How Does a Flyback Converter Work?

Detailed Explanation

The flyback converter is the standard isolated topology in power electronics — used everywhere from phone chargers and offline mains-powered supplies to any DC-DC application where the input and output must have no direct electrical connection. Unlike the buck and boost converters already covered in this cluster, which store energy in a simple inductor, a flyback converter stores energy in a transformer — and that one change is what makes isolation possible. See the SMPS topic for the full set of switching-converter guides.

Energy Storage, Not Direct Transfer

A flyback transformer is not really a transformer in the classical sense (which transfers energy continuously through mutual coupling while both windings conduct simultaneously). It behaves more like two coupled inductors that never conduct at the same time:

Switch on: Current flows into the primary winding, and the transformer's core stores energy as a magnetic field (via its magnetising inductance) — exactly like a boost converter's inductor charging. Critically, the secondary-side rectifier diode is reverse-biased during this phase (by the dot-convention polarity of the windings), so no current flows in the secondary at all while the primary is being charged.

Switch off: The primary current is interrupted. The stored magnetic field, unable to collapse instantaneously, forces current to flow — but now on the secondary side, since the abrupt polarity reversal forward-biases the secondary rectifier diode. The stored energy is delivered to the output capacitor and load.

This "store, then release to the other side" behaviour — energy transfer happens in two distinct phases, never simultaneously — is the defining characteristic of a flyback converter, and is why it's classified as an energy-storage (not direct-transfer) isolated topology.

The Turns Ratio: An Extra Design Variable

Because energy crosses to a separate winding, the flyback topology has a design freedom the non-isolated topologies don't: the transformer's turns ratio. In continuous conduction mode (CCM), the voltage relationship is:

Vout = Vin × (N2 / N1) × D / (1 − D)

where N2/N1 is the secondary-to-primary turns ratio and D is the duty cycle. Compare this to a simple boost converter's Vout = Vin / (1 − D) — the turns ratio term lets a flyback design reach a required output voltage with a duty cycle that stays in a practical, well-controlled range (typically 0.3–0.5), rather than being forced toward the extreme duty cycles that pure inductor-based topologies need for large voltage ratios.

Continuous vs Discontinuous Conduction Mode

Discontinuous conduction mode (DCM): The transformer's magnetising current falls completely to zero before the next switching cycle begins. Simpler to control (no right-half-plane zero in the small-signal model) and common in lower-power designs (a few watts to tens of watts), but has higher peak currents for a given output power.

Continuous conduction mode (CCM): Magnetising current never reaches zero — some energy remains in the transformer at the start of each cycle. Lower peak currents for the same output power (better for higher-power designs), but the control loop is more complex to stabilise due to the right-half-plane zero characteristic of CCM flyback converters.

Most flyback designs below approximately 30–50 W operate in DCM for control simplicity; higher-power designs typically move to CCM or a boundary-mode (transition-mode) control scheme to manage peak currents.

The Leakage Inductance Problem and Snubber Design

No physical transformer achieves perfect coupling — some fraction of the primary's magnetic flux (leakage inductance) never reaches the secondary winding at all. This leakage energy has nowhere productive to go: it can't transfer to the secondary (it isn't coupled), and when the primary switch turns off, this leakage inductance's collapsing field generates a sharp voltage spike across the switch, on top of the reflected output voltage.

Left unaddressed, this spike can exceed the primary MOSFET's drain-source breakdown voltage rating, driving repeated avalanche breakdown events that degrade or destroy the switch. The standard mitigation is an RCD snubber — a diode that catches the leakage spike, a capacitor that absorbs its energy, and a resistor in parallel with the capacitor that bleeds that energy off as heat between cycles. Snubber component selection is a direct trade-off between clamping voltage (lower is safer for the MOSFET) and power dissipated in the snubber resistor (lower clamping voltage generally means more snubber loss) — see the manufacturer application notes in the References section for the standard design equations.

Feedback Isolation

Because the whole point of a flyback design is galvanic isolation, the feedback signal that regulates the output cannot cross back to the primary-side controller through a direct wire without defeating that isolation. The standard solution pairs a TL431 shunt regulator on the secondary side (comparing the actual output voltage against an internal reference) with an optocoupler: the TL431 circuit drives an LED proportional to the output error, and a phototransistor on the primary side — with no electrical connection to the LED, only an optical one across the isolation barrier — feeds this signal to the primary-side PWM controller. This TL431 + optocoupler combination is the near-universal standard for isolated feedback in flyback and other isolated topologies; a digital isolator (see optocoupler vs digital isolator) is a less common alternative for this specific analog control-loop application, more often seen in isolated digital communication than in analog voltage-error feedback.

Multi-Output Designs and Cross-Regulation

A flyback transformer can have multiple secondary windings, each producing a different output voltage from the same primary switching cycle — a common way to generate, for example, a 5 V and a 12 V rail from one converter. The limitation is cross-regulation: only one output (typically the highest-current or most critical rail) is directly sensed by the feedback loop; the other outputs follow it based on their turns ratio, but their actual regulation accuracy is looser and depends on how their individual load currents vary relative to the regulated output. Multi-output flyback designs are appropriate when the secondary (unregulated) outputs can tolerate several percent of additional voltage variation; where a secondary output needs tight regulation independent of the primary rail's loading, an additional local linear or switching post-regulator on that output is the standard fix.

Design Considerations

• Choose the turns ratio around your worst-case input range, not just nominal. The turns ratio must be selected so the required duty cycle stays in a well-controlled range (typically 0.3–0.5) across the full specified input voltage range, not just at a single nominal input voltage — an offline flyback design in particular must handle a wide universal-input range (typically 85–265 VAC rectified).
• Snubber design is a trade-off, not a one-time calculation. Component values that minimise MOSFET voltage stress increase snubber power dissipation, and vice versa. Validate the actual switch voltage waveform on the bench across line and load extremes — leakage inductance varies with transformer construction and is difficult to predict precisely from a datasheet alone.
• Budget for a bias/auxiliary winding if the design needs housekeeping power. Many flyback controllers need their own low-voltage supply to start up and run — a small auxiliary winding on the same transformer core, rectified locally, is the standard way to provide this without a separate supply.
• Isolation creep age and clearance are a compliance requirement, not just good practice. For any design where the isolation barrier is safety-relevant (mains-referenced primary, medical or safety-rated equipment), the PCB layout must maintain the creepage and clearance distances specified by the relevant safety standard between primary- and secondary-side copper — see the Compliance topic for the regulatory framework this sits within.
• Zeus Design designs isolated power supplies, including flyback converters for offline and safety-isolated DC-DC applications, from transformer specification through PCB layout and compliance.

Common Mistakes

• Undersizing the snubber and relying on the MOSFET's avalanche rating as a substitute. Some MOSFETs are rated for repetitive avalanche events, but relying on this as the primary spike-management mechanism (instead of a properly designed snubber) trades a known, controllable loss mechanism for cumulative device stress that shortens the switch's service life — size the snubber to do the actual clamping job.
• Selecting a turns ratio based only on nominal input voltage. A flyback design that works at nominal input but pushes duty cycle to an extreme (very low or very close to the maximum the controller supports) at the input range's edges will show degraded regulation, higher peak currents, or outright instability at those extremes. Verify duty cycle across the full specified input range during design, not just at nominal.
• Forgetting the bias/auxiliary winding start-up sequencing. A controller that depends on an auxiliary winding for its own supply needs an alternative start-up path (a start-up resistor from the high-voltage rail, or a dedicated start-up IC) before the converter is switching and that winding has any output — omitting this leaves the design unable to start from a cold power-up.
• Treating multi-output cross-regulation as tighter than it actually is. Assuming an unregulated secondary output on a multi-output flyback design will track the regulated output as precisely as a dedicated regulator would, without accounting for load-dependent cross-regulation error, leads to out-of-spec voltage on the secondary rails under real, uneven load conditions.