Semiconductors and Switching
Why MOSFETs Heat and Switch Slowly
Explore conduction losses, gate charge, the Miller region and body diode.
You are applying “Why MOSFETs Heat and Switch Slowly” to a real system and need the concept, assumptions and verification boundary to stay clear.
Follow power through a switching MOSFET: channel resistance creates conduction loss, overlapping voltage and current create transition loss, gate charge demands driver current, and the body diode carries important reverse-current and commutation stress.
- Select or check first
- Loss is an operating-point calculation. A data-sheet headline such as low on-resistance describes a specified test condition, not every circuit. Junction temperature, gate-to-source voltage, drain current, switching frequency, bus voltage, commutation path and parasitic inductance all change the result.
- Controlling assumption
- Circuit state, waveform, source impedance, ratings, temperature and protection topology match the equipment being analysed.
- Top risk
- Threshold voltage is not an on-state drive specification. Gate-threshold voltage is measured at a small stated drain current. It marks the onset of channel conduction; it does not promise the low R_DS(on) needed at load current.
- Proof
- Check the exact manufacturer data, units and limits, then verify the result by calculation, measurement or a controlled acceptance test.
Start with the basics Optional foundation before “Why MOSFETs Heat and Switch Slowly”
Electrical-engineering basics
Electrical behaviour becomes easier to reason about when every voltage has a reference and every current has a complete return path.
- Voltage
- Voltage is potential difference between two points. A voltage statement is incomplete unless both points or its reference are known.
- Current
- Current is charge flow through a path. It does not disappear after a component; in a complete circuit it returns to the source.
- Resistance and impedance
- Resistance opposes steady current. Impedance extends that idea to changing signals and can depend on frequency through capacitance and inductance.
- Power and heat
- Electrical power is the rate of energy transfer. Some becomes useful output and some becomes heat that must stay within component and enclosure limits.
- Rating versus operating point
- A rating is a limit or qualified condition, not a target. The operating point must remain inside all relevant voltage, current, power, temperature and timing limits.
How to use this article: Draw the source, load, reference and return path before calculating. Treat exposed or stored electrical energy as hazardous until isolation is proved.
On this page
A MOSFET may look like a nearly perfect switch, yet it generates heat both while fully on and while moving between on and off. The gate also needs charge to be moved every cycle, and an intrinsic body diode can become part of the power-current path. A reliable design therefore joins the device data, driver, load waveform, layout and cooling arrangement rather than choosing a transistor from current rating alone.
Three loss mechanisms belong to different parts of the cycle
Diagram description
The image separates current through on-resistance, simultaneous drain voltage and current during transitions, and charge moved by the gate driver. It notes that gate-drive energy is shared across the driver and gate path rather than all heating the die.
| Mechanism | First-order screen | What the screen omits |
|---|---|---|
| Channel conduction | P_cond ≈ I_rms²R_DS(on) for the conducting interval | Temperature-dependent resistance, duty cycle, current ripple and package interconnect |
| Turn-on and turn-off | E_sw = ∫v_DS(t)i_D(t)dt per transition | Nonlinear capacitance, diode recovery, stray inductance and waveform ringing |
| Gate drive | P_gate ≈ Q_GV_drivef_sw | Different turn-on/off rails, driver output loss and energy recovered or dissipated elsewhere |
| Body-diode conduction | Average of v_Fi_D over its conduction interval | Reverse recovery, temperature, dead time and commutation topology |
| Output capacitance | Energy associated with charging and discharging C_oss | C_oss is nonlinear; use the data-sheet E_oss/Q_oss information where supplied |
On-resistance converts current into heat
First-order channel conduction loss
P_cond ≈ I_rms² × R_DS(on)(T_J, V_GS) × D
Use RMS current over the relevant interval. D is shown explicitly only when I_rms was calculated for the on-interval rather than over the whole cycle. The resistance must be adjusted from its specified test value using the manufacturer's temperature and gate-drive information.
- P_cond
- average channel conduction loss, W
- I_rms
- root-mean-square drain current for the stated interval, A
- R_DS(on)
- drain-to-source on-resistance at the stated conditions, Ω
- T_J
- junction temperature, °C
- V_GS
- gate-to-source drive voltage, V
- D
- conducting fraction, dimensionless
If 8 A RMS flows for half the cycle and the hot on-resistance is 25 mΩ, the interval-based estimate is 8² × 0.025 × 0.5 = 0.80 W. Substituting a colder headline resistance would understate the heat.
The driver controls switching by moving charge
Diagram description
A qualitative turn-on plot shows gate voltage rising, pausing through a Miller region while drain voltage falls, and then rising again. The adjacent note emphasizes driver current.
Before the channel carries the load current, the driver charges the gate-source region. During the drain-voltage transition, gate current is diverted into the gate-drain capacitance; the gate voltage may change slowly through the Miller region while V_DS changes substantially. Afterward, more charge raises the gate to its final drive voltage. Turn-off reverses the sequence.
Charge and transition-time screen
t ≈ ΔQ / I_G; P_gate ≈ Q_G × V_drive × f_sw
The first relation is useful when driver current is reasonably controlled over a data-sheet charge segment. The second estimates energy drawn from a single-ended drive supply; it is not MOSFET-die loss. Q_G must come from conditions relevant to the intended drain voltage and current.
- t
- approximate duration of the charge segment, s
- ΔQ
- charge moved in that segment, C
- I_G
- average gate current in the segment, A
- Q_G
- total gate charge for the stated conditions, C
- V_drive
- gate-drive supply swing, V
- f_sw
- switching frequency, Hz
Moving 20 nC through an average 0.5 A path takes about 40 ns for that segment. At 100 kHz and a 10 V drive, 40 nC corresponds to about 0.04 W drawn per gate from the drive supply under the simplified model.
The transition loss is the area under instantaneous power
Measured switching energy
E_on = ∫_turn-on v_DS(t)i_D(t)dt; P_sw ≈ f_sw(E_on + E_off)
This definition remains valid when the edges are not straight. Manufacturer energy values apply only to their stated circuit, temperature, gate resistance, current and voltage, so use them as a starting point and verify representative hardware.
| Trace or condition | Question answered |
|---|---|
| V_GS at the device pins | Did the gate reach a guaranteed on state without exceeding its absolute limit? |
| V_DS and I_D during both edges | How large is their overlap and is there overshoot? |
| Switch-node ringing | Are layout inductance and commutation exciting unsafe or noisy transients? |
| Case or board temperature after equilibrium | Does the thermal path support the calculated loss? |
| Worst supply, load and ambient | Was the design tested at the condition that creates maximum combined stress? |
The body diode is a circuit element, not decorative symbol detail
Diagram description
The lower body diode carries the illustrated freewheel path during dead time. The diagram identifies its anode at source and cathode at drain for the shown conventional silicon N-channel MOSFET.
An inductive load current cannot change instantaneously. In a half-bridge dead-time interval it may commutate into a body diode, producing a diode drop and loss. When the opposite switch turns on, stored charge in a silicon body diode can create reverse-recovery current, extra turn-on loss and voltage stress. Some topologies intentionally minimize diode time by synchronous conduction; the necessary dead time must still prevent both switches conducting together.
| Item | Design use |
|---|---|
| V_DSS and transient behavior | Block the maximum bus plus credible overshoot with margin |
| R_DS(on) versus V_GS and temperature | Estimate hot conduction loss at the available drive |
| Gate charge and Q_GD | Select driver current and estimate transition timing |
| Body-diode V_SD and Q_rr | Estimate dead-time loss and hard-commutation stress |
| E_oss or Q_oss | Account for nonlinear output-capacitance energy |
| Safe operating area | Check linear-mode, pulse and fault trajectories—not only steady on-state |
| Thermal impedance and package limits | Relate pulse/average loss to junction temperature for the actual board or heatsink |
| Absolute maximum V_GS | Protect the thin gate oxide against ringing and bias error |
Thermal path screen
T_J ≈ T_reference + P_loss × θ_J-reference
Use only the thermal metric whose reference and mounting conditions match the design. Transient thermal impedance is needed for pulses; steady θ values do not describe every layout, interface or airflow condition.
- T_reference
- measured case, board or ambient reference temperature, °C
- P_loss
- average loss assigned to the device for the model, W
- θ_J-reference
- applicable junction-to-reference thermal resistance, °C/W
Why a larger MOSFET is not automatically the cooler choice
A larger die often lowers on-resistance but increases charge and capacitance. That can reduce conduction loss while increasing driver demand, transition time or output-capacitance energy. At low switching frequency the resistance benefit may dominate; at high frequency a device with somewhat higher resistance and lower charge may produce less total system loss. Compare at the same required voltage rating, thermal condition and gate drive.
The distinction between instantaneous power and accumulated energy from Voltage, Current, Power and Energy Are Different is the basis for every loss calculation here. Use the nonlinear operating-point cautions in Resistance, Nonlinear Loads and Heat when deciding which resistance or voltage value belongs in a screen.
One-minute questions
Questions to answer before you apply this lesson
What should I select or check first for Why MOSFETs Heat and Switch Slowly?
Loss is an operating-point calculation. A data-sheet headline such as low on-resistance describes a specified test condition, not every circuit. Junction temperature, gate-to-source voltage, drain current, switching frequency, bus voltage, commutation path and parasitic inductance all change the result.
Which assumption can change the conclusion?
Circuit state, waveform, source impedance, ratings, temperature and protection topology match the equipment being analysed.
What can go wrong if I copy the example directly?
Threshold voltage is not an on-state drive specification. Gate-threshold voltage is measured at a small stated drain current. It marks the onset of channel conduction; it does not promise the low R_DS(on) needed at load current.
How do I verify the result in the real system?
Check the exact manufacturer data, units and limits, then verify the result by calculation, measurement or a controlled acceptance test.
Sources and further reading
- Gate Drive for Power MOSFETs in Switching ApplicationsInfineon Technologies · 2026-10-13Manufacturer application note reviewed for gate charge, Miller behavior, driver impedance, switching waveforms, parasitic inductance and body-diode recovery.
- Fundamentals of MOSFET and IGBT Gate Driver CircuitsTexas Instruments · 2026-10-13Semiconductor-manufacturer application report reviewed for gate-drive energy, source/sink paths, ground-referenced and high-side drive, and layout-dependent behavior.
- CoolSiC MOSFET 650 V M1 Trench Power DeviceInfineon Technologies · 2026-10-13Product-family application note reviewed to confirm that gate-charge shape, plateau behavior and body-diode loss differ by device technology; its numeric comparisons are not generalized.