How to Read a MOSFET Datasheet Without Being Misled

Key points

• Every value is conditional — the condition matters more than the value
• RDS(on) rises 40–70% from 25 °C to 125 °C, so the 25 °C number is a marketing number
• A 'max' limit is tested at 100%; a 'typ' curve is not a guarantee
• The switching curves are usually taken at one bus voltage and one gate resistor — yours will differ

The datasheet is a contract written in fine print

Every number on a MOSFET datasheet is a measurement made under specific conditions, and the conditions are printed next to the number in a smaller font than the number itself. Most bad component decisions come from reading the large font.

Here is a worked example. A device is described as “5.5 mΩ, 100 V, 80 A”. Read carefully:

What the headline says	What the condition says	What it means for you
RDS(on) 5.5 mΩ	at VGS = 10 V, ID = 40 A, TC = 25 °C	At 125 °C it is nearer 8 mΩ, and at VGS = 4.5 V it may be 40% higher again
ID 80 A	at TC = 25 °C, silicon limit	At TC = 100 °C the continuous rating is typically 50–60 A
VDSS 100 V	at ID = 250 µA, 25 °C	Genuine — breakdown rises slightly at cold, so this one is safe
PD 150 W	at TC = 25 °C	Meaningless without a heatsink held at 25 °C, which no real heatsink is

The device is not misrepresented. The headline is simply the best case, and the design must be done at the worst case.

Six numbers that decide the design

1. RDS(on) at your gate voltage and at temperature. Look for a table row that gives both a typical and a maximum at 125 °C. If only a 25 °C typical exists, assume 1.5× for your loss calculation and verify on samples.

2. Total gate charge Qg at your drive voltage. Not Qgs, not Qgd alone — Qg sets drive loss and the switching speed your driver can achieve.

3. Coss and Eoss at your bus voltage. Capacitance falls steeply with voltage, so the value at 25 V tells you little about a 400 V design. Eoss is the loss-relevant integrated figure.

4. VGS(th) range, not typical. Threshold spread across a production distribution is wide; a part with 1.2 V typical may reach 2.5 V maximum, which matters for logic-level drive and for shoot-through margin in a bridge.

5. Thermal resistance RθJC and RθJA. These bound the thermal problem. RθJA on a surface-mount part is measured on a specific test board and is optimistic for your board unless your copper matches theirs.

6. SOA or avalanche rating, if your circuit has inductive transients. If neither is published, the part has no stated capability for those events.

Four ways a datasheet flatters a part

These are not fabrications — they are legitimate measurement choices that make comparison harder than it looks.

Test conditions that flatter

RDS(on) quoted at VGS = 10 V and ID = 20 A looks excellent. If your design drives the gate at 4.5 V from a logic-level source, you are not buying the number you read. Always match the gate voltage first, then compare.

Temperature that flatters

Room-temperature values are the default across the industry, and they are typically 40–70% optimistic on resistance. A part that looks 20% better than a competitor at 25 °C is often identical at 125 °C, because the temperature coefficient differs more than the room-temperature value.

Sampling that flatters

Datasheets distinguish between parameters tested at 100%, parameters sampled per lot, and parameters characterised during development only. Reverse recovery time, for instance, is almost never 100% tested — it needs a defined di/dt and costs throughput. If your design depends on a tight Trr window, ask what the production distribution looks like, not what the typical curve shows.

Axes that flatter

Logarithmic axes are legitimate. A curve that plots RDS(on) from 1 mΩ to 10 Ω compresses the region you care about. Before comparing two curves, check the axis ranges and the test conditions printed in the caption.

Reading the switching behaviour section

The switching curves in a good datasheet include gate charge, output capacitance, transfer characteristics and — most usefully — switching time or energy against gate resistance.

What most designers want from this section is the answer to one question: at my bus voltage and current, how much energy is lost per switching event? A well-made datasheet gives Eon and Eoff curves, or enough information to derive them. A catalogue datasheet gives a single “td(on), tr, td(off), tf” table at one condition, which is nearly useless for loss calculation.

If your supplier can send the characterisation report behind the curves — test board, loop inductance, gate resistor values — take it. It tells you what the part does in a circuit that resembles yours rather than in a fixture.

What to ask a supplier before you commit

Four questions separate a catalogue supplier from one who will support a design:

1. Is RDS(on) max specified at 125 °C, and at what gate voltage?
2. Is Trr, and at what junction temperature?
3. Which parameters are 100% tested versus sampled?
4. Can you send the reliability report for this specific part number, with sample size and conditions?

A supplier who answers all four in writing is worth a second-source discussion. One who answers “it is in the datasheet” when the datasheet contains a typical curve is telling you the limits are not guaranteed.

Our low voltage MOSFET and high voltage MOSFET pages list the ordering codes and the conditions behind each headline value; the glossary explains each parameter with its practical consequence, and the reliability testing page describes what our reports contain.

This article describes general manufacturing practice for machined and fabricated components. It is not a substitute for a drawing-specific engineering review — tolerances, material selection and process capability depend on the individual part.