The RCD Clamp: Protecting the Primary Switch Without Wasting Power
When the primary switch of a flyback turns off, the current in the primary inductance has nowhere to go except through the leakage path, and the drain voltage rises until something clamps it. Without a clamp, the drain voltage can exceed the rating of the MOSFET within a few switching cycles and the device fails. The clamp is therefore not an optional refinement; it is the circuit that makes the topology usable, and its design determines both reliability and efficiency.
The most common clamp is the RCD network: a diode, a capacitor and a resistor connected from the drain to the positive bus. When the switch turns off, the diode conducts the leakage current into the capacitor and the capacitor voltage rises. The resistor then discharges the capacitor between cycles, dissipating the captured energy as heat. The diode must be fast, because it sees the full drain voltage and must recover before the next on-time.
The clamp voltage is a tradeoff. A low clamp voltage keeps the drain stress low but forces the resistor to dissipate more energy, because the capacitor is held at a lower voltage and the leakage current flows for longer. A high clamp voltage reduces the dissipation but pushes the drain closer to the device rating. Designers usually set the clamp so that the worst-case drain voltage, at maximum input and maximum load, stays below about eighty to ninety percent of the switch rating.
Alternatives exist for higher efficiency. An active clamp uses an auxiliary switch and a capacitor to recover part of the leakage energy instead of dissipating it, which improves efficiency and reduces the drain spike at the cost of a more complex control scheme. A transient voltage suppressor or a Zener can clamp the spike with a sharper characteristic, but it must absorb the leakage energy repeatedly and is less forgiving of production spread.
An inadequate clamp shows up in specific ways. The MOSFET fails at high line voltage, where the input plus reflected voltage leaves the least headroom. It fails at startup, when the control loop is not yet regulating and the peak current is higher than in steady state. It fails at low temperature, where the switch is slower and the turn-off spike is larger. Each of these points to the clamp budget rather than to a random component failure.
The clamp is invisible on the datasheet, but it is part of what separates a driver that survives a high line and a surge from one that fails in the field. For a product like the PV100, the input range of 200 to 240 volts is narrow, which keeps the drain stress predictable and allows a clamp design with reasonable margin. A product rated for a wider input range must budget more headroom, and that budget shows up in the choice of switch and clamp rather than in the marketing text.
The clamp capacitor value matters. Too small and the clamp voltage rises sharply with every pulse; too large and the capacitor stores energy that the resistor must dissipate, lowering efficiency. The resistor is chosen from the leakage energy and the permitted voltage rise, and its power rating must include the worst case at maximum input voltage and full load.
For a production design, the clamp also interacts with transformer tolerances. Leakage inductance varies from unit to unit, so the clamp must handle the worst sample rather than the typical one. A design tuned to the average transformer will fail on the tail of the distribution, which is why margin in the drain rating is not optional.
Synchronous Rectification on a 24V Output: Gains, Timing and Risks
Inside the Flyback Transformer: Energy Storage, Leakage and Construction
Related Article