NCE30H15K vs HYG024N03LR1D vs IRLR7843: Full Comparison

The three datasheets behind it are the NCE30H15K at v1.0, the HUAYI HYG024N03LR1D at V1.0 (2019) and the Infineon IRLR7843 (document PD-94638A).

Two rows in that table deserve a second look, and neither is the current rating. The avalanche rows disagree in kind, not just in size — two parts publish a test current and one does not. And the thermal rows disagree by a factor of nearly two and a half.

Start with the resistance, because that is the number people shop on:

On the row a 5V driver actually uses, the HUAYI part wins by 1.4mΩ. That is a real advantage and it is not the whole story — because the same current has to leave through a thermal path that is more than twice as bad.

Run the two together at 20A and the ordering flips: 3.74°C of junction rise for the HYG part against 2.30°C for the NCE part. Lower resistance, hotter die.

But why does the resistance ranking mislead?

Back-calculate the implied hot resistance for all three and the spread collapses — 5.9mΩ, 5.9mΩ and 5.6mΩ, from three different manufacturers and three different current ratings.

The cold resistance is what you buy. The hot resistance is what you get, and it converges.

When to Use HYG024N03LR1D

Use it when your gate driver is a real 4.5V or 5V rail, your load is clamped, and the duty cycle is low enough that 57W of dissipation is not the binding constraint. It is 100% avalanche tested and its body diode recovers in 23.5ns — the fastest of the three.

When to Use IRLR7843

Use it when thermals or current headroom decide the design: 140W of dissipation, 161A and 1.05°C/W. Its 34nC of gate charge at a 4.5V endpoint suits a weak or slow driver.

That endpoint also makes it the most directly comparable of the three if your driver never leaves 5V.

When NOT to Swap at All

Stay with the NCE30H15K when the load is inductive and unclamped. Nothing here matches 1700mJ at 58.5A of avalanche current, and that pairing is the reason this part gets designed in.

Why E_AS Alone Can't Pick a Replacement

An avalanche energy figure without its test current is half a specification. A 1700mJ rating means one thing at 58.5A and something else entirely at 12A.

How do you compare two numbers that were measured under different conditions?

The Infineon part sits furthest left, and that is the whole point. Its 1440mJ is 85% of the NCE part's energy, but it is published alongside a 12A avalanche current.

Avalanche capability falls as avalanche current rises. A large energy at a small current says little about surviving a 50A inductive dump.

The HYG part goes the other way: a modest 338.8mJ, but its 0.3mH test inductance implies a test current near 47.5A once you apply E_AS = ½LI². Different energy, comparable current, very different claim.

Pinout & Layout Notes

The NCE part and the HUAYI part publish the same package code, TO-252-2L, so the land pattern carries over unchanged. The Infineon part is the D-Pak outline, which shares the TO-252 footprint — verify the pin marking on the reel before committing a production run.

One layout rule applies to all three: the tab is the drain, not a ground. Solder it to a dedicated copper pad with thermal vias into an inner plane. Tie it to a ground plane and you short the supply through the device.

And the thermal numbers travel with the part, not the board. The same copper gives a 2.6°C/W part and a 1.05°C/W part very different junction temperatures at the same dissipation.

Other Alternatives Worth Knowing

NCE40H12K — same family, 40V, and a lower guaranteed 10V resistance than the 30H15K (4.0mΩ against 4.0mΩ at matching test current, but with a 4.5V row that reads 7.0mΩ). Reach for it when the rail is 24V, not when the driver is weak.

NCE30H12K — the family's 120A step. It publishes no 4.5V row at all, so it is a 10V-drive replacement only. Its avalanche budget is 350mJ against the 15K's 1700mJ.

NCE30H10K — the entry step, 100A and 5.5mΩ at 10V. Adequate for clamped loads on a real 10V rail, and the lowest-current option in the family.

NCE0115K — the voltage move rather than the current move: same TO-252-2L outline, 100V and 15A, for rails this family otherwise cannot reach.

Frequently Asked Questions

Q1: Can I drop a HYG024N03LR1D into a NCE30H15K socket?

A: Mechanically yes — both publish TO-252-2L. Electrically it depends on your load. The HYG part has lower resistance at both 10V and 4.5V, but also 2.6°C/W of thermal resistance against 1.15°C/W and a 57W dissipation ceiling against 130W. On a clamped, low-duty load it works; on a sustained 40A load the junction runs hotter.

Q2: Which of the three has the lowest conduction loss?

A: The HYG024N03LR1D at both 10V and 4.5V — on paper. 2.8mΩ and 3.6mΩ maximum against the NCE part's 4.0mΩ and 5.0mΩ. But conduction loss is only half the thermal problem: at 20A the HYG part's junction rises 3.74°C against the NCE part's 2.30°C, because R_θJC is more than twice as high.

Q3: Is 1440mJ of avalanche energy close to 1700mJ?

A: Not comparable as published. Infineon lists the 1440mJ alongside a 12A avalanche current; the NCE sheet reaches 1700mJ at 58.5A. Avalanche capability falls as avalanche current rises, so the energy figures describe very different tests. Read E_AS and its current as a pair, always.

Q4: My driver is 4.5V. Which one should I fit?

A: The HYG024N03LR1D or the IRLR7843, depending on duty cycle. Both publish 4.5V rows — 3.6mΩ and 4.0mΩ — against the NCE part's 5.0mΩ. If the load is continuous or the ambient is warm, the IRLR7843's 1.05°C/W path is the tiebreaker. If it is a clamped, intermittent load, the HYG part's lower resistance wins.

Q5: Why do all three show a 70% ratio between the two current rows?

A: Because it is the same thermal convention, not a shared design. Each sheet solves its own R_θJC between a 100°C case and a 175°C junction, then divides by its own hot resistance. Three manufacturers, three dice, one calculation — 105/150, 70/100 and 113/161 all land at 70%.

Q6: Does the lower-resistance part also run cooler?

A: No — and this is the trap in shopping by R_DS(on) alone. The HYG part's resistance is 28% lower at 4.5V, but its junction-to-case thermal resistance is 126% higher. The two effects do not cancel: the part with the better resistance row runs hotter at every current.

Q7: What is the implied hot resistance, and why does it matter?

A: It is the resistance a sheet's own 100°C current row implies, derived from its thermal resistance. All three land between 5.6 and 5.9mΩ despite cold values spread from 2.8 to 4.0mΩ. Design with the cold maximum and your loss estimate is optimistic by the multiplier in the comparison table.

Q8: When should I keep the NCE30H15K instead of replacing it?

A: When the load is inductive and unclamped, or when you need the 600A pulse rating. 1700mJ at a 58.5A avalanche current is the reason this part exists, and neither alternative matches that pairing. If the failure that brought you here was a FET dying on turn-off, the avalanche rows are the ones to compare — not the resistance rows.