As the saying goes, "An ounce of prevention is worth a pound of cure." For electronics engineers, selecting components at the start of a project is essential to ensuring success. Power MOSFETs are among the most commonly used components, but selecting the right MOSFET requires considering many factors, from choosing N- or P-channel and package type to voltage rating and on-resistance. The following summarizes 10 rules to help with MOSFET selection.
1. Step 1: P-channel or N-channel?
Power MOSFETs are available as N-channel and P-channel types. Choose between them based on the application. N-channel MOSFETs offer more device options and lower cost. P-channel MOSFETs are less common and typically more expensive. If the MOSFET source is not connected to the system reference ground, an N-channel MOSFET requires a floating gate drive, transformer drive, or bootstrap drive; the driver circuitry becomes more complex. A P-channel MOSFET can be driven directly, simplifying the drive.
Typical scenarios to consider:
- Laptops, desktops, servers: fan speed control for CPU and system cooling; printer paper-feed motor drives; small appliances such as vacuum cleaners, air purifiers, and fans all use bridge topologies where each bridge leg may use either P-channel or N-channel MOSFETs.
- 48 V telecom systems: hot-swap MOSFETs on the high side can be implemented with either P-channel or N-channel MOSFETs.
- Laptop input protection and load switches: using two back-to-back MOSFETs for reverse-blocking. N-channel solutions usually require a controller with an integrated charge pump, while P-channel devices can be driven directly in some architectures.
2. Step 2: Choose the package
After selecting channel type, choose the package. Key package selection considerations:
(1) Thermal rise and thermal design
Different package sizes have different thermal resistances and power dissipation. Consider system cooling conditions and ambient environment, such as forced air, heatsink shape and size constraints, and whether the enclosure is sealed. The goal is to select a MOSFET whose parameters and package allow acceptable temperature rise and system efficiency.
If other constraints prevent adequate cooling, parallel MOSFETs can be used to share dissipation, e.g., in PFC applications, electric vehicle motor controllers, or synchronous-rectified power stages. If paralleled devices are not an option, choose devices with better performance or larger/new packages, such as replacing TO-220 with TO-247 in some AC/DC supplies, or adopting DFN 8x8 packages in some power supplies.
(2) System size constraints
Some systems are constrained by PCB size and internal height. Telecom module power supplies may use DFN 5x6 or DFN 3x3 packages due to height limits. Ultra-thin AC/DC supplies or enclosures may not accommodate TO-247 height; PCB mounting or bending leads increases assembly complexity. For high-capacity battery protection boards with strict size constraints, chip-scale CSP packages are increasingly used to improve thermal performance while minimizing footprint.
(3) Manufacturing process
TO-220 is available as exposed-metal tab and fully plastic variants. The exposed-metal tab has lower thermal resistance and better cooling but requires insulated mounting hardware, increasing assembly complexity and cost. Fully plastic versions are easier to assemble but have higher thermal resistance. To reduce screw-based assembly steps, some systems now use clips to secure MOSFETs to heatsinks, and vendors have adapted packages accordingly to lower part height.
(4) Cost control
Through-hole parts were common historically, but rising labor costs have driven a shift to surface-mount packages. Although SMT soldering cost per part can be higher, automation lowers overall costs. Cost-sensitive applications such as desktop motherboards often use DPAK because of its low cost. Package selection should reflect company manufacturing style and product requirements.
3. Step 3: Choose BVDSS (drain-source breakdown voltage)
Selecting MOSFET voltage rating often seems straightforward because input voltages are usually fixed and product line parts have fixed ratings. For example, AC adapters and chargers with 90–265 VAC input often use 600 V or 650 V MOSFETs on the primary side; motherboard rails at 19 V often use 30 V MOSFETs.
Data sheets specify BVDSS under defined test conditions, and BVDSS has a positive temperature coefficient. Many sources warn that if transient VDS spikes exceed BVDSS, even for a few or tens of nanoseconds, the MOSFET may avalanche and be damaged.
Unlike bipolar transistors and IGBTs, power MOSFETs can tolerate avalanche energy, and major manufacturers commonly perform 100% avalanche energy testing on production lines. Avalanche typically occurs at approximately 1.2–1.3 times BVDSS and the event duration is often microseconds or milliseconds. Therefore, very short spikes of a few or tens of nanoseconds that are below the device avalanche voltage typically do not cause damage.
Why then is derating recommended, such as 5%, 10%, or even 20% below BVDSS? The reason is manufacturability and reliability in mass production. Real-world parameters vary, and multiple worst-case factors can combine, especially at high temperature where device characteristics shift. Derating and design margin reduce the risk of failures under extreme conditions.
4. Step 4: Choose VTH based on drive voltage
Different systems use different gate drive voltages. AC/DC supplies commonly use 12 V drive; laptop DC/DC converters may use 5 V drive. Select a MOSFET with an appropriate threshold voltage VTH for the available gate drive. VTH is specified under defined test conditions and has a negative temperature coefficient. Different VGS values correspond to different RDSON values. Consider temperature variation so the MOSFET is fully enhanced during conduction while avoiding mis-triggering from gate spikes or coupling during turn-off.
5. Step 5: Select RDSON, note: not by current alone
Engineers often focus on RDSON because it directly affects conduction loss: lower RDSON means lower conduction loss, higher efficiency, and lower temperature rise. Engineers may reuse parts from previous projects or component libraries without careful RDSON selection. If device temperature rise is too low, cost considerations might push selection of higher RDSON parts; conversely, if temperature rise is too high, choose lower RDSON devices or optimize gate drive and thermal design.
If starting a new project, use a power allocation method to choose RDSON:
- From input voltage range, output voltage/current, efficiency, switching frequency, and drive voltage, estimate system losses.
- Estimate stray losses in the power loop, static losses of non-power components, IC quiescent and drive losses. A rule of thumb is these account for about 10%–15% of total loss. If the power loop has current-sense resistors, include their loss. Subtract these estimated losses from total loss to get the remaining loss attributable to power devices and magnetic components.
- Allocate the remaining loss between power semiconductors and magnetic components. If uncertain, divide evenly among components, yielding per-MOSFET power dissipation.
- Split each MOSFET's dissipation between switching loss and conduction loss. If uncertain, assume a 50/50 split. From conduction loss and the RMS current through the MOSFET, compute the allowable RDSON at operating junction temperature. Data-sheet RDSON values are given at TJ = 25°C and have a positive temperature coefficient; convert the operating RDSON to the 25°C equivalent for selection.
- Use the 25°C RDSON target to select candidate MOSFET models, then adjust upward or downward based on actual device parameters.
Many reference designs incorrectly choose MOSFETs by data-sheet current ratings with margin. MOSFET current ratings are specified under TC = 25°C and do not account for switching losses, so selecting by current alone is not a reliable method. For designs requiring short-circuit or surge checks, verify pulse drain current and duration on the data sheet, but note this is separate from RDSON selection.
6. Step 6: Switching characteristics
Switching losses are determined by parameters such as Crss, Coss, Ciss, Qg, Qgd, and Qoss. Qg affects driver losses, which are dissipated in the driver IC rather than the MOSFET. After selecting a MOSFET based on RDSON, consult data-sheet switching parameters to calculate switching losses.
7. Step 7: Thermal design and verification
Using the MOSFET data sheet and anticipated operating conditions, calculate conduction and switching losses, then determine junction temperature from total power loss and ambient conditions. Verify that the junction temperature remains within design limits under worst-case conditions and adjust as needed.
If total loss exceeds the allocated budget, select a different MOSFET, review RDSON tradeoffs, or iterate on switching loss estimates. This process typically loops through steps 5 and 6 until a model that meets design requirements is found.
If no single device matches requirements, consider:
- Using multiple MOSFETs in parallel to share dissipation.
- Reallocating losses to transformers, inductors, or other components while ensuring their temperature rise remains acceptable.
- Changing thermal management or increasing heatsink size if system permits.
- Altering circuit topology or switching frequency, e.g., interleaved PFC, LLC, or other soft-switching approaches.
8. Step 8: Verify diode characteristics
In bridge topologies such as full-bridge, half-bridge, LLC, and the low-side of buck converters, body diode reverse-recovery can be an issue. Prefer MOSFETs with integrated fast-recovery diodes. If the device does not include a fast-recovery diode, check reverse-recovery parameters: Irrm, Qrr, trr, and the trr1/trr2 profile. For example, trr less than 250 ns is often desirable. These parameters influence turn-off voltage spikes, efficiency, and reliability. Poor reverse recovery can cause shoot-through during transitions to capacitive mode, particularly in LLC startup or short-circuit events, potentially damaging devices. If the controller provides capacitive-mode protection, reverse recovery may be less critical.
9. Step 9: Avalanche energy, UIS, and dv/dt
Refer to detailed references for avalanche energy and test conditions. Outside of flyback converters and some motor-drive applications, pure voltage-clamped avalanche events are less common. Often dynamic avalanche occurs due to a combination of diode reverse recovery, high dv/dt, overtemperature, and large currents, which can lead to destruction. Consider avalanche energy (EAS), unidirectional surge (UIS) capability, and dv/dt robustness in applications where these stresses may occur.
10. Step 10: Other parameters
Consider additional parameters such as internal RG value, behavior when used as a load switch or hot-swap device operating in the linear region, SOA characteristics, and EMI-related parameters.
