RC snubber calculator

Every switching converter has a node where the switch, the diode and the inductor meet. When the switch turns off, that node jumps from 0 V to the supply in a few nanoseconds. The loop the current flows through has a little inductance (the traces, the package leads, the bond wires), and the switch has some capacitance across it. Those two form a tuned circuit. Each edge kicks it, and it rings.

TI's SLVA255 shows a boost converter whose switch node should peak at 5.7 V but rings to 9.8 V at 217 MHz. That overshoot can exceed the switch's voltage rating, and the ringing radiates. An RC snubber across the switch gives the ringing current a resistor to burn its energy in.

You can't measure a few nanohenries with a meter, but you can measure what they do. The method comes from TI SLVA255 and Cornell Dubilier's snubber paper:

1. Capture the ringing on the scope and read its frequency, f1 (one over the period).
2. Solder a small ceramic capacitor, Cadd, straight across the switch. Try values until the ringing frequency drops to roughly half.
3. Read the new frequency, f2.

The frequency goes as 1/√(LC), so (f1/f2)² = (Cpar + Cadd)/Cpar. That gives Cpar = Cadd / ((f1/f2)² − 1), and then Lpar = 1/((2π f1)² Cpar). In TI's example, 300 pF moved the ringing from 217 MHz to 113 MHz. The calculator gets Cpar = 112 pF and Lpar = 4.82 nH.

Measure carefully or you will measure your probe. TI used a probe and scope of at least 500 MHz, about twice the ringing frequency, with the probe's ground loop kept as small as it could be. A 15 cm ground clip has its own inductance, and it can ring at a frequency close to the one you are chasing.

With Lpar and Cpar known, the resistor that damps the ringing best sits near the loop's characteristic impedance, Z0 = √(Lpar / Cpar). TI SLVA255 uses Rs = Z0. TI SLPA010 treats the node as a parallel RLC circuit and uses half of Z0. The calculator offers both. For the capacitor, SLVA255 uses at least the Cadd that halved the frequency, which is three times Cpar.

For the 12 V buck node in worked example 2, Z0 = 3.53 Ω, so Rs = 3.3 Ω and Cs = 3 × 282 pF = 845 pF, rounded up to 1 nF. In the model the overshoot falls from 85 % to 26 %, and the node settles within 5 % in 7.27 ns instead of 59.6 ns.

The best resistor value is broad. Halve it or double it and the snubber still does most of its work, which is why a 5 % E12 part is fine.

The snubber capacitor swings through the full node voltage on every edge. On the rising edge it charges through Rs and half of Cs × V² turns into heat. On the falling edge it discharges through Rs and the other half goes the same way. So the average loss is Cs × V² × fsw, and the value of Rs does not change it. Cornell Dubilier's paper uses this form; its example of 390 pF at 160 V and 100 kHz works out to 998 mW.

TI's two notes write the loss as ½ × Cs × V² × fsw, which is the energy taken on one edge. Size the resistor for the full figure. It is the safer number, and it is the one you will measure as heat.

For the resistor package, the calculator picks the smallest chip size whose rating at 70 °C, from Vishay's D/CRCW e3 data sheet, is at least twice the loss. 72 mW needs a 1206 part (250 mW). Check the peak current too: the whole step lands across Rs at the edge, 12 V / 3.3 Ω = 3.64 A for a few nanoseconds.

The snubber only works if its own loop is smaller than the loop it is damping. Put Rs and Cs right at the switch: from the drain (or the switch-node pin of a regulator IC) to the ground pad next to the source, with a via straight down to the ground plane. Parked a few centimetres away at the end of a thin trace, the snubber adds inductance of its own and the node rings anyway.

• Fix the layout first. A shorter loop lowers Lpar, which raises the ringing frequency and shrinks its energy. Only then add a snubber, and leave pads for it on the first board even if you never fit it.
• Use C0G (NP0) ceramic for Cs. X7R loses much of its capacitance at high voltage, so the snubber would be weaker than it reads on the label. Check that its voltage rating is above the peak of the ringing, not only the supply.
• Use a thick-film chip resistor or a carbon composition part for Rs. A spiral-cut film resistor behaves like an inductor at 100 MHz.
• Keep the scope probe's ground short (Figure 3) and use bandwidth of at least twice the ringing frequency, or you will measure the probe.
• After fitting the snubber, feel or measure the resistor's temperature at full load. If it runs hot, Cs is too big for that switching frequency.
• A gate resistor is the other knob. Slowing the switch's turn-on also cuts the ringing, at the cost of more switching loss in the switch itself.

The snubber is an RC network, so the RC time constant calculator shows how fast Cs charges through Rs. The ringing itself is a series resonance: see the lessons on series RLC circuits and Q factor. For the switch, read the MOSFET as a switch, and for a whole converter, how to make 5 V from one 1.5 V cell.

Sources: Texas Instruments, SLVA255, Minimizing Ringing at the Switch Node of a Boost Converter (September 2006), pp. 2-4; Texas Instruments, SLPA010, Ringing Reduction Techniques for NexFET High Performance MOSFETs (November 2011), pp. 11-13; Cornell Dubilier, Rudy Severns, Design of Snubbers for Power Circuits, pp. 4-10; Vishay, D/CRCW e3 thick film chip resistors, document 20035 (revised 14 April 2026), pp. 1-2. The ringing plots are our own model of those formulas. Checked 23 October 2026.