Fullscreen HTML5Med Res (??MB)HTML4

Vintage Radio By Ian Batty Welcome To The Jet Age: Pye’s Excellent C-2 Jetliner Transistor Radio Pye’s C-2 “Jetliner” meets all the obvious criteria for a successful portable radio. It’s good-looking, has loads of audio output, picks up distant stations with ease, runs on almost-flat batteries and offers a tuning meter to precisely locate the “sweet spot”. But it’s what’s under the hood that’s really interesting. A N ADDENDUM to the landmark 1960 “Mullard Reference Manual of Transistor Circuits” described a portable radio using the (then) new family of alloy-diffused transistors, the OC169/170. This design had a sensitivity of 20µV/m and considering that a ferrite rod antenna has a “loss” of some 10:1 (20dB in voltage terms), this implied a basic sensitivity of about 2µV. At the time, the Mullard circuit demonstrated that transistor portables had developed to a point where they could compete with valve sets and win the contest. Very few valve radios could get anywhere near this figure without an RF amplifier stage. Both the Bush TR82C Mk.2 and the Kriesler 41/47 (described in this column in September & December 2013) adopted the basic Mullard design. The TR82C, in particular, achieved out92  Silicon Chip standing sensitivity but suffered from excessive noise on its broadcast band. As with the above two sets, Pye’s Jetliner follows the iconic “Mullard Design”. And like the 41-47, the Jetliner uses a PCB (the TR82C used a metal chassis and point-to-point wiring). PCB construction often means restricted access to the circuit for servicing. Most sets, including the Jetliner, mount the board “component side” up, leaving the connecting tracks on the “inside” of the case. The Pye service data helpfully includes a component layout diagram. You can download the circuit and service details from Kevin Chant’s excellent website at www. kevinchant.com Circuit Description Fig.1 shows the circuit details of the Pye Jetliner. Like the TR82C and 41-47, it follows the design that had become standard for the time: a selfoscillating mixer (TR1), two IF stages (TR2 & TR3), a diode demodulator and a transformer-coupled audio driver (TR4) feeding a push-pull transformercoupled output stage (TR5 & TR6). It uses six transistors (seven really), so it’s the standard “trannie” that we all know. The transistors used in mine are Philips/Mullard germanium PNP types – alloy-diffused in the RF/IF section and alloyed-junction in the audio. Bottom-coupled IF coils Whoever put this circuit together threw away the conventional handbook when it came to designing the 455kHz IF strip. That’s because it uses separate bottom-coupled IF coils in each stage (rather than conventional IF transformers). siliconchip.com.au Fig.1: the Pye Jetliner’s circuit uses a self-oscillator mixer (TR1), two IF stages (TR2 & TR3), a diode demodulator (D2), a transformer-coupled audio driver (TR4) and a push-pull transformer-coupled output stage (TR5 & TR6). D1 is connected to the mixer’s output and provides AGC. siliconchip.com.au FROM CONVERTER IF TRANSFORMER TO IF C3 SECONDARY TUNING C1 PRIMARY TUNING PRIMARY CURRENT C2 BOTTOM COUPLING SECONDARY CURRENT Fig.2: a bottom-coupled IF circuit. It uses two single-winding IF coils in separate cans and C2 couples the energy from the primary coil to the secondary coil. Basically, a conventional IF transformer uses primary and secondary windings, both tuned to the IF (intermediate frequency). They are placed close enough so that their magnetic fields interact and couple energy from the primary to the secondary. Their exact characteristics depend on the inductance of each winding and the spacing between them. It’s possible to calculate the mutual inductance between them (ie, the degree of coupling), along with the primary-to-secondary voltage ratio and the total bandwidth. However, while this method works well, calculations are laborious and transistor circuits require a low-impedance tapping on the tuned secondary for maximum power transfer. A less intuitive (but simpler) connection uses bottom coupling. In this case, the two coils can be in separate metal cans and the calculations are greatly simplified. This design works just as well as the traditional “primary-plus-secondary” version but it does require two separate coil (and can) assemblies. If you’re familiar with valve circuits, top and bottom coupling may seem commonplace but this is the first time I’ve seen the technique used in a transistor set. It’s a clever technique for several reasons. First, although it involves an extra coil can, each IF transformer has only a single slug that’s adjusted from the top. So there’s no need to get to both sides of the PCB for alignment adjustments, as would be the case with the conventional IF transformers. Second, getting the exact degree of inductive coupling needed between two coils in the one can is an exacting piece of electrical and physical design. With bottom (or top) coupling, the coils are simply wound individually. The degree of coupling is then determined by a simple formula that specifies the coupling capacitor’s value. Finally, there’s no confusion over correct slug positions: either of the two peaks is correct. Coupling circuits Before going further, let’s digress and take a generalised look at coupling circuits, so that we can better understand how the Pye Jetliner’s circuit works. September 2014  93 L SMOOTHED DC OUT FROM RECTIFIER C1 C2 (a) ‘PI’ FILTER AS USED IN MAINS POWER SUPPLY FROM POWER AMPLIFIER L C1 TO ANTENNA C2 (b) ‘PI’ FILTER AS USED IN TRANSMITTER OUTPUT FROM POWER AMPLIFIER C1 TO ANTENNA L C2 (c) CAPACITIVE DIVIDER AS USED IN TRANSMITTER OUTPUT Fig.3: (a) shows the conventional Pi filter configuration, (b) shows how it’s used for RF impedance matching (C2 many times larger than C1) and (c) shows a reconfigured version with a capacitive voltage divider as used in the Pye Jetliner (again C2 is much larger than C1). Basically, we need to match a highimpedance tuned circuit to a transistor’s low input impedance. Transmitter circuits also need to match into lowimpedance antenna feedlines, usually 50 ohms. As well, load impedances may be less than the feedline, requiring a step-up in impedance matching. Although tapped coils can be used, it’s easier to use some kind of capacitive voltage divider. This removes the “cut and try” method often needed at very high frequencies, where a coil may be only two or three turns and the exact tap location can be difficult to determine. Most of us are familiar with the Pi-filter configuration that’s used in mains-derived power supplies to smooth pulsating DC. What’s not so obvious is that it can also be used in a tuned circuit to match impedances. Fig.3(a) shows the conventional Pi-filter configuration, while Fig.3(b) shows how it can be for RF impedance matching. Finally, Fig.3(c) shows a reworked version with a capacitive voltage divider, as used in the Pye Jetliner. In a conventional power supply 94  Silicon Chip Pi-filter, C1 and C2 are often of equal values, eg, 8µF in vintage radio sets. However, in the RF version (Fig.3(b)), C2 is usually several times larger than C1, so that C2’s lower circuit impedance matches the antenna impedance. C1, on the other hand, provides a highimpedance load as required by the power amplifier’s output stage. Similarly, in the capacitive divider (Fig.3(c)), C2 is much larger in value than C1. The design calculations are simple and any desired impedance step-down is easily achieved. The capacitive voltage divider has an additional bonus: in the Jetliner, the mixer’s collector voltage is blocked by the “top” capacitor. As a result, the bias network can apply bias directly to the first IF stage, as this point is also isolated from DC ground by the “bottom” capacitor. So we achieve resonance, impedance matching and DC blocking with just three components and no coil tappings. Back to the Jetliner circuit Unlike most ‘broadcast-only’ transistor sets, the Jetliner uses a tuning gang with identical aerial and oscillator sections. In fact, it’s quite unusual to see this in a Japanese-manufactured ‘polyvaricon’ that uses a sheet plastic dielectric rather than air-spacing. As in its valve predecessors with identical tuning-gang sections, this means a that padder capacitor must be added to the oscillator provide tracking. This is the 315pF capacitor (C5) coupling the tuning gang to the top of the tuned oscillator coil (ie, just to the right of the 2N374/AF116n transistor – see Fig.1). The mixer uses collector-emitter feedback, thereby reducing the amount of local oscillator radiation that’s fed back out through the antenna rod. This design also includes an OA91 damping diode between the DC collector load of the first IF amplifier (2N373/AF117n) and the mixer’s collector circuit. This diode considerably improves the performance of the AGC (automatic gain control) on strong signals. The mixer’s output (ie, from TR1’s collector) feeds the untapped primary of the first IF transformer and it’s here that some thoughtful design work becomes apparent. Conventional broadcast-band IF amplifiers use tappings on the IF transformers to match impedances, especially on the secondary winding. This is necessary to match the low base impedances of the IF amplifier transistors and the low impedance of the demodulator diode. By contrast, in this circuit, the first IF transformer’s secondary is tuned by C9 (330pF) and C10 (5.6n) connected in series (giving 310pF). Importantly, C10’s low reactance provides a good match for the first IF transistor’s base impedance. But it’s even more complicated than that! The original Pye circuit drawing depicts the first IF transformer as the usual “two coils in the one can” configuration, coupled by their mutual magnetic fields. In reality, L3 and L4 are individual inductors in separate coil cans. They are bottom-coupled via the 33nF capacitor (C8) that appears to be a simple bypass. In reality, the IF signal circulating in L3’s resonant circuit is fed through capacitor C8, raising one end above signal ground. The signal at this end is in turn coupled through to L4 to create a signal current in its resonant circuit. The use of bottom coupling also explains the unusual connection of L4’s ‘cold’ end. Why not just connect siliconchip.com.au it straight to ground? Because there would be no signal introduced into L4’s tuned circuit; that’s why. Second IF stage The second IF stage is simplicity itself. The signal from the first IF amplifier (2N373) is fed to a single tuned IF coil and then coupled via a capacitive divider into the base of the second IF transistor (also a 2N373). This divider circuit uses the same component count as a tapped-inductor version but is easier to manufacture because there are no coil tappings. There’s also no need for a separate, low-impedance secondary winding on the IF coil to match into the second IF transistor’s base. The final IF transformer uses a tuned primary but also includes a low-impedance, untuned secondary to drive demodulator diode D2 (OA90). The two IF amplifier transistors operate in a similar manner to the IF amplifiers used in most other sets. The first IF stage (TR2) operates with a collector current of about 0.5mA. This allows the AGC to reduce its collector current effectively, to lower the gain as required. TR2’s emitter is connected to ground via an 820Ω resistor and a small meter labelled “Radicator”. This is a 500µA meter with a righthand zero and it functions as a signal-strength indicator. With the set is turned off, the needle rests at the righthand end of its travel. Conversely, when the set is on and there is full emitter current through TR2 (ie, no station tuned), the needle swings fully left, indicating “no signal”. When a station is being tuned, TR2’s emitter current falls due to AGC action and the meter swings to the right, towards the “maximum signal” position. In practice, it’s just a matter of tuning the station for a maximum reading on the meter. This signal-strength meter circuit is a common design and works equally well with both valve and transistor IF amplifier stages. Both types draw maximum current with no signal and minimum current with maximum signal. This is why these meters commonly indicate maximum signal strength when the power is off. All the RF/IF transistors are AF116/117 (or 2N374/2N373) alloydiffused types. Their feedback capacitance is low enough that no neutralisation is needed at 455kHz. siliconchip.com.au This photo shows the component side of the PCB but note that the heatsinks for the output transistors and the bias diode (at right) have yet to be riveted together again following transistor replacement. The demodulator (D2) is a conventional OA90 diode. The demodulated audio is fed via a voltage divider to the volume control, while the AGC voltage is derived via R14 and C21 and fed back to the bias network for the first IF amplifier (TR2). The diode’s output is positive-going, so it “bucks” the negative bias applied to TR2’s base, thereby reducing the transistor’s collector current and lowering its gain. The stronger the signal, the greater the reduction in TR2’s collector current and the greater the reduction in gain. As with all AGC systems, the net effect is to keep the audio signal fairly constant with varying RF signal strengths. However, the amount of control we can apply to a single IF stage is limited; eventually the transistor will be almost completely cut off and there will be no further gain reduction. It’s not practical to control a selfoscillating mixer’s collector current for AGC, as this would force the local oscillator off frequency. However, it is possible to apply damping to the primary of the first IF coil and thus reduce the converter’s overall gain. In the Pye Jetliner, that’s done using the auxiliary AGC diode (D1). As shown in Fig.1, this diode (OA91/1N60-A) has its cathode connected to the DC supply for the first IF amplifier, while its anode connects directly to the mixer’s output (ie, as fed to the first IF coil’s primary). With no AGC action (ie, little or no signal), the TR2’s collector current pulls D1’s cathode down to about 2.5V. This is about 2V more positive than its anode, so the diode is reverse biased and does not conduct. Conversely, as the AGC takes effect (and TR2 draws less current), the D1’s cathode voltage rises, eventually becoming less posi- tive than its anode. When that happens, D1 begins to conduct and this damps (or reduces) the signal at the converter’s collector. As a result, the mixer’s output is effectively reduced and this significantly improves the overall AGC action. According to Mullard, the AGC range improves from about 35dB (ie, input signal increase for a 6dB audio output increase) without the diode to over 55dB with the diode in circuit. Audio stages The audio driver stage (TR4) is biased in a similar manner to the IF amplifiers and works identically. However, it uses a larger emitter bypass capacitor and this is necessary to ensure that it is effective at audio frequencies. TR4 drives the primary of transformer T1 which operates as a phase splitter. Its centre-tapped secondary drives a Class-B push-pull output stage based on transistors TR5 & TR6 and these in turn drive the centre-tapped primary winding of speaker transformer T2. T2’s secondary then drives either two parallel-connected loudspeakers or a set of headphones via a headphone socket. Resistor R24 provides feedback from the output of transformer T2 to TR4’s emitter to minimise distortion. Note that the output stage dispenses with the usual voltage divider or voltage divider-plus-thermistor arrangement for thermal stability. Instead, a series resistor feeds a diode-connected transistor (TR7) and this reduces the bias applied to the output stage as the temperature rises. But that’s not all it does, as we shall see. A diode for bias? Unfortunately, both the Bush TR82C September 2014  95 tors. This gives tight thermal coupling so that the transistor-connected diode will respond to output transistor temperature variations. Even the best thermistors, separately mounted flat on a circuit board, cannot match this degree of bias voltage response. Transistor manufacturing tolerances mean that some form of bias adjustment is needed. As a result, the Jetliner provides a jumper to select one of two bias values. This jumper either places resistor R28 in parallel with R29 or a series combination of R28 & R27 in parallel with R29. Finally, emitter resistors R25 & R26 provide some local feedback and help balance differing gains in the two output transistors. Getting it going The PCB has been lifted free of the case here, revealing the two loudspeakers and the dial-drive mechanism. Note that the two dial pointers must be aligned with the case slots during reassembly. and the Kriesler 41/47 suffer from increasing distortion with falling battery supply voltage, due to decreasing output stage bias. This is a common fault in many transistor radios, especially those using germanium transistors. It’s common to see a thermistor used in the output stage bias circuit but, in many cases, this only compensates for ambient temperature changes and cannot counteract falling bias with falling battery voltage. Worse, thermistors are often mounted on the circuit board and cannot compensate for overheating in an output stage that’s being run at high volume. The Ferris M134 portable car radio was notorious for blown OC72 output transistors caused by just this problem. In many sets, increasing crossover distortion as the batteries age can be so bad that owners will discard batteries before they are truly “flat”. As a result, I’ve actually modified some of these sets for family and friends to improve performance. Unlike the TR82C and the 41/47, the Jetliner uses a semiconductor “diode” in the bias network. Well, it’s not really a diode. Instead, it’s a diode-connected transistor (TR7), which has its base directly connected to its collector. The reason for doing this is straightforward. In operation, a simple germanium diode begins to conduct at around 0.2V but its forward voltage rises quite rapidly with current. This 96  Silicon Chip means that a varying battery voltage would pass a varying current through a resistor in series with the diode and the diode’s forward voltage would change accordingly. Connecting TR7’s base to its collector brings in transistor action. As soon as base current begins to flow, it will cause a larger collector current. This means that even a small increase in base voltage will cause a significant rise in total current, so the device acts as a diode with a sharper ‘knee’ than using the base-emitter junction alone. This device not only delivers the required bias voltage but also has the same voltage-vs-temperature coefficient as the output transistors. As the temperature increases, the voltage across it reduces slightly to ensure that the correct bias is applied to the output transistors to ensure thermal stability. As a result, the Jetliner (and sets with the same bias circuit design) delivers good audio performance until the batteries are almost dead flat. On test, the set easily delivered 50mW at under 10% distortion with “flat” batteries supplying just 3V, ie, half the nominated 6V supply voltage (4 x 1.5V cells). The actual circuit specifies either of two bias transistors (AV-2 or OA675), depending on the actual output transistors used. The transistor is fitted with a “flag” heatsink that’s riveted to those used for the two output transis- As it came to me, the set was almost dead. A common problem with old transistor sets is no output at all due to corroded/tarnished contacts on headphone jacks and power switches. By contrast, this set worked but its performance was extremely weak. What was strange was that the signal meter indicated a “strong signal”, with the pointer stuck at the righthand end of its travel. That just had to be wrong but it was also a clue as to the fault. It didn’t take long to find the cause – a bad solder joint between resistor R7 and the meter. And since R7 is transistor TR2’s emitter resistor, this upset the operation of the first IF amplifier stage. Once this joint had been resoldered, the set leapt into action. However, I was unhappy with the performance of the output transistors, so I raided my junkbox and replaced them. I then tested the set and found that I was able to tune stations from one end of the broadcast band to the other. Considering my country location, it was a good result and the Pye Jetliner seemed to be a pretty sensitive set. A note on circuit board removal and replacement: the two dial pointers sit in a channel moulding behind the dial inserts and cannot be removed unless they are set to the top end of the band, so that they align with a couple of slots in the case. Similarly, on replacement, the two pointers must be lined up with these slots, as shown on one of the photos. A bit about noise figures It was time to pop the set onto the test bench to find out just how sensisiliconchip.com.au This photo shows the needle positions on the ‘Radicator’ for various signal strengths at 1400kHz & 600kHz. tive it really was and take a few noise measurements. Before doing that though, I applied contact cleaner to the switches and volume pot, and then gave it a quick alignment check. I measure sensitivity for a 20dB signal-to-noise ratio at 50mW output and at 30% modulation. To meet this 20dB requirement, I first set the input signal for 50mW output. I then turn the modulation off and (hopefully) get only 0.5mW of output. This gives me a S/N power ratio of 20dB. If the noise-only signal is above 0.5mW, the volume control is turned down until the output reaches this level. I then turn the modulation back on and increase the input signal to get the 50mW standard output. In practice, it’s common to juggle the volume control and signal generator output to get 50mW output at 20dB S/N ratio. You may wonder why I don’t simply detune the signal generator or turn it off, as this would give less noise. The reason is that the 20dB figure must be the “on signal” ratio, ie, the ratio of the audio output to the noise in the received signal. How good is it? The frequency response of this set from the volume control to the loudspeaker is 140Hz to 25kHz ±3 dB. So the high end is pretty “snappy” but a few more henries in the transformer windings would have given a better bottom end. Unfortunately, the IF channel (as in most broadcast-band AM radios) is the bottleneck. From antenna to speaker, the frequency response is 140Hz to about 2.8kHz, confirmed by an IF selectivity of -3dB at ±2.8 kHz. At 60dB down, the selectivity is about ±15kHz. The audio performance is pretty siliconchip.com.au good, with a total harmonic distortion of just 3.5% for a 1kHz signal at 10mW output. At 50mW output, the distortion is still just 4%. It rises to around 7% just as the amplifier begins to clip at 250mW output. The transistor-diode’s biasing superiority shows up with a low battery. At 3V (only 0.75V per cell), distortion is still only around 4.2% for 10mW output and is still under 5% at 50mW. Sets with resistor-biased output stages simply can’t match this performance. The Jetliner’s RF sensitivity is out­ standing and is under 3µV (ie, at the antenna terminal) for an output of 50mW. However, at this level, the signal-to-noise ratio is only 13dB at 1400kHz. The sensitivity falls to about 5.5µV if the gain is reduced to give the standard 20dB S/N ratio at this frequency. It’s about the same at 600kHz, ie, 5µV for 50mW output at 20dB S/N ratio. In operation, this set produces a 50mW output at a field strength of 20µV/m at 1400kHz and 55µV/m at 600kHz (both at full gain). The required field strength rises to 50µV/m at 1400kHz to deliver a 20dB S/N ratio though. Where the Jetliner shines is in the IF channel. In fact, its sensitivity is four times better than the Kriesler 41/47’s when taken at the input to the first IF stage – about 50µV compared to 200µV. Since both sets use the same transistors, it seems that the Jetliner’s IF transformers and its improved coupling circuits are the secret. AGC checks Checking the AGC system revealed that the AGC control held the output to a 6dB rise for a signal increase of some 33dB. However, Mullard quoted 55dB with the specified AGC diode (D1), so what was going on? Further checks showed that transistor TR2’s collector voltage was only going to about 4V on full signal, during which the output was becoming distorted. Suspecting a fault in this stage, I tried shorting TR2’s base to ground. This should have turned the transistor off and allowed its collector voltage to rise to at least 4.5V but this didn’t happen. Even with the set turned off, there was still some measurable resistance between TR2’s collector and ground and the logical suspect was the .047μF (47nF) bypass capacitor (C12). This was one of those “red-tipped” highvalue ceramics that’s been notorious for leakage. On removal, it gave a resistance reading of about 10kΩ so it was effectively forming a voltage divider with the 3.9kΩ collector resistor (R8). And that in turn was preventing the AGC diode’s cathode from rising high enough to obtain forward bias. A new “greencap” capacitor fixed the problem. Shorting TR2’s base to ground now resulted in its collector voltage rising to about 4.6V, as expected. More importantly, the AGC circuit was now working correctly with the set now handling a 60dB signal increase for an output power increase of just over 6dB (well in line with the original Mullard specification). So leaky ceramic capacitors are a cause of trouble in low-voltage transistor radios. It’s not just the electrolytic types that can cause problems. Summary The Jetliner is not a pocket set; instead its size and dual-speaker design make it a “picnic portable”. Its sensitivity is one of its main features and the signal-strength indicator makes it possible to accurately tune stations. It’s a simpler set than the Bush TR82C and although the circuit is similar to that used in the Kriesler 41/47, its performance is much better. In fact, its performance is excellent. It meets the manufacturer’s impressive specifications for sensitivity and low-battery performance and my only reservation is the quoted output of 500mW, which I was unable to obtain. Some manufacturers do indicate up to 750mW output (with a 6V supply) for an OC74 push-pull output stage but the alternative 2N217 transistors appear to be lower-rated than the OC74s. It’s possible that the 500mW figure quoted is for an OC74-equipped model. Finally, note that the original circuit diagram for the Pye Jetliner shows TR7’s base connected to its emitter. The base should in fact go to the collector (so that it correctly functions as a bias diode) and the circuit reproduced here has been corrected. Further reading If you’ve not already done so, take a look at Kevin Chant’s excellent website at www.kevinchant.com It’s a free resource for (mainly) Australian vintage radios, and includes circuits, photos SC and parts information. September 2014  97