Fullscreen HTML5Med Res (??MB)HTML4

AUDIO OUT AUDIO OUT L R By Jake Rothman Microphone Preamplifier (for Vocoder) – Part 3 Fig.22. The assembled microphone preamplifier PCB – note the links for NPN version. IN+ IN GND IN– + SK1 R29 R4 REX3 L2 R1 POT3 POT1-2 (link wiper and CCW at pot) SK2 D 3 TR1 c C4 R6 e C2 C 5 D 6 D5 D4 R8 R10 R14 R13 R16 C7 VR2 VC1 C22 C21 C 10 R11 R13 R22 L3 R15 R5 C3 C24 + D2 D1 b TR2 REX2 R3 R9 R7 b c Gain pot C6 C23 e L1 Most small components are mounted on the PCB, which is a standard, easy-to-repair, double-sided plated-through-hole design available from the PE PCB Service C1 R2 R 28 Construction (code AO-JUL21). Mike Grindle has created a clear mirror-imaged symmetrical design. The overlay is shown in Fig.21 and the capacitors are dual outline, so either standard radial or expensive axial types can be fitted. Stuffing the board should be in the following order: resistors, chip sockets, small capacitors, preset, transistor holders, big capacitors, Molex connectors, and finally the trimmer capacitor, C8 C20 R20 R 17 IC1 R 18 C19 + C 11 C9 SK3 C13 + C 18 R19 IC2 R26 R25 R21 C 19 R27 C15 +48V –15V GND +15V Power C12 R23 L5 Output R24 C14 + L4 C 16 SK4 Transformer out Unbalanced out GND Transformer return + Balanced input just go noisy. Reverse-polarised tantalum capacitors may go short circuit, killing connected components. So, pay attention to polarity and component orientation! + to carry out the construction and testing of our microphone preamplifier. The circuit is mainly direct coupled and has lots of polarised components, so mistakes can cause damage. For example, if D1 is reversed, the transistor base will be connected to the power rail and the transistor and op amp may be destroyed or less obviously, + W e’re now in a position Fig.21. Microphone preamplifier PCB overlay – take care with links and orientations if using different transistors to the BFW16A. Important – in red boxes only use one link: solid for NPN and dashed for PNP version of the microphone preamplifier. 50 Practical Electronics | July | 2021 n Fig.23. Occasionally the input capacitors C1/C2 (here C2, at bottom) may have values 20% apart. In this case, an extra 2.2µF capacitor (C24, shown here above C2) was added in parallel to get good matching and hence high CMRR at 50Hz. which has a delicate adjustment screw. Several views of the completed PCB are shown in Fig.22 to Fig.25. The big, mechanically stressed components are all panel mounted and hard-wired to the board using 7/0.1 stranded wire. No PCB-mounted pots or external plug connectors here, which often cause broken soldered joints when used on a studio floor. I just love diecast boxes (see Fig.26), especially those by Hammond and Eddystone and use them for most of my small-batch builds because they are easy to cut and drill (much nicer than plastic or steel). Surrey Electronics always used them for their range of microphone preamplifiers and broadcast products. A couple of other assembly points: XLR connectors have earthing tags which are useful for enclosure earthing (see Fig.27). Note how the output transformer is mounted off the board using a capacitor clip shown in Fig.28. Fig.24. Note the green 82µH inductor, which looks like a strange resistor. The odd things that look like back-up batteries are Plessey Castanet tantalum capacitors. Fig.25. A thermal link was tried to ensure both transistors drifted together to minimise DC offsets. It was not worth it in this case (but is worthwhile on synthesisers). Testing away. A solution used at Kemo Filters is to put 560Ω resistors in series with any grounded input pins or those connected to the output pin, such as in buffers, as shown in Fig.29. Once powered up, check all transistor voltages and make sure the outputs of the op amps are all at zero volts, apart from the IC2 outputs which should be +0.2V approximately (due to the input bias current flowing through the 680kΩ resistors R11 and R12). If the DC conditions aren’t right, the audio won’t be right. DC conditions At the testing stage – if possible – use a tracking plus and minus dual supply. This should be current limited to approximately 100mA. It’s not nice when expensive components go up in smoke or eyeballs are damaged from flying chips when errors and high currents are combined. The circuit runs at a standard supply of ±15V, with the transistors taking 4.2mA each and the 5532s 9mA. The total current consumption is 22 to 25mA. (The 5532s and transistors can work on an absolute maximum of ±22V supply, taking 30mA.) An anomaly with the 5532 is that it can latch up if the rails don’t rise together at turn-on. This seems to be exacerbated when capacitors are connected to the input pins, which is the situation here. I was puzzled because other op amps don’t do this. When I switched the power supply from independent to tracked, this issue went Fig.26. A single-channel microphone preamplifier fits in this die-cast box, size 187 × 120 × 55mm. (Note the PCB shown above is a slightly different prototype). Practical Electronics | July | 2021 Power supply Naturally, low noise can only be achieved if the power supply is low noise. A standard audio LM317/337 dual-rail circuit will do the job. The 48V supply is more specialised requiring a voltage doubler perched on top of the positive rail. My LM317 circuit (see Audio Out, PE March 2019) is suitable – it has low noise and while you can certainly build your own, I do still have some PCBs – see my Audio Out Shop advert on p.43 for contact details. Alternatively, a TL783 regulator will work. It’s best to keep the mains transformer in a separate screened (earthed metal) box to minimise mains buzz. The circuit needs very low impedance supplies, since the power supply rejection ratio of the input transistor stages is very low. The emitters and collectors are connected to the power rails only Fig.27. Metal cases can be earthed off XLR sockets via an earthing tag on the rear of the socket connected to the PSU 0V. 51 – 5 5 3 4/ 2 + R R IN F – 5 5 3 4/ 2 + 0 V Fig.28. The output transformer is neatly clamped in a 45mm capacitor clip. Fig.29. Inserting 560Ω resistors to prevent latch-up with 5532/4 op amps. by resistors, providing an easy entry route for supply ripple. The best way to ensure a low-impedance supply is by connecting directly to the outputs of the regulators. In one design, I had 22Ω decoupling resistors feeding the board and this increased the distortion into a 600Ω load by 0.5% at low frequencies. both sections wired in parallel), and finally 74dB (×5000) with the gain control fully clockwise. In the middle of rotation, it was 39dB. Check the clipping is symmetrical and the response is flat from 20Hz to 20kHz. Noise check Selection of the fittest has always been part of audio evolution, especially input transistors, which is why I recommend using sockets for circuits where transistors need to be selected for noise level to screen out poor devices (see Fig.33). Use a ‘scope to look at both outputs of IC2 with the gain control (VR1) set to maximum, that way you can see the noise level of each transistor and pick the best ones by comparison. Do turn the power off when swapping the transistors and leave the input unloaded to achieve maximum noise for this test. Gain check Check the overall preamplifier gain by feeding a signal generator into the input via a small transformer or other balancing method. I use the video isolation transformer shown in Fig.30. The gain should be around 15dB (×5.6) with no Rg and 20dB (×10) with Rg = 3.69kΩ (actual value of the Blore Edwards 5kΩ dual pot with Square wave check Applying a 1kHz square wave is always a good idea to check for ringing. This is necessary to optimise the Zobel network on any transformers used (See Fig.41 next month). I also found a bit of ringing on the input chokes L1 and L2; putting 1.3kΩ resistors across them on the spare inductor holes as shown in Fig.31 damped it out. These resistors are labelled R28 and R29 on the latest board but are not shown on the circuit diagram (Fig.18). The noise should also be checked by putting two terminating resistors of 75Ω on the input pins joined together (pins 2 and 3 on the XLR socket) as shown in Fig.32. This will give a source resistance of 150Ω, mimicking a low impedance moving-coil microphone. At full gain, the noise on the output should be around 12mVpk-pk, with the low-cut filter in. (The low frequency noise is inaudible and makes it harder to read as the trace jumps about). With a metal housing, battery power and careful selection of transistors it should be possible to get the noise level lower. I’ve built various ‘resistor terminating units’ into male XLR plugs. I always recommend using metal XLR plugs to provide screening. Fig.30. When testing balanced-input circuits it is necessary to use an isolating/ balancing transformer between the signal generator output and preamplifier input. Common-mode rejection ratio (CMRR ) check The CMRR can be checked by driving both inputs together in phase to see what comes out of the output, ideally nothing (see Fig.32). Unfortunately, the CMRR varies with the gain setting, a disadvantage of transformerless techniques. Because of this, I suggest checking the CMRR with the gain at around 30-40dB (two o’clock on the gain control) with an input of around 1V to 3Vpk-pk. Because the gain control (Rg) only boosts gain on common-mode signals, both outputs from IC1 will remain almost unchanged at around 5Vpk-pk until the last bit of the maximum gain region. This is why the middle gain range is suggested for CMRR adjustment. If the unit is to be used at a fixed gain, say in a broadcast application, the CMRR should be optimised for this. First adjust trimmer VR2 at 1kHz for minimum output, then do a null at 9kHz with the trimmer capacitor VC1. The low-frequency CMRR at 50Hz can be optimised by putting 220nF to 2.2µF padder capacitors (C23 and C24) alternately across input capacitors C1 and C2 until a null is achieved (see Fig.23). I suggest these frequencies because most of the noise that’s a problem in the field is mains related, I np u t f r o m si gnal gener ato r f o r C M R R test . Si gnal gener ato r s o u r ce i mp edance 1% 1% 3 1 2 M al e X L R p l u g (r ear vi ew ) Fig.31. To reduce a minor amount of ringing on square waves, the input filter chokes L1 and L2 can be damped by wiring 1.3kΩ resistors across them (R28 and R29). 52 Fig.32. XLR termination unit provides the correct source impedance of 150Ω and enables a signal generator to be connected for CMRR testing. Fig.33. Transistors should be mounted in sockets so noisy ones can be easily removed. (It’s also good for finding unknown gems in one’s parts box) Transistors from Hitachi, Toshiba, Sanyo and Rohm seem to be the best. Only home constructors have time for this, so it is possible to improve on ready-made units. Practical Electronics | July | 2021 mainly switching noise from power supplies and lighting controllers. Surprisingly, I found the adjustment range for the trimmer was better if C13 (the trimmer parallel capacitor) was reduced to 120pF when the circuit was transferred to PCB. (There may be some stray capacitance associated with the board layout, but I doubt it’d be over 100pF. More investigation is needed!) On my boards I used C13 values ranging from 68pF to 180pF to keep the null of VC1 in the middle position. In the parts list I specified an 80pF trimmer, but I now recommend using a larger value, such as 250pF. Mods and embellishments Vinyl virtues This circuit would make an effective moving-coil (MC) pick-up amplifier with RIAA equalisation incorporated, but I would have to convert my record deck to balanced output. This would make it non-standard in the Hi-Fi world, but ideally, all audio equipment should be balanced – even electric guitars! The board could be used for an unbalanced stereo moving-coil pickup head amplifier by just using the first stages and omitting IC2 and its associated components. + 48 V + 48 V O N 1N 4148 B S170 B al anced i np u t 470 nF 10 nF 0 V Fig.34. Switching on phantom power can produce ear-splitting cracks unless the rise time of the voltage is slowed. This can be done with a 100Ω resistor and 470µF capacitor or the capacitor multiplier circuit shown above. ‘standard’ 15% C reverse-log-law pot give too abrupt a jump in gain at the limit of the rotation, and is therefore not optimal. A better law is the 8% RD law produced by Alps. (se Fig.35) That Corporation decided 2.5% was best, avoiding the sudden up-lift in gain. Such a pot would have to be specially made and be expensive. I should point out here that cermet and wire-wound pots have bad rotational Phantom power noise, the wire-wound type sounds like Switching phantom power can make a loud undoing a zip! If you want to read up in noise. It is best to slug the rise with a big more detail on potentiometer law details, electrolytic capacitor or MOSFET capacithen see my articles from Audio Out, Notance-multiplier circuit, as shown in Fig.34. vember 2015 and April 2018. The alternative to pots is switched gain, Gain switch which enables each step to be optimised Instrumentation-type amplifiers (as used and allows the use of low-noise metal-film in this design) use reverse-log-law gain resistors. Cadac mixers use a 21-way switch control pots. Unfortunately, they are difand Neve preamplifiers often use mulficult to source. – but it gets worse! The tiway switches. The table opposite shows resistor values required for 2dB steps for the biggest 29-way switches available and Fig.36 shows its construction. If you have fewer steps, it’s best to have the smallest increments on the highest gain settings. That Corporation makes a special SMT chip for use with their 1570 microphone amplifier chip that includes electronic switching and a servo to control DC clicks, called the 5171. This has to be connected to a microcontroller, but it does give 1dB steps. Anyone fancy designing a rotary encoder circuit for it? Do note, however, Fig.35. Alps reverse log pot tapers. This shows that both chips are expenvoltage/resistance vs rotation. At 50% rotation, a linear sive (1570, £6 / 5171, £11 potentiometer split point would be would be 50% / 50%, from Mouser). That said, I taper C is 82% / 18% and RD is 92% / 8%. Practical Electronics | July | 2021 Fig.36. Test your soldering skills: a 29-way Painton switch with resistors attached, all different values. Note the ring of 22swg tinned copper wire that forms one terminal of the gain control; the other is the wiper. I often repurpose these high-quality specialist switches from old, scrapped test equipment. do worry about using single-sourced specialist chips, but anyone can make up a switch with a load of resistors on it. Resistor values for stepped gain switch Step 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 Gain (dB) 15 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 72 74 Ideal Resistance open 3.69kΩ 2.7kΩ 1.88kΩ 1.6kΩ 1.17kΩ 790Ω 600Ω 470Ω 390Ω 320Ω 220Ω 178Ω 138Ω 109Ω 85Ω 66Ω 51Ω 40.4Ω 31.9Ω 23.6Ω 18Ω 13.2Ω 9.8Ω 7.1Ω 5.1Ω 2.7Ω 1Ω short E24 value open 3.6kΩ 2.7kΩ 1.8kΩ 1.6kΩ 1.1kΩ 750Ω 620Ω 470Ω 390Ω 300Ω 220Ω 180Ω 130Ω 110Ω 82Ω 68Ω 51Ω 39Ω 30Ω 24Ω 18Ω 13Ω 10Ω 6.8Ω 5.1Ω 2.7Ω 1Ω short Next month We’ll conclude in Part 4 with optional modifications, including the use of transformers. 53