NCE0102 vs NCE0103Y vs NCE0110AK: Side-by-Side Comparison

Sticker current vs the honest board-level continuous line — where the 0103Y's name oversells:

Honest continuous on a real board (R_θJA with typical copper):

What does the honest ladder look like, then? The 0103Y's 3A sticker is 50% higher than the 0102's 2A — but its honest continuous line isn't any better.

The 0102 burns 0.92W at its own rating (26% margin under its 1.25W wall); the 0103Y burns 1.44W at its rating (4% margin under a 1.5W wall).

The AK is the only one whose sticker points at a real ladder rung.

Key Differences

1. The package is the real spec — SOT-23 sheds heat from the board, TO-252 from a tab

Why does a 40W case rating live on the same die class as a 1.25W SOT-23 rating? Because the TO-252 gives heat a second exit: R_θJC 3.8°C/W through a tab that is the drain.

Both SOT-23 parts publish R_θJA 100°C/W and no case path — every watt must leave through the leads and the copper around them.

The AK's 10A is a design conversation; the SOT-23 stickers are a 25°C arithmetic trick.

2. The sticker ladder and the honest ladder disagree — and that is the buying lesson

By name, the ladder reads 2A → 3A → 10A. By honest continuous current on a typical board it reads ~1A → ~1A → 3–5A.

The 0103Y is not a step up from the 0102 for continuous duty — it's a pulse part (20A I_DM) wearing a continuous-current costume. The AK is the first real step.

3. Datasheet maturity mirrors market positioning

NCE0102 ships as v4.2 with E_AS 2.45mJ published and a 150°C junction.

NCE0103Y ships as v1.0 with no E_AS row and no 100°C current row; NCE0110AK ships as v3.0 with everything — E_AS 20mJ at 100% UIS, trr, Qrr.

When a datasheet is silent, treat the budget as unquantified — not as free.

4. Gate drive: all three are logic-friendly, with different guarantees

Thresholds are 1.2–2.5V (0102) versus 1.0–2.0V (0103Y and AK) — all conduct from 3.3V, none guarantee R_DS(on) below 4.5V.

The honest rows: 0102 guarantees at 2A test current, 0103Y at 3A, the AK at 8A for its 4.5V row.

Q_g climbs 12 → 20 → 21.8nC — still light enough for a plain MCU pin at modest frequency.

When to Choose Each Part

✅ Choose NCE0102 when the load is under ~1A and the gate budget is small. Lowest gate charge (12nC), a published E_AS (2.45mJ), and the most honest small-die pricing. 48V fans, relay coils, logic-to-load interfaces at modest duty.

✅ Choose NCE0103Y when the load pulses. The 20A I_DM rating and low 1.0–2.0V threshold are the reasons to buy it — motor inrush, solenoid bangs, hot-plug events. Don't buy it for a continuous 3A it can't deliver; at steady ~1A the 0102 is cheaper.

✅ Choose NCE0110AK when current is real and continuous. 2–5A on 24–48V rails with a copper pour under the tab — and when you want a spec sheet that publishes its avalanche and diode data. This is the family member that survives contact with an actual load.

❌ Skip all three when your rail is under ~30V and current is the problem. A 30V SOP-8 part at 17mΩ (NCE0110AS) or lower carries more current at less loss — the 100V rating is only worth its price when the rail or the ringing demands it.

Two heat philosophies: the SOT-23 parts dump heat sideways through leads and hope; the TO-252 drops it through a tab into copper you control. That single difference is why the AK's 10A sticker is a ladder rung and the SOT-23 stickers are arithmetic.

Frequently Asked Questions

Q1: Which of the three should I pick for my design?

A: Match the package to the duty, not the sticker. Sub-1A continuous on a 24–48V rail — NCE0102. Pulsed or inrush-heavy loads — NCE0103Y (20A I_DM). Real continuous current of 2–5A with copper to spare — NCE0110AK. If the rail is under ~30V, none of them: a 30V low-R_DS(on) part is the honest answer.

Q2: Isn't the 3A NCE0103Y better than the 2A NCE0102?

A: For pulses, yes — for continuous duty, no. At its own rating the 0103Y burns 1.44W against a 1.5W wall (96%), while the 0102 burns 0.92W against 1.25W (74%). Both honest continuous lines sit near ~1A on a typical board. Buy the 0103Y for the 20A pulse and lower threshold, not for a 3A it cannot deliver.

Q3: How much more current can the TO-252 actually carry?

A: Roughly 3–5A continuous versus ~1A for the SOT-23 pair — with the right pad. The AK's 10A rating needs a case at or below 132°C, which requires the tab soldered to a copper pour with thermal vias. Engineers de-rate up to 40% on minimum footprints; board-level thermal models suggest roughly 50°C/W on a 10cm² double-sided pour.

Q4: Can all three be driven from 3.3V or 5V logic?

A: They conduct from 3.3V; none guarantee R_DS(on) below 4.5V. Thresholds are 1.2–2.5V (0102) and 1.0–2.0V (0103Y, AK). For light loads 3.3V works; for guaranteed low resistance, drive at 4.5V or 10V — and remember the AK's 4.5V row is tested at 8A, not 10A.

Q5: The datasheets differ in what they publish. Does it matter?

A: Yes — silence is not a free budget. The 0102 publishes E_AS 2.45mJ; the 0103Y publishes no avalanche value and no 100°C current row; the AK publishes E_AS 20mJ at 100% UIS plus body-diode trr and Qrr. If your design touches inductive energy or wants SR-adjacent duty, the AK's v3.0 completeness is a real buying reason.

Q6: Are these parts pin-compatible with each other?

A: No — the SOT-23 pair and the TO-252 have different pin maps. The SOT-23 parts are 1 = gate, 2 = source, 3 = drain. The TO-252 is 1 = gate, 2 = drain (wide lead), 3 = source, with the tab also drain — and that tab is electrically live, so a layout change is mandatory, never a drop-in swap.

Q7: Which one should I stock for repairs on 48V boards?

A: The NCE0110AK covers the most boards, with the 0102 as the small-signal companion. TO-252 is the workhorse power footprint of 48V telecom and POE gear; the 0102 handles the SOT-23 positions that die from over-voltage. Together they cover most 100V failure positions a repair bench sees.

Q8: Why does the family use the same “0110” name for different dies?

A: NCE suffixes carry package and variant, not die identity. The NCE0110AS (SOP-8, 17mΩ) and NCE0110AK (TO-252, 110mΩ) share the prefix and the 10A sticker but are different products — 6.5× apart in R_DS(on). Always verify the suffix and the R_DS(on) row against your layout before a BOM swap.