Murphy 62B Repair
|
|
Not a refurbishment as
such, but a continuation of a job I
started in 2014 (click to read) to make my Murphy 62B into
a reliable working receiver. |
 |
|
Anyone familiar with this
type of receiver can see that it's been got-at. The smaller knobs
have been changed and a multi-turn pot fitted below a new meter.
I don't like the knobs but the S-Meter is a useful addition,
as is the new control which I believe is for fine tuning SSB.
I need to check on the exact version of the 62B as there are
two, the AP67757 and the AP67757A. I initially chose "A"
but later I discovered my example has some features of the earlier
version without the "A". Below are the appropriate
circuit diagrams in three sections. Click
the picture above to see the full manual which includes all versions
of the receiver.
Preceeding the serial number
and engraved at a later date are the letters "EMD/R",
which Andy, G8JAC suggests is "Refurbished at the Devonport
Dockyard" (Electronic Maintenance Dockyard) |
 |
|
AP67757A RF UNITNote that these sets of drawings are not quite right
for my 62B variant which is coded AP677571. Mine is a mixture
of AP67757 and AP67757A of which the latter carries a cross-mod
control and a radar suppression circuit AND uses all glass valves
not the B8B types used in the non A variant. For completeness
I've copied the A versions first (Fig 66) then the non-A versions
(Fig 65) below each example. Slightly confusing....
Note that this receiver would probably have been
used on a ship with this type of low
pass filter helping to reduce interference.
|
AP67757 RF UNIT
 |
AP67757A
IF UNIT
 |
AP67757 IF UNIT
 |
AP67757A
OUTPUT & PSU
 |
AP67757 OUTPUT & PSU
 |
|
I'd been quite happy with
the work I'd carried out over five years ago on this old set
and it was now residing on display at the end of my work bench
which was probably one of the reasons I decided to renew my acquaintance
with it because, if it had been elsewhere, the sheer weight would
have put me off tackling it. I needed to temporarily unscrew
my desk lamp and some other fittings before dragging it out and
removing its outer case. I then noticed the odd Belling Lee mains
connector. The previous owner had obviously modified the set
and in doing so had replaced the Plessey mains plug (a dangerous
2-pin affair using its shell as earth). I recall having repaired
a similar radio for a customer many years ago and had fitted
a much safer IEC plug (like the ones used in a PC). I must have
had a proper mains lead with a miniature Belling Lee plug back
in 2011 but I couldn't find it so made a new mains lead with
a new connector which I recalled I had in a box labelled "power
connectors".
As enthusiasts like to know
what type of valves are used in this type of equipment, I've
included the table below, where the columns cover the sub-assembly
viz. 1 is RF Unit, 2 is the IF Unit and 3 is the Output/PSU.
The radio is the 62B/AP67757A which I assume is very similar
to my example the AP677571. Note that the 6BA6W is similar to
the EF93, and the 6AL5WA is like the EB91. |
|
Reference |
Type |
Equivalent |
|
Reference |
Type |
Equivalent |
|
Reference |
Type |
Equivalent |
|
V101 |
CV4009 |
EF91 |
|
V201 |
CV4015 |
EF92 |
|
V301 |
CV4009 |
6BA6W |
|
V102 |
CV4009 |
6BA6W |
|
V202 |
CV4015 |
EF92 |
|
V302 |
CV4043 |
6BW6 |
|
V103 |
CV2128 |
EF91 |
|
V203 |
CV4015 |
EF92 |
|
V303 |
CV4005 |
EZ90 |
|
V104 |
CV4014 |
EF91 |
|
V204 |
CV4007 |
6AL5W |
|
V304 |
CV4005 |
EZ90 |
|
- |
- |
- |
|
V205 |
CV4007 |
6AL5W |
|
V305 |
CV4100 |
0A2WA |
|
MR101 |
CV448 |
OA91 |
|
V206 |
CV4015 |
EF92 |
|
- |
- |
- |
|
|
I plugged in the mains
lead, poked a long wire into the aerial socket, and switched
on. The dial lamps came on and after a short while the S-Meter
registered half scale but there was no sound at all... just silence
no matter which switches I pressed or turned. Switching to the
long wave setting and tuning to Radio 4 on 198KHz showed a nice
peak at the S-Meter which on tapping gently proved to be a bit
sticky. Looking at the rear top panel I found a 2-pin socket
marked "speaker" but poking wires from my HRO speaker
proved this output was defunct. After a search I found a pair
of low impedance headphones and plugged them in and waggled the
volume control only to be deafened by a terrible ear-splitting
crackling noise. I removed the phones within a few milliseconds
and waggled the pot. It gradually got a little quieter but refused
to get much better.. but at least I could adjust the volume setting
then gingerly put the phones back on. It was obvious that all
was not as I'd left the receiver years ago. Tuning across Radio
4 showed the IF response was lumpy and after switching to 80
metres (drat-not quite covered!) indicated the set was as deaf
as a post. All the valve shields seemed warm enough so my guess
is a resistor or two, and probably a few condensers need replacing.
Pictures of the 62B follow. The exterior views are not especially
interesting. |
|
Not a lot to see in the
underside view, below. A bit boring.. which may sum up the 62B,
although the mechnical designers must have had a field day with
this series of receiver? |
 |
|
A nice feature
At each rear corner is fitted
a ball bearing race (now devoid of lubrication) which enable
the receiver to be moved on a workbench without gouging the surface.
The outer case does not extend
to cover the underside of the receiver. |
 |
|
|
|
|
 |
 |
|
This example carries light
rust and a grey deposit that smells like "essence of government
surplus shops" and difficult to remove from ones fingers..
Nicely laid out and you can see the odd plug and socket which
are fitted so that complete assemblies can be detached for remedial
work. Not obvious in these views is the large turret (carrying
the RF coils) and the heavy gears etc for turning it.
When I last worked on the receiver
I'd overlooked some sloppiness in tuning. In a receiver of this
quality I imagine there's provision for sorting this out. |
 |
Below the IF area you
can access the power supply valves.
Below is the power supply component
layout. |
|
 |
|
All the RF coils in the
62B receiver are mounted in screened boxes fitted into its turret. |
 |
|
I decided to test the receiver
using a signal generator rather than a random wire, especially
because the wire aerial I have is suspiciously poor when connected
to my SDR. I then discovered that I couldn't find a suitable
RF plug for mating with the socket fitted to the 62B. Clearly,
one desirable change, other than fitting an IEC mains connector
is to fit either an SO259 or a BNC socket for the aerial. However,
by poking a wander plug into the aerial socket I was able to
fit a BNC lead between the 62B and my HP
signal generator. The latter fired up and produced 600KHz
AM which was detectable with the 62B set to the same frequency.
The S-Meter read +20dB with the HP set to +20dBm and dropped
to a just detectable audio output at -20dBm. The meter showed
the same reluctance to move freely without tapping so that's
another job I need to tackle.
Looking at the various controls on the
front panel I noticed the unusual one labelled "Anti-Cross-Mod".
The 62B RF amplifier valve, specified as a CV4014 is an M8083,
a special version of the CV138 or EF91, a common variable-mu
RF pentode (V101 above) which is
operated in accordance with the values of resistors used to set
its electrode voltages. It's important to operate an amplifying
valve so that it works in a perfectly linear fashion otherwise
its output will carry not only a stronger version of its input
signal, but harmonics of this and, more importantly, combinations
of several input signals if more than one is visible to its control
grid. It's important to realise that although the overall receiver
bandwidth may be quite narrow, the RF amplifier bandwidth relies
on the quality of its tuning coil and other factors, so that
its control grid will see many strong unwanted signals, perhaps
in the whole shortwave broadcast band to which the receiver is
tuned. If the valve characteristic is not perfectly linear the
anode current will carry a whole host of small currents associated
with products of many input signals, not just the current associated
with the desired signal. One critical result might be to activate
the receivers AGC line when we are trying to maximise weak signal
amplification.
The specification requirement for the
62B clearly must have asked for some provision for weak signal
reception in the presence of interference. Whether this was strong
adjacent channel broadcasts or something found in a ships overall
electronics package I can't say, but this may have been radar
signals or powerful local RF radio communications transmissions.
The 62B was earmarked for receiving domestic-type broadcasts
and playing these to a ships company so interruptions from the
ships radio tansmitters or noise from radar transmissions would
be really annoying. So, the special anti-cross-mod control was
fitted in order to fine tune the linearity of the RF amplifier
to minimise interference. In fact, studying the circuit of V101
shows it was not fed with the set's AGC voltage, instead relying
on the bias voltage set by RV101 and its cathode resistors. You'll
also note the connection to V101 suppressor grid brought out
to SK101 (described in the manual as being used by R.I.S. outfits
which I understand is "Radar Interference Suppression",
in that a pulse is provided from an external equipment for muting
the RF amplifier for very small periods to eliminate pulse noise). |
 |
|
Now to diagnosing the
dead loudspeaker and what a clever idea to make the various modules
extractable from the chassis so easily. I initially forgot to
remove the knobs but once this was done, and four leads unplugged,
the IF amplifier just slipped out of the chassis. I followed
a yellow wire from the speaker to a Jones plug and found Pins
3 & 5 measured 3 ohms/0.3mH so the loudspeaker appears to
be fine. Next, I removed the RF gain control knob, slackened
the two securing scews and slid out the power supply module from
the chassis. Dead easy... and checked the loudspeaker on/off
switch. Mechanically perfect but it measured open circuit in
both ON and OFF. Clearly a build up of oxide or dirt was preventing
it from working. |
 |
Upper left, the loudspeaker
switch in the OFF position and right, in the ON position but
both positions gave open-circuit across the switch terminals.
I thought the action might be
lazy, but the switch appears to work normally, that is mechanically..
electrically not.
The lower switch is a ganged
pair of identical type and this also read open circuit before
I plugged the receiver into the mains supply. In that case however,
whetting current soon fixed switch continuity.
Now that the power module has
been extracted from the chassis I'll clean both switches then
fit an IEC connector in place of the Belling Lee type fitted
previously.
Below is the detached IF module.
|
 |
|
 |
 |
|
Above, a top view of the
power supply module and below, the underside.. so easy to work
on. |
 |
|
I fitted an IEC mains socket
(see below), a new 2Mohm volume control (RV224) and swapped several
resistors that read high. These were R206,100Kohm (open circuit
V202- g2 ground connection); R228, 330Kohm (412Kohm BFO-g2);
R226, 68Kohm (81Kohm V205 detector); R225, 680Kohm (800Kohm V205
detector); R312, 3.3Kohm (oddly 612 ohm and lost its markings
from excessive heating, HT feed). The LS switch turned out to
have an open circuit at each of its terminals (tinned solder
tags screwed to copper terminals). Reassembly was straightforward
and after jiggling the local oscillator valve the receiver worked
well enough to pull in Radio 4. The dial indicated a peak at
about 210KHz (with a very wide tuning coverage) and loss of treble
when tuned in precisely so alignment is definitely needed. |
|
|
 |
The original 2-pin Plessey
plug had been replaced years ago with this miniature Belling
Lee type (same mounting requirement, but rather flimsy) which
is easily changed for an IEC type. I used a clip-in type because
the flange holes on the other type are in an awkward position.
The power supply circuit is
shown above, where you can see a pair
of filter condensers C309/C310, which decouple the mains input
to the chassis. With the old 2-pin plug it's not an uncommon
practice to fit a live mains lead whilst holding the receiver
case. If the shell connection did not make perfect contact with
the connector screw before one of the 2-pins touched one of the
mating pins you'd get a nasty shock. On board a ship the receiver
would almost certainly be grounded so a shock would be a rare
occurrence. The leading earth pin on the IEC connector grounds
the chassis before live pins make contact. |
|
|
Disconcertingly the new volume
control was nearly as crackly as the original and I noted 4 volts
DC at the input to the pot which probably means C229 is leaking.
This is rated at 0.005uF x 1000 volts (note that these high voltage
condensers were often used in critical parts of circuits known
to be prone to failure), As C229 is readily accessible I'll swap
it tomorrow. |
|
The next morning, before
I cut C229, I measured the voltage at its end again and it was
sitting at 10 volts which I imagined was due to C229 being colder,
but as I looked the meter suddenly shot up to 220 volts which
didn't make sense so I peered at the live end of C229 and noticed
a red insulated wire (from R204, 4.7Kohm) touching the anchoring
tag. I moved it away with a screwdriver and the voltage changed
from 220 to 0.5 volts. Maybe C229 is ever so slightly leaky but
certainly not responsible for the elevated voltages. Over the
last 60 years or so pressure of the insulation against the sharp
solder tag had cut through the plastic. As the new volume control
is over 30 years old, having arrived in a job lot of about 100
Radiospares pots in little boxes, and having a gap in the top,
I was able to squirt in some switch cleaner and, after a few
waggles, the crackling went away and audio control was perfect.
In fact any slightly odd sounds and the loss of treble when Radio
4 was tuned in had disappeared leaving Radio 4 perfectly clear
and undistorted. I waggled the cable harness to the larger chassis
plug and I could hear sharp crackling sounds so no doubt there's
another potential short lurking in the wiring. |
 |
|
|
Now that most of the problems
with this old receiver are fixed I decided to check the IF tuning
to see if this would help explain its deafness. Sure enough the
best IF response was about 497KHz and, after setting all the
IF transformers to peak at 500KHz in the 1KHz bandwidth position,
reception was transformed. The High and Low BFO settings worked
fine on SSB but the tuning was far too imprecise for comfortable
listening. This is probably due to wear and poor lubrication
in the mechanism which is a very complicated affair. I then remembered
the added multi-turn pot at the top right of the front panel
and found it was perfect for tuning SSB. This extra control transforms
the receiver and, with this, 40m LSB signals can be precisely
tuned and received loud and clear. Without this extra multi-turn
pot the tuning backlash is so horrendous it makes it impossible
to resolve SSB.
Switching back to long waves,
and using a length of wire as an aerial, Radio 4 measured a stonking
S9+50dB on the S-Meter whereas previously it registered only
a measily S5. Switching between the three bandwidth settings
showed all was not ideal however; with best reception in each
setting requiring a retune to see a peak on the S-Meter. In the
past I've discovered that it's virtually impossible to properly
align an IF strip using an AM signal generator and audio wattmeter.
You definitely need a spectrum analyser to do this properly otherwise,
nine times out of ten, the receiver characteristics result in
each IF transformer ending up off-frequency. This also happens
because of the way the designers produced the different bandwidths
(here they just added 470pF across each
transformer secondary). In this set as is the case with many
that use a crystal filter, you need to align around the crystal
response. The 1KHz setting in the 62B. If you're lucky the crystal
filter will be close to 500KHz but if not it may be better to
align at whatever frequency the crystal filter dictates, then
check the wider bandwidth settings match this and finally tweak
the local oscillator to match the dial calibrations. Not exactly
the final step though as you also need to tweak the BFO to match
the response of the crystal filter because the three settings
(Tune, High and Low) are preset and not adjustable from the front
panel. All this can take an inordinate amount of time but with
a satisfying end result.
One thing of note I discovered
was the excellent rejection of the receiver to out of band signals.
Usually I'm lazy and use the aerial terminal for injection of
the IF test signal, but even with 1000mV of 500KHz at the aerial,
the IF couldn't be adjusted comfortably and I had to connect
the signal generator via a small capacitor to the front of the
IF amplifier. Surprisingly, I couldn't see an IF trap in the
RF amplifier so the rejection must be due to the excellent screening
from the chassis and the turret design.
Sadly 80m isn't covered in this
model, with the nearest frequency being 3.9MHz, and now that
I've got this far I'm starting to lose interest because the 62B
is a bit boring. I'm sorely tempted to just stick the covers
back on and shove it back into its slot at the end of my work
bench and find something more interesting to fiddle with... but
at least I can listen to Radio 4 on the 62B whilst I'm working.
After all.. that was the main reason for introducing this particular
version of the B40... |
 |
 |
|
Now that I've removed
a side panel, my interest has returned somewhat. This drum is
very similar to that used in my
favourite receiver, the R206. This one could well be a later
development of the same drum? |
 |
These drum contacts are
pretty good and, compared with those R206 rhodium plated contacts,
would be a lot cheaper to produce and repair.
Below.. the RF amplifier chassis
(and the drawing from the manual) which is very accessible especially
with the receiver on its side. You can see how the drum simplifies
wiring and layout. |
|
 |
 |
A quick check of the various
resistors above revealed most were high in value as per normal.
A quick check of the condensers indicated C120 was short-circuit.
This brings me to something obvious (and sort-of helpful?). Each
of the three main assemblies, the RF Unit, IF Unit and AF &
Power Unit have components etc numbered in accordance with their
assembly, starting with 1, 2 or 3 respectively. C120 is therefore
within the RF amplifier chassis seen in the drawing above (top
left). Below the row of those five 0.1uF condensers you can see
three resistors numbered R124 (430), R117 (4.7K) and R115 (220)
which I measured as 56K, 58K and 72 respectively. The picture
shows these as 3.9K, 5.6K and 62. I looked at the circuit for
the AP67757/A.. where's C120? After puzzling over this I eventually
realised I'd assumed my B40 was the /A but it's not, it's the
version without the /A and sure enough, after swapping to the
correct set of three circuits (now in place above) there was
C120, and following the connection (confusingly) from SK102/8
to PL202/8 (and PL205/8) I see it's shorted to ground by SW206c
(AGC ON/OFF). Hence C2120 is grounded when AGC is OFF. The mystifying
resistor values are no longer mystifying because the non- /A
has different values viz. R124 (47K), R117(4.7K) and R115(220)...
at least, partially explained except for R115 which you can clearly
see (in the photo) is 61 ohms. Perhaps this is/was the subject
of a later mod?
What's really confusing though
is the valve complement. The /A version of the 62B has "modern"
B7G and B9A-based valves and the non /A uses the rather unusual
B8B type of valve. Until I resolve the puzzle I'll assume my
62B is the non /A with /A valves.
|
|
|
|
|
|
Having exposed the RF coilsets
I guess it will be worth tweaking the trimmers and cores to align
RF input with dial readings. Before doing this I'd like to see
if I can sort out sloppiness in tuning and also check IF response
which varies between bandwidth settings. I tried a quick check
to see roughly what the IF response was like. I'd noticed several
tuning points after peaking everything and sure enough, when
I looked at the response it had three equal peaks in all three
bandwith settings. Most significant was the narrow range, supposed
to be 1KHz which had steep sides but with three equal peaks with
troughs either side of the centre peak. The response to 500KHz
was nowhere near the centre of the trace but after twiddling
the crystal trimmer to see its effect I found at one critical
point the width of the response reduced to a quarter. The picture
below shows the response with the tracking generator set to -20dBM
and connected via a 220pF capacitor to V201 control grid, with
V204b (the detector) being the output monitoring point. This
is certainly not ideal but good enough to investigate what's
possible. Below, the scan is 200KHz so one horizontal division
is 20KHz, making the response about 1KHz, which is correct for
the narrow setting. I measured bandwidths of around 4.5KHz and
12KHz at the other settings. |
 |
|
With the IF correctly
on 500KHz I then checked the low frequency band to see if the
dial readings corresponded with input frequencies. Rather than
using a signal generator at this stage I used three long wave
broadcasts viz 252KHz, 198KHz and 162KHz. The results weren't
too bad and with some work I can carry out decent alignment later.
At this point I decided to read
the manual to see what was suggested by way of alignment back
in days the receiver was being used. Of course the alignment
needed to be carried out with test gear available back in the
1950s in Naval dockyard repair shops and, as this wasn't a patch
on modern test gear, much better results can (theoretically)
be gained nowadays. I say theoretically because overall degradation
in components will show up when modern test equipment is used.
Coils, condensers and resistors will have changed over the last
60 odd years and. as it's not a viable task to fit new parts,
the optimum final results may be not quite as good as the designers
intended.
The next step is to make a special
tool with which to adjust the IF transformers for what the manual
refers to as "final touch-up". The IFTs use what appear
to be Tufnol screws for adjustment of their cores and these are
locked with something like 7/16 inch nuts. These screws are sticky
in operation and not all are accessible without angling the adjusting
tool. Below are the scans from the three bandwidth settings.
I'd read the manual and noticed a second adjusting screw for
the crystal filter and, following the guide given in the manual
regarding the changing shape of this filter as the adjustments
are made, experimented with the effect of twiddling both screws
rather than, as previously, just to one on the side of the can. |
 |
 |
 |
|
A word about the response
curves above: You'll note in the initial (blurry) scan the vertical
divisions are 4dB with a baseline of -88dBm, giving only 14dB
to the output signal peak, but the three newer scans have vertical
divisions of 10dB and a baseline of -110dBm giving better than
35-36dB. Readings could have been slightly better if I'd used
a VBW/RBW of 300Hz or less. Note the "centre" frequency
of 501.666KHz. This seemed to be the natural resonant frequency
of the crystal in the filter and because of this I aligned the
IF amplifier to that frequency rather than 500KHz. The manual
does obliquely refer to this technique which will give a flat
response rather than a top carrying a few peaks and troughs which
I'd noticed when aligning on 500KHz.
I then tackled alignment of
the 6 wavebands. A tip here is to carry out all core adjusting
at the low frequency end of all wavebands before using the trimmers
for all the high frequency ends, otherwise you'll spend an inordinate
amount of time spinning the dial from one end to the other and
back.
After carrying out this job
I tuned to 40m and tweaked the band edge to 7.000MHz on the dial.
Further tests showed again the sticky tuning mechanism noted
before when trying to resolve SSB. I now have a suspicion that
the cause might be a seizing tuning condenser. The thing is not
readily accessible so I'll leave this for now, but another possibility
is oscillator pulling, something that plagued my R206 when using
it as a second IF on 24-26MHz for my nuvistor converter. Something
else I noted.. oddly on the long waveband Radio 4 peaks on different
frequencies at each of the three bandwidth settings. This does
not seem to happen on the medium waveband. A clue might be the
very high signal strength of Radio4 with my aerial.
A coupe of other minor faults
to sort out.. the limiter control kills the audio and the cross-mod
control seems to do nothing. I'd guess bad diodes are the reason. |
Click the picture to see a better picture of
the tuning mechanism. |
One of the annoying things about
this example of the B40 is imprecise tuning. According to the
manual you should not see backlash of more than +/- one division
at the logging scale. I've found a sort of pulling effect in
the tuning which I've come across before in a few receivers.
It can be due to either stiffness in the tuning mechanism or
oscillator pulling. The latter would be due to the local oscillator
shifting, perhaps on account of a voltage change as the set is
tuned to a strong signal. Also possible is a shift in the IF
response resulting from a significant change in AGC voltage.
The latter be checked by switching off AGC, and IF pulling due
to a strong signal can be tested by reducing the RF gain. (see
later for explanations)
Hopefully poor lubrication is
the problem because this can be sorted out. |
|
|
I decided to bite the
bullet and dismantle enough of the 62B to access the tuning mechanism
to see if anything was wrong, or indeed if everything was as
the designers left it. To start I removed the knobs so I could
detach the front panel. Here, I found a major problem. The wavechange
knob is secured by two grub screws. The smaller one came loose,
but the second larger screw refused to turn. I think it may have
been previously attacked because the slot was damaged and my
only recourse was to drill it out. Fortunately the grub screw
(about 0BA) was not hardened steel and after a pilot drill had
penetrated to the square part of the spindle a second and a third
larger drill removed the old grub screw and the knob slipped
off. I tapped the hole for M6 and I'll refit it later. After
removing the screws holding the front I pulled it away. The various
tuning gears were completely dry so I removed the tuning flywheel
and log disk and applied WD40 to the various gears followed by
cleaning and then grease. I also found the dial window assembly
was seized in place. After oiling the securing nuts this was
movable, but I might return to this later to make it less stiff.
After greasing I found the dial moved much more freely and without
backlash. The latter is dealt with by a split sprung gear which
I'd also greased. |
 |
|
For those unfamiliar with the
B40, the tuning mechanism is really clever. As the tuning knob
is turned clockwise the dial carrying the frequency markings
rotates, and simultaneously the curved plate carrying the apertures
though which the markings are visible rises, so that only the
correct markings are visible. This is similar to the method used
in the AR77 except each of the six wavebands
on the B40 has its own aperture, making six apertures compared
with only one for the AR77, which switches into place to show
the selected waveband. There's a long chain that drives the tuning
gears and this is fitted with a tensioner so that backlash is
minimised. Also coupled to the tuning knob is a heavy flywheel
and a disk carrying 0-100 for logging. A clever mechanical gizmo
rises at each end of the tuning range to act as an endstop and
between the flywheel and the main tuning shaft is a friction
clutch set by a large locknut to prevent mechanical damage if
the knob is spun rapidly. Click
here or the picture above to see some detail. |
|
|
|
I fitted a temporary tuning knob
having a mounting bush that fitted over the original nut used
to hold the flywheel and turned my attention to the tuning condenser.
I noticed that a tuning direction change would result in a slight
knock coming from insde the screening cover mounted over the
tuning condenser. I found this cover is held in place by four
6BA scews, just about accessible. I removed these using a magnetic
screwdriver and tried to lift off the cover. I found it's only
detachable when the rear of the tuning spindle coupler is precisely
vertical. The tuning condenser is 4 gang and held in place by
what?? there are wired connections to the turret tuner and earthing
braids to the chassis, but nothing securing the frame to the
chassis other than three loose screws mounting it to springs.
When the spindle turns one way the various wires stretch to their
limit before the vanes move, then reversing spindle rotation,
the frame rotates, compressing the connections, again before
the vanes rotate. |
|
The right hand end carries the flexible
coupler but the left hand end has nothing coupled to it. As you
can see the centre shaft is a ceramic material. The left end
of the tuning condenser frame is secured by a single 4BA screw
into a springy mount. The right end has two 4BA screws and all
seem to be very loose. For some reason the cover has a pair of
shims (you can just see the RH one) which I must ensure are in
place when I refit it. I need to tighten the loose screws, lubricate/clean
any moving parts and confirm the thing is steady then confirm
there are no broken wiring connections to the stators. |
 |
|
 |
These are the loose screws.
At the left is a springy mount onto which the upper screw should
be a tight fit but its so loose the whole frame can be lifted.
On the right I can't see anything securing the frame to the chassis.
The bracket held by the RH screw is not secured to the bracket
held by the LH screw.All it does is prevent the front of the
frame fifting more than an eighth of an inch. Clearly the whole
assembly was designed to be relatively flexible to prevent damage
from mechanical shock. |
 |
|
|
So there it is.. the reason
for severe tuning backlash and the difficulty in tuning 40m SSB!
Fixing the problem was easier
than I thought because in the rear panel directly behind the
loose screw was a hole through which you can insert a screwdriver
to tighten the wayward screw. Clearly someone had recognised
the fault in earlier examples and a quick fix was possible. Refitting
the screen which I'd detached to investigate the problem was
tricky. Each end of the screen is supported on metal plates which
might be shims fitted to raise the height of the screen. There's
only just enough room to use a long thin magnetised screwdiver
to replace the 6BA screws. The front was easier and I loosely
fitted the first two screws, but in order to align the rear of
the cover you need to use a mandrel, such as a thin phillips
screwdriver, to line up the holes through the cover, the shim
and the chassis. Once this is right you can line up the other
holes and fit a screw. Once this is loosely fitted, extract the
mandrel and fit the last screw, then tighten all four.
I can now tune SSB much more
easily, but there is still a little backlash due the the action
of the chain used to turn the tuning condenser, although further
work to lubricate this and its tensioner might improve things.
I also looked at the S-Meter. Taking it apart revealed some black
flakes stuck to the cylinder around which is the coil. I found
that blowing through the coil area cleared the needle from sticking.
The black flakes had come from the back of the magnet whose paint
was in poor condition.
Looking back to the second of
the two tuning problems. The pulling effect is now resolved having
tightened one loose screw, but the strange effect in the IF response
curves seems to be from the "Miller Effect". This was
suggeseted by Michael G8MOB, then Pete G4GJL sent me an excerpt
from the G2DAF receiver write up which describes the effect in
detail, below. The two increasing peaks that appear when the
input test signal is increased are the result of extra bias from
the AGC circuit affecting the careful settings of the IF transformers.
This effect also appears when tuning across a strong broadcast
signal such as Radio 4 on 198KHz when a long wire aerial is used.
In this case the signal is spread across the range 192 to 200KHz
instead of peaking properly at 198KHz. My guess is its just a
foible of the B40 and should remain so for authenticity. In fact,
as AGC is also applied to the second RF amplifier I'd expect
the S-Meter to get fooled as well. The solution, especially when
encountering very strong SSB is to switch off AGC and reduce
the RF gain control. I say this because I noticed strong 40m
SSB was tricky to tune correctly with a rubbery feel to the tuning,
but as resolving SSB wasn't on the cards when the receiver was
in the design stage I can at least not blame the designer for
this failing. Click
the excerpt below to read the whole book. |
 |
|
I now need to carry out
a final check on alignment after reassembling everything because
I noticed that after fixing the tuning condenser the dial no
longer read correctly. This might be partly explained because
I applied switch cleaner to all the moving parts and probably
slightly altered the local oscillator current paths. The other
reason is that I freed up the cylinder carrying the dial markers.
This was seized and possibly off-centre. There's also the noise
limiter to fix because it kills audio when switched on.
A word about the response curves
shown below.Firstly, these do not show actual power levels. For
example the first curve shows a peak of -75dBm and a baseline
of circa -104dBm. The difference between these readings is purely
indicative, which is really what we're after. The spectrum analyser
is feeding a swept signal (50KHz) centred about 500KHz at a level
of 0dBm from a 50 ohm source via a 220pF capacitor into the plug
connecting to C201 and hence the tuning coil at the grid of V201.
This is easier than feeding the anode of the mixer which would
be slightly better but awkward to do as it's at the opposite
side of the chassis from the IF Unit. Output to the spectrum
analyser is taken from the anode of the detector V204b via a
20pF capacitor anchored to an adjoining tagstrip, but the signal
goes through a high impedance probe. This is necessary for three
reasons... (a) to isolate the 50 ohm input of the spectrum analyser
from any inadvertent high voltage, (b) to avoid damping the high
impedance receiver output and (c) to minimise any detuning of
the IF circuit due to the capacitance of the probe. The
probe has an indeterminate loss of about 30dB, but will remain
at a constant level during measurements. The tracking generator
50 ohm output feeds the IF Unit via 220pF, again to avoid damping
the input, and at 500KHz this will have an impedance of about
1.5Kohm, so you can see the 0dBm at 50 ohms (=225mV) translates
to something like 7mV at the input to the IF amplifier. This
equates to a loss of about 30dB, so the effective tracking generator
output will be circa -30dBm.
Summarising.. the IF input is
-30dBm and the IF output is seen as -75dBm. The latter figure
appears a lot smaller than it actually is, by the loss through
the probe (30dB) and the loss due to the 20pF capacitor. Although
this capacitor has an impedance of about 16Kohm at 500KHz the
probe input impedance can be assumed at least 5Mohm. The detector
anode will be recorded as 1.25mV referenced to 50 ohms, but if
the detector circuit impedance is for example 5Mohm, the anode
voltage will really be circa 400mV. Very roughly (and this ignores
the state of the probe batteries which could easily increase
its loss by a further 10dB) the actual numbers involved will
be 7mV IF input and (at least) 400mV IF output, which makes sense... |
 |
Receiver B/W Setting 1KHz
Horiz: 5KHz per division
Response: +/-1KHz
Centre:500.05KHz |
|
 |
Receiver B/W Setting 3KHz
Horiz: 5KHz per division
Response: +/-2.5KHz
Centre: 500.05KHz |
|
 |
Receiver B/W Setting 8KHz
Horiz: 5KHz per division
Response: +/-7.5KHz
Centre: 500.05KHz |
|
 |
Receiver with BFO on
Horiz: 5KHz per division
Frequency: 500KHz
Low was 498.13KHz
High was 501.13KHz |
|
|
The curves show that with modern
test equipment the old receiver can be set up to produce excellent
results. The initial basic (best) results using a signal generator
and wattmeter were pretty poor, with multiple responses from
strong signals and hopeless crystal filter performance, all no
doubt aggravated by interactive effects of signal strength and
detuning (Miller Effect).
During the tests I did not bother
to disable the local oscillator and therefore the receiver was
actually trying to work on its lowest waveband. Using older test
gear it was customary to disable the local oscillator but I'd
rather leave it running to permit realistic noise levels to be
seen. Whilst this doesn't materially affect the shape of the
response curves, disabling the local oscillator would certainly
reduce the level of the noise floor (measured at about -104dBm)
to possibly -130dBm or better. Comparing with historic response
curves, the results above could therefore be significantly better
than the 104dBm-75dBm= 29dB indicated, perhaps by at least another
20dB, to something like 50dB.
The front of the case is now
back in place and the knobs all re-fitted. Intially I'd tweaked
the IF alignment as this may have changed following the inevitable
bumping around and slight earth current chages around the tuning
condenser from tightening and the use of switch cleaner on its
moving parts.
After the final tweaking, the
IF was centred nearer to 500KHz (about 500.05KHz) than previously
(501.67KHz) and the IF amplifiers appeared to develop 29dB above
the noise floor rather than 25dB previously because I used 100Hz
rather than 1KHz for RBW/VBW. The figures for bandwidth match
the numbers in the B40 manual. The centre BFO frequency (used
in TUNE) should ideally be dead accurate for zero-beating a signal
and conveniently around +/- 1.5KHz (CW, HIGH and LOW) for resolving
USB and LSB but these are not critical. |
|
|
|
|
|
|
proceeding |
|
|
