Fullscreen HTML5Med Res (??MB)HTML4

Using the Buck-Boost LED Driver By Tim Blythman as a Charger or Voltage Converter The High-Power Buck/Boost LED Driver design (June 2022; siliconchip.au/ Article/15340) is a versatile module for driving large LED panels, but it can do much more. This article examines some of its other uses and applications, including charging batteries and converting between different DC voltages. T he High-Power Buck-Boost LED Driver was designed to provide a current-limited output from a voltage that might be above or below the available input voltage. That makes it ideal for driving constant-­ c urrent devices such as bright white LEDs. But it isn’t just a one-trick pony; far from it. It’s a switchmode design that can operate in both boost (increasing voltage) and buck (decreasing voltage) modes with a smooth transition blending between the two. Dedicated circuitry reduces the output voltage when the load current rises above a set threshold. The target design specification was for it to deliver at least 6.5A at the nominal 12V of the LED panels that we had procured. But the LM5118 chip that controls the Driver can operate over a much wider voltage range, as can the other main components, such as the Mosfets that perform the switching. The PCB and other components limit it to handling an input current of 10A and about 8A at the output. Since it can regulate both voltage and current over a wide range, the Driver can be used for many other purposes rather than just driving LEDs. In the same vein that a laboratory PSU is often pressed into service as a battery charger, you can also use the Driver as such. Adding a beefy mains-powered DC Applications for the Buck/Boost LED Driver ∎ Driving high-brightness LEDs/LED arrays ∎ Charging/maintaining a ‘house battery’ in a caravan or boat ∎ Making a portable charger with an internal SLA or Li-ion battery ∎ Powering 12V accessories from a 24V battery or a laptop charger ∎ Powering 24V accessories from a 12V battery ∎ Powering/charging a laptop from a 12V battery (eg, in a car) ∎ Providing a regulated 12V DC supply from a 12V battery ∎ Recharging a backup power battery from a car during a blackout ∎ As a high-current USB power source (eg, to run multiple devices at once) from a 12V battery ∎ Providing a high-current, low-voltage rail within a device that has a higher voltage rail ∎ Powering 12/24V DC equipment directly from a solar panel 54 Silicon Chip Australia's electronics magazine supply to the Driver’s input will allow that, with a few provisos, which we’ll discuss shortly. The beauty of the Driver is its wide input voltage range, meaning that many types of supply can be used. Common laptop power supplies produce around 19V and would be ideal for feeding the Driver, especially as this means a lower current demand on the supply for a ~12V output. This article will also look at options such as solar panels and other battery voltages. We will present some charts based on measurements we made to guide you in setting up the Driver for these sorts of applications. In particular, we’ll look at typical settings and what they mean across the Driver’s operating range, including efficiency. We’ve reproduced the entire Driver circuit in Fig.1 to assist you in following our reasoning and explanation. We’ll also mention a few of the subtler points that may need to be addressed along the way, such as extra parts that may need to be added. Fig.2 shows the most basic way of connecting a battery to the board for charging, with a supply at CON1 and a battery directly connected to CON2. However, we strongly suggest some of the improved alternatives discussed later. We’ve also included digital oscilloscope grabs Scope 1 to 3 to demonstrate siliconchip.com.au Fig.1: the circuit of the Buck/Boost LED Driver, reproduced to aid in how to use it. It’s based around an LM5118 buck/ boost controller chip and uses a bridge of Mosfets, schottky diodes and inductors to perform voltage conversion. the operation of the Driver in its three main modes (buck, buck/boost and boost). Soft current limiting One important point to consider when using the Driver is that it does not have a ‘brick wall’ current-limiting response. Our very early prototypes considered this option, but suffered from instability and oscillation when the current limiting was active. The final design has a softer response, leading to the sort of curves seen in Fig.3. We plotted that with the Driver’s current limit trimpot (VR2) set to three arbitrary positions across its scale, including its minimum. As mentioned in the original article, 1.8A is the minimum current limit threshold. The output voltage has been set to 14V, in the typical charging range for a 12V lead-acid battery. This setup is a good starting point for charging such batteries. This graph was produced by connecting our Arduino Load (also from the June issue; siliconchip.au/ Article/15341) to the Driver and stepping through its 16 load levels. siliconchip.com.au As described in the panel at the end of this article, we achieved higher load currents by connecting a second Load to the first. The result is a variation across what would be the operating range of the battery, with more current flowing initially upon charging a flat battery. We ran the tests used to plot Fig.3 with both 12V and 15V at the input but the results were indistinguishable. It is reassuring that the behaviour will be consistent when powered from the typical range of a 12V battery. This means that you can use a range of different power sources for charging, including another 12V battery! As shown in Fig.3, the current delivery increases as the voltage drops further below 10V. But any healthy 12V battery should have a terminal voltage of at least 10.5V at rest, and probably higher; if your battery is measuring 10V or less, you will want to do something about that before you go about charging it. For charging batteries, we suggest that the output fuse of the Driver (F2) be sized not much larger than the set current limit, to prevent damage to Fig.2: these are the most basic connections for using the Buck/Boost LED Driver to charge a battery. But they are only really suitable for when you are actively monitoring the battery. A few additions need to be made to turn it into a proper battery charger. Australia's electronics magazine October 2022  55 both the battery and Driver in case of a battery fault. Battery charge leakage Scope 1: this scope grab shows buck-only operation, delivering an 8V DC output from a 17V DC input. The blue trace is the output voltage, red the gate of Q1, green the gate of Q2 and yellow/brown Q2’s drain. In this case, only Q1 is being driven as no boost action is required. Note how Q1’s gate ‘floats’ during the off-time, but it never gets high enough (>17V) for Q1 to conduct. Q2’s drain also floats after the inductors’ magnetic fields have fully discharged. Scope 2: this is similar to Scope 1 but with a 13V DC output, close enough to the 17V input that it is now in buck/boost mode. Both gates (Q1, red and Q2, green) are now switching on, with Q2 switching on for a fraction of the time that Q1 is on. The inductor magnetic fields don’t discharge as quickly as in Scope 1, but it is still operating in ‘discontinuous mode’ as the load is relatively light. Scope 3: with the output voltage set to 20V, the unit is now operating in pure boost mode, where both Q1 and Q2 are switched on simultaneously and for the same period. As soon as they switch off, energy stored in the inductors pegs Q2’s drain voltage one schottky diode drop above the output voltage as the inductors feed energy into the output. The output filter capacitors sustain the load current between these pulses. Fig.3: these three curves demonstrate the ‘soft’ current limiting characteristics of the Buck/Boost LED Driver. They show its behaviour at three different current limit settings. The voltage drops off quickly once the current limit is exceeded, but it’s hardly a ‘brick wall’. 56 Silicon Chip Australia's electronics magazine Depending on how you configure the Driver, it may be that the charged battery (at CON2) remains connected while no power is available at the Driver input (CON1). Our tests show that such a state will not damage the Driver or the battery. Most of the circuit is isolated from the downstream battery by diodes D1 and D2. However, in this condition, there is a constant load on the battery at CON2 of around 5.6mA due to the voltage sense divider formed from the 1kW resistor, 5kW trimpot and 220W resistor. 5.6mA is consistent with 1.23V being present across the 220W resistor, which is expected when the Driver is operating normally. IC2, the current shunt monitor, has high impedance inputs when its supply is absent, so it does not present any further load. The only other possible load is via the 1kW resistor back into IC1’s FB pin (pin 8), and we did not detect any current from this in our tests. While the 5.6mA load would take a long time to discharge a large battery, it is not ideal. We have two suggested approaches to eliminate it. The simplest is to fit a suitably rated schottky diode between CON2 and the charged battery; this will naturally drop some voltage between the Driver output and battery, but you can compensate by increasing the output voltage a little. This arrangement is shown in Fig.4. Even at the minimum current setting, such a diode will typically dissipate 1W or more. So you will need to use a chunky diode. You might be tempted to use several in parallel, but it’s hard to guarantee current sharing with such an arrangement. A TO-220 schottky diode with a small heatsink would be a better solution (eg, Altronics Z0065 or Jaycar ZR1029). A better solution, if slightly more complicated, is to add a 10A automotive relay to only connect the charged battery if a suitable supply voltage is present. This is shown in Fig.5. The relay coil is connected in parallel with the Driver’s supply at CON1. Be sure to check the polarity in case the relay is the type that has an integral diode. The normally-open contact is connected between CON2’s ‘+’ terminal and the charged battery’s positive. siliconchip.com.au For the terminal numbers shown on typical automotive relays, the 85 pin should connect to the Driver’s ground and the 86 pin to the CON1’s + terminal. The common 30 pin should connect to CON2’s + terminal, with the 87 pin going to the charged battery positive. The disadvantage of the relay approach is that the power consumed by its coil will reduce overall efficiency, but possibly not as much as the schottky diode approach, depending on the coil power. Automotive relays typically have a coil power on the order of 2W, so the relay is a more attractive option at higher current levels. The diode approach is probably more efficient for lower currents, but remember that its forward voltage will make setting the correct charge voltage harder. Fig.4: a high-current schottky diode should be added to prevent the battery from being drained by the parasitic load of the Driver when the input supply is cut off. This is not necessary if you will always disconnect the battery after charging, though. Charging stages The bare Driver module is essentially stateless; what it does is based only on the prevailing conditions. Because it has voltage and current limits, it can provide float or bulk/absorption charging, but it will charge continuously as long as it has power. So unless you only want float charging, some thought is required to ensure it will not damage the battery. You can bulk charge a battery using the Driver by setting its output to the appropriate bulk charge voltage (eg, around 14-14.4V for a 12V lead-acid battery). But you need to limit the charging time somehow, as batteries can be damaged by charging at this voltage for extended periods. Check your battery’s manufacturer data for its limits. Because we think the Driver will be handy for charging batteries, we have developed a low-cost add-on board described starting on page 60 of this issue. This board’s primary job is to reduce the Driver’s output voltage after the bulk and absorption charging phases have finished so that it switches to float charging for the remainder of the time it is powered. It does this by monitoring the output current and voltages. It determines that the bulk charging stage has ended once the voltage has stopped rising and the charge current starts to drop off. The absorption phase ends (and float charging begins) once the charge current has reduced to about 10% of the siliconchip.com.au Fig.5: a relay can disconnect the battery when the input power is off instead of a schottky diode. This is more efficient at higher charging currents, although it is more costly and involves extra wiring. It also limits the input voltage range. current at the end of the bulk stage. It also includes a timer to terminate absorption if it takes too long. As it can draw a little power from the battery, this timer is not reset if the input power is briefly lost (eg, if the vehicle engine/alternator is switched off, then restarted). This add-on board only has a couple dozen components, fits right on top of the Driver and provides a convenient charge display, plus some extra adjustments. We strongly recommend using it if you want to use the Driver for unattended fast battery charging. See that article for more details on how it works, how to build it and the adjustments and indications it provides. Charging setup If you’re using the add-on board mentioned above, see that article for instructions on setting it up as a charger. Otherwise, the rest of this section applies. To set up the Driver for battery charging, set the voltage to the required charge voltage of the battery; around 13-13.8V is typical for float charging a standard lead-acid type battery, or 14-14.6V for bulk charging. The current limit you choose may depend on your battery (especially for a smaller type), power source and Australia's electronics magazine wiring. In any case, remember that the actual current delivered may vary slightly, especially if the battery is flat and the Driver is providing a much lower than nominal voltage. Allow 10% to 20% extra current when charging a flat battery. One way to handle this is to set the current while the battery is close to flat. Also remember to change fuse F2 to have a trip current just above this setpoint. The next nominal value just above the maximum charging rate (when flat) is a good starting point. This will help avoid runaway conditions if the battery is excessively discharged. Remember to add a diode or relay as described earlier if you don’t want the battery to self-discharge back through the Driver. Efficiency The Driver itself is a source of some inefficiency. The data sheet for the LM5118 (IC1 in Fig.1) has a graph that shows efficiencies between 80% and 95%, varying with input current and voltage. With a 12V supply and 14V output setpoint, we measured a no-load supply current of about 35mA. At 1.8A, this amounts to about 2% of the supply power being dissipated, limiting maximum efficiency to 98%. October 2022  57 Aside from this quiescent current, the main offenders regarding losses are the diodes and inductors; in practice, these are the components that heat up the most during operation. We ran some simple load tests to determine the overall efficiency for some likely configurations. The first test used a 12V input and 12V output, followed by a 24V input and 12V output and a 12V input feeding a 24V output. The results are shown in Fig.6. These cover the most common operating regimes of the Driver: with the input and output voltages similar (hybrid mode), with the input much higher than output (buck mode) and the input lower than output (boost mode). What isn’t obvious from the graphs is that the quiescent current is lower for higher input voltages and higher for higher output voltages. The highest we saw was 47mA at 12V for a 24V output (564mW), compared to 34mA at 12V for 12V output (408mW) and 12mA with a 24V input for a 12V output (288mW). As is typical, the Driver is more efficient when reducing the voltage. Unsurprisingly, the hybrid mode that occurs when the input and output voltages are similar has an efficiency between that of the buck and boost modes. Our measurements show that the efficiency ranges quoted in the data sheet are correct, at least for meaningful current outputs. The buck mode doesn’t suffer from the drop in efficiency at higher currents of the other modes, so having a higher input voltage is beneficial. Solar power You might think that the Driver’s wide input range would be well We used a laptop power supply like this Jaycar MP3346 for our tests. The Driver adds a fully adjustable voltage output with current limiting. The Driver can also run from power sources like batteries and car accessory sockets, to name a few. suited to taking power from a solar panel. For example, a nominally 12V solar panel can vary up to 22V under no-load conditions and will typically have its maximum power point (MPP) at around 17V. It might even deliver less than 12V under low-light or heavy load conditions. We did a few brief tests to test this theory using a 40W solar panel charging a 12V battery with a 1.8A current limit. The basic outcome is that it will work, but it is probably not the best way to do it. It certainly won’t work as well as a good MPPT solar charge regulator. All solar panels vary their output voltage depending on load, and the first thing we found was that the Driver would rapidly oscillate as it would switch on and draw current, causing the solar panel voltage to drop. This triggered the UVLO (under-­ voltage lockout), decreasing the load and causing the solar panel voltage to rise, repeating the cycle. Overcoming this was straightforward; we simply connected a 1000μF electrolytic capacitor across the input at the Driver’s CON1. If doing this, Fig.6: efficiency plots for three different common voltage conversion scenarios. The Driver is most efficient when the output voltage is below the input voltage and least efficient when the output voltage is higher. However, it’s above 80% efficient in virtually all cases. ensure that such a capacitor is rated to handle the open-circuit solar panel voltage, which might be near double nominal voltage. We also tried a 4700μF capacitor. It worked well too, and larger values should also. But this is not the main limitation. Since the Driver primarily strives to deliver the target voltage, it does not fare well under lower light conditions. Any time the outgoing power demand exceeds the available incoming power (minus losses), the input voltage sags, the UVLO activates and no power is delivered to the battery. This is in contrast to a purposely-­ designed solar charge regulator, which modulates its output to provide at least some current based on the power available. In practice, using the Driver this way worked well in full sunlight, but as soon as some cloud cover appeared, the output current dropped to nothing, with brief bursts of activity as the capacitor charged up. Low-light conditions (such as first thing in the morning) will typically be when the demand for charging current is the highest, so there is a definite mismatch in needs against capabilities. On the other hand, if you want to use the Driver to directly power equipment from a solar panel, this behaviour is probably preferred. The device will operate at its rated voltage and current, or not at all. Charging a battery from a solar panel Fig.7: the pinout for a Type-A USB socket. 58 Silicon Chip Australia's electronics magazine siliconchip.com.au via the Driver will definitely need a diode between CON2 and the battery (as described earlier), as a solar panel will spend most of its time (overnight, at least) not providing any charge at all. A relay will not work in this situation as there will be long periods when the solar voltage will be high enough to trigger (or at least hold in) the relay while not having enough power to allow charging. So, in brief, the Driver can work as a solar charge regulator in a pinch, but it won’t be very good at it. That is not surprising, as it wasn’t designed with that in mind. As a USB 5V power source While not envisaged in the original design, the Driver could deliver a regulated 5V for powering USB devices with a minor change. The default divider chain gives a nominal output voltage range of 7V to 34V. To achieve lower output voltages, the 1kW resistor at the top of the divider chain (in green at lower right in the schematic, Fig.1, and in Figs.2, 4 & 5) can be replaced with a 0W jumper. We have not tested this configuration, but expect it will be a stable modification as it does not unduly change the impedance seen by the FB pin. Also note that this will reduce the maximum output voltage to around 29V. You would then need to wire up the Driver’s output to one or more USB sockets (probably several if you intend to pull multiple amps). The pinout of a Type-A socket is shown in Fig.7; the D+ and D- pins can be left disconnected. Test it with something you don’t care about first (such as an old USB drive), as reversed polarity could easily damage a device. Final notes In the original Driver article, we mentioned that it makes sense to change the UVLO divider if you are using a 24V battery to the values mentioned. This is to shut off the Driver if the battery gets too flat. If you want another threshold, keep the lower resistor around 10kW and modify the upper resistor to put 1.23V at the divider at the threshold voltage. Also remember that JP1 is available to control the Driver too. So far, we haven’t had any of our prototypes fail, so we’re happy that it’s a robust design. But the oscillating behaviour we have seen when the siliconchip.com.au Modifying the Arduino Programmable Load to monitor external loads The Arduino Programmable Load project from June was invaluable in developing and testing the Driver. We also used it extensively to collect the data presented in this article. But you might note that we were testing with currents and voltages much higher than a single Arduino Load can handle. The higher-voltage tests (up to 24V) were made possible by connecting a 70W LED panel in series between the Driver’s output and the Load’s input, to drop around 12V at up to 6A safely. We found that this worked well, with both the LED panels and Arduino Load operating within their respective limits. But handling higher currents was a bit trickier. We made a very simple modification to the Arduino Load that allowed us to connect further loads downstream of the 47W resistors built into the Arduino Load. This change allows the current sunk by the external load to be measured and reported by the Arduino Load. Some of our tests used the LED panels, but we also used a second Load downstream of the first. This allowed us to test the Driver at much higher currents than the Arduino Load could otherwise handle. Of course, we made sure the wiring used could handle the necessary currents. A downstream load can simply connect between the VPS and GND rails, meaning that current from a power source connected to CON1 flows through the 15mW shunt and through the secondary load via the VPS rail to GND. Since it passes through the shunt, any current it sinks is also measured by the Load. To do this, we simply soldered a set of screw terminals to the PCB using component lead off-cuts. Refer to our photos and diagram to see the change. Note the terminal polarity; the negative terminals are the two that are closest together. Keep in mind that the Arduino Load still has a 6.67A measuring limit, and the screw terminals themselves should not carry more than 10A. This modification also means that the Arduino Load can be used as a load monitor if none of the 47W loads are active. The output of the serial terminal will sim- By adding another two-way terminal to the Arduino Load, as ply be the prevailing current due to any shown here, you can connect two downstream loads and the voltage level in parallel to handle double the as measured at CON1. current. It’s also possible to connect We have also revised the Arduino Load a high-power LED array in series PCB with provision for this extra terminal, with the load to increase its voltagehandling capability. available in our Online Shop. supply voltage is near the UVLO voltage might not be good for connected devices. So if your setup does have the possibility of operating near the UVLO voltage, make sure that the supply wiring has low resistance and check that connected devices will be unaffected by UVLO dropouts. Conclusion The Driver’s wide input range allows it to be a versatile battery charger, especially if you build the Charge Controller add-on board described on the next page of this issue. It is not the best choice as a solar charge controller, but it might come in handy if a regulator is needed to Australia's electronics magazine power some equipment directly from a solar panel. It’s particularly suited to working and converting between different voltages and is most efficient when stepping the voltage down. However, it can seamlessly work with widely varying input voltages. As the Driver is more efficient when the input voltage is higher than the output, common laptop power supplies that deliver 19V are a good choice for powering a 12V system via the Driver. If you want to power the Driver from a vehicle supply, see the DC Filter article in the November 2022 issue, which will protect the Driver from the damaging voltage spikes that are common in automotive supplies. SC October 2022  59