The 200V NCE family hides its structure in plain sight: the smallest package carries lower resistance than the middle one — 79mΩ in SOP-8 beats 300mΩ in TO-252 — and the real current lives on the TO-220's heatsink tab at 41mΩ.
One-line selection: tight board, sub-2.5A duty → NCE0203S. Pulse current with a tab to cool it → NCE0208KA. Real amps, counted avalanche, a bolted heatsink → NCE0240. And if your rail can survive 100V, none of these three should be on your BOM at all.
| Parameter | NCE0203S | NCE0208KA | NCE0240 |
|---|---|---|---|
| Package | SOP-8 (no pad) | TO-252-2L (DPAK) | TO-220-3L |
| Datasheet revision | v1.1 | V3.0 (2026.6) | V2.0 (2026.6) |
| VDS | 200V | 200V | 200V |
| ID sticker (25°C) | 3.9A | 8A | 40A |
| ID @ 100°C | 2.8A | 6.4A | 35A |
| IDM | 30A | 32A | 160A |
| RDS(on) @ 10V (test) | 79mΩ max (3.7A) | 300mΩ max (4.5A) | 41mΩ max (20A) |
| VGS(th) min / typ / max | 2.0 / 3.0 / 4.0V | 1.0 / 1.7 / 2.5V | 2.0 / 3.2 / 4.0V |
| Gate-drive guarantee | 10V only | 10V only | 10V only |
| Qg (VGS = 10V) | 38nC (100V, 2.2A) | 31.2nC (100V, 5A) | 139nC (100V, 20A) |
| Ciss (test point) | 4200pF (25V) | 1087pF (100V) | 5817pF (100V) |
| EAS avalanche | Not published | Not published | 100mJ, 100% UIS |
| Body-diode trr | Not published | Not published | 42ns (66nC Qrr) |
| PD (case 25°C) | 3W | 78W | 283W |
| RθJC | 41.7°C/W | 1.6°C/W | 0.53°C/W |
| TJ max | 150°C | 150°C | 175°C |
| Honest board-level current | ~2–2.5A | ~3A (8A needs case cooling) | ~9.5A bare board, 20A+ on a heatsink |
One reading of the table: the family is not three versions of one part. It is three dies in three thermal worlds, sharing only the 200V rating and the 10V drive rule.
The RDS(on) ladder — die generation first, package second:
Gate-charge ladder — the drive tax scales with the die:
What does the honest current ladder look like, then? It runs the opposite direction of the sticker ladder: ~2.5A, ~3A, then a jump to 9.5A bare-board or 20A+ bolted down.
Every one of the three stickers is a temperature promise with a different cooling assumption built in — 3W of budget in the no-pad SOP-8, a 78W case in the DPAK, and a 283W case that only exists once the TO-220 meets metal.
RθJC tells the story: 41.7°C/W with no exposed pad, 1.6°C/W through a DPAK tab, 0.53°C/W through a bolted TO-220 — an 80× range in the heat path. The die is only half the part; the other half is the metal you attach, or fail to attach.
The 0208KA's 300mΩ is an older die. The 0203S proves a newer 200V cell can hit 79mΩ in a tiny no-pad package, and the 0240's flagship die adds die size on top: 41mΩ. Don't read resistance as a ranking of packages — read it as a ranking of die generations.
Two of the three publish no avalanche row and no body-diode recovery row. The 0240 publishes both — E_AS 100mJ with conditions and 100% UIS testing, plus a 42ns trr, plus a 175°C junction rating. When a design decision needs numbers, only the flagship has all of them.
Gate charge goes 31.2 → 38 → 139nC, and threshold goes 1.7 → 3.0 → 3.2V typ. The 0208KA is the easiest to drive but its low threshold is the half-bridge liability; the 0240 needs a genuine gate driver — an MCU pin won't charge 139nC fast enough.
Three questions separate the family cleanly. If a 100V rail survives your worst case, the tax makes all three wrong; inside 200V, the heat path and the current decide.
How do you choose between three parts that share a voltage class? You don't compare the parts — you compare the jobs, and the heat each job can move.
Choose the NCE0203S when board space is the scarce resource. Crowded UPS aux boards, sub-2.5A 200V positions, and designs where a low 79mΩ die matters more than a tab you can't use anyway.
Choose the NCE0208KA when the job is pulses or a repair slot. The 78W case and the tab carry intermittent current a no-pad SOP-8 never could, and 200V TO-252 boards in the field recognize the footprint instantly.
Choose the NCE0240 when the design needs real amps or a counted avalanche budget. 20A+ continuous on a bolted heatsink, motor edges that avalanche, and body-diode timing you can price — the flagship is the only one that ships all the numbers.
Skip all three when the voltage class is wrong. A rail that can survive 100V should carry the NCE0110AK (110mΩ) or NCE0140KA (17mΩ, E_AS 520mJ) — same or better current at a fraction of the loss. The 200V family is a tax you pay only when ringing forces it.
A: Let the heat path decide. Board-tight and under ~2.5A — the NCE0203S. Pulses, a tab, or a repair slot — the NCE0208KA. Real amps, counted avalanche, or a heatsink you are willing to bolt on — the NCE0240. And check first whether a 100V part survives your rail, because then none of these belongs on the BOM.
A: Because RDS(on) is a die property, and these are three different dies. The SOP-8 NCE0203S (79mΩ) uses a newer high-density cell than the older TO-252 die in the NCE0208KA (300mΩ), and the TO-220 NCE0240 (41mΩ) adds die size on top. Package decides how much heat leaves; the die decides how much loss rides along.
A: About 2–2.5A, 3A, and then a jump. The 3.9A SOP-8 burns 1.2W of its 3W budget at the sticker; the 8A DPAK burns 19.2W and needs a case at or below ~119°C; the 40A TO-220 burns 65.6W and needs a case below ~140°C — bare board it is ~9.5A. Stickers are cooling promises, not wire ratings.
A: 31.2nC, 38nC, and 139nC — roughly a 4.5× spread. The NCE0208KA is easiest to drive, but its 1.7V typ threshold is the induced-turn-on liability in half-bridges. The NCE0240 needs a real driver delivering amps of peak current — an MCU pin charging 139nC in microseconds destroys the switching budget.
A: Only one of the three has a counted budget: the NCE0240 at E_AS 100mJ, 100% UIS tested, with stated conditions. The NCE0203S and NCE0208KA publish no avalanche row at all — clamp their inductive edges or pick the flagship. And if the rail can survive 100V, the NCE0140KA carries E_AS 520mJ, over five times the flagship's budget.
A: Three packages, three disciplines. The SOP-8 is pins 1–3 source, 4 gate, 5–8 drain with no pad — heat exits through lead pours. The TO-252 is 1 gate, 2 drain, 3 source with a live drain tab needing an isolated pour and vias. The TO-220 is 1 gate, 2 drain, 3 source with a live tab needing mica or ceramic isolation, a shoulder washer, and 0.6–0.9 N·m torque.
A: Not visually — but the markings and the clones need a check. Genuine parts mark “0203”, “NCE0208KA”, and “NCE0240”, and the packages cannot be confused. The VBsemi-compatible line (NCE0203S-VB, NCE0208KA-VB, NCE0240-VB) carries similar codes with different dies, and the 0240 market still floats pre-2026 stock quoting 480mJ. We batch-test at each part's published condition and match the sheet before shipping.
A: When the voltage class is wrong in either direction. Rail ringing that stays comfortably under ~100V belongs on the 100V line (NCE0110AK, NCE0140KA) at a fraction of the loss. Rails that need 150V-class margin sit on the NCE1505S/1540K/1540KA line. The 200V family exists for the middle ground where 100V dies and 150V is not enough — price the tax, then pay it only if the rail forces you.





