Fullscreen HTML5Med Res (??MB)HTML4

A Switchmode Replacement for 78xx regulators By Tim Blythman The 78xx series of three-terminal linear regulators started as the LM109/309 in 1969. So they’ve been around for about 50 years, and they are undeniably still useful today. Their biggest disadvantage is inefficiency, especially with a large input/output voltage difference. If only there was an efficient, drop-in alternative! W Of course, there are plenty of switch- efficiency at higher currents and volte have been using 78xx series linear regulators since the mode ICs which do a similar job, but age differentials. It’s built on a board first issue of SILICON CHIP in they almost always require quite a few that’s roughly the same size as a TONovember 1987 and we still use them extra ‘support’ components, possibly 220 package and has the same three including a bulky inductor. And some- connecting leads. And it’s relativeextensively today. There is no doubt that they are a times selecting the right components ly inexpensive and doesn’t use very simple and effective way of getting is a bit of a ‘black art’. Even then, the many components. However, we must point out that a well-controlled fixed voltage sup- result may not match the performance ply between 3.3V and 24V. They’re of a 78xx; for example, the allowable sometimes, a linear regulator is precheap, they’re available everywhere range of input voltages may be more ferred, mainly because its output does limited. not have switching artefacts (such as and they’re easy to use. This article describes a switchmode high-frequency ripple). Linear regulaFor example, our 45V Bench Supply from October and November 2019 regulator that can be used as a direct tors may also have better line and load (siliconchip.com.au/Article/12014) replacement for a 78xx type regulator regulation. Switchmode regulators are used three 78xx series positive regula- in most cases, but with much greater continually improving in this regard, but we understand that there tors and one 7905 -5V regulawill always be cases where a tor to provide regulated rails Features & specifications linear regulator is required. for its circuitry. The ideal solution is often But being linear devices, • Input voltage: 4-30V to combine a switchmode they can be inefficient, and • Output voltage: 2-24V pre-regulator with a lowthis causes two major prob- • Output current: up to 1A dropout linear post-regulalems. Not only is much of • Quiescent current: around 80µA tor. That gives you the best the supplied energy wasted, • Efficiency: typically 90-96% of both worlds. Our Hybrid but it must be adequately • Dropout voltage: 0.5V Bench Supply from Aprilremoved from the device to • Size: equivalent to a TO-220 package semiconductor device June 2014 used this apprevent overheating. In oth- • Heatsinking: not required proach; see siliconchip.com. er words, more inefficiency • PWM frequency: 500kHz; lower at light loads au/Series/241 means more dissipation and • External capacitance required: 1µF+ at input, 22µF+ at output Thus, in the space taken up more dissipation means more • Other features: under-voltage lockout (4V), thermal shutdown, over-current/short circuit protection by two TO-220 parts, you can heatsinking is required. 38  Silicon Chip Australia’s electronics magazine siliconchip.com.au even implement such a hybrid regulator arrangement by using our design and then passing its output to a discrete linear regulator. The latter should ideally be a low-dropout type, but a 78xx could be used if maximum efficiency is not required. The IC at the centre of this design can deliver any voltage from 2V to 24V, with the output voltage of our Regulator set by just one resistor value. So this design can replace not just one part, but many. You might also be wondering about parts like the ubiquitous LM317 adjustable regulators. They have a different pinout to the 78xx series, so it isn’t possible to make a one-size-fits-all solution that addresses both of these families. While it is possible to fit this device in place of an LM317 in many cases, you would need to make some changes to the surrounding circuitry, including deleting the external resistors which set its voltage. Our replacement device The 78xx we know and love is the one we find in a TO-220 package. This version alone appeared in half a dozen circuits that we published last year. There are also variants in the smaller TO-92 package (the 78Lxx) and SMD TO-252 (78Mxx in surface-mounting D-PAK) packages. It’s the TO-220 package that we’re targeting, because if you can get away with one of the smaller variants, the chances are that you don’t have too much dissipation to worry about. Also, it’s harder to cram the necessary parts into the smaller spaces that these packages offer. If your intended application has a 78xx bolted onto a chunky heatsink, then you’re going to benefit most from our upgrade. And that’s precisely what this project is; a drop-in replacement regulator for the hot, inefficient IC that’s wasting energy on your design. Our design is easily adaptable for Our switchmode Regulator has a very similar outline to the 78xx linear regulator it is intended to replace. With careful choice of parts, the thickness The design can be kept much the same too. If you We wanted our Regulator to be as have space available, you may wish to close as possible to a direct substiuse a larger inductor or larger capacitors tute for the 78XX in a TO-220 packto improve its performance. many voltages; it can be used in place of a 7833 (3.3V), 7805, 7806, 7808, 7809, 7810, 7812, 7815 or 7824. It might also be suitable to replace one of the many low-dropout three-terminal fixed regulators on the market (although their pinouts don’t always match the 78xx). age, and the first item we considered was the size. The body of a TO-220 part is around 10mm x 15mm; a minuscule size for a PCB. But it would not be a dropin replacement if it doesn’t fit in the same space. We decided to leave off the tab mounting hole, since our design will How switchmode regulators work There are many types of switchmode regulators in use today. This includes step-down (buck), boost, flyback, buck/boost, SEPIC, resonant and fully isolated types. But step-down/buck is probably the most common configuration and is also, in a sense, the simplest (with boost not far behind). This is a step-down/buck design. A linear regulator reduces its output voltage by simply introducing a controlled resistance in series with the load. If the input voltage is twice the output voltage, this means that 50% of the power going into the regulator is turned into heat. That means poor efficiency. If your aim was to vary the power to something like a LED or lamp, which only responds to the average current, you could get much greater efficiency by applying the full input voltage to it but only doing so 50% of the time. This could be done using pulse width modulation (PWM), and indeed that is how most DC lamp dimmers and simple motor speed controllers work. The voltage is normally switched by a transistor, with the transistor either fully off (and passing no current) or fully on (dropping no voltage). Little power is lost in the switching element, with real-world efficiencies coming quite close to 100%. But such an arrangement is not suitable for powering ICs or other devices which expect a more-or-less constant supply voltage. Thus, to get a similar efficiency to the PWM approach when siliconchip.com.au powering such devices, we need to ‘filter out’ the rapidly changing part of the waveform (the AC component), giving us just an average voltage level (the DC component). An LC low-pass filter is a simple way to do this. We can’t use an RC filter since we would have half the voltage across the resistor, so efficiency would be no better than a linear regulator. But with an LC filter, energy is stored in both elements (the inductor and the capacitor). Most of that is returned later, so losses and heating are minimal. In the case of the inductor, excess energy is briefly stored in its magnetic field. One way to think of this approach is that applying pulses of voltage to an inductor forms something like a constant current source. At the same time, the capacitor makes the load impedance very low at high frequencies, resulting in a fairly unchanging voltage across the load, despite the pulses applied by the transistor. There will still be some amount of ripple present at the load, but with the correct choice of components, it can be reduced to a manageable amount. In fact, the amount of tolerable ripple dictates the required switchmode frequency and capacitor and inductor values. The best way to reduce ripple is to use the largest inductor and capacitor values possible. In practice, size is an issue, particularly with inductors, so we are forced to compromise (too large an inductor can also affect the regulator’s response to load transients). Australia’s electronics magazine August 2020  39 CON1 OUTPUT GND INPUT 3 2 L1 22 H /1A 4 1 3 1 F 35V 2 1 F 6.3V X7R SC 2020 VIN EN VOUT 6 7 REG1 BOOST MCP16311 VFB VCC AGND 8 100nF R1 52.3k 1 F 1 6.3V X7R PGND 5 10k HIGH EFFICIENCY SWITCHMODE REGULATOR (5V) Fig.1: the circuit of the Regulator is just about straight out of the MCP16311 data sheet, except that the input and output capacitors are lower than recommended. That’s because these are supplemented by external capacitance on the host board. The values in red need to change for a different output voltage. 40  Silicon Chip package (eight-pin micro small outline package). We found a device that came in this package, with a good compromise of most of the features we wanted. By the way, MSOP packages have varying pin pitch, sometimes 0.635mm (the same as SSOP and TSSOP) and sometimes even smaller, at 0.5mm. But they’re also narrower than SSOP and TSSOP, so are one of the most compact packages that can be hand-soldered without too much difficulty. Switchmode operation If you aren’t familiar with the operation of switchmode regulators, see our panel “How switchmode regulators work”. This also explains some 6 5 OUTPUT VOLTS not need to dissipate anywhere near as much heat. So there is no need to attach it to a heatsink. While this does also remove the option of using a mounting screw to secure the part, our Regulator uses sturdy pin headers which are thicker than the leads on most discrete parts. If absolutely necessary, silicone sealant or other adhesive can be used to provide mechanical support. In any case, the Regulator PCB with all its parts is typically around half the weight of a TO-220 device, so the mechanical stresses will be less. With a PCB size set, we started looking for the best switchmode regulator IC to use. We needed to choose one which we could fit on this small PCB, including all the required supporting components. We found it difficult to find suitable parts that could work up to the nominal 35V input that the 78xx series can tolerate. In the end, we settled for a part with a 30V rating, as this covers most use cases. We considered using a device in a user-friendly SOIC-8 SMD package, but one of these would take up around a quarter of the available space on the PCB. Other parts we found came in QFN (quad flat no-lead) and DFN (dual flat no-lead) packages, but we decided that these would be too difficult for many people to solder. You need a reflow oven or hot air station to have much chance of success. So we limited our search to parts with leads. A decent compromise between size and ease of soldering is the MSOP-8 of the other design considerations we had to take into account. While sorting through the (huge number of) switchmode regulator ICs that are available, we looked at several features. Firstly, high-frequency operation is necessary. This means that a lower inductor value is needed, which reduces its physical size. A higher frequency also means less ripple and noise. We also looked for parts which operate synchronously, rather than requiring an external diode. While it is only one extra part, the diode does carry a fair amount of current, so choosing a synchronous part means that some space and dissipation is saved. The voltage drop across the low-side Mosfet (which replaces the diode’s function in synchronous designs) is less than that of the diode. Ultimately, we settled on the Microchip MCP16311. It has a switching frequency of 500kHz and operates synchronously with a minimum number of external components for an adjustable output. As noted earlier, it can operate with up to 30V on its input. We initially tried to lay out the PCB using 3216-size (1206 imperial) passive components. These measure 3.2mm x 1.6mm, but were too large, so we switched to 2012-size (0805 imperial) parts measuring 2.0 x 1.2mm. These save a significant amount of space on the PCB, but aren’t too much harder than 3216-size parts to solder. The footprints that we’ve provided on the PCB are actually a tiny bit larg- SWITCHMODE 4 7805 3 2 1 0 0 1 2 3 4 5 6 7 8 9 INPUT VOLTS Fig.2: the Switchmode Regulator does not operate with an input supply below 4V. At 4V and above, though, it has a much lower dropout voltage than the 7805 and attains a 5V output with only 5.5V at its input (ie, 0.5V dropout). The 7805 needs nearly 7V on its input before it is in regulation. Australia’s electronics magazine siliconchip.com.au er than 3216/1206 parts, so you might be able to use the slightly larger 1206 parts anyway. The circuit The circuit for our design is shown in Fig.1, with the components for a 5V output. IC1 is the MCP16311 integrated switchmode controller. It works with 4.4-30V at its input (pin 4, VIN) and can deliver 2-24V at up to 1A. Pin 3, the enable (EN) input, is tied to VIN so that the IC is enabled as long as there is a sufficiently high supply voltage. The input supply is bypassed by a 1µF capacitor. While this is less than the recommended capacitance in IC1’s data sheet, any application using a 7805 requires an external bypass capacitor anyway, which will supplement the capacitance fitted to the PCB. Pins 5 and 8 are connected to ground, with pin 5 being the highcurrent return for the synchronous switch, while pin 8 is the low-current reference ground to which the output voltage is referred. Both are connected to large copper pours on the PCB. IC1 has an internal low-voltage regulator for its control circuitry, which should be bypassed by a 1µF capacitor connected between pin 2 and ground. This pin sits around 5V, so a 6.3V capacitor is adequate. Pin 1 is connected to IC1’s internal regulator feedback circuitry. The voltage at pin 1 is compared to a precision 0.8V reference, so this pin should be Desired Vout R1 (E96) Nominal Vout R1 (E24) L1 3.3V 31.6k 3.328V 30k 3.2V 15µH (eg, SRN6028-150M) 5V 52.3k 4.984V 51kV 4.88V 22µH (eg, SRN6028-220M) 6V 64.9k 5.992V 62k 5.76V 27µH (eg, ASPI-6045S-270M) 8V 88.7k 7.896V 91k 8.08V 39µH (eg, ASPI-6045S-390M) 9V 102k 8.96V 100k 8.8V 39µH (eg, ASPI-6045S-390M) 10V 115k 10V 110k 9.6V 47µH (eg, SRN6028-470M) 12V 140k 12V 130k 11.2V 56µH (eg, SRN6045TA-560M) 15V 178k 15.04V 180k 15.2V 68µH (eg, TYS6045680M-10) 24V 287k 23.76V 300k 24.8V 120µH (eg, SRN6045TA-121M*) * current rating is 850mA, so don’t draw more than this (the output voltage may drop before reaching that level). For more current, you can probably get away with a 100µH inductor, part code ASPIAIG-S6055-101M. Table 1: Component choices connected to the midpoint of a voltage divider between the output and ground. The ratio of this divider sets what fraction of the output voltage is seen at pin 1 and thus dictates the output voltage. The MCP16311 data sheet recommends a 10k resistor for the lower part of the divider, so changing the output voltage is simply a case of setting the upper resistor. For a 5V output, the upper resistor should ideally be 52.5k. While From left to right, a 3.3V Regulator, a 5V Regulator and a 12V Regulator. Note that the inductor needed is much larger for higher voltage versions. This version is only 6mm thick, which is more than the 5mm of many 78xx regulators, but still slim enough to fit in most places where one would be used, especially as no heatsink is normally required. siliconchip.com.au Nominal Vout Australia’s electronics magazine changing this resistance will set a different output voltage, for optimum performance, other components must be adjusted too. In practice, 52.3kis the closest commonly available value, from the E96 (96 values per decade) series. This gives a nominal 4.984V output. For comparison, a 51k resistor (found in the more common E24 series) would give a nominal 4.88V output. Unless you need a precision voltage reference, either of these would be close enough for most 5V supplies. You probably should not use a switchmode device as a precision reference anyway! Pin 6 is the switch (SW) terminal, which is connected to the two internal Mosfets. One switches the output to ground (pin 5), the other to VIN (pin 4). A non-synchronous part would require an external diode (typically a schottky diode) in place of the lower transistor, to allow inductor current to circulate while the upper Mosfet is off. Between the switch terminal and the output is an LC low-pass filter comprising a series inductor and a capacitor to ground. Like the input capacitor, we’re using a lower capacitor value than recommended in the view that more external capacitance will be fitted. However, it would be possible to fit a higher capacitance in the space available if necessary. The output of the LC filter is fed to August 2020  41 Setting the output voltage The MCP16311 data sheet recommends different inductors for different output voltages. The rule-of-thumb value is 4.5µH per volt at the output. In choosing an inductor, keep an eye on the DC resistance specification too. Values around 100mare recommended, meaning that the inductor will drop 0.1V, dissipating 100mW when the regulator is supplying 1A. If you are planning to run your regulator near 1A, this will probably be the biggest loss. Another critical point is the voltage rating of the output filter capacitor. You need a 6.3V or higher rating for a 5V output, but we’ve specified 50V for all capacitors to keep things simple. Advanced constructors may wish to use devices with a lower voltage rating but higher capacitance, as long as they still have a sufficient voltage rating for their particular role. Table 1 shows some choices for both the top resistor value (from both the 42  Silicon Chip TOP VIEW 1 F 18105201 1 BOTTOM VIEW L1 1 52.3k 10k 1 F 100nF REG1 1 1 F REG1 1 1 F CON1 100nF L1 1 R1 R1 52.3k 10k 1 F 18105201 CON1 the output pin of CON1, which forms the interface with the external circuitry; its other two pins are connected to the VIN pin of IC1 and the ground pour. This output voltage is also fed to the upper resistor in the feedback voltage divider mentioned earlier. The final component on the board is a 100nF capacitor between pins 6 (switch or SW) and pin 7 (BOOST). Because the internal high-side Mosfet is an N-channel device for maximum efficiency, it needs its gate to be brought above its source to conduct. As the source is connected to the SW pin, a voltage above SW (and possibly above VIN) is required to drive its gate. An internal charge pump provides this higher voltage, which is stored in this 100nF capacitor until it is needed to switch the Mosfet. The overall operation is as follows. IC1 produces a pulse-width modulated (PWM) square wave at the SW pin (pin 6) which is filtered by the LC circuit. The output voltage is monitored by the voltage divider connected to pin 1, which causes IC1 to change its PWM duty cycle to maintain the desired output voltage. With a light load at its output, IC1 can also ‘drop’ or skip PWM cycles, reducing power consumption. Three-pin header CON1 has 0.1in (2.54mm) spacing, to match a TO-220 package. 1 F 1 (WITHOUT LABELS) Fig.3: we’ve shown the component overlays same size (above) as we IN GND OUT OUT GND IN normally do but thought a veryBOTTOM VIEW (300%) TOP VIEW (300%) much-enlarged view (at right) would help you with assembly. Inductor L1 is fitted to the top side of the PCB, opposite to the other parts. It is easiest to solder IC1 first, as access to its pins is not as good once the surrounding parts are in place. The part that controls the output voltage is resistor R1 at upper left. Here it is a 52.3kresistor, to give a 5V output. Pin header CON1 can be fitted to either side, depending on the needs of your application. This can be fitted last, so you can test fit the board before soldering it. E24 and E96 series) and also a suggested inductor value. Note that the E24 resistor values do not allow for a high level of accuracy, but may still be close enough, depending on your application. Performance Naturally, we ran some tests to ensure that the Regulator has equivalent performance to its linear predecessor. As per the data sheet recommendations, we connected around 10µF extra capacitance at the input and 22µF at the output. Efficiency is very high compared to a linear device. We connected our prototype 5V Regulator to an 8 load (a wirewound power resistor), drawing a nominal 625mA. For low values of input voltage (up to around 12V), efficiency was 96%, dropping off above 12V. This agrees well with the information in the MCP16311 data sheet. Our calculations suggest that well over half of these losses are simply due to dissipation in the inductor’s DC resistance. Hence, the importance of low DC resistance in this part. During this test, we noted the Regulator was warming up above ambient, but was never too hot to touch. Another quick measurement indicated that the quiescent current of the Regulator (under no-load conditions) is around 80µA, close to the Australia’s electronics magazine value from the data sheet, and a lot less than a 78xx regulator at around 5mA (60 times higher!). Fig.2 shows how the output voltage varies with the input voltage, comparing the Regulator with the expected performance of a standard 7805. This also indicates the dropout voltage. Interestingly, the 7805 passes more voltage at very low input voltages. This is not unexpected, as the MCP16311 does not even come out of the under-voltage lockout until its input reaches around 4V. Once it starts up, it has a much lower dropout, needing an input of only 5.5V to supply 5V at the output; a dropout voltage of around 0.5V. On the other hand, the 7805 is not regulating correctly until its input reaches around 7V; a 2V dropout. In battery-powered applications, both the lower quiescent current and the low dropout voltages are big advantages. Not only does the higher efficiency mean that less energy is wasted, but the Regulator is also capable of operating with much lower input voltages, making better use of the same battery. One advantage of the MCP16311’s low-voltage shutdown feature is that in a battery situation, the 7805 would continue to pass current, completely flattening the battery (which could be fatal if it’s a rechargeable type), while the MCP16311 will switch off when siliconchip.com.au Again reproduced very much larger than in real life, these photos show front and back of the regulator – in this case set up to replace a 7805 (5V) regulator. Changing the regulation voltage is as simple as changing R1 and L1 to suit. its input gets too low, preventing this. Since the output is below 5V by the time the input reaches 4V, the connected circuit will probably not be operating to specification anyway. Scope1-Scope4 show more details of the circuit’s performance. Scope1 shows that the Regulator takes around 350µs to start up, which is quick enough for most applications. Scope2 shows output ripple. This is one area in which the 7805 will be superior, although this small amount of ripple is tolerable for most applications. Scope3 and Scope4 show the response to load and line changes; the output varies by around ±150mV for a 625mA load step, recovering in less than 100µs, while line regulation is around 1%, ie, an output variation of around 17mV for an input ripple of 1.88V. Construction Taking note of what is described above, choose your components before starting construction. Many of the components are quite small, and their marking will be barely legible. The capacitors will probably be unmarked, so take care not to mix them up (or lose them!). Check that you have the appropriate tools for working with small surfacemounted components. At a minimum, we recommend a fine-tipped soldering iron (preferably siliconchip.com.au with adjustable temperature), a pair of fine-tipped tweezers, a magnifier as well as some flux paste and solder braid (wick). Something to secure the very small PCB would be handy. If you don’t have a PCB clamping tool, then Blu-Tack may be sufficient. The Regulator is built on a double-sided PCB coded 18105201 which measures 15 x 10mm and is 0.6mm thick (a standard PCB is 1.6mm thick, which would make the device 1mm thicker). Refer to the PCB overlay diagram (Fig.3) during construction, to see which parts go where. IC1 has the finest pins, so start by fitting it. Check and confirm where the pin 1 dot is and align it with the markings on the PCB. If you have CON1 at the bottom then IC1’s pin 1 is at lower left. If you cannot find a dot, then the part may have a chamfer along one edge; this edge should be closest to CON1. IC1’s pins are on a 0.65mm pitch with only a 0.2mm spacing. You will probably bridge some pins while soldering it, so the solder braid is essential. Apply some flux to the pads and hold the IC in place with the tweezers. Solder one pad down (or one even one side if your iron tip is broad). Check and double-check that all the pins are entirely within their pads; if they are not, then they may short to adjacent pads even after any solder bridges are removed. Also check that the part is flat. Once you are sure of this, solder the pins on the other side. Don’t be concerned about bridges; in this case, they are almost inevitable. Just ensure that each pin is soldered to its correct pad in some fashion. With the IC soldered in place on both sides, you can clean up any bridges. Apply some more flux paste to the pins and press the braid against the pins with the soldering iron on one side. Gently draw the braid away from the part. It should draw up any excess solder, leaving a clean fillet. Inspect this with the magnifier and compare it to our photo above. Apart from IC1, none of the parts are polarised, so do not be concerned about the orientation after IC1 is installed. Follow with the 100nF capacitor which goes near IC1’s pins 1 and 8. Apply flux to the pads and solder Australia’s electronics magazine Parts list – ‘78XX’ 1 double-sided PCB coded 18105201, 15 x 10 x 0.6mm 1 3-pin right-angle header, 2.54mm pitch (or straight header, depending on application) (CON1) 1 22µH 6mm x 6mm 1.1A inductor* (eg, BOURNS SRN6028-220M) 1 MCP16311 switchmode IC, MSOP8 package (Digi-key, Mouser) Capacitors (all X7R SMD ceramics, size 2012/0805) 3 1µF 50V^ 1 100nF 50V (code 105) (code 104) Resistors (all 1% SMD size 2012/0805) 1 52.3kW (R1)* (code 5232) 1 10kW (code 1002) * parts for 5V output; see Table 1 for other voltages ^ increase to 2.2µF if an external lowESR input bypass capacitor of at least 1µF is not possible one lead only. Confirm that the part is flat and square within the pads before soldering the other lead. Go back and retouch the first lead with some fresh solder or a bit of extra flux. Use the same technique to fit the three 1µF capacitors. While they don’t all need to be 50V types, the price difference is small, so it’s easier to just use 50V types for all three as stated in the parts list. That also makes assembly easier since you don’t have to worry about which one goes where. The two remaining parts on this side are the resistors. Fortunately, these are usually marked so are more difficult to mix up. The 10kresistor will be marked as 103 or 1002. The other resistor value will vary depending on your selected output voltage. For the 52.3kresistor we’ve recommended for a 5V output, expect a code of 5232. The last component, inductor L1, is on the other side of the PCB. So now is a good time to clean up any flux residue on the top before flipping the PCB over. If you don’t have a dedicated flux solvent, isopropyl alcohol may work (assuming you can get some at a reasonable price… even metho is getting hard to find!). In any case, take care, as many of these cleaning substances are flammable. Allow the PCB to thoroughly dry before resuming soldering. L1 is a larger part and will generally have more thermal mass, so may August 2020  43 Scope1: this shows the response of the Regulator to having 8V applied with an 8  load (625mA). Its startup time is limited mostly by having to charge the output capacitance, which would be the case for most regulator circuits. require more heat. We’ve sized the pads for a nominal 6mm x 6mm part although up to 8mm x 8mm may fit. In this case, you may need to apply heat to the inductor leads. The technique is much the same as for other two-lead parts. Apply flux, solder one lead, check that the part is where you would like it and then solder the remaining lead. Then clean up the flux that’s been applied to this side of the PCB. You may need to install straight or right-angle headers for CON1, depending on how you wish to use the Regulator. We’ve fitted right-angle headers to our units to make them install just like a TO-220 device. This is also ideal for use on a breadboard. If using right-angle headers, check which side is the best fit (they can be soldered on either side), in case space is tight in your application. We fitted the headers at the rear (IC1 side) of the PCB by removing the pins from the plastic frame and threading them into the frame from the other side. This allows the pins to be held in position while soldering. This arrangement can also be used to mount the Regulator flat against the PCB by bending the leads a further 90°, just as you would for a discrete part, but a more rigid option would be to mount a straight header at the back. This may not work if you have components very close to where the Regu44  Silicon Chip Scope2: under the same test conditions as Scope1, we’ve zoomed into the output waveform after it has had time to stabilise to show the output ripple. We see around 50mV of ripple at the MCP16311’s 500kHz PWM frequency; more output filter capacitance would reduce this. This ripple is the main downside of using a switchmode regulator. lator will need to mount, but will be a lot more secure as the shorter leads will not be able to flex as much. Testing One of the worst things that could happen is that R1 is open circuit, which would mean that IC1 is not able to regulate the output as it cannot see any voltage at its output; effectively, the input voltage will appear at the output. If this is your first foray into surface mounted parts, you might want to double-check your soldering against our photos. You should ideally also test that the Regulator works correctly before deploying it to your circuit. 3-pin header CON1 will make it easy to plug into a breadboard or use jumper wires to rig up a test circuit. Note that the front of the Regulator is the side with the inductor and the CON1 header and pin 1 markings are on this side. Apply 4-30V to pin 1 of CON1 (with respect to GND at pin 2). Use a currentlimited supply if possible (eg, a bench supply) or series resistor to limit the current; this will minimise damage in the event of a fault with the circuit. You should be able to measure the desired output voltage at pin 3. You may also like to load the output (for example, with a resistor) to check that the circuit works under load. If it works as expected, you are ready to Australia’s electronics magazine solder it into your final circuit. Installation Because it is intended to replace a single component, the Regulator could be used in any number of designs, so we can only offer general advice. Any design using a 78xx or similar should have separate bypass and filter capacitors already included. We’ve put some modest capacitance on the Regulator PCB, but as mentioned earlier, not as much as recommended by the MCP16311 datasheet; mostly due to space considerations. The MCP16311 should ideally have at least 2µF at its input and 20µF at its output; we’ve provided around 1µF For some variants, we squeezed in slightly larger 3216/ 1206-sized capacitors across the input and output pins. It’s generally easier to get these larger valued or higher-rated parts in the larger part sizes, so it is worth considering if space is not critical. siliconchip.com.au Scope3: here we connected a 68  load to the Regulator and switched a second 8  load in and out using a Mosfet (gate voltage in blue, the yellow trace is the supply voltage). Thus the output current jumps from 75mA to 700mA and back. The green trace shows the output voltage, which in all cases stays within 200mV of the setpoint. More output capacitance will stabilise this further. for each. Thus an extra 1µF on the input and at least 22µF at the output is recommended. One option to add more capacitance directly to the Regulator PCB is to stack capacitors vertically. We’ve even seen part manufacturers do this to create discrete capacitors with more capacitance! You might also be able to get discrete capacitors with a higher value that will fit onto the board, depending on the actual input and output voltages you’ll be using. Check what parts are available in the 2012/0805 size (or 3216/1206 size, if you’re willing to jam them in). We recommend that you stick with types having an X5R, X6S or X7R dielectric. For example, 2.2µF 50V X5R capacitors are available in 2012/0805 size, if you can’t fit a 1µF external ceramic bypass capacitor on your host board. We’ve also built some variants with larger 1206 (3216 metric) sized input and output capacitors; you can see these in the photos. On the other hand, if your design can tolerate some ripple at the output, then you may be able to reduce the output capacitance below the recommended value. Just be careful to check that this doesn’t affect stability under the range of load conditions the regulator will experience. When fitting the Regulator to your siliconchip.com.au Scope4: the same 8  load as before but with the input supply being fed from an AC transformer and bridge rectifier with a 1000µ µF filter capacitor. Around 2V of ripple (at 100Hz) from the supply produces less than 20mV of ripple at the Regulator’s output, an attenuation of around 100 times. PCB, keep in mind that there are bare component leads on the back of the Regulator PCB which may short against (for example) the existing 78xx mounting hole. Some insulating tape (eg, polyimide) applied to the PCB should be sufficient to avoid problems here. Under low load conditions, thermal dissipation will be quite low, so you could probably even seal the entire part in heatshrink tubing, although we haven’t tested this. Alternatively, if you have space, extend the headers pins of CON1 so that there is clearance between the Regulator PCB and the PCB underneath. If your design is subject to vibration, some neutral-cure silicone sealant between the two will reduce mechanical fatigue. If you are using the right-angle mounting arrangement, then you will also lose the option to mechanically secure the Regulator PCB because it lacks a mounting hole. You should also ensure clearance between the Regulator PCB and any case parts that might short against the components on the Regulator PCB. Again, some tape and sealant may be required to maintain clearance and insulation. If you have space, the right angle connector CON1 can be mounted at the front (rather than the back) of the PCB. This will increase the clearance behind it. SC Australia’s electronics magazine EVERY ARTICLE IN EVERY BACK ISSUE OF Nov 1987 Dec 2019 CAN NOW BE YOURS FOREVER INDIGITAL FORMAT YOURS(PDF) FOREVER IN DIGITAL (PDF) FORMA A Storing 30+ years of SILICON CHIP magazines takes up a lot of space. Now you can save all that space and still have all the issues available. Or maybe you simply want the convenience of searchable files plus index - so you can find that feature or article you want without trawling through back issues! The choice is yours. The digital edition PDFs are supplied on a quality metal USB flash drive, at least 32GB. Each flash drive contains a fiveyear block (60 issues), covering: Nov 87 - Dec 94 Jan 00 - Dec 04 Jan 10 - Dec 14 Jan 95 - Dec 99 Jan 05 - Dec 09 Jan 15 - Dec 19 Each five-year block is priced at just $100, and yes, current subscribers receive the normal 10% discount. If you order the entire collection, the 6th block is FREE (ie, pay for five, the sixth is a bonus!). All PDFs are high resolution (some early editions excepted). Save the files to your PC hard disk, and the USB Flash Drive can be used over and over! For more information, or to place an order, visit www.siliconchip.com.au/ shop/digital_pdfs AUGUST 2020 45