NCE30H14K vs IRLR7843 vs NCE30H15K: Full Comparison

On the raw numbers the IRLR7843 looks strictly better — more current, more avalanche energy, a lower junction-to-case path and a published threshold coefficient. It is also a different device in the ways that matter.

Three of the rows above are measured under conditions that flatter one part or the other. The sections below take them one at a time, because that is where a swap actually goes wrong.

When to Use the IRLR7843

Take the IRLR7843 when you need more current in the same footprint and your driver can reach 4.5V cleanly. At 161A continuous and 113A at a 100°C case it carries more than this part does, and the 1.05°C/W path moves heat out of the die faster.

It also publishes the one number this family does not: a threshold temperature coefficient of −5.4 mV/°C. That lets you compute the gate margin at operating temperature instead of assuming the 25°C figure holds.

At a 125°C junction its 2.3V maximum threshold falls to about 1.76V. Run the same arithmetic on the NCE30H14K and you cannot — the coefficient is not on the sheet.

When to Use the NCE30H15K

Take the NCE30H15K when the load is inductive and unclamped. Its 1700mJ avalanche rating is more than four times this part's and its R_DS(on) maximum is 4.0mΩ against this part's 3.0mΩ.

You are trading conduction loss for avalanche headroom, and both sheets print the test inductance (0.5mH) so the trade is checkable. That makes the two NCE parts directly comparable with each other in a way neither is comparable with the Infineon device.

It is also the lower-risk swap. Same manufacturer, same footprint, same pinout, same test conditions — only the die changes.

When Not to Use This Class at All

None of these three is the right answer above roughly 100kHz. All of them carry between 50nC and 80nC of gate charge, and switching loss overtakes conduction loss well before the low on-resistance pays for itself.

Below about 5A, the low on-resistance is also wasted. A smaller die in the same footprint costs less and switches faster, and the copper you were going to give the tab is better spent elsewhere.

And if the rail is above 30V, none of this applies. Step up the voltage class rather than leaning on avalanche ratings to absorb the margin.

Pinout & Layout Notes

The pinout is identical across all three, so a swap needs no layout change. Pin 1 gate, pin 2 the drain tab, pin 3 source. What differs is the silkscreen.

Two layout notes follow from that. The tab carries drain current and the heat path, so the copper that spreads heat is also the switching node — keep the loop area small. And with 410pF to 430pF of reverse transfer capacitance on these parts, the gate return is the sensitive trace.

A gate trace that runs alongside the switch node couples the drain swing straight back into the gate. That is the mechanism the next section is about.

Avalanche Ratings: Why the Two Sheets Describe Different Events

The IRLR7843 publishes an avalanche energy and an avalanche current. The NCE30H14K publishes only the energy. That single missing number is what makes the two ratings impossible to compare.

A 1440mJ rating sounds like four times the headroom of a 400mJ rating. It is not, because the events are four hundred times apart in energy and forty times apart in inductance.

The NCE sheet prints the inductance, so you can derive the current. The Infineon sheet prints the current, so you can derive the inductance. Neither prints both — but only one of them leaves you unable to reconstruct the event.

The bar chart is misleading on purpose. Bar length says nothing about which part survives your load — the two NCE bars share a test inductance and can be read against each other, and the Infineon bar cannot be read against either.

Other Alternatives Worth Knowing

NCE30H12K — 120A with 79nC of gate charge and the same 1.25°C/W package path. If 140A is more die than the design needs, this is the nearest step down and changes the driver almost not at all.

NCE3095K — 95A at 4.4mΩ maximum and 38.4nC. Worth noting because it beats the 80A part in the same family on both resistance and gate charge, so the ladder is not a straight upgrade path.

NCE3080KA — 80A, and the only part in the family specified at a 5V drive rail rather than 4.5V. Its threshold window is the family's tightest at 0.7–1.4V, which helps when paralleling and hurts in a half-bridge.

NCE3050K — 50A at 32.3nC, and the only part in the family that publishes a dv/dt rating. If a logic pin has to drive the gate, this is where to look.

Frequently Asked Questions

Q1: Is the IRLR7843 a drop-in replacement for the NCE30H14K?

A: Electrically yes, on the footprint and pinout — mechanically no. Both are TO-252 with pin 1 gate, pin 2 drain tab and pin 3 source. The gate charge and the reverse transfer capacitance differ enough that the driver and the gate loop should be re-checked.

Q2: Which has lower on-resistance?

A: The NCE30H14K on the printed maximums — 3.0mΩ against 3.3mΩ. But the two were measured at different drain currents, 20A against 15A. Resistance rises with current, so the Infineon figure is the flattered one.

Q3: Is the IRLR7843's avalanche rating really better?

A: It is a different rating, not a larger one. At 1440mJ and a published 12A avalanche current the implied test inductance is around 20mH. The NCE30H14K's 400mJ at 0.5mH implies about 40A. Short and sharp against long and gentle.

Q4: Why does only one datasheet publish the threshold temperature coefficient?

A: It is optional, and it is the number that decides hot-state noise margin. The IRLR7843 gives −5.4 mV/°C, so its 2.3V maximum threshold falls to about 1.76V at 125°C. Without a coefficient you can only bound the NCE part at its 25°C figure.

Q5: Does the different switching test load matter?

A: Yes — the fall times are not comparable. The IRLR7843 is characterized into a clamped inductive load, the NCE30H14K into a resistive one. Different topology, different waveform, different number.

Q6: Which replacement should I keep in stock?

A: Depends on whether your loads are inductive. The NCE30H15K is the safer swap because it shares the test conditions and adds avalanche headroom. The IRLR7843 is the one to stock when you need current and a datasheet that lets you compute hot-state margins.

Q7: Can I parallel two NCE30H14K instead of moving up?

A: Yes, and the gate charge is the cost. Two devices means two gate networks and 160nC to drive per cycle. Give each its own gate resistor, keep drain and source paths symmetric, and couple them through shared copper.

Q8: What disqualifies this part class for my design?

A: Frequency, current or voltage outside the window. Above roughly 100kHz the gate charge dominates. Below about 5A the low on-resistance is wasted. Above 30V the class does not apply at all.

Q9: What should I compare when substituting a MOSFET?

A: Package and thermal path first, then ratings, then switching parameters. Check the package style before the electricals — a part that relies on a copper plane for cooling will not behave the same way in a different footprint. Then voltage, current, R_DS(on), threshold and capacitances.

Q10: How close is close enough on on-resistance?

A: Within about 20% counts as a fairly close match. The NCE30H14K at 3.0mΩ and the IRLR7843 at 3.3mΩ are 10% apart, well inside that band — though they were measured at different drain currents, so the real gap is wider than printed.