Fullscreen HTML5Med Res (??MB)HTML4

0-30V 0-2A Part 1 by John Clarke bench supply Every workshop or laboratory needs an adjustable voltage, current-limited DC power source. This 0-30V Supply includes adjustable current limiting up to 2A with voltage and current metering, plus load switching. Most of the parts are commonly available; the two harder-to-get parts and the PCB are available from Silicon Chip. B ench power supplies are necessary for any workshop, powering electronic circuits and other loads such as small motors, LEDs and testing circuits. They are even pretty handy for charging batteries and the like. Looking back through our power supply projects, we haven’t published a basic workhorse supply like the one presented here that suits most workbench applications. We have published several dual tracking supplies and higher-current single output supplies, but 0-30V at up to 2A is sufficient for many applications. This being a simpler, cheaper design also makes it suitable for relative beginners to build. Our Supply includes metering that shows the voltage and the current being drawn from it. A load switch is used to connect or isolate the load when required, with an indicator LED to show when the output is on. The current limit can be adjusted from 28 Silicon Chip near zero to 2A to protect circuitry from excess current should there be a fault. A current limit indicator LED is also included. Load switching is over-ridden if the heatsink gets too hot, in which case the output is disconnected. In this case, the load indicator LED will remain off regardless of the load switch position. Our power supply includes power- up and power-down circuitry that protects the load as the Supply is switched on and off. This ensures the voltage from the regulator is fully settled before being applied to the load. Similarly, the load is disconnected quickly at power-off, well before the output drops significantly, preventing unexpected voltages from being applied to your load. Features & Specifications ∎ ∎ ∎ ∎ ∎ ∎ ∎ ∎ ∎ ∎ ∎ ∎ ∎ Easy to build using mostly standard components Low noise output Excellent regulation Output voltage: 0-30V Current limit: 0-2A (non-foldback) with indicator Regulation method: linear Load regulation: better than 0.5%, 0-2A Output noise & ripple: <8mV RMS, <50mV peak-to-peak <at> 2A Meters: voltage (100mV resolution), current (10mA resolution) Voltage adjustment: single-turn or multi-turn knob Load disconnect: load switch, load indicator Over-temperature protection: disconnects load when heatsink reaches 60°C Other features: short circuit protection, clean switch on and off Australia's electronics magazine siliconchip.com.au Scope 1: the Supply’s output voltage only dropped by 58mV with a 2A load step and recovered in about 300ms. Another valuable feature of our power supply is that you can adjust the output right down to 0V. Some very basic supplies will only go down to about 1.2V and there are times when that isn’t low enough. For example, if you are testing a circuit that runs from a single 1.2-1.5V cell and want to see how the circuit behaves when powered from a discharged cell at or below 1V. For the voltage adjustment, you can use a standard potentiometer. However, we recommend getting a multiturn potentiometer, especially if you want fine adjustment at low voltage settings. More on that later. The Supply is housed in a folded metal enclosure with an aluminium base and ventilated steel top cover. The front panel has the mains power switch, knobs to adjust the output voltage and current limit, the load switch, the two indicator LEDs and the voltage and current meters. There is just the mains power input socket and a heatsink on the rear panel. Scope 2: output noise and ripple with no load (top), 2A load (middle) and 1.92A current limited (bottom). Performance As this Supply uses linear regulation, it has excellent load regulation, clean current limiting and low output noise and ripple. Load regulation is tested by setting the voltage to a fixed level and changing the load resistance so that the output current rapidly swings between two extremes. With the output set to 16V, it dropped by less than 100mV when the load changed from 0A to 2A at the output terminals. When measured directly on the PCB, the voltage drop was 60% less. So most of the voltage drop is due to the wires from the PCB to the terminals on the front panel. We set the oscilloscope to monitor the AC voltage so that only the sudden changes in voltage are shown. Scope 1 shows what happens with a sudden load change. This revealed that the output momentarily dropped by 58mV when the load jumped from 0A to 2A. Similarly, when the 2A load was released, there was a positive shift Fig.1: the basic regulator arrangement is essentially the standard LM317 application from its data sheet but with current booster transistor Q1 added to increase the maximum output current and improve heat dissipation. As REG1 draws more current, the voltage across the 33W resistor at its input rises until Q1’s base-emitter junction becomes forwardbiased, and Q1 takes over delivering the load current. siliconchip.com.au Australia's electronics magazine of 34mV before recovery. Note the waveform does not show the DC voltage change, just the momentary shift in voltage from 16V. There is no visible change in voltage when the oscilloscope is set to show DC voltage at 2V/div so that the full DC voltage can be seen. That’s because 58mV and 34mV are only 0.4% and 0.2% of the output voltage, respectively. Output noise We measured the output noise and ripple under three different conditions: with the Supply unloaded, at 2A load and with the current limit active just below 2A. All three results showed low levels of noise and ripple. Scope 2 shows the output noise and ripple at 16V with no load for the top waveform, a 2A load for the middle waveform and current limited at 1.92A for the lower blue waveform. There is no discernible difference between the loaded and unloaded waveforms. However, there is a little more ripple for the current-limited waveform as current limiting is taking over from voltage regulation. Operating principles The basic circuit for our power supply (Fig.1) is based on an adjustable three-terminal regulator (REG1) and current boost transistor (Q1). REG1 is an LM317 that, in its standard arrangement, can deliver a voltage ranging from about 1.2V up to 37V at 1.5A. The regulator has internal protection such as current limiting, thermal shutdown and safe operating area (SOA) protection. The output voltage is set using October 2022  29 resistors connected between the output and adjust pins (R2; 100W) and between adjust and ground (VR1). The resistor between the adjust and output pins sets the quiescent current of the regulator, which needs to be at least 12mA if it is to maintain regulation when the output is otherwise unloaded. When the adjust terminal is connected to ground, the output voltage equals the reference voltage, which appears between the output and adjust pins. This is between 1.2V and 1.3V, depending on tolerances in the regulator manufacturing. For our circuit, the resistance is set at 100W to provide the 12mA minimum load current for the worst-case specification when the regulator reference is 1.2V. There is a minimal current of typically 50μA flowing out of the adjust terminal, which is small enough that it can usually be ignored. The output voltage calculation then simplifies to the following equation: Vout = Vref × (1 + VR1 ÷ R2). If you need to include the adjust terminal current, that current, multiplied by the VR1 resistance, adds to the output voltage. What the simplified circuit of Fig.1 does not show is that, in the full circuit, the lower end of VR1 is connected to a negative supply that is greater in magnitude than Vref. That way, the output can be adjusted down to 0V. With the reference voltage cancelled out, the output voltage calculation simplifies to Vout = Vref × VR1 ÷ R2. Current boosting As shown in Fig.1, REG1 is used in conjunction with PNP power transistor Q1. This transistor supplies the bulk of the load current but with the output voltage controlled by REG1. The input voltage is applied to the base of Q1 and the regulator input via a 33W resistor. As current is drawn from the output, it also flows through the 33W resistor, so the voltage across it rises. When 18mA flow is reached, the voltage between the base and emitter is 0.6V. At this point, transistor Q1 starts to conduct and bypasses extra current around REG1. The result is that the circuit can supply more current than the 1.5A limit of the LM317, while the LM317 remains in control of the output voltage. However, we do lose the over-­ current shutdown feature provided by 30 Silicon Chip Parts List – 30V 2A Bench Supply 1 double-sided PCB coded 04105221, 76 × 140mm (main board) 1 double-sided PCB coded 04105222, 56 × 61mm (front panel control board) 1 vented metal instrument case, 160 × 180 × 70mm [Jaycar HB5446] 1 30V 2A transformer (T1) [Jaycar MM2005] 1 current and voltage meter [Core Electronics 018-05-VAM-100V10A-BL] 1 fan type heatsink, 72mm high [Altronics H0522, Jaycar HH8572] 1 SPDT 10A, 24V DC coil relay (RLY1) [Altronics S4162C, Jaycar SY4067] 1 IEC male chassis connector with integral fuse holder [Altronics P8324, Jaycar PP4004] 1 1A M205 fast-blow fuse (F1) 1 rubber boot for IEC chassis connector [Altronics H1474, Jaycar PM4016] 1 DPST neon illuminated mains-rated switch (S1) [Altronics S3217, Jaycar SK0995] 1 SPDT toggle switch (S2) [Altronics S1310, Jaycar ST0335] 1 normally-closed 60°C thermal cutout (TH1) [Jaycar ST3821] 1 red binding post [Altronics P9252, Jaycar PT0453] 1 black binding post [Altronics P9254, Jaycar PT0454] 1 green binding post [Altronics P9250, Jaycar PT0455] 1 silicone insulating washer for TO-3P package devices 1 silicone insulating washer and bush for TO-220 package devices 2 4-way pluggable terminal sockets, 5.08mm spacing (CON1, CON2) [Altronics P2574, Jaycar HM3114] 2 4-way screw terminal plugs (for CON1 & CON2) [Altronics P2514, Jaycar HM3124] 2 14-pin IDC boxed headers (CON3, CON4) [Altronics P5014] 2 14-pin IDC line sockets (for CON3 & CON4) [Altronics P5314] 1 3-way screw terminal with 5.08mm spacing (CON5) 2 2-pin vertical polarised headers, 2.54mm spacing (CON6, CON7) [Altronics P5492, Jaycar HM3412] 1 2-pin polarised header plug (for CON7) [Altronics P5472 and 2 x P5470A, Jaycar HM3422] 1 8-pin DIL IC socket (optional; for IC1) 2 5mm LED bezels 1 knob to suit VR1 1 knob to suit VR3 10 1mm PC pins (add 12 if using them for all test points) Wire & cable 1 150mm length of 14-way ribbon cable 1 150mm length of brown Active wire stripped from three-core 7.5A mains cable 1 150mm length of blue Neutral wire stripped from three-core 7.5A mains cable 1 150mm length of green/yellow Earth wire stripped from three-core 7.5A mains cable 4 100mm lengths of 7.5A hookup wire (assorted colours) 2 150mm lengths of 7.5A hookup wire (one red, one black) Hardware etc 4 M4 × 10mm panhead machine screws 4 M4 hex nuts 4 M4 star washers 4 6.35mm-long M3-tapped Nylon spacers 8 M3 × 5mm panhead machine screws 2 M3 × 20mm panhead machine screws (for Q1 and REG1) 4 M3 × 15mm panhead machine screws 1 M3 flat steel washer 6 M3 Nylon washers 6 M3 hex nuts 2 small M3.5-threaded right-angle brackets [Jaycar HP0872, pack of 8] 2 crimp eyelets (Earth connections to chassis) Australia's electronics magazine siliconchip.com.au 4 blue female spade crimp connectors (connections to mains on/off switch) 5 150mm cable ties 3 100mm cable ties 1 50mm length of 25mm diameter heatshrink tubing 1 50mm length of 6mm diameter heatshrink tubing 1 50mm length of 3mm diameter heatshrink tubing 1 small tube of thermal paste Semiconductors 1 TL072P dual op amp, DIP-8 (IC1) [Altronics Z2872, Jaycar ZL3072] 1 INA282AIDR or INA282AQDRQ1 shunt monitor, SOIC-8 (IC2) [SC6578] 1 LM317T three-terminal adjustable regulator, TO-220 (REG1) [Altronics Z0545, Jaycar ZV1615] 1 LM336-2.5 voltage reference, TO-92 (REG2) [Altronics Z0557, Jaycar ZV1624] 1 TIP36C PNP 100V 25A power transistor, TO-3P (Q1) [Altronics Z1137, Jaycar ZT2294] 1 2N7000 N-channel Mosfet, TO-92 (Q2) [Altronics Z1555, Jaycar ZT2400] 3 BC547 45V 100mA NPN transistors, TO-92 (Q3-Q5) 1 BC327 45V 500mA PNP transistor, TO-92 (Q6) 2 5mm high-brightness red LEDs (LED1, LED2) 1 33V 1W zener diode (ZD1) [1N4752] 2 12V 1W zener diodes (ZD2, ZD3) [1N4742] 1 BR106, PB1004 or KBPC1006 bridge rectifier (BR1) [Altronics Z0085/Z0085A, Jaycar ZR1320] 6 1N4004 400V 1A diodes (D1, D3, D4, D7, D8, D10) 1 1N5404 400V 3A diode (D2) 3 1N4148 75V 200mA signal diodes (D5, D6, D9) Capacitors 3 4700μF 50V radial electrolytic 1 2200μF 35V radial electrolytic 1 1000μF 16V radial electrolytic 1 47μF 16V radial electrolytic 1 10μF 50V non-polarised/bipolar radial electrolytic 1 10μF 35V/50V/63V radial electrolytic 2 10μF 16V radial electrolytic 1 1μF 16V radial electrolytic 1 1μF multi-layer ceramic 4 100nF 63V/100V MKT polyester Potentiometers 1 16mm 5kW linear single-gang potentiometers (VR1●) [Altronics R2224, Jaycar RP7508] 1 16mm 10kW linear single-gang potentiometers (VR3) [Altronics R2225, Jaycar RP7510] 2 5kW multi-turn top-adjust trimpots (VR2●, VR4) [Altronics R2380A, Jaycar RT4648] 1 500W multi-turn top-adjust trimpot (VR5) [Altronics R2374A, Jaycar RT4642] 2 10kW multi-turn top-adjust trimpots (VR6, VR7) [Altronics R2382A, Jaycar RT4650] ● alternatively and ideally, replace VR1 with a 2.5kW multi-turn pot [Bourns 3590S-2-252L – element14 2519607; Digi-Key 3590S-2-252L-ND; Mouser 652-3590S-2-252L] and delete VR2 Resistors (all 1/2W, 1% unless otherwise stated) 2 100kW 1 33kW 4 10kW 2 4.7kW 1 3.3kW 1 2.2kW 1W 5% 1 2.2kW 2 1kW 1 330W 4 100W 1 33W 1 20mW 1W M3216/1206-size SMD resistor [Vishay WSLP1206R0200FEA or similar – element14 1853240; Digi-Key WSLP-.02CT-ND; Mouser 71-WSLP1206R0200FEA; part of SC6578] siliconchip.com.au Australia's electronics magazine the LM317, limiting the output to 1.5A. But that’s what we need to get a higher output current. We use extra circuitry to add back current limiting, with the advantage of being able to adjust the limit over the 0-2A range. This boost circuit includes a hidden bonus in that it prevents the regulator from shutting down due to high power dissipation (assuming Q1 has sufficient heatsinking). This way, the circuit can supply the full 2A across the entire voltage range. Without the boost transistor, the regulator would shut down when there is high dissipation, ie, high current at low output voltages. For example, if the regulator output voltage is 12V, the input is 32V and there is a 1A current flow, the regulator (without Q1) will be dissipating (32V − 12V) × 1A = 20W. The specifications for the device package show a 5°C/W temperature rise between the case and junction. Thus, at 20W, the junction temperature will rise 100°C above the case (20W × 5°C/W). For a case temperature of 25°C, the junction will be at 125°C and the device will shut down. So the Supply wouldn’t be able to provide 1A at 12V without shutting down. By adding the boost transistor, REG1 is only handling 18mA and dissipating about 360mW in this case (18mA × [32V − 12V]) and the junction will only be 1.8°C above the case temperature. The dissipation is instead handled by Q1. Its junction temperature will not be anywhere near as high as the regulator, as it has a much lower junction-to-case thermal resistance of 1°C/W. So at 20W, its junction will only be 20°C above the case temperature. Using a large enough heatsink, we can maintain the case temperature at a reasonably low value. We do lose the thermal shutdown feature of the LM317 as a consequence of directing the primary current through Q1. The junction temperature for REG1 will essentially follow the temperature of the heatsink. To solve this, we attach a separate thermal switch to the heatsink to provide an over-temperature shutdown. It opens at 60°C, disconnecting the power supply load and allowing the transistor to cool. We haven’t mentioned the capacitors in Fig.1. The bank of three 4700μF capacitors at the input smooths out the ripple from the pulsating DC derived October 2022  31 Fig.2: the complete Supply circuit. Note how many signals are routed to CON3, then via a ribbon cable to CON4 on the front panel control board, and in some cases, back through the cable to another pin on CON3. from rectified AC. This is required to keep the regulator’s input voltage at least 2.5V above the output to maintain voltage regulation. The capacitor between REG1’s ADJ pin and ground reduces ripple and noise at the regulator output, while the capacitor between Vout and GND prevents oscillation and improves transient response. Diode D1 protects REG1 from the capacitor discharging through REG1 if the output is short-­ circuited. 32 Silicon Chip Full circuit details The whole circuit is shown in Fig.2. Power for the Supply is derived from the mains via transformer T1. T1’s primary winding is supplied with 230V AC via fuse F1 and power switch S1. The secondary winding between the 0V and 24V taps of T1 is fullwave rectified by bridge rectifier BR1 and filtered using three 4700μF 50V capacitors to produce a nominal 32V DC. Typically, the DC voltage is higher than this as the mains is usually above Australia's electronics magazine 230V AC, and the transformer is not usually heavily loaded. This filtered voltage is applied to the emitter of transistor Q1. The output of the regulator and the collector of Q1 are applied to the load via the normally-open contact of relay RLY1. The relay control circuitry will be described later. Bringing the output to 0V The circuitry around REG1 differs from that shown in Fig.1 in that, siliconchip.com.au instead of connecting to GND, VR1 is connected to the output of op amp IC1a. IC1a produces a negative voltage below ground, to cancel out the reference voltage of REG1. Setting IC1a’s output negative by the same siliconchip.com.au magnitude as REG1’s reference voltage will allow the output to go to zero. A negative voltage is derived via the 30V tapping on the secondary of the mains transformer to produce a -8V supply. This is achieved by diode D3 Australia's electronics magazine that half-wave rectifies the AC voltage, and a 1000μF capacitor filters it. Diode D4 prevents this supply from going above 0V by more than 0.6V and prevents significant reverse polarity from being applied to the capacitor October 2022  33 Everything fits neatly into the fairly compact and attractive instrument case. You can see transistor Q1 at left, attached to the case opposite the heatsink, with the thermal switch above it. The blue multi-turn voltage adjustment pot is also clearly visible. when the power is switched off. By all appearances, the -8V supply should work. But there is a hidden problem: unless the main 32V supply derived from the bridge rectifier has sufficient load, the -8V supply will not be available. This is because, under light load situations, there is no current path for the -8V supply current through D3 to flow back through the bridge rectifier. The only way is blocked by the diode in the bridge between the 24V tap and the ground supply rail. With the -8V supply, current only flows during the parts of the mains cycle when the 24V and 30V taps produce a negative voltage with respect to the 0V end of the windings. So current has to flow through the diode in BR1 that connects from the 0V transformer tapping and positive supply, then through the load on the main supply and -8V supply and back to the 30V tap via D3, as shown in Fig.3. If the load on the main supply is insufficient to maintain the -8V supply, its magnitude will drop while the voltage applied to the main supply from the 30V tapping will increase. This is resolved by adding a 2.2kW 1W resistor from BR1’s positive terminal to ground, setting a minimum load so the -8V rail is always available. The -8V supply provides a bias current for REG2, an LM336-2.5V shunt regulator. It produces a regulated negative supply with its positive terminal connected to ground, and its negative terminal connects to the -8V supply 34 Silicon Chip via a smaller 2.2kW current biasing resistor. As a result, the voltage at its negative terminal is a stable -2.49V even with temperature variations due to diodes D5 and D6 providing temperature compensation. Trimpot VR7 is adjusted until there is very close to -2.49V across REG2. This reference voltage is bypassed with a 10μF capacitor. Trimpot VR6 connects across the -2.49V reference to provide an adjustable negative voltage to offset the reference voltage produced by REG1. This negative reference is obtained from the wiper of VR6, which is adjusted to provide a fixed voltage between -1.2 to -1.3V to counter REG1’s reference voltage between its output and adjust pins. The wiper of VR6 connects to the non-inverting input of IC1a. IC1a acts as a unity-gain buffer, where the output voltage follows the input. IC1a’s output then sinks 12-13mA from REG1 at the lower end of VR1. With VR6 correctly set, REG1’s output is zero when VR1 is fully anticlockwise. Current monitoring Fig.3: the negative supply generator used to adjust REG1’s minimum output voltage (among other purposes) seems straightforward, but there’s a trick to it. The load resistance on the main rectifier (the ‘resistor’ at upper right) must be low enough for the current to flow through the path shown in red. Otherwise, the negative supply drifts positive. We ensure this is the case by adding a 1W ‘dummy’ load resistor across the positive supply. Australia's electronics magazine IC2 measures the current drawn by the load. This measurement, in conjunction with op amp IC1b and Mosfet Q2, is used to provide current limiting. IC2 is a current monitor that measures the voltage drop across the 20mW shunt in the GND supply line. The voltages at either end of the shunt are applied to pins 1 and 8 of IC2, which amplifies the difference by a factor of 50. We selected the shunt so that the pin 5 output of IC2 provides 1V per 1A of output current. There is a 20mV voltage drop across the 20mW shunt at 1A, which, when multiplied by 50, gives 1V. But note that IC2’s output voltage is with respect to the -2.49V reference rather than the 0V rail. The calibration is linear, so IC2 will deliver 2V above the -2.49V reference for a 2A current flow or proportionally lower values at intermediate currents. siliconchip.com.au There isn’t much on the rear panel – just the heatsink and IEC mains power input. Note how the heatsink hangs down below the bottom of the case as it is slightly taller. We get around this by making the case’s feet taller. For current limiting, we compare the current measured by IC2 with the maximum set current level. The current setting for limiting is provided by a voltage divider across the -2.49V supply. The main adjustment is potentiometer VR3, with VR4 and VR5 setting the maximum and minimum current range limits. Ignoring VR5 for the moment, VR4 is set so that when VR3 is set fully clockwise, the voltage at its wiper will be 2V above the -2.49V reference, corresponding to a 2A current limit. VR5 provides a small voltage offset above the -2.49V reference. It is used to set the minimum setting of VR3 to match the output of IC2 when there is no load current. Typically, IC2’s output will always be above the -2.49V reference due to the small standby current drawn by the reference, IC1, IC2 and the meters. Also, there will be an offset voltage inherent to IC2 even with no current flow. VR5 allows us to dial out this voltage so that the voltage between test point TP10 (at the top of VR5) and TP3 (at the wiper of VR3) ranges between 0V and 2V, matching the 0-2A current limit range. If the VR5 adjustment is made carefully, that will also allow VR3 to be rotated fully anticlockwise without entering current limiting when there is no load. The current limit setting voltage from VR3’s wiper is applied to the inverting input of IC1b via a 1kW siliconchip.com.au resistor. This voltage is compared with the output from IC2, which goes to the pin 5 inverting input of IC1b via a 10kW resistor. When IC2’s output is lower than the setting for VR3, IC1b’s output (pin 7) is pulled low, towards its pin 4 supply (-8V). In this case, current limiting indicator LED1 is reverse-biased, so the gate of Mosfet Q2 is held at its source voltage, with no current flowing through the Mosfet. When the output from IC2 goes above the threshold set by VR3, the output of IC1b begins to go high, lighting LED1 via the 1kW resistor between Q2’s gate and source pins. This also starts to switch on Q2 as its gate voltage rises. The channel of Mosfet Q2 then begins to conduct, pulling the adjust terminal of REG1 down to reduce its output voltage. Note that the adjust terminal is isolated from the voltage setting resistance of VR1 via a 330W resistor. This allows Q2 to drive the adjust terminal without being loaded by the voltage setting resistance. The 100nF capacitor between pin 5 of IC1b and the drain of Q2 acts as a compensation capacitor for the current limiting feedback, preventing it from coming on too rapidly, possibly leading to oscillation. Compensation for the op amp is also provided using a 1μF capacitor between the pin 6 inverting input and the pin 7 output. While this capacitor Australia's electronics magazine could be as low as 47pF to prevent oscillation, the 1μF value gives better output ripple reduction when the supply is in current limiting. Load switching As mentioned previously, we use a relay to switch the Supply’s output to the load. This relay (RLY1) allows the circuitry to disconnect the load during power-up, power-down or if the heatsink gets too hot. Disconnecting the load when power is first applied, and when it is switched off, prevents unexpected voltages from being applied to the load. This circuit section comprises diodes D7 and D8, transistors Q3 to Q6 and associated components, plus RLY1. We use the 18V transformer tap to derive a 25V supply. Diode D7 halfwave rectifies the AC, and a 2200μF capacitor filters the resulting voltage to a relatively smooth 25V DC or so. The positive power supply for op amp IC1 is taken from this rail via a 100W resistor. As the negative supply for IC1 is from the -8V rail, ZD1 is included to ensure that the overall supply to IC1 does not exceed 33V. Diode D8 also provides half-wave rectification of the 18V tapping, but this is not filtered so that we have a pulsating voltage. This way, the voltage from diode D8 will immediately cease when power is disconnected, allowing us to quickly detect when the power is switched off. October 2022  35 When power is applied, the positive voltage at D8’s cathode switches on transistor Q3 for half of every mains cycle. With our 50Hz mains, the positive excursion is over a 10ms period. Q3 discharges the 1μF capacitor via a 100W resistor each time it is switched on; this capacitor begins to charge via a 100kW resistor from the 25V supply during the negative half of the waveform. This capacitor will stay mostly discharged, provided that Q3 repeatedly discharges it every 10ms. Potentiometer options We have provided the option of using a standard single turn (300° rotation) potentiometer for VR1, which adjusts the Supply output voltage. In this case, it’s a 5kW linear potentiometer connected in parallel with a 5kW trimpot. This is the cheapest option, but not the best. The alternative is to use a 2.5kW multi-turn potentiometer, making it easier to adjust the output voltage, especially for low values. While we are using a potentiometer for the voltage adjustment, it is used as a variable resistance (or rheostat) rather than as a potentiometer. With a potentiometer, the wiper can produce a range of voltages between the voltages applied at the two ends of the potentiometer’s track. The wiper and just one end of the potentiometer are used to produce a variable resistance. The unconnected end of the potentiometer is often connected to the wiper, but this does not alter the resistance-versus-rotation law. When using a standard 300° potentiometer to adjust the voltage over a 0-30V range, a slight adjustment causes the output voltage to change quickly. So, for example, a 0.3V change is made with each 1% (3°) of rotation. So to change the voltage by 1V, just over 3% of rotation (10°) is required. Another problem is that while the physical end stops are 300° apart, the actual resistance element generally only changes over a 270° range, further ‘squashing up’ the adjustment range. Also, we don’t use a 2.5kW single-turn pot since they are difficult to obtain and rather expensive. Instead, we use a 5kW linear pot in parallel with a 5kW resistance to provide an overall 2.5kW range. This means that the plot of resistance vs rotation is not linear, further exacerbating the adjustment sensitivity for low voltage values, as shown in the plot below. The cyan line is for a 2.5kW linear pot, while the red line plots the resistance law for the 5kW pot in parallel with a 5kW resistance. The parallel resistances do not provide a linear change in resistance versus rotation, with the largest difference being near the ends of the pot rotation, making accurate low-­voltage adjustment even more difficult. For the first 10% of rotation, the linear 2.5kW pot changes resistance by 250W, while the 5kW pot and 5kW parallel resistance changes by nearly 500W. At half rotation, the 2.5kW pot measures 1.25kW (half the total resistance value), while the 5kW pot gives 1.67kW (2/3 of the resistance value). At 90% rotation, the 2.5kW pot is at 2.25kW (90% of the total resistance), while the 5kW pot gives 2.37kW (95% of the resistance). This non-linearity causes the adjustment at the low end to be much coarser than in the middle of the range. This plot shows the difference in resistance vs rotation for a regular 2.5kW pot and a 5kW pot shunted with a fixed resistance. They start and end at the same points, but the shunted pot’s resistance law is not linear. If you can get the multi-turn 2.5kW potentiometer to use for the output voltage adjustment, you’ll be able to set the output voltage much more easily and accurately. 36 Silicon Chip Somewhat similarly, transistor Q4 controls the charge on the 47μF capacitor. When Q4 is off, it allows the 47μF capacitor connected to TP8 to charge via the 100kW and 100W resistors. Q4 remains off, provided that the 1μF capacitor connecting to Q4’s base is discharged. So when there is an output from the transformer, the 47μF capacitor charges up. The base of Q5 needs to be above 13.2V to switch on due to the voltages across diode D9 and zener diode ZD2, the latter being biased via a 3.3kW resistor from the 25V supply. As a result, when power is first applied, there is a five-second delay before the 47μF capacitor charges enough to switch Q5 on. But when the power switch is flicked off, within a few tens of milliseconds, the 1μF capacitor at Q4’s base charges enough to switch it on, discharging the 47μF capacitor and switching Q5 off. When Q5 is on, it pulls current from the base of PNP transistor Q6 via a 4.7kW current-limiting resistor. The current from Q6 flows through the load switch (S2), then through thermal switch TH1 and to the relay coil. So the load is only connected by RLY1 when Q6 is on, S2 is on and thermal switch TH1 is not open. To put it another way, the load is disconnected during power-up, power-­ down, when S2 is off or when the temperature of TH1 is too high. Diode D10 clamps the negative voltage when the relay coil is switched off. By the way, we sneakily reuse the 12V supply from zener diode ZD2 to power IC2, the INA282 shunt monitor. Metering The voltmeter and ammeter connect to the regulated output of the Supply. The voltmeter measures the voltage before the relay contact. The shunt for current measurements is in the negative supply line; it has a very low resistance, so there is a minimal voltage drop across it. The meter is supplied from the 25V positive rail and uses the MI- terminal as its ground. Next month We have now described what our new Supply can do and how it works. Next month’s follow-up article will have the assembly details for the two PCBs, chassis assembly instructions and wiring details. SC Australia's electronics magazine siliconchip.com.au