Restoration of a DST100

A task not for the weak or faint-hearted

See the receiver write-up

 This example of the DST100 is a MkIII* made by R.A.P.and probably dates to 1942. The star in the code means it was tropicalised. Some electrical work has been carried out on it in the last 30 years but since then it appears to have been stored in slightly damp conditions resulting in light oxidation to aluminium and rusting to various steel parts, however it's in very good overall condition. The restoration which will sort out the slight corrosion, fix any obvious problems and make the set usable will include construction of a power supply much like the original "Rectifier No.8".

 

 

 

 
 Above some dating evidence. Both meters were made by M.I.P.Ltd, a company dating back perhaps to Edwardian times but although I found many meters for sale bearing that name, I haven't yet discovered who they were. Below are pictures of the slow motion knob which reduces even further the 24:1 main drive reduction. Below left you can see a twist of around 30 degrees and below this the reason for the distorted mounting. The slow motion knob is fitted to a lever which is attached to a spring behind the front panel enabling the mechanism to be pushed away from the outer ring of the dial. This then allows rapid tuning with a flywheel effect. the knob is positioned such that careless handling of the set can place its 100 pound odd of weight onto the lever with a high mechanical advantage on the securing screw. The result is distortion of the lever.The lever is a peculiar shape and flattening it is tricky. Below right is the refitted knob. Below this you can see the repair to the friction material. The original rubber had almost entirely disappeared and what was left hardened with age. I have a large collection of VCR rubber tyres left over from my days of repairing these things and I found a packet of half a dozen and four of these slid snugly onto the bush providing perfectly smooth drive for the outer rim of the dial.

 

 

 

 

Although the overall paint finish is presentable the four handles were pretty rusty. The picture on the right shows one that was refitted after cleaning. The 2BA securing nuts used for three of the handles are all fairly accessible so all can be readily detached. The larger centre handle and the top of the IF/AF section are secured by 2BA screws the reason being that nuts would have been in an inaccessible metal channel. One or two handles needed to be bent back into shape because they'd been distorted due to rough handling of the set and one or two of the securing nuts were peculiar sizes. The handle secured by two screws was very difficult to refit because of manufacturing tolerances. The handle was slightly too wide and had to be bent to fit the screws, an almost impossible task. After trying for an hour I had to use extra long screws at an angle. As these tightened up the handle gradually bent to fit the screw holes. The large number of the screws used to secure the panels around the set were a mess. The threads were originally 4BA but some metric or US replacements had been fitted and in one or two cases I had to drill out broken screws and retap the threads. You can see below the odd shape of the loose handle. This had been bent to fit the screw holes and because it was fitted with two studs it was easy to fit.

 

 

 

 
 

 

 Above you can see oxidation of the various metal parts.... Something I've never come across before is that all the 16 major parts you see, not just the valves are plugged into octal valve-holders. This includes all the IF transformer assemblies.

Below is the turret tuner using curved perspex covers and colour coded in-line with colours used on the plastic cursor forming part of the dial assembly. The R206 had rhodium-plated contacts on its drum and rumour has it that this resulted in most examples being scrapped for recovery of this precious metal. The DST100 coils are not fitted into copper boxes unlike those of the R206. This may have been to simplify coil adjustment because, once fitted into metal boxes their inductances would have been a lot different.

 

 

 Counting valves, one mustn't overlook this bunch above. These are fitted upside under the RF compartment. The metal cover hides the three gang tuning condenser.

Below you can see that many old wax covered condensers have been replaced with those yellow plastic things or is this evidence of modifications? Several old condensers still remain. All these will be leaky but in some applications the leak will not be critical (click to see).

I've developed a method of restoring the wax-covered condensers by stuffing with surface-mount chip capacitors (click to see). Why did I choose surface-mount types? Because I found a suppliers' bargain basement item of 0.1uF x 500 volt examples for 2p each and I bought a few hundred. Lots of resistors have been changed as well. Those smooth cylindrical types have brass ends which tarnish and result in high resistance values. The larger types with a single turn of wire wrapped round each end of the carbon rod are usually much more reliable. Strangely, unlike later resistors, values of resistors in most radios from the 1920s and early 30s are often very close to their markings.

It's pleasing to see ceramic valveholders. Sometimes WW2 bakelite examples can develop a low resistance between their sockets.

That giant ex-TV electrolytic is connected as an extra HT smoothing capacitor (I'll remove it shortly). Top right is a small transformer taped to the chassis... why?

 
With the refurbishment of any WW2 equipment I aim get the thing roughly into working order before spending a lot of time on the restoration, but as this example appeared to be complete, and with no obviously unserviceable parts, I reckoned it was worth expending effort making a contemporary-looking power supply before powering up the receiver... now shown below Click to see it

 

 I finished the replica power supply and connected it to the receiver, removed the PSU HT fuse and turned it on. The front panel lamp came on and you could see heaters glowing. The LT voltage measured about 6.7 volts, a little high but I can move the mains wiring to the +10 volts tapping on the LT transformer and reduce this to nearer 6.3 volts. I connected a variable HT supply and slowly wound it up. I could hear sparking and cracking noises and spotted what looked like a short in the 6V6 below the HT current meter. It was just a broken wire at the solder tag on the meter and sparks reflecting in the glass envelope. I fitted a new solder tag and refitted the wire. The HT current read over 100mA at something less than 200 volts but I could hear correctly sounding hissing in headphones. The various controls seemed to change the nature of the hissing but as I rotated the tone control anti-clockwise the HT current dropped sharply so there's probably a duff condenser there. I set the tone control to minimum and increased the HT to 235 volts at which point the HT current read 80mA. Turning the tone control fully clockwise increased this to 110mA so obviously there's something amiss.

I connected a long wire aerial and found Radio 4 on long waves. Operating the various controls indicated that there are alignment and overloading problems (the latter may be poor AVC due to condenser leaks) because the signal didn't tune in cleanly. Best reception was off-tune and some controls did unexpected things, still... all-told very promising first results.

 

 

 Most valved receivers have an audio amplifier feeding an output stage and this is running very often in Class A with a significant standing current. Class A gives excellent audio quality but relies on correct biasing to avoid exceeding the maximum power dissipation of the valve. Usually a cathode resistor serves to set the anode current as current passing through this resistor (anode + screen) will produce a negative grid voltage with respect to its cathode. Sometimes the cathode is grounded and a negative bias is supplied through the grid leak but this is relatively uncommon.

The problem arises when the coupling condenser (in this case C11N) develops a leak allowing current to pass from the amplifier valve anode to the grid circuit of the output valve. In this example the grid leak is R11K (100Kohm) to ground and the auto-bias is R24A (270 ohm) to ground. The amplifier anode load is R1J (250Kohm), R23A is 3Kohm and the pot VR6A is 500Kohm.

Roughly speaking, if C11N is 1Mohm it will result in about +6 volts or so on the grid of the 6V6. The auto-bias resulted originally in about minus 11 volts. The extra positive voltage from the leak will attempt to raise the 6V6 anode current by about 30mA and R24A will counteract this, but the anode current will anyway rise and continue to rise as C11N leakage gets worse until the 6V6 degrades.

 I noticed the ON/OFF switch leaves the 6V6 audio output stage powered up. Why is this? Maybe it's to keep a load on the HT rail to prevent the voltage rising to its peak level of around 350 volts and shortening the life of condensers and valves, but I wonder if it was done to preserve the HT current meter as it was a later mod done on the III and III* models. Suddenly turning on say 350 volts could result in a surge which might damage the meter. Turning back to the strange effect of the tone control (with reference to the circuit diagram) which is VR6A. If C11N was leaky then current would flow through R23 and forward bias the 6V6. With the full resistance of VR6A's 500Kohms in circuit R23 (3Kohm) and R1K (250Kohm) would reduce the forward bias and explain the reason for the 30mA difference.
 

 

 Here we have the first three duff condensers which I renewed. These were mounted on a small dismountable tagstrip and are C5B, C5C and C11N. C5B measured 1.25uF and had an ESR of over 20 ohms (it should be 25uF and less than half an ohm), C5C was open circuit and C11N, probably the worst offender in ALL valve radio receivers (ie. connecting the audio amp anode to the output valve grid) was not easy to measure. A capacitance meter failed to show a reading and an ohm meter gave varying results from 200Kohm to a couple of Megohms. The ONLY way to test these WW2 condensers is to apply HT across them and check their leakage. I wired a 200Kohm resistor in series with it and measured 235 volts across the resistor with 275 volts across the pair. If the condenser had been OK it should have resulted in virtually zero volts across the resistor. As the anode load of V8 is 250Kohm this means that most of the anode voltage would appear across the grid bias components of the 6V6.

Rotating the tone control enabled its high resistance to offset most of the positive bias on the 6V6 grid.

I fitted new 25uF capacitors into the old electrolytic cases and a 100nF x 500 volt chip capacitor into the old 0.1uF case.

 I fitted the stuffed condensers back in the receiver and tried again. The tone control now works a treat with the HT current stable.I've now noticed several minor (hopefully minor) problems. The main tuning dial appears to be a large number of degrees out. The noise limiter virtually kills any audio. The BFO seems to be maybe 10KHz out. The HT current seems too low.. never more than 90mA with the incoming HT line at 260 volts and it drops by 10mA if the RF sensitivity is minimum and that control has lost its end stops (see later). Switching the power to OFF leaves 50mA standing current which may be a bit high? A annoying feature is that if you select AVC you turn off the S-meter and probably vice-versa. There are about half a dozen or so wax covered condensers still lurking around and I'll be amazed if any are serviceable. What else...

 

 

Above, wrapped in tape, is what seems to be a replacement output transformer, although both the original and the new one are wired up, so this needs investigation. One popular mod was to fit a loudspeaker output transformer in place of that for high impedance headphones and 600 ohm line. Maybe the taping was unwillingness to drill holes and compromise the authenticity of the chassis?

Once I was happy that the set wasn't going to explode I continued investigations using the new power supply instead of the variable HT supply. All wavebands work although I suspect the local oscillator stops oscillating at the end of one range. The quiescent HT current is on the low side at 90mA instead of 110mA so maybe a valve or two has low emission or perhaps one or more screen resistor has gone high in value, or of course just as likely, the screen decoupling condenser is drawing current. That odd high-speed relay with the yellow wires is used for part of the single/double superhet switching (see later).

 
 Above a large 32uF x 375V Daly smoothing condenser taped to the chassis ( I later removed this and fitted a 22uF x 400V modern capacitor which had 50 times less volume). The original set had a pair of squarish cased condensers fitted either side of the LFC. These must have long since expired and been replaced by a pair of tubular types fitted with clamps to the chassis. The circuit shows three smoothing condensers, C14B=8uF, C22A and C22B=32uF. The replacements are marked 8uF + 32uF and one has a disconnected wire so must in turn have expired to be replaced by the big one. Below are lots of "new" parts. Most of the old condensers that can ruin the performance of the set still remain, so why has this area been messed with? I suspect these parts are associated with the AVC circuit where, because of really high resistor values, any leaky condensers would have killed the AVC action resulting in severe overloading of the RF stages. Then again, I suspect these were the easy ones to swap.. some are buried in the chassis and haven't been touched. The old high value resistors often go very high in value which explains the new ones below.

 

 I looked again at the dial being out of kilter, also I'm pretty sure it rotated properly after I'd repaired the extra-slow motion knob but since then the outer ring only rotates a short way before going clonk and stopping. I removed the centre dial (the disk carrying the markings) and you can see a plunger that pops up (pictures below) and prevents the tuning from continuing. I think this is a special end-stop designed to prevent damage to the mechanism. Rotating the dial backwards it clonks into the plunger again but the plunger can then be pressed down allowing further rotation which the feels as if the main tuning condenser has reached its limit. I might need to detach the large metal screen that covers this so I can see exactly what's going on but this simple procedure looks tricky because the valve top cap leads are threaded through it.

The main tuning dial has a slider which, at dial limits, engages with the end of the plunger and can then move an inch or so into rubber bushes which are hardened and brittle so need replacing. These bushes seem to be ordinary rubber grommets about 3/8 inch diameter so I can easily sort these out. I suspect the plunger isn't returning to allow dial rotation so I'll need to discover why this is. With an ordinary receiver I'm sure this would be easy but the sheer size and incredible weight of the DST100 makes this a real effort. Tuning to Range D, I can hear a weak signal in the narrow tuning range which sounds like Radio Solent. Is this a true signal or an image? Radio Solent for Dorset operates on 1359KHz on 800 watts and the Hampshire station on 999KHz also at 800 watts. Range D covers 780KHz to 1.9MHz which is confusing. Clearly the tuning is limited to a range somewhere in the middle of the band.

 

   To access this part of the dial mechanism is easy. First remove the perspex plate carrying the waveband information (three 4BA screws and washers).The centre of the dial carrying the frequency markings has a centre disk that should then be unscrewed revealing three 4BA screws . Remove these screws and lift off the brass dial (as above). This gives you access to the plunger (left), Here you can see the plunger is only partly retracted. This plunger pops out to prevent flywheel dial action from wrecking the tuning condenser couplings and gears. It's supposed to pop back in as soon as the dial is turned the opposite way but if it sticks in the out position the dial can only turn through 180 degrees.

 I didn't want to remove the main cast tuning dial for fear of upsetting the complicated gearing and stringing behind the panel, so what follows is with the dial in-situ. Armed with a strong lamp and some lubricating oil I set to on the assembly. The end-stop plunger was soon fixed. This is a brass piston about 1/4 inch diameter that moves inside a bush to protrude against a brass pulley that is able to slide within a brass slot fixed to the main cast dial. Lack of lubrication between the piston and its bush was part of the problem. You can see it's still partly sticking out in the picture above after oiling.

At each end of the slot in which the brass pulley slides is a rubber grommet against which the brass pulley strikes when it hits the plunger. I removed the remains of both grommets. The pulley is supposed to slide over a small arc but was jamming in its slot. I found that once the grommets had been removed the pulley could be pulled out but the hole for the grommet is almost an interference fit round the pulley. The slot could then be cleaned of residual bits of the old grommets. In my example the pulley wasn't running smoothly in the slot even after cleaning everything but, by scraping one edge of the slot and removing a minute amount of brass, the pulley fitted perfectly. The pulley has a protrusion that comes into contact with the plunger at each end of the tuning range and this protrusion had a ring of hardened and cracked rubber stuck to it. I scraped this off and fitted a tiny rubber tyre from my old VCR tyre collection. The tyre which acts as a shock absorber has to be of smaller diameter than the hole through which the pully fits. The whole exercise could be carried out easier with the cast dial removed but this wasn't essential.

 
   
 

 Top left the pulley extracted from the slot and now minus its perished rubber sleeve.

Top right, tightly fitted with a new sleeve (note: the sleeve was slid close to the end before refitting the pulley).

Left the pulley refitted and a pair of new grommets in place.

 The last job was to figure out the reason for a tightness in the gearing just before the end-stop plunger operated and some reluctance for the plunger to reliably fully retract. The plunger is activated by a raised section or bump on one of the tuning gears. The bump pushes on a spring loaded brass bar carrying a pip. Contact between the bump and the pip precisely levers the bar against a pivot causing the other end of the bar to push the plunger piston through its bush. This mechanism carries a spring arranged to retract the piston immediately the tuning dial is spun the opposite way. There's an adjusting screw near the pivot (accessed from the front panel) and this and the pivot point needed lubricating because the bar was jamming. Once oil had been applied to these final parts of the mechanism the end-stop feature worked perfectly. Finally, I applied oil to the collection of tuning gears as these were dry.

 
 

 Awkward to photograph... this picture shows the pivot point of the bar opposite the end carrying the plunger and below the pip is the adjusting screw which sets the amount the plunger protrudes from the front panel when it's activated.

The gear wheel above the pip carries a short ramped section about 2mm in height. This pushes on the pip and operates the lever to which its attached after 12 turns from either end of full travel. The tuning condenser turns through 180 degrees hence the quoted 24:1 basic tuning ratio.

By pressing on the bar against its spring tension with your finger, just next to the pip, it should push out the endstop piston then smoothly return when pressure is released.

 I've only described a small part of the complicated tuning arrangement because everything else seems to work OK, but behind the front panel there are several pulleys, cords and springs which are used to link subsidiary controls to the set's tuning. Thankfully, the Working Instructions cover restringing because at first glance it would seem an impossible task to replace the cords without a detailed explanation. Thinking about it the DST100 mechanical design team must have employed a clock maker to help and you can just about see in the above picture that the gears are anti-backlash types comprising twin gears coupled together with springs. The small pinion gear has a hyperbolic shape which presumably can take up any slack.

The design of the tuning system means that you can tune to a station then use the logging scale to note the dial position. The precise logging scale is very desirable for intercept receivers so a listener can set the dial and wait for a transmission. It would be interesting to see the Requirement Specification for this receiver which may have been written as early as 1935.

 Now that some of the mechanical problems had been resolved I needed to figure out why the dial readings are out of kilter. I selected Range F and tuned in Radio 4. This station broadcasts on 198KHz but the dial read 170 which is about 45 degrees out. The problem is mechanical though, because the end-stops read well into the dial readings rather than the lines at the end of the scales. The outer dial can be rotated by 120 degrees to the alternative positions so its not that the dial is set wrongly. It must be due to the position of the dial on the main spindle. Peering behind the front panel, amidst the gearwheels, I spotted a small gearwheel mounted on the centre spindle alongside another larger one. The small gear is secured by two setscrews which I was able to loosen. Because this type of screw is relatively fragile it's important to use a screwdriver that perfectly fits its slot. Once this was done the gear was stiff enough on the shaft to drive the mechanism but, by turning the dial plate with a little force this could be reset on the shaft. After a couple of attempts I was able to get both endstops to line up with the end of scale lines marked on the dial. Carefully tightening the screws fixed the dial in place and tuning to Radio 4 showed it roughly in the right place... close enough to enable the RF and oscillator coils to be reset.

 

 
 Left, you can just make out the small gear with its pair of setscrews that hold the dial to the main tuning shaft.

 I checked the frequency coverage of Range F and it was 116KHz to 310KHz so not too much wrong. Range G covered 48KHz to 128KHz, again not too bad. However, I noted that reception of the signal generator was scratchy and during proceedings the RF gain control decided to fail completely which didn't help. The bandwidth control seemed a bit dodgy and the BFO was miles out. The receiver is a double superhet on all but the lowest range but switching the bandwidth control to its broadest position arranges the circuitry to switch to single conversion. This means that there's some complicated switching involved and hence a strong possibility that faults can materialise. In fact, under the chassis is a high speed switching relay which carries out some of the circuit changes.

The HT current was reading 40mA with HT off and only 70mA with HT on and the reduction of the latter might be related to a reduction in the heater voltage which had been slightly high. I've now changed the LT transformer primary tap so I need to check the heater voltages at the valve pins because a low voltage will reduce valve emission. I checked the LT and it read 6.4 volts at the front dial lamp. The new HT current reading of 40mA as opposed to 50mA with the higher LT indicates a slightly low emission 6V6 and the overall shortage of HT current suggests a couple of other low emission valves.

 

 

 

 The duff RF gain control which is marked 5,000 ohms. There's no maker's name, which often suggests foreign made. I found an unused wirewound pot from WW2 having the same value and this worked fine.

The control knob had been turned too roughly, breaking the endstops and allowing the wiper to fall onto the very thin bakelite insulator between the resistance wire winding. The thin bakelite had given way allowing the wiper to jam against the inside of the wire and then to cut through it.

Below, you can see the difficulty working on a set of this size and incredible weight. Fortunately my workshop benches have angle iron supports running under their tops  because they were designed to support large TV sets, some of which (a 29" Sony and a 32" Toshiba) were even heavier than the DST100.

I've heard it said that the set was designed by John Scott Taggart. I can understand the reasoning behind this because of the number and type of controls used in the DST100 to winkle out the weakest stations, a technique used by JST in his range of receiver designs. Click the picture below to see an example of his work, the ST300 which is one of his simplest designs.

 

 The next stage before tackling RF alignment is to replace faulty parts. Leading the list are condensers. Looking into the RF section I can see that all bar two dodgy condensers have been replaced. This must have been done with the RF unit removed from the case. The original electrolytic condenser decoupling the RF stage gain control is still present which is rather odd. Why change twenty paper condensers and not bother with the electrolytic?

As the thing connects across the gain control it's simple to test, revealing "open circuit". It's slightly puzzling because the condenser is located with the parts for the 2nd mixer behind the front panel but I soon discovered a wire running from V1A at the rear, through a grommet, along the top side of the RF chassis, then back through another grommet to the front of the set. It was quite a long way down but I was able to twist it off the tag strip and lift it out. You can see below the before and after stuffing readings. I fitted a 47uF capacitor because I have a box of these.

 

 

 
 I planned to change the old condenser after lifting out the tuning drum, but I soon discovered that this is only possible once the RF chassis is removed from the case, a job I wasn't prepared to tackle unless absolutely necessary. The problem is the waveband selector whose screws aren't accessible prevents the drum from lifting out with the drum in place, a sort of catch 22... It might be possible to move the front panel an inch or so by removing the handles which are bolted through the front panel into the RF chassis casting and this might just be enough to separate the two drum gearwheels. I did remove the wavechange knob so was able to lubricate the mechanism as well as the drum rear bearing which I'd detached.

 

 

 Above shows the wavechange mechanism after detaching the knob. The starting handle-shaped thing has a large pip on the back which runs around the bearing surface when pulled out against spring tension and rotated until it drops into a hole and locks the drum. Fitting the parts back together is not quite straightforward because the small gear can rub on the front of the drum. I used a shim between the back of the small gearwheel and the front surface of the drum. With the shim in place the spindle for the main knob protrudes much further allowing the knob to be fitted securely. I'll need to clean the gears and apply fresh grease. A quick test showed the set was working slightly better but I noticed it was easy to get audio distortion with some control settings. I suspect this is primarily due to the remaining paper condensers still fitted to the IF/audio chassis. After further work on the old set I looked at the twin output transformers to see why there were two. It seems a second transforms the 300 ohm line output to loudspeaker impedance. I tried my headphones on the output but couldn't really hear anything. As I had a second pair of headphones I tried those and found the output louder but not what I'd have expected. Without thinking I plugged these headphones into the normal socket and nearly deafened myself such was the volume. In fact, turning down the volume control left the audio still pretty loud but only just acceptable. I tried the first pair and the audio was very low and distorted. The second pair gave undistorted audio so I'm afraid the first headphones are u/s. In fact I'd used them with a metal detector a week ago and wondered why the audio was very low. Anyway, all the DST100 audio distortion was due to the headphones. It was odd really. For example tuning to Radio 4 produced undistorted audio on the signal skirts and very distorted audio as the station was tuned in.

I tackled some more of the old condensers and give the findings below...

 

 

 I tested these two condensers by connecting HT across each with a 120Kohm series resistor. If a condenser is perfect no DC current will pass through it and you'll see no voltage across a series resistor.

The 0.04uF 350v condenser passed 1mA at a voltage of 300 so I saw 122 volts across the 120Kohm resistor and 178 volts across the condenser. At 200 volts the current was 0.76mA and at 100 volts 0.43mA. The DC resistance of the old condenser was therefore 178Kohm, 143Kohm and 112Kohm at the different test voltages.

The 0.1uF 350v condenser passed 0.23mA at 300 volts, 0.21mA at 200 volts and 0.11mA at 100 volts.

The DC resistance of the old condenser was therefore 1.2Mohm, 833Kohm and 791Kohm at the different test voltages.

 

 Here you see the same test carried out after the contents of one of the old condensers had been replaced with a 100nF chip capacitor. The voltage was 300 across the new capacitor in series with a 120Kohm resistor. The reading shows a voltage of 4mV across 120Kohm or a current of 33nA and this could well be due to the conductivity of the old case rather than capacitor leakage.This represents a DC resistance for the capacitor circuit of 10,000Mohm although I'm ignoring the resistance of the multi-meter.

HT is applied to one end of the condenser and a 120Kohm resistor to the other end. The other end of the resistor connects to HT negative and the voltmeter connects across the resistor.

 As I was twiddling the dial and listening to Radio Solent the audio began to fade and suddenly the reassuring hisses, pops and crackles just faded away. Turning up the volume control resulted in a low hiss so it sounds as if something has failed. Maybe a valve heater has gone out or a screen resistor gone open circuit? Maybe one of those dratted old condensers still remaining under the chassis has gone short-circuit... It happened again the next day but I think it may be dirty contacts on the turret tuner drum (I later found it was an intermittent connection in the last 110KHz IF trasformer). More experiments showed that position 4 on the selectivity control gives good results but none of the others work anything like as well. At this point I decided to change the audio output transformer because it was really unsuited to both modern headphones and loudspeakers. The last owner's idea of trying to convert a 300 ohm output to loudspeaker wasn't at all successful so I removed the original and the piggy-back transformers and looked in my junk box for an alternative. I found one marked 5,500 ohms/2.7 ohms and checked the 6V6 spec. This gave its anode load as 5,000 ohms and at 45mA/250 volts, about 4.5 watts output or around 40% efficiency. After wiring in the substitute the audio output sounded very much better, and for the first time drove a loudspeaker at quite a respectable volume with plenty of bass. At least I now won't get deafened when wearing headphones and switching the waveband selector. Another reason for removing the old transformer was to gain access to a number of old condensers and I removed and stuffed several as well as changing resistor R6G which read 70Kohm instead of 50Kohm. Whilst the output transformer was removed and access was easier to that area of the chassis, I also replaced another old condenser, this time C20A, 0.025uF which connects to a 110KHz filter choke. It had a steady relatively high leakage when tested at various HT voltages. I fitted a new yellow capacitor.

During bench testing I'd noticed several times that the set was sensitive to vibration so tapped various things under the main chassis. I found a bunch of half a dozen grounding wires at the new yellow capacitors where a pair were dry jointed. After heating and applying fresh solder the joints were made good.

I'm now almost at the stage where I can investigate various odd characteristics of the old receiver such as the bandwidth settings and overall alignment as everything now appears to work tolerably well. Before I tackle things further I need to read up about the receiver operation and figure out where all the components are fitted. Being a double superhet the second IF strip needs to be aligned, then the fixed frequency oscillator, the first IF strip and then the 6 wavebands. Fortunately the "Working Instructions" document clearly states how to go about alignment. The first test was to inject 110KHz at the tail end of the IF strip and adjust the trimmers. Two problems were found straight away. The trimmers which are compression condensers were seized and some were located directly underneath the end of the frame of the case and adjustment was impossible with them seized unless I made a special tool to free them first. I tried removing the metalwork carrying the handle but although I removed the four screws , without slackening off dozens of others the end metalwork couldn't be removed.

Below: pictures of the four 110KHz IF transformers with inductance/capacitance measurements

 
   

 IFT4, top coil=4.75mH, capacitance 429/600pF (111 to 94KHz)

 IFT4, bottom coil=4.72mH, capacitance 418/700pF (113 to 87KHz)
   

 IFT5, top coil=4.83mH capacitance 418/700pF (112 to 86KHz)

Resistor R2H 500K =859K

 IFT5, top coil=4.79mH capacitance 418/700pF(112 to 87KHz)

Condenser C17A 500pF=420pF
   

 IFT6, top coil=4.91mH capacitance 415/600pF (111 to 93KHz)

 IFT6, top coil=4.95mH capacitance 400/575pF (113 to 94KHz)
   

 IFT7, top coil=4.92mH capacitance 460/685pF (105 to 87KHz)

 IFT7, top coil=4.96mH capacitance 400/600pF (113 to 92KHz)
 I found that I could hear a lot of signal noise but not the modulating audio tone. The noise was loudest at the broadest selectivity settings and dropped to zero at the narrowest setting. My guess is the 110KHz IF strip is miles out of adjustment with each transformer nowhere near set to 110KHz. As the selectivity is broadened the IF strip begins to take in the skirts of the input test signal but of the audio tone very little can be resolved because this is outside the range of the detector which is swamped by noise. One stage crackled a lot and I managed to unplug its transformer. Unplugging the transformers from their octal bases is not too easy. I discovered that one of the 350pF tuning condensers had broken away and as I tried to fix it back the other end also broke away. Whether this is a weakness in the condenser I don't know, but there are lots of these fitted in the 110KHz transformer assemblies. I suppose I'll have to unplug them all to check the fixed tuning condensers and to free their seized compression trimmers. I might also need to measure all the fixed tuning condensers to check they're not putting their associated coil outside the 110KHz tuning range. See results above.

 
 

 Most of the IF transformers are fitted inside these heavy plated copper cans. The bases have three slots arranged to align the trimmers with holes in the can tops. The tops are only lightly soldered in place. One pushed out and needed refitting and a second had been refitted by a previous owner but 90 degrees out so the trimmers were partly masked by metal. It's almost impossible to rotate the cans once the pegs on the valveholder skirt are engaged with the slots in the can.

Another problem is the IF assemblies can rotate slightly as the trimmers are adjusted and in one or two instances this resulted in shorts between the assembly and the can. I need to fit a paper sleeve inside each can to prevent shorts.

Maybe the cans were easily removable in 1942 but they're not any longer. Replacement is difficult because they are now stiff to turn but needs to be done properly because some trimmers will be inaccessible if their adjusting hole isn't lined up.

 I unplugged four of the 110KHz IF transformers. Not an easy job as with the RF unit in situ (which it has to be for testing) you can't extract the can without unplugging the transformer and you can't easily unplug the transformer without first removing the screening can. Another Catch 22. The first thing you notice is the absence of a paper sleeve to prevent shorts between the assembly and the can. I should have fitted something before re-plugging them all and then to find a couple were intermittent probably due to proximity of internal bits to the can. The good news is I managed to tune nearly all the IF coils. Because of the design method of decreasing selectivity you cannot peak the coils accurately in anything other than the "Sharp" setting, and in my receiver the switch seems to have an intermittent fault in that setting. Something related to Sod's Law...At least now, as you turn the selectivity switch the noise level progressively increases instead of being random.

Midway through testing there was a loud crack and the HT meter read full-scale. Turning off the toggle switch fortunately killed the short. Anything like this with a hundred weight of receiver is difficult to sort out, but I connected a variable HT PSU after checking for an HT short and finding none. Cranking up the voltage revealed the problem. An unsleeved coax cable was touching a large carbon resistor and by merely bending the resistor leads to move it half an inch from the exposed coax braid, all was well.

   A problem waiting to happen. The uninsulated braid shorted to the 10Kohm resistor. The clue to the nature of the fault was a loud crack whenever the HT on/off switch was pressed to ON. A bad condenser would have resulted in smoke.

Aligning the 110KHz transformers proved to be difficult. The problem was twofold: the tuning range of the trimmers didn't allow peaking and there were intermittent problems. After attempting to rectify what I thought was a bad transformer module I suddenly realised my signal generator was faulty. I had the receiver tuned to 40 meter CW and this was nice and lively, but superimposed was an intermittent 110KHz signal. The penny dropped when I realised that the CW was stable. Checking the signal generator on an oscilloscope showed the 110KHz output was randomly switching off and on.

See what was involved

 Having got the Wavetech signal generator running again I turned my attention to several intermittent faults in the DST100. One of these is the bandwidth switch where the narrowest setting, designed for 1000Hz reception wasn't working unless the knob was turned slightly clockwise. Looking at the yaxley switch contacts nothing was obviously wrong with it so I sprayed a switch cleaner onto it and waggled the knob backwards and forwards. This improved things a little and by probing the contacts with a wooden knitting needle I found the rogue contact. As luck would have it, this is buried deep in the wiring but after some prodding at least the Sharp setting was usable with care. This is important because the IF tuning needs to be completed in the narrowest bandwidth range because various clever switchery places components in the IF wiring to broaden the bandwidth settings. Once I've established that all the trimmers are tuning to a peak I can use my spectrum analyser to check the IF response, but before this there are still at least two other intermittent faults to rectify. One is around the second mixer and I found a loose screw that might account for it. Another seems to be the grid lead to the VP41 which disappears under the screening box over the tuning condenser. I removed the screen which is held in place by four screws but oddly I could find no reason for the intermittent complete loss of signal. The ECH35 glass is loose as is its top cap but I don't think this is serious. I did find two of the dreaded wax covered condensers decoupling the AVC feeds. The DST100 is very unusual as it has two separate AVC lines. The feed to the VP41 read several volts of negative bias with Radio 4 tuned in but that to the ECH35 was around minus 0.8 volts.

Below the RF cover and a couple of valve screens removed. VP41 on the left, ECH35 and a pair of 6J5G triodes. A previous owner's wiring replacing the original harness between the front-end and IF/audio unit.

 

 A bit more investigation and I'm down to just some minor problems. Listening on 40m sideband I noticed that the receiver would produce a sort of warbling noise if the case was banged. I discovered that the first local oscillator, a 6J5 triode was mounted in an anti-microphonic octal holder so it could wobble slightly and this resulted in the capacitance between the valve's anode and its screening can changing with the result that the oscillation frequency changed in sympathy. I suppose two solutions are possible. Either change the 6J5 for a metal or a metallised version if such exists or to pack something between the can and the valve to stop the wobbling. A metal 6J5 might be the answer but would probably require lots of retracking so I added two strips of neoprene to jam against the valve to stop it moving. Another puzzle was that the four under-chassis trimmers for 110KHz adjustment didn't peak the signal, but this was because these are used only on the lowest frequency band.

What became apparent when I was testing reception was this set is miles away from a single-knob type. To get the best reception needs adjustment of the RF and IF gain controls and, if the signal is really weak, the reaction control. For an intercept receiver it seems to me to be pretty good. The specification includes an excellent value for image suppression and 2nd order problems, but for narrow CW reception of weak signals and probably weak SSB you'd need to use the reaction control as well as careful gain setting to balance the level of the BFO. It will be interesting to see how it performs using a spectrum analyser. This will be the next step.

It was indeed the next step and in the exercise I discovered the set had a faulty VP41 RF amplifier. I'd already detached the front-end screening can to see if there was a dry joint around the VP41 grid lead, but everything looked OK. The set was still being plagued with intermittent faults and by gently tapping the VP41 whilst looking at the S-meter it was apparent the gain was shifting maybe 10dB or more so I hunted around for a replacement. I found what looks like a new VP41 in a pre-war Murphy radio and after plugging this in the set is really transformed. The continuous crackling and gain changes have disappeared and a strange effect causing the set to behave as a narrow band FM receiver has vanished. I also swapped the 6J5G oscillator valve for a metal type because the rubber packing around the valve to stop warbling had caused the set to howl with feedback. The long wave frequency shifted a few KHz when I changed the valve because of the additional capacitance from anode to ground, but it's stable. I also stuck a paper insulator inside one of the IF cans and this stopped intermittent shorting when the rather stiff compression trimmers were adjusted. Below are some pictures of the IF response (not very good pictures as the receiver was in front of the spectrum analyser)

 

 
 Above are pictures showing the "Sharp" setting response to a 110KHz input signal connected to the IF amplifier and monitored at the grid of the detector valve. Top with the IF gain set at maximum and bottom at minimum showing a variation of about 8dB.

 

 To change the receive bandwidth the DST100 designers used a strange and expensive method where capacitors and resistors are inserted into the IF transformer circuitry by a complicated yaxley switch. Admittedly it saves on a couple of crystal filters such as are used in the R206 but the response isn't very traditional. Above is the rather strange response for setting "3", which is much the same general shape as the other settings. The shape of the curve is due to the method of tuning the IF transformers by continuous peaking at 110KHz rather than tuning each transformer at a slightly different frequency (known as stagger tuning). It's possible though that the values of specific additional components selected by the bandwidth switch were originally designed to stagger the transformer tunings, but this appears not to be the case unless the condensers have drifted miles from their original values (which is certainly the case with fixed resistors).

As the bandwidth is increased, each step adds 6dB of gain. The spectrum analyser tracking generator is set at a centre frequency of 110KHz with a sweep or span of 50KHz so each vertical division represents 5KHz. In the Sharp setting (top picture of the three above) the received signal is at 110KHz with a +/- 5 KHz bandwidth at -55dB and +/- 2KHz at -6dB.

In normal reception mode there are no less than seven independent controls affecting overall gain. These controls are RF gain, IF gain, Regeneration, IF Reaction (preset), Bandwidth, Tone and Noise Suppression. The aim I suppose being to give an intercept operator the maximum flexibility to hear very weak enemy transmissions. You can also elect to read signal strength or switch on automatic volume control (but not simultaneously both). There's an aerial peaking control and a very fine tuner for the local oscillator. Basically the set's design was almost certainly concerned with weak signal reception and to this end there's a lot of overall gain.

Before reassembling all the metalwork I may change some more of the old condensers, especially to improve the dual AVC actions.

 Alignment of this old receiver is not at all straightforward. There are several interdependent factors to consider. Because it's a double superhet there are two mixers, one of which has a fixed frequency oscillator that's preset with beehive trimmer VCZ via a hole in the side of the chassis. This is used to convert the 2MHz IF to 110KHz so must be either 2.110MHz or 1.890MHz. I think this oscillator should ideally be the first adjustment one should make once the receiver is working well enough. I set about this by using a signal generator but unfortunately my accurate Wavetech has developed another problem. It's actually the same problem I fixed earlier but must be due to a failing component whose effect I'd repaired (overheating of a voltage regulator which might have been due to a bad capacitor). I used my analogue signal generator coupled to a frequency counter to set a 2MHz test signal. With this I aligned a number of the IF transformers to precisely 2.000MHz. I noticed that the 2MHz mixer oscillator wasn't adjusted correctly so the 110KHz IF strip wasn't aligned properly to the 2MHz IF strip. The oscillator is contained in a metal can under the chassis and directly underneath the second ECH35. I tried adjusting the beehive trimmer but it was set fully in and had jammed fully closed. Because of its position one cannot pull the top free so I used a screwdriver to try and lift the top whilst turning it anti-clockwise. Eventually I had the top free but the oscillator was now no longer oscillating and all attempts to get it going failed.

 
   This is a view of the oscillator screening can. Someone has detached this in the past so they could renew some condensers. Not really an exhaustive job because there's a tubular wax covered type still buried inside the can but to change it would have been a major task so it was left.

 
   Under the metal can you'll see the oscillator coil and some capacitors and resistors. I started by testing the ECH35 thinking it may have failed during testing due to degrading emmision, however it passed at 60%. I then found a couple of bad resistors. Two 40Kohm measured 82Kohm and 70Kohm. I changed the worst which fed HT to the ECH35 triode anode but left the grid leak. Checking showed that the oscillator was still dead so I began tests with an ohm-meter. The circuit is a type of Colpitts oscillator with HT on the coil but I soon discovered no link between the coil and HT. Using a strong torch I noticed a wire underneath the coil seemed to be in free orbit and the penny dropped... whilst turning the jammed behive trimmer the screwdriver had cut the coil connection. I managed to rerieve the loose end and tin the litz wires. Reconnecting the coil proved the oscillator was now working and seemed to tune correctly.
 Whilst testing the set I'd noticed an intermittent fault whose source remained elusive but after getting the 2MHz oscillator running I tackled the wide bandwidth setting which hadn't previously appeared to work and soon located the location of the intertmittent. This was inside the can of an IF transformer primarily used in the wide setting and below you can see a dry solder joint at the trimmer condenser. The DST100 is a strange receiver because, when you switch to the Broad bandwidth position on all but the lowest frequency band the set is turned from a double to single superhet using only the 2MHz IF amplifier. Turn the bandwidth knob to other settings and the 110KHz IF strip is turned on with the DST100 now a double superhet. On the lowest band (G) the set is again a single superhet, but this time using not the 2MHz IF strip, but the 110KHz IF strip. Now that the 2MHz oscillator was running at much the right frequency I was able to switch between "Broad" and the adjacent setting "5" and hear the same incoming broadcast station and, by tweaking the 2MHz oscillator, I was able to get the main tuning almost the same for both bandwidth settings.

 

 

 Above is the IF transformer that's used in the broad bandwidth setting with the dry joint (at the lower pin of the upper trimmer).

Although plug-in IF transformers seem a jolly good idea, in practice they are very annoying. Before the assembly can be unplugged its screening can needs to be removed. Anyone that's had to remove an oil filter from a car engine will understand the difficulting in removing it. Because the plug-in assemblies are packed close together it's tricky gripping the can and twisting it off. Having done this (and including threading the top cap lead where this is fitted) the assembly needs to be very carefully unplugged to avoid damaging the coil connections. All mine had loose screws so were not very rigid and the bakelite bases are below the lip of the metalwork for securing the can. It's a question of carefully jiggling the thing and pressing on the centre pip if this is accessible (some are hidden). Once the various screws are tightened it's a lot easier refitting them. Adding paper to the inside of the cans helps to avoid shorts, but these IF transformers were the source of many intermittents. The trimmers are rather stiff and the whole assembly will twist slightly when some are adjusted, not to mention the fact that several are not easily accessible, being under the end of the case where the lifting handle is located. The 2MHz IF transformers, like the one shown above use air spaced trimmers as you can see. Those for the 110KHz transformers have solid dielectric compression trimmers and one of the pairs of each is at HT potential.

At this point I'll raise the subject of the purpose of the DST100. It was said that it was designed specifically for CW and MCW reception and not for receiving AM speech (or music for that matter). In fact the set will receive both speech and music quite well. I imagine that this is due in no small measure to degrading of the various parts used to modify the IF transformer responses. Capacitors and resistors will drift in value and this will modify their intended effects. The specification of the set was probably written to meet a particular purpose and that almost certainly for weak signal reception of transmissions not intended to be heard outside a limited range. At the time the receiver was proposed morse code was used extensively either in plain language or in code and had been since the days before WW1. Secret government organisations in the UK needed receivers like the DST100 for both eavesdropping on the enemy or for listening out for their own covert operators using low powered equipment using makeshift aerials such as the B2. Although the DST100 is essentially an Army receiver it seems to have been used by organisations reporting to the Admiralty. Another receiver used by the RN and probably replaced by the DST100 for eavesdropping was the Marconi 730.

The DST100 apparently failed to attract the "Y Service" possibly because sufficient quantities couldn't be manufactured to supply demand so the AR88 and the HRO were used instead. In my opinion, neither of these two receivers can match the overall performance of the DST100 which I suspect was quickly adopted by secret Naval communications centres listening out for enemy ships and submarines. In fact, during WW2 it was not unusual for the different military services to demand completely different equipments tailored to their own specifications. As a typical example of this see also the 76 Receiver which represents a range of equipment specified and used by the Fleet Air Arm.

  

 pending

Read about the new power supply

Return to Reception