Power device loss calculator

How it works

This tool adds up the losses of the power devices in a converter, one device at a time, and solves the junction temperatures with the on-resistance rising as the junction warms. It covers a synchronous buck, a boost, a continuous-conduction PFC and a half-bridge or inverter leg for silicon MOSFETs and IGBTs, SiC MOSFETs and GaN HEMTs, then lines the chosen device up against an alternative at the same operating point. Efficiency here counts the devices only; inductors, capacitors, drivers and control are outside it.

1. Find the currents in each device. Each converter shape gives the average and RMS current of every device from the duty cycle and the inductor ripple. In a buck the high-side switch carries the mean inductor current for the duty D and the low-side device for 1 − D; the RMS current squared includes the ripple, the peak-to-peak ripple divided by the square root of 12. The boost is the mirror image on the input current. The PFC averages the same idea over the line cycle, and the inverter leg uses the sinusoidal PWM averages of the Infineon note, with the modulation index and the power factor.buck: D = Vout / Vin; I_rms^2 (switch) = D x (Iout^2 + ripple^2 / 12); I_rms^2 (low side) = (1 − D) x (Iout^2 + ripple^2 / 12) leg: switch I_rms^2 = Ipk^2 x (1/8 + M x cos(phi) / (3 x pi)); diode I_rms^2 = Ipk^2 x (1/8 − M x cos(phi) / (3 x pi))
2. Conduction loss. A MOSFET or HEMT is a resistor: its loss is the on-resistance at the junction temperature times the RMS current squared, with the on-resistance scaled by a linear temperature coefficient. With synchronous rectification the channel also carries the reverse current. Without it, and for an IGBT, the reverse current goes through a diode modelled as a fixed drop plus a slope. An IGBT's forward path is likewise a fixed saturation voltage times the average current plus a slope times the RMS current squared.MOSFET: P_cond = R_ds(on, 25 °C) x (1 + alpha x (Tj − 25 °C)) x I_rms^2 / n IGBT or diode: P_cond = V0 x I_avg + R x I_rms^2
3. Switching loss. Three ways to get the energy of each edge. From the data sheet energies Eon and Eoff, measured at a reference voltage and current, scaled in proportion to both. From the switching times, as a linear overlap of voltage and current. Or from the gate charge, plateau and threshold voltages, loop resistance and drive voltage, which give the times (the method of the Texas Instruments plateau-voltage note). In a buck the high-side switch turns on at the lower valley current and off at the upper peak.energy: E = E_ref x (V / V_ref) x (I / I_ref); times: E_on = 0.5 x V x I_on x t_r, E_off = 0.5 x V x I_off x t_f charge: t_r = (Qgs2 / (Vdr − 0.5 x (Vpl + Vth)) + Qgd / (Vdr − Vpl)) x Rg; t_f = (Qgs2 / (0.5 x (Vpl + Vth)) + Qgd / Vpl) x Rg; P_sw = f x (E_on + E_off)
4. Capacitance, recovery, dead time and gate drive. The output capacitance is charged and discharged each period, costing the stored energy at the bus voltage, scaled from the data sheet energy by a power of the voltage ratio. The body diode of the freewheeling device recovers its charge at each hard turn-on of the other (a GaN HEMT has none). During the dead time the freewheeling device conducts through its reverse drop. Gate drive costs the gate charge times the drive swing times the frequency.P_oss = E_oss(V_ref) x (V / V_ref)^k x f; P_rr = V x Qrr x (I / I_ref) x f; P_dead = f x t_dead x sum(V_sd(i) x i); P_gate = Qg x V_drive x f
5. Solve the junction temperature. The heatsink sits above ambient by the total loss of the stage times its thermal resistance. Each device sits above the heatsink by its own loss times the case-to-heatsink and junction-to-case resistances. Because the losses depend on the junction temperature, the tool repeats the calculation until the temperatures stop changing, and flags thermal runaway if they do not.T_heatsink = T_ambient + R_sa x P_total; Tj = T_heatsink + P_device x (R_cs + R_jc / n); repeat until Tj settles
6. Efficiency and comparison. Efficiency counts only the device losses. The load sweep keeps the relative ripple fixed, so it does not model the discontinuous mode or burst mode of a real converter at light load. The alternative device is run through the same operating point and thermal path with its own figures.efficiency = P_out / (P_out + sum of device losses)

Sources

The methods and default figures rest on these public sources. The wording here is our own; read the source for its full text.

• MOSFET Power Losses Calculation Using the Data-Sheet Parameters (Application Note V1.1), Infineon Technologies (Dušan Graovac, Marco Pürschel, Andreas Kiep), 31 July 2006. Conduction loss with the temperature coefficient, the linear switching model with reverse recovery, and the converter currents for the buck, boost and three-phase drive.
• An Accurate Approach for Calculating the Efficiency of a Synchronous Buck Converter Using the MOSFET Plateau Voltage (SLVAEQ9) (opens in a new tab), Texas Instruments (Xing He, Daniel Li, Vincent Zhang), July 2020. The switching loss from the gate charge, plateau voltage and loop resistance; gate drive, output capacitance, body diode reverse recovery and dead-time loss.
• Calculating Efficiency (SLVA390A) (opens in a new tab), Texas Instruments (Arvind Raj), February 2010, revised March 2020. RMS currents in a buck's switches with the ripple term; the ripple adds 0.375% to the RMS current at 30% ripple, which the tests check.
• COSS Losses in 600 V GaN Power Semiconductors in Soft-Switched, High- and Very-High-Frequency Power Converters (Zulauf, Park, Liang, Surakitbovorn, Rivas-Davila) (opens in a new tab), IEEE Transactions on Power Electronics, 2018. Why the output capacitance of a GaN device costs loss that grows with frequency and stored energy.
• Fundamentals of Power Electronics, third edition (Robert W. Erickson and Dragan Maksimović) (opens in a new tab), Springer, 2020. The textbook treatment of conduction and switching loss in converters and of the loss budget.