Keywords: PWM dead time, motor drive, shoot-through, dead time compensation
Keywords: PWM dead time, motor drive, shoot-through, dead time compensation
Introduction
In the world of motor drive design, where engineers obsess over MOSFET selection, gate driver specifications, and control algorithm optimization, one parameter often goes underappreciated until it causes a spectacular failure: PWM dead time. This seemingly minor setting — the brief delay between turning off one MOSFET and turning on its complementary partner — can mean the difference between a reliable, efficient motor drive and a smoking pile of silicon.
Dead time exists to prevent shoot-through, the catastrophic condition where both the high-side and low-side MOSFETs in a half-bridge conduct simultaneously, creating a direct short from the DC bus to ground. But dead time is not simply a safety margin to be maximized — excessive dead time degrades motor performance, increases torque ripple, introduces current distortion, and reduces efficiency. Finding the optimal dead time and implementing effective compensation strategies is one of the most nuanced aspects of motor drive design.
This article explores PWM dead time in comprehensive detail — from the physics of MOSFET switching that necessitates it, through the calculation of optimal dead time values, to advanced compensation algorithms that mitigate its negative effects. Whether you are designing a 1kW drone ESC or a 100kW electric vehicle inverter, understanding dead time is essential for achieving the performance and reliability your application demands.
![Image Placeholder 1: Waveform diagram showing MOSFET gate signals during dead time, with current flowing through body diode]
Why Dead Time Matters: The Physics of Shoot-Through
To understand why dead time is necessary, we must examine what happens during a MOSFET switching transition. Consider a half-bridge leg where the high-side MOSFET is on and the low-side is off. When the controller commands a transition (turn off high-side, turn on low-side), several things happen in sequence:
Turn-Off Delay (td_off)
When the gate drive voltage goes low, the MOSFET does not stop conducting immediately. The gate must discharge from its driven voltage (typically 10-12V) down to the Miller plateau voltage — the point where the drain current begins to fall. This discharge time depends on the gate drive sink current, the gate resistance, and the gate-to-drain (Miller) capacitance. For a typical power MOSFET with 50nC gate charge and a 2A gate driver, td_off is 20-50ns.
Current Fall Time (tf)
After the gate voltage reaches the Miller plateau, the drain current falls from its full value to zero. The fall time depends on the MOSFET's transconductance and the gate drive speed — typically 10-30ns for modern low-voltage MOSFETs and 50-100ns for high-voltage devices.
Turn-On Delay (td_on)
Meanwhile, when the complementary MOSFET is commanded to turn on, its gate must charge from 0V up to the threshold voltage and then to the Miller plateau before drain current can flow. This charging time is similar in magnitude to the turn-off delay — 20-50ns for low-voltage devices.
The Shoot-Through Window
If both MOSFETs are commanded simultaneously (one off, one on), there is a window where the turning-off MOSFET is still conducting while the turning-on MOSFET begins to conduct. This overlap — the shoot-through window — creates a low-impedance path from the DC bus to ground through both MOSFETs. The shoot-through current can reach hundreds of amps in nanoseconds, far exceeding the MOSFET's safe operating area.
The magnitude of the shoot-through current is limited only by the DC bus voltage and the combined on-resistance of both MOSFETs plus parasitic resistance. For a 48V bus with two 10mΩ MOSFETs, the theoretical shoot-through current is 2,400A — though in practice, parasitic inductance limits the rate of current rise. Even so, 100-500A shoot-through currents are common, and they dissipate enormous power: at 48V and 200A, the instantaneous power is 9.6kW, concentrated in two small silicon chips. Without dead time, repeated shoot-through events at 20kHz will destroy the MOSFETs within seconds.
Calculating Optimal Dead Time
The minimum dead time must exceed the worst-case switching time mismatch — the scenario where the turning-off MOSFET is slowest and the turning-on MOSFET is fastest. The formula is:
t_dead_min = td_off_max + tf_max - td_on_min
Where: - td_off_max = maximum turn-off delay (worst-case temperature, voltage, and current) - tf_max = maximum current fall time - td_on_min = minimum turn-on delay (best-case conditions)
However, several additional factors must be considered:
Temperature Variation
MOSFET switching times vary with temperature. At 125°C junction temperature, the threshold voltage decreases by approximately 1-1.5V, and the transconductance changes. The net effect is that turn-off delay typically increases by 20-40% at high temperature, while turn-on delay may decrease slightly. The dead time calculation must use the high-temperature turn-off delay and the low-temperature turn-on delay to ensure safety across the entire operating range.
Gate Drive Tolerance
Gate driver ICs have propagation delay tolerances — typically ±20-50ns. In a half-bridge driver, the high-side and low-side channels may have different delays, and this mismatch must be added to the dead time. Some gate drivers specify a "delay matching" parameter that quantifies the maximum mismatch between channels — this is typically 10-30ns.
Gate Resistor Tolerance
Gate resistors have ±1% to ±5% tolerance, which affects the gate charge/discharge rate and thus the switching times. For designs with external gate resistors, the tolerance contributes to switching time variation.
Miller Capacitance and dv/dt Induced Turn-On
When one MOSFET turns off, the rapid voltage transition (dv/dt) at the switch node couples through the Miller capacitance (Cgd) into the gate of the complementary MOSFET. If the gate is not held firmly off (low impedance), this dv/dt can raise the gate voltage above the threshold, causing a brief unintended turn-on. This "Miller turn-on" or "Cdv/dt turn-on" is another form of shoot-through that dead time helps prevent — by keeping the complementary MOSFET's gate actively held low during the transition.
Practical Dead Time Values
Based on these considerations, typical dead time values are:
| Application | Bus Voltage | MOSFET Type | Typical Dead Time |
|---|---|---|---|
| Low-voltage BLDC (12-24V) | 12-24V | 30-60V, low Qg | 200-400ns |
| Medium-voltage (48V) | 48-54V | 60-100V | 300-500ns |
| High-voltage (e-bike, industrial) | 300-400V | 600V, superjunction | 500-1000ns |
| Very high voltage (EV inverter) | 400-800V | 650-1200V SiC | 200-500ns |
Silicon Carbide (SiC) MOSFETs deserve special mention: despite operating at much higher voltages, their switching times are significantly faster than silicon MOSFETs, allowing shorter dead times. However, SiC devices are more sensitive to shoot-through due to their higher current density, so the dead time optimization is more critical.
![Image Placeholder 2: Graph showing the relationship between dead time and total harmonic distortion (THD) in motor phase current]
The Consequences of Excessive Dead Time
While too little dead time causes shoot-through, too much dead time creates a different set of problems that are less dramatic but equally detrimental to motor drive performance.
Current Distortion and Torque Ripple
During dead time, both MOSFETs are off, and the motor current flows through one of the body diodes. The body diode has a forward voltage drop (Vf) of 0.7-1.2V — significantly higher than the Rds_on voltage drop (typically 0.1-0.3V). This extra voltage drop creates a voltage error in the applied motor voltage. The error is always in the same direction (it always opposes the desired voltage), creating a systematic distortion in the current waveform.
For a 48V bus with 500ns dead time at 20kHz, the voltage error is:
V_error = Vf_diode × (t_dead / T_pwm) = 1.0V × (500ns / 50μs) = 0.01V
This seems small, but it accumulates. The dead time distortion manifests as 5th and 7th harmonics in the motor current (for a fundamental at the electrical frequency). These harmonics produce torque ripple at 6× the fundamental frequency, causing vibration, acoustic noise, and reduced motor efficiency. In FOC systems, the distortion appears as a disturbance in the Iq and Id current loops, degrading the controller's ability to maintain smooth torque.
Efficiency Loss
During dead time, the current flows through the body diode, which has approximately 3-5× higher conduction loss than the MOSFET channel. For a motor drive operating at 20kHz with 500ns dead time, the dead time represents 1% of each switching period. While this seems negligible, the diode conduction loss during dead time can contribute 0.5-2% of total drive efficiency loss — significant in battery-powered applications.
Low-Speed Performance Degradation
At low motor speeds, the electrical frequency is low, and the voltage command is small. In this regime, the dead time voltage error represents a larger fraction of the commanded voltage. At very low speeds, the dead time distortion can exceed the fundamental voltage, causing severe current distortion, torque pulsation, and audible noise. This is a well-known problem in FOC drives and is the primary motivation for dead time compensation.
Dead Time Compensation Algorithms
To mitigate the negative effects of dead time, motor drives implement dead time compensation — algorithms that modify the PWM duty cycle to cancel out the voltage error caused by dead time. Several approaches exist, ranging from simple to sophisticated.
Simple Voltage Compensation
The simplest compensation adds (or subtracts, depending on current direction) a fixed voltage offset to the commanded duty cycle to account for the dead time voltage error:
V_compensated = V_commanded + sign(I) × Vf_diode × (t_dead / T_pwm)
This approach is easy to implement and effectively cancels the average dead time error. However, it does not account for the current-dependent nature of the diode forward voltage (Vf varies with current) or the temperature dependence of the MOSFET switching times. The simple compensation is typically 70-80% effective at reducing dead time distortion.
Current-Dependent Compensation
A more advanced approach uses the measured phase current to calculate the actual diode forward voltage and the actual voltage error:
V_error = f(I) × t_dead / T_pwm
Where f(I) is a function that models the diode Vf as a function of current. This can be a lookup table or a polynomial approximation. The diode Vf follows the Shockley equation: Vf = n × Vt × ln(I/Is + 1), but in practice, a linearized model (Vf = Vf0 + R_diode × I) is sufficient.
This approach achieves 85-95% compensation accuracy and is the most commonly used method in commercial motor drives. It requires knowledge of the phase current magnitude, which is available in any FOC system.
Adaptive Dead Time Compensation
The most sophisticated compensation algorithms adaptively tune the compensation parameters based on observed motor response. These methods use observers or harmonic analysis to detect the residual dead time distortion and adjust the compensation in real time.
One popular approach uses a disturbance observer that estimates the total voltage error (including dead time and inverter nonlinearities) from the difference between the commanded and measured current. The observer typically uses a low-pass filter to extract the slow-varying voltage error and applies it as a feedforward correction to the PWM duty cycle.
Another approach analyzes the current in the synchronous (dq) reference frame. In this frame, the 5th and 7th harmonics caused by dead time appear as 6th harmonics. A resonant controller tuned to the 6th harmonic frequency can detect and cancel the dead time distortion without requiring an explicit model of the inverter nonlinearities.
Hardware-Based Active Dead Time Control
Some advanced gate driver ICs implement active dead time control, where the driver monitors the actual MOSFET drain voltage and adjusts the dead time in real time based on the measured switching transient. The ADuM4135 from Analog Devices and the UCC21750 from Texas Instruments are examples of isolated gate drivers with active dead time optimization.
These drivers detect when the MOSFET has actually turned off (by sensing the drain voltage transition) and immediately enable the complementary MOSFET — minimizing the dead time to the theoretical minimum on every switching cycle. This eliminates the need for software compensation and achieves near-zero dead time distortion.
Effect of Dead Time on Total Harmonic Distortion
The relationship between dead time and current THD is well-documented. For a typical 48V motor drive operating at 20kHz, the following measurements illustrate the impact:
| Dead Time (ns) | THD at Rated Speed (%) | THD at 10% Speed (%) | Efficiency Loss (%) |
|---|---|---|---|
| 200 | 2.1 | 8.5 | 0.3 |
| 400 | 3.5 | 15.2 | 0.8 |
| 600 | 5.2 | 24.7 | 1.5 |
| 800 | 7.1 | 38.3 | 2.4 |
| 1000 | 9.0 | 52.1 | 3.6 |
The data shows that THD increases approximately linearly with dead time at rated speed, but exponentially at low speeds. At 10% speed with 1000ns dead time, the THD exceeds 50% — the current waveform is more distortion than fundamental. This underscores the importance of dead time minimization and compensation for low-speed motor operation.
![Image Placeholder 3: FFT spectrum comparison showing 5th and 7th harmonic magnitudes for compensated vs uncompensated dead time]
Practical Dead Time Tuning
Setting the dead time on a real motor drive involves a systematic process:
Step 1: Calculate Minimum Dead Time
Using the MOSFET datasheet values and the gate driver specifications, calculate the theoretical minimum dead time:
t_dead = (td_off_max - td_on_min) + t_margin
Where t_margin accounts for gate driver delay mismatch, gate resistor tolerance, and PCB parasitic effects. A typical margin is 50-100ns.
Step 2: Verify with Double Pulse Testing
Before running the motor, verify the switching behavior using a double pulse test. This involves applying two gate pulses to the MOSFET and observing the switching transients with an oscilloscope. The test reveals the actual turn-on and turn-off delays under specific voltage and current conditions, allowing you to confirm that the calculated dead time is sufficient.
Step 3: Test at Extreme Conditions
Test the drive at the extremes of its operating range: - Maximum temperature (typically 85-105°C ambient for industrial drives) - Maximum bus voltage - Maximum load current - Minimum load current (light load can be worst-case for switching times due to reduced gate drive current)
Monitor the MOSFET gate signals and switch node voltage with an oscilloscope to confirm that no shoot-through occurs under any condition.
Step 4: Implement Compensation
With the hardware dead time set to the minimum safe value, implement software dead time compensation to minimize the current distortion. Use the current-dependent compensation method, with parameters tuned for your specific MOSFETs and gate drivers.
Step 5: Verify with FFT Analysis
Run the motor at various speeds and loads, and capture the phase current waveform. Perform an FFT analysis to measure the THD and the magnitude of the 5th and 7th harmonics. If the compensation is effective, these harmonics should be reduced by 80-90% compared to the uncompensated case.
Common Dead Time Pitfalls
Using the Datasheet Typical Value
MOSFET datasheets specify switching times with "typical" and "maximum" values. Never use the typical value for dead time calculation — always use the maximum turn-off delay and minimum turn-on delay. The typical values are measured under idealized conditions that may not reflect your circuit's actual operating environment.
Ignoring Temperature Effects
Switching times change significantly with temperature. A dead time that is safe at 25°C may be insufficient at 100°C. Always calculate dead time using the high-temperature switching specifications, or add a temperature derating factor of 1.3-1.5× to the room-temperature value.
Forgetting About the Bootstrap Circuit
In bootstrap-based high-side gate drives, the bootstrap capacitor voltage drops during the dead time (it must supply the gate drive current for the high-side MOSFET). If the dead time is very short and the high-side duty cycle is very high, the bootstrap capacitor may not have enough time to recharge. This is not directly a dead time issue, but it constrains the minimum dead time in bootstrap-based designs.
FAQ
What is shoot-through in a motor drive, and how does dead time prevent it?
Shoot-through occurs when both the high-side and low-side MOSFETs in a half-bridge are simultaneously on, creating a direct short circuit from the DC bus to ground. This happens because MOSFETs have finite turn-off times — after the gate signal goes low, the MOSFET continues to conduct for 20-100ns. Dead time inserts a deliberate gap (typically 200-1000ns) between turning off one MOSFET and turning on the complementary one, ensuring that the first MOSFET has fully stopped conducting before the second one begins.How do I know if my dead time is too short?
Symptoms of insufficient dead time include: unexpectedly high MOSFET temperature (due to shoot-through current spikes), audible buzzing or whining from the motor drive, current spikes visible on the switch node voltage waveform, and in severe cases, MOSFET failure. You can verify by probing the switch node voltage with an oscilloscope — during dead time, the switch node should briefly float (showing a body diode drop) before the complementary MOSFET turns on. If you see a brief voltage spike in the wrong direction, shoot-through is occurring.Can dead time be too long?
Yes. Excessive dead time forces current through the MOSFET body diodes for longer periods, increasing conduction losses (body diode Vf is 0.7-1.2V vs 0.1-0.3V for Rds_on). More importantly, the voltage error during dead time distorts the motor current waveform, introducing 5th and 7th harmonics that cause torque ripple, vibration, and acoustic noise. At low motor speeds, excessive dead time can make smooth operation impossible — the current distortion exceeds the fundamental current. Dead time should be minimized to the safe minimum and compensated in software.Does dead time need to be different for different MOSFETs?
Yes, the optimal dead time depends on the MOSFET's switching characteristics, which vary with the MOSFET's voltage rating, die size, and technology. Higher-voltage MOSFETs (600V superjunction) have longer switching times and need more dead time (500-1000ns) than low-voltage MOSFETs (30-100V, 200-400ns). SiC MOSFETs switch faster than silicon and can use shorter dead times, but they are more sensitive to shoot-through. Always calculate dead time based on the specific MOSFET datasheet values, not generic rules of thumb.What is active dead time compensation in gate driver ICs?
Active dead time compensation (also called predictive dead time or adaptive dead time) is a feature in advanced gate driver ICs that monitors the actual MOSFET drain voltage during switching and adjusts the dead time in real time. Instead of using a fixed dead time that must accommodate worst-case conditions, the driver detects when the MOSFET has actually turned off and immediately enables the complementary MOSFET. This minimizes dead time on every cycle, reducing both shoot-through risk and current distortion. Examples include the Analog Devices ADuM4135 and TI UCC21750.How does dead time affect FOC performance specifically?
Dead time creates a voltage error that appears as a disturbance in the FOC current loops. In the dq reference frame, this disturbance manifests primarily as a 6th harmonic (from the 5th and 7th harmonics in the abc frame). The PI controllers in the Iq and Id loops cannot fully reject this disturbance because it is at a frequency above the controller bandwidth. The result is torque ripple at 6× the electrical frequency, which causes speed ripple, vibration, and acoustic noise. Dead time compensation is essential for high-performance FOC drives to achieve smooth torque output, especially at low speeds.References
- Texas Instruments, "UCC21750 Isolated Gate Driver with Active Dead Time," Datasheet, https://www.ti.com/lit/ds/symlink/ucc21750.pdf
- Analog Devices, "ADuM4135 5 kV rms Isolated Gate Driver with Active Dead Time," https://www.analog.com/en/products/adum4135.html
- J. W. Choi and S. K. Sul, "A New Compensation Strategy Reducing Voltage/Current Distortion in PWM VSI Systems Operating with Low Output Voltages," IEEE Transactions on Industry Applications, vol. 31, no. 5, 1995. https://ieeexplore.ieee.org/document/469851
- N. Mohan, T. M. Undeland, and W. P. Robbins, "Power Electronics: Converters, Applications, and Design," 3rd Edition, Wiley, 2003.
- Infineon Technologies, "Dead Time Calculation for Motor Drive Applications — Application Note AN_201709_PL52_003," https://www.infineon.com/dgdl/Infineon-Dead_time_calculation-ApplicationNotes-v01_00-EN.pdf
- STMicroelectronics, "Dead Time in STM32 Advanced Timer for Motor Control — Application Note AN4790," https://www.st.com/resource/en/application_note/an4790-advancedtimer-deadtime-stmicroelectronics.pdf
Meta Description: Master PWM dead time setting in motor drive design. Learn why dead time prevents shoot-through, how to calculate optimal values, implement compensation algorithms, and minimize current distortion for smooth motor control.
