Fullscreen HTML5Med Res (??MB)HTML4

Vintage Radio Columbia TR-1000 Six-Transistor Portable Radio This attractive set uses a pretty standard (for the time) six-transistor circuit. It has a few quirks, though, such as a relatively low maximum audio output power, unusual transistor bypassing and a slightly odd audio feedback configuration. By Ian Batty T he Regency TR-1, released in 1954 (see April 2013; siliconchip.au/ Article/3761), was the first practical transistor radio made in significant numbers. So Columbia’s 1957 release of the TR-1000 was well into the heyday of the transistor radio. It seemed everybody in electronics was offering a transistor radio. While most buyers would not know much about how the radios worked, they probably assumed that six transistors were better than four, but maybe thought that seven would break the bank. So, six it was. But manufacturers would still need a way to make their offering stand out. Another small black brick (like the TR-1) was not going to do it. Some 50 years earlier, Kodak’s Brownie camera had introduced affordable photography, initially to children. By the 1940s, small, affordable cameras were ubiquitous. They also represented ‘go-anywhere’ convenience, and makers of portable radios took notice. It seems Columbia intended to cash in on the camera vibe, putting its TR-1000 into a handy leather case. Radiomuseum lists 423 offerings from Columbia, but only five transistor sets. The TR-1000 was made by Roland Radio Corporation, with the only real difference from their model 71-483 being the type of output transistors. The TR-1000 thus was an early ‘badge-engineered’ design adopted by Columbia. The TR-1000 circuit (Fig.1 overleaf) looks pretty much like any set of the era. It has a converter with emitter feedback (TR1), two intermediate frequency (IF) stages (TR2 & TR3), a diode for demodulation and automatic gain control (AGC; D1), an audio driver (TR4) and a push-pull Class-B output (TR5 & TR6). It uses single-point grounding for the IF stages, with the base and collector circuits bypassed to the emitter rather than to chassis ground. Ensuring that the base, emitter and transformer cold ends share the same AC reference point prevents emitter degeneration and eliminates signal loss or unintended feedback, without requiring a high-value emitter bypass With the right-hand knob removed, you can see the concentric shafts underneath. The D-shaped shaft in the middle is the drive shaft. The outer sleeve rotates more slowly. siliconchip.com.au Australia's electronics magazine February 2026  95 capacitor. This technique reduces the component count and is most commonly used at HF to UHF, where every component’s lead inductance must be at a minimum. Eliminating the emitter bypass removes the possibility of its lead inductance affecting the circuit. Controls The tuning gang has trimmer capacitors at both ends. While that’s unusual, it does leave one pair easily accessible, ensuring that the antenna/local oscillator (LO) trimmers are accessible and that the top end can be aligned without needing special tools. The tuning dial drives the tuning capacitor via a planetary reduction gear for precise tuning. The photo on page 95 shows the bright metal driveshaft, concentric to the tuning gang shaft, on the right. The frequency indicator disc fits the tuning gang shaft and is viewed through the transparent, knurled tuning knob. A few resistors in the set (R2: 33kW, R5: 2.2kW and R16: 33kW) are ±5% types. It’s not clear why just these three, especially R5 (the decoupling resistor for the converter), have a tighter than typical specification for the time (±10% was more common, and ±20% was not unheard of). It’s especially odd given the very wide spread of transistor parameters at this early stage of development. The audio section uses negative feedback via 33kW resistor R16, but 96 Silicon Chip this behaved in a peculiar way during testing. More on this later. The SAMS circuit shows an earphone socket, with the usual cutting-­ off of signal to the speaker when a plug is inserted. My set lacked this refinement, with a pop rivet filling the hole in the case. The SAMS circuit was not super-legible, and features some oddities, especially with capacitor numbering and notation. Capacitors C1 through C4 are electrolytics. It’s customary to begin component numbering at upper left, yet C1 is the second from the right, then comes C2, well to the left as the AGC audio filter for the first IF amplifier. The paper/ceramic capacitors (C5 through C20) are numbered according to their location, but the IF bypass capacitors (C11a, C11b, C13a & C13b) are multi-capacitor assemblies. Paper/ceramic capacitor values are given in picofarads, thus SAMS’s C12 is 10000pF (10nF), but C14 is “.05”, presumably meaning 0.05μF (50nF). Circuit description The Photofact circuit’s component numbering is peculiar. Capacitors C1 to C4 appear to the right and centre. The first fixed capacitor in the diagram was then C5. I have renumbered all components to conform to accepted drawing practice. As is common with a circuit containing only PNP transistors, the battery supply (9V) is positive to ground, Australia's electronics magazine making all circuit voltages negative. Converter TR1 uses collector-­emitter feedback, continuing the design used in the first “trannie”, Regency’s TR-1. The tuning gang, with its cut plate design guaranteeing accurate tracking between the oscillator frequency and the tuned signal, eliminates the need for a padder capacitor. The low forward bias supplied by R2/R1, in combination with the high emitter resistor R3, is only about 0.1V. This confirms the converter is operating in the Class-C mode that is vital to the conversion process. In Class C, the transistor is conducting for less than 180° of the signal cycle, compared to close to 180° for Class B, more than 180° for Class AB and 360° for Class A. By biasing the transistor so that it only conducts in short pulses, its nonlinear behaviour is emphasised. The short conduction bursts act like a ‘sampling’ of the LO and RF signals, which naturally generates the frequency products. We don’t want faithful amplification of either input on its own; we want the intermodulation products, including the downmixed IF signal that’s later extracted. TR1, a 2N411, feeds the tapped, tuned primary of the first IF transformer, L3. Its untuned, untapped secondary feeds first IF amplifier TR2, a 2N409. This stage works with minimal bias, supplied via R6 (100kW), for a collector current of around 0.7mA. This siliconchip.com.au allows the demodulator’s filtered DC output to control TR2’s gain via the AGC function. Capacitor C10 (3μF) bypasses the audio component of the demodulator’s output, while dual capacitor C11a/ C11b bypasses the collector circuit and the base circuit directly to TR2’s emitter. As noted above, this single-point technique is more effective than the usual bypassing directly to ground. TR2 is neutralised by feedback from the second IF transformer via 4.7pF capacitor C12. TR2 drives the tapped, tuned primary of the second IF transformer, L4, with the signal from its untuned, untapped secondary passing to the second IF amplifier, TR3. This stage (also a 2N409), as in most six-transistor sets, works with fixed bias and is also neutralised by 4.7pF capacitor C14. TR3 feeds the tuned, tapped primary of the third IF transformer, L5. Its untuned, untapped secondary feeds demodulator/AGC diode D1, a 1N60. All three front-end transistors are alloyed-junction germanium types. These use the same construction as the Philips/Mullard OC44/OC45, but with lower frequency specifications. The audio signal is developed across 5kW volume control R14, and the IF component is filtered out by 20nF capacitor C15. The AGC signal is fed back to the base of the first IF stage (TR2) via 3.3kW resistor R9. The audio signal then goes to the base of the audio driver transistor, TR4 (a 2N405), which uses ‘combination bias’ – a resistive divider at the base, plus an emitter resistor for stabilisation. There’s a top-cut capacitor (C19, 2nF) from TR4’s collector to ground, reducing noise and making the ultimate sound less shrill. TR4 drives the interstage/phase-splitting transformer, T1. Signals from T1’s secondaries feed the bases of the output transistors, TR5/TR6, both 2N407s. These drive the speaker transformer, T2, which then drives the speaker. Capacitor C20 (50nF), placed across T2 primary, adds further ‘top cut’. These days we would use complementary output transistors (PNP and NPN), but in the 1950s, only germanium PNP types were readily available. Early germanium NPN devices did exist, but they were generally inferior in performance. As a result, the preferred arrangement was a phase-splitter transformer driving two identical PNP output transistors in push-pull. Feedback is taken from T2’s secondary and fed, via 39kW resistor R21, to the base of audio driver TR4. Transistor specifications Apart from special types, it’s rare to see valves with a maximum frequency rating. The 6BE6 miniature pentagrid, common in broadcast radios, has been used in FM receivers in the 88-108MHz band. The miniature triode-­pentodes 6U8/6BL8 worked as converters in VHF-band TV tuners. Yet the TR-1000’s germanium converter and IF amplifier transistors would struggle to operate into the middle of the shortwave band, as would the OC44/45 types we’re more familiar with. Philips’ introduction of the alloy-­ diffused OC169~OC171 family offered receiver operation up to 50MHz. The alloy-diffused technology matured with the AF186, able to operate up to 820MHz. The ‘all-diffused’ Mesa and planar transistors that succeeded them easily exceeded 1000MHz (1GHz). But even within manufacturing technologies, maximum operating frequencies vary widely, so we need an explicit ‘frequency rating’ for a transistor. Transistors are specified for high-­ frequency operation in several ways, Fig.1: the TR-1000 circuit includes six alloyed-junction PNP germanium transistors and one point-contact germanium diode. TR1 is the mixer/oscillator with emitter feedback, TR2 & TR3 are the IF gain stages and TR4 is the audio preamplifier which drives phase-splitter transformer T1. The audio output pair, TR5/TR6, drives the loudspeaker or earphones via matching transformer T2. The components have been renumbered to conform to accepted drawing practices. siliconchip.com.au Australia's electronics magazine February 2026  97 often depending on their manufacturer. The most useful specification is the transition frequency, ft. This is calculated by plotting common-­emitter current gain (hfe, beta or β) against frequency. The point where β drops to unity is the transition frequency. Two other specifications exist: the β cutoff frequency (fβ), where common-­ emitter current gain falls to 70% of its mid-band value, and the alpha cutoff frequency (fα), where the common-­ base current gain (hfb, α) falls to 70% – see Fig.2. The transition frequency is the most useful. In practice, the common base fα figure is close to ft. Thus, a common-­ base circuit will operate satisfactorily up to ft. For the common-emitter configuration, say we use a transistor with ft = 1GHz (1000MHz) and hfe = 50. It will have a common-emitter gain of around 1.0 at 1000MHz (ie, at ft), but a gain of around 50 at 20MHz and any frequency below that. This raises a confusing question. The high-gain audio BC109 (β Fig.2: a plot of transistor current gain (common base & common emitter) versus frequency. = 240~900) has ft = 350MHz, while the low-gain BF115 RF amplifier (β = 45~165) has ft = 230MHz. Why bother with the BF115? For a BC109 with ft = 350MHz and hfe = 900, its fβ is just 360kHz (350MHz ÷ 900) – its gain will progressively drop with increasing operating frequency above that. For a BF115 (ft = 230MHz, β = 165), the fall begins at 1.4MHz. While these are clearly the worst cases, the best cases put the BC109 at 1.4MHz, and the BF115 at 5MHz, before their common-­ emitter gain starts to drop off. There’s another reason for preferring the BF115. As explained in The History of Transistors, Part 2 (April 2022; siliconchip.au/Article/15272), an internal resistance exists within the base region. This intrinsic resistance, rbb’, acts as does any resistance: it is a source of electrical noise according to the Stefan-Boltzmann Law. A high rbb’ acts as significant noise source within the transistor. In contrast, a low rbb’ will result in a lessnoisy transistor, and RF transistors – with their relatively low current gains – usually have low values of rbb’ when compared to low-frequency (‘audio’) types. So while a BC109 could replace a BF115 RF amplifier, the result would be a markedly higher noise figure. If you’ve ever seen an audio preamplifier with an RF transistor in the first stage, the explanation should now be clear. The first stage in any processing system typically sets the noise performance for the entire equipment. Low Output Transformer 1st Audio 3rd IFT Driver Transformer Outputs 2nd IF 2nd IFT 1st IFT Oscillator Coil 1st IF Antenna Tuner Oscillator Tuner Converter Ferrite Rod Antenna 98 Silicon Chip Australia's electronics magazine siliconchip.com.au noise is preferable to high gain, as gain can readily be made up following the input stage. Assembly It’s built on a metal chassis with point-to-point hand wiring. The transistors, with their leads offset to conform to the TO-40 packages, are all socketed. The components are packed tightly, with three of the electrolytic capacitors on top of the chassis, connecting via sleeved wires to the underside through holes. Some components in the converter section are placed on an ‘outrigger’ tag strip. It isn’t as packed as Astor’s Mickey OZ valve mantel set, but make sure you have some patience in reserve if you tackle a TR-1000. Cleaning the set up I bought this set on eBay quite a few years back. It’s one of those situations we probably all experience: a rush of blood to the head, purchase, delivery, storage on the ‘get to it one day’ shelf. It was complete, and in good condition, only missing the leather carry strap. The original battery connector had been removed, and a 9V ‘snap’ connector added. Turning it on gave some results, but its performance was pretty poor, failing to give more than about 40mW of audio with a strong signal. The AGC bypass capacitor, C10, was suspect. I’ve had trouble with electros in this position before. With bias resistor R6 at 100kW, even a few microamps of leakage in C10 would upset the AGC action. I replaced it, but to no effect. So I connected my audio generator to the top of the volume pot. Try as I might, this radio refused to give more than around 40mW before clipping. Time for some maths. From a 9V supply, you’ll get a maximum of 18V peak-to-peak across output transformer T2’s primary. That’s an RMS value of about 6.4V. Now, with T2’s primary at 840W, I get a maximum possible 48mW of audio power. So it seems the radio was OK – it’s a good example of ‘know your beast’. That is, don’t assume that every radio works and performs the same as every other one. I checked the SAMS literature, but there was no confirmation of my calculation. Given the maximum of 40mW at clipping, I did all sensitivity measurements at 20mW output. It aligned OK, with the problem being that the IF slugs are only accessible from the underside, and they are partly obscured by components. The slugs have very small slots, maybe 5mm or less, so care is needed when adjusting them. The audio feedback was a puzzle, as it didn’t seem to work. After a bit of faffing about, I discovered that putting a 5kW resistor in series with my audio generator allowed the feedback to take effect. This circuit, unusually, uses shunt voltage feedback via 33kW resistor R16 to the base of TR4, but relies on feeding an input of some 2kW impedance. My audio generator’s output impedance of only 50W was defeating the feedback by putting TR4’s base pretty The top (left) and bottom (right) of the Columbia TR1000 chassis. We haven’t marked it on the photos, but the demodulator diode (D1) can be seen in the right-hand photo at upper right (it has a glass body with a light blue cathode stripe). siliconchip.com.au Australia's electronics magazine February 2026  99 much at AC ground, in terms of impedance. So this is another case of ‘know your beast’. With a 5kW resistor in series with the audio generator, I simply set the audio generator to give 20mW of output, then used a millivoltmeter to measure the actual signal at the base of TR4. level rise; that’s good for a single-stage AGC system. Total harmonic distortion (THD) measured about 4% at 20mW output, and the maximum output was 40mW at clipping, with 10% THD. At 10mW, THD was still 3%. The output waveform appeared asymmetrical, with one half-cycle reduced in amplitude. How good is it? To try to explain this asymmetFor this first generation of six-­ ric waveform, I tested the gain of the transistor sets, it’s pretty good aside 2N407 output transistors, TR5 and from the low maximum audio output. TR6, and got very different gain (β) For 20mW output, it needed readings of 60 and 140, respectively. 350μV/m at 600kHz and 170μV/m This explained the asymmetry. at 1400kHz, with signal+noise:noise The alloyed-junction 2N406~8 ratios better than 20dB. Scaling up to (audio) and 2N409~12 (converter/IF) 50mW out, this is equivalent to around types were released in two different 550μV/m and 270μV/m. Raytheon’s packages: the offset-lead TO-40 and contemporary T-2500, using seven the triangular layout TO-1. Searchtransistors, was only about four times ing my junk box unearthed TO-1 ver(at 600kHz) and twice (at 1400kHz) as sions of these transistors, some of sensitive. which tested well. Substitution did Its RF bandwidth measured as give improved audio performance, ±1.5kHz for -3dB and ±23kHz for and roughly doubled the sensitivity. -60dB. The audio response from Ultimately, though, I left the TO-40 antenna to speaker measured as package transistors in place. Many 150~2600Hz at -3dB; from volume other radios of the era use the eloncontrol to speaker, it was 115~8000Hz. gated cases from the Regency TR-1/ The AGC action showed a 37dB sig- grown-junction era, or shiny cylinnal increase for a 6dB output audio drical ‘bullet’ cases. I reckon the black Versatile TO-40s make the TR-1000 – should you ever see inside one – distinctive. Special handling All knobs, and the frequency indicator disc, come off with finger pressure. Be aware that the knurled tuning knob fits the small, inner bright metal shaft that drives the planetary reduction, and that the station indicator disc fits the larger, outer brass shaft. The frequency indicator shaft is not keyed, so you will probably need to carefully twist it to give a correct frequency indication. Conclusion This set is unusual enough to belong in any collection of transistor radios from the 1950s. From that time of rapidly evolving designs, Columbia’s TR-1000 is a ‘must have’. Radiomuseum has useful information on this set at www.radiomuseum. org/r/cbs_columb_tr_1000.html The circuit appears in SAMS Photofact Folder 5, set 405, and is of better quality. I could not find it in any free online catalog. The SAMS website charges US$15 (+post) for the service SC sheets: www.samswebsite.com Battery Checker This tool lets you check the condition of most common batteries, such as Li-ion, LiPo, SLA, 9V batteries, AA, AAA, C & D cells; the list goes on. It’s simple to use – just connect the battery to the terminals and its details will be displayed on the OLED readout. Versatile Battery Checker Complete Kit (SC7465, $65+post) Includes all parts and the case required to build the Versatile Battery Checker, except the optional programming header, batteries and glue See the article in the May 2025 issue for more details: siliconchip.au/Article/18121 100 Silicon Chip Australia's electronics magazine siliconchip.com.au