An inductor specified as 10 µH ±20% is not a 10 µH inductor — it is a component that may deliver 12 µH or 8 µH in your assembly. In a buck converter that tolerance translates directly into ripple current, peak current, saturation margin and, in a current-mode loop, into a change in the plant gain. Treating tolerance as a purchasing detail is one of the quiet causes of a design that passes on the bench and fails on the line.
1. Where the tolerance comes from
- Core AL tolerance: ±25% is typical for ungapped ferrite; ±3–10% for a ground gap.
- Turn count: an integer, so a 12-turn design that needs 11.6 turns has an inherent 3% error.
- Winding position and spread: a few percent, more on toroids than on bobbins.
- Temperature: ferrite permeability changes 10–25% from 25 °C to 100 °C; powder cores change far less.
- DC bias: a 20–40% drop at rated current is normal, and it is often not in the tolerance figure at all.
2. Ripple and peak current move with inductance
In a buck converter, ripple current is inversely proportional to inductance: ΔI = (Vin − Vout) × D / (L × f). A −20% inductance error gives +25% ripple. At a 30% nominal ripple that becomes 37.5%, and the peak current rises accordingly. If the saturation point was sized with only 15% headroom, the low-tolerance unit saturates on the worst-case line and load combination.
The fix is not simply to buy tighter parts. Design the converter at the minimum inductance of the tolerance band and verify at the maximum, because the maximum sets the loop response and the minimum sets the saturation risk.
3. Tolerance versus loop stability
In peak current-mode control the inner loop gain scales with inductance and with the current-sense gain. A ±20% swing moves the crossover frequency by the same proportion. Designs with a phase margin under 45° at nominal can oscillate at the tolerance extreme. If the controller has slope compensation, check it against the minimum inductance — insufficient slope compensation at low L is where sub-harmonic oscillation appears.
4. What tolerance actually costs
| Specification | Typical tolerance | Cost impact | When to use |
|---|---|---|---|
| Ungapped ferrite, standard | ±25% | Baseline | Filter inductors, non-critical chokes |
| Gapped ferrite, ground gap | ±10% | +5–15% | Most switching converters |
| Gapped ferrite, measured per piece | ±5% | +15–25% | Resonant tanks, precision filters |
| Powder core, distributed gap | ±8–15% | Baseline | Bias-heavy inductors, PFC |
Note that a tight AL tolerance does not fix temperature drift or bias roll-off. If the converter needs stable inductance across temperature, the material choice does more than the tolerance callout.
5. Specifying it correctly on the drawing
- State the test frequency (1 kHz, 10 kHz, 100 kHz give different readings on the same part).
- State the DC bias current at which inductance is measured, or specify both zero-bias and rated-bias values.
- State the temperature at which the value applies.
- State the measurement method: LCR meter series or parallel model, and the allowed fixture.
- Give separate limits for minimum (saturation) and maximum (loop) if the application needs them.
Shaanxi SiRui Technology measures inductance at the specified frequency and bias for every Magnetic Core series wound part, and can supply ±5% measured gapped cores where a resonant or precision application requires it.
Send your tolerance and test-condition requirement to sales@sirui-tech.com.