Buck-Boost Regulators: How They Work
A conceptual/reference doc on buck-boost DC-DC regulators — operating principle, the two dominant topologies (single-switch inverting vs. four-switch non-inverting), conduction modes, and how buck-boost compares to adjacent converter families (SEPIC, Ćuk). Written as a standalone power-electronics reference; not tied to a specific IC or datasheet.
1. What a Buck-Boost Regulator Is
A buck-boost regulator is a switching DC-DC converter that can produce an output voltage either above or below its input voltage from the same circuit, by varying duty cycle alone. This is what distinguishes it from a plain buck (step-down only) or boost (step-up only) converter: a battery that starts above the target rail and sags below it over its discharge curve — a common scenario in single/dual-cell Li-ion systems — needs a regulator that can step down early in the discharge and step up late in the discharge without switching topologies.
All buck-boost variants share the same core mechanism: an inductor is alternately connected to the input source (charging, storing energy in its magnetic field) and then to the output (discharging, delivering that stored energy to the load). The ratio of charge-time to discharge-time — the duty cycle, D — sets the output voltage. Where the topologies differ is how the inductor gets switched between those two states, and that difference determines whether the output is inverted, how many switches/diodes are needed, and how smoothly the converter transitions between step-down and step-up behavior.
2. Topology 1: Single-Switch Inverting Buck-Boost
The classical buck-boost is the simplest possible version: one inductor, one switch, one diode, and output capacitance — the same parts count as a basic buck or boost converter, just rearranged.
L
Vin ──┬────)))────┬──── to load (negative rail)
│ │
SW ▼│ (diode, cathode toward SW node)
│ │
GND Cout ── load ── GND
Switch ON (duration D·T): The switch connects the inductor's far end to ground. Current ramps up linearly through the inductor, drawing energy from the input. The diode is reverse-biased, so the output capacitor alone supplies the load during this interval.
Switch OFF (duration (1-D)·T): The switch opens. The inductor, resisting the change in current, forces its voltage to swing negative until the diode forward-biases, and the stored inductor energy discharges into the output capacitor and load. Because current is now being pulled from the switch node down through the diode into the capacitor's positive terminal being tied to ground-referenced output, the resulting output voltage is negative with respect to the input's ground.
Result: the output is inverted. This is the defining property of this topology (also called an "inverting buck-boost") — it is not a side effect of poor design, it's structural. Input and output share a common ground rail but opposite-polarity rails relative to it.
In continuous conduction mode (CCM — inductor current never reaches zero), the ideal steady-state transfer function is:
$$V_{out} = -\frac{D}{1-D} , V_{in}$$
At D = 0.5, |Vout| = Vin. Below D = 0.5 the magnitude steps down; above D = 0.5 it steps up — the same duty-cycle knob covers both regimes, which is the whole point of the topology.
The main practical limitation is exactly what falls out of the equation: the output is negative-polarity and floats relative to input ground, which is a poor fit for systems where downstream logic expects a positive rail referenced to the same ground as the source. This restricts the single-switch inverting topology to applications that specifically want a negative rail (op-amp bipolar supplies, bias generation) or ground-referenced-load designs are re-architected around the inversion.
3. Topology 2: Non-Inverting Four-Switch Buck-Boost
Most modern buck-boost regulator ICs — the ones used to keep a positive output rail steady across a battery discharge curve — use the four-switch (also called H-bridge or synchronous) topology instead, specifically to avoid the polarity inversion.
Q1 L Q3
Vin ──┤├──┬───)))───┬──┤├── Vout
│ │
Q2 Q4
│ │
GND GND
Q1/Q2 form a synchronous buck half-bridge on the input side of the inductor; Q3/Q4 form a synchronous boost half-bridge on the output side. Because both half-bridges share the same inductor, the converter can be operated as a buck, a boost, or a true buck-boost simply by choosing which pair switches and which pair is held static — this is what keeps the output positive and ground-referenced in every mode.
The controller selects among three operating modes based on the relationship between Vin and Vout:
| Mode | Condition | Q1 | Q2 | Q3 | Q4 | Behavior |
|---|---|---|---|---|---|---|
| Buck | Vin comfortably > Vout | Switches at D | Switches at (1-D) | Held ON | Held OFF | Acts as a plain synchronous buck; Q3/Q4 side is static, so switching losses are confined to Q1/Q2 |
| Boost | Vin comfortably < Vout | Held ON | Held OFF | Switches at D | Switches at (1-D) | Acts as a plain synchronous boost; Q1/Q2 side is static |
| Buck-boost (transition) | Vin ≈ Vout | Switches at D | Switches at (1-D) | Switches at D | Switches at (1-D) | All four switches active every cycle — this is the only mode that behaves like the classical single-switch buck-boost |
Approximate duty-cycle relationships per mode (CCM, ideal, lossless):
- Buck mode: D = Vout / Vin
- Boost mode: D = (Vout − Vin) / Vout
- Buck-boost mode: D = Vout / (Vin + Vout)
Running all four switches continuously (always in "buck-boost mode") is possible but wasteful — it roughly doubles switching losses versus letting the converter drop into pure buck or pure boost mode whenever Vin and Vout are far enough apart to allow it. Real controllers therefore keep the converter in buck or boost mode as much as possible and only activate all four switches in the narrow band where Vin ≈ Vout, since that's the only region where a single half-bridge can't do the whole job in one switching cycle. Locking one pair of switches fully on/off when it isn't needed is what gets four-switch buck-boost regulators close to plain-buck or plain-boost efficiency outside the transition band, rather than paying buck-boost-mode switching losses everywhere.
Mode-transition hysteresis: Because Vin and Vout are rarely perfectly stable (battery sag, load transients), a controller that switched modes exactly at Vin = Vout would "chatter" between modes near that crossover. Practical controllers add hysteresis bands around the buck↔buck-boost and buck-boost↔boost thresholds, and many precompute the duty cycle needed in the next mode before switching to it, so the output doesn't glitch during the transition.
4. Conduction Modes: CCM vs. DCM
Independent of which topology is used, a buck-boost stage can operate in one of two conduction regimes depending on load current and inductor value:
Continuous conduction mode (CCM): Inductor current stays above zero throughout the switching cycle — it ramps up during the ON interval and back down during the OFF interval, but never fully depletes before the next cycle starts. CCM gives lower peak/RMS current for a given output power, smoother output ripple, and a duty-cycle-only (load-independent) transfer function — the equations above assume CCM.
Discontinuous conduction mode (DCM): At light load, or with a small inductor, current ramps down to zero and stays there for part of the cycle before the next ON interval begins. DCM allows a physically smaller inductor and can improve light-load efficiency, but the transfer function becomes load-dependent, peak/RMS currents are higher for the same average power, and output ripple increases.
The boundary between the two regimes is set by the inductor's peak current relative to twice the average current it must supply. At the CCM/DCM boundary:
$$I_{L,peak} = \frac{V_{in}}{L} D T_s = \frac{V_{out}}{L}(1-D) T_s$$
Below this boundary current (lighter load or smaller L), the converter falls into DCM; above it, CCM. This is the same reason buck and boost converters spec a minimum load or minimum inductance for CCM operation — buck-boost inherits the identical constraint on each half of its switching cycle.
Inductor ripple current, the other core sizing equation, comes directly from the two slopes the inductor current follows each cycle:
- ON interval slope:
m1 = Vin / L - OFF interval slope:
m2 = -Vout / L
Choosing L to target roughly 20–40% peak-to-peak ripple relative to average inductor current is a common starting point in datasheet design procedures — smaller L gives faster transient response and a physically smaller part, larger L gives lower ripple, lower RMS loss, and better CCM margin at light load.
Output voltage ripple in buck-boost topologies is set by the fact that the output capacitor supplies the entire load current during the switch-ON interval (there's no direct input-to-output current path the way there is in a buck converter), which is why buck-boost and SEPIC output capacitors need higher ripple-current ratings than an equivalent buck design:
$$\Delta v_C = \frac{I_{out} , D , T_s}{C_{out}}$$
5. Comparison to Adjacent Converter Families
Buck-boost is not the only topology that can step a voltage up or down; it's simply the one that dominates general-purpose regulator ICs. SEPIC and Ćuk converters solve the same up/down problem with different tradeoffs:
| Topology | Order | Output polarity | Output ripple character | Practical note |
|---|---|---|---|---|
| Single-switch buck-boost | 2nd | Inverted | Pulsating (like buck) | Simplest part count; inversion is the dealbreaker for most designs |
| Four-switch buck-boost | 2nd (per active half-bridge) | Non-inverted | Pulsating | Dominant choice for battery-rail regulation; most complex driving/control, but fast and simple to regulate |
| SEPIC | 4th | Non-inverted | Pulsating | Avoids inversion with only one switch, but 4th-order dynamics make the control loop slower and harder to compensate; needs a coupling capacitor rated for high RMS current |
| Ćuk | 4th | Inverted | Non-pulsating (can be near-zero ripple on both ports) | Best ripple performance of the group, but shares SEPIC's slow 4th-order control dynamics and adds inversion back |
The practical reason four-switch buck-boost dominates general-purpose regulator ICs over SEPIC/Ćuk despite needing more switches: SEPIC and Ćuk are fourth-order systems (two inductors, two main energy-transfer capacitors), which makes their control loops intrinsically slower and harder to compensate for tight, fast voltage regulation. Buck-boost — in either topology — stays second-order per active stage, which is what lets it regulate quickly enough for mainstream point-of-load and battery-management applications. SEPIC/Ćuk remain attractive in niche cases that specifically want single-switch simplicity (SEPIC) or ultra-low ripple (Ćuk) and can tolerate slower loop response.
6. Component Stress and Design Considerations
A few things fall directly out of the topology and are worth carrying into any component-selection exercise:
- Switch/diode voltage stress: In both buck-boost topologies, the switching node swings to roughly
Vin + |Vout|(inverting) or must block the higher of Vin/Vout plus margin (four-switch) — higher than themax(Vin, Vout)stress seen in a plain buck or boost, so devices need correspondingly higher voltage ratings. - Inductor RMS current: Because the same inductor carries both the input-charging current and the output-discharging current every cycle (rather than a buck's more direct input-to-output path), buck-boost inductor RMS current is higher than a buck or boost inductor delivering the same output power — this shows up directly as higher I²R loss unless the inductor is sized up accordingly.
- Output capacitor RMS current: As noted in §4, the output cap supplies the full load current during the switch-ON interval, so its ripple-current rating needs headroom that a buck converter's output cap (which only ever sees the inductor ripple, not the full load current) does not need.
- Efficiency vs. mode: For the four-switch topology, efficiency is generally best deep in buck or boost mode (only one half-bridge switching) and dips somewhat in the buck-boost transition band (both half-bridges switching) — this is the direct consequence of the "roughly double switching loss" tradeoff described in §3, and is a normal, expected dip rather than a design flaw.
7. Typical Applications
- Battery-powered systems where the rail must stay regulated across a full discharge curve that starts above and ends below the target output (e.g., 2-cell to 4-cell Li-ion packs regulating to a 5V or 12V system rail).
- Automotive electronics, where a nominal 12V rail can sag well below regulation during cold-crank events and needs to stay in regulation through the sag.
- Solar/energy-harvesting front ends, where source voltage varies with panel illumination and load conditions and can cross the target output voltage in either direction.
- LED drivers needing constant current across an LED string voltage that may be above or below the supply rail depending on LED count and forward voltage.
- Negative-rail generation (bias supplies, op-amp split rails) — specifically the domain of the single-switch inverting topology from §2, where the inversion is a feature rather than a limitation.
Sources
- Monolithic Power — Buck-Boost Converters
- All About Circuits — All About the Inverting Buck-Boost Converter
- Texas Instruments — Multimode Control for a Four-Switch Buck-Boost Converter (SLYT765)
- Texas Instruments — Using a Buck Converter in an Inverting Buck-Boost Topology (SLYT286)
- Electrical Engineering Codidact — Why do DC/DC switching controllers seem to favour the buck-boost topology over similar ones like Ćuk, SEPIC and Zeta?
- University of Warwick, EE2E11 — Buck-Boost Converter (CCM/DCM equations, ripple derivations)
- ResearchGate — The Basic Topology of "Buck-Boost Four Switch Converter"