Fullscreen HTML5Med Res (??MB)HTML4

AUDIO OUT AUDIO OUT L R By Jake Rothman Making a Transistor Radio – Part 2 L ast month, we introduced Reflex radio The active radio The next stage is a major upgrade; we are going to add audio amplification to the crystal set using a transistor and a battery (Fig.14). All this does is make the crystal set a bit louder. It was just sufficient for me to hear it. I was disappointed to find the original circuits suggested in the Ladybird book did not work properly, since the biasing was all out. The main problem was that the diode had a DC voltage on it. This was fixed by adding a coupling capacitor and re-jigging the bias values to get the output to sit at half-rail, as shown in Fig.15. This could be made insensitive to transistor Hfe variations by including a 1kΩ emitter resistor. This should be bypassed with a capacitor of around 22µF, as shown in Fig.16. Replacing the 2.7kΩ resistor and crystal earpiece with the telephone earpiece sounded much better (see circuit in Fig.17). The crystal types do have a rather ‘tinny’ response. 1 0kΩ O A8 1 10 µ F .7kΩ – O C7 1 + Given the simple design we are working with, there was only one simple way to get a usable signal and that was to build a regenerative or reflex radio circuit, as shown in Fig.18. Although this design has both gain and selectivity, it only needs one transistor, which is effectively used twice; combining both an RF amplifier and audio amplifier stage together. The audio load is the collector resistor R2, while the RF load is the choke L2. In the old days, a transistor was much pricier than an inductor. Today the situation is reversed, and at least two transistors would be used, but with a doubling of power consumption. Positive feedback is also applied, ‘regeneration’ or ‘reaction’ as it is know in wireless parlance. This is used to ‘peak’ the tuned circuit. As with all positive feedback, there is a real risk that too much can result in oscillation, causing the radio to act like a transmitter! The two-diode network is for detecting the audio and is basically a voltage doubler circuit. The output of this is fed back into the transistors input again. Germanium diodes work best here, since silicon types can cause instability. The original Ladybird reflex construction is shown in Fig.19. 9 V 5 0 turns + N C 5 0 0 pF 10kΩ 10kΩ Fig.15. Adding a coupling capacitor and adjusting biasing on Fig.14. –9 V 68kΩ 10 µ F 4.7kΩ O C7 1 –5 V k + the Ladybird Transistor Radio and got as far as building a simple crystal set. This month, we go active and build the full transistor version, plus we’ll suggest some simple upgrades to improve performance but still retain the spirit of the germanium-based electronics. 10kΩ 1kΩ + 0 V Fig.16. Crystal set with improved singlestage amplifier. Adding an emitter resistor and a bypass capacitor to the circuit in Fig.15 gives better transistor variation tolerance. Component notes a) b) .7kΩ 1 0kΩ Aerial O A8 1 O C7 1 Crystal earpiece 5 0 0 pF – O A8 1 9 V + 19 0 µ H .7kΩ – 5 0 turns O C7 1 9 V + 5 0 0 pF kΩ For a retrospective project like this, the components have to look the part – no anonymous black surface-mount devices here. Sourcing classic glass-cased diodes is a good start. Screws Fig.14. Adding a single-stage amplifier to the crystal set: a) transistor with no bias; b) biased transistor, but the base voltage stops the diode conducting. Solid brass No.4 or No.6, 3/8-inch (10mm long × 2mm wide) screws with a countersunk head are used. These are normally only available in slotted head. The cups are No.6 (11mm across with 3.5mm hole). I was worried about the lacquer spoiling the contact, but this was 52 Practical Electronics | April | 2021 N C E arth 71 .4kΩ impedance V C1 5 0 0 pF R 2 4.7kΩ R F positive feedback L 1 19 0 µ H + 10 µ F V C2 10 pF 00Ω resistance 1.8 R 1 0kΩ C3 10 µ F + 4 5 turns C2 4 7 0 pF 5 turns 4.7kΩ R F input 0 V R F AF + R F D2 O A8 1 O C4 5 C1 10 nF Fig.17. Adding a telephone earpiece; this sounds better than the crystal unit. not the case. I used No. 5/6 cups with No.4 screws and had no problems. The brass-plated types do work but may oxidise with time. Solid brass cups are available, but expensive. Audio O utput L 2 2. 5 mH R F C R epanco CH 5 Audio input Detector D1 O A8 1 E arth Fig.18. Reflex radio circuit – the transistor is used twice, note audio (red) and radio (blue) frequency pathways. Transformers One way of improving the sensitivity of the crystal set is to feed a crystal earpiece via a step-down 20kΩ-to-1kΩ driver transformer, such as the LT44 shown in Fig.20, to increase the load impedance further. The inductance of this resonates with the earpiece’s capacitance to further boost output. This transformer also works well with the old balanced-armature telephone receivers illustrated in Fig.21. These have an AC impedance of 2.4kΩ and a DC resistance of 300Ω. Using the transformer in the collector of the transistor amp with the telephone earpiece finally increased output to a usable level. Output transformer The output transformer is that hobbyist’s favourite, the Eagle LT700. This has an impedance ratio of 1.2kΩ to 3.2Ω. The output of the Ladybird Radio is very low, 0 V 1 L T4 4 0kΩ 1kΩ 81 L T4 4 0kΩ 1kΩ R 1 1 0 1 00 CT N C 10 earpiece 10kΩ –9 V CT N C B ias 68kΩ 71 Crystal earpiece + n 1 0kΩ –9 V Aerial –9 V M agnetic earpiece 1kΩ + 0 V Fig.20. Adding an audio transformer to the crystal set increases output. in the order of about 50mW, just over 1V peak to peak into 3Ω. If double output power is needed, an LT726, or the bigger LT730 could be used, which have 500Ω primaries. Their secondaries can drive 3Ω or 8Ω speakers. If you can get one, the Repanco or RS T/T4 is the best, since it is designed to have DC flowing through the core. Overseas readers could use the Xicon 42TU400-RC from Mouser. The standing current will have to be increased to 14mA for these transformers by reducing bias resistor RBias (Fig.20) to 51kΩ, which will be too much for a PP3 battery. I decided the original LT700 was overall the best for the job and put it back after trying the others. I also tried a Philips 80mm 150Ω high-impedance speaker wired directly, and that worked well. Germanium transistors Watch out for types that are too ‘good’. Using non-stabilised bias (as in Fig.23) caused most transistors to turn hard-on. The low-gain OC70 might have been worked better here. I dropped bias resistor Fig.19. Construction of the reflex circuit. Practical Electronics | April | 2021 Fig.21. Telephone earpiece – use of an LT44 transformer also improves output. 53 –9 V 100kΩ L T7 0 0 P rimary impedance 1. kΩ N C Ω –8 . 8 V + 10 µ F AC18 8 5 . 5 mA 10kΩ CW L og Ω V olume + 4 7 0 µ F 6 .3 V 0 V Fig.22. Uprated and re-biased output amplifier circuit to avoid thermal runaway. log volume potentiometer because I didn’t have any 10kΩ types. I then had to reduce bias resistor R7 to 100kΩ. This set the standing current to around 5.5mA. The output was 42mW into 3Ω. These bias improvements add six extra components to the whole radio. I see now why George Dobbs omitted them in his original circuit (Fig.23), too much complexity for children. The modified Ladybird circuit is shown in Fig.24. Radio-frequency transistor The radio stage specified an OC45. I only had an old one which seemed to be rather noisy. I tried a ‘new old stock’ lower noise Toshiba 2SA53 (Fig.25) which reduced it, but there was still an additional noise. Shorting out the tuned circuit revealed a steady hiss. Initially, I didn’t bother adding emitter stabilisation to this stage, since the intermediate frequency (IF) amplifier transistors specified have less Hfe spread, typically 50, a cutoff frequency (F t) of 4-6MHz. Proper RF types such as the OC44, may again be too good (Hfe = 80, Ft = 8-12MHz). I reckon George Dobbs was inspired by some very similar circuits by A Sapciyan’s Pocket Reflex Receiver in the August 1968 issue of Radio Constructor magazine (see p.27) and the Reflex-3 in the December 1968 issue (p.302) issue. These used an AF114 or AF124 transistor, but I would be worried about audio frequency noise using an alloy-diffused germanium device. Endless mods Putting in proper potential divider bias (R1, R12) reduced the RF performance, because its shunt impedance damped the tuned circuit. A load resistor (R11) and coupling capacitor (C11) were also required for the detector circuit to avoid its DC output upsetting the bias. Isolating the diodes from + R5 from 33kΩ to 4.7kΩ to get half rail 4.5V across the collector load resistor, R6. Better than tweaking, make the circuit immune to current gain (Hfe) spreads by putting in a decoupled emitter resistor: 1kΩ and 22µF in the first audio stage –9 V Aerial and 22Ω and 470µF R 1 R 3 7 mA 0kΩ in the output. 1kΩ 1mA 5 . 5 mA 0 . 5 mA R 2 00kΩ C4 C3 4.7kΩ V C2 I noticed stabili10 0 µ F + 10 µ F 0 . 5 - 10 pF sation was applied L T7 0 0 R 4 1 0kΩ in the audio ampli1. kΩ V C1 R 7 L 2 R 6 L 1 5 0 0 pF fier on page 44 of N C 5 mH 4.7kΩ 1 0kΩ 19 0 µ H 100kΩ R F C Learn About SimR epanco CH 2/ CH 5 Ω 4 5 turns –4 . 5 V –8 . 8 V ple Electronics. In – TR 2 –1. 5 V C2 O C7 1 C5 5 turns the radio section, 4 7 0 pF 10 µ F TR 1 9 V TR 3 I found reducing O C4 5 P P 3 O C7 1 the 390kΩ resis+ V R 1 tor to 300kΩ gave D2 D1 CW C1 10kΩ O A8 1 O A8 1 10 nF R 5 more accurate bi4.7kΩ kΩ L og asing with 3.5V 4.7kΩ across the collec0 V tor load resistor R2. B ias tweak values shown in red red After a few hours, I E arth found that the output transistor OC71 Fig.23. Full original Ladybird radio circuit. Making it as simple as possible using just base bias made it dependent on went into thermal transistor characteristics, which unfortunately made it unlikely to work without adjusting resistors, shown in red. runaway and became permanently leaky. Interesting–9 V ly, this process was Aerial C4 R 3 7 mA kicked off by the R 1 10 0 µ F 1kΩ 1mA 5 . 5 mA 0 . 5 mA R 2 C9 * 00kΩ RF emissions from C3 4.7kΩ V C2 + + 220 µ F 10 µ F 0 . 5 - 10 pF touching a nearby L T7 0 0 LED lamp. I vaguely 1. kΩ V C1 L 2 R 4 R 6 R 7 remember my OC71 L 1 5 0 0 pF N C 5 mH 68kΩ 4.7kΩ 100kΩ 19 0 µ H R F C getting hot in 1972, R epanco CH 2 Ω 4 5 turns –5 V which denotes –8 . 8 V TR 2 – O C7 1 –1. 5 V C2 death in germaC5 5 turns 4 7 0 pF 10 µ F TR 3 TR 1 nium transistors. 9 V AC18 8 O C4 5 P P 3 Runaway is quite + R 8 * C6 * l i k e l y t o o c c u r, V R 1 D2 D1 CW 1kΩ C1 10 nF 4.7kΩ O A7 0 O A7 0 since the trans4 7 nF C8 * R 5 L og + + 4 7 0 µ F 10kΩ former secondary R 9 * C7 * Ω 22µ F has a DC resistance 0 V of only 60Ω. I reM odified components in red Added components marked with an asterisk ‘ * ’ placed the OC71 E arth with a higher power AC188 (Fig.22) Also, I used a 4.7kΩ Fig.24. Modified full Ladybird circuit. C1 is increased for better bass and C6 is added for RF filtering. + + + 54 Practical Electronics | April | 2021 Radio-frequency choke (RFC) The original design used a Repanco 2.5mH CH1. Bourns also do a suitable type, the 2.4mH pie-wound 4666-RC from Mouser, shown in Fig.29. 5mH Repanco CH2 types are still obtainable, as shown in Fig.30. However, these seemed to measure low at 3.5mH, which is fine for the Ladybird circuit. Ferrite rod Fig.25. Long-wave version of RF stage using a 2SA53 transistor. Tuning capacitor the DC base current removed their forward voltage bias, further reducing sensitivity, but also reducing noise (it wasn’t the transistor after all). Five extra components and slightly worse RF performance, but better audio quality. The final modified circuit is shown in Fig.26. With this design, using the 8-inch LW (long-wave) ferrite rod meant the trimmer capacitor didn’t peak up the gain, it seemed to reduce it. The layout was also more critical. L2 needed to be at right angles to the ferrite rod to reduce coupling, with the output being reduced if it was in-line. I realised there was a unique synergy to the original simple circuit. It is well worth adjusting the bias resistor R1 for the occasional deviant transistor. Finally, a 220µF bypass capacitor (C9) should be added across the power rail so the radio does not burst into oscillation as the battery ages. This is good practice in any circuit. This new circuit with the other mods in the audio stages is at the limit for screw and cup construction (see Fig.34). Maybe a PCB is now needed. The Jackson Dilecon solid-dielectric type is no longer available. Only the single-vane 100pF type is still sold, as shown in Fig.27. However, a Plessey variable capacitor with a total capacitance of 500pF (364pF and 186pF) shown in Fig.28 can still be obtained for about £5.00. This is what I had to use. It’s superior anyway, having slow-motion drive and ball bearings. Also, air-spaced capacitors have lower losses than solid-dielectric types. Since the radio can only get a couple of stations, a 50pF trimmer capacitor in parallel with a fixed 150pF capacitor could be used to save cash if you’re happy to listen to only one station. Trimmer cap I used a 40pF mica ‘postage-stamp’ trimmer, although 10pF was specified, the unit I used could go low enough to just about 1pF. The tags have to be bent 90 degrees to lie flat for clamping with the cups. –9 V R 1 68kΩ R 2 4.7kΩ C3 10 µ F + V C1 3 6 5 pF TR 1 2SA5 3 C11 10 µ F C1 4 7 nF 5 turns R 12 10kΩ C4 10 0 µ F C2 4 7 0 pF + L 1 4 . 5 mH L ongwave ferrite rod It’s sad, but I now know why my childhood radio didn’t work. There were too many little things wrong which at the time I didn’t know how to test for. Also, like many beginners I built the whole thing in one go without testing along the way. If I had had an older mentor with a multimeter it would have worked, because the biasing errors would have been spotted. I guess George Dobbs was using the surplus OC71s which had very low Hfe and were the cheapest transistor in the UK at the time. I got my mum to buy full-spec devices from Electro Value Ltd. Emitter resistors should have been used to make the design less Hfe dependent, or O utput L 2 R epanco CH 3 7 . 5 mH –5 V Sob story R 3 1kΩ 1mA V C2 1 – 4 0 pF I just used a 75mm (three-inch) piece of standard 10mm (3/8-inch) ferrite rod with 50 turns of 36 SWG enamelled wire tapped at five turns and secured with Sellotape. This gave a total winding inductance of around 190µH. This can vary according to the grade of ferrite used. I had no idea what mine was, so this is another area where tweaking must be employed to get full tuning range. One Ladybird design fault is the ferrite rod bare wire-loop mounting technique which caused two shorted turns. I used a pair of 9.5mm plastic ‘P’ clips. Hand-winding coils was normal for electronics people back in the 1970s. For those who dread doing it, I have made some ready wound coils available. I recommend a longer eight-inch rod for better reception if you’re not using an external aerial. + D2 O A7 0 D1 O A7 0 + 0 .8 8 V R 10 1kΩ + C10 22µ F R 11 1kΩ 0 V Fig.26. Modified RF circuit with lower noise. Practical Electronics | April | 2021 Fig.27. The original Jackson Dilecon tuning capacitor. 55 Fig.29. Mouser sell a suitable RFC (RF choke); this one is from Bourns. Note the multi-pie windings to reduce parasitic capacitance. Fig.28. The recommended Plessey tuning capacitor. Still available and a mechanical work of art. Note a rightangle bracket is needed on the front with a 4BA bolt for mounting. an additional section written on tweaking the bias resistors to get the right voltages. The mechanical solder-less construction worked surprisingly well though. I was amazed how sensitive the simple reflex circuit was when the positive feedback trimmer capacitor was peaked up just before oscillation occurred. It even worked well with no aerial and earth. With just the ferrite rod alone, BBC Radio 5 Live filled the workshop. However, apart from about ten whistles and power supply buzzes, that was all I could get. There was plenty of sensitivity, but little selectivity. The finished radio is shown in Fig.31. Fig.31. The finished radio – original version. Fig.30. Repanco CH2 RFC; this uses wave winding and Litz wire to minimise losses. Home service I suspect most PE readers would prefer a long-wave version to get Radio 4 198kHz and possibly our French readers need LW as well. I was surprised that I received a station unknown to me (in mid-Wales): RTE Radio 1 from Ireland on 252kHz. In the evening, I even got a French station. To get LW requires a few changes to the tuned circuit. I used a ready-wound 4.5mH LW coil with an 8-inch ferrite rod. This consists of two separate windings which have to be joined together, as shown in Fig.32. The smaller section of the tuning capacitor was disconnected, using only the 360pF part. A reduction in tuning capacitance is also needed for the ready-wound MW coil. For LW, C1 was increased to 47nF and L2, the RFC, to 7.5mH (a CH3 type). This modified radio is shown in Fig.33 with an old sensitive Goodmans 3Ω loudspeaker. The final version with the bias mods for the reflex stage is shown in Fig.34. They’re a bit more complex than the original Ladybird design, but then I can’t resist trying to improve everything. (I did, however, draw the line at modifying it for silicon semiconductors.) Consuming a miserly 7mA, I could listen to news all day with just a PP3 battery. It was also a fantastic electromagnetic interference ‘sniffer’, picking up emissions from my lamps and test gear ‘unseen’ by normal radios. Components All components marked with a ‘*’ are for modifications from the original design. Availability I can supply all parts or kits: Tel 01597 829102 Email jrothman1962<at>gmail.com Fig.32. Long-wave aerial coil; note how the two separate windings are linked underneath to get tap. 56 Also, a very good supplier is Birketts, who have been supporting the Ladybird and other George Dobbs designs for years. Best to phone Thursdays when Judy Birkett is often in. They don’t operate online, which for this project seems entirely appropriate! Practical Electronics | April | 2021 Fig.33. The completed radio, LW version, with main bias modifications only. VC1 365 + 186 pF dual-gang Plessey with slow-motion drive. VC2 mica trimmer 10pF, or a plastic-film Transistors TR1 OC45 or *2SA53 radio-frequency germanium TR2, TR3 OC71 (TR3 * AC188 or AC153) audio-frequency germanium Diodes D1, D2 OA81 or * OA70, OA85, CG92, OA91, IN60 germanium point contact Fig.34. The completed radio, LW version with all bias modifications. Note how L2 is twisted to reduce coupling to the long ferrite rod. Inductors L1 Ferrite rod 3/8-inch diameter 3 to 8 inches (longer is better) 1 yard / metre of 36swg enamelled copper wire plus sticky tape MW or LW coils with coupling winding on formers. L2 Repanco choke 5mH CH2 for MW or * 7.5mH CH3 for LW. I got my brass screws and cups from: Marches Ironmongery Llandrindod Wells LD1 6DF 01597 823 822 http://bit.ly/pe-apr21-ao1 http://bit.ly/pe-apr21-ao2 J Birkett, Radio Component Suppliers, 25 The Strait, Lincoln LN2 1JF. Tel 01522 520 767 Resistors Use big old-fashioned half or one-watt types. I mainly used Electrosil tin-oxide TR5 types with long leads. For visual authenticity use carbon composition devices; 10% tolerance is good enough. R1 390kΩ (*300kΩ original, 68kΩ final modification) R2 4.7kΩ R3 1kΩ R4 150kΩ (*68kΩ) R5 33kΩ (*4.7kΩ original, 10k final mod) R6 4.7kΩ R7 150kΩ (*100kΩ original) R8 *1kΩ R9 *22Ω R10 *1k R11 *10kΩ R12 *10kΩ VR1 10kΩ log (*4.7kΩ log) Your best bet since MAPLIN Chock-a-Block with Stock Visit: www.cricklewoodelectronics.com O r phone our friendly knowledgeable staff on 0 2 0 8 4 5 2 0 1 6 1 Components • Audio • Video • Connectors • Cables Arduino • Test Equipment etc, etc Capacitors. Axial types fit best; I used the blue Mullard 017 electrolytics for appearance. Tolerance or voltage is uncritical. C1 10nF ceramic disc (*47n C296 polyester) C2 470pF silvered mica or ceramic C3,5 10µF 25V C4 100µF 10V C6* 10n ceramic C7, 10* 22µF 25V C8* 470µF 6.3V C9* 220µF 10V C11* 10µF 10V tantalum Practical Electronics | April | 2021 V is it o u r S h o p , C a ll o r B u y o n lin e a t: w w w .c r ic k le w o o d e le c tr o n ic s .c o m 0 2 0 8 4 5 2 0 1 6 1 V is it o u r s h o p a t: 4 0 -4 2 C r ic k le w o o d B r o a d w a y Lo n d o n NW 2 3 E T 57