The 480mJ That Never Died

We pulled the official NCE0240 PDF from ncepower.com the same week we checked the distributor pages. Several distributor pages and the LMO240 clone listing still carry the pre-2026 figures — 480mJ, 220W, 0.68°C/W. The revision line on the official v2.0 says it plainly: “Update the parameters and curves,” June 2026.

The Leiditech LMO240, a same-parameter clone of the NCE0240, lists the old numbers too. So does a chunk of the stock floating around the market.

How does that happen? Aggregator pages are written once and never revisited. A distributor uploads the spec in 2022, the manufacturer improves the measurement and issues v2.0 in 2026, and the page stays frozen at the old revision — forever.

Nobody is lying. The pages just stopped being true.

The same part, two ways to quote it:

Which one is “better” is not even the right question. 100mJ at V_DD 50V with L 0.5mH is a stated, testable condition; 480mJ with no condition you can find is a number looking for a home.

Every Amp Sticker Is a Cooling Promise

The 200V class taught us the same lesson three times in one week, because NCE's 200V family ships in three packages with three very different stories:

• NCE0203S — SOP-8, 3.9A sticker, 3W budget, no exposed pad.
• NCE0208KA — TO-252, 8A sticker, 78W case rating.
• NCE0240 — TO-220, 40A sticker, 283W case rating.

Run the conduction math at the sticker and the truth shows up fast:

The 40A sticker is real — if you bolt the tab to something that keeps the case below 140°C. On a bare 1-inch pad, the 40°C/W junction-to-ambient path caps the NCE0240 around 9.5A before the 175°C junction ceiling arrives.

Forum veterans said it for years and it's still true: front-page drain-current numbers are “wildly optimistic” compared with what a board can actually pull. The sticker is a cooling promise with an unstated heatsink assumption.

The Voltage Tax Is Real — and Die Generations Keep Beating It

200V silicon costs about 2.7× the on-resistance of 100V silicon at the same die generation. The NCE0110AK (100V, TO-252) is 110mΩ; the NCE0208KA (200V, TO-252) is 300mΩ. Same footprint, same package, double the voltage rating, nearly three times the loss.

So why would anyone buy 200V? Because a rail that rings past 100V kills the 100V part on the second power cycle — and a dead switch costs more than a few milliohms.

Then the family does something that breaks the clean story: the SOP-8 NCE0203S hits 79mΩ at 200V — lower than the TO-252 NCE0208KA's 300mΩ. Die generation beats the voltage tax, because a newer, denser cell design shrinks the penalty that older silicon pays.

The lesson for a BOM owner: don't assume the voltage class alone tells you the loss. Same class, three dies, three resistance worlds — check the actual row on the actual sheet.

No Row Means No Budget

Here's the quietest trap in the 200V family: two of the three parts publish no avalanche row at all. No E_AS number, no 100% UIS claim — just a feature bullet on one of them saying avalanche is “fully characterized.”

Characterized is not a budget. If a datasheet publishes no E_AS value, you cannot design unclamped inductive edges against that part. The die may survive them; the sheet gives you no number to count on.

The flagship NCE0240 is the one that publishes: E_AS 100mJ with its test conditions, a 42ns body-diode recovery row, and a 175°C junction rating. That is the difference between a part you can engineer with and a part you have to hope about.

Same for body-diode recovery — the two smaller parts publish no t_rr. When the design needs that timing number, only the flagship has it.

10V Gate Drive Is a Feature, Not a Limitation

All three 200V parts publish their R_DS(on) at 10V only. None has a 4.5V row. To a 3.3V-era designer that reads as a limitation — and for one of them, the threshold makes it genuinely interesting.

The NCE0208KA's threshold is 1.0–2.5V — low enough to start conducting early and low enough to be dangerous in a half-bridge, where dv/dt through C_gd can lift the gate past threshold and cause shoot-through.

The NCE0203S and NCE0240 sit at 2–4V thresholds with a 10V drive rule. Treat 10V drive as part of the specification, like the voltage rating itself.

A half-enhanced channel at 5V runs hotter and slower than the sheet promises — and against a 3W budget, that's a fast way to a field failure.

The Three-Check Habit

What does the habit buy you? One less field failure, one less wrong BOM line, one less argument with a supplier about whose number is right.

We run the same checks on every lot we stock — the official PDF revision, the conditions behind the numbers, and the test at the published R_DS(on) condition — because in Shenzhen distribution, the stale sheet and the clone die are both part of the everyday market.

The 200V family sits on one shelf: NCE0203S in SOP-8, NCE0208KA in TO-252, NCE0240 in TO-220 — genuine NCE, current revisions, batch-tested. Same-day shipping from Shenzhen, and honest answers about which of the three your rail actually needs.