ZVS Induction Heater Coil, Capacitor and Heating Calculator

Size a ZVS induction heater work coil and tank capacitor. Calculate loaded resonance, skin depth, RMS currents, workpiece power and heating time through the Curie transition.

#capacitor ESR #Curie transition #induction heating #Q factor #resonance #thermal ODE #ZVS

A ZVS induction heater can draw a modest DC current while hundreds of amperes circulate between its work coil and tank capacitor. This induction heater calculator connects those electrical stresses to the power actually absorbed by a cylindrical metal workpiece and its predicted heating time. It is intended for engineers, students and builders comparing coil dimensions, capacitor banks and heating conditions.

Start with the supply voltage, actual capacitor bank capacitance and its ESR at operating frequency. Enter the work coil's inside diameter, winding length, turns and copper tube dimensions. Include the complete copper lead length and an estimate of its stray inductance. The workpiece should fit concentrically inside the coil, with its heated length close to the winding length. The drawing updates to those dimensions; drag the diameter handle to explore the fit.

As a work coil inductance calculator, the tool applies the Nagaoka finite-length correction rather than treating every winding as an infinitely long solenoid. It then represents the workpiece as a loosely coupled shorted secondary. Coupling depends strongly on clearance, overlap and the surrounding metal. The flux linkage adjustment lets you calibrate the estimate against a measured loaded coil. Without that calibration, absorbed power is a design estimate.

Material inputs matter as much as geometry. Supply resistivity at 20 °C, its temperature coefficient, effective relative permeability and a mean heat capacity appropriate to your temperature range. For nonmagnetic conductors, set relative permeability to one. For steel, the model brings permeability smoothly to unity at the entered Curie temperature. The induction heating skin depth calculator therefore shows why the same workpiece can respond very differently before and after it becomes nonmagnetic.

The ZVS induction heater coil calculator solves loaded parallel resonance at each temperature, including copper and capacitor losses. It reports coil and capacitor RMS currents, tank peak voltage, electrical efficiency and the capacitor bank's reactive duty. The maximum current and reactive rating over the heating sweep are the quantities to compare with component data. The optional design frequency gives an ideal starting capacitance at the initial workpiece temperature.

Press play to follow the heating transient. The coloured section shows current penetration, while the exposed metal changes its incandescent colour with computed mean temperature. The thermal integration includes convection and radiation and stops at the requested target or simulation duration. A target not reached within that duration is reported explicitly.

This model covers a solid cylindrical workpiece and a cooled single-layer coil. It does not predict hardening depth, melting, local hot spots, startup or MOSFET switching transients. Validate coupling, material properties, cooling and capacitor ratings for the actual assembly before selecting hardware.

How it is calculated

The empty coil is a single cylindrical current sheet with Nagaoka's finite-length factor (Nagaoka, 1908, Eqs. I–II, printed p. 316). Complete elliptic integrals are evaluated by 512-panel quadrature. With mean radius r and winding length ℓ, L = μ₀πr²N²K/ℓ; lead inductance is added separately.

The workpiece is a shorted, one-turn cylindrical secondary. This implements the transformer and skin-effect approach described in onsemi AN-9012, printed pp. 1–2. The external secondary inductance uses the same current-sheet expression with one turn. The adjustable coupling estimate is k = k_link(r_work/r_coil)√(ℓ_short/ℓ_long), and M = k√(L_winding L_secondary). This geometry estimate is an assumption to calibrate, not a measured coupling law.

At every temperature, ρ = ρ₂₀[1 + α(T − 20 °C)]. Effective relative permeability falls by a smooth cubic from the entered value at T_C − ΔT_C to unity at T_C, and remains unity above it. Skin depth is δ = √[2ρ/(ωμ₀μᵣ)]. The surface impedance gives R₂ = ρπd/(ℓδ) and Z₂ = (1 + j)R₂ + jωL₂. Reflected impedance is Z_ref = ω²M²/Z₂. Copper resistance uses the tube's outer conducting annulus of depth min(δ_Cu, wall), multiplied by the entered proximity factor; lead resistance is included.

A bracketed root solve finds zero imaginary part of the total parallel admittance, including capacitor ESR. For a stiff supply and sinusoidal current-fed push-pull operation, each feed choke has zero mean voltage: the half-sine drain waveform has mean V_peak/π. Thus the ideal differential tank peak is πV_dc and its RMS value is πV_dc/√2. Branch currents follow their complex impedances. Separate I²R terms give workpiece, copper and capacitor losses; capacitor reactive rating is I_cap²/(ωC), reported in kV·A, numerically kvar. Total loaded Q is the capacitor reactive duty divided by the sum of workpiece, copper and capacitor ESR losses. Its minimum over the entire temperature sweep guards the sinusoidal-voltage assumption, independently of playback time. The design capacitance cancels the coil susceptance at the requested frequency and initial temperature; that recommendation neglects capacitor ESR.

The thermal ODE integrates m c_p dT/dt = P_work(T) − hA(T − T_air) − εσA(T_K⁴ − T_air,K⁴), plus cumulative absorbed and lost energy. The temperature grid concentrates samples in the Curie transition and uses shape-preserving interpolation; an adaptive Dormand–Prince pass locates the target crossing or duration limit. A second pass evaluates dense solver output on a uniform grid over that actual run interval, supplemented by finely subdivided adaptive intervals; playback uses shape-preserving interpolation of this history. The image uses |J(x)|/|J_surface| = exp(−x/δ); the dashed circle marks one skin depth. Incandescent colour is estimated from Planck radiance at three visible wavelengths, with fixed display exposure.

Assumptions

A centered solid cylindrical workpiece, a single-layer round copper tube coil, and approximately equal heated and winding lengths. The whole workpiece is represented by one temperature. Density, mean heat capacity, thermal conductivity, convection coefficient and emissivity are user inputs. The starting steel-like values are illustrative engineering assumptions, not a certified alloy dataset.

The supply maintains its entered voltage and can furnish the displayed DC current. The coil stays at the entered copper temperature through cooling. The driver efficiency is an allowance for switches and feed chokes; it does not alter the ideal tank voltage. Proximity loss and coupling require calibration for an actual assembly. Annealed copper uses ρ₂₀ = 1.724 × 10⁻⁸ Ω·m and α = 0.00393 K⁻¹ (standard IACS values). Hysteresis loss, saturation and permeability's field dependence are omitted.

Where it applies

The workpiece surface-impedance and exponential-profile approximation needs diameter ≥ 8δ throughout the temperature sweep; a condition flags smaller parts. It does not solve the weak-skin cylindrical Bessel problem. Finite conductor pitch, discrete turns, lead layout and end fields introduce additional inductance error beyond the Nagaoka current-sheet solution. Treat the geometric coupling estimate as a starting point and replace its adjustment using measured loaded impedance.

The heating time is a mean-temperature estimate, not a surface or core temperature prediction for hardening. Biot number and radial diffusion time conditions flag conditions needing a spatial thermal model. Phase changes, latent heat, alloy-specific Curie/heat-capacity anomalies, temperature gradients, support conduction, coil-to-part radiation and oxidation changes in emissivity are outside the model. Set a suitable average heat capacity and temperature coefficient over the intended range; the linear resistivity law is not an alloy property curve.

The default temperature sweep has 240 intervals. An adaptive solve first finds the actual run duration; 1200 uniform history intervals then span that duration, even when the requested maximum duration is much longer. Subdivided adaptive intervals add samples around rapid changes so a long unreached-target tail cannot erase the earlier transient. The uniform history resolution is adjustable under Advanced; adaptive intervals receive at least four subdivisions, eight at the default setting. Both ODE passes use adaptive steps, and separate convergence checks cover stopping time, playback temperature and playback power. The end section is fitted to its own scale. The 151 × 151 current-field image samples the analytic skin profile and blends cells; cold skin layers thinner than a display cell are not spatially resolved. No thermal diffusion field is inferred from its colour. Optical colour is illustrative, not calibrated pyrometry.

The low-Q warning uses the worst capacitor-reactive-duty/total-real-loss ratio across the specified temperature sweep, including capacitor ESR. The sweep extends to the requested target even if it is not reached within the duration limit, so this guard is conservative. Low Q, supply current limiting, imperfect gate timing and feed-choke saturation invalidate the π-voltage assumption. Tank voltage, circulating current and capacitor reactive duty can be much larger than the DC supply values. Check the capacitor maker's AC voltage, RMS current, ESR, frequency and cooling ratings together; a DC voltage marking alone is insufficient. The small circuit omits the gate network, protection and parasitics. This tool does not establish startup, ZVS margin, MOSFET overshoot or a working protection design.

Reference cases verify published inductance coefficients and skin depths. Complete heater power and time predictions have not been validated against a measured assembly; coupling and material data dominate that uncertainty.

References

  • H. Nagaoka, Note on the Self-inductance of Solenoids, 1908, pp. 314–325. https://doi.org/10.11429/ptmps1907.4.16_314 (open source s3; scanned numerical table on PDF p. 12).
  • onsemi / Fairchild, AN-9012, Induction Heating System Topology Review, rev. 1.0.4 (2013), printed pp. 1–4. https://www.onsemi.com/download/application-notes/pdf/an-9012.pdf (open source s2).
  • ROBOTERM, What is induction heating: exponential penetration, the eight-skin-depth diameter criterion, and copper skin-depth reference values at 40 °C. https://www.roboterm.cz/en/induction/about-induction/what-is-induction-heating (open source s4).
  • Rudnev, Loveless and Cook, Handbook of Induction Heating, 2nd edition, CRC Press, 2017. https://doi.org/10.1201/9781315117485 — further reading on coupled electromagnetic and thermal heating; no licensed figures or tables reproduced.
  • E. J. Davies, Conduction and Induction Heating, IEE Power Engineering Series 11, 1990. https://doi.org/10.1049/PBPO011E — further reading on workpiece impedance and heat transfer.
  • L. A. Morley, Mine Power Systems, Volume I, U.S. Bureau of Mines, 1982, Table 3.2, printed p. 77: annealed copper resistivity and temperature coefficient. https://stacks.cdc.gov/view/cdc/235133

Inputs

InputSymbolUnitDefault
Supply voltageV_dcV48
Tank capacitanceC_tuF2
Capacitor bank ESRR_capmohm1.5
Capacitor design frequencyf_reqkHz100
Work coil turnsN_c6
Coil inside diameterD_inmm40
Winding lengthl_coilmm45
Copper tube outside diameterd_tubemm6
Copper tube wall thicknesswall_cmm1
Total lead lengthl_leadmm100
Copper temperatureT_cu°C40
Coil proximity loss multiplier
AC loss beyond isolated skin effect; use measured coil resistance when available.
F_prox2
Workpiece diameterD_wmm24
Workpiece lengthl_workmm45
Flux linkage adjustment
Multiplies diameter and overlap estimates of coupling; calibrate against measured loaded inductance.
k_link0.85
Workpiece resistivity at 20 °Crho_20uohm*m0.16
Resistivity temperature coefficientalpha_r1/K0.005
Relative permeability below transition
Use effective permeability at the actual field level; set 1 for nonmagnetic metals.
mu_cold70
Curie temperatureT_curie°C770
Curie transition widthdT_curieK80
Densityrho_masskg/m^37850
Mean specific heat capacity
Use the mean heat capacity over the intended temperature range.
c_heatJ/(kg*K)600
Thermal conductivityk_thermW/(m*K)35
Initial workpiece temperatureT_start°C20
Target temperatureT_goal°C850
Ambient temperatureT_air°C20
Convection coefficienth_convW/(m^2*K)12
Surface emissivityemiss0.8
Heating simulation durationt_maxs600
Temperature sweep intervalsn_temp240
Elapsed heating timets180
Lead and stray inductanceL_strayuH0.1
Driver and feed efficiency allowanceeta_driver0.9
Uniform history intervals
Adaptive time intervals are also subdivided to preserve sharp transients.
n_history1200

Results

With the default inputs above, the calc gives:

ResultValue
Heating outcomeTarget reached
Time to target / duration limit367.7 s
Temperature at elapsed time794.2 °C
Work coil inductance1.142 uH
Loaded inductance1.048 uH
Loaded resonant frequency109.9 kHz
Ideal bank for design frequency2.269 uF
Estimated coupling coefficient0.443
Tank peak voltage150.8 V
Tank RMS voltage106.6 V
Coil RMS current147.3 A
Power into workpiece353.9 W
Copper loss207.4 W
Capacitor ESR loss32.5 W
Tank to workpiece efficiency0.596
DC to workpiece efficiency0.536
Estimated supply current13.7 A
Workpiece skin depth1.34 mm
Relative permeability1
Workpiece resistivity0.7793 uohm·m
Convection and radiation loss291.5 W
Final temperature in simulation850 °C
Heated mass0.16 kg
Maximum bank RMS current147.3 A
Maximum bank reactive duty (kvar)15.7 kV·A
Minimum loaded tank Q17.9
Maximum coil cooling load207.5 W

Checked against sources

Each case below was calculated by this calc and agrees with the value from its source (within the tolerance where one is shown).

Nagaoka published 45° coefficient

Source: Nagaoka (1908), DOI 10.11429/ptmps1907.4.16_314, printed p. 325 / PDF p. 12, scanned table: α = 45°, coefficient 0.688423. Here mean diameter and length are both 46 mm, so α = atan(D/ℓ) = 45°. Value read from the original table image.

Inputs: Winding length = 46 mm; the others as above.

ResultSourceThis calc
Coil Nagaoka coefficient0.68840.6884

Published copper penetration at 10 kHz

Source: ROBOTERM, What is induction heating, copper at 40 °C: 10,000 Hz → 0.7 mm (rounded to 0.1 mm). The same page states 86.5% of heating lies within one skin depth in the high-skin limit. The tolerances follow the published rounding, not a fit of resistivity.

Inputs: Workpiece resistivity at 20 °C = 0.01724 uohm·m; Resistivity temperature coefficient = 0.00393 1/K; Relative permeability below transition = 1; Initial workpiece temperature = 40 °C; Target temperature = 100 °C; Capacitor design frequency = 10 kHz; the others as above.

ResultSourceThis calc
Skin depth at design frequency7 × 10⁻⁴ ± 5 × 10⁻⁵ m6.863 × 10⁻⁴ m
Fraction of high-skin heating within one skin depth0.865 ± 0.0010.8647

Published copper penetration at 50 kHz

Source: ROBOTERM, What is induction heating, copper at 40 °C: 50,000 Hz → 0.3 mm. Absolute tolerance 0.05 mm reflects the source's one-decimal rounding.

Inputs: Workpiece resistivity at 20 °C = 0.01724 uohm·m; Resistivity temperature coefficient = 0.00393 1/K; Relative permeability below transition = 1; Initial workpiece temperature = 40 °C; Target temperature = 100 °C; Capacitor design frequency = 50 kHz; the others as above.

ResultSourceThis calc
Skin depth at design frequency3 × 10⁻⁴ ± 5 × 10⁻⁵ m3.069 × 10⁻⁴ m

Related Resources

Responses

Your email address will not be published. Required fields are marked *

L ading...