWM-2 was produced, Collins dealers were well versed in the
art of persuading purchasers to also buy a wide range of expensive
accessories, to increase the overall profit in the sale.
17. Glossary