Analysing the PR155 circuits
|
|
I re-read this page after around 12 months.
Because of other commitments (ie. lift repairs and trying to
organise retirement from these plus tackling my R1475) I'd shelved
the Plessey receiver (till later). I found the text really confusing
so I'll try to explain the reason for this in case you'd like
to plough through the report.
When I first bought this set I tried it out briefly,
found it worked after a fashion, mentally noted some puzzling
features then put it away.
Essentially the PR155G is a bit different to the
standard PR155. Because it was a "special" version
supplied to MoD that might explain its differences. In carrying
out the changes I reckon the designers made a few design errors.
Instead of using standard PR155 modules they were told to design
and develop new versions of some. Maybe this was to save money
on parts or most likely (I believe) a ploy to keep the PR155
team alive and well. I suppose a change to overcome certain design
weaknesses might have been an excuse but whatever they did introduced
further errors. One really annoying difficulty with the PR155G,
as far as modifying it for ham radio use, was the difficulty
of adding SSB reception.
As I checked the operation of my example (below),
and without in-depth knowledge of the operation, I made some
mistakes. Fortunately this came to light as I tried to get the
set working properly on LSB and USB (these modes were never included
in the PR155G). Not only are design weaknesses exposed in the
"G" version, but the Technical Authors responsible
for documentation slipped up as well.
From the evidence someone has tried before to get
the set working (eg a blown SN7272)
Read on.... |
|
I decided to undertake this task (ie making it work
as I'd like) even though nowhere (as yet) have I found an accurate
circuit diagram of the PR155, either as an accurate document
or reflecting my own PR155G variant. This variant I understand
was put together for users not requiring anything other than
CW and AM reception, but as it's very likely that all the features
are available within the various modules within the chassis for
including SSB I plan to modify my example of the "G"
variant to add USB and LSB. Before this is possible I need to
figure out the wiring for my particular "G" variant
because it would appear that the technical author responsible
for the "G" user manual forgot to change several parts
of the circuit diagrams from the "B" variant (or whatever)
he had at his disposal. As you'll read below I was somewhat caught
out because the module circuits in the B and G versions are different
and most likely, as changes were made, circuits within the modules
of different ages were modified. Changes, partly perhaps due
to feedback from military users, or even in-house improvements
suggested by the design and development team at Plessey. "PDS"
or Post Development Support was a very lucrative thing for Plessey
as essentially it was "cost plus" and could include
all manner of activities.
I expect the Plessey engineers were instructed to
design "standard" modules which could be used in the
different variants. As one of my aims was to add LSB and USB
to the mode switch in place of settings "5" and "6"
I examined the earlier "B" variant.. The circuits of
B modules that are specific to LSB and USB handling are shown
below. I imagine that certain users may have requested improvements
in performance so it's possible that the odd module differs from
standard. As I was a design engineer at Plessey, albeit in computer
development, during the time the PR155 was being developed I
feel somewhat familiar with their design processes.
Below is the "standard" Module 11 which
is used for generating the oscillator for upper and lower SSB.
Apart from this is the BFO oscillator which needs to be tunable.
That will probably be the same for all variants offering CW.
Wrong! Module 11 is completely different in my "G"
variant.
Why is the one shown so complicated? My guess is a
PDS team was put together after the receiver was launched and
the engineers did what engineers liked to do... make things more
complicated. |
 |
|
Above is the version of Module 11 that is included
in the B variant manual.The natural frequency of the oscillator
for LSB is governed by a set of components around VT1-VT3 together
with the negative tuning voltage introduced at Pin 3 and modified
by additional components introduced by selecting USB via Pin
2. Supply voltage is at Pin 4 and ground at Pin 1. The RF output
is a coax connection from VT14 and it uses loads of transistors
and zener diodes! |
 |
|
Above is the "G" version which is
dramatically simpler. Selection of LSB and USB is achieved you
might suspect is as before by selecting Pin 2 or Pin 3. Transistors
VT1 and VT2 are crystal controlled to 10.6 and 10.8MHz. |
|
Next is Module 12.. again the first example
using a divide by ten TTL chip not fitted in the G variant. |
 |
|
A quick glance at the circuit above shows this
circuit divides an incoming 1MHz signal at PL28 (and output at
SKT 23) by 10 producing a 100KHz signal presented at CAL OUT
which is used when the CAL setting is selected at the mode switch
S2. Minus 15v applied at "-15 CAL" or at "-15v
SSB" resulting in 100KHz (SKT 21) and CAL OUT (SKT 22). |
 |
|
Again, the design of Module 12 above is simpler
in the G variant, but uses transformers which may have been a
bit expensive and awkward to source.
Below Module 7. Click to see in larger format. This
carries twin boards in module 7 whilst the G variant has two
modules labelled "7A" and "7B" located either
side of module 8. |
 |
|
For my purpose ie. the addition of SSB... the
fact that a product detector is used for CW means that given
a fair wind the Module 7 fitted in my "G" variant will
be good for SSB. All that's needed is for the mode switch to
direct LSB or USB from Module 11 to Module 7b. |
 |
|
Above is the G variant (separate Modules 7A
and 7B) which has only relatively minor changes to the version
in module 7. One interesting point to note is the flip flop added
at the BFO input. This may have been included because the BFO
running at 100KHz was reported to be troublesome.
My receiver has a long-standing fault. The RF gain
control doesn't work properly and faint pencil markings suggest
the previous owner just put up with the problem, however the
main task is to add LSB and USB and to do this means tracing
my "G" S2 wiring and produce a circuit of it.
Below is the circuit of S2 (comprising 4 poles) given
in the "G" manual. |
|
The markings at S2 are almost correct for the G variant.
Oddly the circuit includes use of both LSB and USB at S2AF. As
I do not plan to rewire S2 I'll be adding USB and SSB to the
unused sections of S2 (ie the pair labelled "5" and
"6").
As far as I can tell the two connections to Module
11, Pins 2 and 3 select LSB and USB respectively. S2 position
1 which is not marked on the front panel apparently selects USB,
with CAL, CW and AM using LSB.
S2AB/2 turns on Module 11 for a CAL signal at 100KHz
and turns on the BFO at S2AB/3. It apparently also turns on Module
11 for unmarked switch position S2AB1/1.
S2BF sends either position 1, CAL and CW to Module
7 product detector plus AM to Module 7 AM detector.
S2BB handles the BFO and CAL. |
 |
|
 |
First what I believe to be the required changes
to wiring in my G variant before tackling the SSB mod.
Switching around the two wires to Pins 2 and 3 of
Module 11 rationalises LSB and USB.
In addition, I found later, is a change to the RF
gain control circuitry. |
|
|
Next the wiring for the G variant with LSB and
USB added at S2/5 and S2/6 |
 |
|
|
Below is a block diagram of the PR155 (with
tech author errors). Click to see it in larger format. |
 |
|
The only thing stopping me from proceeding with
modifications is the fact that there are several potential faults
with my example of the receiver as well as a few parts of the
wiring I'm not happy about. Another puzzle is the use of an extra
connection at S2AB & S2BB which I've labelled as "8".
At first sight this appears to be a duplicate of "1".
Sure enough, when I examined the switch connections 1 and 8 were
basically the same except the postion on the wafer demanded a
second pip on the rotor.
What I couldn't figure out was why Pin1/8 was connected
when S2 pointer wasn't labelled at the first position. Operating
the receiver however proved that position 1 was wired for USB
(I intend to switch around the output wires as mentioned previously).
All that was necessary therefore was to copy position 1 for all
four wafers to position 5 and sure enough a set of crocodile
leads proved this was so. Next, I'll solder a set of four short
wires to replace the leads. Then I'll work out how to wire position
6. Ostensibly the wiring will echo that for position 5 except
I'll need to pick up the alternate connection to the oscillator
module.
Quite a lot happened. Firstly I got USB working as
planned merely by connecting tag 5 on each wafer to tag 1. For
an odd reason Position 1 although marked with a "-"
was wired for USB. However as I tuned around to test this on
14MHz there were a number of crackles and not long afterwards
only AM was working. Maybe I'd upset something? I then discovered
that if I switched off AGC I could set the RF gain cotrol to
a very critical setting and recover CW etc. I decided to remove
various modules and check their components. Apart from a few
small electrolytic capacitors I found the resistors were not
too accurate but not too far out to explain the fault. During
my checks I must have shorted something vital and after detaching
module 8 I found a bad transistor. I swapped this for a BC108
and AM was restored so back to the original fault.
It was only when studying the circuit boards against
the diagrams to try and locate the fault that I found lots of
changes. Let me explain further... I have two manuals.. one is
rather poor quality and the other is pretty good so having wrongly
assumed the circuits in each were the same I'd settled on the
better copies. Alas it's the set of poorer copies that better
reflect my circuit boards in the modules.
I'm now of the opinion the fault which occurred during
testing lies in the AGC or RF gain control area as a careful
setting of the latter restores operation.
The contruction of the receiver is interesting. Loads
of coax leads fitted with plugs and sockets with the modules
being soldered in place. The latter seemed very awkward to remove
but I quickly discovered the circuit board pins were soldered
to sliding solder tags. One heats each joint and merely sldes
the mating solder tag to one side. Each module is secured to
the chassis by a couple of small screws. Remove their nuts (6BA
socket) and the module slides out. Remove two more nuts and the
board can be slid from the outer case. A very clever idea as
it eliminates bad contacts if the modules were to be fitted in
sockets.
Below is module 8 from the G variant. I'll also show
the RF gain control wiring. |
 |
|
The next topic covers the RF gain control. I believe
this circuit is bad. Either it has an early spec that doesn't
work or it was wired incorrectly in the factory because operation
is hopeless and calculation of voltages proves it cannot work. |
|
The RF gain control wiring is simple enough
taking a feed from the neg 15v supply. However the results in
my receiver are bad because in AM mode only the last fraction
of the pot produces sound and in CW etc the position of the pot
is extremely critical.
Something is wrong in the wiring or in one of the
connected modules.
Variants other than G are very different with the
pot connecting into Pin 5 of the AGC module rather than directly
to the amplifiers. See below. |
 |
|
|
The AGC switch has a manual setting labelled
"OFF" where the RF gain pot wiper connects to four
amplifier modules, and "FAST" and "SLOW"
when the AGC module 8 deals with overall control. One would think
that if the AGC module works for AM then turning AGC off would
enable the RF gain pot to work properly (which it doesn't). The
receiver fault was initially everything working except incorrect
manual RF gain, followed by only AM working with only fast and
slow selected. It's possible that one of the four AGC controlled
modules is a rogue, or the receiver has a factory error.
Next I measured the AGC voltage at various switch
settings. I found that AM gave realistic AGC voltages but CW/SSB
were clearly bad. But of course, thinking about things, AGC will
be poor if there's a lack of signal ie. Radio 4 gives me a good
AGC voltage in AM mode but poor in CW/SSB modes. By carefully
adjusting the RF gain pot Radio 4 appears in CW and SSB modes
and I noticed very faint pencil marks at this setting so the
last owner must have been using the set and found this RF gain
setting to work. However CW/SSB do not work with AGC on. |
 |
|
Above for variants other than G, Module 8 circuitry
includes an input from the RF gain pot at Pin 5.
At this point I decided to read the intructions ie.
the PR155G manual. This revealed a few points I hadn't appreciated
but cleared up a few puzzles. The G variant BFO actually runs,
not at 100KHz but at 200KHz. The same goes for the calibration
signal injected at Module 7B. It explains the "flip-flop"
note on the circuit diagram of Module 7B which converts those
signals to 100KHz. The reason for the change (B to G) might have
been associated with stray pick-up in the third IF amplifier
which runs at 100KHz. |
 |
From the description in the manual I sketched
out this block diagram to help figure out where the two problems
reside ie. the strange RF gain operation (signal output restricted
to only a few degrees of rotation) plus the deafness in CW/SSB
using AGC. By very carefully setting the RF gain pot I can hear
plenty of signal for CW/SSB settings but the level is so high
it completely over-rides the BFO and the 10.6/10.8MHz oscillators.
One complication in fault-finding is the totally screened
construction. One option is to make up a circuit to intercept
signals in the various coax cables many of which have break points.
I need to find suitable connectors (vaguely similar to SMB) to
do this properly.
Below is the circuit for Module 5 which uses the 10.6/10.8MHz
oscillator output from Module 11. |
|
 |
|
Above you can see the AGC input at Pin 3 and a second
input at Pin 2 marked "Reference". This connects to
Pin 2 on modules 4 and 7A also. Module 5 decription fails to
mention this signal but from Module 7 circuit you can see it's
a neg 4.7 volt reference voltage marked "REF". What's
the purpose of REF? At this point we need to note that the supply
line is minus 15 volts so when you're looking at logical events
you have to bear this in mind. So, if the AGC voltage on Module
7 drops below neg 4.7 volts diode D2 conducts and pulls down
the local AGC voltage to around neg 5 volts. On Module 5 the
reference voltage clamps the local AGC voltage to a maximum of
neg 4.7 volts although it can drop to a level less than this
based on whatever the incoming AGC voltage if this is less than
neg 4.7 volts. Module 4, 1st IF amplifier and 2nd mixer which
is also AGC controlled, and in this case the AGC voltage is also
clamped to around neg 5 volts by the neg 4.7 volt REF input to
this.
I finally got around to clearing my workshop of lift
repair stuff and was able to move the PR155 onto the test bench
where I was able to use a signal generator. This repeated the
results I was getting earlier with my standard Radio 4 broadcast,
but I also noticed a further problem. With the mode set to AM
I found very little by way of recovered audio when inputting
a 5MHz AM signal. As the level increased the tuning dip clearly
got deeper but audio disappeared. I tried 30uV to 100mV and no
audio to speak of. Could there be a problem with Module 7B? One
way of checking is to look at the 100KHz RF output on the back
panel as this is the same signal that if fed to the detector
circuits. I decided to check this 100KHz output with my spectrum
analyser and found a few interesting things. One was that with
AGC off the RF gain control operated over only the final 10%
of its track. As the control was turned anti-clockwise the receiver
instantly cut off which surely must be wrong and, as this may
be a clue to all the odd problems, it's time I investigated further.
The obvious thing to check was the potentiometer although I'd
previously looked at the control voltage from its wiper which
seemed continuously variable. The pot was fine with a resistance
of 69 to 402 ohms to ground. This suggests an error because the
resistor to ground should be 120 ohms.
The manual gain control, RV1, is active only when
AGC is turned off so takes the place of the AGC line at the four
amplifier modules. The action of a voltage on the AGC line whether
derived from the incoming signal or the wiper of the RF gain
pot is to produce a degree of attenuation at a switching diode
in Modules 4, 5 and 7. In Module 1 the action is similar but
more complicated. See below.
As far as I can see none of the four AGC control circuits
should switch off the receiver as the RF gain control reduces
overall gain, so it's possible that one of the four has a long-standing
fault (or the RF gain circuit components are bad). I decided
to remove the AGC connections and monitor the voltage with the
RF gain at either end of its travel. Module 1 reduces the voltage
from neg 10 to neg 8 volts. The other modules do not change this.
I calculated the voltages at RV1 using the book values of the
components and found the max/min voltages should be neg 1.55
to neg 6.199 and not neg 1.5 to neg 10.5 so maybe a resistor
is high. Given the correct voltages and including Module 1 current
draw (which increases the AGC voltage by about 2 volts.. don't
forget we're working with a minus 15v power supply). I reckon
the AGC line should range from neg 1.5 to neg 4.2 volts and this
pefectly fits in with the observed voltages when AGC is active. |
 |
|
Can it be as simple as this?
The RF gain control is hidden by S4 and awkward to
get at but, using the component values in the manual, I calculated
the voltages shown opposite.
These actually measured neg 10.5v and neg 1.5v with
Module 1 reducing the higher value to neg 8.5v.
The picture of S4 is very poor but you can see the
switch looks a bit odd and for the life of me I can't see why
the designer chose such a switch because a simple 3-pole 3-way
would have sufficed.
I only discovered this because in trying to access
the resistors at the RF gain pot I inadvertently broke the switch
wafer leaving repair with superglue or replacement as the options.
Once the old rotary switch was moved I found the (hidden)
resistors at the RF gain pot were covered in verdigris and NOT
marked as shown in the diagram opposite. |
 |
|
|
I took several pictures of S4 and RV1 and found several
strange things. Firstly S4 includes a number of unused tags not
part of the switching arrangement that seem to be anchor points
for components. Secondly R10 seems to have been replaced by two
220 ohm resistors in series. The 680 ohm resistor is present
but connects with a capacitor marked 47uF which must be C3 to
ground (ie. C3 is not decoupling the wiper). Resistor R11 appears
to be marked 300 ohms. |
 |
I need to check my findings because it's difficult
to see the parts involved in the confined space.
The green electrolytic is marked 47uF (C3) and the
blue wire goes to the wiper of RV1 (below left). The other end
of C3 is grounded to the tag carrying the black ground wire.
Below the green wire is the AGC feed to Pin 3 of the
AGC controlled amplifier modules. Yellow goes to Module 8 Pin
3 and the blue plus pink you can see on the left picture (RV1
wiper). |
|
|
|
 |
Above left is the 1Kohm RF gain control pot.
This is shunted by that 100 ohm resistor and connects to a 680
ohm resistor to ground and a pair of 220 ohm resistors in series
to neg 15 volts (that purple wire seen left).
Nothing like the circuit in the manual.
One 220 ohm is shown left with the second connecting
to the solder tag and just visible.
Those are solder tags either side of the tag carrying
the white/orange wire .
Left, C3 the electrolytic positive end in pink sleeving
connects to a solder tag which has a wire linked to the grounded
(black wire) switch wiper hidden by the capacitor.
The white/green wire goes to Module 8.
The purple wire is from neg 15 volts and connects
to the 220 ohm resistor.
Again I need to check these before wiring up a new
circuit. |
|
|
Looking at the revised circuit taken from the
receiver my first thought is perhaps the area, which is not easy
to get, at was wired incorrectly.
The variation and level of control voltage looks to
be completely wrong bearing in mind the AGC line is clamped in,
for example the detector Module 7.
In that amplifier module the manual control voltage
of between say neg 8.4v and neg 9.5v will see a zener voltage
of neg 4.7v.
That means manual control of that module through RV1
will be clamped.. certainly not right and certainly explaining
the strange action of the RF gain control.
This doesn't quite explain the very limited control
exercised by RV1 (maybe no more than 10% of the pots travel)
because all of the gain signal is clamped.. however.. read on.
I plan to rewire the area using a new switch, but
the old pot, with resistors shown in the manual but allowing
for changes to be made if necessary by fitting the new resistors
at the top of the space available. |
 |
|
 |
|
|
|
A new 4-pole 3-way switch will be fitted and
wired as opposite and including parts located in the immediate
area such as the RF gain control, RV1 with its resistors etc. |
|
|
|
Gain pot RV1 wiper wired to S4 as shown, RV1 min setting
goes to ground via new 120 ohm resistor, RV1 max setting goes
to neg 15 volt via replacement 680 ohm resistor and shunting
RV1 is a new 560 ohm resistor. Connecting to the blue neg 15
volt connection is a grounded (to black wire) 50uF x 25v capacitor.capacitor
(= C3). As in the original design I'll use some of the unused
solder tags (the unused pole) on the switch to anchor components.
RF gain pot resistors will be positioned for easy
changing. |
 |
|
|
Subject to checking... the wires in the switch area
exit in a harness so awkward to follow but are as follows:-
AGC line to amplifier modules (MX/3) = green
Ground (pos 15v) = black
Module 8/3 = yellow
AGC amplifier Module 8/4 = white/green
AGC fast to Module 8/2 = white/orange
Neg 15 volt supply line = purple
RV1 wiper = blue |
 |
|
|
I removed the two parts of the broken switch wafer
and cut the resistors from RV1 which is held half way down the
fron panel by a blue wire that goes onto a tightly bound harness.
Much to my surprise the resistors were not what I'd expected.
The shunt was 100 ohms as expected but the others were 22 not
220 ohms and the other was 68 not 680 ohms. I've redrawn
the circuit below to see exactly what this means. |
|
The current draw via the pot is 50mA so there's neg
11.6v dropping to neg 9.6 across the 68 ohm resistor and neg
8v dropping to neg 6v
In Module 7 amplifier module the manual control voltage
of between say neg 3.4v and neg 8v will see a zener voltage of
neg 4.7v .
That means manual control (except for the extreme
clockwise setting) of that module through RV1 will be clamped..
certainly not right and certainly explaining the strange action
of the RF gain control.
This explains the very limited control exercised by
RV1 (maybe no more than 10% of the pots travel) because nearly
all of the gain signal is clamped..
From experimentation I'd found Module 1 has a large
effect on the overall acction of RV1. Module 1 in fact has a
preset gain setting for amplification which dramatically affects
RV1 setting such that any fiddling with the former (or component
ageing) will mess up the action of RV1.
I plan to rewire the area using a new switch, but
the old pot, with resistors shown in the manual but allowing
for changes to be made if necessary by fitting the new resistors
at the top of the space available. |
 |
|
|
Up to now I've been analysing the G variant but, for
interest I looked at the B variant in respect of RF gain control.
Thereby lies the answer to the G variants problems. The system
is entirely different. In the B variant the AGC line is basically
taken over by the manual control setting. The answer to the G
problems I guess will be to set the RF gain to match the actual
G variant circuitry. Module 1 for example draws current from
the RF gain pot circuit. The designers must have chopped and
changed their ideas because the manual shows a current draw through
the pot circuit of about 13mA but in my example (with completely
different resistors) the draw is 50mA. The higher current is
probably used to lessen the effect of current draw at Module
1 and will therefore require less fiddly adjustment of individual
module gain settings. My plan to make the resistors accessible
is going to help. I guess another way is to experiment by setting
up variable resistors external to the receiver and work out how
to give RV1 a smooth gain adjustment. It was only after removing
the old switch and the pot I discovered the pot was u/s. AGC
action within the modules even with AGC off had masked the bad
pot. I fitted a new 1K pot and things improved dramatically,
although I had to add an extra 330 ohms in series with the 680
ohm resistor to get smooth gain variation over the full range
of the pot. The circuit now has 120 ohm to ground, 560 ohm shunting
RV1 and 1010 ohms to neg 15v. |
|
Once the problems with the RF gain control and AGC
switch had been fixed I tackled the deaf CW/SSB fault. These
modes had worked after a fashion but only at a fraction of the
audio of AM. I checked both the SSB and the BFO local oscillators
and these were 496 and 660mV respectively so the problem appeared
to be in the product detector on Module 7B. I detached this and
looked at the board. I'd already swapped the electrolytics so
looked at the board under magnification. A thin coax cable running
from the front to the back of the board was corroded with signs
of vedigris. After testing continuity I discovered the centre
conductor had a break close to the pin end of the board. |
 |
|
 |
I also decided to check the board from
Module 7B carrying the flip-flop and started by measuring the
5-volt zener diode voltage used for powering the SN7472 chip.
Applying neg 8v with a current limiter set to low current I found
the zener voltage was only around 3.4 volts and the chip was
slightly warm so removed the chip and tested it. It tested as
faulty on my trusty test meter so went off to search for a new
one. I eventually found three examples. The first also tested
as faulty but the second was good. The 7472 is very unusual,
being a flip flop with AND gates although these aren't used.
An SN7474 would have been a simpler choice.
After fitting the new chip and a new length of coax
I re-fitted Module 7B and CW output was transformed with audio
now being consistent with that of AM and the BFO able to handle
the strongest signals. Previously the BFO heterodyne could only
be heard when the RF gain was turned right down.
SSB reception however was only possible by switching
to manual RF gain and carefully adjusting the level. The G variant
of the receiver is odd because it uses 200KHz signals for BFO
and CAL. It should also use 200KHz for SSB. This frequency is
halved by the SN7472 circuit (discussed above) unlike the other
variants which use 100KHz. The 200KHz SSB and CAL signals come
from Module 12, the design of which is completely different in
the G variant to accommodate the 200KHz frequencies. At first
sight (because of very congested construction at mode switch
S2) I cannot identify the termination of 200KHz SSB cable from
Module 12. It should be at output "21" with CAL at
"22". The drawing in the manual shows "12"
which is an error. The block diagram shows this connected to
the CAL setting of S2. Cable "22" is shown going to
the OFF/SBY/ON/CAL switch. I need to look at the two signals
with either a scope or a spectrum analyser as CAL should be rich
in harmonics.
Suffice it to say CAL doesn't work properly either! |
|
|
If Cable 21 does in fact carry a clean 200KHz
carrier it should provide correct operation of the product detector
in Module 7B and should therefore be used for SSB as well. Ostensibly
both LSB and USB at S2 settings 5 and 6 should echo those of
the CW setting but for some reason SSB isn't working.
Something I'd overlooked now seems to be the source
of deafness on the SSB issue. Module 12 is supposed to develop
a 200KHz signal which is derived from an internal 1MHz oscillator.
The signal is present but measures around 496mV compared with
the BFO of 660mV. My guess is the SN7472 circuit cobbled together
to halve these 200KHz inputs cannot deal with the lower voltage.
To prove a point I removed the pcb from Module 12 and tested
it on the bench. It's a weird arrangement of transistors and
coils that are designed to mix an internal 800KHz with 1MHz and
turn out 200KHz. I saw the output at less than 500mV but although
most of the resistors measured a little high nothing seemed amiss.
I did discover that adding 68nF across a tuning coil bumped up
the output to over 600mV (although it was an odd shape with alternate
cycles of slightly different levels) but refitting Module 12
resulted in the lower level once again. On a hunch I unplugged
the 1Mhz signal and the output went up to the 600mV I'd seen
on the bench and SSB worked proving the low level problem I'd
surmised. One puzzle is how the 200KHz signal could be derived
without the mixing process (ie. without the 1MHz input). It must
be to do with T3 being sharply resonant at 200KHz. I need to
repeat the bench experiment and ensure the 800KHz oscillator
is in fact on the correct frequency and of sufficient amplitude.
Below I've copied its schematic. |
|
|
|
In other variants Module 12 has a divide by
10 chip which produces a 100KHz signal used in the product detector
to resolve SSB. It's a pity that version of Module 12 couldn't
have been amended to incorporate a selectable divide by five
option which is actually embodied in the SN7490 used for the
divide by ten. The corresponding 100KHz output from the above
is dependent on the SN7490 output plus a filter followed by a
pair of 270 ohm resistors. The SN7490 output is 2.4v to 3.4v
and this results in say 1.2v to 1.7v neglecting the effect of
the low pass filter (100uH with 10nF). If we add the effect of
the filter I reckon the output is between 800mV and 1100mV. Module
7B with the flip-flop supplies about 2.4 to 3.4v into the product
detector via an 820 ohm resistor.
The answer would seem to be to raise Module 12 output
to at least the level of the BFO (600mV) which works fine. Sadly
the Plessey engineers failed to record RF levels in the manual.
Perhaps this was because the levels were not very well defined?
I removed the pcb from Module 12 for the second time
tested all the transistors which were all OK and then powered
it up from a 15 volt supply. I noticed that a mod had been made
to the 1MHz input. This was to add an attenuator or matching
circuit comprising a series 33 ohm resistor with a 120 ohm grounding
resistor before C1. I fed in a one volt 1MHz signal from my signal
generator and looked at the 200KHz output. My scope couldn't
produce a sensible trace so I instead used my spectrum analyser.
The result (below left) was a very broad signal at 200KHz with
lots of noise. After prodding around to no avail I happend to
touch the top of transistor TR2 and magically the trace cleared
up to a decent 200KHz spike. How does one use a component in
place of one's finger? The answer was to solder a 100pF capacitor
from TR2 collector to ground (effectively in parallel with C4
(1000pF). The result is below right. I used a high impedance
scope probe to check the signals so ignore the levels shown below.
Later I noticed a tiny tuning slug that may have had the same
result.
I then reduced the 1MHz input and everything was OK
down to about 350mV. I checked and found the 1MHz signal from
the receiver was much greater than this, so refitted Module 12.
Will SSB now work properly? |
Module 12 output before tuning the input stage
(ie. without finger)
|
Module 12 output after tuning the input stage
(ie. with finger or extra capacitor)
|
|
|
|
|
pending |
|
|
