There are two fundamental methods to control the bridge MOSFETs in a non-inverting, buck-boost converter that uses a fixed-frequency PWM. The first method stores energy in the inductor by switching on both Q1 and Q4 simultaneously (Fig. 2). The stored energy is then diverted to the output capacitor by turning on both Q2 and Q3, after Q1 and Q4 have been turned off. This is analogous to the operation of traditional inverting buck-boost and flyback converters. The input-to-output DC transfer function is given by:

With this control method, the inductor must be able to store all of the energy transferred from the input to the output.

The second control method facilitates true buck conversion when the input voltage is higher than the output voltage, and true boost conversion when the input voltage is lower than the output voltage (Fig. 3).

This method is used most by newer non-inverting, buck-boost controllers. In the buck mode design, the duty cycle through Q1 and Q2 is modulated to control the output voltage; Q3 and Q4 are off. As the input voltage decreases and approaches the output voltage, the duty cycle of Q1 approaches 100 percent. Below this point, the converter operates in the boost mode, where Q1 remains on and Q2 remains off, while Q3 and Q4 are modulated. The input-to-output DC transfer function for this control method is given by:

At the boundary condition where the input voltage equals the output voltage, Q1 and Q3 are both on, while Q2 and Q4 are both off. Along with lowering the inductor's energy storage demands, this approach also reduces the ripple current on both the input and output capacitors.

Chips such as the UC3638, on the other hand, were designed for PWM motor drive and amplifier applications. However, the UC3638's programmable oscillator and the structure of its two PWM comparators are well-suited for controlling the full-bridge, buck-boost converter in either of the two previously described operation modes. The oscillator's functional block diagram and PWM comparators are shown in Fig. 4.

Its frequency, as well as average and peak-to-peak magnitudes can be programmed using external components. In operation, the oscillator's triangle-wave output is presented to two separate PWM comparators. One comparator offsets the error amplifier's output with a positive dead- band voltage, while the other uses a negative offset. This dead-band offset is externally adjustable also.

The oscillator's internal waveforms and PWM comparators are shown in Fig. 5. If we apply the second (and preferred) control method, the OUTB signal is used to control the low-side buck MOSFET (Q2 in Fig. 2). The OUTA signal is used to control the low-side boost MOSFET (Q4). The complement of the OUTB waveform drives the high-side buck MOSFET (Q1). Likewise, the complement of OUTA drives Q3. If we set the dead-band slightly higher than the oscillator's peak-to-peak magnitude, the transition from buck to boost mode is smooth, and the error amplifier's response will have no discontinuity. You should check the feedback loop's stability at the input range extremes; the power stage's small-signal response is different in the buck and boost modes of operation.

Similarly, you should evaluate each component at the input range extremes when you design the power stage for this topology. The peak-to-peak inductor ripple current actually is zero at the transition from buck to boost mode. The current increases as the input voltage goes further into either buck or boost mode. Figure 6 shows how the loss in each MOSFET varies with input voltage. The buck MOSFETs (Q1 and Q2) must have a voltage rating higher than the maximum input voltage. In the buck mode, Q4 is on (chief loss is conduction loss), and Q3 is off. In the boost mode, Q1 will be on (chief loss is conduction loss) and Q2 will be off. The boost MOSFETs (Q3 and Q4) just need to be rated higher than the output voltage. Note that Q1 and Q2 will experience both switching and conduction losses during buck mode operation; the same holds true for Q3 and Q4 during boost mode operation.

Although the oscillator and comparators of the PWM motor controller work well for this topology, the same cannot be said for internal MOSFET drivers. With a little creativity and a handful of low-cost external parts, the drive signals from the UC3638 can be used to drive low-side n-channel MOSFETs and high-side p-channel MOSFETs (see Fig. 7). This gate drive circuit is designed to introduce delay, which eliminates cross-conduction in both the buck and boost MOSFETs. Driving high-side n-channel MOSFETs presents more of a challenge.

The ripple current in the output capacitor is worst during the boost mode, where the capacitor must conduct the full load current during part of the switching cycle. This may force you to use a small secondary output filter. In the buck mode, the ripple current in the output capacitor is reduced to the triangular AC current in the inductor. Similarly, the ripple current of the input capacitor is worst during buck mode, and is significantly less during boost mode operation.

This design (for an automotive stereo application) generates a 15-volt, 1.75-amp output from a DC input range of 10 to 40 volts, and occupies less than two square inches of board space. Dual packages combining a p-channel FET and n-channel FET are used for both the buck and boost MOSFETs. The oscillator is programmed for 100 kHz operation. A zener diode clamps the output of the error amplifier. This limits the maximum duty cycle in boost mode, and prevents lock-up (100 percent duty cycle).

The supply's efficiency is plotted in Fig. 8. The circuit efficiency peaks at around 93 percent. It is highest when the input and output voltages are close to the same value. This circuit's performance exceeds that of a typical SEPIC or flyback converter. You can further improve the efficiency, particularly at higher input voltages, by adding a circuit that uses output versus input voltage for bias power.

For more detailed information about advanced PWM motor controllers, visit www.ti.com/sc/device/uc3868. For integrated four-switch, buck boost converters, visit www.ti.com/sc/device/tps63000.

References
[1] "No need to fear: SEPIC outperforms the flyback," by Betten, John & Kollman, Robert, Planet Analog, 01/25/06; http://www.planetanalog.com/showArticle.jhtml?articleID=177103753
[2] The so-called "buck-boost" is somewhat of a misnomer, in that a traditional buck-boost converter generates a negative output from a positive input.

About the Author
Brian King is an applications engineer and a Member, Group Technical Staff in the Power Management Group at Texas Instruments. He holds a BSEE and MSEE from the University of Arkansas.