Output Capacitor Ripple and ESR Heating Calculator

The load takes the DC; the output capacitor takes everything else. For a continuous-inductor output that is the triangular ripple current, and it stresses the capacitor twice: its charge swings the voltage through the capacitance, and its current drops voltage across the ESR and heats it. Which term dominates decides whether you need more microfarads or a better dielectric - they are different purchases.

Triangular ripple current dividing between the load and the output capacitor LC + ESRloadIL = DC + triangleDC onlytriangle onlycapacitor currentdI, zero mean dV = dI/(8fC) + dI x ESR, 90 deg apart
The load will only take DC, so the whole triangle is forced into the capacitor. Its charge works against the capacitance and its current against the ESR - two ripple mechanisms with two different cures.

Inputs — enter your values

From the inductor ripple calculator, or measured. The whole triangle lands in the capacitor.

Effective value at the DC bias - class 2 ceramics can lose over half their nameplate capacitance at rated voltage.

At the switching frequency, from the impedance curve - not the 120 Hz datasheet headline.

Case to local ambient. Zero skips the self-heating estimate.

Results — computed for you

Conservative sum of the two components; the true value sits between the larger component and this sum because they are 90 degrees apart.

dI/(8fC) - shrinks with more capacitance or higher frequency.

dI x ESR - immune to more capacitance. Only a better capacitor technology reduces it.

dI/sqrt(12). Check it against the capacitor ripple-current rating at your frequency and temperature.

Electrolytic life halves roughly every 10 K - this number is why ripple rating matters.

Choosing something to solve for makes Output voltage ripple, peak to peak an editable target and computes the chosen input from it.

Formula

Ic,rms = dI / sqrt(12) dVc = dI / (8 f C) dVesr = dI x ESR dVpp = dVc + dVesr (conservative: components are 90 deg apart) P = Ic,rms^2 x ESR, dT = P x Rth
dI peak-to-peak triangular inductor ripple (A)
f, C switching frequency (Hz) and capacitance (F)
ESR equivalent series resistance at f (Ohm)
Rth capacitor thermal resistance, case to ambient (K/W)

What this model assumes, and where it stops

Assumptions

  • Continuous-inductor output: the capacitor sees a clean triangle. Buck outputs qualify; boost and buck-boost outputs see a pulsed diode current instead, with far harsher RMS - this page is not for them.
  • All the ripple current enters this capacitor - any second stage or parallel bank shares it in inverse impedance ratio.
  • ESR and capacitance are the effective values at the switching frequency and DC bias.
  • ESL is neglected. Above a few hundred kilohertz the inductive spike can rival the ESR step.

Limitations

  • The total shown is the arithmetic sum of two components that peak a quarter period apart, so it overstates the true ripple by up to about 25% in the mixed regime. The two components individually are exact.
  • Boost and buck-boost output capacitors carry the discontinuous diode current: RMS of order Iout x sqrt(D/(1-D)), several times harsher than the triangle here. Use the PWM pulse RMS page for that waveform.
  • Load transients, not steady-state ripple, usually set the output capacitance in point-of-load converters - this page does not size for transients.
  • Ceramic capacitance falls with DC bias and ESR varies strongly with frequency and temperature; feed effective values, not nameplate ones.

When you need a 3D field solution instead

Closed-form models like the one above hold on idealised geometry. These are the cases where they stop being good enough and a full 3D electromagnetic and thermal solution is the only way to get a trustworthy answer:

  • Current sharing in a paralleled capacitor bank, where trace and terminal inductance decide which capacitor takes the ripple.
  • Hot-spot temperature inside a large can, where the winding-to-case gradient adds to the case-to-ambient rise calculated here.
  • High-frequency loop layout: the effective ESL of the placement often matters more than either term this page computes.

Common questions

Should I fix ripple with more capacitance or lower ESR?

Look at the split this page reports. The capacitive term dI/(8fC) shrinks with more microfarads or higher frequency; the ESR term dI x ESR does not care how many microfarads you add. With a 2 mOhm polymer at 200 kHz the ESR carries about a quarter of the ripple; with a 50 mOhm electrolytic it carries over 90%, and adding capacitance is nearly pointless. Ceramics are the opposite - almost purely capacitive.

What does a capacitor ripple current rating actually protect?

The electrolyte. Ripple current times ESR is heat generated inside the can, and electrolytic lifetime roughly halves for every 10 K of extra core temperature. The rating is the current at which self-heating stays inside the life budget at rated ambient. This page computes the actual dissipation and rise so you can compare like with like - at your frequency, where ESR is far below the 120 Hz datasheet value.

Why does my measured ripple differ from the calculation?

Three usual causes. The two components here are 90 degrees apart, so the arithmetic sum shown overstates the true peak-to-peak by up to about 25% in the mixed regime. ESL adds a switching-edge spike this model omits entirely - often the biggest term on a scope at high frequency. And ceramic capacitors lose a large fraction of nameplate capacitance at DC bias, so the effective C is smaller than the label.

References

  • Erickson & Maksimovic - Fundamentals of Power Electronics, 3rd ed. - output filter ripple analysis
  • TI SLVA630 - Output ripple voltage for buck switching regulator
  • Nichicon technical notes - Ripple current, self-heating and lifetime of aluminium electrolytic capacitors

Related calculators