The CJD ELF/LF Receiver
|
|
That meter was odd.. the zeroing
screw had been tweaked and resetting it proved simple, but later
when testing I found the coil was open circuit. It's NATO code
is 6625-99-971-7538.
This outline sketch shows a
complete CJD receiver and you'll note a panel mounted above the
main receiver. This box carries some essential parts of the receiving
set-up, such as the power supply and their electronic assemblies.
Whether the basic receiver chassis
can be encouraged to work as a self-contained item remains to
be seen.
Click the
box to see more.
The basic receiver is pretty
heavy, weighing in at 97 pounds which is not quite as heavy as
another Navy receiver, the DST100 which
manages 110 pounds. Anything designed for the Royal Navy
needs to be solid enough to withstand the effects of the pyrotechnics
used at sea and I guess the CJD needs to be depth charge proof. |
|
|
The chief reason for developing
this receiver was obliquely referred to in my True
Story concerning the British nuclear deterrent as it was
in the 1960s. Only low frequency broadcasts can penetrate deep
seas and if one tunes across the VLF band many strange broadcasts
can be identified. These if listened to with a BFO turned on
seem to be jingly rather than carrying speech. Because intelligence
can be transmitted only commensurate with the basic carrier frequency,
once you get to the low end of VLF means it's typically teletype
style signalling that has to be used.
This CJD model can receive AM,
CW, facsimile plus single and double sideband. It can be tuned
normally or at the flick of a switch, and some carefully fiddling,
a synthesised oscillator can be used in place of its tunable
local oscillator.
The receiver above carries the
NATO code 5820-99-916-4903 and to the best of my knowledge contains
the front end and the majority of the IF amplifier. The plain
fronted unit with its code ending in 04 contains the power supply,
a final IF amplifier stage, detector, AGC, and audio output circuitry.
Oddly, because of its history the CJD uses a mixture of valves
and transistors. Whether this is just a matter of timing I don't
know, but valves are much more robust when it comes to EMP as
well as having superior technical aspects compared with early
transistors. The DST100 designed well before the CJD uses an
ancient RF amplifier valve (CV21=VP41) just because it works
better than newer types for pragmatic reasons I won't go into
here. |
The important detail for
a listener are the five wavebands covered by this set. The range
is 10KHz to 200KHz. The top end conveniently covers Radio 4 on
198KHz. Was this deliberate? It delivers 10mW for a signal of
1uV which is pretty good. What isn't run of the mill is its IF
amplifier (61.5KHz and 21.5KHz) with the lower frequency amplifier
more akin to hi-fi audio than RF. The circuit configuration has
the highest frequency range (Band 5 carrying Radio 4) as a dual
superhet using 61.5KHz followed by 21.5KHz, but as a single superhet
using an IF of 21.5KHz for the other wavebands. |
Band
|
KHz
|
1 |
10 to 23 |
2 |
23 to 41 |
3 |
41 to 71 |
4 |
71 to 119 |
5 |
120 to 200 |
|
|
|
Above, never mind the
width.. look at the quality.. two drums which are connected via
complicated bits of stuff to a 3-gang tuning condenser and something
else? The tuning dial has a rotatable assembly a bit like that
used in the CR100 that carries the KHz markings for the 5 wavebands
and a steel strip threaded around the drums. This strip carries
a pointer indicating the tuned frequency. It's clearly never
going to go wrong even if a bomb dropped on it unlike the really
flimsy arrangements used in Army kit of similar vintage. Compared
with the RA17 the above is an order of magnitude mechanically
more advanced. Even over-engineered one might be inclined to
think... but then again because of the reliance on ELF kit for
launching WW3 missiles maybe not...
Now, how difficult will it be
to reproduce the upper box. Using intelligence collected from
a number of sources it seems a circuit diagram is not too easy
to lay ones hands on. I heard one solution for copying the RF
section of the upper box revolves around a chip (LM373
probably the H, or can version which is easier to use) introduced
in the 1960s and improved in the early 70s. I've now ordered
one, whose date is yet unknown, from a supplier in Italy. It
arrived in mid-September 2021. The original circuit used a simple
diode detector for AM and a ring modulator for CW and SSB. I
also have a circuit diagram for a test unit used for the CJD
power unit from which I've gleaned most of the connections to
the flying lead emerging from the rear of the receiver box. An
adjacent coax connector probably has the IF feed. |
|
Pins numbered: Right to left, top to bottom
: B F L R V Z DD JJ NN, D J N T X BB FF LL, A E K P U Y CC HH
MM, C H M S W AA EE KK
So, for example, the mains circuits are the 4 pins
on the extreme right, top to bottom B, D, A and C
No problem trying to figure out the mating half
as Pete G4GJL kindly sent me one. |
PIN |
FUNCTION |
NOTE |
B |
Mains on/off |
Mains N |
F |
6.3VAC |
- |
L |
AGC long/short |
Long=+24V |
R |
minus 55V |
- |
V |
BFO on/off |
On=ground |
Z |
Screen |
For Pin U |
DD |
Ground |
- |
JJ |
Ground |
- |
NN |
Spare |
- |
|
PIN |
FUNCTION |
NOTE |
D |
Fuse F1 |
from Pin B |
J |
Rx Ready |
Relay ON=0 |
N |
600 ohm o/p |
- |
T |
Audio out |
Wiper of pot |
X |
+235V |
- |
BB |
minus 43V |
- |
FF |
+150V |
- |
LL |
Ground |
- |
- |
- |
- |
|
PIN |
FUNCTION |
NOTE |
A |
Mains on/off |
Mains L |
E |
6.3VAC |
- |
K |
+24V rough |
Lamps |
P |
600 ohm o/p |
- |
U |
Audio in |
End of pot |
Y |
Screen |
For Pin T |
CC |
+24V smooth |
- |
HH |
Ground |
- |
MM |
AGC1 |
- |
|
PIN |
FUNCTION |
NOTE |
C |
Fuse F2 |
from Pin A |
H |
+24V var |
Lamp dim |
M |
AGC on/off |
Off=+24V |
S |
+20V |
- |
W |
+20V |
- |
AA |
RF level |
To meter |
EE |
AGC3 |
AGC to RF |
KK |
AGC2 |
- |
- |
- |
- |
|
|
Information is beginning
to arrive from Pete G4GJL (thanks for the pair of 34-way connectors),
Peter G8BBQ and Paul G8GJA so what seemed like a hopeless task
is now starting to look quite promising and my £230 ("inc
delivery") investment may be saved...
Provisional pinning for 34-way
connector above.
The drawing below is that of
the test set for the CJD receiver which I've reproduced to indicate
the power supply requirements, hence a key to designing a suitable
PSU for the receiver sub-assembly plus the ancillary parts located
within the upper panel. Allowance for the latter needs to be
added to the power requirements below because the test set connects
only to the receiver sub-assembly. |
|
Voltage |
Test Load |
Load Wattage |
Current |
Notes |
6.3VAC |
4.7 ohm |
10W |
1.34A |
Valve heaters |
20V |
130 ohm |
3W |
154mA |
Transistor power |
24V rough |
220 ohm |
5W |
110mA |
Dial lamps |
24V variable |
220 ohm |
5W |
110mA |
Dial lamps as above |
24V smooth |
100 ohm |
10W |
240mA |
Relay power |
minus 43V |
150 ohm to -55V |
1W |
80mA to -55V |
Oven power + AGC |
minus 55V |
47 ohm |
40W |
1.18A |
Oven power etc |
150V |
10 Kohm |
6W |
15mA |
Valve local oscillator |
235V |
4.7 Kohm |
30W |
50mA |
Valve HT |
|
The receiver design is
slightly puzzling, almost as if it were the result of a committee
decision. It's either fitted into a cabinet for bench mounting,
called CJD(1) or as a five receiver setup CJD(2) or CJD(3) in
a floor-standing rack. The main receiver(s) carrying the controls
is either powered by a completely separate power supply mounted
above the receiver CJD(1) or in a rack with each receiver carrying
a small chassis bolted to the rear of its chassis. Almost as
if the designers ran out of space the PSU part is accompanied
by extra receiver circuitry again either in the box above the
receiver as CJD(1), or bolted onto the rear CJD (2) or (3). These
differences make some of the documentation a bit confusing.
To clear up which bits are where,
the main receiver includes a synthesiser made up from four parts
viz.a Phase Demodulator, Ring Modulator, Frequency Divider and
Harmonic Frequency Selector, secondly an IF Amplifier, thirdly
a BFO, fourthly the RF front end using a set of valves and an
oven for stabilising frequency, and it seems an audio amplifier
driving the loudspeaker. The parts not fitted into the main receiver
are the Power Supply with its various voltage stabilisers, the
AM/SSB/CW Detectors, an Audio Amplifier specifically for driving
a 600 ohm line and something called a Receiver Gain Equaliser
(developing AGC voltages from its own IF amplifier). These items
except the PSU wound components and those requiring heatsinks,
are fitted onto three printed circuit boards. Most of the parts
in the main receiver are sub-assemblies which can be removed
from their main chassis.
One of the slightly mystifying
features of the receiver is control of its overall sensitivity.
Not content with various gain controls and powerful AGC, hidden
under the rear of the chassis is a switch for enabling or disabling
the first RF amplifier. AGC is derived in the outboard circuit
which accepts the IF output of 21.5KHz from the IF amplifier
where it is also processed to derive AM, CW or SSB using the
BFO output as appropriate. Following this, also outboard is the
AF amplifier used, not for the internal speaker but, for broadcasting
around the vessel. Communication between the outboard circuits
including the power supply and the main receiver chassis is via
a cable with a 34-way connector (as covered previously) and a
pair of coax leads (IF + BFO). |
|
Next topic to investigate, now
I have the requirements (above), is the practical implementation
of a suitable power supply.
Below is the original design
which has slightly awkward transformer voltages. |
|
From an intial view of
the requirement I envisage using three mains transformers including
an HT transformer rated at a minimum of say 30VA to provide 235
volts (at 50mA=11.75W) and a stabilised 150 volts run from the
235 volt supply (15mA=3.5W). Heater supply for the valves from
the same HT transformer at 6.3 volts AC (at 1.34A = 8.5VA). This
has a secondary winding of 185V in the schematic above and will
produce a peak voltage of about 260 volts DC.
The second transformer has a
secondary of 70V providing a peak of 98 volts DC. This needs
to supply 55 volts at 1180mA and 43 volts 300mA (1480mA=145W
peak) and so will be much heavier than the first. The diodes
in the original design were CV7313 which are similar to the BYX38-600
or the common P600K 6A diode. A suitable alternative for these
however will be a standard full wave bridge rectifier such as
the SB2510. I'll need to search through my transformer collection
to identify something suitable for these two but the other lower
voltages aren't a problem and a small, junk box, transformer
can be used. Of course I'll use modern regulators to simplify
the design of the above. |
The IF and audio circuitry
will be based on the LM373H chip, an early device used typically
in the circuitry shown below, taken from a 1972 article by K4DHC.
As you can see this includes a BFO circuit which is not required
as this is already included in the main CJD case. Also included
within the main CJD receiver case are the various switches and
controls. All interconnections are carried via the 34-way connector.
Below is based on an IF of 455KHz but of course the CJD final
IF is much less at 21.5KHz whose frequency will match the filter
shown below. |
|
It will be a case of interfacing
the existing circuits into those required by the LM373H. The
filter shown in the sketch above will need to be in keeping with
the method used in the IF amplifier carried on the main chassis
and maybe a simple affair as it looks from the markings on the
front panel as if the IF response will be shaped before the detector
in the LM373H. I suspect the amount of audio power required for
the main chassis will need to be tailored to suit the small loudspeaker.
Mode switching is carried out via small 24-volt relays.
Below, I've marked the 34-way
connector table in purple where connections are required for
the new detector/audio circuit. Note there are four AGC connections
associated with the front panel AGC switch positions "Short"
and "Long" plus a separate feed to the valved front
end. Markings in red are for power supply connections. Keeping
the main chassis to its original design will unfortunately mean
mixing mains connections, DC and AC supplies plus low level signals
and control connections. |
PIN |
FUNCTION |
NOTE |
B |
Mains on/off |
Mains N |
F |
6.3VAC |
- |
L |
AGC long/short |
Long=+24V |
R |
minus 55V |
- |
V |
BFO on/off |
On=ground |
Z |
Screen |
For Pin U |
DD |
Ground |
- |
JJ |
Ground |
- |
NN |
Spare |
- |
|
PIN |
FUNCTION |
NOTE |
D |
Fuse F1 |
from Pin B |
J |
Rx Ready |
Relay ON=0 |
N |
600 ohm o/p |
- |
T |
Audio out |
Wiper of pot |
X |
+235V |
- |
BB |
minus 43V |
Oven control |
FF |
+150V |
- |
LL |
Ground |
- |
- |
- |
- |
|
PIN |
FUNCTION |
NOTE |
A |
Mains on/off |
Mains L |
E |
6.3VAC |
- |
K |
+24V rough |
Lamps |
P |
600 ohm o/p |
- |
U |
Audio in |
End of pot |
Y |
Screen |
For Pin T |
CC |
+24V smooth |
- |
HH |
Ground |
- |
MM |
AGC1 |
- |
|
PIN |
FUNCTION |
NOTE |
C |
Fuse F2 |
from Pin A |
H |
+24V var |
Lamp dim |
M |
AGC on/off |
Off=+24V |
S |
+20V |
- |
W |
+20V |
- |
AA |
RF level |
To meter |
EE |
AGC3 |
AGC to RF |
KK |
AGC2 |
- |
- |
- |
- |
|
|
I'll need to incorporate
a 12 volt regulator with the LM373H as the 24 volt rail is too
high for its 18 volt max rating. I understand the CJD design
has two printed circuit boards covering this area viz. an audio
power amplifier and the second for AM/SSB/CW detection and development
of AGC to control the RF and IF circuits. As the LM373H is designed
as a complete IF amplifier I may need to provide a suitable attenuator
at its input and because IF shaping is within the main chassis
its filter characterisitics can be relatively broad, perhaps
even as minimal as a simple coupling capacitor. For example the
impedance of a 100nF capacitor at 21.5KHz is 74ohms. From a perusal
of the collection of pages I have in hand it appears the IF output
will be around 1mV. |
Below is a set of drawings
(kindly supplied by Paul G8GJA) showing how the external detector,
AGC and audio circuitry connects to the CJD main chassis. You
can see in the first drawing, on the left the SSB/CW and AM detectors
selectable by a pair of relays; centre the provision for remote
volume control and the audio amplifiers, (the lower) one on the
main chassis for its loudspeaker and headphone sockets and the
second for 600 ohm output. Click the
picture to see more detail.
|
This next picture shows
the detector circuitry in detail where you may recognise the
straightforward envelope detector for AM plus a diode ring using
the BFO signal from the main chassis for SSB/CW detection. In
order to keep switching of sensitive signals local to this circuit
a pair of relays is used. Click
to see more detail.
|
This third picture shows
the AGC circuitry. For some technical reason perhaps a completely
separate IF amplifier is used to develop AGC. The reason is probably
to simplify the application of AGC to SSB/CW reception. In a
standard superhet AGC and AM detection are closely coupled but
in this case the two functions are governed by two different
circuits. As with the detector circuitry, a pair of relays is
used to keep sensitive signals local. Click
the picture to see more detail.
|
My intention is to use
the LM373 device, as described previously, to handle both detector
and AGC functions although I'll need to add some circuitry to
provide the delay feature and a fixed potentiometer to provide
different degrees of feedback (AGC1, 2 and 3). I can see the
need for some experimentation when it comes to exact voltage
levels associated with the new detector circuit and AGC output,
but this might be simply achieved by using preset pots.
Physical implemention of the
external features needs some thought. In the CJD rack the PSU
is carried on a chassis which also carries a cardframe for the
detector etc on printed circuit boards. |
|
|
|
I can understand having
an external power supply but I'm puzzled about why the designers
couldn't find room in the main receiver for the detector, AGC
and external 600 ohm audio line driver. Maybe there was a very
strict limit on the case dimensions, bearing in mind the requirement
for co-locating a number of receivers? There's also the unduly
complex design of tuning. This is very reminiscent of the much
earlier CR100 which must have influenced the writers of the procurement
spec. The use of valves in the front end almost certainly would
have been the result of MoD concerns about EMP. If not EMP, the
design was initiated in the days when transistors were limited
in performance and predominantly using germanium, although this
surely would have had little bearing considering the low frequencies
involved. |
Building the Power Supply
|
The voltages used in the
CJD Receiver are a mixed bunch and a search through my transformer
collection yielded a couple that should do.The first is a standard
HT/LT transformer (T1), once removed from a receiver, which has
a 256V-0-256V secondary (W3) plus 6.3V (W1) and 5.0V LT (W2)
windings. The second is an LT transformer (T2) carrying three
24V windings (W1/W2/W3). Power outputs are taken from the Test
Supplement. |
Voltage |
Current |
Notes |
Source |
6.3VAC |
1.5A |
Valve heaters |
T1 W1 |
20V stabilised |
150mA |
Transistor power |
T2 W3 |
24V unsmoothed |
200mA |
Dial lamps |
T2 W3 |
22V smooth |
200mA |
Relay power |
T2 W3 |
minus 43V |
80mA |
AGC |
T2 W1+W2 +T1 W2 |
minus 55V |
1.18A |
Oven power etc |
T2 W1+W2 +T1 W2 |
150V |
15mA |
Valve local oscillator |
T1 W3 |
235V |
80mA |
Valve HT |
T1 W3 |
|
|
As I develop the circuit for
the PSU I'll edit the drawing below which initially showed the
original design shown earlier. |
|
I've decided to build
the PSU on a handy piece of scrap steel. It's not aluminium so
it takes longer to work, but for heavy transformers it's pretty
good and I did get 65% in "O-Level" Metalwork back
in 1957. |
|
For appearance and safety
all the transformer wires will be under the chassis and terminated
on tagstrips even though it meant lots of drilling cutting and
filing. It probably doesn't matter too much but I arranged the
two cores to be at right angles. I'll drill as many holes as
necessary before mounting the two transformers.
The chassis measures 11"
x 7.5" x 2". I'll be fitting an IEC connector and an
umbilical cable will terminate in Pete's (G4GJL) 34-way connector
to mate with the CJD receiver.
The RF circuitry will be mounted
using a combination of veroboard plus components soldered directly
to tin sheets. The latter allows for very easy modifications
and the material is free from our recycling bin. |
|
I've realised that with
some fiddling the CJD receiver, once powered from its basic supplies,
will produce an IF output of 21.5KHz which if fed into an SDR
will prove its basic functionality. That way, as I usually do
when faced with a fairly major job is to check on the practicalities
of continuing and completing the job without wasting too much
effort. In fact I could have taken a shortcut and powered the
receiver from bench supplies. The 55 volt supply heats the oven
running at something over 1 Amp until the temperature reaches
69C. At this temperature a "Ready" lamp comes on and
a relay operates. The relay adds an additional 1.2Kohm resistor
into the series regulator drive. I'm unsure about this feature
but it may be used to merely reduce the dissipation of the 55
volt regulator or, alternatively to reduce the voltage to a level
where the oven current of maybe a couple of hundred mA allows
the temperature to just remain constant. I can delay the design
of the new 55 volt circuit until I've made some measurents.
A possible critical factor might
be the grounding arrangements. For example, mains wiring is carried
in the 34-way cable connecting to the on/off switch and fuses
on the receiver front panel using Pins A, B, C and D. Grounds
are Pins DD, HH, JJ, and LL. These latter connections cater for
safety earth as well as return paths for DC power and signal
earths. In addition there are two coax cables used which have
screen grounds. Poor choice of grounding will result in modulation
hum. A prudent change to the original design (without modifying
the main chassis) will be to use a mains on/off relay in the
PSU which is driven from the receiver on/off switch. This, with
the addition of a mains on/off switch plus a fuse in the PSU,
removes the requirement for mains connections in the umbilical
cable. This is slightly more complicated than it first seems
because, if the PSU is turned on via the front panel switch via
a relay, power must already be available for the relay coil.
In commercial equipment this power is usually provided by a small
standby PSU so I'll add a small low voltage transformer/rectifier
for this purpose. I fitted two mains switches, one to enable
testing without the main receiver and the second, for normal
use, to turn on the standby mains transformer for the relay coil
supply. Mains power will be provided by a small relay operated
from Pins B and D in the umbilical cable (which will now carry
24 volts for the relay coil). I'll add a blue LED to indicate
standby mode and a green LED to indicate the main receiver is
switched on. A feature shown in the drawing below is to bar the
PSU test facility if control has been passed to the CJD receiver
ie. testing of the external circuitry can only be done if the
standby switch is turned off. |
|
A couple of views above
showing construction awaiting wiring up. The two transformers
are pretty heavy and conservatively rated for the job but a junk
box hasn't got ideal bits and pieces hence its name. The taller
transformer seems to have three 24 volt windings plus a winding
for 142 volts, the other just a standard mains radio transformer.
I'll need to test the 24 volt windings to measure their power
output so that the 55 volt requirement of over 1A can be achieved
without overheating of windings. |
|
Above is the circuit diagram
for the mains wiring of the new power supply which removes mains
from the umbilical cable and allows testing, and below the progress
so far (most of the AC wiring). The standby transformer has a
primary winding of about 2Kohm and produces 18VAC off load and
is ideal for a relay with a 24VDC coil. The small LED works from
the standby transformer secondary using a 5.6Kohm resistor to
set the current at about 3 to 4mA. The 6.3VAC (yellow) wires
will be anchored later when I'm wiring the 34 termination tags
for the umbilical cable. I fitted the PSU control relay between
the two large transformers.. |
|
|
This is the terminal board (whose
position is shown in the picture above) to which the umbilical
cable to the main receiver will be wired. One option is to terminate
this in a chassis mounted socket but it will be more flexible
to use a mating cable of about twelve inches that will allow
the power supply to be mounted away from the receiver in case
work on the latter needs to be done whilst powered up. I found
I'd missed out "V" so used the pin next to "B". |
|
As work on the PSU progressed
I discovered one of the windings on the LT transformer was linked
to the mains primary winding so was unsuitable for providing
the 150V supply. I also found the HT was a bit high at 325V.
Initially I added a bleed resistor and this together with a small
LF choke dropped the HT slightly. I then added a 1Kohm resistor
in series with the LF choke bring the HT down to 300V, still
a bit high. I then decided to solve both this and the missing
150V by adding an OA2 stabiliser fed with a 7.5Kohm feed resistor.
This worked and brought the HT down to 250V. Once the receiver
draws its HT current this should drop further to the correct
voltage. The downside of all this was having to fit a B7G valve
holder. I'd also fitted the HT choke and smoothing condenser
on the upper surface of the chassis. I need to fit a pair of
coax leads terminated with BNC plugs for the IF and BFO inputs. |
Scant information and
lack of familiarity is a problem with this CJD receiver, as it
is with many projects concerned with working on an old piece
of electronic equipment. The CJD receiver designers had limited
semiconductor devices at their disposal so developing a modern
substitute for the missing parts is interesting. I'm looking
at the power supply for the oven and reminded of an
exercise many years ago concerned with regulating something
like 80 volts for a teleprinter circuit. In those far off days
the designer had seen that a huge stud-mounted 80 volt zener
was available and had used this to stabilise the supply. Alas,
although the voltage was nicely stabilised when the teleprinter
was running and drawing power, the zener diode was expected to
handle much the same amount of power when the teleprinter wasn't
running. The engineer hadn't read the small print in the zener
diode spec about "safe operating area" so things went
from bad to worse once a batch of equipments had been delivered.
Of course the blame was placed on a "bad batch" of
zener diodes and all ended well but it was silly for a programmer
to be appointed to run the project! Make a programming mistake
and it's dead easy to just issue a patch but a hardware design
error impacts on printed circuit layout and component changes.
Looking at the slightly mystifying
-55V section of the PSU (above) it would seem that the designer
(or most likely the designer's boss?) had noticed a heating problem
and added a relay (RL2A) to circumvent this. It's worth mentioning
that even today, it's not easy to find a suitable linear stabiliser
to develop minus 55 volts, so is it possible to simplify matters?
The circuit diagram refers to a series stabiliser which I first
assumed was a large PNP transistor, but I spotted a drawing showing
a "TR1" mounted on a heatsink with its collector grounded..
meaning it's probably an NPN device. This would need to handle
around 1.5A and with a bridge rectifier output of say 85 volts
the transistor dissipation would be quite high.. not as high
as it first seems (45W) because circuit resistances come into
play and at full current draw the dissipation could be as low
as say 20W... but with a germanium transistor this is still pretty
high.
The problem here is the designer
needed to use worst case parameters such as the maximum transformer
output and stabiliser circuit parameters once the full current
demand had dropped hence the addition of the relay which is apparently
used to reduce the stabiliser dissipation once the full current
has decreased. Nowadays the choice of stabiliser is only slightly
easier. A silicon power transistor (2N3055) or even an IGBT with
a suitable heatsink could be used, but before I can complete
the design I need to see how my particular transformer/bridge
rectifier works with the oven. Below are the results of testing
the 55 volt PSU. Clearly lots of power is available from the
bridge rectifier as we need only about 1.13A at 55V.. Using the
figures given below the source resistance of the transformer
and rectifier is around 14 or 15 ohms. |
LOAD |
OPEN |
21 OHMS |
27 OHMS |
47 OHMS |
VOLTAGE |
88 VOLTS |
51.4 VOLTS |
58 VOLTS |
64 VOLTS |
CURRENT |
0A |
2.45A |
2.13A |
1.36A |
POWER |
0W |
126W |
124W |
87W |
|
This might be the circuit
of the original oven power supply?
In order to reduce the zener
dissipation once the oven has reached its correct temperature
the extra 1.2Kohm resistor is switched into circuit. Once the
oven turns on current through TR2 rises and TR2 demands more
base current. The voltage from the bridge rectifier supplying
Vin drops, transistor TR1 then turns off, relay RLA deactivates
and switch RLA1 shorts out the 1.2Kohm resistor.
With a junk box transformer
the voltages will be different to those from the original transformer
and component values will be different.
A bench test will allow me to
work out new values.
Transistors will be selected
for ease of mechanical mounting so TR2 can be something like
a 2N3055 or BUT11 with TR1 2N5415 or BD244C.
Another drawing I spotted shows
two multi-turn pots meaning worst-case calcs were not essential. |
|
|
Here are some rough and ready
calculations if the general form of the above circuit is realised
(max oven current = 1.3A) Ref. pictures below whose voltages
are used as a general guide. After testing I'll revise the relay
switching zener value to suit the bridge off-load voltage. The
initial figure of 88 volts compared with voltages on-load reflects
the internal resistance of the transformer secondaries plus that
of the bridge rectifier together with the performance of the
reservoir capacitor.
At full load the transistor
will dissipate something like 1.3A x (68V-55V) = 17W.
The combined zener and base
current will be (68V-55V)/330 = 39mA. Assume the zener draws
5mA leaving 34mA for TR2 base current. Zener dissipation being
55 x 5mA = 0.3W
With a gain of say 38, TR2 emitter
will manage 38 x 34mA = 1.3A. When the oven switches off the
PSU voltages rises to its off-load level of 88V and TR1 turns
on bringing the 1.5K resistor into circuit reducing the current
to (88-55)/1.83K= 18mA. The zener will sink most of this.. say
16mA and dissipate about 1W.
If the 1.5K resistor were not
to be brought into the circuit the total current would be (88V-55V)/330
= 100mA causing the 55V zener to dissipate 5.5W. |
|
All these figures are close enough
to try the layout above and make some measurements. Note the
relay is slugged by the 47uF capacitor which will give a measure
of hysteresis to its operation. In practice I'd use a combination
of zener diodes in series to improve their power handling. One
unknown is the gain of TR2 which is possibly the reason the designers
used a couple of potentiometers in their implementation. I selected
a 2N3055 and a 2N5415 for TR2 and TR1 respectively and I'll using
a 680uF reservoir and a 33uF smoothing capacitor.
Below you can see a little more
progress as the oven supply is being fitted top right on the
second picture. It seems there's less and less space becoming
available for the electronics section! |
|
|
Oven supply test results
LOAD |
OPEN |
OPEN** |
150 OHMS |
100 OHMS |
50 OHMS |
INPUT VOLTS |
88 |
83 |
68 |
68 |
68 |
INPUT CURRENT |
15mA |
30mA |
420mA |
640mA |
1.14A |
INPUT POWER |
1.3W |
2.5W |
28.6W |
43.5W |
77.5W |
OUTPUT VOLTS |
58 |
59 |
57 |
57 |
57 |
OUTPUT CURRENT |
0mA |
0mA |
380mA |
570mA |
1.14A |
OUTPUT POWER |
0W |
0W |
21.7W |
32.5W |
65W |
|
With the circuit wired
up I connected a variable DC power supply across the bridge rectifier
and cranked it up until RLA switched off. This happened at about
87 volts. I found that without its capacitor the relay tended
to oscillate at the switching point. Using different load resistors
I found the results listed above. I used a 24V zener instead
of the original design's 27V and a 24V + 33V for the 55V zener
hence the elevated output voltage. I'll substitute a 39V + 15V
for these zeners later to reduce the power dissipations. I used
a 48V relay with a 4.8K coil in series with 1.2K. Once the testing
is moved to the chassis supply I'll check the off load voltage
to see if the 24V zener needs changing. With the 57V output the
relay switching point is a couple of volts adrift also.
** I guess heating of the various
components resulted in the changes particularly the open circuit
voltage. This effect should be reduced once the 43 volt supply
is wired up.
Besides the oven supply the
minus 55V rail also supplies the minus 43V for the AGC and RF
level indication circuitry. This latter area is due for redesign
once I start the analysis using the LM373H so its requirement
of up to about 280mA is presently fluid. The 43 volt supply test
figure shows a 150 ohm load across the 43 to 55 volt supplies
rather than 43V to ground which is a bit odd. Given a 43V zener
and a 360 ohm feed resistor to the 55V supply the zener off-load
current is 12/360= 33mA so the current drain might allow for
5mA zener current leaving only 28mA max load so I'll use a 330
ohm feed resistor to the zener and test using a load of say 1.2Kohm. |
Testing the 55 volt power
supply with a bench DC supply which worked but then with the
PSU transformer proved the difference in voltages between open
circuit and loading to 1A was insufficent to guarantee the relay
switching (because my circuit uses a lower AC voltage with lower
transformer resistances) so I decided to try the obvious solution
which surely the original designers should have adopted rather
than the relay circuit with its adjustments namely using a Darlington
arrangement. I fitted a small NPN 2SC2236 transistor into the
2N3055 base circuit. A quick test showed it worked fine with
the output dropping from 54.3 to 54.2 volts with a 50 ohm load
drawing about 1.06A. The zener voltage remained constant using
a 1.5Kohm feed resistor and inserting a 330 ohm resistor into
the driver emitter so I could check its current, gave 4.5V representing
13.6mA zener current with the 50 ohm load. This represents a
gain for the 2N3055 of about 1046/14=77. The off-load drain through
the zener(s) is 10mA so its dissipation is only 550mW. |
|
|
The next stage was tidying
up the new circuit and testing it further with the correct transformer/rectifier.
Off load the rectifier produced 75V dropping to 64V with a 50
ohm load at the -55V output. The output voltage was 54.4V off
load and 53.9V on load. The 2N3055 was dissipating about 10W
with the 50 ohm load.
I used a bridge rectifier for
the 24V output and with 2200uF the winding produced 27VDC. The
normal load for supplies derived from this amounts to 550mA.
If I understand correctly the only stabilised output is 20V at
150mA. I'm unsure of the various 24V-based rails as they appear
to carry different voltages in the connector cable/test set compared
with the power supply markings. As the raw DC output isn't too
far from 24V I guess some feeds can be taken to this rather than
the stabilised level, but anyway I fitted a 24V stabilised output
and I fitted a series resistor of 27ohm as per the original circuit
for the 22V output.
To follow the original design
I need to figure out how to easily provide the Receiver Ready
signal which is used to light the front panel lamp. This is done
in the 55V area where a rise in the off-load voltage is detected
once the oven has warmed up and turned off. The voltage across
the series regulator transistor could be used as this goes from
20V off load to 9V on full load. Something like a small relay
with a 5V coil fed via a 10V zener diode might work, however
when I looked at detailed relay specs I realised that once triggered
a typical relay would not unreliably unlatch whilst the coil
voltage was at all positive hence the reason for the designers
use of a PNP transistor and zener diode. It seems the primary
use of their relay was to provide signal to the Ready Lamp and
the power saving was just a bonus. In fact reducing the main
zener current aided the voltage increase and hence made the relay
more reliable. My problem is the overall resistance of the 55
volt supply being much lower than theirs reduces the voltage
swing between full load and no load making a more sensitive circuit
necessary. Why couldn't the designers use the oven control circuitry
to light the Rx Ready Lamp? I guess the additional zener resistance
was important to them after all because of their simple voltage
stabiliser drawing too much power in standby.
After experimenting I discovered
a typical relay indeed failed to turn off as the 2N3055 VCE dropped
from 20V to 10V as sufficient holding current was always present.
This meant that from say 24V to 5V the relay stayed on, but adding
a small zener diode in series with the coil resulted in only
a very small change being necessary for the relay to drop out.
For the moment I'm using a SPCO relay with a 24V coil at 2.4Kohm
and a 6.8V zener diode. The relay kicks in at 16V and drops out
as soon as a 50 ohm load is applied to the minus 55V output.
I'm using a red LED on the front of the chassis to indicate the
oven is on with the NC contact used as shown above. A ground
at the Rx Ready wire should light the red lamp on the receiver
front panel. Once everything is connected it might be necessary
to change the 6.8V zener to say 4.7V because residual 55 and
43 volt standby current may reduce the measured off-load voltage
from 75V to a bit less thus failing to turn the relay on. |
|
Above is a view of the
new 34-way umbilical cable to be used between the power supply
and the main receiver chassis. I had enough different coloured
stranded wires to make the job easier than using a single colour
used in typical equipments of this vintage. The key problem with
constructing this equipment is preventing accidental damage from
wiring errors because the wiring carries a mixture of valve power
supplies including HT, low voltage rails for transistors, general
control wires and even mains wiring (although I've in fact designed
out the need for mains wiring for safety reasons). Apart from
stranded single wires two signal wires for use in the volume
control area are carried by miniature screened cable. Wires handling
HT have better insulation and those carrying a degree of current
are heavier. At this stage as you can see the soldered connections
are awaiting insulation.
Below is a picture showing current
progress (usually termed a "rats nest" method). Note
centre-top a bleed resistor for discharging the HT which I'd
inadvertently found twice to be present several days after bench
testing. The cable above will be connected to the tagboard. As
yet, none of the addtional circuitry needed to handle demodulation
etc is present. My general idea is to power the main receiver
and monitor the IF output if this is possible before proceeding
with the LM373 design. That loose relay will soon be soldered
in place with a red LED (top left) to indicate oven current. |
|
For my own benefit I'm recording
the wiring detail below because my computer screen is next to
the bench where I'm doing the wiring. |
PIN |
FUNCTION |
WIRE |
B |
Mains on/off |
WHITE |
F |
6.3VAC |
BROWN |
L |
AGC long/short |
PURPLE/RED |
R |
minus 55V |
GREYISH BROWN |
V |
BFO on/off |
WHITE/BLACK |
Z |
Screen (U) |
SCREEN+BLUE |
DD |
Ground |
BLUE |
JJ |
Ground |
BLUE |
NN |
Spare |
BLACK |
|
PIN |
FUNCTION |
WIRE |
D |
Fuse F1 |
WHITE |
J |
Rx Ready |
ORANGE/BLACK |
N |
600 ohm o/p |
YELLOW/BLACK |
T |
Audio out (Y) |
SCREENED BLUE |
X |
+235V |
RED/BROWN |
BB |
minus 43V |
YELLOW |
FF |
+150V |
DARK PINK |
LL |
Ground |
BLUE |
- |
- |
- |
|
PIN |
FUNCTION |
WIRE |
A |
Mains on/off |
PURPLE |
E |
6.3VAC |
BROWN |
K |
+24V rough |
GREY |
P |
600 ohm o/p |
YELLOW/RED |
U |
Audio in (Z) |
SCREENED RED |
Y |
Screen (T) |
SCREEN+RED |
CC |
+24V smooth |
WHITE/RED1 |
HH |
Ground |
BLUE |
MM |
AGC1 |
ORANGE/RED |
|
PIN |
FUNCTION |
WIRE |
C |
Fuse F2 |
PURPLE |
H |
+24V var |
PINK |
M |
AGC on/off |
GREEN/WHITE |
S |
+20V |
ORANGE |
W |
+20V |
ORANGE |
AA |
RF level |
GREY/BLACK |
EE |
AGC3 |
WHITE/RED2 |
KK |
AGC2 |
GREEN/RED |
- |
- |
- |
|
|
|
The umbilical cable is
wired in place and connections tested. Plugging it into the receiver
and turning on the power resulted in a fairly loud 100Hz hum
from the loudspeaker because none of the RF/AF circuit is in
place. The HT was OK at the rectifiers and measured around 240Vat
the output pin which slowly dropped to 190V (presumably as the
valves warmed up and suggests either too much resistance in the
HT supply or too much HT drain in the receiver. Maybe the floating
AGC connections are the reason as these should be somewhat negative?
The 150V supply also dropped to around 135V following the HT
drop. LT voltages seemed about right. There's a few things to
test. I could remove the speaker hum by grounding the audio pin
and see if I can get any indications on the front-panel meter.
Alas the meter which should
have registered the HT for example refused to budge and after
removing it I discovered an open circuit coil. I did notice the
voltage across the rear connections rose from under a volt to
13V in one particular switch setting. Is that the problem or
is it just a false reading on my high impedance multi-meter?
I need to discover the reason for the meter failure before I
fit an alternative. |
|
Lots of the 34-way connector
pins were bent and the worst snapped off but this turned out
to be unwired and marked as "spare".
Once I'm happy with the wiring
I intend to tidy it up somewhat. |
|
|
Eventually, if all goes
well, the chassis will be mounted on a panel to be located above
the main receiver chassis much like the original equipment layout. |
pending
|
|