Fullscreen HTML5Med Res (??MB)HTML4

AUDIO OUT AUDIO OUT L R By Jake Rothman Theremin Audio Amplifier – Part 3 L ast month, we completed a silicon-transistor-based amplifier fo the PE Theremin, but I concluded the article with a promise to look at a germanium-transistor version. This actually turned into quite a fascinating project, as you will see. Lost values When I get a break from eyeball-stressing SMT (surface-mount technology) industrial jobs, it’s a relief to get back to some old-fashioned electronics. I recently got really fed up with unmarked SMT components. It started with capacitors, but now even resistors are being supplied unmarked, as shown in Fig.1. This makes fault finding on pre-production prototypes very difficult. When the unit fails a few years later it can’t be fixed without full service data, and who supplies that today? delivery of components rendered obsolete by RoHS directives, my childhood electronics excitement came back. Behold! – there were unopened packs of 1974 Siemens germanium transistors. There was also a bundle of Siemens books, including one called Design Examples of Semiconductor Circuits, from 1969. The section titled, ‘Audio-frequency Amplifiers’ on p.7 compared the use of germanium output transistors with the then relatively new silicon planar transistors. It said the germanium types were superior for low voltage, which is to be expected, since their turn-on voltage is around 0.1V as opposed to silicon’s 0.6V. What intrigued me, however, was the statement that they had better linearity, which should mean lower distortion. The amplifier designer and Bradford University academic Arthur Bailey also said the same in his Wireless World article, ‘High Performance Transistor Amplifier’ (see p.543, November 1966). I’m always a bit wary about ‘linearity’ and other spurious audio claims, but there is only one way to be sure – build and compare. On pages 14 and 15, the Siemens book had two almost identical four-transistor amplifier circuits for a 9V supply and 8Ω speakers, see Fig.3. The main difference was one used the germanium output transistors (AC153K and AC176K) in the box and the other used silicon (BC140 and BC160). Biasing requirements accounted for the other circuit differences. The claimed output power was 1.1W for the germanium and 0.7W for the silicon. This sort of germanium amplifier was used in Robert’s Radios, such as the R505, until the early 1970s. They still sound good today – and they don’t eat batteries. Maybe their designers were onto something. Nice old junk Sifting through some musty boxes (Fig.2) courtesy of yet another lock-down Fig.1. It started with ceramic chip capacitors. Now, even some resistors have no markings, such as these Vishay CRCW080522K0FKEA resistors from RS. Just black and blue rectangles (R84 and R15). This is what I have to put up with in my professional life – I am a human, not a pick-and-place machine! 46 Fig.2. A good dollop of old ‘proper’ components with wires and markings. Siemens germanium output transistors and Philips ‘Minipoco’ 1% tolerance lead-foil polystyrene capacitors. Still 1% accurate after 35 years. Practical Electronics | January | 2021 A udio input – 9 V Note: A C 153K and A C 17 6 K transistors are germ anium , the rest are silicon. Ω 0Ω O utput 1. 1W 15m A total I q Max supply current 19 0 m A A C 153K 0Ω Ω + B A 10 3 or 1N4 14 8 . kΩ 0kΩ 1µ F 50 0 µ F 0. Ω NT C 1 0Ω 0. Ω 10Ω B C 16 8 330 pF + n Comparison circuit Fig.3. The Siemens recommended circuit for their transistors. It’s very efficient but needs matched output devices. Also, the signal reference 0V and speaker reference 0V are different, which gives poor power supply rejection ratio. We won’t build these. A C 17 6 K 10 µ F + 100kΩ B C 258 1. kΩ + 250 µ F 0 V kΩ 10 µ F + 9 V + Ω 0Ω Note: all transistors are silicon. 0Ω B C 14 0 3x + O utput 0 . 7 W 22m A total I q Max supply current 150 m A 50 0 µ F Power supply Since germanium NPN transistors are rare, the circuit was inverted to use a negative power rail and positive earth. This way only one NPN transistor was needed, the rest are PNP devices, normal Ico = grounded em itter leaka ge current 0 . 15m A <at> 25° C , typically doub les ev ery 5° C 1Ω B A 10 3 To perform a practical germanium vs silicon comparison, I decided to use the PE Theremin Amplifier circuit and PCB from November/December 2020’s Audio Out. I installed transistor sockets on the board – always a good idea when trying unusual parts. I used cut strips of SIL turned-pin IC sockets for this (see Fig.4), since real transistor sockets are expensive. I decided to go for a standard 9V, 8Ω set-up to see if I could replicate the Siemens amplifiers’ results. There is no point in trying to go for a micro-power design with germanium, since their high leakage currents prohibit this. When experimenting with germanium transistors it is important to check the leakage current (I cbo) of old stock devices. This is the current that flows between collector and emitter with the base open-circuit. I use a Peak DCA75 Analyser for this (Fig.5), but a 5V power source and multimeter, as shown in Fig.6, will also do the job. . kΩ 0kΩ A udio input 1µ F 2. 2nF + 250 µ F 100kΩ 4 . 5V O C 7 1 ( typical type) + B C 10 8 B C 16 0 B ase lef t open hence the ‘ o’ in Ico 3. 3nF 1. kΩ – mA 1Ω B C 17 8 + . kΩ 0 V 10Ω 0 V Fig.6. Circuit for checking leakage current. Fig.4. When trying out unusual transistors, use SIL sockets for easy substitution during R&D. Practical Electronics | January | 2021 Fig.5. It is important to check the leakage currents of germanium transistors. Anything above 0.25mA is suspect, except for power types, where the limit is 2mA. 47 1. 7 m A R 5 1. kΩ R 6 1. kΩ C 4 4 7 0 µ F 16 V + C 3 22µ F 10 V + – 5. 4 V C lip R 14 Not used T R 4 * B C 14 3 * T R 4 / 5 with T O 5 clip- on heatsink s 7 20 m W output R 7 1kΩ T R 3 B C 327 + D1 B A T 8 6 VR 2 kΩ R 3 100kΩ C 11 10 µ F 6 V R 11 0. Ω R 12 0. Ω C 7 4 7 0 µ F 10 V – 4 . 2V + R 2 47kΩ R 13 10Ω I q set C 9 Not used R 4 47Ω VR 1 00Ω VR 1: DC m id- point adj ust R 8 1 0Ω C 2 4 7 µ F 6 V R 9 47Ω + – 9 V – 4 .8 V T R 1 B C 557 B D2 R ed A udio input I q = 29 m A Ω C 6 10 0 nF T R 5* B C 138 + + R 1 4.7kΩ C 8 10 0 0 µ F 16 V + 10 m A T R 2 B C 557 B C 10 Not used C 1 10 µ F 10 V T ant R 10 10Ω C 5 10 0 µ F 10 V 0 V P ositive earth Fig.7. To start the development of the germanium amplifier, a silicon amplifier was developed using a negative power rail with positive earth. This way, only one NPN device was needed, with the rest being PNP types, paving the way for step-by-step substitution with germanium devices. creating your own ‘Robert’s Radio sound’ then I’ve even got some old alnico magnet Celestion 6×4-inch speakers for sale). for germanium. Of course, when doing this ‘mirroring’, all the electrolytic capacitor and diodes have to be reversed too. This is no problem with the silicon circuit, since the transistors cost peanuts, whatever the polarity. Later I’ll use this positive-earth amplifier design with some germanium fuzz circuits and Mullard LP1171/69 radio modules, as used in the Robert’s R600. (If you fancy e b underside c T O 5 B C 138 / 4 0 underside R 6 pin vi ew 1. kΩ e b c ( case) + C 1 10 µ F 10 V T ant R 1 4.7kΩ C 4 + 4 7 0 µ F 16 V 1. 7 m A R 5 1. kΩ + R 14 Not used C 9 Not used VR 2 kΩ + C 11 10 µ F 6 V R 12 0. Ω + C 8 10 0 0 µ F 16 V + + 9 V C lips at 19 0 m A supply current * T R 4 / 5 with T O 5 clip- on heatsink s R 11 0. Ω + 4 . 2V 7 20 m W output C 7 4 7 0 µ F 10 R 13 10Ω C 6 10 0 nF Ω + VR 1 00Ω R 8 1 0Ω C 2 4 7 µ F 6 V R 9 47Ω T R 5* B C 14 3 I q = 29 m A S tandard silicon am plif ier with negative earth I q set R 4 47Ω VR 1: DC m id- point adj ust R 7 1kΩ T R 3 B C 337 D1 B A T 8 6 R 3 100kΩ T R 4 * B C 138 + 4 .8 V T R 1 B C 54 9 C C lip R 2 47kΩ 10 m A T R 2 B C 54 9 C D2 R ed A udio input R 10 10Ω C 3 22µ F ? ? V C 10 Not used + 5. 4 V The first task is to get the silicon version working. That way any transistor blowups will cost 5p rather than possibly £1. To do this, we’ll take the original circuit (Fig.9, p.65, PE November 2020) and + T O 9 2 B C 54 9 C pin vi ew Germanium gestation C 5 10 0 µ F 10 V 0 V Fig.8. The silicon circuit with standard negative earth, a useful medium-power amplifier. 48 scale up the currents by a factor of three for the lower speaker impedance of 8Ω, rather than the original 25Ω. This entails dividing most resistor values by three. Capacitors will have to be increased by three to take into account the lower impedances. Finally, the power rails and all other polarised components are flipped. There are then resistor tweaks to bring the biasing into line. This new circuit is shown in Fig.7. An interesting high-frequency instability occurred where the frequency of oscillation was so high my 40MHz ‘scope couldn’t see it, but it was enough to completely mess-up the DC conditions and drive one output transistor to partly cut off. The clue was when I put my finger near R6 and the problem just cured itself, a sign of VHF oscillations. Ironically, removing stabilising capacitor C10 fixed it. When the new silicon circuit was rebuilt using a standard positive rail (Fig.8) the instability vanished and there was no problem with C10. I have heard that there is more likely to be problems with silicon PNPs; they have more junction capacitance modulation (Early effect) than NPNs. The germanium transistors were then gradually plugged in, checking along the way with a multimeter, ‘scope and signal generator. The only significant changes were to the DC biasing, with R2 being changed from 47kΩ to 15kΩ. Of course, the output current bias (Vbe multiplier) transistor has to be germanium if the output devices are. This can be an AC153 or a special low-voltage transistor designed/selected especially for the job, such as an AC169. Using an NKT214 didn’t work because the Iq could not be turned down to zero. Do note that some AC169s only have two wires, they are essentially just a 0.13V diode. The metal-cased three-wire ones have to be checked on a transistor tester to determine the connections. Germanium-alloy junction transistors are generally ten-times slower than silicon planar types, giving inferior square-wave and high-frequency response. With TR1 and TR2 (in Fig.9) being 2MHz-Ft (frequency at which gain falls to unity) NKT214 audio types, the –3dB point was about 15kHz, fine for guitar and AM radio. The square-wave response at 1kHz 2V pk-pk output had overshoot on it from TR2. Connecting a 33pF capacitor (C10) removed this. The high-frequency response was brought up to Hi-Fi (40kHz bandwidth) standards by putting in germanium 2SA12 RF transistors, as used in AM radio oscillators and leaving C10 out. These have an Ft of around 10MHz. Finally, we had a germanium amplifier with good performance. The circuit is shown in Fig.9. Practical Electronics | January | 2021 A C 17 6 K / A C 153K , X O 4 packa ge R ed dot ( NK T 214 only) b Mounting hole e c underside e T O 1 packa ge NK T 214 / 2S A 12/ A C 153 c b 1. 7 m A R 5 1. kΩ 17 m A R 6 1. kΩ * Use C 10 with NK T 214 f or T R 2 R 1 4.7kΩ VR 1 00Ω + – 9 V C lips at 19 0 m A supply current www.poscope.com/epe Germ anium am plif ier with positive earth 8 8 0 m W output R 7 C lip 0Ω T R 3 A C 16 9 + VR 2 kΩ C 11 10 µ F 6 V R 11 0. Ω R 12 0. Ω – 4 . 3V R 8 1 0Ω C 2 4 7 µ F 6 V R 9 47Ω T R 5 A C 17 6 K on heatsink C 7 4 7 0 µ F 10 V R 13 10Ω I q set R 4 47Ω VR 1: DC m id- point adj ust 3m A D1 C G9 2 or O A 9 1 R 3 100kΩ C 9 Not used T R 4 A C 153K on heatsink I q = 22m A + R 2 1 kΩ T R 2 NK T 214 or 2S A 12 C 8 10 0 0 µ F 16 V + – 4 . 5V D2 5. 1V A udio input R 14 Not used T R 1 NK T 214 or 2S A 12 R 10 10Ω C 3 22µ F 10 V C 10 * 33pF – 4 .7 V C 4 4 7 0 µ F 16 V + C 6 10 0 nF Ω + + C 1 10 µ F 10 V T ant + c I ndent C 5 10 0 µ F 10 V 0 V P ositive earth Fig.9. The final circuit using germanium transistors. Output transistor selection For the silicon amplifier, standard TO5 metal-can devices were used; in this case BC138 and BC143 – just because they were in the drawer. They can comfortably deliver output powers up to 2W with standard clip-on heatsinks. The germanium equivalents are the AC153K (PNP) and AC176K (NPN) shown in Fig10. The type with the K suffix was used because they have a convenient hole to mount on a metal bracket for heatsinking. The XO4 and TO1 germanium cases are electrically isolated, the heat passing from the junction to the case via a white paste filling of aluminium oxide and silicone grease. There are few germanium complementary pairs available. Alternatives are AC128/176 and AC188/187. For higher powers up to 6W the AD161/162 are often used (Fig.11). These TO66 devices are the germanium equivalent of the silicon TO126 BD135 and BD136 types. Next month That’s all for this month. In Part 4, we’ll finish with the component options and examine the frequency responses. Fig.10. The Siemens AC176K and AC153K complementary output transistors. These are still used today in the very expensive EMS VCS3 synthesiser for their ‘musical’ sound. Practical Electronics | January | 2021 Fig.11. The Mullard AD161/2 output transistors (right). Very popular in the 1970s. Robert’s used them in their RM50 table radio along with the Celestion speaker in Fig.18. Note their small TO66 case compared to the standard TO3 (left). - USB - Ethernet - Web server - Modbus - CNC (Mach3/4) - IO - PWM - Encoders - LCD - Analog inputs - Compact PLC - up to 256 - up to 32 microsteps microsteps - 50 V / 6 A - 30 V / 2.5 A - USB configuration - Isolated PoScope Mega1+ PoScope Mega50 - up to 50MS/s - resolution up to 12bit - Lowest power consumption - Smallest and lightest - 7 in 1: Oscilloscope, FFT, X/Y, Recorder, Logic Analyzer, Protocol decoder, Signal generator 49 Make it with Micromite Phil Boyce – hands on with the mighty PIC-powered, BASIC microcontroller Part 24: Counting pulses, rotary encoders and a digital safe T his month, we are going to explore three useful MMBASIC commands that relate to counting digital pulses. Using these built-in commands greatly simplifies tasks such as measuring the frequency of a signal, measuring the time period of a pulse cycle, or simply counting the number of pulses present on an input pin. Linked to these topics is something that follows on nicely from last month, where we used a rotary potentiometer to generate a voltage between 0V and 3.3V, which in turn was used to control the position of a servo actuator arm. Several readers contacted us and asked if there was an alternative to a potentiometer; something that can instead be continually turned in either direction (unlike a potentiometer which is mechanically limited – typically to around 270°‚ not even a complete 360° revolution). This is where rotaryencoders come into play, and if you have never used one before, they can be an extremely useful input device. To make this a fun, we will show you how to use a rotary encoder to simulate entering a digital safe combination number. What exactly are digital pulses? Digital pulses comprise transitions between low and high logic levels on a signal line. The result is a digital signal. These can vary considerably, yet are often simply drawn as the examples shown in Fig.1. The pulses in a digital signal can be repetitive and symmetrical; an example is a square-wave – see Fig.1a. Here they are shown with a fixed frequency (ie, the length of a pulse cycle is constant), and with a duty cycle of 50% (meaning that for 50% of the pulse cycle, the logic level Questions? Please email Phil at: contactus<at>micromite.org 50 is high, and for the remaining P ulse cycle time it is at a low logic level). We have seen, and used, square-waves earlier in the 1b series when we used a PWM signal to drive a piezo sounder to generate musical notes. 1c Pulses can also be repetitive but non-symmetrical ie, where the frequency is constant, but Fig.1. Examples of various digital signals. 1a represents the duty cycle is not 50% – a square wave, meaning a duty cycle of 50%. 1b see Fig.1b. A great example represents a digital signal with the same frequency (ie, of this is a servomotor signal fixed time-cycle) as 1a, but with a lower duty-cycle. 1c (ie, PWM signal) where, as is a random signal with a varying frequency (and cycle we saw last month, the duty time period) and with varying duty-cycles. cycle controls the position of a servo motor actuator arm. Digital pulses that the boundaries of each pulse cycle are may also be random (ie, non-repetitive indicated by the dotted lines in Fig.1a and and non-symmetrical) – see Fig.1c where Fig.1b. In terms of logic-level transitions, neither the frequency nor the duty cycle a pulse cycle in a digital signal is shown is constant. in Fig.1 as starting with a low-to-high Consider each of these digital signals transition followed by a high-to-low, and being drawn as a graph where the x-axis finishes on the next low-to-high transition represents time, and the y-axis represents (which is effectively the start of the next the digital logic level; ie, either a logic pulse cycle, and so on…). Understanding high (3.3V) or logic low (0V). The point this concept of a pulse cycle allows us here is that at any moment in time, the to measure two important parameters – state of the digital signal can be considered time period and frequency – by simply as either a low or a high logic state. detecting logic-level transitions. If we were to use a Micromite input pin to read the logic level of a digital Pulse time-period signal, then by detecting the transitions So, having just explained what a pulse between the low/high states within the cycle is in terms of logic-level transitions, signal, we can start to measure certain it now makes it easy to explain the pulse parameters about the digital signal. We’ll time period (sometimes referred to as ‘cycle discuss the theory first and then look at time’). The time period is simply the time the MMBASIC commands that simplify taken to complete one pulse cycle. In other the whole process. words, relating to the square-wave pulse cycle highlighted in bold in Fig.1, it is the time taken between two consecutive lowPulse cycle to-high logic-level transitions. More on A pulse cycle within a digital signal can this later when we discuss the MMBASIC be considered as a ‘complete wave’ that command to measure this timing. simply comprises a high-level logic pulse and also a low-level logic pulse (it doesn’t matter which comes first). Fig.1a has one Signal frequency square-wave pulse cycle highlighted in Frequency is often regarded as the number bold to make this a little clearer. Note too of pulses per second, and it is measured Practical Electronics | January | 2021