The CJD Receiver Part-2

My initial attempt at getting my CJD Receiver to work ran into trouble.

Below are links to the sections of the CJD receiver documentation that I have. These are incomplete but hopefully will prove adequate for my purposes (like having a jigsaw with half the parts missing). One reason for adding these is I will be able to access them on my workshop computer.

First section of the CJD manual, pages 1 to 32,	BR2407-1	Part 1 Ch1 to 4 + Part 2 Ch1
Second section of the CJD manual, pages 33 to 43	BR2407-3	Part 1 Ch5
Third section of the CJD manual, pages 44 to 97	BR2407-2	Part 2 Ch2 to 5
Fourth section of the CJD manual, pages 98 to 123	BR2407-4	Diagrams + circuits
BFO	BR2407-6	Circuit
Power supply rectifiers & regulators	BR2407-8	Circuit board
Power supply mains & transformer	BR2407-9	Mains input
Power supply sub-assembly	BR2307-10	Mains input + other parts less pcb
2nd RF/1st Mixer/2nd Mixer/Local oscillator	BR2407-11	Centre of front end simplified circuit
Synthesiser sub-assembly	BR2407-13	Block diagram
Harmonic frequency selector	BR2407-14	Part circuit

The choice of chassis for my power supply and receiver detector and audio sections proved to be too small, leading to overcrowding of circuitry and difficulties of testing etc. The new design will be made using available materials and past experience of constructing power supplies. Because a lot of the exercise may include changes then the design will need to allow flexibility for changing components and wiring modifications. For example, although my previous plan was to use an LM373 chip but there may be an easier solution using an existing commercial circuit board suitably modified to match the requirements of this receiver.

First the new chassis.... I can think of a couple of contenders. A wooden baseboard with a metal front panel or a traditional metal chassis using formed steel sheet. One of the main drivers affecting the choice is the electrical interface between the receiver and chassis which is a multi-way cable carrying power and RF and control signals pictured below. A simple solution is to provide an aperture in the PSU baseboard through which the receiver multi-way connector can be passed. Another driver is the method of bringing together the PSU and the receiver and of course the ease I can make changes and finally of course the choice of materials to hand.

The original PSU was far too cramped at 8 x 11 inches so fitting a wooden baseboard roughly 15 x 18 inches minus a cable aperture say 2 x 2 inches gives me an improvement of better than three times. 266 compared with 88 square inches.

Material 20mm chipboard.

The steel front panel will be matched in width to the CJD front panel.

Leaving an open cable aperture will allow the PSU to be moved away giving access to the top of the receiver chassis.

I might add a sheet of steel onto the baseboard.

Using available materials this base will form the foundation for the new power supply and detection circuitry.

As a test meter is already present on the receiver front panel not much if anything will need to be fitted to the front panel here.

Overall height is around 4.5 inches.

Here's the progress so far. Slightly more complicated than the original design because I've included a test feature which enables one to check the various voltages before turning on the receiver.

Using a wooden baseboard makes it a simple matter to rearrange things and add the odd terminal strip.

What I'm attempting to do in summary is to duplicate the mains power supply comprising a transformer with various voltage regulation circuits (mounted on a printed circuit board) plus a couple of other circuit boards carrying the radio detector circuitry and audio processing circuitry ("radio parts" in the sketch above). These all interface with the main radio chassis via a multi-way connector.

As most restorers will not have all the appropriate CJD documentation and those that have limited information, as I did, find it a bit confusing I'll add useful information as I discover it. For example the complete receiver uses a chassis mounted above it carrying not only the complicated power supply, but a couple of important receiver sections. These are located on two printed circuit boards with a third board carring smaller power supply components. To identify these I've added the final four digits of their NATO codes.

4933= power supply parts, 4934= the audio amplifier parts and 4935 IF amplifier/detector/AGC parts.

Heavy PSU parts are coded 5136

To see how the radio parts fit into the overall circuitry see the picture below. The section in the lower part of the drawing shows the final audio amplifier and loudspeaker plus jack socket located in the main receiver chassis whilst the areas to the left and right are located on circuit boards named "Equaliser gain panel" and "Audio amplifier panel" respectively.

Below is the Equaliser gain panel circuit

Below is the Audio amplifier panel

My plan is to replicate the functions of the two panels using an LM383 device.

My other task is to figure out why the meter is open circuit (ie find out why a high voltage was present in one position) plus to work out why the PSU got extremely hot and smoked after a couple of tries of connecting it up. Below is a table giving the receivers power requirements.

Voltage	Current	Notes
6.3VAC	1.5A	Valve heaters
20V stabilised	150mA	Transistor power
24V unsmoothed	200mA	Dial lamps
22V smooth	200mA	Relay power
minus 43V	80mA	AGC
minus 55V	1.18A	Oven power etc
150V	15mA	Valve local oscillator
235V	80mA	Valve HT

Before proceeding I'll just show you an article from 1972 regarding the LM373 which is the device I'm planning to use for the detector circuits (re-my previous page on this) Just click here to see the article published in CQ Magazine over 50 years ago.

This device was being marketed around the date of my example of the CJD receiver. I bought one on Ebay three years ago and put it away for safekeeping. Fortunately I've just found it after an hour or so.. in the first place I'd looked in a jiffy bag containing connector parts kindly supplied by Paul, G8GJA.

Here's the spec...

Repairing the meter circuit

I decided to fix the meter circuit. When I first acquired the receiver the meter was damaged so maybe that was the reason the receiver was scrapped? I found one of the ranges had a relatively high voltage at the meter leads. I managed to find a junk box replacement but first I need to fix the fault. Circuit diagrams are incomplete so I measured the resistance across the leads of each meter switch position. See below.. the green/brown lead is infrequently grounded. When current flow through a shunt is applicable grounding isn't common.

SETTING	OVEN	CAL	LOCK	RF LEVEL	AF LEVEL	SYN	IF2	IF1	RF1	HT2	HT1
OHMS	0.3	2.7M	3.6M	OPEN	24K	3.3	8.7	13.6	10.2K	2.1K	2.2K
OHMS	73K	750K	GND	OPEN	OPEN	3.7K	3.7K	3.7K	5M	GND	GND
NOTE	1R105	5R43	PLE15 PLA-J	PLA21 9PLA-AA	9PLA-N	1R106	1R103	1R104	1R107	1R109	1R110

Many readings are taken across meter shunts connected to the meter switch and, as the meter isn't protected, a fault say in the multi-way connector circuit can ruin it.

The open RF level is probably fed via the unplugged muli-way connector.

Below are details taken from the handbook.

Before long I'll need to gain access to the circuitry and on the topside you can see a couple of screening panels. Under the left panel secured by 29 screws are the circuits for the IF amplifier and on the right, secured by a mere three screws is the RF front end circuits with their 3-gang tuning condenser.

Above, a view of the receiver front end with its valves.

On the right is a view of the panel above the test meter.

From left to right, top

Resistors measuring 3.7M ??, 8.5, 1.5, 12.7 and 6.23K (R111)

and horizontal (R110) 2.18K

Bottom,

Resistors 10.2K and 2.07K, a diode then 10.09K, 15, 10.21K

and coil 0.1 ohm

Now, upended to remove covers on the underside...

Above left, the trimmers for the various tuning ranges and on the right, under the cover is a piece of fibre board insulation material.

Removing this reveals another metal panel and a worrying feature... only a single securing crew. In fact during the dismantling I'd noticed about 50% of the washers were missing and also of four large securing screws for the valved front end, none are tight. Three seem cross-threaded with one bent. I'm not the first to be investigating this receiver.

Finally (left), under the cover at the base of this module you can see three printed circuit board sockets and, on the right a view under the last cover. Click to see a larger picture.

Now I'm going to investigate the problem I'd had with the power supply overheating which, at the time, I'd reckoned it was due to a fault in the oven circuit.

One slim possibility is incorrect wiring between the PSU and the multi-way connector as the pin numbering is extremely odd as you can see here.

Measuring the resistance between each plug pin and ground revealed that many were high resistance and others much as expected.

I've listed the results in the tables below after the summary of pin allocations.

The final and most important step is to confirm the wiring is correct. All the taboard colours matched OK, but I'd unsoldered the wire from the HT connection to the plug.

Thankfully there were no errors.

The unsoldered HT lead must have been due to a leaky HT feed?

PIN	FUNCTION	NOTE	OHMS
B	Mains on/off	Mains N	HIGH
F	6.3VAC	-	0
L	AGC long/short	Long=+24V	HIGH
R	minus 55V	-	34K
V	BFO on/off	On=ground	0
Z	Screen	For Pin U	30
DD	Ground	-	0
JJ	Ground	-	HIGH
NN	Spare	-	HIGH

PIN	FUNCTION	NOTE	OHMS
D	Fuse F1	from Pin B	HIGH
J	Rx Ready	Relay ON=0	HIGH
N	600 ohm o/p	-	HIGH
T	Audio out	Audio for 600 ohm	HIGH
X	+235V	-	400K
BB	minus 43V	-	73K
FF	+150V	-	107
LL	Ground	-	0
-	-	-

PIN	FUNCTION	NOTE	OHMS
A	Mains on/off	Mains L	HIGH
E	6.3VAC	-	1.0
K	+24V rough	Lamps	30
P	600 ohm o/p	-	HIGH
U	Audio for CJD speaker	End of pot	HIGH
Y	Screen	For Pin T	HIGH
CC	+24V smooth	-	227
HH	Ground	-	0
MM	AGC1	-	HIGH

PIN	FUNCTION	NOTE	OHMS
C	Fuse F2	from Pin A	HIGH
H	+24V var	Lamp dim	HIGH
M	AGC on/off	Off=+24V	HIGH
S	+20V	-	3.6K
W	+20V	-	3.6K
AA	RF level	To meter	HIGH
EE	AGC3	AGC to RF	HIGH
KK	AGC2	-	HIGH
-	-	-	-

The next step is to power the HT line with a variable supply then, after checking for odd current, power the valve heaters and see if the extra HT current draw is sensible, something I should have done before plugging in the custom supply. In fact I can even monitor the IF output and see if that is present.

All went well initially. HT current less the valve heater current with the voltage measured at 237volts was only 0.5mA which I hope suggests the HT condensers are in good order. Adding the heater voltage of 6.3 volts AC resulted in a current draw of 9mA at 240 volts. The 150 volt draw was 1.5mA with the valve heaters warmed up and correctly drawing 1.4 Amps. Moving onto the oven circuit the current measured 1.6 Amps. Testing the 24 smooth circuit resulted in 30mA, the 24 volt rough drew 32mA and the 24 variable input drew nothing. The 20 volt circuit drew 20mA and at this point I realised I'd applied not neg 55 volts but plus 55 volts to the oven circuit. My only reference to this circuitry is a half page drawing showing the neg 55 volt input going nowhere as the other half of the circuit diagram is missing. I now need to check if I've damaged something in the oven control area.

Oddly, none of the measurements I made, either resistance or current draw, explain the overheating transformer and smoke when I tested the power supply 18 months previously.

This is the new improved design for the power supply and includes several switches which are not used in a normal operating environment for the receiver.

One reason is I do not like the idea of running AC mains through the umblical cable so the on/off switch on the front panel of the receiver carries current through a 24 volt relay coil. In my first design the PSO ON relay had a DC coil which meant that the transformer T3 needed a rectifier and capacitor.

To operate the receiver the Standby switch is turned ON followed by the CJD ON/OFF switch. The other switch (T1/T2) which I've called the Test switch is inhibited via S2. That means operating the Test switch with the Standby switch set to OFF will not turn on the three power supply transformers.

This means that the various PSU voltages can be checked (ideally)with the umbilical cable unplugged by turning on the Test switch. I haven't added an interlock to prevent the Test switch turning on the receiver PSU with the umbilical cable plugged in. This could be done by adding another relay but this adds yet more complication to the wiring.

After modifying my new power supply to use a 24AC relay (T3) and do away with the need for providing 24VDC, the only relay with a 24VAC coil I had turned out to be faulty with an open circuit coil. I failed to find a suitable alernative in my junk box so put back the rectifier and smoothing capacitor after finding a nice 12VDC relay with decent terminals. Next I tested this and it worked fine but I'd made a mistake in the mains wiring so the connections to the pair of switches (Test and Standby) need to be sorted out. Not too difficult as I'd omitted a wire from T2 to the live side of the transformers. Now the transformers powered up I checked the HT. This was unloaded and sitting at 350 volts but adding a dummy load of 3.6Kohm reduced this to 207 volts at 57mA. The reason for the steep drop was the secondary circuit resistance. The choke resistance is marked as 300 ohm and the HT secondary winding measured 230 ohms. I should probably later remove the choke and fit a suitable resistor so the output is at the level designed for the receiver.

Next I checked the 24 volt circuit. This is using a different transformer than my first circuit and it measured 24VAC. Switching on with the 24 volt and 20 volt regulators connected at first resulted in a 24 volt feed to the regulators however I hadn't yet fitted a reservoir capacitor. I fitted a 2200uF capacitor and switched on. There was a loud crack and the 7824 chip exploded. This was slightly puzzling as the transformer output measured 24VAC RMS meaning the maximum smoothed DC should be something like 35 volts and the 7824 should be good for 40 volts so why did it explode? The junk box surprisingly failed to yield a replacement so...

I decided to change the overall design of the oven and 24 volt supplies. The new oven transformer having twin 43 volt windings makes it easier to use full-wave rectification which will result in a peak voltage of about 60 volts dropping under load. Previously I'd used a 2N3055 series pass regulator (with a transformer having too high an output for the job) and this I'll now use to regulate the 24 volt outputs.

Despite my careful planning things didn't turn out as I'd expected. Loading the neg 55 volt supply with a handy 27 ohms reduced the rectified terminal voltage to 43 volts and it was apparent that a 2N3055 regulator will be required. A simple change might be to add a spare AC winding to the 43 VAC transformer rather than wastefully use the pair of 43 volt windings in series (=86 volts).

A suitable spare is a 5 volt winding at the HT transformer. This would increase the output, at first sight, to 48 volts but still not enough so using 86 volts may be the best bet as the transformer should be able to supply the desired current. Wasted power will be (86V-55V) x 1.2A=37W. The other option is to locate a different transformer. I wonder if a suitable filament lamp in series with the AC winding would reduce the semiconductor losses?

VOLTAGE
235
150
-55
-43
24S
24R
24V
20
6.3

TEST LOAD OHMS
4700
10000
47
150*
100
220
220
130
4.7

CURRENT mA
50
15
1180
80*
240 110
110
154
1340

* means the load/current is via the neg 55 volt supply

The table above is constructed from the McMichael Radio Test Set parameters shown here.

I gave up on the twin 43 volt transformer. Despite trying various schemes nothing provided me with a sensible 55 volt output at full load so back to the junk box. I collected several potential candidates and chose one which supplied a 20-0-20 volts plus 15 volts with thick enough copper to provide a couple of amps. The fun started when I came to its mains tapping selector which gave options of 100/120/220 and 240 volt inputs. I cut off the selector and measured the winding resistances. By trial and error I managed to put the 100 and 120 volt windings in series the wrong way round. The incorrect phasing gave me 240 volts DC output at my test rectifier bridge together with a loud hum and some smoke. Clearly the bad phasing was giving me a step up from the expected 55 (x root two) volts to 240 volts. I switched round one of the primary windings and tried again. The output voltage at the capacitor had fallen slightly and measured about 180 volts instead of 55 x root two volts. After puzzling and trying again the penny dropped. The overcharged capacitor was stuck at 180 volts (cutting off the rectifier diodes) and not discharging. I discharged the capacitor and tried again. The voltage was now a slightly rising 74 volts. This relates to (40+15) times root two minus a few volts at the rectifier. My dummy load of 27 ohms got exceedingly hot with the terminal voltage at a little over 58 volts... (= 2.1 amps) perfect for the 2N3055 series pass regulator.. In fact the maximum load will be 47 ohms so plenty of power will be available with any surplus dissipated in the regulator.

Next... is the 24 volt supply man enough? From the test results above I need no more than 700mA and tests with a 27 ohm dummy load (=880mA) showed the outputs were just about acceptable.

I've copied the potentially useful information below from another page on my website.

For interest I weighed the twin 43 volt and its 20-0-20 volt +15 volt replacement. The first was 1400 grams and the second 1930 grams. The table gives their power ratings as about 56VA and 101VA respectively. My tests drew circa 43V x 43V over 28.5 ohms = 65VA and 58V x 58V over 28.5 ohms = 118VA respectively. These figures line up pretty well with the table above once I'd measured the load plus its leads.

I screwed down the new transformer and connected up the remains of the regulator taken from the (too small) chassis. This comprised the 2N3055 transistor fitted to a heatsink plus a 1Kohm resistor and a pair of zener diodes rated at 39 volts and 15 volts to give me 54 volts. The transistor tested OK when I checked it with my multimeter but applying the 73 volts from the new transformer resulted in smoke and the pair of zeners going short-circuit. I was puzzled once again. First the 24 volt regulator failing for no reason and now my series pass regulator failing as well. Were the zeners being over-run? The current should be restricted to (73-54) volts over 1000 ohms = 19mA. The zeners would dissipate 15 volts x 19mA = 285mW and 39volts x 19mA = 741mW and both were rated at 5W. I checked the transistor again but realised the 1000 ohm resistor from base to collector was giving me a false reading and sure enough, removing this showed the 2N3055 base connection was open circuit. I fitted a new transistor and a pair of 27 volt zeners and the regulator worked perfectly supplying 54 volts. All I can think is I'd inadvertently connected the zeners back-to-front. This would result in a current of 73 volts over 1000 ohms = 73mA. But that would be 2.8W dissipation at the 39 volt zener and surely not enough to render the pair of zeners short-circuit? I'm still puzzled.... but at least I now have the oven supply. I just need to add the extra zener to the output to produce my 43 volt supply. I'm not going to bother reducing the heat losses under no-load conditions as in the original circuit.

Below, the original simplified circuit of the 55 volt oven supply and then with the Darlington transistor.

Testing the new circuit under load I realised the original designers had indeed needed to overcome the problem of the oven voltage load. As the load current increases the 2N3055 base current increases thus reducing the zener voltage to a point where the load voltage drops below the required 55 volts. In fact there's a balance needed to be achieved taking worst case figures for zener dissipation and load current. In fact this was very similar to the first exercise I'd had to deal with when I'd started in industry although dealing with germanium transistors. The design associated with turning solid state switches on and off needed a lot of manipulation of my slide rule. A simple solution is to use a Darlington transistor which has a larger current gain than a 2N3055 (gain = 20 to 70). I selected a BD651 (gain = 750) and this worked fine with the zener supply voltage constant under load and the output remaining relatively steady with a dummy load of 47 ohms.

Below... progress so far. Note the array of LEDs on the front panel. One for each of the supply voltages plus test!

The next step is to protect the various AC mains connections with plastic covers before these are accidentally touched. The wooden baseboard lends itself to making additions such as this as well as moving things around to make best use of space. Note at the bottom of the picture my ceramic tipped heatproof tweezers which are perfect for holding wires as they're soldered.

I wired up a set of LEDs along the front panel. There are quite a lot of power supply voltages required to run the receiver and I needed to fit no less than nine LEDs for these plus one to indicate a TEST mode. The receiver has its own on/off switch which was designed to turn on and off the vessel's mains supply which I rejected this as being a little unsafe. Rather than running 240 volts AC through the umbilical cable I'd fitted a standby transformer which swapped the 240 volt connections for a low voltage supply used to turn on a 12-volt relay. The relay in turn switches on all the mains transformers in the power supply. The TEST switch turns the low voltage transformer on or off. When in the OFF setting it readies the mains on/off switch on the receiver front panel. TEST ON opens the low voltage transformer mains input and TEST OFF turns on the low voltage transformer. TEST OFF illuminates the blue LED on the front panel.

At this point I connected together the ground connections of the various supply voltages.

Now that the power supply is basically ready I need to fully integrate the new design for the detector circuitry.

For example I'll need to drill the PSU front panel for some extra controls.

In the original published design on which I plan to base my own there are several controls viz. a switch for selecting AM or CW/SSB, master IF and audio gain controls, an auto/manual AGC switch. and a BFO tuning control. Looking at the picture above you can see several similar controls, and its worth pointing out that the PSU box for the CJD receiver has no front panel user adjustable controls. However, the CJD receiver has a number of controls whose connections are carried via the umbilical cable. To make the new design fully testable though I might need to add a TEST feature which allows these remote connections to be duplicated. I'll also need to add a regulator or two to provide working voltages for the new IF circuitry.

Below is an overview of the LM373. Click to see the full spec.

Above.. BFO ON/OFF is carried by Pin V with ground = ON. This implies that without a ground connection (BFO=OFF) AM is received. In the CJD detector this signal (ground) selects either the product detector or the AM detector via a pair of 24 volt relays. This method will be used in the redeigned detector but will require a TEST switch to operate the relays.See drawing below

Above the CJD 600 ohm broadcast level and the AGC selector switch used to operate a pair of relays in the external detector circuitry in the PSU box.

Control signals and audio in and out of the detector circuits in the PSU box are carried via the 34-way umbilical cable but RF is carried via coax cables. These latter are the IF output from the CJD receiver plus the BFO signal.

The BFO RF connection shown here on the right is carried from the CJD receiver.

The IF signal here however comes from the IF/AGC amplifier circuit shown below.

A detailed diagram of this area is shown below.

Above, switching is again carried out via relays operated from CJD front panel knobs via the umbilical cable.

Below is the way the receiver audio is controlled. There are two circuits provided. One is direct to the CJD internal loudspeaker and the other for on-board ship broadcasting.

First pass at the LM373 circuit for the CJD receiver. Note that S1 and S2 are changeover relays driven by the CJD receiver controls via the umbilical cable.

You'll note the component marked "Filter" which is used to define the bandpass of the detector. In the published circuit this is a CFS455J which has an input/output impedance of 2000 ohms.

As the incoming IF signal is already well defined by the main receiver as shown below the LM373 filter probably isn't too important, however I did check on one of the Net calculators to see how easy it might be to add a half decent filter..

On the right is a 3rd Order filter and then a 4th Order more complex design.

Below is the shape using a single coil and capacitor which is my preferred option for initial testing, because I can readily find the parts.

Building the RF detector will be quite an easy task as I intend to use my favourite method of construction. This needs a flat piece of of tinplate onto which will be soldered the parts (the LM373, 12 volt regulator and audio amplifier together with suitable relays). Before building an experimental circuit I looked for and found a more comprehensive datasheet. Click to see this.

Below, I've shown the two layouts (AM plus CW/SSB)

Looking at the published AM circuit from K4DHC it seems he used the recommended component values as depicted here and omitting the optional AGC theshold adjustment.

Again, looking at the published CW/SSB circuit from K4DHC it seems he also used the recommended component values as depicted here and omitting the optional local oscillator and signal nulling adjustments.

Missed is the 50nF tone correction capacitor at the audio output which might well be omitted to boost HF audio components for older constructors.

I need to interface the circuit more precisely to the CJD BFO output by adding an attenuator reducing the input from 700mV to around 25mV.

I might have to also deal with the IF signal amplitude from the CJD which is here shown as being terminated by 50 ohms.

One aspect I've glossed over is the design of the AGC system within the CJD receiver. The LM373 has an integral AGC circuit whose operation is predefined. In order to send an AGC signal back to the CJD circuits I'll need to emulate the circuitry in the original design. This needs to be able to handle the LONG/SHORT and ON/OFF signals at the umbilical cable and generate the AGC 1/2/3 outputs fed back. A relay is fitted for AUTO/MANUAL AGC. Another relay is probably going to be needed for handling LONG/SHORT.

Looking again at the AGC section of the LM373 and the practical aspects of the CJD receiver there's a problem. Traditionally in a communications receiver AGC is produced by rectifying the RF level at the tail end of the IF amplifier and indeed the CJD circuitry housed in the PSU does this. The LM373 however has a different method which relies not on the RF level, but the audio output. From my experience of signals in the VLF range audio may not be a suitable parameter on which to base AGC.

Looking at the drawings above, Pin 1 is the point at which the gain of the LM373 is governed. Its done by applying a filtered average voltage proportional to the audio output, although as you can see, in both circuits the AGC can be modified either by switching in a pot to apply a DC voltage to Pin 1 for CW/SSB or adding a pot to modify the audio-derived voltage from (almost) full to completely off.

One way to handle the CJD requirement is to merely add a standard diode detector to rectify the 21.5KHz IF signal and use this to operate on Pin 1, maybe adding the facility to select either this or audio-derived AGC. In fact, it's then a straightforward matter to derive the required CJD inputs of RF Level (AA) and AGC 1/2/3 (MM/KK/EE).

The drawing below shows the IF stages and the AGC path. A potential source for the RF signal for the above solution is either Pin 9 or Pin 4 and would avoid adding a separate amplification stage sharing Pin 2.

See the drawing showing the section of the original CJD circuitry below the LM373 picture.

This drawing is a bit blurry but Diode D1 is the classical envelope RF detector (driven from the LM373 and the required signals mentioned above are shown.

A relay, driven from Pin L, is used to select Long/Short AGC with Long set with its 24 volt coil active.

A second relay is used to turn off the AGC at VT8 with its active state being OFF

I can imagine the end result will need some experimentation to get it working as the original designers' intended.

Note that this circuit uses the neg 43 volt supply so that the AGC signals are negative with respect to ground. The line marked "AGC Detector" is merely the +20 volt DC supply.

Below, almost completed design

COMPONENTS LISTS

R1	3.9K	R7	1K	R13	6.2K	R19	33K
R2	1K	R8	47K	R14	220	R20	12K
R3	20K	R9	47K	R15	15K	R21	3.3K
R4	10K	R10	1K	R16	150K	R22	15K
R5	5.1K	R11	470K	R17	1M
R6	27K	R12	3.3K	R18	2.2M

C1	10uF	C7	100nF	C13	1uF	VR1	10K
C2	100nF	C8	10nF	C14	100nF	VR2	5K
C3	25uF	C9	100nF	C15	100nF	VR3	100K
C4	10uF	C10	30nF	C16	100nF
C5	1uF	C11	10nF	C17	100nF
C6	10nF	C12	22uF

Now that I had enough detail to start building the above circuit I constructed what you see below. The 600 ohm amplifier isn't needed (except for completeness of the overall system) so that can be left until later. The audio output level from the LM373 is defined as 120mV so I might need to deal with this once testing gets underway as I'm unsure of the CJD requirement. Also, I might need a 21.5KHz tuned circuit at the input to the LM373 and perhaps a better bandpass fillter for best operation.

I counted a total of 22 fixed resistors, 3 pots and 17 capacitors plus maybe a few more as the style of construction uses capacitors as component anchor points (eg. C14-17). In addition I need to construct the 21.5KHz bandpass filter.

Below... the detector assembly almost complete. Power requirements are neg 43 volts and 20 volts with an on-board 12 volt regulator for the LM373 device. The coil, using a core from a scrap flat screen TV pcb, is rewound to give 1.8mH on my latest Chinese component meter, and the 33nF tuning capacitor marked selected for 30nF so I can add extra capacitance to centre the response. I used 2.2K resistors to provide a similar impedance to the filter specified in published circuits.

The picture below that of the detector shows the PSU with the detector roughly in position.

Above... general view of the PSU not far from complete and waiting for the detector assembly

I suppose the next step is to test the detector circuit to prove its operation. I plan to do this with a pair of bench power supplies and my spectrum analyser to confirm the thing is resonant at 21.5KHz then a pair of signal generators to check AM and CW/SSB performance. In fact that idea wasn't exactly straightforward because many signal generators do not go down to 23.5KHz. Fortunately, once I'd checked my TinySA did this comfortably and allowed a sensible modulation frequency. Looking at its spec however revealed it was supposed to go down to 0.1MHz so I'll need to confirm the indicated frequency really does get generated!!

Panic stations

I connected up a power supply and an oscilloscope and my TinySA to no avail. Firstly the connection to the TinySA was intermittent so I swapped the BNC cable.. still intermittent and I suspect its a solder joint in the TinySC. However the main problem was nothing at all from the LM373. I modified the filter by inserting a couple of 100nF series capacitors as the filter was grounding pins 4 and 9. Still nothing from the LM373.. was it a bad one? Maybe my relay was wrongly wired? But no, a check revealed all was well. Before abandoning the LM373 I looked at the K4DHC design and spotted a drawing showing the lead opposite the tab on the 10-pin chip was marked "1" but the device spec appeared to show it was "10". After a few more searches I realised I'd copied the published error.. oops. Have I wrecked the LM373? I rewired it and powered it up and it now started producing signals but of course one of the functions may now be damaged. In fact my scope did show odd results but connecting headphones to C4 gave me a tone correcponding to the modulation frequency. The proof will be to inject a BFO signal and see if that produces a correct CW output. I'll be using my reliable Black Star audio signal generator as this has a 10:1 vernier tuning control, works from 10Hz to 100KHz nicely covering the required 21.5KHz.

I've corrected the latest drawings viz the LM373 connections.

Now to fix the TinySA. I opened it up and pulled off a metal concealing the soldering to the output socket. It was fine so put it back together and tried the lead. The short length seemed to have a break which responded to pressure so I'll ditch it and use a BNC lead with an appropriate adaptor.

Success.. the LM373 seems to be working properly except the output needed a 100nF capacitor to ground to remove RF feedthrough but once in place I could vary the signal generator around zero beat. I'm not sure about the operation of the basic 21.5KHz bandpass filter with my simple test configuration so might look at this later.

The various figures were noted as follows. Maximum output at 21.5KHz input resulted from -12dBm (circa 50mV RMS) across the 10K pot turned to max. The BFO input was about 1 volt RMS and the audio output was enough for headphone listening. The spec tells me this should be from 50 to 120mV RMS. I've yet to check the AGC characteristics and the AGC circuitry as I'll need a second bench PSU for the neg 43 volt supply.

Unfortunately it seems the LM373 is badly compromised as when I selected Manual Gain via the relay I noticed the 12 volt supply drew up to an amp, the regulator became exceedingly hot as did the LM373 itself. Before abandoning this device and using the original circuits I'd like to see exactly what's going on.

Very odd! There was a tiny solder whisker across the supply decoupling capacitor that came and went. I removed this and the excessive current disappeared and I found AM and CW could be received. I'll need to move the next stage of experimentation to my workshop so I can carry out better testing and prove or disprove the LM373 performance.

One of the problems of testing is the low IF of 21.5KHz which is well below the coverage of most RF signal generators. I have several options. The TinySA does produce low frequency signals but its outside its full spec. My Marconi TF2008 works down, unbelievably to zero frequency but at 21.5KHz its ease of tuning isn't ideal. I also used an audio signal generator but as a BFO only because one can't modulate it. During testing I found my Rigol DSA815TG worked well at 21.5KHz.

I looked on Ebay for a suitable signal generator and found nothing to fit the bill so I looked on my own website in case I'd actually got something suitable and was very surprised to find my Wavetek 2407 went down to 10KHz. I checked its spec in case I'd made a mistake, but no it went down to 10KHz so it will be ideal because it's extremely stable.. so cross fingers it works properly as it has a history of going wrong.

Next, I'll look again at the LM373 filter because when I tested this it had a very poor response. Whether this is due to a matching problem, I'm not sure. I'm using 330nF input and output coupling capacitors and these are miles different in terms of impedance to the 2Kohm I was aiming for at 21.5KHz. I think around 3.3nF might be better.

Sure enough, by replacing the 330nF (=22 ohms) input capacitor with 1nF (7Kohm) I managed to get a filter response which produced a half-decent shape.

I fitted the filter to the circuit and it retained its shape but became much less responsive. Note that the coil is 1.8mH.

Fig A

Fig B

Above at Fig A... the response of the filter on the bench, detached from the circuit.

Above right Fig B... the filter inserted into the circuit.

Right Fig C... a typical 21.5KHz signal injected into the input along with the tracking generator The true amplitude is only relative because I'm using my home made probe to avoid loading the LM373, but has a loss at this frequency of around 10dB

See the pictures later on down the page.

I'm monitoring the signal at the relativly high impedance Pin 9 of the LM373.

Output was via the grey 330nF capacitor but this is now 1nF, and input from the tracking generator via the green 2.2nF capacitor. The coil measured 1.81mH and the black tuning capacitor 3nF. Result in Fig A above.

Below.. testing the simple filter. The "T7" tester is used to measure the inductor and capacitors.

Fig C

When I looked at Fig A and C above I realised that in CW mode the filter would be best centred slightly lower in frequency as the BFO range of +/- 3KHz is quite significant in the VLF range so I added by trial and error (with the Wavetek set to 18.5 and 24.5KHz) an extra 1nF capacitor across the tuning capacitor shifting the response leftwards. Using AM I could hear a tone at the audio output but setting the AGC relay didn't have much effect.

This section is to prove my probe is suitable for making low frequency measurements because, as I see it, the receiver needs to provide relatively very wide bandwidths at the IF of 21.5KHz. At 465KHz reception of a broadcast signal would require a minimum of say 6KHz bandwidth or 1.3%. At 21.5KHz I need to ensure 3KHz bandwidth or 14%.

It's been some time since I tested my probe so decided to check it.

Centre frequency set to 21.5KHz with each horizontal division 1KHz.

Tracking generator set to -20dBm

Right with the probe turned off.

Below the probe turned on. Loss = 10.38dB

Below right with its external input series 390kohm resistor shorted out, loss = 9.11dB.

The 390Kohm resistor fitted for HT protection accounts for only 1.27dB.