Fullscreen HTML5Med Res (??MB)HTML4

AUDIO OUT AUDIO OUT L R By Jake Rothman Microphone Preamplifier (for Vocoder) – Part 2 this new design for a microphone preamplifier. We offered it as a suitable design for a vocoder, but in fact it is a general purpose, very highquality circuit that will work for most microphone applications – and at a fraction of the cost of commercial models. This month, we’ll complete the design and start to look at construction issues, which will be completed next month with the PCB and build options. A question of balance In microphone preamplifi er circuits, balanced lines have always been used where low-noise pick-up is required. The system involves two signals anti-phase to each other surrounded by an earthed screen that carries no current, unlike say an unbalanced single-core guitar lead. When a signal appears in phase (common mode) on both wires it is rejected by the circuit which only looks for the voltage difference. Transformer-based microphone preamplifiers have a floating balanced input with a common-mode (CM) voltage of over 100V. If a transformer is not used, then the system has to be electronically XLR so cke t ext ernal vi ew 2 + 17V XLR so cke t so ld er vi ew . 1 3 + V + B alanced input – 3 Output V – Set g ain Fig.8. The most-basic balanced microphone preamplifier, the differential op amp amplifier. Too noisy for studio use. To vary the gain, two resistors have to be varied using a dual-gang pot. + 2SC 2362K Output 100µ F XLR input 1 0V 48 2SC 2362K 5534 + 2 0V B alanced input – T L071 1 3 2 – . Set g ain XLR input 1 balanced. The differential B alanced amplifier configuration is input the standard approach, + and it can reject CM sigG ain = 1 + (2Rf / Rg ) XLR – input nals of up to around 10V. R 3 – 1 A single op amp differ+ 2 R ential amplifier shown in Rf – 0V Fig.8 will work, but it is Rg too noisy, mainly due to + Rf Output R its input resistors. Most microphone amplifiers use R – the instrumentation ampli0V fier circuit shown in Fig.9, + where two separate amplifiers with gain feed the inputs of a unity-gain dif- Fig.9. The instrumentation amplifier configuration. The ferential amplifier. A great basis of most transformerless microphone preamplifiers. feature of this approach is The gain is set by one resistor Rg. that a single gain-control resistor (Rg) sets the gain of both the inshown in Fig.11. Other companies, such put amplifiers. as Solid State Logic (SSL) and Neotek The input amplifiers are often just took this further by wrapping the input single low-noise transistors, such as transistors within a feedback loop of an in Fig.10. This arrangement has high op amp. The inverting input of this op distortion at high gains because they amp was often directly coupled to the are simple common-emitter (CE) stagcollector of the transistor, effectively es with a low effective emitter resistor creating a current input or ‘virtual earth’ for linearisation. In their mixers, manu– see Fig.12. facturer Mackie got round this by using This topology is referred to as the ‘cura complementary follower pair (CFP) rent feedback instrumentation amplifier’, + L ast month, we introduced 0V + – A nti-log 3 G ain 2 1. 1mA – 17V 0V Fig.10. Replacing the input op amps with low-noise low Rbb input transistors enables low source impedances to be amplified with minimum noise. The distortion produced by the single transistors is in the order of 1% at high gains and levels. Practical Electronics | June | 2021 + 15V . . 14. 4V 330pF 5. 8 mA 330pF Rg G ain P osi tive input 1000µ F 6V N eg ative input . + n A -log 2SA 1312 2SA 1312 MP SA 06 MP SA 06 . . Fig.13. An integrated CFIA solution is available, the That 1512 from Profusion. I’ll be developing a circuit for it, hopefully with no big electrolytic capacitor on the gain pot. . – 4. 7V – 1. 18 mA T L071 + – 4. 7V Output . – 15V 0V Fig.11. The complementary follower pair microphone preamplifier produces ten-times less distortion. It’s difficult to get enough current flowing in the input transistors to get a low optimum source impedance. in these ICs benefits from the inherent matching of transistors available to the IC designer but is not suitable for discrete circuits. You can of course use these single-sourced chips if you can get them. But I regret using some of them in some products I designed in or CFIA, which is what we will use here. This topology was further refined by adding cascodes and current mirrors and integrated into specialist chips such as the Solid State Music/Intel SSM2015,2016, 2017, 2019 and the That 1510 and 1512 series. This topology Fig.12. The current feedback instrumentation amplifier – op amps linearise the input transistors. V + RC B alanced input C urrent into op amp – + 1 0V + – 0V 3 2 XLR input V – Rf N eg ative f eed back R R – + Rg Set g ain R D if f erential amplif ier R V + RC 0V C urrent into op amp – + 0V V – N eg ative f eed back Practical Electronics | June | 2021 Rf Output the past because now they can be difficult to fix if the ICs are unavailable. The That chips (Fig.13) are still available from Profusion, however, and they have kindly sent me some free samples which I will evaluate soon. The killer phantom High-quality condenser microphones are powered by 48V applied to each conductor of the balanced line via a transformer centre-tap or two 6.8k resistors, as shown in Fig.14. The power rides on top of the audio and is rejected by the differential input of the amplifier. Since the voltage is applied without extra wires, it is called ‘phantom power’. (Note that the 6.8k resistors must be at least 0.5W rated, 1%.) This 48V system was originally proposed by Neumann in 1966 because the capacitive diaphragm assembly needs a high polarising voltage and there were plenty of 48V power supplies used for phone systems at the time. It subsequently became universally adopted. 48V is relatively high compared to today’s solid-state electronics and it can cause considerable damage by reverse biasing delicate base-emitter transistor junctions, making them permanently noisy. This situation has been described as the ‘phantom power menace’ by the Audio Engineering Society and I spent much time in 1996 replacing the BC109C input transistors in the famous EMI mixing desk from Abbey Road originally used for the Beatles recordings. I’ve only recently had to change an SSM2019 chip in an ironically named Mackie ‘Spike’ microphone USB interface because the Zener diode clamping the phantom power spike had failed open circuit. The cause of the killer spikes is the plugging and unplugging of the cables in conjunction with the big DC blocking capacitors. Shorts in the cables make the situation even worse. The solution is a total of six diodes to steer the spikes away and current-limiting resistance of at least 5 in the lines, as shown in Fig15. Don’t put too much resistance 49 P hantom pow er 0V Red XLR input W hite 3 – 1 + OE P X18 7B Microphone transf ormer Ratio: 1: 9 . 13 H ig h input imped ance microphone amplif ier B lack + 10-47µ F XLR input 1 B lue Rf resi st ance . . V + 1% 3 2 . . 0V 1% R V – 10-47µ F G reen . 0. 5W 1% XLR input 1 + 3 – 2 R + Output D if f erential low -imped ance microphone amplif ier R 0V in, since this will increase noise. Surprisingly, cheaper ordinary diodes are better than Schottkys or Zeners in having less leakage noise and capacitance. One trick I used to do to improve the linearity and headroom on CE and CFP designs was to use 48V for the positive supply rather than 15V. If it’s there, you might as well use it. This was a technique used in Tim McCormick’s article Putting Mic Amplifiers on the Line in Electronics World + Wireless World in May 1992. I don’t see a theoretical reason for the technique to be beneficial with the CFIA topology used here because the collector voltage swing is small. The larger resistor values could give greater consistency of collector current however, so provision to connection to the 48V rail is made on my PCB for experimenters. The use of constant-current sources may introduce additional sources of noise. RFI protection Transistor base-emitter junctions are prone to demodulate RFI (radio frequency system here but found I couldn’t get axial inductors with the necessary low DC resistance of <10 . I had to use radial inductors which were less ‘adjustable’. In the end, I resorted to a trimmer capacitor on the differential amp. By putting the coils touching each other on the PCB, I could get around 25% cross coupling. Another pair of inductors are used on the emitter circuit to stabilise the amplifier at high gain. This is a trick that was used on the famous 990 discrete op amp. Their DC resistance also defines the minimum emitter resistance, and hence the maximum gain for the stage. interference); they are diodes after all. So, in addition, we need RF filter chokes and capacitors. All these extra components are not generally needed with microphone transformers, which offer excellent RF rejection. But it is a general trend in electronics to replace expensive single components with multiple ‘jelly bean’ components. The final block diagram is shown in Fig.16. Wonky windings I always like to offer readers a few anarchic analogue anecdotes (AAAs) I’ve picked up along the way. Here’s one from when I was a test engineer at Brook Siren Systems in 1987 when I was setting up their four-microphone preamplifier splitters in a box, the MSR604. I found their design engineer Stan Gould had come up with an ingenious system for trimming the 10kHz CMRR at 10kHz. This involved wiggling the input inductors relative to each other to achieve a null due to the variable magnetic cross coupling (Fig.17). I wanted to use this Fig.16. Block diagram of the full microphone preamplifier. T R1 The full monty Now we’ve looked at the sections, we can put together the full circuit, as shown in Fig.18. (Do see the circuit notes in Fig.18.) Since we are using dual op amps, just as with logic gates, there is always one spare left over. In this case we will use IC2b for a special negative resistance generator circuit to minimise distortion when using an output transformer. Again, this D if f erential amplif ier w ith C MT T trim Low -noise g ain block 1 + 48 V phantom pow er 1 0V + – 3 2 RF f iltering P hantom protection IC 2a + Low -noise g ain block 2 RV 1 g ain control 0V T R2 IC 1 N eg ative f eed back 50 U nbalanced Output – IC 1 N eg ative f eed back XLR input 1N 4148 Fig.15. (above) The ‘Phantom Menace’ protection scheme (apologies to George Lucas and Star Wars fans). Fig.14. (left) Phantom power application to microphone balanced line inputs. a) Using transformer centre tap and b) Via 6.8kΩ resistors used in the transformerless designs. (Note that these resistors have to be rated at ≥0.5W because the dissipation can be greater than 0.25W if the microphone cable develops a short.) – R . 0. 5W 1% 0V R + + D C blocki ng capacitors B R1 1A 100V B ase -emitter reve rse -bias protection 0V + 48 V 1N 4148 + 0V Input transi st ors can be d isc rete or in a an IC Output – . resi st ance 2 + 48 V phantom pow er + 48 V P SU + 0V N eg ative resi st ance f or transf ormer output IC 2b – B alanced Output R-se nse Practical Electronics | June | 2021 Fig.17. Brook Siren Systems microphone input filtering – cross-coupling between coils (the above are not resistors!) was tweaked by moving them with a Bourns trimming tool to obtain a maximum CMRR null at 10kHz. Banish the electrolytics was another ‘AAA’ I picked up from work in the early 1990s, this time from Calrec who made mixers in Hebden Bridge. We’ll cover some of the theory of this technique next month in Part 3. Output transformers are much better at preventing earth loop hum than balanced output amplifiers which inherently centre their signals around signal ground (0V). I like my studio gear to have the best of both worlds; electronically balanced inputs (where you don’t need an earth connection) and floating transformer balanced outputs, where any earthing situation can be tolerated. I often have a cull of wet electrolytic capacitors in any professional audio design because of their short life expectancy and high leakage currents. They can often be replaced with expensive film and tantalum capacitors, but the massive anti-scratch capacitor in series with the gain control is often difficult to source. Since the minimum resistance is less than 10 at maximum gain, the capacitor has to have a reactance at low frequencies of a lot less. For a −3dB point at 20Hz the value has to be 800µF, and for top quality gear, −3dB would not be V + + 48 V input N P N * 0V P N P * V – Rex * * 6 R11 C 1 22µ F 50V L1 4. 7mH 0V + – + 7V / – 7V 4mA / – 4mA C 23* * * 1µ F R4 0V R5 D 3* * * * 1N 4148 60µ A D 2 1N 4002 T ransf ormer return current C 10 39 pF 2. 18 V / – 2. 18 V R9 L3 8 2µ H . + 2 + R20 P N P * N P N * V R2 C 7 330nF + 7V / – 7V 4mA / – 4mA R14 L2 4. 7mH 6 R6 60µ A D 5 1N 4002 4. 2mA / – 4. 2mA IC 1b 5 5532 + Practical Electronics | June | 2021 V – V C 1 10k H z 8 0pF C MRR trim * T R1/ 2 B lack: N P N B FW 16A (also black vo ltag es/ currents) Red : P N P B C 143 (also red vo ltag es/ currents) For exa mple, w ith N P N the vo ltag e across R16 is + 2. 18 V at the output of IC 1b compared to T R2’ s emitter, but – 2. 18 V if P N P is use d . 7 + 0. 15V C 11 39 pF * * Rex : exp eriment – se e text * * * C 23/ 24: optional low -f req uency (50/ 60H z) C MRR pad d er capacitors R16 * * * * D 3/ D 6: reve rse f or P N P transi st ors 2. 18 V / – 2. 18 V R10 L4 8 2µ H . V + Fig.18. Circuit diagram of the microphone preamplifier. Complex, but most parts are fairly cheap. C 13 220pF R24 Circuit notes 0. 22mA / – 0. 22mA N P N * 0V D 6* * * * 1N 4148 C 14 100µ F 25V L5 8 2µ H (Optional f or low output imped ance) z C MRR trim – T R2* 0V – 2V / + 2V P N P * R3 C 24* * * 1µ F V – 1kH – 1. 3V + C 2 22µ F 50V 0V 4 0V R8 D 4 1N 4002 1 R22 R12 + 8 /– 8 V U nbalanced output R23 0V V – R2 . 0. 5W – 8 IC 2a 3 5532 R17 V + V + R19 C 8 ***** 330µ F 6V V – Rex * * T o output transf ormer R21 0. 22mA / – 0. 22mA 0V C 12 270pF R15 V + C 4 470pF 0V C 15 330µ F 6. 3V R26 . R25* + C 5 470pF L1, L2 loose ly mag netically coupled – T R1* 0V – 2V / + 2V 4. 2mA / – 4. 2mA C 3 470pF + V – 3 2 8 1 IC 1a 3 5532 + 0. 15V + 4 R13 N P N * 1 2 – 1. 3V + XLR input D 1 1N 4002 + 7 V + C 6 330nF P N P * B alanced input + 8 /– 8 V IC 2b 5 5532 * Select f or output transf ormer use d R7 R1 . 0. 5W R27 – + C 9 ***** 330µ F 6V R18 S1 g ang ed w ith V R1 * * * * * C 8 / 9 : reve rse connection f or P N P transi st ors S1 + V CC C W V R1 C 16* * A ll 100nF ceramic C 17* C 19 * A ntilog G ain control C 18 * C 20* C 21 + 10µ F 25V C 22 + 10µ F 25V + V 15V 0V – V 15V 0V – V CC 51 they don’t drift. These capacitors not only increase bass loss because of their reactance, but they can also boost low-frequency noise due to the increase in effective source impedance. Thus, they have to be big, necessitating a large physical film capacitor of at least 6.8µF. I’ve settled on 22µF 50V metal-cased solid tantalum types because again, I have a big stock and never had any failures. In this design they are polarised by the −1.3V on the transistor bases. Prototype construction The prototype was built on Veroboard shown in Fig.19. A double-sided plated-through hole PCB with masses of Fig.19. Veroboard prototype. A mess after the abortive attempt to add a servo to get rid of the big ground plane is really to be electrolytic capacitor on the gain control. Note the input transistors in sockets for picking the quietest expected for this quality level specimens – coming next month, a nice neat PCB! and a design will shortly be coming from our erstwhile but this is often unnecessary in practice. designer Mike Grindle. As usual, solconsidered top notch. Making the miniIf you want to upgrade the servo amp der up in height order, links, diodes, mum resistance larger is not a good idea, you can use the low-noise LT1012CN8, resistors, chip sockets, transistors, small because it increases the noise. Since the but it’s £5. The servo can be checked caps, connectors then big caps. Take polarity is undefined, two capacitors by ensuring the voltage across the gain Mark Nelson’s advice and use 60/40 (C8 and C9) are needed back-to-back, pot is less than 6mV and its output is leaded solder for reliability, especially halving the value of capacitance, but not saturated; it will typically be a few when using old mil-spec tantalum calowering the tantalum distortion. The volts either way. I did a further check pacitors and other NOS components. I voltage at each emitter is −1.9V; thus, if by putting very unequal transistors in always specify ‘leaded finish’ for my the ends of the capacitors are groundto see if it still zeroed. The servo is very boards, it’s cheaper and better. Don’t pay ed via pull-down resistors R17 and R18 sluggish, so allow at least a minute for rip-off EU/UK prices for leaded solder they will be just sufficiently polarised. it to settle. either. I use 3% 60/40 AMI solder made The expensive capacitor solution uses When building such a circuit I just in Canada and available from Mouser, Plessey Castanet cup wet-tantalum types pick from my stock of good old TL071 Part No. 13288. of the biggest value I could find, 750µF op amps. The low offset ones 3V (see Fig.7). This will still give bass T R1 go into servos and the high loss at very high gains. But the moment L1 offset ones into AC coupled the pot is backed off a bit, the bass will circuits. This servo pot noise return. Normally, the response is −1dB R5 eliminator was great in theat 20Hz and 25kHz. ory, but in practice it had a Input transi st ors fatal flaw. If you rotated the Servo servitude T R2 gain control fast, like a musiThe theoretical solution to getting rid L2 cian rather than an engineer, of the gain capacitor is to engineer a it scratched as the sluggish servo circuit that keeps the voltage on From R6 input servo tried to catch up. It also both ends of the pot the same so that circuitry occasionally latched up if the 0V no current flows through it. Building preamplifier was switched on the circuit with worst-case variations 1µ F Servo with the gain control at maxin transistors showed up to 500mV imum. There wasn’t enough difference was possible. This could V + . 7 voltage across the pot for it be avoided by matching the transis2 – to sense and start-up. I extors; for example, by using an LM394 6 IC 3 V R1 T L071 3 perimented with it for two or SSM2210 dual device, but at high + + 22µ F* N oise days solid and found it was 4 cost. Matching two separate transistors . f ilter V – + a waste of time – for now. can work, but thermal coupling is re1µ F 22µ F* * or a si ng le Rg g ain 10µ F bipolar quired to prevent drifts. control 0V I tried a differential integrator servo Input capacitors with IC3 to sense the difference and feed It’s worth matching the input a correction signal back to the input capacitors C1 and C2 to main- Fig.20. The servo senses the voltage across the gain shown in Fig.20. The noise contribution tain low-frequency CMRR. potentiometer (Rg) and drives it to zero. In practice, this from the output is surprisingly low. It can This only works with film circuit was flawed. I think I need two separate servos (one be filtered further by use of capacitors, and tantalum types because for each transistor) or a true differential output servo. 52 Practical Electronics | June | 2021 Parts list Here is a complete parts list to help start your microphone preamplifier build, ready for next month’s PCB. Some of the trickier parts will be available from me – see comments in list and check my PE stock clearance ad next month for bargains. Since this circuit is dependent on absolute symmetry for good CMRR, it’s worth matching all the components that are duplicated on each side (marked * below). I can supply these if required. SSL used to match its input capacitors to 1%. Semiconductors TR1*, TR2* BFW16A (NPN), or BC143 (PNP) or similar low-Rbb, low-noise transistors D1 to D4 UF4002, 1N4002 D5, D6 1N4148 IC1, IC2 NE5532 low-noise op amp Resistors All 1% 0.25W metal-film MR25 or similar, except where shown. R1, R2 R3, R4,24 R5, R6 R7, R8,25 R9, R10 6.8k 100k 22k 2k 3k R11, R12 680k R13, R14 560 R15, R16 10k R17, R18 22k R19, R20, R21 1k R22 820 R23 47 R25 2k select for output transformer used) R26 1.2k R27 10 VR1 VR2 Capacitors C1*, C2* C3, C4, C5 C6, C7 C8, C9 C10, C11 C12 270pF 5k reverse log C or Blore Edwards AB CTS 45 series dual 5k RLOG with switch (see my PE advert next month or Tayda for reverse log pots). 500 cermet trimmer 6.8µF to 22µF 50V plastic-film, tantalum or low-leakage electrolytic (in order of preference) 470pF ceramic 330nF polyester 330µF 6.3V or larger tantalum or low-leakage electrolytic (in order of preference) 39pF 5mm 5% NP0 ceramic 2% polystyrene C13 220pF C14, C15 C16 to C20 C21, C22 VC2 C23, C24 Inductors L1, L2 L3, L4, L5 2% polystyrene 100µF 20V axial or radial electrolytic (PCB fits both) 100nF X7R 5mm ceramic 10µF 25V 5mm 5.5pF to 80pF swing Philips plastic-film trimmer (see my PE advert next month) 1µF polyester (optional) 4.7mH 10 radial inductor Murata 8RB (Mouser 22R475C or see my PE advert next mont) 82µH 1.3 axial inductor TDK or similar (Mouser 778F820J-TR-RC, or see my PE advert next mont) Miscellaneous TO5 Winslow Adaptics W3437G from CPC (part SC09471) transistor holders 8-pin DIL chip socket (2 off) Next month In Part 3, we will move from a veroboard rat’s nest (Fig.19) to a low-noise, clean PCB design. We will add a balanced floating output and of course run through testing and some distortion-reduction techniques. eptsoft Directory of Education, Lifestyle and Leisure Suppliers Over three-hundred Hobbies and Special Interest Suppliers with online links to their websites for you to explore. Your best bet since MAPLIN Chock-a-Block with Stock Visit: www.cricklewoodelectronics.com Or phone our f riend ly kn ow led g eable st af f on 020 8452 0161 Components • Audio • Video • Connectors • Cables Arduino • Test Equipment etc, etc Promoted into the Education Community through emailing and Free Education Resources for Nurseries | Academies | Primary Schools | Secondary Schools | Further and Higher Education | Special and Independent Schools. To add your business email info<at>eptsoft.com Visit our Shop, Call or Buy online at: www.eptsoft.com www.cricklewoodelectronics.com 020 8452 0161 Visit our shop at: 40-42 Cricklewood Broadway London NW2 3ET Practical Electronics | June | 2021 epts oft.indd 1 53 12:00 04/02/2021