Most rechargeable battery-powered systems include a switch that connects the load either to the battery or to a source of charging power. This switch allows a system with a depleted battery to operate immediately when plugged in, running on adapter power while the battery is charging.
The simplest and least expensive switch is a diode-OR connection. The load connects to each power source (battery and adapter) through separate Schottky diodes, so power is applied by the higher voltage of the two sources.
The drawback to this approach is the power dissipation (PD = Ibatt × Vdiode) and voltage drop (Vdiode = 350 mV at 0.5 A for the PMEG2010AEH) incurred when the battery services the load. Such losses may not be significant for high-voltage multicell batteries. But for one-cell Li+ batteries or two- to four-cell nickel-metal-hydride (NiMH) batteries, the power loss and diode drop across the blocking diode are considerable.
An alternative technique using a battery-charging IC with a Power OK (POK) output, like the MAX8814, switches loads with a voltage drop of only 45 mV at 0.5 A, an improvement of 305 mV (Fig. 1). This technique cuts power dissipation by 152.5 mW (175 mW – 22.5 mW) compared with the diode-OR connection. Performance improves at lower currents. At 100 mA, for instance, a diode drops about 270 mV but the alternative circuit drops only 10 mV.
 |
|
| Figure 1. | Using a Power OK (POK) output this circuit switches between battery voltage and Vdc In without the losses associated with a conventional diode-OR connection. |
This circuit switches loads without intervention from a microcontroller or system software. When operating on batteries and with Vdc In disconnected, U1’s POK output is high. This condition connects load and battery by turning on Q4 and Q3. Node 1 is biased to the battery voltage through R2, which maintains Q1 and Q2 off. When Vdc In connects to a dc source, Q1 and Q2 are held off because C1 raises node 1 to Vbatt + Vdc.
High voltage at the gates of Q1 and Q2 is generated instantly when Vdc In is applied. To prevent damage at the POK input, Q5 is configured as a source follower (voltage buffer). With its gate biased to the battery voltage, POK sees no voltage higher than the battery voltage. When POK goes low, current flows through Q5 and pulls down the gates of Q1 and Q2, turning them on. Vdc In services the load, and U1 charges the battery. C1 and R1 provide a short delay that allows Q3 time to turn off, preventing unregulated current from flowing to the battery.
 |
|
| Figure 2. | This scope screen shows the performance of the POK-based circuit as the load is switched from dc power to battery and back to dc power. Channel 1 is the voltage across the load. Channel 2 is the dc supply voltage. Channel 3 is the POK output, and Channel 4 is the battery current. |
When Vdc In is removed, POK becomes a high-impedance node, and battery current flows through the body diode of Q3. The load voltage is Vbatt – Vdiode. Because Q5’s gate is biased at the battery voltage, Q5 conducts until POK reaches a voltage level sufficient to service the load by turning on Q4 and Q3. Figure 2 shows the circuit’s response (with battery installed) when Vdc In is applied and removed.
 |
|
| Figure 3. | Designers can also use a battery-charger IC without a POK output. In this circuit, U3 provides a POK output by comparing Vdc In to the battery voltage. |
A modified circuit is suitable for battery-charging ICs such as the MAX1507, which does not provide a POK output (Fig. 3). The comparator (U3) supplies a POK output by comparing Vdc In to the battery voltage. This modified circuit’s response is very similar to that of the original circuit (Fig. 4).
 |
|
| Figure 4. | The circuit in Figure 3 delivers similar performance to that of the original POK-based circuit. |
