Fullscreen HTML5Med Res (??MB)HTML4

Vintage Radio The TRF-One AM Radio based on a vintage IC By Dr Hugo Holden In April 1969, Electronics Australia published a radio design using the then-new LM372 AM radio integrated circuit (IC). 55 years later, the design is still valid, although the chip can be somewhat difficult to obtain. Despite that, intrigued by the design, I decided to build a modern version. T he single IC radio has always been a source of excitement and intrigue for radio constructions. The notion that nearly all the work can be done inside a single chip package is very appealing. This did not escape the attention of Jim Rowe in 1969, when RF-­capable ICs were making their debut. The siliconchip.com.au result was his “Micro-Plus” radio receiver design, published in the April issue of EA that year. The idea of a TRF (tuned radio frequency) radio is as old as the notion of radio itself. It involves a tuned resonant circuit consisting of an inductor and a capacitor; in a radio application, it is typically tuned by Australia's electronics magazine a variable capacitor. The tuned frequency range is usually the medium wave (MW) band, typically from 530kHz to 1600kHz, sometimes to 1700kHz. The most basic form of a TRF radio was the crystal set. In that case, the tuned circuit’s output was simply rectified by a diode to recover the February 2025  95 Fig.1: the LM372 IC from the late 1960s contains 14 NPN transistors, nine diodes and 17 resistors. It was intended to be the IF gain stage, detector and AGC circuit of an AM radio, but someone realised an antenna could be coupled directly to the input for MW reception. EA’s Micro-Plus radio used just eight components besides the LM372, battery and earphones (most of them capacitors). transmitted radio carrier’s amplitude modulation (AM). That audio signal could be sufficient to drive a high-­ sensitivity earphone or a crystal earpiece without any active amplification, so no power supply was needed! If a power supply was available, amplification could be added to the circuit to get a higher volume level, eg, to drive a loudspeaker. Later, multi-stage TRF radios were designed with high selectivity; then superhet radios came along with excellent selectivity and from that point on, TRF sets fell out of favour. I wanted to revisit the design of a single-gang variable-capacitor tuned The LM372 was not designed as the crux of an AM medium-wave TRF radio, but was pressed into service for that application. Rather, it was intended as an IF (intermediate frequency) amplifier with AGC (automatic gain control) – see Fig.1. However, it turned out that little needed to be added to the chip to make it function as a complete radio. Other MW band-single IC radios have been designed based on the ZN414 IC. Also, single-chip FM radios came along using the Phillips TDA7000, including popular radio kits sold by Dick Smith Electronics in the 1990s. The LM372 came in a TO-99 metal can package and has three internal functional blocks (see Fig.1). The gain stage typically amplifies the signal by 2360 times (67dB), while the precision detector stage has a gain of three times Photo 1: I made the radio’s case from phenolic material (left) and white Bramite (right). The latter was an Australian product that was no longer manufactured. Here the panels have already been cut to size, with the holes drilled and countersunk. Photo 2: the phenolic base with the rubber feet, spacers & other hardware. The two wires emerging from the base go to the AA cell holders underneath. 96 Silicon Chip TRF circuit, perhaps because I built these as a boy and had good success with them. I made several superhet radios as an adult, including those based on PLLs (phase-locked loops), and some FM (frequency modulation) radios too. Still, I retain a fondness for those early TRF sets. National Semiconductor’s LM372 IC Australia's electronics magazine siliconchip.com.au Fig.2: like EA’s 1969 Micro-Plus design, I have coupled a ferrite rod winding to the pin 2 input of IC1 with a series capacitor. The 1969 design was pretty minimalist, using just one active device (an NPN transistor) besides the LM372 IC, while mine adds a vintage op amp and two transistors for more overall gain and more power delivered to the speaker with less battery drain. or 10dB. So, a 50μV signal input modulated by 80% will produce an audio output signal of around 280mV RMS, or 800mV peak-to-peak. The AGC stage has an enormous control range of 60dB, with a threshold of 50μV. Therefore, using this IC as the basis of a TRF AM radio, the output level could be expected to be reasonably constant even if the signal level from the antenna increased from 50μV to 50mV. That means, tuning across the MW band, weak and strong signals would come in at a fairly uniform volume, more so than your typical AM radio. The LM372 has long been discontinued, but I found some for sale on eBay, so I snapped them up. They are presently hard to find, but a few are still for sale on eBay. At the time of writing, this listing offers two units (see Screen 1): siliconchip.au/link/abtp This IC has an input impedance of around 3kW, much lower than the ZN414. So when used in this application, unlike the ZN414, it requires a tap on the main resonant circuit, or a small coupling coil, to avoid damping the main tuned circuit. Designing a new circuit The circuit I designed around the LM372 IC is shown in Fig.2. As it is based on the same chip, it bears some similarities to the Micro-Plus from EA, April 1969, but it is my own original design. The main difference is that the Micro-Plus used a single-­ transistor Class-A amplifier whereas I have incorporated a preamplifier stage based on an op amp (IC2) plus a more powerful and efficient push-pull Class-AB amplifier based on NPN transistor Q1 and PNP transistor Q2. For a small battery-operated radio, it is always important to consider the power consumption. The LM372 can’t drive a speaker directly, so I decided to use a vintage Fairchild 741H op amp, also in a TO-99 metal can package, to provide a further voltage gain of 10 times. It drives a complimentary emitter-­ f ollower transistor output stage with simple diode biasing. Also, I decided to settle for a modest power output of 150-180mW (depending on whether I used the 32W or 40W speaker) so the output transistors would not require heatsinks. A 470μF capacitor stops DC being applied to the speaker. The 741H IC, 2N3053 NPN and 2N4036 PNP transistors are also available from eBay sellers. The physical bodies of the metal TO-5 cased 2N3053 and 2N4036 transistors act as heatsinks for the transistors inside them. They are better Photos 3 & 4: at this stage, I had soldered all the passives and sockets to the PCB, and then by mounting these high-quality metal AA cell holders on the underside of the base, the battery can easily be replaced when it goes flat. Once the cells are installed, a metal bar goes across them so they can’t fall out. siliconchip.com.au Australia's electronics magazine February 2025  97 Screen 1: it is challenging to find LM372s for sale these days, but there are a few around. This listing on eBay is probably your best bet (siliconchip.au/link/ abtp), but only two are available. The price is not bad, considering the original price and how long these have been obsolete. than epoxy-cased transistors in this respect. The temperature rise of each transistor body at full continuous sine wave output power is 10-13°C above ambient. In normal use listening to the radio, they never get noticeably warm. The manufacturer did not recommend the two 47W resistors in series with the LM372 input pins 2 (RF input) and 3 (gain stage input). Still, reading the Electronics Australia article, they had some difficulty with HF stability. So I decided to add them as a precaution. I also paid attention to the design of the PCB tracks around the input pins of the LM372. I provided double RF bypassing on the supply rail with high-quality 100nF axial ceramic capacitors. 1N5819 schottky diode D1 is included in case somebody installed the battery cells backward, so the LM372 and LM741 ICs would not be destroyed. I had vintage 40W and 32W speakers to test to see if they were suitable, along with a vintage National tuning dial that was made in 1943. I made the PCB with iron-on film, etched with ferric chloride. I added eyelet tags to connect wires to it and 0.9mm gold-plated pins and single connectors to couple in the signal from the ferrite rod’s coupling coil. Ultimately, I removed the tags and just used the eyelet part for the PCB connections. I stuck with all axial-leaded parts to give it a vintage theme. It pays to be mindful of the quality and appearance of the components. For example, the green 100nF 100V ceramic capacitors I used are high-quality vintage parts made by Corning Glass Works. I also used some ‘tropical fish’ capacitors to throw in a splash of colour. The IC and transistor sockets are high-quality types with gold-plated pins. It is a shame to have to solder to the pins of a very rare part like the LM372, or a vintage 741, for that matter. I found that grounding the body of the LM372 helps improve the stability, because it is such a high-gain arrangement in a very small package. I made a springy earth clamp out of brass that I screwed to one of the variable capacitor mounts with a collar – see Photos 7 & 8. I silver-plated the brass with an interesting product from the UK that is used to restore tea pots with a silvered finish. Mechanical construction In making an original or unique radio, ideally, you want it to look good and be long-lasting. So I never scrimp on materials and spend plenty of time to ensure that cut edges are smooth and polished. I also ensure that all the holes are in the correct positions, with perfect countersinking, so the screw heads sit flush where necessary. I had quite a lot of 10mm-thick brown phenolic material left over from other projects, some perforated aluminium mesh, and some white insulating material called Bramite – see Photo 1. Bramite is a uniquely Australian insulating panel material once used on household fuse boxes. It is practically unobtainable now. It is fantastically heat resistant, incredibly strong and machines well. I buy the 10mm-thick brown phenolic insulating panels from the markets at Akihabara in Tokyo. A local plastics company helped by planing the Bramite panel down to 5.5mm thick. All the hardware in this set is made from either stainless steel or nickel-­ plated brass. The hookup wire is Teflon covered. To keep the holes neat, I marked them with a micrometer edge, then a hand-held spike and started them with a 1mm drill bit in a hand pin chuck. I then drilled 1.5mm pilot holes and checked that everything fit together correctly. To ensure the CS screws that attach the front panel to the base and sides were all in the correct positions, I Photos 5 & 6: the completed radio chassis. The dial visible in the righthand photo is a vintage unit from 1943, while I made the speaker grille on the right from a scrap of perforated aluminium. 98 Silicon Chip Australia's electronics magazine siliconchip.com.au initially glued it together with some small dots of weak glue and used the holes in the panels used as a template to start the drill holes into the 10mm-thick phenolic material, so they all were in perfect registration. I used a metal strap to prevent the batteries from falling out of their holders. These Keystone cell holders (visible in Photo 4) are far superior to the usual Nylon AA holders that often stretch, harden or crack over time. The holders are retained by 4-40 UNC machine screws, with threads tapped the entire thickness of the 10mm thick phenolic base. The rubber feet are door stoppers. I machined spacers from ¼in brass tube that fit inside them, and they are attached to the base with 6-32 UNC screws passing into threaded holes in the baseplate. The variable capacitor is mounted on two nickel-plated brass spacers that attach it directly to the PCB. The vintage 40W loudspeaker was rusty and required rubbing down, treatment with Fertan and re-painting with Holts Auto Spray Paint. When the 365pF variable capacitor is fully meshed and the coil on the rod positioned to tune 530kHz, at the high end, the radio tunes to 2MHz. This Japanese-made variable capacitor does not have an additional trimmer capacitor on it, and there is nothing directly loading that point to add any capacitance there. It was made for the American market, with a ¼in shaft, and its body holes are pre-threaded with 6-32 UNC, rather than the usual metric threads found on Japanese parts. The finished radio is 200mm wide and about 150mm tall, including the rubber feet. At the upper end, by 18kHz, the first change in the amplifier’s output waveform, rather than amplitude loss, is slew-rate limiting by the 741 op amp. Performance The sinewave simply becomes trianThe main problem that a radio with gular, and the amplitude drops as the just one tuned circuit has is reduced frequency rises further. selectivity compared with a superhet This radio could almost be regarded or a TRF type, as they both have more as a hifi AM receiver. A tone correctuned stages. In other words, isolating tion capacitor is required to roll off the a station is harder if it’s close in fre- higher audio frequencies a little for a quency to another station. balanced sound. After some listening In this radio, this concern is some- tests, I found that an 820pF capacitor what offset by the very high-Q ferrite across the 100kW feedback resistor rod coil (shown in Photo 5) and the gave the best result. low loading on this by the coupling While there are better modern op coil, which I spaced away a little from amps than the 741, with output stages the Earthy end of the primary tuned that can swing closer to the supply circuit. Note that the input resistance rails (to gain more power output before of the LM372 is around 3kW. clipping), the internally frequency-­ Also, unlike most TRF radios, this compensated 741 is totally deaf to radio has a very high gain and a phe- radio frequencies and very stable, too, nomenally effective AGC. Weak and so it suits the application well. local stations appear with a similar Using the radio in an outdoor patio volume. Therefore, the performance is area, the 150mW of audio power is super lively, with many stations com- plenty, and it easily receives many AM ing in at a similar volume. stations with loud, crystal-clear outThe rod antenna is also deaf to elec- put with the volume control at about tric field noise. I used 30 AWG wire half or less. The radio exceeded my (0.254mm diameter) for the rod coil. expectations for an AM radio based I tried Litz wire but could not detect on a single tuned circuit. It is a very any difference in the Q compared to pleasant radio to listen to, and I find the 30 AWG wire. The large, high-­ myself using it most days. It was also permeability rod means there are fewer a fun exercise to design and build it! turns on the coil (just 46) than most If you want to build a similar set, you MW transistor radio coils. can download the PCB pattern from Due to the absence of transformer siliconchip.au/Shop/10/394 coupling in the audio stages, the freI also have dimensional drawings quency response of the audio circuit for the case, the PCB component layis flat, being about 3dB down at 50Hz out and some other details in the PDF (not that the small speaker could repro- at www.worldphaco.com/uploads/ SC duce such low frequencies very well). THE_TRF-ONE.pdf Photos 7 & 8: here you can see the spring-loaded grounding clamp I made to ground the TO-99 metal package of the LM372 radio IC. It attaches to the grounded metal post of the tuning gang. siliconchip.com.au Australia's electronics magazine February 2025  99