Search
Filters

Techniques for High and Low Side Drive Control of Transistor Modules

High-Side and Low-Side Drive Control Tips for Transistor Modules: What Actually Works in the Field

Driving the high side and low side of a transistor module in a half-bridge or full-bridge configuration sounds simple on paper. Low side is easy — ground-referenced gate drive, no tricks needed. High side is where things get ugly. The gate floats on top of a switching node that moves at full bus voltage, which means your gate driver has to survive hundreds of volts of common-mode noise while still delivering a clean, fast signal. Get this wrong and you get shoot-through, false turn-on, or destroyed gate oxide. Here is how experienced engineers handle it.

The Fundamental Problem With High-Side Drive

When the low-side transistor is on, the source terminal of the high-side device sits at ground potential — no problem. But when the low side turns off and the high side turns on, that source terminal jumps to the full bus voltage. The gate driver now has to float at bus potential and still produce a gate-to-source voltage of 15 to 20 volts. That is a floating power supply problem, and it is the single biggest headache in bridge topologies.

The low side is not without issues either. Ground bounce from high di/dt currents in the power loop can shift the local ground potential by several volts. This shift appears directly across the gate-emitter terminals and can cause unintended turn-on or turn-off. So both sides demand careful attention — just in different ways.

High-Side Drive Techniques That Do Not Fail in Practice

Bootstrap Circuits: Still the Workhorse But Do It Right

Bootstrap is the most common high-side drive method, and for good reason — it is cheap, simple, and effective when done correctly. A bootstrap capacitor charges through a diode when the low side is on, storing enough energy to drive the high-side gate during the on-time. The catch is that the high side cannot stay on indefinitely. The capacitor drains through the gate leakage and the driver quiescent current, so you need a minimum off-time to recharge it.

Most engineers get this wrong by using a diode that is too slow or a capacitor that is too small. Use a fast recovery diode — not a standard rectifier — and size the bootstrap capacitor at least 10 times the gate charge of the module. For large IGBT modules with gate charges in the microcoulomb range, a 10 microfarad bootstrap capacitor is not overkill. Place the bootstrap diode and capacitor as close to the gate driver IC as possible. Long traces add inductance, and that inductance limits the charging current during the brief low-side on-time.

Isolated Gate Drivers: When Bootstrap Is Not Enough

Bootstrap breaks down when you need high duty cycles, slow switching frequencies, or when the high side must stay on for extended periods. In those cases, an isolated gate driver with its own floating power supply is the answer. The isolation can be achieved through transformers, capacitors, or optocouplers, each with trade-offs.

Transformer-coupled drivers offer the fastest signal propagation and the best noise immunity. They work well at high PWM frequencies but require a minimum pulse width to transfer energy across the isolation barrier. Capacitive isolation drivers are more tolerant of very narrow pulses and work well for MOSFET modules at high frequencies. Optocoupler-based drivers are the slowest and degrade over time, but they are cheap and adequate for low-frequency IGBT applications.

The key rule with any isolated driver: the isolated power supply must be able to deliver peak gate current without voltage droop. A weak isolated supply causes slow switching, which increases loss and heat. Check the transient response of the isolated supply under full load before you commit to the design.

Low-Side Drive Is Not As Easy As You Think

Ground Bounce Is the Silent Killer on the Low Side

Everyone focuses on the high side, but ground bounce on the low side kills modules just as fast. When the low-side transistor switches off, the rapid current change induces a voltage spike on the ground plane due to parasitic inductance. That spike appears between the gate driver ground and the module emitter, effectively adding or subtracting voltage from the gate-emitter terminal.

The fix is straightforward but requires discipline. Use a Kelvin connection for the gate driver ground — a dedicated trace that returns directly to the module emitter pin without sharing any path with the power ground. Keep this trace short and wide. Add a local decoupling capacitor right at the gate driver ground pin. And never route the gate signal trace parallel to the power current path — even a few millimeters of parallel run creates enough mutual inductance to inject noise into the gate signal.

Turn-Off Speed Matters More Than Turn-On Speed on the Low Side

Most designers obsess over fast turn-on to reduce switching loss. But on the low side, slow turn-off is what causes real damage. When the low-side transistor turns off slowly, it spends too much time in the active region where both voltage and current are high. That is where the power dissipation peaks. A slow turn-off also gives the high-side transistor more time to turn on before the low side is fully off, increasing shoot-through risk.

Use a low-value gate turn-off resistor — typically 1 to 5 ohms — to actively pull the gate low. For IGBT modules, an active clamp that forces the gate negative during turn-off speeds up the process dramatically and reduces tail current. For MOSFET modules, a strong gate driver capable of sinking 5 to 10 amps is essential for clean, fast turn-off.

Dead Time Management: The Parameter That Separates Good Designs From Bad Ones

Dead time is the short interval when both high-side and low-side transistors are off. Too little dead time and you get shoot-through — both devices conduct simultaneously, creating a direct short across the bus. Too much dead time and the body diode of the opposite transistor conducts, increasing loss and distorting the output waveform.

Adaptive Dead Time Beats Fixed Dead Time Every Time

Fixed dead time is a compromise that works at one operating point but fails at others. At high bus voltage, the transistor turns off slower because the stored charge takes longer to remove. At high temperature, turn-off gets even slower. A fixed dead time that is safe at low voltage and room temperature may be dangerously short at high voltage and high temperature.

Adaptive dead time adjusts the off-interval based on operating conditions. Some gate driver ICs include desaturation detection — if the high-side device does not turn off within a preset window, the driver extends the dead time automatically. This is a lifesaver in motor drive applications where load conditions change constantly. If your driver IC does not support adaptive dead time, implement it in your controller firmware by monitoring the voltage across the transistor during turn-off and adjusting the next cycle accordingly.

Measuring Actual Dead Time, Not Just Setting It

Setting dead time in firmware is not the same as having the right dead time at the module pins.y. Propagation delays in the gate driver, mismatched rise and fall times between high and low side, and PCB trace delays all add up. The actual dead time at the module can be 50 to 200 nanoseconds different from what you programmed.

Measure it with an oscilloscope. Probe both gate-emitter voltages simultaneously on the same time base. Measure the interval from when one gate falls below the threshold to when the other gate rises above threshold. Do this at maximum bus voltage, maximum temperature, and minimum gate drive voltage. If the measured dead time is less than 200 nanoseconds for IGBT modules or less than 50 nanoseconds for MOSFET modules, you are in the danger zone.

Layout Practices That Make High-Side and Low-Side Drive Work Together

The power loop and the gate loop must be treated as separate systems. The power loop carries hundreds of amps and generates massive dI/dt. The gate loop carries milliamps but must remain clean. If these loops share any area on the PCB, noise from the power loop couples directly into the gate signal.

Place the gate driver IC between the high-side and low-side modules, as close to both as possible. This minimizes the gate loop area for each side. Use separate ground returns for the high-side driver and the low-side driver, joining them only at the DC bus capacitor negative terminal. This star-ground approach prevents high-side switching noise from contaminating the low-side gate reference.

For high current modules, use multiple parallel gate resistors instead of a single resistor. This distributes the gate current evenly and reduces local heating. And always include a gate-emitter resistor — typically 10 to 100 kilohms — on each transistor to ensure it stays off when the driver is unpowered or in fault state. A floating gate is an invitation for false turn-on.


footer Upper Image