Fullscreen HTML5Med Res (??MB)HTML4

Vintage Radio Building a 1970s Little General By Fred Lever The Little General is a classic superhet AM radio design published in the April 1940 issue of Radio & Hobbies magazine. Some time ago, I built one using parts that were available in 1946, but I decided to see what improvements could be made using parts from the 1970s. M y classic post-war styled (1946) Little General, shown in Photo 1, used octal valves and contemporary parts. The set was quite heavy and bulky by today’s standards at 280 × 200 × 200mm and 5kg, but for 1946, that was typical of what a radio enthusiast could achieve. By 1976, 30 years later, electronics and components had greatly advanced due to the advent of TV. So I decided to build a new Little General using valves and parts that were available in 1976. For inspiration, I went through my valve box and, out of dozens of TV types, found a 6CS6 pentagrid, a 6EH7 frame grid pentode and a 6DX8 triode/pentode output valve. All were new old stock (NOS), still in their original boxes. With a twogang mini condenser, a suitable aerial coil and an oscillator coil, the 6CS6 could be the tuner/converter and the 6EH7 could be used as an IF amplification stage. That IF signal would then be applied to a diode and filter to demodulate the 94 Silicon Chip AM and eliminate the remaining RF signal. As the 6DX8 is a triode/pentode, I could use the triode section as a diode and the pentode as the audio amplifier. To keep things compact, four of the new (in 1976) silicon diodes can work as a bridge rectifier in the power supply in place of a 6V4 valve rectifier. That eases the heater draw and allows a simpler transformer with just two secondary windings. The resulting set (see the lead photo) measures 230 × 150 × 140mm and weighs approximately 2kg, so it is much more compact than my 1946 style model, at 4.8L versus 11.2L. The more modern miniature valves draw much less power, reducing the size of the required power supply. Parts like IF transformers and valves are about ¼ of the size of the 1946 version. The performance is similar. had several types of mini intermediate-­ frequency transformers (IFTs) to choose from. However, I was short aerial and oscillator coils. Still, I had a box full of assorted unknown coils to go through at a later stage. I wanted to settle the size of the transformers first as they are the biggest parts on the chassis and affect the layout more than the small pieces. Using three valves, I needed 1.3A at Design process I will now go through the design process. Having selected the valves, a look through the junk boxes showed I Australia's electronics magazine Photo 1: this 1940s-style Little General radio I built earlier works well, but it’s hardly compact and fairly hefty at 5kg. siliconchip.com.au Photo 3: I placed the components on the chassis to get a rough layout, marked the locations, drilled and cut the holes and then painted it. Here it is ready to start having parts mounted to it. 6.3V for the heaters and about 30mA at 250V for the plates. That works out to about 16W, so a 20W transformer would be suitable. I had a discarded soldering iron transformer specified as “22 watts” on the sticker. I dismantled the transformer, leaving the mains primary winding on the bobbin. I replaced the soldering iron’s 30V secondary with a 6.3V winding for the heaters and a 240V winding for the HT. I then restacked the transformer, tested it with dummy loads and finally, varnished it. I had a Jaycar AS3025 90 × 50mm 8W general-purpose rectangular speaker on hand, so I decided to use that for the radio. The speaker transformer needed to reflect the 8W impedance of the loudspeaker to a higher value for the plate of the 6DX8. I had a Jaycar MM2006 2W 12V mains transformer Photo 2: the aerial coil (left) and oscillator coil (right) look a bit messy, but they tune over the required ranges. siliconchip.com.au Photo 4: I riveted the valve sockets and mains transformer to the chassis. Most other components were mounted less permanently later, via bolts or on tag strips. that I wired to a 6DX8 in a bench test circuit configured as a class-A audio amplifier to see how it performed. The transformer did an OK job of passing a couple of watts from the valve to the speaker. The impedance of the primary circuit, at maximum power transfer, was around 12kW. I dismantled the transformer, stripped off the original tapped secondary and wound back on a single secondary with a turns ratio that matched the 8W speaker to 12kW. I reassembled the transformer with a slight air gap in the lamination stack, tested it again, then dunked it in varnish. Tuning coils The mixer stage needed tuning and oscillator coils that would give a continuous frequency differential matching the intermediate frequency (IF) while adjusting the tuning gang. For example, if the tuning coil tuned from 500kHz to 1700kHz over the full rotation of the tuning gang, the oscillator coil would need to tune from 955kHz to 2155kHz, ie, 455kHz above the tuning coil (assuming a 455kHz IF). Both coils needed to be adjustable, with ferrite cores, and inductances to suit the broadcast tuning range. Note that the required ratio on the tuning coil is about 3:1, while it’s closer to 2:1 for the oscillator coil. The mini gang I intended to use had equal aerial and oscillator capacitance sections. That means series ‘padding’ of the oscillator gang capacitance is Australia's electronics magazine needed to compress the oscillator range from 3:1 to 2:1. I scratched about in my coil junk box and found nothing that looked like an oscillator coil, but I did locate a rough-looking complete ferrite core coil on a ½” (12.7mm) tube with a tuning winding and a small primary. I measured the inductances as 0.1mH for the big coil and 0.01mH for the other. When I hooked it to the gang and tested for resonance, I found it tuned from 600kHz to 1800kHz, and screwing the core in and out made a big difference to the range. That was good enough for the aerial coil. For the oscillator coil, I had a spare blank portion of a ¼” (6.35mm) IFT former left over from previous projects, so I wound on about 250 turns of scrap Litz wire and measured its inductance as 0.08mH. I added a 30-turn tickler coil of 0.01mH. I tested its resonance and, with 150pF in series with the coil, I had a tuning range of 950kHz to 2300kHz that also varied a fair bit by moving the slug in and out. Those two coils were good enough to start testing. I found a pair of mini IF cans marked “L128” and checked their resonance. Both coils resonated at around 440kHz with measurements of 1.43mH and 23W. The four adjusting cores worked on both, so they looked good to go. Building the set I dropped the parts gathered so far onto a sheet of paper and outlined a June 2025  95 Photos 5 & 6: these photos show the underside of the chassis (left) and top (right) partway through construction. Most of the larger parts are in place, with the smaller components and wiring to do. likely layout. That layout provided a template for the chassis. The chassis is so small that some light gauge sheet (from a computer case) sufficed. I centre-punched the holes and used drills and hole saws to make the cutouts. I made a few adjustments, like slotting the control spindle holes so I could drop those parts in and out easily. I sprayed a light undercoat on the inside and a light coat of white paint on the outside, giving the result shown in Photo 3. I then started mounting parts on the chassis, pop riveting some parts permanently into place, like the valve sockets and tag strips, as shown in Photo 4. Next, I mounted the heavy parts, followed by lighter parts like the coils and gang. I mounted the tuning gang using some spacers to lift the shaft to the centre height of the speaker. I left the actual dial drum for later and used a large knob to move the gang spindle temporarily. I also bolted a pot shaft to the chassis for a string drive. That shaft bush and nut were later secured to a strip of Bakelite on which the tuning coils were mounted. I made access holes in the chassis front panel for the slugs of the tuning coils. I fitted some tag strips underneath and squeezed another tag strip on the top of the chassis behind the speaker. On that, I mounted the filter capacitors and three 4.7kW PW5 ceramic resistors in parallel for ~1.5kW total to use in the HT filter. I pushed the resistors hard up against the transformer as a heat sink. Underneath, I mounted a small MB4 bridge rectifier on a tag strip and wired the HT through the filters to the 6DX8. I then completed the mains and the heater wiring to the sockets. It was time to power it up and road test the 6DX8 with the new power supply. The audio stage Having completed the power supply and 6DX8 wiring, I increased the AC input voltage in small steps using a variac to reform the electrolytic capacitors. Nothing smoked, and I measured 313V DC at the rectifier output and 6.6V AC on the heaters. The three 4.7kW 5W HT dropper resistors lowered the 313V DC to 284V DC. I had wired the 6DX8 with a 330W bias resistor, keeping in mind the plate rating of 18mA, and measured a 24V drop across the HT resistor and 6V bias, both indicating about 20mA being drawn. The audio stage tested OK with an input sensitivity of 0.5V for clipping and no audible hum. The IF stage I wired in the 6EH7 and 6CS6 and fluked the oscillator tickler coil phasing, allowing the oscillator to run immediately. I aimed to prove the IF part of the circuit first but encountered Photos 9 & 10: shown adjacent is the set with a temporary tuning knob, while above is the tuning knob string arrangment I came up with. 96 Silicon Chip Australia's electronics magazine siliconchip.com.au Photos 7 & 8: the photo on the left shows the initial stage of under-chassis wiring, while on the right I have added and wired up the smaller components too. problems feeding a 440kHz IF signal through the control grid of the 6CS6. Usually, I just kill the local oscillator and treat the converter as a straight RF valve to pass the IF signal into the control grid and through the IF transformer set. In this case, if I shut down the local oscillator, the 6CS6 valve would not pass a signal from its control grid to the plate! As soon as I unblocked the oscillator grid circuit, the valve would self-bias and work as an RF amplifier. However, if I blasted several volts of 440kHz into the 6CS6 grid, enough passed through the plate that I could at least peak the cores. There were many other problems with making the IF section work, but suffice it to say that after a hard struggle, it worked well. One important lesson I learned was that the 6EH7 needs a separate, stable screen supply, not one shared with the converter. Also, the 6EH7 is a very high-gain valve and needs an AGC bias feedback control on top of a pedestal of self-bias to work stably at all signal strengths. With the IF system working, I had to adjust the tuning and oscillator coils so that, with the tuning gang set anywhere in its range, the oscillator frequency was 440kHz higher than the tuned station frequency. The initial oscillator range of 1000kHz to 2700kHz was too high. I left the coil turns the same but changed the padder capacitor value, added a trimmer on the gang and varied the coil core position. By juggling those three factors, I achieved the desired range. The next job was to make the tuning coil resonate from 500kHz to 1800kHz. With the core set so that good coupling was achieved from primary to secondary, I could not get the bottom frequency under about 650kHz, and then the top was around 2300kHz, both too high, indicating insufficient turns on the coil. I pulled one lead end off the big winding, joined some Litz wire and wound on another 40 turns. I then got a range of 549kHz to 1890kHz, close enough to work. Next, I carefully measured the actual difference in frequency between the two coils at multiple points over the tuning range. My first tests concluded that the variation was about ±20kHz around 450kHz over the tuning range. With a bit more careful adjustment of the coils, I reduced that error to ±5kHz – see Fig.1. As a product of that process, the mean IF value increased to about 455kHz. I deemed that acceptable, as the IFTs have a passband broad enough to encompass the deviation without a significant loss of coupling. With those changes, the set started to act like a real receiver. The volume could be adjusted from zero to Fig.1: this plot shows the difference in the tuning and oscillator coil resonances (vertical axis) as the dial is rotated (horizontal axis). The red plot is what I found initially, with a variation of more than ±20kHz from an average of 450kHz. Some tweaking gave me the blue curve, within about ±5kHz from 455kHz over most of the range, resulting in more consistent performance. siliconchip.com.au Australia's electronics magazine June 2025  97 Fig.2 (above): this is my revised version of the Little General circuit. There are other changes besides the different valve lineup, such as the volume control method (attenuation of the audio signal rather than varying the valve bias) and the oscillator coil arrangement (tapped rather than two coupled windings). Fig.3 (below): the original Little General circuit diagram from Radio & Hobbies, April 1940. You can find all the changes I made in my circuit by comparing the two. Still, the overall configuration (number of valves and purpose) is very similar. 98 Silicon Chip Australia's electronics magazine siliconchip.com.au maximum, and the audio output was level no matter what station it was tuned into. At this stage, the circuit, shown in Fig.2, was pretty much final. You can compare it to the original Little General circuit, Fig.3. The AGC voltage was low on a weak station, around -1V with 4.2V across the IF valve bias resistor. On a strong station, the AGC signal measured -12V and the cathode measured 1.5V, indicating that the valve was throttled, trying to keep a consistent IF signal level. However, the set was full of heterodyne whistles! They led me on another merry chase, trying this and that with little effect. Having run out of ideas, I realised that the set, while very selective, was not that sensitive, needing a fair length of antenna to work. I decided to look at that problem first. Harking back to the 6CS6 not wanting to work as a plain RF amplifier, I tried another 6CS6 valve. For this test, I tuned the receiver with the original valve and settled the RF level so the AGC was –12V. I then swapped the valve for a grubby, well-used XTV chassis 6CS6 (from a different manufacturer), and as it warmed up, without moving anything else, I was amazed to see the AGC climb past -12V and settle at -24V! That was not just double the gain, as the AGC works up a slope throttling the 6EH7, but many times the gain. The AGC system was now working even better, with the 6EH7 operating over a huge bias range, drawing 4mA with no signal and throttling back to around 0.2mA on 2RPH, with a mean level of around 2mA on average stations. The net result was that the audio level was consistent, irrespective of the station signal level. Off-station, the background frying and fizzling from all the suburb rooftop inverters comes up, while on-station, the background noise disappears and stations tune in loudly. I then realised that the whistle problem was also gone! Thinking about this later, I suspect the 6CS6 might not have been the best choice. While it is a pentagrid, the valve was designed to be used as a sync pulse separator. A minor manufacture variation that had no effect on separator use may have a large impact when used for another application like this one. siliconchip.com.au Scope 1: testing the IF response with a swept sinewave fed into the radio reveals that it is pretty symmetrical about the ~450kHz intermediate frequency. Scope 2: the signal from the volume control pot’s wiper with a station tuned in. You can see the lowerfrequency audio signal is overlaid with higherfrequency noise, the remnant of the IF (and possibly RF) signals. Scope 3: the audio signal delivered to the speaker is cleaner than that shown in Scope 2, mainly due to filtering by the 1nF capacitor across the speaker coil. I also tried a second old 6CS6, which worked just as well as its stablemate. Still, no real conclusion can be drawn with a sample of just three valves. I suspect a radio type 6BE6 would be a better choice. Another possibility is that my NOS 6CS6 was simply faulty! Returning to the IF stage I went back to the IF, swept it, and Australia's electronics magazine took some shots of the response. The sweep response was quite symmetrical on either side of 450kHz, as shown in Scope 1. Note that this is an ‘active’ response curve as the AGC is working and limiting the gain. However, the general response is evident. When tuned to a station, after the volume control, I found a signal of over 120mV peak-to-peak with a fair amount of RF still present (Scope 2). June 2025  99 Photos 11 & 12: I turned five-ply timber on a lathe and routed a channel around to hold the string. Note the tension spring on the back of the dial. By the time we get through the 6DX8, and with the help of the top-cut capacitor on the plate, we wind up with a clean audio signal of around 140V peak-to-peak at the plate (Scope 3). Finishing it off The final chassis is not one of my neatest jobs and would benefit from being stripped out and rebuilt, with some parts moved. Placing an electrolytic capacitor next to a hot output valve is not the most sensible move. However, it was good enough to function, and I wanted to press on, finalise the cabinet and dial and get to the end. With a tuning knob spindle already mounted on the chassis, I needed a drum on the tuning capacitor to couple to the spindle. I had nothing in the junk box, so I grabbed a flat scrap of five-ply timber and made a drum about 80mm in diameter. I machined a string groove in the centre of the outer rim. I had a Jaycar ¼in (6.35mm) bore hub (Cat YG2784) that matched the gang shaft and fitted that to the centre of the timber wheel. Next, I drilled holes to thread the string ends through the drum from the rim groove. These short holes emerge at an angle at the back of the drum. One hole allowed one end of a string to be anchored to a wood screw. The string then goes around the drum, down to the spindle, two-and-ahalf times around the spindle and back up to the drum, then down through a second hole, terminated to a spring to maintain some tension on the string. I sketched out a cabinet design made of plywood with a circular dial opening and then looked for something to make a dial bezel. What I needed was something round and shiny. My eye fell on some tin cans in the kitchen recycling bin. I put a can in the lathe and bored the end out of it. Then I swung the tool post around and cut the end off, giving me a ‘chrome’ bezel. The idea for the cabinet was to have the front panel recessed from the front to protect the knobs. Otherwise, it’s a simple box made from five-ply timber with glue fillet joints and a back plate with slots to let air in and form a handle. The dial bezel and a bunch of ¼in (6.35mm) holes for the speaker completed the front panel. The back is then held in with four screws that go into the chassis blocks and two top blocks. One of those also limits the power transformer’s upward movement. With the basic box made, I sanded it down a bit and flowed on a coat of red stain. I repeated that a couple of times, with sanding in between, until I had a reasonably smooth finish. Ultimately, I decided that sticking one’s fingers into the live works to carry it was not a good idea, so I carefully added a flat strap to the top as a proper carry handle. Conclusion It has some flaws, such as parts not quite lined up straight, rat’s nest wiring and values that need optimising. These are properties of prototype radios that would be ironed out in a production run. Still, I am not a manufacturer, so it will do. As with any other scratch-built project, there was far more work involved in getting it to work than this article reveals. Much more detail can be found at the following links (parts 1-3): • siliconchip.au/link/abtk • siliconchip.au/link/abtl SC • siliconchip.au/link/abtm Photos 13-15: the last few steps required before assembly involved making the timber cabinet, which I then stained red. The complete Little General radio was more compact, weighing ~2kg; about half the weight of the radio shown in Photo 1. 100 Silicon Chip Australia's electronics magazine siliconchip.com.au