Eddystone 358X Electrical Testing

See the mechanical tasks

 It's not an easy task to test a receiver when very little is known about it. This 358X is in this category because it has been completely rebuilt and bears only a passing acknowledgement to its Eddystone circuit (given below).

None of the original octal valves are fitted having been replaced with all-glass miniature types. At least a couple of front panel controls have been stripped out and alternatives wired in their places. A mains power supply is fitted which makes things a bit easier, but I did notice the mains plug was pretty old in not having its plastic parts on the live and neutral pins. The 13A plug was so tarnished I found no continuity to the mains transformer primary, but I cleaned the pins and found the transformer primary was OK and there was a mains switch at the audio gain control.

Lots of parts from vintage component days have been used in the "amateur enthusiast improvements" and I expect many will now be in poor condition.

I chose to tackle the testing by first ensuring there were no mains wiring problems and the HT line had a respectable resistance to ground.

 

 Click to see the circuit full size

Component values below

 

 

 

 No shorts and the mains circuit seemed safe so I plugged it in and waited.. the valve heaters glowed and nothing was heard in the loudspeaker I'd plugged into one of the front jack sockets so I waited... the HT measured about 288 volts.

After a few minutes there was a hot resistor smell accompanied by wisps of smoke curling up from around the audio valve area. I turned off the set and upended it to see the components. Nothing seemed amiss and nothing was warm except a metal can held in place by two screws. I detached the can to reveal the wiring for V5 (below).

 

 

Fig 1

 

 Fig 2 shows the circuitry for V5 whose markings are erased but I assume is an EBC80, double-diode-triode.

That waxed condenser was fairly warm as was the yellow condenser, whilst the smoke had probably arisen from that discoloured 4K resistor. The latter condenser 4uFx200V had nearly 300V across it which hadn't helped and that 2.2K measured circa 60K.

Disconnecting the bad condensers didn't help with the lack of audio except I found that touching the anode of V7 with an insulated screwdriver gave a crackle in headphones connected to a front panel jack socket.

 

Fig 2

 Next I'll check the voltages at the 6BW6 (2=g1, 3=k, 4,5=h, 7=a, 8=g2, 9=g3) and find where its control grid connects.

Then look at the EBC80 (1=at, 2=g1, 3=k, 4,5=h, 6=d1, 7=scrn, 8=d2).

The 6BW6 screen grid is connected to a distant large metal decoupling condenser negative terminal (via the yellow/brown wire... Fig 1) and then to a resistor. That same decoupling condenser positive terminal goes to the reservoir condenser positive. Very odd. The 6BW6 cathode goes to a pair of 180 ohm resistors in series to ground and decoupled by the yellow condenser adjacent to the output transformer. The cathode voltage measured zero volts meaning that either the 6BW6 is cut off or the yellow condenser is short circuit or both. I checked and that condenser is certainly bad.

 

Fig 3

I'd noticed that the two jack sockets are wired from different sources and I'd been plugged into the one wired to the output transformer. I swapped the headphones to the other and heard a comforting hissy crackle which crackled more as the volume control was rotated. I touched the aerial socket with my finger and the hissing noise increased in volume which on the face of it is a very good sign. I looked around for a wire to use as an aerial and then there was a cracking noise, the hissy noise vanished and smoke rose from between the front panel and the chassis by the output transformer. I guess from the BFO can which is screwed to the rear of the front panel. I turned off the power and it took ages for the smoke to stop. Next, I'll need to detach the BFO can and investigate its innards. Not an easy job.

The BFO now uses V6, a 6BE6 (1=g1, 2=k, 3,4=h, 5=a, 6=g2+g4, 7=g3) which is a heptode so that's a bit puzzling as the original was an EF39, an ordinary pentode wired as a triode so what's going on? Pound to a penny the smoke is due to another condenser failure.

 

 

 

 

 

 

 Five views of the BFO box. Above (and left), connected to the red wire is a burnt resistor so the culprit will be the associated decoupling condenser. Several condensers will be replaced as removing the BFO box is not easy. Centre top shows a repaired coil but all coils and chokes have continuity. The pink condensers appear to have been pretty hot (being slightly darkened) although these pictures don't show this.

 

 On the right is the original circuit of the BFO but looking at the box contents there are several extra components.

Besides the main tuning coil there's an RF choke.

Two coax connections are present. One goes to V5 and the other to the rotary switch in the tone control position.

My guess is R28 (10K) is the burnt resistor. R42 is 47K. C54 is 15-45pF, C55 is 3-10pF, C56 is 100pF, C53 is 300pF, C57 is 10nF and C58 is 100nF (and probably very leaky).

Cathode o/p is via C52 (3pF) not shown.
  

 

 Because of the very cramped components in the BFO box I can't guarantee this circuit but it certainly looks different to the original.

I wonder if the BFO with its 6BE6 has been changed into a product detector otherwise why not have used a standard triode-connected pentode as original?

The purpose of the RFC looks odd. Perhaps it's a Pi-filter to reduce the 450KHz RF output whilst passing audio (hence that 0.1uF coupling condenser).
  

 
 

 Here are the condensers from the BFO box plus the burnt resistor. I fitted five yellow plastic replacements plus a new 200 ohm.

Although R28 was 10K the burnt one was whole and read 190 ohms and was probably fitted in place of the original when the valve was changed.

With difficulty due to the cramped space I fitted six new capacitors and a new HT feed resistor. I'll rig it up to an LT/HT supply tomorrow and see if it works before refitting it.

I've yet to swap any of the other parts in the set.

I wondered whether the radio would work without the BFO in place. I turned it on and it was silent although the volume control crackled when twiddled then noticed that touching the wire at the end of one of the screened cables unsoldered from the BFO produced a hum in the headphones.

The cable went to the rotary switch in the tone position so I rotated this and the set worked with a hiss in the background. Touching the aerial socket resulted in lots of background noise. I poked in a foot of wire and with the 1250-2100KHz coil plugged in I heard a weak broadcast station at around 1400KHz so I turned on my TinySA and found I could hear a good strong signal at -37dBm so the front end, IF amplifier and AM detector are working.

The coil packs and sockets need cleaning as the set went off and on and crackled when the coil wobbled.

I decided to temporarily fit the BFO using croc clip leads and see what happens before I put it back in the receiver. With the coil for Range E (1250-2100KHz) and using a short length of wire resulted in a very noisy background which disappears with no aerial so the thing is working on AM. I used my TinySA on 1500KHz at 500Hz modulation and this worked well producing a strong output at almost exactly 1500 on the dial and with the BFO switched to ON heard a heterodyne but turning the rotary switch (AM-SSB??) reduced the audio significantly which didn't seem right. I also tested the meter with an ohmeter and found it was open circuit which is really annoying as I thought I'd checked it earlier. I was about to see if the IF was set correctly at 450KHz (or whatever is the crystal frequency) and hoped to use what I believe is the S-Meter. I haven't figured out what the control under the meter is for yet.. maybe its for adjusting the meter sensitivity? See Fig 6 below.

I fitted another meter but rated at 50uA instead of 250uA. I added a pair of 10K resistors (one in each leg) as I'd measured 0.5V across the incoming leads. The meter worked but of course the pot below it clearly wanted to put the pointer at midway (the old one was a centre-zero type after all). I found that signals moved the pointer but not very much so I need to sort this out later. I've now got as far as receiving Radio 4 Longwave on Range H. Everything tunes to a peak and it was at this point I found the primary winding trimmers are sitting at HT (sparks.. not finger).

The BFO seems to tune to 450KHz but I'm not sure whether the product detector is working properly. Now's the time to start replacing bad condensers and fit new resistors as necessary.

 

With no signal and no AVC voltage the EF89 cathode voltage will be say 6.8V (cathode current = 10mA via 680) the 2.5K pot will be set to balance the EF89 standing cathode current and read zero. With a lot of AVC the EF89 cathode current might drop to say 4mA resulting in a cathode voltage of 2.7V which will tend to put a voltage across the meter and its shunt of 6.8-2.7=4.1V **

The meter has a resistance of 960 ohms and to read a fsd of 50uA will require 48mV (960x50/1000) across its terminals. Given that a max signal voltage producing 4.1V between the pot wiper and the EF89 cathode resistor the meter will need a shunt resulting in a meter voltage of 48mV. Say we expect to provide an extra 6mA (10mA-4mA) through the cathode resistor then the shunt will be about 8 ohms (48/6). At the moment a good aerial signal overloads the meter with its 12 ohm shunt so a reduction to 8 ohms would seem OK.

** Because the potentiometer chain draws only 3.5mA any increase in current in the same ball park through the cathode resistor will tend to decrease the wiper voltage, thus will cramp the higher meter readings.

Ideally the pot current should be at least 10 times the supply current but this would be excessive.

 

This is the draft S-Meter circuit. As the original 250uA meter was open circuit I tried a 50uA junk box meter and worked out the new shunt resistor.

 

Fig 6

 

 Here's the second bunch of condensers removed from the set.

I haven't bothered testing them all just yet and frankly I didn't see a dramatic improvement but of course, as I only used my ears, AGC would have masked some of the improvements.
  

 The next step is to improve the S Meter deflection and begin IF alignment. Incidently, I discovered the reason for intermittent loudspeaker operation. Unplugging the low impedance headphones significantly raises the loudspeaker volume. This is because the headphone jack socket is connected directly across V7 470Kohm grid leak (I suppose this isn't as crude as it might be because it does mute the speaker to a large extent). I also looked again at V3 and realised the "triple" structure inside was that of an EABC80 not an EBC80 as I'd initially thought so I've changed the earlier text references to show this. This part of the circuit looks pretty standard with the triode driving the output valve grid although there's lots of screened cables running to a new 2-way rotary switch mounted above the BFO which I'll need to trace. I'll refer to this as the AM/SSB switch.

I tried a long wire aerial and found something odd which I'd noticed earlier but only when AVC was turned off. As the RF gain is increased beyond about half way the audio starts to reduce in volume until it disappears. This happens when AVC is on or off. Presumably the AVC voltage or maybe the setting of the RF gain control is cutting off one of the valves. The original circuit (and, I believe the new circuit also) uses a variable resistor R41 in the joint cathodes of V1, V3 and V4 (RF Amp and 1st + 2nd IF Amps) so grounding the cathode resistors R6, R26 and R29 should maximise gain. I measured the voltage at the wiper of R41 and it went from 12 volts (=min gain) to 0 volts (=max gain).The AVC voltage tracks the RF gain setting reaching about minus 9 volts when the RF gain is max, and also, as the RF gain is increased the S-Meter rises and peaks correctly indicating that the 1st IF amplifier is working correctly, but when the AVC reaches around minus 5 volts the audio level starts to drop and by the time the AVC voltage has peaked the audio has disappeared.

Oddly this RF gain/AVC issue reminds me of a similar problem with another receiver I'm also working on (a Plessey PR155G) which has an equally confusing strange fault in its AGC circuitry.

In the original 358X circuit the AVC voltage is derived from one of V5's diodes (the other being the AM detector) and as the AVC voltage is behaving correctly the circuitry deriving this must be OK. The question is whether the AVC voltage is being used for another circuit or perhaps there's a bad component. Could the problem be in the EABC80 or its circuitry as this produces both the AVC voltage and the audio output? The problem turned out to be easy to solve. The EABC80 triode grid is fed from the AM/SSB switch and its 10Mohm grid leak connects to the anode of D2. I believe D2 is for a noise limiter as such a toggle switch has been fitted to the front panel and labelled "ANL ON/OFF, but as it carries most of the AVC voltage this reacts with the audio amplifier. It carries an AVC voltage of minus 8 volts when the receiver is tuned to Radio 4. Because of this connection, the triode grid is sitting at minus 4 volts and this is cutting off the triode whose anode increases from about 150 volts at no signal to 247 volts at max signal. The triode anode load is 220Kohm and the HT about 255 volts making the triode anode current only 8v/220K= 36uA. To prove the point I connected a 1Mohm resistor between the triode grid and ground. The audio returned and the RF gain control worked correctly. I see that with the 1Mohm in place the triode anode varies between 50 or 60 volts (anode current circa 1mA) depending on incoming signal strength. The problem may be due to the fact that the Product Detector/BFO isn't connected up. I need to trace the EABC80 circuitry and figure out the wiring around the AM/SSB switch plus the new product detector/BFO.

The receiver is now working very well except for a feedback whistle which occurs at most settings of the RF gain control with AVC off. There are several non-paper physically small condensers remaining around the AVC circuitry so I swapped these for new as in this area resistor values are very high and even a small leakage will upset voltages. The Product Detector/BFO seems to be working and I'll fit this back to the front panel then align the IF amplifier to match the crystal filter.

Looking at the volume control I noticed its mains switch was wired in very thin plastic insulated wire which looks to me like a low voltage type and the mains cable has only a red and black conductor pair with a black wire twisted around it for a safety earth. The next job will be to fit an IEC mains socket and replace the thin mains wires with something more suitable.

 
 

 This seems to be the Product Detector/BFO circuit used for resolving SSB.

The 6BE6 is mixing the tunable 450KHz BFO oscillator with the fixed 450KHz IF input signal to produce RF plus audio at its anode.

The anode circuit looks a bit odd because of the way I've drawn it. Basically the RFC forms part of a low pass Pi-filter blocking the 450KHz RF.

I've yet to trace the AM/SSB circuit switching.

 Full of confidence having cracked the latest fault on the old receiver I swapped the mains lead with one I cut off our old washing machine, put back the BFO box after rewiring it into the circuitry, and turned it on. Oh dear... the audio was very distorted, the set whistled at any setting of the RF gain control and didn't like AVC (it whistled and howled at me). The AVC voltage wasn't managing 12 volts but only 1 volt... I sat down and went over what I'd done.. but all seemed in order so could a valve have expired? I picked up the pencilled wiring diagram for the BFO and studied it... There were three wires, HT, LT and an extra ground wire which I'd added instead of relying on the screws holding the box to the front panel (or the coax braid which may introduce hum from the valve heater). Also a black and a white length of coax. Could I have reversed these inadvertently?

I opened the BFO box and removed its 10 screws, but found the coax leads were soldered to the right points, so put it back but with only a single screw and sat down again to study the sketch I'd made. .

It showed coax "WH" and coax "BL"... clearly the white inner for the black cable and the blue inner for the white coax. Then the penny dropped. It meant the white coax and the black coax (the outer sheath not the inner wire colours). I removed the box yet again and swapped over the coax connections, refitted it and... it worked fine. I'd reversed the input and output leads. With AVC ON the RF gain works over its range and the set is stable but with AVC OFF using the RF gain control lets the AVC line rise to 12 volts and the loudspeaker whistles and howls as soon as a strong station is tuned in. I can tune to a weak French longwave broadcast and the set is pretty good. AVC OFF gives me extra readability of a weak station as the RF gain is increased from zero to about three quarters before distorting. My guess is there's so much gain the IF stages are unstable. Maybe my best bet is to set up the IF response to give me a wider response (=less IF gain)?

The problem is probably not new. The RF gain control is active with AVC on as well as AVC off so restricting the RF gain by adding a fixed resistor will degrade the receiver sensitivity when AVC is on. The AVC switch is single-pole and cannot perform a double function unless supplemented by a semiconductor or two and that has to be beyond the pale....

Below.. a first pass at tracing the AVC circuitry. Note the 470Kohm at the EABC80 triode grid.

The factory mod for the noise limiter is shown on the right.

Next, I'll add the AM/SSB switch wiring and the volume control.

   
 

 I'm looking again at the provision of AVC to the EABC80 triode. Was this a wiring error or was the chap that swapped all the valves following a set of published modifications and, if the latter, did this suggest the audio AVC mod? Is it a practical thing to do must be the question and if so are there examples of this in other receivers? Pete, G4GJL found a rare example in the B40 series so I checked this and found the B40 uses a CV303 (=EF22 variable-mu pentode) audio amplifier connected to the AVC line but of course the rule is AVC can only be connected to variable-mu valves. In typical WW2 receivers RF and IF amplifiers a 6K7 is used with AVC with the 6J7 (sharp cut-off) pentode a straight amplifier. The 358X RF and IF amplifier valve type to which AVC is applied is the EF89 which is variable-mu but the EABC80 triode is not, therefore it won't work with AVC. So the 10M grid leak connected to the AVC line serves no purpose.. all it does is alter the triode anode voltage and with an increasingly large negative AVC voltage it will rapidly cut off the valve clipping the audio and eventually killing it, so I'll connect the grid leak to ground instead.

I did this and removed the shunt resistor and this proved to be the cure so the 10M grid leak should have been grounded.

I then decided to listen on a shortwave band. 40m was too noisy so I tried 20m using a ropey long wire in my conservatory but the background noise from local switchmode power supplies was dreadful. I tried the noise limiter and to my surprise the interference disappeared like magic. Tuning around 14MHz wasn't very successful so I checked the dial calibration and became very puzzled.

 

 

 When one is aligning any superhet receiver it's important that the design of the local oscillator is clear otherwise it will be impossible to align dial readings with incoming frequencies. Looking at the picture above, (for Range B) which shows the only way I could tune the coil, you can see the spike from the local oscillator measured at the aerial socket reads about 13.55MHz and the dial pointer reads 14MHz. The mixer is generating an IF of 450KHz so the oscillator is on the low side of the signal with an image tunable at 13.10MHz. Is this correct, when more often than not the local oscillator works on the high side of the incoming signal with the chief reason being the tuning gang sizes.

In this receiver all three sections of the tuning condenser are given as 271.5pF. This means that it's not straightforward to tune an oscillator on the low side of the signal and accurately track across the dial. The easiest way of tracking, given equal tuning condensers is to use what's known as a padder condenser whose value can be calculated so that the oscillator and RF track across the dial. The other way is to use a tuning condenser having different oscillator and RF sections.

In the 358X each of the coilpacks A, B & C carries a fixed padder, C7, and for the remainder both a fixed plus a variable condenser adjusted to line up the dial readings across the tuning range. The effect of the padders is to reduce the oscillator tuning condenser effective value slightly so that at any point in the range the oscillator will be higher than the incoming signal by an exact amount equal to the IF. So in the picture above why is the oscillator frequency lower than that of the signal? The reason is not easy to explain but may be the result of swapping the oscillator valve with the consequent change of wiring layout. To help determine the root cause of the problem I need to check other coilpacks and also the components used in the local oscillator.

To confirm my understanding read the excerpt shown below. The signal generator display being used back in the 1940s would be somewhat vaguer than modern types so setting it to 32Mc/s would have been a bit hit and miss. This is demonstrated by the expression "lower frequency of the two" refering to 32Mc/s and the image 32.9Mc/s. Using the "lower" indicates that the local oscillator is higher than the signal generator output. Range B and C are the same.

During testing the loudspeaker faded away so I had to use headphones. Is it a bad 6BW6 or has a heater wire come adrift? I swapped the 6BW6 but nothing changed so a wire's come adrift.

 

 

 

 

Range A coil details

 

 

 
  

 

 Left the local oscillator and above the values of local oscillator trimmers C3 and padders C7 used in the coilpacks. Unfortunately the documentation at my disposal doesn't supply the coil details.

Effectively C7 is in series with the tuning condenser so although the RF coils tune with a swing of 271.5pF the oscillator tunes with only 73pF bearing in mind one has to also take into account stray capacitances plus the setting of the trimmer.

 

 As I checked the various coilpacks I found some of the ceramic trimmers were failing. This is a common problem after a long period of disuse when either the top surface cracks or the solder fixing the screw to the top surface fails.

 

 

 

 I checked around the ECH81 circuitry after comparing the internal capacitances between this new valve and the original ECH35.

In fact the ECH35 has much higher values of capacitance (anode-cathode of 9pF compared with about 3pF for the new valve) and the wiring looked pretty much as original so I decided to check the tuning condenser was fully unmeshed with a dial reading of 100. It was fine so I opened the coilpack and removed the oscillator fixed trimmer, marked 20pF (note the coil details ... above do not include this fixed trimmer and so might be a manufacturing error!).

This measured a little under 25pF which is slightly more than its +/-20% marking says. In fact I think this capacitor was a poor choice, unless specially selected from a batch, because at 32MHz 4pF has a large effect (let's say 29 to 35MHz) and at 24.7pF the max frequency will be 28.79MHz making accurate calibration impossible.

The solution was easy. I selected an 18pF suflex capacitor, which measured 19.2pF, and fitted this (opposite).

The oscillator then tuned up to 32.66MHz so I was able to set the variable trimmer to meet the requirement of 32.450MHz. The image is now higher at 32.9MHz and the tuning range is correctly 32 to 22MHz conforming to the coil tuning graph.

 

 

 After fitting the Range A coil and adjusting it to meet the spec (32-22MHz) I checked the image level and found it was reasonable then tuned to the ten meter band where I was able to find several SSB stations. Next, I need to sort out Range B which also fails to tune correctly.

The Range B coil (the highest range to have a dial scale) was slightly puzzling. The indicated frequency was too low by around a quarter of an inch because the oscillator couldn't be persuaded to go high enough to match the marked frequencies. It was possible though to use the oscillator on the low side of test signals, but this would make it impossible to track the oscillator to the RF tuning (and the dial markings). Unlike the Range A coil there was no fixed trimmer to explain the problem, but Range B does have a physically large padding condenser, listed and marked as 0.0046uF +/- 5%. If this was too high in value it might prevent the coil from reaching 22.450MHz so I removed it and found it measured 0.0048uF (the top end of its tolerance). I swapped it for a set of three suflex capacitors equalling about 0.0046uF and re-checked the max frequency (ie with the variable trimmer at min capacity). Surprisingly the frequency seemed too low for the 200pF change in padder but I found this was because the screening cover wasn't fitted. Putting it over the coils pushed the oscillator to just over 22.500MHz. Checking the opposite end of the dial I then found the lowest frequency of the range at 9MHz was too low, but remembering the comment in the handbook about the coils being "very carefully adjusted before leaving the factory" I noticed the turn at the top of the coil was loose enough to allow it to be slipped upwards or downwards thus reducing or increasing the inductance. Bearing in mind that coil inductance should be modified for accurate alignment at the low end I pushed the end turn up to the next notch on its former. This increased the oscillator frequency up to 9.450MHz matching the dial reading for 9MHz. Returning to the high end a tiny adjustment of the trimmer re-aligned the dial at 22MHz.

A final check at the 14MHz dial reading revealed it was accurate enough to hear CW signals when the tuning knob was moved very slightly anti-clockwise. I did notice however that although it was easy enough to resolve SSB with the BFO switched on, with the product detector switched in it was quite tricky and hum was audible so this area needs some attention.

 

 I now need to do a quick check on the remaining coils.

I can understand why the receiver is so heavy. It needs all that weight to anchor it to the bench when swapping coils. I can imagine the electrical designers telling the mechanical engineer to make sure the cabinet weighed at least 20 pounds.

As a matter of general interest I tested all the old condensers removed from the radio using the receiver HT of 250 volts and a digital voltmeter.

The higher the voltage across the 1.2M series resistor the worse the condenser leakage.

 

 

 There are a couple of things about this receiver that annoy me. One is the BFO isn't quite strong enough and secondly, if I tune to a really strong signal, although the AVC copes admirably, switching this off and using the manual RF gain control results in instability. Because the RF gain is used with AVC any change to this will affect reception in both AVC and none AVC operation. The instability might be due to one or more of the stages of amplification bursting into oscillation when some of the output signal feeds back into the input. My guess is the 1st IF amplifier is the culprit because when I added a 680 ohm resistor (value found by trial and error) from the RF gain control wiper to the EF89 cathode it stopped the instability.

There are two ways used to control the receiver gain. The method that works well is AVC where rectification of the final IF signal produces a negative voltage which is applied to the grids of the RF amplifier, plus the 1st and 2nd IF amplifiers (all of which have variable mu characteristics enabling gain change through grid bias changes).

The manual gain control which is used both with AVC and by itself, puts a resistor into the cathode circuits of the same three stages. The aim being to introduce a higher positive voltage on the valve cathodes whilst under manual control leaving their control grids at ground potential (when AVC is off). Hence the grids of the three valves are more and more negative as the cathode resistance is increased by virtue of "automatic bias". The three amplifier valves in question however each have a fixed cathode resistor in place. These resistors already produce automatic bias in terms of the valves DC operating conditions thus setting their operating conditions in terms of anode curren. Adding an extra fixed resistance (ie. the setting of the RF gain control) changes the operating condition of each the three amplifier valves. For instability to occur it would seem that at least one valve isn't working properly in the presence of a very strong signal. Not only am I seeing instability, but once that is eliminated by altering the setting of the gain control severe distortion to the audio is apparent. In fact I'm now wondering if I'm seeing the same sort of fault when AVC was applied to the audio amplifier. In other words one or more of the amplifier valves is operating close to cut off? The way to check this is to measure the anode current of each valve as the RF gain control is varied when AVC is turned off.

 

 

 

 

 The only valve whose measurements were worthy of note was the EF183 RF amplifier. This has a 360 ohm cathode resistor which enables me to check the combined anode and screen grid current. Much to my surprise I found something like 0.2mV with Radio 4 longwave tuned in which equates to a mere 0.5uA current meaning the valve is virtually cut off. Comparing the specs of the original EF39 and the new EF183 with a screen grid voltage of 50v reveals an anode current of 10uA at minus 10V bias for the former compared with about 1uA for the latter. Looking at the EF183 spec it would seem that raising the screen grid voltage from circa 50V to 100V would increase the anode current from less than 1uA to 200uA. The feedback issue with its howling noise arises when the RF gain is turned to minimum and I'd already noticed it was worst with no RF gain decoupling to ground. I'd fitted an 0.5uF which reduced the frequency of the howl to about 300Hz so being curious I fitted a 47uF electrolytic.. and the howling disappeared.

When the receiver is operating with no signal everything is fine and the EF183 is drawing sensible current so I left it there. The only instance of howling noise I can find now is if the crystal filter is switched in, RF gain set at max, AVC off and Radio 4 tuned in to its peak. I can live with that and any audio distortion from trying to receive a very strong signal with no AVC.

Something I've been putting off for weeks was the 6BW6 heater going out. In order to continue testing I had to wrap a wire around the grounded heater pin and connect this to neighbouring chassis. The B9A valve socket pins were badly corroded and the only alternative was to replace it which I did yesterday. I also looked at the problem of the RF amplifier misbehaving when using manual RF gain control with no AVC. This only happens when tuning really strong signals so I ended up dropping the anode feed resistor from 22Kohm to 330 ohm which allowed a cathode current at max signal with max RF gain and AVC off of 15uA instead of close to zero.

A point of note however is the noise limiter which works exceedingly well in suppressing local interference from switching power supplies.

 

 

 This is the response of the AVC circuit when an IF signal is connected to the aerial socket.

It's only a representative response because there's an IF trap involved and the AVC circuit is slow to react. Also I was using my high impedance probe whose gain, although quite level is uncalibrated (and the battery is flat).

With a crystal filter I always use the crystal response as a marker for IF tuning and the BFO centre frequency.

In this case the IF peaks were 451.063 KHz and 450.786KHz giving a centre frequency of 449.924KHz and a peak to peak reading of 278Hz which works out at 300Hz bandwidth (the value specified in the user guide).

It took many passes to get this response, in particular to minimise the dip between the peaks. At this point I awas unaware of the phasing trimmer.

 

 This represents the response with the crystal switched out. Only one trimmer setting is valid (ie. the second trimmer in the crystal can which has very little effect with the crystal switched in).

Again the response is not an exact curve because I'm monitoring the AVC line which is heavily damped. One reason for using this is that the whole of the EABC80 circuit is covered by a metal screening cover. Ideally one should monitor the detector diode anode.

In this picture the span is considerably greater at 100KHz per division than the crystal filter scan (2KHz per division). You'll notice the centre frequency of 451KHz matches that of the first scan.

There's a loss of signal amplitude of a little under 9dB with the crystal switched in.

 The final things I did were to move the main smoothing condenser to sit vertically on the chassis behind the tuning condenser and to fit a plastic cover over the HT connections then to swap the Belling Lee connector for a BNC.

 

 

 

 

This is what the receiver handbook says about aligning the crystal filter. Use is made of the audio output socket rather than the AM detector and note the use of the BFO. Interestingly no mention is made of C63 (crystal balance) which presumably was tuned in the factory and was determined to never need touching?

 

 

 At this point I decided to tackle much the same job as before but this time to monitor, not the loudspeaker output as in the receiver user guide, and not using the AVC line as before to monitor the response, but to (correctly) use the signal at the rectifier diode in the EABC80. To do this I twisted the end of an inch of insulated wire to Pin 1 of the valve and connected this to my high impedance probe.

The results are shown below.

 

 

 This is with Narrow Bandwidth selected with AVC off and max RF gain.

To get this balanced shape to the response I adjusted the phasing trimmer accessed via the rear of the crystal filter can.

Note that the centre frequency (governed by the crystal filter) is 451KHz.

 

 

 This is with Normal Bandwidth selected with AVC off and max RF gain.

The only extra adjustment was to the second trimmer in the crystal can which isn't used in the narrow setting.

Note that the centre frequency (governed by the crystal filter) is still 451KHz.

 

 

 This is exactly as above with Normal Bandwidth selected but with AVC on and max RF gain.

 

 

 Finally, this is the BFO signal shown with the RF gain at minimum to lower the noise skirt.

I initially confirmed the BFO tuning control knob was set correctly with the left and right end points giving min and max frequencies at its marking of "5" either side. I then set the control to its midway position of "0" and carefully adjusted the frequency to 451KHz to match the IF response.

The anti-clockwise setting gives 447KHz and clockwise gives about 455KHz or minus/plus 4KHz, nicely balanced.

As in previous measurements this is the BFO signal at the EABC80 rectifier anode.

With a strong broadcast signal the BFO level, with RF gain at max and with AVC on, was very weak. However, using the product detector the beat note was pretty strong.

 

 I was initially puzzled by the results. Although I was getting really nice response curves I noticed that the IF bandwidth looked excessive. The "normal" setting registered 20KHz and "narrow" gave about 10KHz. I reasoned the width may have been because the tracking generator was set at -20dBm or about 22mV. To test this hypothesis I inserted 40dB of attenuation giving 225uV and reduced the RF gain to minimum and sure enough the response narrowed to that shown in the pictures below which are much more realistic. To prove the point I increased the RF gain and both responses maintained their general amplitudes but widened considerably.

Left is the "narrow" crystal filter setting and right is the "normal" setting.

 
   

 

 I then air-tsted the receiver first on 80m then on 40m. The 80m band was very noisy and the only signal I could hear was an AM QSO on 3.6MHz. On 40m LSB signals were loud and clear. I used the RF gain control in normal reception and then the product detector which worked perfectly apart from some hum. Either the valve has developed a heater-cathode leak or there's a measure of ground current in the signal connections.

 

 

  in progress

 Take a look at the coil information

 Return to Reception