DEV Community

朱常
朱常

Posted on

Inductor Selection and Magnetics Design for DC-DC Converters

Inductor Selection and Magnetics Design for DC-DC Converters


Overview
The inductor is the energy-storage element of every buck, boost and buck-boost converter, and its choice directly sets ripple current, output voltage ripple, transient response, efficiency, and physical size. Engineers often treat the inductor as a passive item picked from a datasheet table, but saturation current, DC resistance, AC losses, shielding, and thermal behavior all determine whether the final power supply meets its efficiency and temperature targets. This article explains how to choose an inductance, interpret magnetics datasheets, and avoid common mistakes.
  1. Choosing the Inductance Value
    Inductance is chosen by the desired peak-to-peak ripple current. A common starting point is to set ripple current to 20 to 40 percent of the maximum rated output current (typical). For a buck, L = (VIN - VOUT) x D / (fSW x I_RIPPLE); for a boost, L = VIN x D / (fSW x I_RIPPLE). Higher inductance reduces ripple and output capacitor RMS current, which improves efficiency and lowers noise, but it makes transient response slower because the inductor current slews more slowly. Lower inductance lets the loop respond faster but increases ripple, peak switch current, and core losses. Most integrated converter datasheets give a recommended L range for a given switching frequency and output current.

  2. Saturation Current
    An inductor is rated for a saturation current ISAT - the DC current at which inductance drops by a specified percentage (often 20 to 30 percent). If the peak current (average load current plus half the ripple) exceeds ISAT, the core saturates, inductance collapses, and switch current spikes, potentially destroying the FET. The selection rule is simple: ISAT must exceed the worst-case peak current, including transient and startup margin, with headroom. Many datasheets plot inductance versus DC current; read that curve at your worst-case temperature, because saturation current drops as the core heats up.

  3. Core Materials

  4. Ferrite cores: high permeability, low core loss at high frequency, but saturate sharply and are sensitive to temperature. They are common in shielded power inductors for multi-megahertz converters.

  5. Powdered iron (iron powder) cores: distributed air gap, soft saturation (inductance rolls off gradually), higher core loss. They tolerate high peak current without hard failure and are common in boost and automotive inductors.

  6. Molypermalloy (MPP) and high-flux cores: low loss, stable across temperature, but more expensive; used in high-efficiency and high-reliability designs.

  7. DC Resistance and AC Losses
    Losses in an inductor come from:

  8. Copper DC loss: IL^2 x DCR. DCR is specified in milliohms and is the dominant loss in most designs.

  9. Copper AC loss: skin and proximity effect at the switching frequency, important at high fSW and high ripple.

  10. Core loss: proportional to flux swing and frequency; ferrite cores keep this small at megahertz operation.
    A low-DCR inductor runs cooler and more efficient, but physically larger. Compare loss at your actual operating point, not just the datasheet's typical efficiency number.

  11. Shielded versus Unshielded
    Unshielded (open-core) inductors are cheaper and smaller but radiate stray magnetic field, which can couple into sensitive analog traces and cause EMI. Shielded inductors enclose the winding in a magnetic core, reducing radiated field, at the cost of slightly higher DCR and size. Use shielded inductors near noise-sensitive circuits, and keep unshielded parts away from feedback traces and analog sections.

  12. Reading the Datasheet Curves

  13. Inductance versus DC bias: shows how L falls as current rises; do not pick an L where bias has already knocked it below your design target.

  14. Saturation current versus temperature: ISAT degrades with heat; derate for the ambient inside the enclosure.

  15. DCR at temperature: copper resistance rises about 0.4 percent per degree C; budget for hot-spot DCR.

  16. Rated current (heating current): the current that raises the inductor by a specified temperature rise (often 40 C); this is a thermal limit, not a saturation limit. Both numbers must be respected.

  17. Design Considerations

  18. Calculate peak current (IOUT max + half ripple + transient margin) and choose ISAT above it.

  19. Derate inductance for DC bias and temperature; many inductors lose 20 percent of L at rated current.

  20. Keep the inductor's footprint small but its DCR low; compare size-to-loss trade-offs, not just L value.

  21. Place the inductor so the switch-node loop and input loop are tight; magnetic field coupling into feedback is a common source of jitter.

  22. For high-current multiphase designs, use rated-current (thermal) curves as well as saturation curves; heat spreading matters.

  23. Verify the inductor's self-resonant frequency is well above fSW, or its impedance behavior will deviate from ideal.

  24. Conclusion
    Inductor selection is not a passive afterthought. The L value sets ripple and transient speed; ISAT and core material set robustness; DCR and core loss set efficiency and temperature; shielding sets EMI. Read the bias and saturation curves at your operating point, budget for temperature rise, and you will avoid the most common magnetics failures.

FAQ
Q: Should I use the inductor's rated current or its saturation current to size it?
A: Both must be satisfied. Saturation current ensures L does not collapse; rated heating current ensures the inductor does not overheat. Whichever limit is lower wins.

Q: Why does my inductance read lower than the label?
A: DC-bias derating. Ferrite inductors lose inductance at high DC current; check the bias curve, not the part number value.

Q: Can I use a larger inductor than recommended?
A: Yes, up to a point. Larger L reduces ripple but slows transient response and adds size; the loop must still be stable at the new LC pole.

Q: Is a shielded inductor always necessary?
A: No, but near analog circuits, feedback traces, or radio sections, shielding eliminates field coupling issues at modest cost.

For exact specifications and design guidance, refer to the official inductor manufacturer datasheets and the inductor-selection tables in the chosen converter datasheet.

Top comments (0)