The 100°C Column Is a Calculation, Not a Measurement

Start with one row from the 30H12K datasheet: maximum power dissipation, 120W at a 25°C case. And one more: thermal resistance, junction to case, 1.25°C/W. Divide the second into the 150°C span between a 25°C case and the 175°C junction limit and you get exactly 120W.

Those two rows are the same statement written twice. That matters because the third row — 84A at a 100°C case — is built from the same arithmetic with one extra term.

Work it backwards. At a 100°C case there is (175 − 100) / 1.25 = 60W available. Divide that by the square of 84A and you get 8.5mΩ. That is the resistance the manufacturer expects the die to have when it is hot: 1.9 times the 25°C maximum of 4.5mΩ.

Now run the same back-calculation on all four parts:

Four different dice, four different thermal resistances, and the same haircut at the end:

But why would four separate designs land on the same ratio? Because it is a convention, not a coincidence. Every sheet takes its own thermal resistance, solves for the power available between a 100°C case and a 175°C junction, then divides by its own hot resistance.

The convention shows up outside this family too. Two parts from other manufacturers, in the same package and voltage class, publish the same shape: a 100A device rated 70A at a 100°C case, and a 161A device rated 113A. Three manufacturers, three dice, one calculation.

What All Four Parts Actually Share

All four parts share the same ±20V gate window and the same 175°C junction limit — and the same shared blind spot. Strip away the current ratings and they are closer than the numbers suggest:

That last row of the shared block is the one people trip over. None of the four publishes a steady-state junction-to-ambient resistance. Every thermal number in these sheets is junction-to-case, which means every thermal number starts at the solder joint and assumes you have already solved everything below it.

The gate charge row is where the family actually separates. Note the test conditions rather than the values: the four are specified at 30A, 20A, 30A and 20A of drain current.

The gate voltage endpoint is 10V on all four, so those numbers are broadly comparable — which is not something you can assume across manufacturers.

The Resistance Ladder Is a Die-Generation Ladder

On resistance, the ranking is not what a voltage-class intuition predicts. The 40V part guarantees the same 4.0mΩ maximum as the 30V 150A part, and both beat the 30V 120A part's 4.5mΩ. Voltage class is not the axis this family sorts on; die generation is.

The implied hot resistance from the previous table tells the same story from the other end:

The 15K is the outlier by a wide margin: its die sheds less resistance on the way to temperature than any sibling, including the 40V part.

That single number is why the 15K can publish the family's highest current at the family's best thermal resistance with the family's lowest gate charge.

The practical consequence: the cold maximum is what you buy, and the hot resistance is what you get. An estimate built on the 25°C row is optimistic by the multiplier above — between 1.5 and 2.1 times, depending on which part you picked.

Where the Avalanche Budget Actually Lives

The family's avalanche ratings are 350mJ, 350mJ, 1700mJ and 1080mJ. Read alone, those numbers suggest the 15K is five times more rugged than the two entry parts.

Read with their test conditions, it is more than that. The two entry parts are rated at L = 0.5mH. The 15K reaches its 1700mJ at L = 1mH and 58.5A, and the 40H12K reaches 1080mJ at 1mH and 46.5A.

Apply E_AS = ½LI² to the two parts that publish their avalanche current, and it reproduces the printed energies to within one percent.

That tells you the relation the manufacturer used, and lets you read the third part's test current off the same curve. From 350mJ at 0.5mH, the 30H12K's implied test current is about 37.4A.

So the 15K's advantage is not just a bigger number. It is a bigger number measured at a higher avalanche current, on twice the inductance — and avalanche capability falls as avalanche current rises, which makes that the harder test.

Sizing One of These for Your Own Board — Three Steps

How do you turn a 150A headline into a current you can design with?

Step one is choosing the row that matches your driver. On this family that decision removes half the options immediately: the 30H10K and 30H12K are 10V-drive parts, and only the 30H15K and 40H12K publish a 4.5V number.

Step two is the one people skip. R_θJC of 1.15°C/W tells you nothing about your board. It tells you the temperature difference between the die and the tab. Everything from the tab outward — solder, copper, vias, air — is a second thermal resistance that the datasheet does not supply.

Step three is arithmetic. Take the 30H15K. Its 100°C current row implies a hot resistance of about 5.9mΩ, and that implies a temperature slope of roughly 0.32% per degree from the 4.0mΩ cold maximum.

Put a 40°C/W board under it and a 25°C ambient around it, and the self-heating equation solves near 22A at a 125°C junction.

That is one seventh of the number on the front page. Not because anyone lied — because 150A is a case-temperature figure and 40°C/W is a board figure.

When to Use This Family — and When Not To

Reach for the 30H15K when the load is inductive and unclamped, when your driver is a real 4.5V, or when you want the family's best thermal path. It carries 1700mJ at the highest avalanche current of the four.

Reach for the 40H12K when the rail is 24V or above, or when you want the lowest guaranteed 10V resistance at the 120A class. It also publishes a 4.5V row, though at 7.0mΩ.

Reach for the 30H12K when the rail is 12V or 15V, the load is clamped, and the driver delivers a real 10V. It is the middle of the family and still the right answer for a lot of boards.

Reach for the 30H10K when 100A is enough and cost leads. It is the entry step, and for a clamped load on a 10V rail it does the job.

Look elsewhere when the rail exceeds 40V, the continuous current exceeds roughly 25A, or the design needs a steady-state thermal rating the sheet does not print. At that point the honest answer is a different device class, not a different suffix.

Internal Links

• NCE30H10K — 30V/100A TO-252, the family's entry step
• NCE30H12K — 30V/120A, and what its 84A row implies about hot resistance
• NCE30H15K — 30V/150A with the family's best thermal path and 1700mJ of avalanche
• NCE40H12K — 40V/120A, lower guaranteed resistance than either 30V part
• NCE30H15K pinout and layout — the gate loop and the thermal pad traps
• NCE30H15K replacements — comparing alternatives on the rows that decide

The Habit That Prevents All of This

One habit covers most of the surprises in this family. Before reading a current rating, find the case temperature it assumes. If the row says 25°C, you are reading a die capability. If it says 100°C, you are reading a heatsink requirement.

Then find the thermal resistance, and ask what it is measured between. Junction-to-case means the solder joint is the boundary of the promise. Everything past it is yours.

None of that makes these parts worse. A 30H15K at 20A on a decent board is a comfortable, rugged, cool-running switch. It just will not be a 150A one, and no TO-252 ever will be.

Frequently Asked Questions

Q1: Why do all four parts show a 70% ratio on the 100°C row?

A: Because it is the same thermal convention applied four times. Each sheet solves its own R_θJC between a 100°C case and a 175°C junction, then divides by its own hot resistance. Four dice, four thermal resistances, one calculation — and competitor parts show the same shape.

Q2: Should I design with the typical or the maximum resistance?

A: The maximum, adjusted for temperature. Typical values describe a median unit rather than the one on your board. On the 30H12K and 40H12K the typical figures actually rank the two parts in the opposite order to their guaranteed maxima — and only the maxima are promises.

Q3: Can any of these go on a 24V rail?

A: Only the 40H12K does it comfortably. A 24V rail is exactly 80% of a 30V rating, which is the conventional derating limit and leaves nothing for switching-node overshoot. The 40V part puts the same rail at 60% of its rating.

Q4: What is the difference between R_θJC and R_θJA?

A: Where the measurement stops. Junction-to-case ends at the tab, so it is a property of the part and the solder joint. Junction-to-ambient includes the board and the air, so it is a property of your layout. These sheets publish only the first, which is why the second is your job.

Q5: How do I compare avalanche ratings between parts?

A: Read the energy and the avalanche current together. Capability falls as avalanche current rises, so a large number at a small current is the weaker claim. The 15K's 1700mJ is quoted at 58.5A; a 1440mJ rating quoted at 12A describes a much gentler test.

Q6: When is the front-page current actually achievable?

A: When the case is held at 25°C. That means an infinite heatsink, which is a laboratory condition rather than a board condition. On real copper, expect a single-digit fraction of the headline figure for continuous operation.