McMichael 368

 This survivor from 1936 has arrived here to be put back in service. The set has been stored in damp conditions so the case is in poor shape and there's a good smattering of rust inside, but it's complete and I see no reason that it cannot be sorted out.


 The first thing you'll spot is the 120 volt battery which is probably over 50 years old.. I'll attempt to date it later (note.. it's post WW2). The set uses British valves and remarkably these are complete with intact metallising. This is important as there will be lots of gain and this may result in instabilty if screening is poor. Besides the HT battery there would have been a 2-volt accumulator but examination of the circuit reveals there was no grid bias battery. Instead the HT negative feed establishes bias for the valves. This is pretty important because, without negative grid bias, the valves would draw lots of HT current making the set expensive to run.



Above you can see tell-tale marks from a woodworm infestation. Apparently they've taken a particular liking to the speaker cloth, but in fact this means that interior woodwork has probably suffered. From the colour of the dust inside the case I think the things have long gone, but I'll need to paint on a woodworm treatment fluid as a precaution.

Most battery operated sets had no dial illumination so there's a nice clear dial with markings on paper. Many receivers from this era and to the end of valve sets used luminous dial markings or clever light-guides to illuminate station information. These needed dial lamps which would consume as much power in a battery set as the valves themselves hence doubling the trips to the local shop for recharging the accumulator. Initially I thought the dial had staining but once removed from the chassis I could see it was a strange semi-reflective finish.

One can date an early receiver by studying the station names marked on the dial. Click to see this page on dating a set.





 Before I could remove the chassis from the cabinet I had to detach the four knobs. This is frequently a difficult proposition and in this case it was especially tricky. There are no grub screws, instead the knobs are held in place with internal steel springs which had rusted in place. Two spindles are steel and two are brass making removal of the former extra difficult. There are a few ways of removing stubborn knobs. I tried a method wrapping string around the back and pulling which failed to budge them. I have a pair of pliers with hooked jaws and managed to get these behind one of the knobs. After heaving on this I heard a crack and the thing slid off. The other three knobs were pushed in place leaving too little space for the pliers, but I removed the four chassis securing bolts and this allowed the chassis to be pushed forwards leaving just enough space for the pliers. To help deal with the rust I'd dripped pentrating oil onto the shafts and this helped to get the other three loose without damage.

Once the chassis was ready for removal I unsoldered the speaker leads after marking these because oddly there are six separate wires.



 In the picture above you can see the dial cord on the right. Instead of the usual string it looks like pyjama cord or old fashioned bootlace. No chance of this breaking.. which is a good thing because the tuning condenser has seized solid and if it had been standard dial cord it would have snapped.



 The dial is interesting because it uses two pointers. The upper one is a standard pointer coupled to a slow motion drive made from a pair of brass gears and the second a pointer screwed to the tuning condenser shaft. The latter rotates through only 90 degrees so one side carries medium wave and the other long wave markings (these are wavelengths in metres as is usual in the UK). You will just be able to discern in red in the inner long wave scale the marking "Aircraft". Curious listeners to long waves might appreciate that these wavelengths are still in use today where short range MCW broadcasts identify airfields. For example, our local airport in Christchurch, which is named "Bournemouth" has a transmitter on 339KHz or 885 metres sending the callsign "BIA". No doubt standing for the grandiose title "Bournemouth International Airport".






 This receiver uses old British-based valves, and there's a good selection. From left to right...the RF amplifier and frequency changer/oscillator is a relatively rare TP22, a triode pentode with a 9-pin base. The IF amplifier, a B7-based VP210 is a variable mu pentode, the audio amplifier, detector and AGC rectifier is an HL21/DD using onlt a B5 base and the audio output valve, a B7-based QP230 which is a slightly exotic double pentode for push-pull operation. All have 2 volt filaments consuming a total of around 750mA. Anode currents for these valves are pretty low and in total would only drain 5 to 10mA from the HT battery depending on the strength of the tuned broadcast and audio output setting.



 The tuning condenser has three ganged sections, usually indicating an RF amplifier stage in front of the mixer, and there are two sets of RF coils, not because the set has a separate RF amplifier, but it has bandpass circuits aimed at providing decent reception of distant stations in the presence of strong local broadcasts. Also the low IF of only around 125KHz demands fairly sharp tuning to reduce image reception. Tuning condensers like the example above have a section which is adjustable so that perfectly tracked RF-Oscillator ganging can be maintained across the tuning range otherwise too much of an image may be leaked through resulting in a beat whistle spoiling reception. In this example you can see the nearest section has its tuning capacity reduced by the bending of sections of the aluminium vane.


 The first job is to sort out the tuning mechanism. Turning the spindle fails to turn the dial pointer because a number of the parts are seized due to rust. Initially I tried to lubricate the moving parts without dismantling everything. The mechanism is overly complicated and looking between the tuning condenser and the rear of the dial revealed something amiss and it was only by removing the dial that I could examine the problem. You can see the remains of a flexible coupler fastened to the tuning drum.




 The other part of the coupler is fitted to the dial pointer gears via a collar held in place by a grub screw. The design of the old coupler allows for variations in the fitting of the dial and gears.


 Looking around for a part suitable for making a new coupler I found an old solder reel that had the correct diameter... Below the reel drilled and held in place on the tuning drum and then fitted with the second part.




 The new coupler isn't very flexible so I drilled out the fixing holes to enable it to be accurately lined up with the front panel holding the dial. Once the driving hole was central I tightened the fixing screws.

  The tuning spindle pulley was slipping because the rubber washer was hardened and worn to a shiny finish so I found a couple of rubber VCR pulleys which will provide sufficient friction for the drive belt.



 The re-sleeved tuning pulley.


 To centralise the tuning mechanism the tuning condenser fixing screws need to be slackened then tightened when the dial had been fitted.



 The gearwheels have an anti-backlash spring.




The first components I checked were the electrolytic capacitors. The replacement Radiospares is marked 8uF (HT smoothing, C19) and checked out as pretty well perfect but the original 50uF (grid bias smoothing, C20) was open circuit. Both of these may have been important if the receiver was used with a battery eliminator to reduce hum. Click the link to see a collection of these.



 The next problem, after the tuning mechanism fault, was the push-pull driver transformer. This had an open circuit in its output winding. Oddly the primary winding which is fed by a capacitor that invariably would be expected to have leaked HT current and damaged the winding, was OK, measuring the correct value of 700 ohms, but one of the two centre-tapped secondary windings was open.circuit. This would have resulted in low audio volume. As the secondary winding is closest to the transformer core and its the inner winding that's failed the transformer is u/s. I need to find a replacement having a centre-tapped secondary winding measuring 2,500-0-2,500 ohms and a primary of 700 ohms. The ratio might be anything from 2:1 to 5:1. I could measure this perhaps because half the winding is OK?

I measured the input at the primary winding as 5 volts 1KHz with 15 volts at the centre tap to the good end of the secondary.. so the transformer is a 3:1 or 1:3 whichever way you want to express the gain. What about a replacement? Back in 1936 it wouldn't have been a problem, but over 80 years later the choice is quite restricted. There are basically two avenues.. one is to source a close match to an expensive specialist transformer and the other to use a small mains transformer. Looking at the parameters. First the construction of the transformer. Most interstage transformers used extremely thin wire (in fact the size of the wire is likely to be the reason for failure as it can fuse open circuit if a problem occurs such as a bad condenser or even damp conditions followed by corrosion) and can accommodate a huge number of turns resulting in a high inductance. The end result is a DC resistance and an AC impedance for the primary and secondary windings.

Starting with the primary winding. Once this has been determined the secondary winding will automatically be specified once the step up ratio is set. For a 3:1 step up the secondary will have three times the number of turns as the primary. Its resistance and impedance will then be decided by the size of the wire.

Looking at this specific receiver circuit you'll see an 0.1uF blocking condenser in series with the primary winding so no DC current will flow. At this point we should consider this condenser as a source of a problem because it needs to be completely leakage free and that is a tall order for an 80 year old condenser so it must be replaced, but maybe not because the primary winding is OK. Now the secondary winding connections... these connect directly to the grids of the QP230 with the centre tap connected via 100Kohm to HT negative. Looking at the circuit diagram you'll notice there is no grid bias battery, but instead a resistor chain which will pass HT current. Within a very short time, once the filaments are heated, each valve will establish anode (and screen current if applicable). The current will flow through the bias chain and very soon will settle to a steady state value determined by the HT current together with the bias applicable to each valve. What in fact happens is a surge in HT current which lasts for only a short duration with the amount of current flowing mainly determined by the zero bias characteristics of each valve.

The most important valve as far as the interstage transformer is concerned is the QP230. With zero bias the valve's pair of anode currents can total 50mA. The bias chain equals 920 ohm so assuming say an instantaneous HT current of 55mA the bias applied to the QP230 will be about minus 50 volts. Now the QP230 will be cut off. The current flow through the bias chain drops and the bias reduces to say minus 8 volts resulting in around 3mA of anode current. The end result or steady state HT current will eventually in short order stabilise at say 6mA. During this phase I wonder if the interstage transformer is stressed? I guess if there's a strong crackle from a dirty switch there may be a large voltage swing and conceivably this might encourage the QP230 to draw grid current, but the grid feed is via a 100Kohm resistor which would limit its effects? All told, I think the transformer just failed from corrosion in its copper windings brought on by damp conditions, as evidenced by the rusty chassis.

The old transformer is said to have a primary DC resistance of 700 ohms and across the whole secondary winding 5 Kohms. Assuming a ratio of 1:3 it looks like the secondary is wound from thinner wire than the primary winding having twice the resistance of the primary wire. The metal core is not very large in cross section being about 20% the area of a 50 Hz transformer core so I initially thought the inductances of the windings may not be especially high, but of course depends on how much wire is used (see below for a picture).

I put a 32nF capacitor across the primary winding and measured the resulting resonant frequency, repeating for the good half of the secondary. I found the primary resonance was 130Hz and the secondary 50Hz, giving me 47H and 316H respectively. Allowing say 1nF for winding self-capacity these figures drop to 35H and 214H. These seem pretty high so I then checked the windings in series with the same capacitor, and ignoring the windings self capacity, I got 44H for the primary and 390H for the half-secondary which are in the same ball park. No doubt, if I allowed for the self capacitances of the windings the final numbers would probably be closer. In fact it's relatively easy to measure self capacitance of coils using a square wave and oscilloscope. Having got a rough idea of the winding inductances one could calculate their impedances although the results will vary significantly depending on the value of their self capacitance. Trying a few numbers an impedance of at least 300 Kohm would result for the primary winding impedance.

 There are some transformers that might be suitable for the interstage transformer but I'm not sure about their winding resistances or their winding inductances so I tried an experiment using three small 1.5VA transformers designed for 440 volt to twin 12 volt windings. These have a 10Kohm primary winding.

Feeding in 5 volts RMS at 1KHz gave me about 225mV output across the two 12 v windings in series which I connected to a 12 volt winding of a second transformer. This resulted in about 7 volts output. Not a gain of three but at least a possible decent match to the valves.

Next I tried using a small 230 volt to 2x 20v transformer to drive the two 440 volt and this should produce a voltage gain... this worked OK giving me a gain of between 2 and 3 and the resistance of the primary winding being only about 1Kohm compared with 10Kohm for the 440v transformer means its a much closer match to the original, so I will need to test the transformers in the receiver and see how this compares with bench tests.


 Below, the output compared with the input showing a step up of two and below this a picture of the two outputs with a 90 degree difference in phase. The input resistance is about 1,000 ohms and the output 10Kohm + 10Kohm.



 Views of the new interstage transformer mounted adjacent to the QP230 socket.


 Above.. I treated the chassis with a rust remover which left the patina....

 Now, the output transformer. this is just as old as the interstage transformer and has a centre-tapped primary winding totalling 750 ohms and a secondary DC resistance of 0.2 ohm (see below for a picture).

Sure enough, both primary windings were open circuit. Thinking about why this should be so, I remembered the electrolytic condenser wired across the bias chain. This is measuring open circuit, but if it were to have failed initially showing a serious leak the QP230 would be zero-biased and draw loads of current. This would result in a transformer dissipation of about 2 watts if 50mA was drawn through the 750 ohm winding.

I'll try a 115v-0-115v mains transformer in its place.

The loudspeaker is a Rola marked as having 8 ohm impedance. The QP230 output valve has an anode-to-anode output impedance of 16 to 18 Kohm so the output transformer needs a step down ratio of something short of 2,000:1 and of course it needs to be centre-tapped for the HT feed. A typical 6V6 output load is around 5 Kohm (ratio of 625:1) suggesting the QP230 output transformer has a lot more turns and therefore a higher resistance than typically found making it more susceptible to accidental damage.


If I was faint-hearted I'd have ditched this receiver as too hard to fix and it was even more likely when I'd removed the duff output transformer from the loudspeaker and discovered the loudspeaker was also open circuit. I was interested in tracing the fault so dismantled the thing. It was clear that some jigs are needed to assemble the cone and the magnet assembly because of the tiny air gap between the cone and metalwork. There was some corrosion in the wire from the cone to the coil, but I cleaned some to bare copper and found the fault to be in the coil itself. This has two layers of thin enamelled copper wire and there must be a break in the inner layer so at this point I decided a cheap replacement permanent magnet speaker will have to do.

Possibly the output transformer had developed a short between primary and secondary and placed a high voltage on the speaker coil?

I tried a mains transformer with twin 115 volt inputs and 12 volts output... not a very good match in theory but I'll see if it works. I mounted the new output transformer on a metal strip and fixed it between the tuning condenser and the QP230 using existing chassis holes and wired it up. Next, I removed the audio coupling condenser and tested it for leakage. With 200 volts across the condenser in series with a 100Kohm resistor, I found about 190 volts across the resistor. If the condenser was good this voltage would have been close to zero. I also changed the bias decoupling condenser marked 50uF at 12 volts working whose positive terminal connects to chassis. I fitted 100uF at 63 volt working.




 I then connected a variable HT supply to the HT leads and checked the total HT leakage. This was about 1mA at 120 volts. I also tested the bias supply and found it changed quite a bit as the HT was varied, settling to minus 6.8 volts with the HT at 120 volts.

I checked the valve filaments and all were intact so it's still a goer... Next was to replace the valves, connect up a loudspeaker and connect the low tension for the valve filaments. These consume about 700mA so I set my PSU to 2 volts and set the current limit at 500mA then carefully increased this whilst watching the voltage. At something under 700mA the voltage stabilised at exactly 2 volts. Then I connected a variable HT supply whilst monitoring the current, stopping at 120 volts with the current at around 10mA.

With the receiver of this vintage it's possible to check the output stage simply by touching the top cap of the triode amplifier. This produced a strong hum in the speaker (proving the new transformers are servicable). Note that just because a valve carries a top cap doesn't necessarily mean that this is the control grid connection. In fact the VP210 (below) has its anode connected to the top cap. I'd temporarily fitted a pair of knobs to the tuner and wavechange switch but adjusting either didn't have much effect on the faint hiss/hum from the speaker. Touching a long wire aerial to the front end resulted in only faint crackling noises so I connected a signal generator to look for the IF. After some experimentation I found I could hear a signal at 110KHz and to hear this the signal generator was running 1000mV.


 This is the first IF transformer which follows the TP22 frequency changer and is designed to run at precisely 128.5KHz.

The two rusty screws had been fully tightened. These are compression trimmers and their excessive capacity is the reason for the 110KHz response. By unscrewing these, together with those on the second transformer I was able to shift the receiver response to the correct IF of 128.5KHz, and was able to reduce the generator output to something like 20mV from its original 1000mV. In doing this I found the automatic volume control was doing a good job of maintaining a constant audio level.

It became apparent that the anode coil in each transformer peaked nicely but neither of the two grid coils tuned. Looking at the circuit diagram there are a few critical components that might explain this. One is the decoupling condenser C5 which provides a return path to ground from both IF amplifier grid coils.

 Having proved the receiver was essentially working I tried to hear a test signal on either the long or medium waveband. It became obvious that the wavechange switch needs cleaning but in a working position I was able to hear frequencies from about 550KHz up to about 1450KHz but nothing at all on the long wave setting so I set the generator to 828KHz corresponding to a strong medium wave broadcaster and found I could faintly hear it but only with a long wire connected to the top of the input coil adjacent to the right of the dial. Tuning across the station told me the IF response was very lumpy indicating the same tuning problem as when adjusting the IF tuning.

The very poor sensitivity can be due to a variety of things... chief culprits being the various decoupling condensers C4, C10, C11 and two RF couping condensers C12 and C13. There's also the tuning condenser trimmers... these may have been screwed tight like those on top of the IF transformers? That might explain why the highest frequency I could hear was 1450KHz. The dial shows Bournemouth which was close to 200 meters in 1936. 200 meters corresponds to 1500KHz and usually sets tuned slightly higher than this. Another reason the waveband is too narrow is the oscillator padder condenser which is marked as 1,081pF. If this has gone down in capacity, even by a small amount, the tuning dial range would be reduced. Why no long wave reception? Probably the wavechange switch which is an extremely simple affair carrying only three single pole switches.. Preselector, RF tuning and Oscillator.

I fitted new capacitors in place of C4, C5, C10, C11, C12, C16 and C17.... C5 (0.1uF) actually measured 15uF and the others about double their marked values, but connecting them to 200 volts via 100Kohm resistor showed all were leaky with between 70 and 160 volts across the resistor (meaning all had leaks of around 2mA. This would have resulted in low valve screen voltages and also reduced signal strengths. C16 and C17 across the output transformer would reduce the potential audio output. After re-testing the receiver and re-aligning the IF to 128.5KHz, not much had changed except I could now hear a very weak Radio 4 on 198KHz so thankfully the oscillator is working OK on medium and long waves.

It's also possible that one or more valves needs replacing, however, tweaking the LT up to 2.2 volts does not alter the audio at the loudspeaker which tends to indicate their emissions are satisfactory..




Above are the faulty transformers.. upper left is the interstage transformer and upper right the output transformer. Ordinarily these would be awkward to replace, but because they are both push-pull transformers it's impossible to obtain a replacement from their era.



Opposite is a selection of the leaky condensers replaced by modern equivalents. 
 Worryingly the IF secondaries are still not tuning and the front end seems also to be completely flat with no change in signal strengths when adjusting the tuning condenser trimmers. Clearly there are at least two faults remaining, although feeding 20mV into the TH22 I see 200mV at the detector diode of the HL21DD socket and tuning to 198KHz I see a gain of 200 which is presumably coming from the frequency changer and the IF amplifier. It looks like I'm going to have to make some resistance checks of the coils and then remove the IF cans and inspect the circuitry.



  It was easy to detect the fault. With the can screwed to the chassis my meter measured 42 ohms across the anode winding and 43Mohm across the grid winding.

Each complete can and transformer was easy to remove as only two 4BA nuts hold them in place and the four connecting wires, like most in this set, are only touch-soldered in place. However it was difficult to remove the coil assembly from the can because the metal bar at the bottom of the coil former is bent by machine after passing through slots in the aluminium can. The two slots needed to be cut and their edges bent back then two nuts removed from the top where the trimmer assembly is fitted.

In the centre of the ceramic top there's a screw sealed by resin. The screw holds the trimmer assembly to a wooden plug which was jammed into the coil former. Over the years the wooden plug shrunk allowing the trimmer assembly to turn independently of the coil former.

I suspect vibration from the loudspeaker produced sufficient movement of the coil former for the black connecting wires to rub against the coils.



 Here's a view of the break. In fact it was a design fault.. although I can't say exactly when the radio stopped working but the problem was slowly moving to the point where an IF coil went open circuit. Interestingly, one of the can securing nuts was missing.. was this an echo of a previous repairer who'd given up?

The IF coils comprise two tuned windings inside the cans and separated by a few centimetres . These transformers were never intended to be repaired after manufacture because the fixing screws are bent through the sides of aluminium cans and the cans have to be cut to remove the coil assemblies.. you'll note the plural...

 Both transformers had an identical fault (same wire.. same position), hence my diagnosis as a design flaw. There are four cotton covered wires stretched between the trimmers at the top and the pins at the bottom. These were in contact with the coils and at that point you could see green verdigris where friction between the cotton and the coil had removed the coil insulation. At the crossing points of the thin wires and the connecting wires a small cardboard strip was glued in place, but in a few instances the connecting wire was outside the protection of the strip and pressing on the coil. The coils are made from single stranded wire rather than better quality Litz wire so were easy to fix... it was the outer turns on each coil that had fractured so no real problem making the repairs.. I was careful to bend the black connecting wires away from the coils before putting back in their cans.



 Here you can see the simple repair.

Note the dark discolouration on both coils. This is transference of pigment from the black wire as well as some verdigris. It's spread over a couple of millimetres because the coil former is held loosely to the fixed trimmer assembly to which four black wires are connected and any vibration would have wobbled the coil former but not the can itself. The fixed trimmer assembly wires then rubbed on the outer surface of the coils. Any damp in the air would be absorbed by the cotton covering then, once bare copper was exposed from friction, verdigris was produced resulting in eventual fracture of the thin wire.


 In the second transformer, the paper strip did not quite cover the coil and again wear followed by damp and corrosion resulted in the break.

To avoid a repeat of the problem I superglued the coil former to the trimmer assembly. The junction between the two is via a wooden plug in the top of the coil former. The plug had shrunk slightly so the coil former was loose.

I applied superglue around the circumference of the wooden plug on each transformer then bent the black wires away from the coils.

When I'd finished all four coils had much the same DC resistance of about 42 ohms.

 After re-fitting the two IF transformers the radio turned from deaf to very noisy. I re-aligned the IF to 128.5KHz and this time all four trimmers were able to peak the signal and attaching a long wire aerial brought in dozens of broadcasts. These sets are tricky to align because of their tendency to burst into oscillation and I found the trimmers on the variable condenser need looking at as their springs which open and reduce the capacity are too weak and need bending outwards.

 Next, I'll set aside the chassis and take a look at the cabinet. It needs a woodworm treatment first. Here's a set of pictures taken as work proceeded.










 When you're restoring the cabinet of an old radio it's not usually a good idea to strip off the old finish. An old radio like other "antiques" needs to retain it's patina, When faced with hundreds of woodworm holes you have a problem but although I think the blighters have long gone I still brushed on lots of woodworm treatment fluid. This set isn't too bad and if you want to see a worse case look here and scroll down to the Model 100 phone, then see it's refurbishment.

The only difficulty I met with this cabinet was removing the old speaker cloth. This was glued to the rear surface of the cabinet front and on top of the cloth was the speaker baffle. This was not only secured by six completely rusted screws, but wedged in place by two side cheeks each held in place with four more rusty screws.

I eventually removed 14 screws, the baffle and the cloth. A couple of the screws sheared off and one had to be chiselled out.

I have a box of various speaker cloths so I'll pick out the best looking after I've finished work the cabinet exterior.

The panels on this cabinet are in fair condition with only the top almost devoid of paint, a not uncommon condition because owners often plonked a large plant on the top and then over-watered it so the plant pot overflowed destroying the finish. I rubbed the panels down with 800 grade emery cloth and oil then brushed on a wax paint. I tried filling some of the woodworm holes but it was a waste of time because they remained visible.

Below.. methods of fixing back loose veneer and re-gluing a loose top and then after preparation for reassembly.






 Originally the cloth was stuck to the rear of the front panel but it's easier to stick it to the speaker baffle then adjust the tension using staples.

The glass clips in place.

Shortly the chassis will be fitted but first I'll complete testing and replacement of old power supply wiring.



 Now, back to the receiver chassis. I'd noticed a few minor problems during initial work... the tuning condenser preset weren't working properly and there was an intermittent fault which reduced the audio output whilst drawing an extra 25mA or so of HT current. The top of the tuning condenser had some patches of rust and the trimmer washers were heavily corroded so I removed the parts.




 Removing the adjusting nuts and the phospor bronze strips revealed a problem. Corrosion was preventing sufficiently low resistance contact for the trimmers to operate.

Once the rust had been cleaned away and new washers fitted all three trimmers worked perfectly allowing alignment of the medium waveband. These trimmers are adjusted once the dial has been set to 214 meters. The coil inductances are designed such that once medium waves are aligned using the trimmers and the long wave oscillator trimmer adjusted correctly both medium wave and long wave alignment is excellent.


 Before final alignment of medium and long waves was completed I set the IF amplifier to exactly 128.5KHz. The official alignment details advise adjusting the four trimmers on the IF coils for maximum audio output, but I found a double hump when tuning across Radio 4 Long Wave so I used the spectrum analyser and found the response as initially set up was pretty poor. I believe this is due to the action of the automatic volume control circuit which is very sensitive and results in distortion of the IF response. However, using the spectrum analyser and after twiddling the trimmers several times the IF ampifier suddenly took on a decent shape with a single tuning peak and excellent skirts... below. The auto setting was used and this resulted in odd cursor settings. The response is better than +/- 13.5KHz @ -50dB or +/- 5KHz @ -10dB. The 3dB bandwidth looks about 7KHz which is a trifle low for high fidelity but ideal for night-time listening on a noisy and crowded medium waveband.


 During alignment the intermittent fault got really annoying. I traced this to the output valve which drew an extra 25mA HT if tapped gently. I substituted a QP22B valve in place of the QP230 and the intermittent magically disappeared. The total HT current read about 10mA at 120 volts and LT 675mA at exactly 2.0 volts. With a long wire aerial the medium and long wavebands are now filled with stations. Time to return the receiver to its refurbished cabinet and consider a suitable power supply.



 As HT batteries and accumulators are not readily available the best bet for powering this old receiver will be a mains power supply. Back in the 1930s it was not uncommon to buy a battery eliminator to use in place of an HT battery and sometimes it was possible to also replace the accumulator but this would be a relatively rare option because of annoying hum from inadequate smoothing. Nowadays it's possible to generate a 2 volt DC supply with no hum. What isn't common these days is the availability of a mains transformer suitable for directly producing a 120 volt HT voltage so one must use a voltage multiplier. Conveniently, transformers salvaged from equipments requiring several voltages may be at hand in most electronic workshops, and with their secondaries wired in series will produce, using a voltage doubler, a voltage suitable for the HT for this sort of receiver. The peak DC output from the transformer shown below will be root 2 times (27 + 14 + 12) volts = 74 volts. Using a simple voltage doubler this will around 148 volts. This can be fed through a choke salvaged from a switching power and result in an HT voltage under a load drawing say 10mA of about 130 volts.

The filament supply can be developed using a second transformer having a rating of say 10VA. I selected a 6 volt transformer whose output is rectified by a full wave bridge and fed to an adjustable voltage regulator set to exactly 2 volts by the 500 ohm pot. The 750mA drawn by the receiver poses no problem currentwise and the PSU delivers a rock steady 2 volts with no ripple. The 317T regulator has a small heatsink to limit its temperature.

The power ON LED is in series with a 12Kohm resistor. This also serves two other purposes.. to draw some current to limit the HT voltage and to drain any voltage when the PSU is turned off. The LT output is prevented from rising too high by the inclusion of a 3.3 volt zener diode which will blow the LT fuse if for any reason the voltage rises too high. I fitted a 2 amp fuse in the LT circuit and a 100mA fuse in the HT circuit.



 The outputs are carried by sockets for the HT and screws for the accumulator spade terminals. To the right of the terminals you can see the (ten turn) adjusting screw for the LT voltage. Note that the McMichael requires an isolated HT negative feed to develop correct bias voltages for its valves.




 I tested the completed receiver using the new PSU and found it was so sensitive it picked up stations without an aerial.

 Return to Reception