Fullscreen HTML5Med Res (??MB)HTML4

Vintage Electronics The BC-221 US-Military Frequency Meter from 1941 Domestic valve radios were calibrated at the factory and might be tweaked occasionally during service. But military radios operate in rough conditions, out in the field, and need to tune accurately across thousands of channels. How were they kept in calibration? The BC-221 was the secret weapon. By Ian Batty W orking on VHF and UHF aircraft radios back in my RAAF days in the mid-1960s was simple. Everything operated at defined channel frequencies and was crystal-controlled. The crew needed only to switch to the appropriate channel, and the equipment would do the rest. This demanded one crystal unit per channel in both the receiver and transmitter; a total of 16 in the eight-­ channel American AN/ARC-3, and 44 in the British TR16440! Even by the 1960s, crystals were still expensive and labour-intensive to produce. During WWII, millions would have been needed for the many tens of thousands of military radios of all kinds. Up until the late 1940s, equipment as diverse as the United States’ Marines TBY Squad Radio (September 2020; siliconchip.au/Article/14580) and the British Wireless Set No. 38 MkIII used a calibration-frequency 96 Silicon Chip crystal oscillator to provide ‘markers’ at specific intervals over the operational band. The SCR-274 “Command” sets, working at HF, used a single crystal in each tuneable transmitter, which, in concert with a ‘Magic Eye’ tube, confirmed the calibration at one point in the operating band. This was far from ideal, though. Firstly, the HF band spans 3-30MHz, with the range of around 3-15MHz most commonly used. Allowing 1kHz channel spacing, that demands around 12,000 channels. The famous SCR536/ BC611 “handietalkie” used any one of 50 crystal-controlled channels in its allocated band of 3.6~6MHz, but had to be returned to a depot to change channels. So, general-purpose HF transmitters, receivers and transmitter-­ receivers, such as our Wireless Set No. 19, could only be continuously tuned. Few of these sets provided any internal Australia's electronics magazine calibration, so operators were in the position of ‘set and hope’. Sets were routinely brought back to company headquarters, field depots, or major repair centres, so it was possible to provide frequency calibration then. But we still have the problem of about 12,000 possible frequency allocations. The solution was to provide a highly accurate and stable signal generator/ receiver with a detailed calibration chart. This would give technicians a reference that was accurate to a few hundred hertz when calibrated against its own internal crystal oscillator. The SCR-211/BC-221 The BC-221 Frequency Meter (BC = Basic Component) was a part of the SCR-211 parent set (SCR = Set, Complete Radio). Interestingly, in this case, it was the only part of that set. Still, the Meter itself is the BC-221, not the SCR-211. The SCR/BC system was siliconchip.com.au ultimately replaced by the JETDS system in 1943, but the BC-221 was never allocated a JETDS designator. The BC-221 comprises a variable-­ frequency oscillator for setting reference frequencies, a heterodyne demodulator for measuring transmitter frequencies, and a crystal calibrator for setting either its own calibration, or that of external equipment. The Technical Manual for the initial -A, -B and -C types of the BC-221 is dated 1941, while the definitive TM 11-300 of 1944 lists 25 variants. The instrument was made in the tens of thousands, and was also made by countries other than the USA – Louis Muelstee’s Wireless For The Warrior website has a Russian example (www.wftw.nl/russian221.html). The BC-221 was used right down at field level, not just in depots, thanks to it being compact and battery-­powered. It was designed to be carried by radio operators, technicians and small unit signal sections, so they could keep their sets on-frequency without having to ship them back to a depot. The box is built like a brick, with shock-resistant mounts, big batteries, and the headphones-plug-as-power-­ switch trick to conserve them. By consulting the calibration book and zero-beating against the internal crystal, an operator in a forward area could check his transmitter or receiver on the spot. That said, depots and workshops did use them too, for aligning radios after major repairs, or as a reference when producing training or calibration manuals. In practice, company headquarters/field workshops would have at least one. Frontline signal units often carried one so their radios could be set and kept in tolerance. Higher-level maintenance depots had them in larger numbers for mass calibration. The BC-221 has two ranges, Low Band (125~250kHz) and High Band (2~4MHz). Each instrument was individually calibrated, with an extensive custom calibration book, bound inside the lid of the case inside the front cover. The Low Band has a calibration point every 100Hz, while the High Band has points at 1kHz spacings. Setting its output frequency to a precision of 100Hz (on the Low Band) or 1kHz (on the High Band) is done using a master dial drum (visible through a window) and a rotary vernier dial. siliconchip.com.au The manual also shows how to measure frequencies between the indicated values. Thus, within the instrument’s limits of calibration, it would be possible to tune a transmitter to 8.130MHz, midway between the chart listings of 8.128MHz and 8.132MHz. Although the oscillator is extremely frequency-stable, its Class-C operation makes its output rich in harmonics. Given its fundamental frequency ranges of 125~250kHz and 2~4MHz, the BC-221’s useful ranges extend from 125kHz to 2MHz (fundamental, 2nd, 4th and 8th harmonics) on the Low Band and 2MHz to 20MHz (fundamental, 2nd, 4th and 5th harmonics) on the High Band. The spectrum sweep in Scope 1 shows the harmonic output for a dial setting of 2MHz. Likewise, the internal reference crystal oscillator’s output is rich in harmonics, ensuring checkpoints beyond 20MHz. The expanded view of the frequency meter. This version used a timber cabinet, but there are also some that have an aluminium-alloy cabinet. Scope 1: both oscillators (the primary and crystal ones) have significant harmonics in their output, as shown here. This is not a flaw but in fact a useful feature, since you can use the harmonics to tune a radio to a multiple of the selected frequency. Why good radios are so important for militaries The battle of Port Arthur/Tsushima in 1905 saw the Japanese Fleet destroy or capture all eleven battleships of the opposing Russian Second Pacific Squadron. How was it possible for the Japanese, who had acquired their first warship less than 50 years previously, to defeat the Russians, with a naval fighting history reaching back to 1696? Both fleets were equipped with Morse code equipment, but the Japanese gear was locally designed, technically superior, more reliable, and used to much greater effect due to thorough operator training at the Yokosuka Training School. So radios can play a decisive role in armed conflict, but only if they are reliable and operated effectively. You need to know what channel or frequency to use to get that communication happening, hence the need for calibration. Australia's electronics magazine December 2025  97 Side views of the BC-211 frequency meter. There were as many as 25 different BC-211 models, each with slight variations to the circuit. Automatic calibration in the 1940s The method used to generate the custom calibration book for each set is most surprising, because it was done by computer – in the early 1940s! Engineers at the Philco Corporation Research Division, Engineering Department created an automatic calibration computer for this task using 126 valves. The BC-221 was calibrated by noting the dial reading for internal heterodyne beats and calculating how many cycles per dial division it was from the previous calibration point. The computer consisted of an automatic calibrator combined with an adding machine (semi-automatic), which recorded the calibration data at 327 points, interpolated between those points, and automatically printed the 3252 different frequency value numbers in each individual calibration book. The time to do all this was 6.5 hours; the actual printing of the values in the book took just 16 minutes. The time for an experienced human to do the work manually, in contrast, was averaged at 16 hours per frequency meter. A mechanical hand automatically turned the dial of the BC-221 to various predetermined settings while measurements were underway. The calculations were carried out It’s interesting to note the arrangement of the chassis (shown from the rear), with the separate boards and the valves placed at different angles. 98 Silicon Chip Australia's electronics magazine by automated complex adding machines of the type then used by banks and finance companies, and the results printed in the calibration book and also saved on paper tape. The adding machines were fitted with solenoids to depress the keys; the valves the computer used were mainly 0A4G cold-cathode thyratrons, similar in function to a modern bistable flip-flop. This description of the calibration was based on information from the webpage at https://jproc.ca/ve3fab/ bc221.html Operating principle Most modern devices that require precise high-frequency signals generate them using phase-locked loop (PLL) circuits. A master oscillator (MO) operates at a fixed frequency and feeds one input of a phase comparator. The phase comparator’s other input is fed by the output of a voltage-­ controlled, variable-frequency oscillator (VCO) via a frequency divider (usually a programmable one). The VCO’s frequency is controlled by the phase comparator’s DC output. If the phase comparator detects a difference between its inputs (VCO and divider), it will ‘steer’ the oscillator until its two frequency inputs are equal. In practice, the VCO will be forced to be in phase with the divider’s output. This is a servo system, similar to how a car’s cruise control can maintain the set speed, even when going uphill. In this case, the “servo” is providing an accurate, fixed frequency, using a reference and frequency divider. siliconchip.com.au The top view of the chassis, showing the master dial drum at the bottom. an external transmitter signal or the internal crystal reference. As you tune the BC-221, you hear a beat tone in the headphones. When the beat tone slows to zero, the VFO and the reference signal are at the same frequency. So, with the BC-221, you sense the difference between the BC-221’s frequency and that of the incoming signal. Instead of an automatic control loop as in the homodyne, the operator listens for the beat note and dials the oscillator until the difference can no longer be perceived. This manual zero-beating method was simple, reliable, and accurate enough that with the aid of the calibration book, operators could set or measure frequencies to within a few hundred hertz, more than sufficient for wartime communications. Accuracy To change frequency, the divider is simply commanded to a different division ratio. The division ratio of the divider, through negative feedback, becomes the frequency multiplication factor. While “PLL” is a modern term, the principle goes back to 1924 as the homodyne (“same power”) system, in contrast to Armstrong’s heterodyne (“different power”). If we heterodyne an AM signal with a pure sinewave of the same frequency (eg, the output of a PLL locked to the carrier), we have a superhet with an intermediate frequency (IF) of 0Hz – all that would be left is the modulation! This type of circuit appeared as early as 1924 as the homodyne, and later as the synchrodyne in Electronics Australia, June 1975. The original homodyne used a weakly oscillating circuit that would ‘pull in’ to an incoming signal of sufficient strength and synchronise to it. Connect a pair of headphones to the valve’s anode lead and off you go. This circuit demonstrated the ‘lock-in’ principle but, since it used a filter to create the control voltage, it was an automatic frequency control (AFC) system, rather than a true PLL. Similar techniques would be widely used in FM tuners before the ready availability of integrated-­ circuit PLLs. The generic homodyne circuit in Fig.1 shows just one tuning component: C1 tunes the oscillator into its siliconchip.com.au ‘capture’ range, and the control voltage does the rest. As the homodyne/ synchrodyne converts directly at the incoming carrier frequency (‘direct conversion’), it offers a frequency response from essentially DC to some upper limit set only by the filter that follows the demodulator. In a homodyne or synchrodyne receiver, the local oscillator is automatically pulled into lock with the incoming signal when the frequencies are close. When that happens, the two are at the same frequency and the audible ‘beat note’ disappears. The BC-221 uses a related principle, but instead of locking, it relies on manual adjustments by the operator. The instrument’s variable-frequency oscillator (VFO) is mixed with either The instrument is highly precise; the following lists the maximum frequency error expected due to each source of imprecision: 1. Small shocks (caused by handling, and the thrust on the dial and pressure on the panel when using the equipment): 100 cycles maximum 2. The action of locking the dial: 30 cycles maximum 3. Warming up: 100 cycles maximum 4. Changing of load on the antenna post: 50 cycles maximum 5. A drop of 10% in battery voltage or a change of 5°C in the surrounding temperature: 325 cycles maximum 6. Error in calibration: 500 cycles maximum 7. Error in crystal frequency: 250 cycles maximum Fig.1: the general concept of a homodyne. Once tuned close to the input signal frequency, the oscillator will ‘lock on’ to it. The output of the mixer is therefore just the modulation signal, with the carrier removed entirely. December 2025  99 Total error: 1355 cycles maximum or 0.034% at 4000 kc. The manual notes that “Actual tests show that the maximum errors can be assumed no greater than 50% of the values given…” In practice, it’s also unlikely that all errors would sum in the same direction. The manual notes that the maximum errors will occur at a frequency of 4MHz and a temperature of -30°C (!). Robert Watson-Watt’s dictum, “Give them the third-best to get on with. Second best takes too long and the best never comes”. That approach provided the radar systems that won the Battle of Britain in 1940. But who doesn’t love going down the rabbit hole? Using an atomic clock reference (the carrier of 774 ABC Melbourne), it is accurate to well under 100 millihertz (0.1Hz). My HP 8656B signal generator came in 4Hz high, and Power supply all following figures are quoted relative The instrument was designed to corrected figures on the HP. for battery operation. Four series-­ On power-up after who-knowsconnected “Number 6” cells supplied how-many-years, with the recomthe 6V valve heaters, while six 22.5V mended supply voltage and after a BA-2 batteries in series supplied the few minutes of warming up, the crys135V HT. tal oscillator came in at 1,000,002Hz I found that -0.5V and +0.5V changes (1.000002MHz), a +2ppm error. Surin heater voltage gave a frequency error prisingly, during the performance of -33Hz and -2Hz, respectively, at measurement process, it settled to 2MHz, while a drop from 135V to 1,000,017Hz, an increase to +17ppm. 100V HT caused an increase of 146Hz Still, not bad after 70+ years. at 2MHz. To measure that tiny 2Hz error, I just It’s easy to get lost in the Jungle of set my BC-211 to exactly 1kHz above Excess Precision – it’s where you chase the reference frequency and sent the down some parameter to the point of resulting audio tone to my frequency undetectablity. I’m reminded of Sir counter. This indicated 983Hz and 998Hz. Subtracting my 1kHz offset gave the -17Hz/-2Hz results. I checked my counter’s calibration as part of the exercise, so I’m OK with claiming my 998Hz measurement. Circuit details Components in the circuit (Fig.2) are numbered in order (with sub-­ numbering when values are shared): the capacitors are #1 to #10-3, resistors are #20-1 to #26, with minor components filling the gaps to #36. The valves are not numbered. Confusingly, the -A circuit shows the variable oscillator with the screen grid enclosing the control grid, and shows a pentode with four grids! This is an erroneous symbology dating from the early days of circuit diagrams. In reality, the ‘77 is a conventional triple-­ grid pentode, with its first (control) grid connected to the band switch and main tuning capacitor. The -B circuit is correct. The VFO, a VT77/‘77 in the initial issue, or a VT116/6SJ7 in later versions, operates as a cathode-­coupled Hartley oscillator. The cathode con- Fig.2: the BC-221-C and -D circuit looks deceptively simple, but this instrument has been carefully designed to minimise drift so that after factory calibration, it will remain very accurate in field service. You just need to refer to the calibration booklet attached to the unit to tune it to just about any frequency. 100 Silicon Chip Australia's electronics magazine siliconchip.com.au Fig.3: this shows how accurately my unit tracked despite its age. Compare the green and dark blue curves to see the benefit of calibration. nects to a tap on the selected tuning coil, and feedback to the grid is taken from the top of the coil. Unusually, there’s no grid bias resistor, and the grid returns to DC ground; the circuit uses cathode bias via resistors #20-2/ #22-2 (5kW/10kW). The circuit is tuned by main capacitor #1 (150pF). There is an individual trimmer for the top end of each band (#3, #4), and a master Corrector trimmer #2 mounted on the front panel. The RF output is taken from the oscillator’s anode; this is electron coupling, ensuring that any changes in the output circuit do not affect the oscillator’s frequency stability. The oscillator output connects to the antenna terminal and also the grid of the heterodyne detector, a 6A7 pentagrid or a VT167/6K8 triode-hexode. The antenna connection allows the set to operate as a low-power signal generator for receiver calibration. As the oscillator is not modulated, exact receiver adjustment is easiest done by calibrating a tuneable transmitter to match a frequency set on the BC-211, then tuning the receiver to the calibrated transmitter. The BC-211 manual details this method in paragraph 11, on page 6 (a link to this is under the References cross-heading). The 6A7/VT167/6K8 converter operates in one of three modes: a. as a direct heterodyne comparator of an incoming transmitter signal and the local VFO b. as a 1MHz ‘marker’ generator for calibrating external equipment siliconchip.com.au c. as a 1MHz ‘marker’ generator for calibrating the BC-221 itself The converter’s anode feeds the audio amplifier, a VT76/’76 in the -A issue, or a VT116/6SJ7 afterwards. This amplifies the converter’s heterodyne output to drive external headphones. As a final nice touch, you have to plug the ‘phones in to apply power to the heater circuit. As you have to remove the headphone jack to close the case, battery life would be preserved during transport even if the power switch was left on. How good is it? It is outstanding, as you can see from Fig.3. For equipment that’s been unused for decades to come in with its calibrator only 17ppm high is outstanding. Its only limitation is the lack of a modulator for the VFO – this would have made it easier to use when lining up receivers. It’s capable of very high accuracy once calibrated, but, as my ‘as found’ results show, you can just check the chart for the frequency you want, dial up and off you go. The dial uses vernier graduations. This scheme allows accurate dial-­ setting to within one-tenth of a minor division without the burden of reading minuscule engravings. The handbook shows clearly how to do this. References TM 11-300 (1944 issue); which counts as the BC-211 (SCR-211) manual: www.qsl.net/zl1bpu/IONO/TM11300.pdf The original 1944 article on automated calibration: siliconchip.au/ link/ac94 (see pages 96~107). For a brief description: https://jproc. SC ca/ve3fab/bc221.html The underside view of the chassis. The cabling is tied with a continuous run of waxed string, known as “looming”. 101