Table 1 shows buck-boost operation in buck (step-down) and boost (step-up) modes.
When VIN > VOUT, the IC regulates in buck mode, with N1 and N2 switching synchronously. When VIN < VOUT, the IC regulates in boost mode, with N3 and N4 switching synchronously.
The critical aspect of this architecture is the smooth transition from one mode of operation to the other, when the regulator input and the output voltages are close in value.
In the transition from buck to buck-boost operation, as the input drops closer to the output, the converter’s duty cycle approaches unity. Namely, the ‘on’ time of N1 gets longer, while the ‘on’ time of N2 gets shorter. At some point, the N2 power transistor may not turn on fast enough within its allowed ‘on’ time, leading to discontinuities in the output voltage and current. Even if N1 and N2 were fast enough, the transistors N3 and N4 are in the ‘off’ state during buck mode and, when called to duty, have a turn-on delay due to the respective charge pumps (not shown), which require time to get back into operation.
The best strategy then is to anticipate the introduction of boost operation at a time when the buck operation is far from its operational limit. For example, when DC1 = 83.3% as shown in Figure 4. The boost operation (CLK2), 180° out-of-phase with buck operation (CLK1), does not interfere with the latter.
Accordingly, as the buck DC1 approaches 83.3%, the two boost transistors (N3 and N4) kick into action. Looking at the current waveform, we see that by equalizing volt x seconds, we have:
VIN x TON4 + (VIN-VOUT) x (TON1 - TON4) = VOUT x TOFF1
By reordering and simplifying, we obtain the characteristic equation of the four-switch buck-boost converter:
where DCx = TONx/T.
By eliminating the boost converter in Figure 2 and utilizing a buck-boost converter for each LED string, the voltage-regulator input voltage is now reduced to the battery voltage, minimizing each regulator’s switching losses.