Fullscreen HTML5Med Res (??MB)HTML4

Vintage Radio US US Marine Marine Corps Corps TBY-8 TBY-8 Squad Squad Radio Radio By Ian Batty Military equipment can be state-of-the-art, or just plain ancient. This radio is a bit of both; it’s seemingly an obsolete design at the time it was fielded, but there are good reasons for the choices made, and it turns out to be an outstanding performer. It’s also a bit different from your usual vintage radio fare. 96 Silicon Chip Australia’s electronics magazine Consider the US Air Force, which fields some of the latest and greatest aviation technology, like the F-35 Lightning II multi-role stealth fighter, and some positively ancient technology, like the B-52 Stratofortress. Some B-52s still in service today were built in the early 1960s! The RAAF is not much different; they also field the thoroughly modern F-35 alongside the positively ancient C-130 Hercules, which first took flight in 1956, over 60 years ago. The common thread here is fitness for purpose. It takes billions of dollars and decades to design new military equipment, so if the old equipment does the job, and can be kept going, it’s often the way to go. Consider the modulated oscillator transmitter and the super-regenerative receiver. These were well-proven if somewhat ‘old hat’ even in the 1930s. That’s when the United States Navy contracted for a new radio set. It was to be “ultra-portable” for use by Marines on foot, to operate well above the commonly-used lower frequencies of the HF band, and to offer Wireless Telegraphy (W/T) for Morse code transmission and Radio Telephony (R/T) for voice transmission. It’s part of the T (transmitters) series, B (portable) subseries, letter Y in order of registration. This class of equipment is now known as a squad radio. As well as being carried on foot, TBYs were also commonly used as ship-to-ship links in convoys and battle groups. The TBY was famously used by specially-recruited Navajo-speaking “Codetalkers”, as the Navajo language had never been documented. Its purely oral tradition, unusual syntax and highly inflected, tonal pronunciation made it unlikely that, even if intercepted, any “codetalked” message could ever be decrypted. It’s one of the few examples of “clear speech” being anything but clear to the enemy. siliconchip.com.au This was the inspiration for the 2002 movie “Windtalkers”. Technical details Condensed Specifications The full height of the antenna is approximately 9ft. Squad radios commonly use battery or generator power, since they need to be able to go where the troops do. As most use directly-heated valves, cathode biasing for each stage is impractical. The most common designs use multi-voltage batteries that include the bias supply. It’s not unusual to see one or two filament and HT supplies along with the negative bias supply. The TBY uses this design, with 1.5V and 3V LT rails, 150V HT and a -7.5V bias battery. Its full circuit is shown in Fig.1. Multi-channel transceiver designs would either use a bank of quartz crystals (rare, bulky and expensive in the 1930s), with one per channel, or a much simpler Variable Frequency Oscillator (VFO) design. If the transmitter were crystal controlled, it would have been possible to use the same crystal for receive and transmit (with a bit of magic between), but it would still have been an intricate design. With no crystal control in the transmitter, however, the receiver would have to be continuously-tuned. The TBY uses a modulated oscillator transmitter, which has the great advantage of simplicity; it only requires one RF stage. But that simple design leads to frequency instability, producing frequency modulation along with the intended amplitude modulation (AM). For receiver performance, nothing could beat Edwin Armstrong’s super-regenerative design in its day, and that’s still true today. So long as a valve can be made to oscillate, it can be used as a super-regenerative demodulator, right up to its maximum operating frequency. While the super-regenerator was good enough in the 1930s, even ham radio operators were abandoning it by the 1950s, gradually pushing the design to higher and higher VHF/UHF bands until finally giving up on it. Its versatility and simplicity, though, did see the use of super-regenerating klystrons in simple radar receivers. If it can oscillate, it can super-regenerate. So we have two mostly deprecated systems from the late 1930s/early 1940s: an unstable, messy modulation transmitter and a primitive, cranky siliconchip.com.au Operating frequency: 28~80MHz in four bands; 131 channels at 400kHz spacing. Transmission/reception: A2 (tonemodulated continuous wave – MCW), A3 (AM – double-sideband full carrier – R/T). Transmit power/operating range: MCW 0.75W, R/T 0.5W. Range up to 3 miles (~5km). Receiver sensitivity: 5µV on bands 1, 2 and 3; better than 15µV on band 4, all for 1mW output at 6dB SNR. Power supply and duration: combined battery, 1.5V “A” supply (RF section), 3V “A” supply (audio section), 150V “B” supply, -7.5V “C” supply. 25 hours operation when new, minimum of 15 hours. Versions: TBY-1 and -2 used fixed antenna mounts of Westinghouse manufacture. TBY-3 not issued. TBY-4 to -8 featured rotatable antenna mount and SO-239 socket for antennas other than the nine-section rod, Colonial Radio manufacture. Metering: indicating meter switchable to RF filament voltage, audio filament voltage, transmitter anode current (loading) indications. Operator-useable rheostats to control audio and RF valve filament voltages. Interfaces: R/T provision for two headsets, Morse key for MCW operation. Channel selection: channels set according to the attached individual calibration chart. Able to be set accurately on any channel an exact multiple of 5MHz. On any other channel, dependent on equipment tuning chart and antenna coupling. One report shows transmitter frequency varying by as much as 100kHz with coupling adjustment. Accessories: Carbon microphone/ dynamic earphones combination, Morse key, dry battery, ten-section rod antenna, 4V accumulator and vibrator power pack, 115V DC/AC mains power pack, canvas carry backpack, 72.5MHz fixed ground plane antenna, timber transit case. Australia’s electronics magazine September 2020  97 receiver. Given the poor opinion most authors have of this combination, I want to find out just how bad (or good!) they can be. The TBY squad radio The TBY (version 1 released in 1938) is a seven-valve, battery-powered squad radio transceiver which can be carried by one person in a backpack. It provides four switched, manually-tuned bands from 28~80MHz and uses a nine-section whip antenna whose length is adjusted (by add- ing or removing sections) to always be roughly one quarter-wavelength at the chosen operating frequency. Completely assembled, the antenna just tops 2.6m! (The red ribbon at the top is not recommended for combat conditions). Tuning is indicated by graduated wheels behind viewing windows. No frequency calibration is provided; operators use reference charts attached to the top cover to select any one of 131 operating channels at 400kHz spacings. The internal 5MHz crystal calibrator’s “marker” signals allow receiver and transmitter calibration at intervals of 12-½ channels. I acquired this one in the 60s at ACE Radio, a disposals company long gone. I was actually not sure what it was, but its design was too good to pass up. Transmitter circuit The transmitter uses Acorn 958A valves (V3 and V4) in a push-pull Hartley circuit. Unlike Class-B audio circuits, this operates in Class-C, Fig.1: circuit diagram for the TBY-8 radio. There is no capacitor C23 shown (but there is a C24), and C23 does not appear in the parts list. Presumably, this was a late change during manufacturing, or a change from a previous version. 98 Silicon Chip Australia’s electronics magazine siliconchip.com.au where the conduction angle is considerably less than 180°, and the control grids are driven sufficiently positive to rectify and create grid current during part of the operating cycle. You’d expect such a brief conduction cycle to create massive distortion, and it does. Class-C can only work with tuned loads (“tank” circuits) that ‘force’ the output to form a sinewave. You can think of the tank circuit as acting like a flywheel, pushed along by anode current pulses; or, as a conventional tuned circuit that only responds to the desired frequency, attenuating the ‘crossover distortion’ harmonics. Class-C operation can give efficiencies exceeding 70%. Simply put, during conduction, the valve operates in heavy saturation with little voltage drop across it and little power wastage. This high efficiency is a boon in battery-powered sets, but it also allows valves to give substantial outputs exceeding three times their anode dissipation limits. The basic Hartley circuit uses a tapped inductor to provide feedback. The TBY’s push-pull transmitter’s anode tuned circuit uses centre-tapped coils to both combine valve currents and provide ‘cross-connected’ feedback. Feedback is provided by 50pF capacitors C15/C16 to the grids of V3/V4, with centre-tapped choke L10 isolating the grids from the RF ground provided by 500pF capacitor C17, which bypasses 5kW grid bias resistor R4. The transmit stage is matched to the antenna via the secondary of the selected tank coil, in combination with matching variable capacitor C13. In operation, meter M1 is switched to the Plate Current position, and C13 adjusted for a centre reading on M1. The intimate coupling between oscillator and antenna makes the TBY’s frequency stability vulnerable to antenna length and capacitive effects between the antenna and other objects. For R/T (voice) transmission, modulation begins with the carbon microphone, powered from the -7.5V bias supply. The microphone current is stepped up by transformer T2 to drive V7, a 1E7 dual pentode. T2’s grid drive to V7 is in anti-phase, so the modulation amplifier works in Class-B push-pull mode, with T3 combining the anode currents of V3 and coupling modulation (via its secondsiliconchip.com.au ary) to the transmitter. V7 receives the full -7.5V grid bias via the driver winding (secondary) of T2. For Morse transition, pushbutton key S101 switches the tertiary winding of T1 to ground, as well as activating transmit/receive relay K1 and keying the transmitter. Grounding T1’s tertiary activates V6’s feedback loop (C25/R12/R13), which is inactive until pin 5 on T1 is connected to ground. V6 oscillates at around 500Hz, feeding the tone to modulator V7, which in turn modulates the transmitter. Receiver circuitry The receiver begins with 959 Acorn pentode V1, operating as a commoncathode RF amplifier. This provides the usual gain and selectivity, but also helps reduce radiation from the oscillating demodulator. Without adequate demodulator isolation, this set would radiate enough energy in receive mode to allow hostile interception and direction-finding. It’s the military version of a flashing “kick me” sign on your back. The antenna circuit uses one of four turret-switched coils (L1-3 & L15), with its secondary tuned by the C1 section of the receiver’s ganged tuning capacitor. Antenna trimmer C2 compensates for antenna capacitance and/ or nearby objects. The amplifier gets grid bias from the bias battery via the antenna coil secondary, from resistive divider R21/R22. V1’s anode load, the primary of L4-6 Australia’s electronics magazine The side of the TBY radio showing where the antenna mount is attached. & L16, couples to its tuned secondary. This secondary forms the Hartley oscillator circuit for V2, the super-regenerative demodulator. The super-regenerator, one of Edwin Armstrong’s four industry-defining patents (regeneration, super-regeneration, the superhet and frequency modulation) achieves astounding sensitivity. How does a single-stage voltage gain approaching a million sound? Heavy feedback, aided by 1MW grid bias resistor R2’s return to V2’s positive anode connection puts V2 into powerful oscillation. The rectified grid current develops a negative voltage across 100pF coupling capacitor C7, counteracting the positive September 2020  99 voltage that would otherwise be present on the grid. This counteraction continues until the negative bias is so strong that the valve cuts off. With no oscillation to maintain it, the cut-off bias across C7 will be discharged according to the time constant of grid resistor R2, coupling capacitor C7 and the positive supply voltage. As the cut-off bias leaks away, V2 will come back into conduction and will again oscillate, re-initiating negative bias across C7. This cycle will repeat at the quenching frequency (22~40kHz in the TBY, depending on the Regen setting). I’ll use 30kHz as my example. You might expect this to simply produce a self-modulated RF output. Indeed, such ‘squegging’ oscillators were used in ultra-simple lifeboat transmitters. But the average anode current of this circuit is very noisy. The quenching frequency exhibits a large amount of phase noise (jitter). It’s related to the conditions at the instant when oscillation is re-initiated. As this exact instant is strongly influenced by valve noise, the average anode current which forms the quench frequency is also noisy as shown in Fig.2. So far, all we have is either a jittery oscillator or a circuit greatly magnifying its own inherent noise. But what if an external, unmodulated signal is fed to the super-regenerator? It will ‘lock’ to the incoming signal and, although the quench frequency will remain at around 30kHz, it will now be very stable. Each new burst of oscillation will be initiated as the incoming signal brings the grid out of cut-off and into active operation, rather than by valve noise. Fig.3 shows that, if the anode current jitter is quieted, the anode current assumes a constant, noiseless DC value. Fig.2 100 Silicon Chip The left-hand side of the TBY radio showing the transmitter section. You can also see one of the 958 Acorn valves (V4) at upper left. Applied signal synchronisation Now, if we supply a modulated signal, the initiation of each oscillatory period is determined by the varying instantaneous input signal amplitude. The incoming modulated signal will influence the quench frequency’s phase. The simplest modern paradigm is that of pulse-width modulation (PWM). But PWM is, in context, a ‘modern’ concept, postdating Armstrong’s invention by over thirty years. It’s why you’ll find incomplete, confusing and elaborate descriptions of the super-regen, including its “strong AVC action”. Fig.4 shows how the modulated input signal is translated into an audiovarying anode current that is amplified and delivered to the headphones. The dotted line is a notional bias voltage that the input signal’s amplitude must exceed to provoke oscillation as the circuit’s highly negative bias ‘leaks off’. Eagle-eyed readers may interpret the input signal’s modulation as suffering from non-linearity. You’d be correct, but the illustration does show that the super-regen can successfully demodulate an AM signal. More on this later. The demodulator circuit connects Fig.3 Fig.4 Australia’s electronics magazine siliconchip.com.au To reduce lead inductance, conventional basing methods were eliminated, and the shortest possible connections made to the external circuit. RCA’s all-glass Acorn valves (named for their envelope shape) set the stage for the next thirty years of valve design: baseless all-glass construction, connecting pins penetrating the envelope, and electrode connections welded directly to the connecting pins. The Acorn base demanded a spacehogging peripheral socket, so the final B7G development had the pins exit the envelope in a circle around a glass button base, with the socket not much larger than the valve’s envelope. How good is it? The right-hand side of the TBY-8 which showcases the receiver section. to the supply via the primary of audio transformer T1 and 500kW potentiometer R8, the regeneration control. In operation, R8 is adjusted so that the receiver just comes into reliable super-regeneration. That gives maximum sensitivity. T1’s secondary feeds audio to 500kW volume control pot R7, and then to preamplifier valve V6. Like V7, this gets a -7.5V grid bias from the battery. V6’s anode drives output valve V7 via audio transformer T2. Transmit/receive switching is managed by relay K1, which responds to the press-to-talk (PTT) switch on the microphone. K1 gets current from the +3V filament supply. Transmit/receive changeover is managed by switching filament power (K1d). HT to the transmitter is also switched (K1e), as early versions of the 968A would continue to oscillate with no filament power applied – anode current alone was sufficient to sustain emission. The circuit contains a lot of RF bypassing not generally seen in AM radios. This is needed for predictable siliconchip.com.au operation in the low VHF band, and – especially in the receiver – to reduce possible radiation from the oscillating demodulator. Acorn valves The 1930s saw an explosion of research into higher and higher radio frequencies. Governments, along with commercial and scientific organisations, joined the race to exploit the revolution. But experimenters quickly discovered the thermionic valve’s limitations. Even ‘modern’ octal-based types, universally preferred for MF (medium frequencies) to low VHF (very high frequencies), struggled to work much past 100MHz. This was due to three principal problems: transit time from cathode to anode, internal capacitance, and lead inductances. Much smaller constructions could reduce transit time and capacitances, and lead inductances by much shorter leads. This was pretty simple to achieve; just reduce the valve’s elements down to the limits of hand assembly. Australia’s electronics magazine I get 1mW in 600W output (775mV) from a 30%-modulated 3µV signal. Given that such a signal carries around 1µV of modulated audio, this set has a voltage gain of about 775,000 from the antenna to headphones. Beat that! In decibels, 1µV into 50W is around 2 × 10-14 watts. We have one milliwatt output, making the power gain around 113dB. Not bad for just four valves. For the demodulator itself, I get around 70mV of audio for a measured input of 3µV (implying 1µV of audio modulation), so the demodulator voltage gain is about 70,000. Like I wrote earlier, in its day, nothing gave more gain than the super-regen, and nothing can today. Note that I’m not quoting a dB figure for the demodulator, as I can’t state the demodulator’s input and output impedances, and dBs should only be calculated with known impedances. A close-up of a 955 Acorn triode valve. Source: https://en.wikipedia. org/wiki/File:955ACORN.jpg September 2020  101 I’m guessing this set has not been used since it was decommissioned, so I wondered how well it had retained its calibration. So I set my HP signal generator to Channel 1 (28MHz) and tuned it in. It came in at 27.85MHz, a calibration error of around 0.5%. That’s excellent long-term stability. Following the instructions, its internal calibrator allowed me to set Channel 6 (30MHz) to correct this error to within 6kHz. That isn’t bad for a calibrator that’s over 70 years old, with no temperature control or adjustment. It sits at 5.000750 MHz, an accuracy of 150 parts per million, and certainly adequate for the application. The receiver still meets specifications, with a sensitivity of 3µV bettering the quoted 5µV for a signal-to-noise ratio (SNR) of 6dB on Bands 1~3, and 10µV or better on Band 4. The receiver -3dB bandwidth varies from ±100kHz at 28MHz to ±200kHz at 80MHz. For -60dB, it’s ±500kHz and 700kHz respectively. That bandwidth sounds woeful compared to a superhet, but remember that superhet designers can specify an IF bandwidth as low as a few 102 Silicon Chip kilohertz, regardless of the incoming signal frequency. Essentially a TRF design, the super-regen must rely on high-Q RF coils to give usefully narrow bandwidths. Since Q = F ÷ df, where Q is the quality factor, F the operating frequency and df the -3dB bandwidth, at 28MHz Q = 140 and at 80MHz, it’s 200. That’s very good for just two tuned circuits, one of which is heavily loaded by the oscillating demodulator. Although the RF amp’s gain is small, its lack of loading allows the antenna circuit to contribute most of the circuit’s selectivity. Signal output is determined by the maximum possible change in anode current pulse width, and it reaches this limit at quite low signal levels. In this way, it’s similar to an FM receiver’s limiter. Starting with a 3µV signal at 28MHz, it needed around 200mV to get a 3dB audio output increase, a range of more than 90dB. For the accepted 20dB SNR, it needed over 10µV, and never achieved much better. You may know of FM’s capture effect, where a signal that’s only a few times stronger than another will ‘blanAustralia’s electronics magazine ket’ the weaker signal. The TBY exhibits significant capture effect with a signal ratio of three times or more. It does produce heterodyning ‘birdies’ if there is any large frequency difference, and this seems to be due to interaction with the 30kHz quench frequency. The circuit is naturally noisy and exhibits poor linearity. A 25µV signal (30% modulation at 400Hz) produced 15% total harmonic distortion (THD). Audio response from the antenna to headphones of 160~600Hz, as determined by the demodulator; from the primary of T1 to the headphones it’s 160~6500Hz. For the microphone input, it exceeds 80Hz to 10kHz. It is capable of demodulating FM, but needs a stronger signal to exploit slope demodulation: at 28MHz, a 25µV signal with ±60kHz deviation produced 1mW output. So while it would receive FM broadcasts with some rebuilding of the Band 4 coils, given the audio top end of under 1kHz you’d be pretty disappointed with the results. Transmitter output exceeded the specifications on all bands, delivering upwards of 1W on some frequencies. And yes, it does produce substantial siliconchip.com.au frequency modulation. Fig.5 shows the carrier and many side frequencies. A pure amplitude-modulated signal will produce the carrier and only two side frequencies: upper and lower. Multiple side frequencies are a frequency modulation signature. Although the TBY’s receiver will demodulate an FM signal, this isn’t much help in demodulating the FMrich signal from another TBY, as you would have to detune your own set to go into slope demodulation with the penalty of lower sensitivity. I was concerned about demodulator radiation, but it appears well-con- Fig.5: shown in greyscale for clarity. siliconchip.com.au trolled. I tried an FM Walkman that tunes down to 76MHz, and could just pick up the radiation with the two sets next to each other. The demodulator’s anode voltage varies with the Regen setting, and the circuit shows typical values. The set under test was powered by an inverter bought off eBay. It works well, but does give a high bias output, as shown in the circuit readings. Usability For equipment designed to be used under the extreme conditions of warfare, the TBY’s simplicity of operation is excellent. Once tuned, all one needs to do is listen or talk, for up to 25 hours. But re-tuning is another matter. Band changing is simple, but actual channel tuning is difficult to the point of being almost impossible. The tuning knobs are small and difficult to operate, and the dials can only be read by looking directly into the windows. Perhaps the original luminous markings would have helped, however, in their now-degraded state they are visible but not readily legible. Australia’s electronics magazine Originally-described as “radio-active”, a Geiger counter registered emissions at the lower end of concern. References • Instruction book for Navy Model TBY-8 Ultra-Portable Very High Frequency Transmitting-Receiving Equipment, 1943, Colonial Radio Corporation, Buffalo NY • Catalog for the models TBY, TBY1 & 2: siliconchip.com.au/link/ab3x • VMARS has heaps of military manuals, including the TBY-8, at siliconchip.com.au/link/ab3u • An extensive description of the radio: siliconchip.com.au/link/ab3v • Complete description and analysis of super-regeneration: Microwave Receivers, Van Voorhis, S. N. Ed., McGraw-Hill, 1948, Chapter 20, Superregenerative Receivers, Hall, G. O. pp 545-578. (MIT Rad. Lab. Vol 23) • Armstrong’s patent, US1424065: https://patents.google.com/patent/ US1424065A • Armstrong’s paper: Some Recent Developments Of Regenerative Circuits, Armstrong, E.H. siliconchip. com.au/link/ab3w SC September 2020  103