Bridge snubber capacitor
How much capacitance at the bridge keeps the voltage spike at turn-off under a limit. When the switches open, the current in the loop's stray inductance has to change, and until it has, the difference goes into the capacitor. In a full bridge it reverses rather than stopping: the tank's current carries on through the opposite diodes back into the bus, a step of twice the peak. The page sizes for the worst case, a turn-off at full current, and shows the ordinary one beside it, where the phase lead turns the bridge off just ahead of the current's zero.
snubber capacitor385 nF160 A off, 440 V bus, 600 V limit
normal turn-off490 V50 A with the phase lead set
stray inductance96.4 nHwires, 10 cm, 10 mm apart
ΔI = 2·I (full bridge: the current reverses) ΔI = I (half bridge)
½·L·ΔI² = ½·C·ΔV² so C = L·ΔI²/(V_max - V_bus)²
rise with a given C = ΔI·√(L/C) ring f = 1/(2π√(L·C))
normal turn-off: the current the phase lead page says the bridge breaks
loop L, worked out from what you entered, never asked for:
two wires (out and back): L = (µ0·l/π)·acosh(s/d)
bus bars: L = µ0·l·h/w
two rings, with C_add clipped on for the second:
L = (1/(2π·f₂)² - 1/(2π·f₁)²) / C_addThe page does not ask for the stray inductance. It is rarely known, so the page works it out from what you can measure with a ruler: how far the bus capacitor is from the bridge, how far apart the two wires run, and how thick they are (or, for bars, their width and the insulation between them). That is the loop the current flows round, out and back, and it comes to about a nanohenry a millimetre for wires an ordinary distance apart, tens of times less for a laminated bus. The formula published calculators use, L·I²/ΔV², is the one-leg case. A full bridge needs four times that, because the step is 2I. Until 03.11.2026 this page used the one-leg form for both. A review caught it by simulating the reference bridge, where the printed capacitance let the spike reach 760 V against a 600 V limit.
With a scope it can be measured instead. The switch rings when it turns off: the loop against the switch's own capacitance. Note that frequency, clip a film capacitor of a known value across the switch, and note the new, lower one. The two frequencies and the one capacitor you added give the inductance, with no need to know the switch's capacitance. A nanofarad or two is a sensible size to add.
The worst case is rare. With the phase lead set, the bridge turns off a little ahead of the current's zero, where the current is a fraction of the peak, and the spike is that much smaller. A turn-off at full current happens when something has gone wrong: lost feedback, a start-up that misses the pole, an overcurrent trip that does not wait for a zero. Size for it anyway, because that is the edge that kills switches.
Placement matters. Across the bus, at the bridge, a film capacitor with short leads is the usual snubber and is all this page assumes. A capacitor across each switch instead adds to the capacitance the bridge has to swing in its dead time, which the phase lead page counts as the snubber term of its zero voltage check. If the switch ever turns on hard, it dumps ½·C·V² into it every edge.
The limit field is the spike you allow, not the switch's rating: the spike rides on top of the bus, so the bus has to sit far enough below the rating to leave room for it. The layout the page opens on is an example, not the reference coil's, which has not been measured. The current it breaks normally is the one the phase lead page prints as “must break at turn-off”; it only changes the ordinary turn-off line.