hallicrafters sx-28a ‘super skyrider’ receiver introduction · with many of its components...

Hallicrafters SX-28A ‘Super Skyrider’ Receiver

Introduction

Around January 8, 1946, the Hallicrafters radio company manufactured Model SX-28A, Serial Number HA-30857 - just one in a very long production run of this remarkable communications receiver, the original SX-28 being announced in July, 1940, with production commencing in August, 1940. It has been estimated that around 16,850 SX-28s and 10,300 SX-28As were manufactured over the 6 years of production (Henry Rogers on the Western Historic Radio Museum website – link below). Some 65 years later, HA-30857 found itself in Ladysmith on

Vancouver Island, BC, and not in the best of sorts, having been unused for several years and with many of its components having suffered the ravages of time, probably scores of well-intentioned re-alignments (many likely inexpertly performed) and less-than-ideal storage conditions. Repair had been attempted by Bill West-Sells, a fellow CVRS1 member who resides on Vancouver Island and who has repaired/restored domestic receivers for decades, but who self-admittedly has little experience fixing

communications receivers. Bill reported to me that the set had been pulling in a couple of the strongest local stations on the Broadcast Band but was virtually ‘deaf’ on all the Shortwave bands, even after him changing-out several waxed paper capacitors and subbing some tubes. Bill therefore recommended the set’s long-time owner contact someone with more experience in repairing communications receivers of this vintage for help. After a few emails and a phone call, the set was eventually delivered to the author in Coquitlam, BC for electronic repairs and a 1 Canadian Vintage Radio Society (http://www.canadianvintageradio.com/)

Hallicrafters SX-28A Gerry O’Hara

2

cosmetic ‘touch-up’ so the owner could once again use the set for shortwave listening and, possibly also on the amateur bands, for which the set is equipped with very useful bandspread for 10m, 20m, 40m and 80m.

Homework

I had never worked on a Hallicrafters set before, let alone the famous SX-28/28A, though I had coveted one of these classic receivers for many a year. Now, I must admit up-front this model is really the only Hallicrafters I had ever liked the look of – certainly many of their post war sets appear flimsy and not particularly interesting to me. So, starting from scratch I decided to gather some background information, schematics and manuals from web-based sources and to see what the SPARC Museum had in their library. My primary reference materials included the following websites and materials downloaded from links provided on them (schematics, manuals, notes, adverts, etc):

http://www.radioblvd.com/SX28Notes.html (great notes on the SX28/28A, lots of background, serial number chronology, etc)

http://antiqueradio.org/halli12.htm - well-illustrated SX-28 restoration article

http://www.qsl.net/wa2whv/hallicrafters.html - includes restoration logs

http://antiqueradio.org/SX-28Gearbox.htm - seminal article on refurbishing an SX-28 gearbox

http://pcbunn.cacr.caltech.edu/jjb/Hallicrafters/SX28/SX-28.htm - has links to SX28/28A schematics, manuals (including the Military version) and Bill Feldman’s alignment procedure

www.tuberadio.com/sx28/sx28.html - good description and photos of dismantling the coilbox on an SX-28A – but is it really necessary? (see later in this article)

(paste the above ‘tuberadio.com’ link to your browser and check out the ‘restoration log’ link – some interesting information here)

The military version of the SX-28A manual (the military designation for the SX-28A is AN/GRR2) contains much more servicing information than the ‘civvie’ one, including tables of resistance measurements and a more complete voltage table (be careful using this though as the voltages provided for V1 through V4 are with the tubes removed from their sockets). Not all of


3

the resistance measurements make sense, for example that from the cathode of V9 to ground with the automatic noise limiter (ANL) control fully advanced should (according to the manual) read 450ohms, whereas it actually reads 33 ohms, which is correct if the schematic is studied. Also note that the schematics do not show component values and these have to be referenced from the component list, which rather annoyingly lists many components as being the same as others, so you have to then look up the cross-referenced part – what a good idea that was… (not!). It’s well worth spending a bit of time writing component values directly onto the schematic before working on the radio. There is a lot of useful material available to the would-be restorer and owner of these fine receivers – start with the links above and then you will find that Google really is your friend…

Overview of the SX-28/28A Receiver

The Hallicrafters SX-28 and SX-28A receivers are 15-tube single-conversion superheterodyne designs (nominal 455kHz IF), covering 550kHz through 42MHz in six bands:

550kHz – 1.6MHz 1.6MHz – 3.0MHz 3.0MHz – 5.8MHz 5.8MHz – 11MHz 11MHz – 21 MHz 21MHz – 42MHz

In addition, an electronic bandspread facility is provided in a separate scale for the following frequency ranges, covering the 80m, 40m, 20m and 10m amateur bands:

3.5MHz – 4.0MHz 7.0MHz – 7.3MHz 14MHz – 14.45MHz 28MHz – 30MHz

The set is designed to receive AM and CW signals, but with care, SSB signals can be received quite well. Controls include main and bandspread tuning, a six-position selectivity control (three positions with the single crystal filter in-circuit), crystal phasing, ANL, BFO, AGC on/off, antenna tune, RF gain, AF gain, tone, a ‘bass boost’ switch and a standby switch.


4

Referring to the block diagram above, the circuit comprises two pentode RF stages (6AB7 and 6SK7 as pre-selectors), the first stage only used on frequencies above 3MHz. A 6SA7 ‘pentagrid’ (heptode) carries out mixer duty, and a second 6SA7 acts as the local oscillator. Two stages of IF amplification are provided, a 6L7 heptode followed by a 6SK7, feeding a 6B8 dual-diode pentode as detector/AGC/S-meter amplifier. Audio from the detector is fed to a 6SC7 dual triode doing duty as a pre-amplifier and phase-splitter, to two 6V6 output tubes in push-pull, giving 8W of audio. A second 6B8 is used as an AGC amplifier/detector for the RF and mixer stages, a second 6AB7 as the ANL noise amplifier and a 6H6 dual diode as the ANL rectifiers (see below). A 6J5 triode is used in the BFO circuit, which is loosely-coupled to the detector circuit with a ‘gimmick’ capacitor. The power supply is conventional, using a 5Z3 rectifier. The circuit includes some interesting innovations: there are two AGC circuits – one amplified, feeding the RF and mixer stages, and the other conventional, controlling the first IF stage. There are also two ANL circuits: the first provides for shunting of the detector diode load with a capacitor to momentarily bypass audio signals on noise peaks, the second amplifies the noise signal at the IF frequency and places an instantaneous negative potential on the 3rd grid of the first IF amplifier (a 6L7 heptode) at a level selected by the operator thus canceling out noise


5

pulses – this is a version of the Lamb2 noise silencer (‘blanker’). The use of permeability-tuned RF transformers using iron-dust slugs was a fairly new innovation at the time, as was the variable-bandwidth IF and single crystal gate circuitry. Even the S-meter circuit is a little novel having its own amplifier stage, and the audio output impressively has a pair of 6V6 tubes in push-pull for great sounding audio, complete with a switchable bass boost circuit. The receiver is very solidly constructed on a stout steel chassis, weighing-in at 78lbs. The thick steel front panel is separate to the main chassis, with some controls being mounted directly to this. The RF circuits are contained in a series of screened compartments beneath the chassis, reducing heat build-up here, however, a temperature-compensating capacitor is included, this being a negative-temperature coefficient ceramic trimmer located in the tuning gang compartment above the chassis where the four RF tubes are located. The gearbox in the SX-28 uses gears to drive both the main and bandspread gangs, however, the SX-28A uses a dial cord to drive the bandspread gang. Cords are also used to operate band indicators on each of the two tuning dials on both models when the bandswitch is operated. Some other differences between the SX-28 and SX-28A include:

- Construction of the RF circuits: in the SX-28A, ‘Hi-Q Micro Set’ coils with phenolic formers were mounted on Paxolin plates, whereas in the SX-28, the coil units were mounted onto the sub-chassis;

- The coil box cover is screwed in place in the SX-28 and is held on with retaining clips in the SX-28A, the cover in this model also having a louvered top;

- The SX-28 has open ‘spoke’-style tuning knobs and the SX-28A was fitted with closed-in ‘webbed’- style tuning knobs.

The SX-28 was introduced in August, 1940 and stayed in production through 1944, with production then swapping to the SX-28A from late-1944/February 1945 through to July 1946, after an estimated total of 27,150 SX-28/SX28A’s were manufactured. The receiver was widely used by the military in WWII and in the post-war period quite rightly became a very sought-after and popular receiver within the amateur community.

2 Lamb, J.J. ‘Noise‐Silencing IF Circuit for Superhet Receivers’, QST, February, 1936, P11, and ‘More Developments in the Noise‐Silencing IF Circuit’, QST, April 1936, pp 16‐19, 78‐86. Also described in US Patent 2,101,549, December 7, 1937.


6

On Arrival – First Checks

Overall, this receiver looked a little forlorn on arrival – like it had not been used for a long time (photo, right). However, it was almost complete, just lacking a couple of original knobs, sporting a mongrel aluminum one (National) on the bandchange switch and a 1960’s skirted variety on the crystal phasing control. It was installed in the correct tabletop cabinet with that distinct and attractive art-deco aluminum strip around the upper shoulders (though 1940 was a little late for that styling?). Underneath, a galvanized steel sheet

had been made to fit the cut-out allowing access to the coilbox without removal of the chassis from the cabinet. The lid of the cabinet is original and fits reasonably well, allowing access to the upper chassis for easy tube replacement. The upper chassis (photo, left) was mildly corroded but was in pretty good overall shape.


7

The set came with a Hallicrafters Model R-46 speaker which is fitted with a 500ohm matching transformer (the SX-28A does not have a low-impedance speaker output, only 500ohm and 5000ohm line outputs). I noted that the bandspread dial was not working due to the bandspread dial cord being absent and that the band indicator mechanism was missing on both the main and bandspread dials. I checked the mains fuse in the fuseholder on the rear apron of the set and found it was fitted with a 10A cartridge (not affording the set a lot of protection!) – I replaced it with a 2A unit and also installed a spare fuse into the clip on top of the chassis.

The receiver came with a spare 6B8 tube (second detector and S-Meter amp), with a note from Bill West-Sells that the one fitted in the chassis was not working although it ‘looked like new’, plus a bagful of wax-paper capacitors that Bill had replaced. On inspecting the underside (photos, above and top of next page), I noted that the replaced capacitors were a mixture of types, manufactures and voltage ratings, many on extended leads. I decided to replace all of these and all remaining waxed-paper capacitors to promote long-term reliability. The electrolytic capacitors in the power supply filter and audio amplifier looked to be the original metal can types. I spot-checked several resistors at random and most tested were found to be close to or within tolerance. I noted that a few resistors had been changed-out already for more


8

modern types. I next checked that all the tubes were in their correct sockets and were of the correct type (often they are not) – indeed, I found that the RF tube fitted in the set was a 6SK7 and not the specified 6AB7, so I swapped that out.

Bill had noted that the set was (just) working, so I decided to power it up at this point and see what happened… well, it did power up, dial lights working, and with no noticeable hum. The heater supply checked out at 6.3vAC and the HT line was at 270vDC, which seemed ok. As Bill noted though, only very strong local stations could be heard (very faintly) at full-volume on the Broadcast Band and nothing at all was heard on the Shortwave bands, with the S-meter not deflecting at all. I thought that it may just be an alignment problem, but decided to replace the wax-paper capacitors and undertake some voltage and resistance checks first, it being likely that Bill (or others before him) had already thought the same and had ‘had a go’ at re-aligning, if so, obviously to no avail.

On peeking into the coilbox (photo, left), I noted that the slots in several of the tuning slugs looked ‘well-worn’ and the upper surface of the ceramic trimmers were coated in a grey (metal-oxide?) deposit - both to be expected after over 65 years of service, mostly in the hands of well-meaning radio amateurs. I gave the below-chassis areas a good clean with a paintbrush and blasts or compressed air, and then got to work…


9

Preliminary Electronic Repairs and Re-alignment

I started by replacing all the waxed-paper capacitors – some 43 in total in this set, including all the ones Bill had already replaced. All of the original capacitors removed tested leaky and/or were out of tolerance. The capacitors in the main chassis areas (IF strip, audio, AGC, ANL) were easy-enough to replace (photo, right), then came the RF section…

The coilbox in the SX28 and SX28A differ significantly, but both pose access problems for the service engineer. The SX-28A has the coils mounted on Paxolin boards, these mounted on brackets fixed to the chassis. These Paxolin boards and the web of wiring around the bandchange switch wafers limit access to several capacitors and resistors in the RF, Mixer and Local Oscillator stages of the receiver, with several articles suggesting that the only way to replace these parts is to dismantle and remove the coilbox sub-chassis from the set. To do this requires much dismantling/wire disconnects, potentially introducing mechanical and electrical problems when re-installing these sub-chassis. I decided to attempt to service the RF units without removing them from the main chassis and found it to be fairly straightforward with a little care, patience and the right tools. Here’s how:

First, loosen the grub screw from the coupling onto the antenna trimmer, pull the antenna trimmer (phenolic) shaft out from the front panel (this needed a little lithium grease on this set to ease its passage). I started in the 2nd RF stage section - looking from the front of the set, that’s the third compartment from the front of the set. On the left-hand side of this compartment, unsolder the wire that comes up from under the Paxolin board to T15. Remove the two screws that fasten the antenna tuning capacitor in place and temporarily push the capacitor back into the 1st RF compartment. Remove the three screws holding the Paxolin board onto its mounting bracket and ease-up the edge of the Paxolin board farthest from the bandspread switch. Clean dust and fluff out from beneath the Paxolin board and replace the waxed-paper capacitor beneath (photo above). Also check the


10

three resistors present and if out of tolerance, replace as necessary (photo, below). Gently push the Paxolin board back into position, replace the screws, replace the antenna trimmer capacitor screws and re-solder the wire to T15. Simple eh? Then move onto the right-hand Paxolin board in the same compartment. This time, unsolder the three wires that enter the compartment through the large grommet from the 1st RF stage compartment, plus the wire that sneaks around the side of the board near the bandchange switch. Again, remove the three screws that hold the board in place and ease it up gently,

replacing the waxed-paper capacitor that lies beneath (this capacitor was actually missing from this set! – only the grounding solder tag remained, photo, below). Gently push the Paxolin board back into position, replace the screws and re-solder the four wires. Next replace the waxed-paper capacitors located near the bandswitch wafers in this compartment. To do this needs a little ‘surgical skill’ and

the right tools – nothing special, just the right ones, some patience and dexterity. I use a good pair of angled tweezers, two pairs of long-nosed pliers, a hemostat, two pairs of snips (one with a long ‘neck’), soldering tools (pointed and hooked ends) and a slim Antex 15 Watt soldering iron. First gently move any wires obscuring access as best you can (temporarily unsolder if necessary, though I did not have to), cut out the old capacitor leads, clean and tin the connection points where the capacitor attaches to, carefully pre-dress the leads on the new part, tin the leads, leaving a small solder ‘blob’ on the end, hold in place with the tweezers and tack-solder the part in place with the slim soldering iron. While you have access to the tube socket, check the resistors present – if out of tolerance, replace as necessary using a similar technique. Once the new component(s) is in place, re-solder the leads (if necessary), then re-dress/attach any leads you moved/


11

disconnected. Then replace the waxed-paper capacitor attached to T18. That’s this compartment completed.

I suggest you then move onto the Mixer stage compartment (second from the front of the set). A similar process applies. Start with lifting the left-hand Paxolin board, again removing the wire connecting from underneath the board (to T21). This reveals two resistors (photo, right) – check and replace if out of tolerance. Gently push the Paxolin board back into position, replace the screws and re-solder the wire. Move onto the

right-hand board. No wires to unsolder this time, just the three screws to remove and then pry the board up to reveal a single waxed-paper capacitor – replace (photo, below). Gently push the Paxolin board back into position and replace the screws. Next replace the two waxed-paper capacitors located near the bandswitch wafers in this compartment using similar techniques and approach as for the 2nd RF stage. Again check the resistors located on the tube socket and replace if out of tolerance.

Next, move on to the Local Oscillator (located nearest the front of the set): there are no waxed-paper capacitors in this compartment, however, it does contain one high-wattage (2W) resistor that is worth checking. This is R6 and it is revealed by lifting the right-hand Paxolin board (photo, below). This entails temporarily removing a single wire that sneaks around the far side of the board and joins to T30. The 6.8kohm resistor measured over 9kohm in in this set and was therefore replaced

by a new 2W part. The associated capacitors (C70 and C86) are silver mica types (and should be ok) but check them anyway while you are in there, plus resistor, R47.

Gently push the Paxolin board back into position, replace the screws and


12

re-solder the wire onto T30. Don’t bother to lift up the left hand board in this compartment as only wires run beneath it. Again check the resistors located on the tube socket and replace if out of tolerance. It is also worth checking the two silver mica capacitors (C68 and C69), the former connects between the grid of V4 and a connection to the bandswitch (wafer nearest the front side of the compartment) and the latter connects between the cathode of V4 and Pin 5 of V3 in the Local Oscillator compartment. C68 was found to be out of tolerance in this set and was replaced.

Finally, now you have perfected your dental surgery techniques, it’s time to tackle the 1st RF stage (photo right). This compartment does not contain any Paxolin boards, the coils being mounted on metal brackets fixed to the chassis. There are several parts located beneath these brackets that should be replaced/checked. This time, looking from the rear of the set, look closely into right-hand side of the 1st RF compartment (the side with the solder connections coming in from the side). Here are located R1, R3 and C14. With some care, these can all be checked and replaced. Then, take a look at the left-hand side of this compartment (the side with the jack socket located on the rear panel). Remove the jack socket to make a little more space for access. There are two waxed-paper capacitors (C15 and C77) and a 2W resistor (R4) hidden beneath the metal bracket here. I disconnected the capacitor connections to the tube socket (C15) and bandswitch wafer (C77) and left the ground connections alone, tucking the old capacitors well-away so they could not contact anything (ie. I physically left them in place but with only their ground ends connected, thus effectively out of circuit). I checked the resistor (R4), which was within tolerance, and replaced both capacitors. I also found that the insulation on the screen grid wire running from R4 to the two 2.7kohm resistors located in the power supply compartment was missing in parts and so this wire was replaced with new stock cloth-covered wire. The space available for all these checks and connections


13

is very limited, so this can take a little time and benefits from careful pre-dressing of the new component leads.

So, that is the coilbox capacitor and resistor service completed (photo, above) – not quite as bad as some folks seem to think eh? Anyway, I decided to power-up the set again at this point – disappointingly not much difference to before to be honest – a few more stations present on the


14

Broadcast Band and a little stronger, but still dead as a Dodo on the Shortwave bands. Time to see what a little preliminary re-alignment would do.

I followed the procedure for IF re-alignment as described in Bill Feldman’s notes rather than that in either the regular of military manual versions. This basically comprises undertaking a ‘rough’ alignment of the IF stages at 455kHz, followed by finding the resonant frequency of the crystal filter, re-aligning to that frequency, setting-up the crystal filter for optimum performance (peak and null) for each of the bandwidth settings – all pretty standard for a receiver with a single crystal filter, then setting-up the RF stage AGC and finally the (Lamb) ANL circuit.

Although more complicated than for most single-conversion receivers, the above would be a fairly straightforward process if everything was working as it should – I soon came to realize that things were not working as they should in this set(!). I could get the first and second IF transformers (T1 and T2) to peak at 455kHz, but the third transformer (T3) refused to peak. I decided to investigate T3, which meant disconnecting it and removing it from the chassis. This necessitates temporarily disconnecting a couple of component leads that obscure the transformer connections – also, make sure you know how to re-connect it if you do this (I suggest that you take a photo or two of the leads as they pass through the chassis as a couple of them cross over). The schematic and text in the Military SX-28A manual3 indicated that the transformer is tuned by two ceramic trimmers (C108/C109), however, I also found a couple of silver mica capacitors wired in parallel with the trimmers inside the transformer can (photo, above – with the old silver mica capacitors removed). I checked these and found both to be significantly out of tolerance, so I replaced them with new capacitors of the value as the old capacitors were marked at (52pF). On re-installing the transformer, I found I could now tune-up T3, though not as satisfactorily as I would have expected (best described as a little ‘jumpy’) – anyway, I left it for now. I then tried to find the

3 The ‘civvie’ manual schematic also indicates two trimmers, however, the text description of the circuit in the ‘civvie’ manual indicates that the transformer is tuned by a combination of ‘lump’ capacitors and trimmers


15

resonant frequency of the crystal per Bill Feldman’s instructions. I soon realized that the crystal phasing control was not working, so I removed the variable capacitor from the front panel to find that the stator and rotor plates were shorting (photo, right) – I carefully cleaned the trimmer with Deoxit, maneuvered the plates into position and checked for correct operation before re-installing on the front panel, wondering how on earth its plates had become so bent? Recommencing setting-up the crystal, I found the resonant frequency to be 458.500kHz (a small secondary peak at 456.350 kHz, a null at 457.680kHz), and the upper slug of T2 re-adjusted for the deepest null between the peaks (while retaining both peaks) – this takes some time to get right. C30 and C29 were then adjusted for peaks in the ‘Xtal Sharp’ and ‘Xtal Mid’ selectivity settings respectively. That completed, I then set-up the RF stage AGC amplifier/detector by adjusting T6 – this peaked quite nicely, with plenty of AGC voltage present to the RF stages.

Setting-up the ANL was next and this proved rather problematic – no response was obtained when trying to adjust T5 per the instructions. I eventually resorted to some resistance checks around the ANL amplifier/detector tube, V9, finding that the cathode resistance to ground was infinity (the Military manual indicated 450ohms with the ANL control switched in and fully advanced). I traced this problem to one half of the DPST switch (S5) ganged to the ANL control being intermittent in operation. I removed the control from the receiver, then, not wishing to disassemble the unit unless I really had to, I squirted a liberal dose of Deoxit into the switch and operated it several times (repeating this operation several times more) – the switch soon regained reliable functionality and was replaced into the receiver. I then found that V9 cathode to ground now measured 33ohms with the ANL control fully advanced, which makes sense, as only R54, the V9 cathode resistor, is present (the Military manual indicates this should read 450ohms – I have no idea why). On re-checking operation of the ANL, I still could not peak T5, though the ‘scope showed that V9 was now definitely amplifying the IF signal applied to its grid. Finally I decided to check-out the compression trimmers in T5 (photo, right). This


16

requires the screening can to be removed, which is a little awkward as the fixing nuts are difficult to get a box-spanner onto and requires the temporary removal of a couple of component wires that obscure access. With the screening can removed, the compression trimmers appeared to be in good shape, albeit dirty, however, the adjustment screws were badly corroded. I disassembled the trimmers and cleaned them up thoroughly, checked the inductors for continuity and looked for any dry joints – all seemed in order. I re-checked operation of the ANL without replacing the screening can and both trimmers on T5 tuned up nicely – must have been poor connections on the adjustment screws. I re-fitted the screening can and re-checked operation again – all was working fine.

With the IF stages, AGC and ANL circuits now working reasonably well, I ran through the RF alignment per the manual instructions (table). Some of the spot frequencies used for tracking adjustments seem a little odd, not being close to the ends of the tuning bands, eg. 7MHz and 10MHz on the 5.8MHz to 11MHz band, still, I used the suggested frequencies. One point not made very clear in the manual is to set the Bandspread dial at ‘100’ on its logging scale (bandspread capacitor plates fully open) before alignment is attempted, otherwise it will not be possible to obtain tracking across the upper part of the Shortwave bands. Before I started, I inspected the iron-dust tuning slugs – they are a type that has coarse threads and a vertical groove(s) into which is inserted a piece(s) of rubber or plastic sleeve to prevent vibration moving the slug in the coil former. All of the rubber/plastic sleeves had hardened with age and were removed, a light coat of Rocol Kilopoise being smeared on the slugs instead for this purpose. While the slugs were removed (one at a time), the threads on the slugs and the coil formers were cleaned using a brush and compressed air. I found that when I obtained tracking at the lower end of the Shortwave bands, I could not reliably obtain tracking at the upper end. On inspecting the trimmers on the bands concerned, I noticed that the upper ceramic surface was smeared with a metallic sheen. I painstakingly cleaned this off each trimmer using a Q-Tip and Deoxit. Doing this cured some, but not all, of this strange effect. The set is fitted with a padder (compression trimmer) on each range that the manual says are ‘fixed padders’ with a warning of ‘do not touch’ (and therefore does not explain how to set them up!). I guessed that someone had probably twiddled these in the past 65 years, so I included them in my alignment procedure – a little trial and error is needed: fully screw in the padder, adjust the coil slug at the low-end marker, adjust the trimmer at the high end marker. If I could not obtain good tracking, loosen the padder slightly, re-adjust the slug at the low end marker and the trimmer at the high end marker, repeating this on all ranges until good alignment could be obtained. The set was now


17

roughly in alignment and pulling in stations reasonably well on the Broadcast Band plus some stations and noise on the Shortwave bands, though the set was not as sensitive as I would have expected and the S-meter was barely moving past S2 even on a very strong local signal.

At this point I thought I would take some light-relief from the pesky electronics and undertake some mechanical work and clean-up the front panel…

Clean-up and Mechanical Repairs

The front panel is a cinch to remove: simply take off all the knobs (none of the grubscrews were seized on this set – a real bonus!), remove all the control retaining nuts and washers, disconnect the two wires from the S-meter, remove the two screws holding the panel to the chassis and pull the panel away. With the panel separated from the chassis, I removed the S-meter and the three escutcheons from the panel (careful not to let the glass fall out of these) – photo, below. I

scrubbed the panel with hot soapy water and then used Novus #2 and then #1 to remove the oxidized top layer of paint on the ‘alligator’ finish and polish the panel. I also cleaned the escutcheons with Novus and the glass inserts with lens cleaner, finally re-assembling the parts onto the panel.

While removed from the panel, I also dismantled the S-meter, cleaned the inside of its case and checked for any dust in the movement – all was clear. I also checked that the meter was working by applying a small current from a battery through a resistor – it seemed fine (note that the meter is a reverse-acting meter). I cleaned the meter ‘glass’ (thin Perspex) with antistatic fluid and then re-assembled the meter and escutcheons to the panel. I washed all the knobs in hot soapy water and then cleaned them with Novus #2 and #1. While the front panel was off, I replaced the wax-paper capacitors located on the RF gain and tone controls (easier access). I installed connection eyelets onto the S-meter leads as these had simply been wrapped around the binding posts on the meter – not very satisfactory.


18

Removing the front panel affords good access to the tuning and bandspread dials, the gearbox and the bandspread dial cord area (photo, above). It was noted previously that the band indicator shutter mechanisms were missing from both of the dials in this set – however, they had been fitted at some time in the past as the pulleys for these cords were still present on the gearbox casing. I removed each of the tuning dials and cleaned by gently brushing away dust using a paintbrush and then using lukewarm, slightly soapy water (testing a corner of the lettering to check it was not water-soluble – it was ok on this set). I then re-strung the bandspread dial using 0.03” dial cord brushed lightly with rosin powder. The dials were replaced and zero set on each. The gearbox was working reasonably well, but a little stiff, however, not sufficient to warrant disassembly of the gearbox (see article by Doug Moore on ‘Phils Old Radios’ website, http://antiqueradio.org/SX-28Gearbox.htm if you need to do this). The gears and shafts were cleaned using lighter fluid and Q-tips and a thin smear of lithium grease applied to the gears. Light machine oil was applied to the non-roller bearings. At this stage the upper chassis surface and gearbox casing were cleaned using lighter fluid and alcohol. The re-assembled front panel was then refitted to the chassis, the knobs replaced and the antenna trimmer shaft re-installed.

Next, all the Yaxley switch wafers were cleaned thoroughly using Deoxit spray and Q-tips (photo, right), and then a spot of red Deoxit fluid applied and the switched operated several times. A squirt of Deoxit spray was also applied to the various control pots.


19

More Electronic Repairs and Final Re-alignment

Having tidied-up the front panel and chassis, and with the bandspread dial now working properly, it was time to re-check some of the issues I had come across in the initial checks and alignment. The recalcitrant S-meter was annoying me, so I removed it from the circuit and replaced it with an external 5mA meter – it behaved the same, so the meter movement seemed to be ok. The meter effectively measures the anode current of the pentode section of the 6B8 (S-meter amplifier), the grid of which is biased negative by the rectified IF signal from T3 (the final IF transformer) – as the grid becomes increasingly negative with a stronger signal, then the tube starts to cut off and reduces the anode current. I had read that the zero adjust pot can give trouble in this circuit, so I removed it, pried-off the back cover and gave it a good clean with Deoxit – it then worked well. After replacing the pot, I checked the 100ohm resistor that is in parallel with the meter/zero adjust pot and that was ok, however the 27kohm 2W resistor was way out of tolerance and the carbon element was found to be loose in its casing (never a good sign), so I changed it out for a new, high-stability 2W part. A little improvement – at least I was now able to adjust the zero reliably. At this point I became suspicious of T3 again, as I tweaked it slightly to see if I could obtain more deflection on the S-meter it started to behave erratically. So out came T3 again – inspection this time revealed the next problem: the soldered joint between the adjusting screw and the rotor had split on both ceramic trimmers (see arrow on photo, right), making intermittent and generally poor connection. The trimmers were removed, cleaned-up and the joints re-soldered. After fitting T3 again, the transformer could be peaked nicely and the S-meter started to operate much better (phew!), though I could only obtain around S9+5 on a very strong signal from my generator. I decided to start checking things in the IF stages. There was no real ‘smoking gun’ evident from voltage and resistance checks, but I decided to change some of the more critical resistors out where they were of marginal tolerance, paying particular attention to the AGC function – all for only a minor improvement in S-meter performance.

Next I thought I would try swapping out a few tubes for some NOS ones from the SPARC stock to see if any improvement could be had – fitting the 6SK7 that was previously fitted in the 1st RF stage in place of the 6SK7G (glass envelope) in the 2nd RF amp perked things up a bit, as did a replacement 6SA7 in the mixer stage. I tried subbing a 6AC7 for the 6AB7 I had fitted in the 1st RF stage (this is a well-publicized ‘mod’ for the SX-28/28A receivers, the 6AC7 having higher


20

transconductance than the 6AB7) – this substitution increased the S-meter readings by over a full S-point at 10MHz without any sign of instability on any band, however, the 6AC7 is a sharp cut-off tube and I therefore suspected that the AGC on the 1st RF stage may not operate satisfactorily. I also tried subbing another 6AB7 for the 6SK7 in the 2nd RF stage (per the final factory mod listed on the SX-28 Notes website), the 6AB7 having a higher transconductance than the 6SK7, thinking this would increased the RF gain while retaining the AGC action (the 6AB7 is listed as a semi-remote cut-off tube, the 6SK7 as remote cut-off) – it did not improve the gain or AGC action, so I re-installed the 6SK7.

Having found what appeared to be the optimal tube compliment in the RF and IF stages, I ran through the alignment process again, this time taking much more time and care at every stage. Following this process, I measured the RF (amplified) AGC - this was giving around -27vDC on a very strong signal, and the IF AGC circuit was providing around -11vDC on a similar strength signal – this seamed healthy. The set was tracking well on all ranges and sensitivity was now more reasonable. Further tweaks were undertaken to the ANL circuit (peaking T5 and re-adjusting the IF wavetrap in this circuit), as well as the crystal filter. Although this alignment procedure worked reasonably well, I decided to use a wobbulator for final set-up of the IF stages. This technique is invaluable for adjusting sets with variable bandwidth and crystal filter controls such as the SX-28A. Sure enough, the wobbulator revealed an asymmetrical response on the wide and medium bandwidth settings that was easily corrected visually by slight tweaking of T1 and T3, giving a reasonably flat-topped response curve on each of these (photos, below - before and after tweaking T3 on the widest bandwidth setting), and a nice, symmetrical single peak on the narrowest (non crystal) setting. The crystal filter was also re-adjusted with the wobbulator set-up, giving a very desirable sharp peak and adjustable deep null in the receiver passband. Lastly the BFO was adjusted to centre on the crystal resonant frequency.

Above, left: widest bandwidth setting after conventional alignment process (per the manual). Above, right: widest bandwidth setting after set-up using the wobbulator to obtain a symmetrical flat-topped response curve


21

In Use and Yet More Electronic Restoration…

With the final alignment completed, I started to use the set – it performed quite well: first, WWV was tuned in on each of 5, 10 and 15MHz and the set sat on frequency with minimal drift after initial warm-up. I then tuned around the Broadcast Band – stations were romping home and the audio quality (and volume) were found to be excellent. On the Shortwave broadcast bands, the set was pulling in many stations. I tried the set on the 20m amateur band and with the audio full-up and the RF gain wound down, I was able to listen to several SSB stations – the bandspread dial is really useful for the amateur bands (just remember to set the main dial to the little round marker above the band you wish to tune first), and may also be used as an un-calibrated vernier tuning dial on other parts of the Shortwave bands – very useful at the higher frequencies. I also checked-out the crystal filter and found it was working well – able to null unwanted carriers adjacent to wanted stations on the wide and medium crystal settings and to peak for ‘single signal’ reception of CW on the narrow crystal setting. I did not have cause to use the ANL, but the circuit was working well so I assumed it would work if

Above, left: minimum non-crystal bandwidth setting after set-up using the wobbulator. Above, right: medium crystal filter bandwidth setting – note deep notch and sharp resonant peak


22

the need arose. Altogether a good experience – and all on a few feet of wire from the eaves of my house to the basement. However, after a few of evenings of coasting around the bands over the Christmas holidays, I noticed that AM stations were starting to sound distorted and the sensitivity was beginning to drop off again/S-meter becoming less responsive. I decided to investigate…

I started with the IF stages and checking the AGC action – I found that some of the resistors that had previously checked within tolerance had drifted out of tolerance during the extended soak testing (this is a good reason to do this test!), so I decided to replace any resistor that was now not within around 5% of its stated value or even marginally within tolerance. This resulted in quite a number of substitutions and for only a slight improvement in performance (though likely quite an improvement in long-term reliability). The AGC action on the RF stages was investigated further and I found that on the Broadcast Band (Range 1), where a much higher level of AGC is applied to the RF stages on Range 1 only: on a strong signal over -25vDC of AGC was being generated, limiting the S-meter action to S9+5db no matter how strong a signal was applied. In order for the S-meter to read S9+40 on a very strong signal on this range, I found that the amplified AGC to the RF stages needed to be limited to around -16vDC. I checked the components around the AGC amplifier – the 250pf coupling capacitor to the grid of V9 was ok, however, the grid bias resistor was reading around 700kohms (it should be 470kohms), though correcting this by temporarily paralleling with a 1Mohm resistor did not change things. The grid bias resistor for V9 is located in the can of T1 and requires the can to be removed to be changed-out. With the possibility of damage to T1 in the removal process and the fact that the high value of the grid resistor was not affecting performance of the AGC, I opted not to change this resistor. I then decided to try a voltage divider on the feed to the RF stage grids on Band 1, using a couple of 470kohm resistors (circled in photo, right) – this succeeded in providing the receiver with optimal AGC and S-meter action on Band 1 (and can be easily reversed if originality is desired). However, the audio distortion was still evident…

I therefore decided to check-out the AF amplifier stage – previously I had only changed out the (paper) grid coupling


23

capacitors to the output tubes. The audio amplifier comprises a 6AC7 dual-triode, one half serving as a pre-amplifier, the second as a phase-splitter feeding the push-pull 6V6GT ‘beam-power’ tetrode output tubes. To gain access to this stage, the choke (Ch2, in the bass-boost circuit) can be removed quite easily. Again, several resistors were found to have drifted out of tolerance and were replaced. The electrolytic capacitors were also checked and C41 (40uF cathode by-pass on the 6SC7 twin-triode) was found to be very leaky and almost zero capacitance, and the 10uF anode decoupler for the 6SC7, while measuring 11uF with a surprisingly low ESR, had high leakage. These were replaced with new electrolytics, wired directly to the tube sockets. I also checked C47 (40uF cathode by-pass capacitor on the 6V6GT output stage) and found it to be within tolerance and with a low ESR, though somewhat leaky. Although a leaky capacitor in this location is not important (it is in parallel with a 220ohm resistor), it was changed-out anyway to increase reliability.

The power supply filter capacitors (30uF each) seemed to be working fine, there being little hum and with a HT voltage only some 30v or so below that indicated in the manual. However, I contacted the owner and agreed to also change these out as a precaution, as failure of these can lead to costly problems, such as the power transformer burning out. I used a twin 32uF 500vDC working can-type capacitor as the replacement part (photo, right of the repaired audio/power supply section). On removal of the two old filter capacitors, I found one (C48, the smoothing capacitor), was almost perfect at 31uF with over 200kohm leakage resistance and an ESR below 1ohm. However, C49 (reservoir capacitor) measured 1.39nF (yes nF), though with a small leakage at 950kohm - my PEAK ESR meter could not detect a component present - amazing that the set worked as well as it did with this capacitor in this state! The anodes of the 6V6GT output tubes are fed from the reservoir capacitor and this


24

acts as both as an audio bypass and regulates the voltage during peaks in current draw – without it working the result was causing the bulk of the distortion I was hearing, exasperated by no decoupling of the driver tube cathode. After these capacitors were replaced the HT voltage was now per the manual and the remnant distortion had gone – what great (and powerful) audio!

Looking at the underside of the chassis I could now see only 6 x 1W resistors that had not been changed-out (as they were within 5% tolerance). I decided that I might as well change these for aesthetics if nothing else! I also decided to re-check the out-of-sight resistors in the RF compartment, changing-out any that were not well within 5% (5 were changed).

For the record, by the end of this second episode of passive component replacement , a total of 55 resistors and 49 capacitors had been changed-out in the set (photo, above).

After another extensive soak test (several evenings and a weekend) and everything holding out well this time, I tried changing a few of the RF-IF tubes again, including changing the 6AC7

Above: under-chassis after all component replacement work was completed


25

tube in the 1st RF amplifier position back to a 6AB7, but no improvement could be effected, so I changed them back. Strangely, although the 6AC7 tube is listed as a sharp cut-off type, it seems to work just fine and is as responsive to the AGC action as the 6AB7. I also undertook one final re-alignment using the wobbulator on the IF, boxed the set up and soak tested it again – enjoying the bandcruising and the superb audio quality.

Speaker

The Model R46 speaker that accompanied the set (not a contemporary model) looked a little ‘worse for wear’ – the paintwork had several scratches, score and ring-marks (underlain by rust). Also, the glass-fiber flocked speaker grill cloth was badly stained, all detracting from the appearance of the set when next to the speaker. In follow-up correspondence with the owner, he requested that I also refinish the speaker.

The original grill ‘cloth’ is not really a cloth, but a glass-fibre flock-covered metal mesh in a buff colour. Unfortunately, a combination of nicotine, something that had been spilled on it years ago and part of the flocking being damaged, meant that a simple re-colouring of the flocking would not look good – at least not in a buff colour. Three choices remained – re-flock, find a suitable cloth (not original-looking but better than the stained mess of the current one) or re-finish in a darker colour. I decided to try to retain the green ‘Hallicrafters Co.’ and ‘Model R46’ number decals on the front of the speaker cabinet by masking over each of the letters prior to spraying, then removing the masking tape and then using a fine brush to touch-up with the paint between the letters. When painted (Plastikote ‘charcoal grey pearl’), the whole cabinet was given a couple of coats of clearcoat to give the same finish to all sides/protect the decals. The speaker grill was given several light coats of satin-finish black spray paint, hiding the mess underneath

and effecting a slight contrast to the cabinet finish, in particular the Hallicrafters ‘h’.

Closure

The SX-28/28A receivers are a classic in many ways – from the art deco, symmetrical styling of the front panel to the solid construction and innovative circuit design within. My only gripe would be that with a little more thought the coilbox could have been much easier to service (replace capacitors and resistors), especially in the 1st RF section, and with better access to the bases of the RF tubes for voltage measurements. Still, the component replacement task in the RF compartments was not as bad as some folks have indicated and should only be necessary very rarely once quality modern components have been fitted – likely not in our lifetime anyway. As I reluctantly boxed the set up for the owner, I knew I had to own one of these sets one day to add to my modest collection of ‘favourite communications receivers’ of the tube era, which currently include an RCA AR77 and AR88LF, RACAL RA-117, Collins 75A-2, Hammarlund SP600 and HQ180, National HRO and NC-183, over 20 Eddystone sets… and counting!

Gerry O’Hara, VE7GUH, Vancouver, January, 2012


26


27

Above: S-Meter connections re-made using eyelets. Below: IF chassis after final round of replacing resistors and paper capacitors. The S-Meter zero pot is on the right – this was removed, its back popped-off and cleaned with Deoxit (inset photo)


28

Above, left: checking electrolytics – here C49 shown measuring 1.39nF (should be 30uF). Above, right: corrosion on the underside of one of the (T3) ceramic trimmer before cleaning. Below: last IF stage and detector/AGC amp area of the chassis after replacing paper capacitors and resistors


29

Above: refinishing the speaker unit. It was sanded down using 400 grit wet and dry, cleaned using lacquer thinners, a thin coat of primer applied, then 3 coats of charcoal grey pearl auto spray paint followed by two coats of clear-coat. Prior to spraying, the decals on the front of the speaker were masked-over very carefully, then the masking tape was removed and a fine brush was used to touch-up with the paint between the letters before the clear-coat was applied. The overall effect is reasonable and certainly an improvement on what it was. Below: re-finished speaker with the original flocked grill re-painted satin black to hide the nicotine staining


30

Above: the SX-28A receiver with R-46 speaker on the workbench in the SPARC museum awaiting pick-up by the owner – yet another set that I just did not want to part with! Note the replacement ‘Ant Trimmer’ knob (a reproduction moulding by Craig Marston, another SPARC museum volunteer). Also, the band change knob is non-original – a moulding for this is in progress (see inset photo on next page)


31


32

Above: using a wobbulator to check the IF response curve (here the narrowest non-crystal setting). Below: set-up for connecting the wobbulator – inject the swept IF signal from the wobbulator into grid 1 of the mixer tube (section 1.1 of the main tuning gang, or access pin 8 of the tube base – the connection to the bandchange switch) and feed the signal from across the diode load resistor (R25) into the scope Y-input.

hallicrafters sx-28a ‘super skyrider’ receiver introduction · with many of its components...

Documents