Short answer: pick a strand diameter at or below about one skin depth of your operating frequency, size the total copper cross-section from your current (3–6 A/mm² is typical at high frequency), divide one by the other to get the strand count, then verify the bunched outer diameter fits your winding space. The skin-depth table and two fully worked examples below do exactly this.
In copper, skin depth in millimetres ≈ 66 / √f, with f in hertz. Keep the strand diameter at or below roughly one skin depth (0.7–1.5 δ works in practice)
| Operating frequency | Skin depth δ (copper) | Typical strand diameter |
|---|---|---|
| 20 kHz | 0.467 mm | 0.30–0.50 mm |
| 50 kHz | 0.296 mm | 0.20–0.30 mm |
| 100 kHz | 0.209 mm | 0.15–0.20 mm |
| 200 kHz | 0.148 mm | 0.10–0.15 mm |
| 400–500 kHz | 0.100–0.105 mm | 0.05–0.10 mm |
| 1 MHz | 0.066 mm | 0.05 mm and finer |
Take the strand diameter from the skin-depth table above, scaled to your exact frequency.
Total copper area = rated current ÷ current density. 3–6 A/mm² is typical at high frequency; tighten it for sealed or high ambient temperature windings.
Strand count = total copper area ÷ area of one strand. One 0.2 mm strand is 0.0314 mm², one 0.1 mm strand is 0.00785 mm². Round up to a standard construction.
Bunched OD ≈ 1.25 × strand dia. × √strand count. Confirm it fits the slot or bobbin and winds cleanly; if not, adjust strand diameter and repeat.
A typical wireless charging transmitter coil specification
| Step | Result |
|---|---|
| Skin depth at 100 kHz | 0.209 mm → choose 0.20 mm strand (≈ 1 δ) |
| Copper section at 4 A/mm² | 12 A ÷ 4 = 3.0 mm² |
| Strand count | 3.0 ÷ 0.0314 = 95.5 → 100 strands × 0.20 mm |
| Check: actual density | 100 × 0.0314 = 3.14 mm² → 3.8 A/mm² ✓ |
| Check: bunched OD | 1.25 × 0.20 × √100 ≈ 2.5 mm ✓ |
A real build we supplied, measured and verified
| Step | Result |
|---|---|
| Skin depth at 440 kHz | 0.100 mm → choose 0.10 mm strand (≈ 1 δ) |
| Strand count | 3500 strands × 0.10 mm, built as re-twisted bundles |
| Total copper section | 3500 × 0.00785 = 27.49 mm² |
| DC resistance | 0.627 mΩ/m |
| Continuous rating | 80 A in a naturally cooled winding (2.9 A/mm²); 90 A absolute max |
| Finished OD | ≈ 7.3 mm |
DC wire tables ignore skin and proximity effect. In a multi-layer winding above tens of kilohertz, AC resistance of a solid or coarse-stranded conductor can be several times its DC value — and your coil heats exactly where the current is.
Strand diameter below the skin depth buys little, and every extra strand adds insulation and cost. Fill factor drops because a growing share of the bundle is enamel, not copper.
Adjacent turns and layers distort each other's field. When the winding has many layers or tight pitch, specify a multi-bundle (re-twisted) construction — bundles twist groups of strands so that each takes its share of the layers.
Thermal class 130 / 155 / 180 / 200 / 220 must match your worst-case hotspot, not your ambient. Served or yarn-covered constructions add outer diameter — include it in the slot check before ordering.
Possible but uncommon. Standard constructions keep one strand diameter throughout — bunching stays uniform and soldering predictable. If a single diameter cannot meet loss and flexibility targets, the usual fix is a multi-bundle (re-twisted) construction, not mixed strands.
When strands are sized near one skin depth, the AC/DC resistance ratio stays close to 1.1–1.3 even in multi-layer windings. The same total copper as a solid wire can see several-fold higher AC resistance at the same frequency. Exact numbers depend on layer count and pitch, which is why ampacity confirmation is worth doing per design.
Yes, as re-twisted bundles. The 3500-strand example above is a production build with a 7.3 mm finished OD. Strand counts of 20–1000+ are routine; the practical limit is usually winding space, not manufacturability.
Operating frequency, rated RMS current, temperature-rise target, available slot or bobbin space, enamel thermal class, and any preference for serving or self-bonding. Send it on the litz wire page and the strand calculation with a quotation usually comes back within 48 hours.
Custom litz wire from drawings or physical samples, trial quantities welcome, strand calculation free