Snubber Circuits and Their PCB Layout

Every switching circuit contains inductance that the designer did not intend. The loop formed by the transistor, the diode, and the capacitor has a parasitic inductance measured in nanohenries, and when the current is interrupted the energy stored in that inductance has to go somewhere. It leaves as ringing, as a voltage spike, or as radiated emission. A snubber gives that energy a controlled path to dissipate.

What a Snubber Does

A snubber is a network that absorbs energy at the moment of switching and releases it as heat. It does three things at once: it reduces the peak voltage across the switch, it damps the ringing that follows the transition, and it slows the rate of change of voltage, which reduces the high frequency content that radiates. It does not eliminate switching loss, and in most designs it adds a small amount of loss in exchange for a large reduction in stress.

Because it acts on the parasitic elements, the snubber is closely coupled to the layout. A snubber placed across the wrong nodes will do nothing useful, and one placed a centimetre away from the device is separated from the ringing it is supposed to damp by the very inductance that caused the problem. Placement is therefore part of the topology, not an afterthought.

RC Snubbers Across a Switch

The simplest form is a resistor and a capacitor in series, connected across the switching device. The capacitor absorbs the high frequency energy and the resistor dissipates it, while the series combination prevents the capacitor from discharging into the switch at turn on. The network is often designed by measuring the ringing frequency with the snubber removed and then choosing a capacitor that halves it, with the resistor set to provide critical damping.

That empirical method works well because it uses the circuit itself as the model. It assumes that the ringing is a simple second order response, which is usually true enough for a first pass. The values then need to be verified at full load and at the extremes of the input voltage range, because the effective parasitic capacitance of the switch changes with voltage.

RCD Clamps and Energy Recovery

An RCD clamp uses a diode, a capacitor, and a resistor to catch the energy that would otherwise spike the drain of a flyback or a forward converter. The diode conducts during the spike, the capacitor stores the charge, and the resistor discharges it. The clamp voltage can be set by the choice of resistor, which allows the designer to trade off the voltage stress against the energy that is dissipated.

The alternative is to recover the energy rather than dissipate it. A resonant or active clamp recycles the leakage energy back into the circuit, which improves efficiency at the cost of complexity and a wider voltage stress on the switch. Where efficiency is the main constraint, that trade is worthwhile; where cost and simplicity dominate, an RCD clamp remains a reasonable solution.

Choosing the Resistor and Capacitor

The capacitor sets how much energy the snubber absorbs and how much it slows the transition. A larger capacitor reduces the ringing more but dissipates more energy in the resistor on every cycle, which lowers efficiency and raises the resistor temperature. The resistor sets the damping and must be able to absorb the energy pulse without changing value, which usually means a surge rated or a wire wound type rather than a thin film part.

Both components see a pulse rather than a steady current, so their ratings must be interpreted. The resistor energy per cycle is roughly half the capacitance times the square of the voltage swing, and the average power is that energy multiplied by the switching frequency. A value that looks acceptable at 20 kilohertz becomes a problem at 200 kilohertz, which is where most modern converters operate.

Layout and Loop Area

The snubber loop must be as small as the power loop it protects. The capacitor should be placed immediately at the device, with the resistor adjacent to it and the return path directly under the connection rather than around the board. Where the snubber is on the opposite side of the board from the switch, vias should be placed in pairs so that the loop area stays small.

Component choice interacts with the layout. A small package with low internal inductance is preferable, and a ceramic capacitor with a short body has less parasitic inductance than a large film part. Where the energy is high enough to require a film capacitor, the physical size grows and the loop with it, which is one reason that the empirical tuning of the snubber should be repeated after any component change.

Loss, Temperature, and Efficiency

A snubber always costs efficiency, and the cost should be quantified rather than assumed to be negligible. Measuring the input power with and without the network, at the same load and the same output voltage, gives the real figure. Where the loss is significant, the design should be reviewed: a better layout that reduces the source of the ringing may allow a smaller snubber, which is usually a better answer than accepting the loss.

Temperature is the practical limit. The resistor runs hotter than any other passive on the board, and its temperature rise must be checked at the worst case ambient and the maximum load. Where the resistor is close to a temperature sensitive component, the thermal coupling should be evaluated, because a resistor operating near its limit will change value and reduce the damping.

Measurement

Verification uses a voltage probe with a short ground connection at the device, a current probe in the power loop, and a clean trigger on the switching edge. Comparing the waveform with and without the snubber shows its effect directly, and the reduction in the peak voltage and in the ringing amplitude is the evidence that it is working.

Measurements should be repeated at the extremes of the operating range and at the highest ambient temperature, because the ringing amplitude changes with voltage and the damping changes with temperature. The relevant suppression measures should be evaluated together with the conducted emissions scan, since a snubber that reduces radiated emission may also reduce the conducted signature, and the two results belong in the same report.

Where a snubber is used on a high volume product, the tuning should be recorded as part of the design documentation. The waveform, the component values, the load, and the ambient temperature together define the state in which the network was optimised, and a later change to any of them can move the circuit out of that state. Recording the design intent prevents a cost reduction exercise from removing a component that was doing real work.

FAQ

How do I choose the initial snubber values? Measure the ringing frequency without a snubber, add a capacitor that halves it, and set the resistor to damp the response. Then tune at full load and at the voltage extremes.

Does a snubber reduce switching loss? No, it converts some of the ringing energy into heat and reduces the voltage stress. The total loss usually increases slightly, which is the price of the reduced stress and emissions.

Why does my snubber run hot? Because the energy per switching cycle multiplied by the frequency is high. A larger capacitor makes it worse, so reduce the capacitance to the smallest value that controls the ringing.