Fullscreen HTML5Med Res (??MB)HTML4

AUDIO OUT AUDIO OUT L R By Jake Rothman New de-thump circuit and power supply mods M ost audio circuits make Original circuit Time The basic function delay AC Power blocks of a relay mutinput supply* ing circuit are shown in R Fig.2. The requirement *Must have Reset Relay minimal is a few seconds delay Driver coil smoothing in turning the relay on (see text) C at power-up and instant turn-off when powered down. First, there is a Fig.2. Block diagram of a typical de-thump circuit. rectifier and a minimally sized smoothing capacitor to facilitate quick discharge when turned off. Then Voltage there comes an RC timing circuit (R4, C2). across relay This drives a Darlington transistor (TR2 0.1s and TR3) to power the relay. Following 26V pull-in that are some refinements, mainly the addition of a timing capacitor discharge circuit (TR1) to deal with a turn-off (possi13V holding bly, quickly followed by a turn-on again). voltage This is basically a reset function so that a 0V full turn-on delay period is always availTurn on Time (after delay) able. Another circuit dodge is the ‘current saver’ circuit in series with the relay. This Fig.3. Graph showing pull-in and hold consists of a big electrolytic capacitor (C3) voltages used for a 24V relay. in parallel with a high-wattage resistor to hold a pulled-in (switched-on) relay (R7) giving a full voltage pulse to pull than to switch it. the relay in. The voltage then drops to a value sufficient to just hold the relay, resulting in half the pull-in current set by Improvements the resistor. This function is shown in the The main weakness in the original graph in Fig.3. The key point here is that circuit was the slow turn-on time, it takes less voltage (and hence current) which resulted in less than expected performance from the current-saver. A current-saver circuit needs R1 Bridge RLY1 rectifier a fast switch-on step to 0.5W (32V holding) – + +42V 30V deliver a good kick to the AC (21V holdingl) D1 relay and pull it in deci1N4001 sively. This was achieved R2 R4 by putting in a compar+ R7 C3 470µF L R ator built around a 741 1W 63V D e -th u m p ZD1 C1 + A u d io (IC1) after the timing cirr e l a y BZY88 47µF 10V 63V cuit. This improvement R5 enabled the holding voltTR2 + C2 age to be reduced from 47µF BC337 R3 10V 21V to 13V, resulting in R6 TR3 BC141 30mA TR1 a relay current reduction BC327 from 30mA to 18mA. The comparator’s output was sufficient to drive the Fig.1. My original, 40-year-old de-thump driver circuit. – 60 + nasty noises (‘thumps’) when they are turned on and off as power rails and DC conditions stabilise. Although it is possible with great care to make quiet circuits, it is usually expedient just to mute the output during the stabilisation periods. This muting is normally accomplished by means of switch out the audio output, most commonly a set of relay contacts. Some low-cost circuits have a pair of transistors to short out the audio output, possibly more reliable than a relay, but giving higher output impedance and distortion. Whatever approach is taken, these switches have to be controlled with a timing sequence generated by a special circuit to successfully mute potential thumps. In last month’s Universal Audio Power Supply there was a relay driver circuit for audio output muting. I designed this circuit, shown in Fig.1, in 1984 for a college project. Although it’s done the job perfectly well in my products for nearly 40 years, it seemed worth revisiting, since there’s no circuit that can’t be improved with regard to reliability, function and power consumption. Here, I’ll show some of the design processes involved in this rather obscure branch of audio circuitry. Finally, I’ll give some extra tweaks to the power supply itself. Practical Electronics | June | 2022 + n virtually nothing. This allows a low-power op R1 BR1 amp to be used for IC1. 56Ω 1A/100V 30mA DIL Bridge – + 0.5W +42V The TL061 is ideal for 30V AC this job, because it has All voltage and R7 9.8mA currents are after FET inputs (often the 1kΩ relay has turned on Texas trade name ‘BiZD1 33V/400mW FET’ is used). Therefore, BZY88 TR2 R5 the loading on the timBFX85 R2 R4 100kΩ (or equiv) 56kΩ 560kΩ C3 ing circuit is minimal, TO5 heatsink 6.8µF 3 + 7 allowing the resistor to 25V 2.2mA 0.18mA +15V +30V IC1 C1 + be increased and the caRLY1 47µF 2 –741 +14V 6 TR1 + C2 24V 63V pacitor size reduced. This BC327 15µF +13V D1 4 700Ω 20V 1N4001 saves cost, since low-valR8 Tant 8.2kΩ ue (<4.7µF) tantalum R3 R6 R9 ZD2 capacitors are cheap. 330kΩ 68kΩ 6.8kΩ 12V L R Now that the currents are 400mW BZY88 18mA reduced, the current-saver circuit impedance can be scaled up, allowing the Fig.4. Improved de-thump circuit using bipolar devices. electrolytic capacitor (C3) output transistor directly without an Back-EMF protection to be replaced with a non-polarised type intervening driver transistor (TR2 in When the relay turns off, protecting the – always a good idea. Also, the Zener Fig.1). The op amp must have a simoutput device from the back voltage as regulator current can be decreased by inple 33V Zener regulator to protect it the magnetic field collapses is usually creasing R7 to 5.6kΩ, allowing a reduction from over-voltage. (The power rail is done by a single diode. However, this in current to the comparator circuit from 43V, while the op amp’s rated maxiapproach slows down the release time around 10mA to 2mA. The total current mum is 36V). An unexpected benefit of of the relay while the back current circonsumption is now 20mA, rather than this arrangement was the hum level on culates. If the back EMF is clamped at 30mA for the original circuit. the relay coil was reduced by a factor say 12V by a Zener (D2) rather than the Since a large TO220 metal-tab MOSof five, down to 500mVpk-pk. This may 0.7V of a normal diode (D1), then the FET consumes no more drive current help in low-noise applications. (I’m not energy is dissipated faster, resulting in than a small type, this avoids the need sure how much hum can couple from a quicker release. for a clip-on heatsink which the bipolar the relay coil into the signal switches. transistor required. If a TO220 bipolar This possible crosstalk path may be FETishisation transistor, such as a TIP31 were used, worth further investigation.) A standard trick I use to improve old cirthen the Hfe would be low, requiring Another improvement was to pocuits is to replace bipolar devices with more drive current than the TO5 type. sition the current-saver circuitry on FET-based devices. Their higher input the input of the relay transistor rathimpedance means lower drive currents. Discharge circuit er than its output, getting rid of the This minimises power consumption and One of the characteristics of JFETs is bulky and unreliable 470µF electroallows capacitor sizes to be reduced. that they are ‘on’ (conducting) when lytic capacitor (C3 in Fig.1). This new they have zero bias voltage on the gate. bipolar-based circuit is shown in Fig.4. Relay driver This characteristic is useful for the The high-wattage resistor (R7) is also The first thing to do is to replace the discharge circuit. The original PNP removed, although its voltage drop bipolar transistor (TR2, Fig.4) with a transistor configuration used couldn’t and the 500mW dissipation is now MOSFET. The drive current required fully discharge the timing capacitor; it transferred to TR2, the output device. is then reduced from about 2.3mA to stopped at about 1.5V. The JFET allows it to fall all the way to 0V, making the operation of this circuit with repeated R1 BR1 56Ω turning on and off more 1A/100V 20mA DIL Bridge – + 0.5W +42V 30V reliable. JFETs do cost AC **TR2 more than bipolar tranR7 1.8mA STP7NK80ZF (Rapid 47-0195) 5.6kΩ IRF510 (Rapid 47-0334) sistors, but the P-channel RFP15N05 J175 is still cheap and N-chan MOSFET TO220 has a low on-resistance. All voltage and currents are after relay has turned on C1 + 47µF 63V R2 R4 180kΩ 2.2MΩ TR1 J175 G *Note: TR1 is symmetrical; S and D can be swapped R3 330kΩ D* S* + R5 100kΩ +22.6V 3 + 7 +33V IC1 2 TL061 – C2 2.2µF +19V 25V Tant R6 150kΩ 4 2.2mA ZD1 33V/400mW BZY88 C3 220nF +31V 6 R8 1MΩ R9 1.2MΩ +16V G D1 1N4001 ZD2 12V 400mW BZY88 Final circuit D TR2** S RLY1 24V 700Ω L R 18mA Fig.5. Final FET version of Fig.4 de-thump circuit, giving reduced power consumption and faster reset. Practical Electronics | June | 2022 The final FET-based circuit is shown in Fig.5, and the Veroboard construction in Fig.6. For higher input voltages, R1 will need to be increased and the relay may be increased to 48V. No high-quality components are needed, generic parts will work 61 +15V Possible damaging currents through input pins if rail lost – IC1a 5532 + Output Big current flow if positive power rail lost –15V +15V 560Ω Resistors prevent input damage – IC1a 5532 Output + 560Ω –15V Fig.6. The de-thump circuit from Fig.5 constructed on Veroboard for testing. Layout is uncritical, and many experienced constructors can build it straight from the circuit diagram. If you don’t mind waiting, a PCB is in preparation. Fig.7. Losing the positive rail can cause the NE5532 op amp to self-destruct. fine. The high input impedance of TR1’s gate makes an ideal control input to connect a circuit to detect dangerous DC offsets on a power amplifier’s output. However, this feature is for another day and another PCB design. Miscellaneous Connectors, Molex or screw Veroboard 125mm x 55mm or bigger Relay, 24V 700Ω Component list Mutual rail shut-down (Final version – Fig.5) Another useful enhancement to the power supply – and any dual-rail power supply – is mutual rail shut-down. This functions Power rail symmetry by turning off the other rail if one rail is I don’t normally like trimmers, but it can shorted out or goes down for some reabe advantageous with some systems, such son. Occasionally, op amps and circuits as synthesisers, to be able set exactly can fail if one rail is lost. The audio engisymmetrical power supplies. It is possineer’s favourite op amp – the NE5532 – is ble, under worst-case tolerances, for the especially prone if it loses its positive rail, power supply outputs shown in Fig.8 to while still having a full negative rail, as be +14.2 and –15.8V. With op amp cirshown in Fig.7. This is particularly bad cuits it’s generally not the total magnitude in voltage follower circuits, where there of the voltage between the positive and is a direct connection from the output negative rails that matters, but how equal to the negative input. Also, this can be they are. The good thing is, we only need made worse if the non-inverting input is one trimmer for symmetry, rather than directly grounded. The total current can two (one for each rail). Of course, this exceed 24mA in these situations, possimeans you could end up with a ±14.5V bly exceeding the package dissipation or ±15.8V power supply, rather than the (780mW for the SOIC8 pack, 1.2W for theoretical ±15V, but usually the quality the DIP). This is why prudent designers of the balance outranks the actual voltput 560Ω resistors in series in these conage value of the outputs. Fig.9 shows the nections. Other circuits with low-impedance Use TVS P6KE20A-E3/54 DC-coupled loads, such LM317 +15V as distortion-cancelling Unregulated DC input, approx 23V 270Ω transformer drivers and 1kΩ 19V 1% TVS power amplifiers can BC337 even burn up. Also, in 1kΩ 3kΩ AC-coupled circuits, electrolytic capacitors 0V can become reversed 1kΩ 3kΩ polarised, causing electrical damage and BC237 19V 1kΩ physical liquid leakage. TVS 1% 270Ω The original SoundUnregulated DC input, approx –23V craft mixer power LM337 –15V supplies had mutual shut-down, but used two positive LM338 Fig.8. Mutual-rail shut-down circuit. Semiconductors IC1 TL061 low power BiFET single op amp TR1 J175 P-channel JFET (Mouser part number 512-J175D26Z) TR2 I used an RFP15N05 N-channel power MOSFET, but any equivalent >50V >1A >1W free-air rating, such as STP7NK80ZF (Rapid 47-0195) or IRF510 Rapid (47-0334) will work. D1 1N4001 ZD1 33V 400mW BZY88CV33 ZD2 10 to 18V 400mW BZY88CV12 BR1 1A 100V DIL bridge rectifier DB102 Rapid 47-2962 Capacitors C1 47µF 63V electrolytic C2 2.2µF 25V tantalum C3 220nF non-polarised, you can use any dielectric Resistors All 5% 0.25W carbon-film R1 R2 R3 R4 R5 R6 R7 R8 R9 62 56Ω 180kΩ 330kΩ 2.2MΩ 100kΩ 150kΩ 5.6kΩ 1MΩ 1.2MΩ Further power supply add-ons regulators. I modified the circuit to work with positive and negative regulators, but the principle is the same; using the variable voltage feature of the regulators to reduce it to the minimum using transistors to pull the adjust pin to ground. This drops the output voltage to 1.6V, which is almost off, and certainly enough to provide protection. The circuit is shown in Fig.8. Practical Electronics | June | 2022 + input LM317 +V 270Ω 10kΩ 3.3kΩ 0V 50kΩ 3.3kΩ Voltage symmetry 10kΩ 270Ω – input LM337 –V Fig.10. Shut-down circuit and rail symmetry trimmer board. Fig.9. Power rail symmetry trimmer circuit – tweak to make the rails equal. trimmer circuit. Note that the 3kΩ resistors (R7 and R8) have to be increased to 3.3kΩ to compensate for the parallel resistance of the trimmer network. Building the add-ons The shut-down circuit and the symmetry trimmer circuit can be built up together on the same bit of stripboard, as shown in Fig.10. Five wires go from this to the main power supply board. PTC/polyswitch fusing As mentioned last month, the LM317/337 short-circuit and over-temperature protection is a bit slow to react. The mains primary fuse can also be slow and possibly be changed by the user to one that has too high a current value. In view of this, it’s well worth adding extra protection in the form of positive temperature coefficient (PTC) self-resetting fuses in the transformer secondary AC lines. This also helps with the mutual shut-down, because the transistor circuit (Fig.8) will only work assuming the regulators are fully functional. If a regulator has an input/output short the mutual shut-down 15-0-15V 50VA Fig.12. Suppression capacitor and polyswitch fuses added to transformer secondaries. won’t work. I chose the 0.5A Raychem Littlefuse RXEF050 (Rapid 26-0720). Using big heatsinks for the regulators, the 0.65A RXEF065 also works. It’s worth spending 50p to protect a £20 transformer. Overvoltage *Good upgrade from previous 2200µF design: F2 Polyswitch Panasonic EEU-FS1V272, rated for 10,000hrs at 105°C, available from Mouser 0.65A AC input 15V 240V AC mains input 6.8µF 160V DC – + + 2700µF* 35V 15V F3 Polyswitch 0.65A + 2700µF* 35V Fig.11. Inserting polyswitch fuses and adding an anti-resonance capacitor to the transformer. The optimum capacitor value depends on the transformer. For most toroids, 1 to 6.8µF 160V does the trick. Practical Electronics | June | 2022 When a regulator fails, like most semiconductors, it often becomes a short circuit. If this occurs from input to output, the high unregulated input voltage will appear on the output and possibly destroy anything connected due to overvoltage. A sensible precaution is to place high wattage (5W) 20V Zener diodes or Transorbs across the output. A Vishay unidirectional transient voltage suppressor (TVS) diode P6KE20A-E3/54 (RS 811-9981) with a 19V breakdown voltage and 500W transient capability will fit in the holes for the 1N4001 start-up diodes, D20 and D21. When conducting these will trip the PTCs. A 40p upgrade that could save 20 op amps. Fine tuning The switching of the rectifiers can cause 100Hz modulated busts of ringing in the transformer (discussed in the Theremin Power Supply article: PE, August 2020). This can be damped down by placing a parallel capacitor after the PTCs, tuned for the transformer used, as shown in Fig.11. A photo of this arrangement is shown in Fig.12. That’s all for now, and it just goes to show a circuit is never finished, only improved. Next month is a total wind up – I’ll be looking at audio transformers! 63