Where the 10A Number Comes From

Open any SOP-8 power MOSFET datasheet and the headline current sits on the first page: I_D = 10A, 25A, even 40A. The fine print says what the number means: at T_A = 25°C, with ideal copper.

Take the NCE0110AS, a 100V/10A SOP-8 part we stock: 17mΩ max at 10V, 70A pulse rating — a genuinely good die. Its package power rating is 3.1W, and its junction-to-ambient resistance is 40°C/W.

Those two numbers are the real datasheet. The 10A headline is a marketing number wearing a measurement condition.

But why do manufacturers publish it that way? Because every competitor does, and because the rating is technically true — at 25°C ambient, on an infinite copper plane, for a single pulse. The condition is the fine print.

The Math That Kills the Dream

Conduction loss is I²R. At 10A through 17mΩ, that's 1.7W of heat sitting inside an 8-pin package.

Multiply by the thermal resistance: 1.7W × 40°C/W = 68°C of junction rise over ambient — before you add the board's own heating, before the enclosure, before summer.

Now check the datasheet's own derating: the NCE0110AS lists 7A at 100°C. The manufacturer is telling you the same thing from the other direction — sustained current, derated.

The same story plays out across the whole class. Diodes' DMT10H009LSS (100V/10A, SO-8) rates 2.5W; Vishay's SI4058DY rates 5.6W with the same 10A headline. Two "10A" parts, double the thermal reality — because package, not die, is what you're actually comparing.

What the Package Can Actually Shed

The SOP-8 has no exposed tab. Its only thermal path is through the drain pins — four thin leads soldered to copper that may or may not exist.

Compare that to a TO-252 (DPAK): the same class of part carries the same die on a large exposed drain tab with a junction-to-case resistance of ~1–2°C/W instead of 40°C/W to air.

From what we see across Shenzhen lots (2025–2026), the boards that fail with SOP-8 parts aren't the ones running 10A — they're the ones running 8A with 1 oz copper and no drain-pin pour. The copper is the hidden half of the rating.

Two ounces of pour under the drain pins, plus a few vias to an inner plane, is the difference between 6A and 9A of real capacity on the same part. The forums converge on this, and so do our failure reviews.

How to Read a MOSFET Datasheet Like a Heat Engineer

Skip the first page. Go to the absolute-maximum ratings table and find three numbers: P_D, the thermal resistance, and the 100°C derated current.

P_D tells you the package's wall. The thermal resistance tells you how far from the wall you are at your current. The 100°C derating tells you what the manufacturer thinks is sustainable.

All three numbers are on every power MOSFET datasheet — and all three are routinely ignored in favor of the headline current.

When the Nameplate Current Is Real

The 10A rating isn't a lie — it's a burst rating with a polite costume. It's real for: inrush at power-up, stall events, hot-plug transients, and any load that returns to idle before the junction catches up.

The 70A pulse rating on the NCE0110AS (160A on the bigger TO-252 parts) is for microsecond-scale events: a capacitor bank charging, a motor starting, a changeover glitch. The die absorbs the energy, then cools back down.

Here's the thing: a load that draws 10A for an hour is not a burst. It's a continuous design, and continuous designs need the package to match.

When to Stop Using SOP-8

Use SOP-8 when: the load stays under ~7A continuous, the board has 2 oz pours under the drain pins, and you need the small footprint on a dense board.

Don't use SOP-8 when: the load is 8A+ sustained, the enclosure runs hot, the board is two-layer with no copper budget — or the part keeps coming back as a burned joint (that's not bad luck, it's the package ceiling).

Frequently Asked Questions

Q1: Why does my "10A" MOSFET overheat at 8A?

A: Because 10A is rated at 25°C with ideal copper, and your board is neither. At 17mΩ, 8A is 1.09W — a 44°C junction rise at 40°C/W before the enclosure adds its own heat. The fix is copper under the drain pins, a lower R_DS(on) part, or a TO-252 package — in that order of effort.

Q2: How do I calculate the real current my SOP-8 part can carry?

A: Work backward from the thermal budget. Pick a target junction temperature (125°C is safe), subtract the ambient, divide by the R_θJA to get allowable watts, then solve I = √(P/R_DS(on)). For the NCE0110AS at 60°C ambient: (125-60)/40 = 1.6W → √(1.6/0.017) ≈ 9.7A — but that assumes the datasheet R_θJA on your actual board, which is optimistic by default.

Q3: Does pouring copper really help?

A: Yes — it's the largest lever you have on an SOP-8. The drain pins are the only thermal path; 2 oz pours under them plus vias to an inner plane can drop the effective junction-to-ambient resistance by a third or more. The same part on a bare 1 oz footprint runs meaningfully hotter than the datasheet curve predicts.

Q4: How much better is a TO-252 for the same current?

A: A different thermal world. The NCE0110AS (SOP-8) sits at 40°C/W junction-to-ambient with 3.1W of package rating; the NCE0125AK (TO-252) has ~2.1°C/W junction-to-case and 70W. The same 10A that means a 68°C rise in SOP-8 is a mild tab temperature on the TO-252.

Q5: So is the 10A rating a lie?

A: No — it's a conditional promise, and the condition is clearly stated. At 25°C, on ideal copper, the part does carry 10A. The rating is real for bursts, inrush, and short excursions; the mistake is reading it as a continuous capability. Every honest MOSFET datasheet includes the derating table that says the same thing we're saying.

Q6: What should I use for 8–15A continuous?

A: A TO-252 part from the same family. For 8–10A, the NCE0115K (15A, 50W); for 10–20A, the NCE0125AK (25A, 70W, 3V drive); for true power stages, the NCE0140KA (40A, 140W, 520mJ avalanche). All three share the 100V class — the package is the upgrade.