Overview: What the Gate-to-Source Resistors Do
Typical application circuits for half-bridge gate driver ICs show a resistor connected from each MOSFET gate directly back to that MOSFET's source terminal - one across the high-side device (gate to floating source/switch node) and one across the low-side device (gate to power ground). They are not optional decoration. They perform three distinct jobs:
- Guaranteed default OFF state. A MOSFET is off when VGS = 0 V. The resistor forces the gate to track its own source whenever the driver output is not actively sourcing current - during power-up, after driver undervoltage lockout (UVLO), on a broken gate trace, or if the driver output goes high impedance.
- Leakage bleed. The driver's low-side output leakage, plus MOSFET gate leakage and PCB surface leakage, has to develop a voltage somewhere. Without a defined DC path, that leakage charges the gate capacitance until the device begins to conduct.
- Transient charge dump. During supply ramp-up, capacitive coupling (Miller charge through CGD when the switch node moves) can inject charge into the gate. The resistor provides a passive, "brainless" discharge path that works before any control logic is alive.
Sizing Rule: Start From Driver Leakage
The resistor must be low enough that the worst-case leakage current cannot lift VGS anywhere near the MOSFET threshold, and high enough that it does not waste drive current or dissipate significant power.
Upper bound (turn-off integrity):
R_pd(max) = V_GS(margin) / I_leak(max)
where:
V_GS(margin) = a fraction of V_GS(th) minimum from the MOSFET datasheet
(a common engineering choice is 20-25% of V_GS(th,min))
I_leak(max) = driver output low-state leakage + MOSFET gate leakage
+ estimated board/contamination leakage, at max Tj
Use the datasheet leakage figure for the driver's output in the LOW state if it is specified. If no such specification exists, the practical starting point is 47 kΩ and adjust from measurement. Note that both VGS(th) and leakage currents degrade with temperature: threshold falls roughly a few mV/°C while leakage rises, so evaluate the ratio at maximum junction temperature, not at 25 °C.
Lower bound (drive efficiency and dissipation):
I_diverted = V_drive / R_pd (steady-state current stolen from the driver)
P_R = V_drive^2 / R_pd * D (D = duty cycle the gate is held high)
| R value | Diverted current at VGS = 15 V | Dissipation at 50% duty | Comment |
|---|---|---|---|
| 1 kΩ | 15 mA | 112 mW | Excessive load on driver and bootstrap supply |
| 10 kΩ | 1.5 mA | 11 mW | Acceptable for high-leakage or noisy designs |
| 47 kΩ | 0.32 mA | 2.4 mW | Common default starting value |
| 100 kΩ | 0.15 mA | 1.1 mW | Only with verified low leakage and clean board |
Values are calculated from the formulas above for a 15 V gate drive; substitute your actual drive rail. Resistor power rating is rarely the limiting factor - drive-current diversion and bootstrap-capacitor droop usually are.
High-Side Specifics: Bootstrap Interaction
The high-side resistor connects gate to the floating source (switch node), so its current is drawn from the bootstrap capacitor while the high-side device is on. Two consequences:
- Bootstrap droop. The resistor discharges CBOOT for the entire high-side on-time. Include this current when sizing the bootstrap capacitor: ΔV = (I_total × ton(max)) / CBOOT, where I_total includes quiescent current of the floating driver section, gate charge per cycle, and Vdrive/Rpd. Keep ΔV well below the driver's high-side UVLO threshold.
- Layout. Return the resistor to the MOSFET source pin (or Kelvin source pad) at the device, not to a distant node. Any inductance shared with the power path injects VGS noise that the resistor is supposed to suppress. Place the resistor and the gate resistor physically adjacent to the FET package.
Why You Cannot Use the Resistor as a Pull-Up for a Default-ON State
A tempting idea: if the required power-up state is high-side OFF and low-side ON, reconnect the low-side gate resistor to the +15 V rail instead of ground, so the low-side device conducts until the microcontroller finishes its boot sequence (which can be tens to hundreds of microseconds).
Do not do this. The pull-up and the series gate resistor form a permanent divider from +15 V to the driver output. When the driver pulls its low-side output (LO) to ground to turn the FET off, the gate does not reach 0 V - it settles at:
V_GS(off) = V_supply * R_gate / (R_gate + R_pullup)
Worked case (assumed values - substitute your own):
V_supply = 15 V, R_gate = 10 ohm, R_pullup = 47 kohm
V_GS(off) = 15 * 10 / 47010 = 3.2 mV (negligible)
But with R_gate = 2.2 kohm and R_pullup = 10 kohm:
V_GS(off) = 15 * 2200 / 12200 = 2.7 V (near or above V_GS(th))
The residual voltage depends entirely on the ratio, and it is worst exactly where designers use large series gate resistors for dv/dt control. Any residual VGS near threshold means the low-side device is not fully off, so every high-side turn-on produces cross-conduction current. There is also a continuous DC path from +15 V through the driver output stage, and the pull-up fights the driver on every turn-off, slowing the turn-off edge and increasing switching loss.
Correct Ways to Define the Power-Up Bridge State
| Approach | Implementation | Trade-off |
|---|---|---|
| Input-side logic | Gate the driver HIN/LIN inputs with discrete logic or an RC + Schmitt buffer that forces the desired state until the controller asserts a "ready" signal | Extra parts and BOM cost; deterministic and works in all conditions |
| Driver shutdown/enable pin | Hold the driver disabled (both outputs low, both FETs off) until the controller releases it | Only valid if a bridge-off state is acceptable during boot; check whether the driver latches |
| Driver internal pull-downs | Select a driver that specifies internal gate-output pull-down resistors active during UVLO | Still add external resistors - internal ones may be inactive with the driver unpowered |
| Controller I/O defaults | Configure the MCU pins as outputs driven to the safe level in the earliest boot code; add external resistors on HIN/LIN to define the level before that | Depends on reset behavior; the external resistors are the part that actually guarantees it |
Note that the input-side pull-down/pull-up on HIN/LIN is a completely different network from the gate-source resistor and does not suffer the divider problem, because the driver's input impedance is high and its output stage still pulls the gate hard to the source.
Verification Procedure
- Static leakage check. Power the driver bias rail with the control inputs held in the OFF state and the DC bus removed. Measure VGS on both devices with a high-impedance (10 MΩ) probe. Anything above a few tens of millivolts means the resistor is too large for the actual leakage.
- Hot check. Repeat at maximum expected ambient/heatsink temperature, where leakage is highest and VGS(th) is lowest.
- Power-up sequencing. With low DC bus voltage and a current-limited supply, scope both gate-source voltages differentially (isolated probe or differential probe for the high side) during bias ramp-up. Look for any transient VGS excursion toward threshold. Include a fast rail rise and a slow, drooping rise.
- Shoot-through screening. Monitor DC bus current with a current probe during switching. Current spikes coincident with the opposite device's turn-on edge indicate incomplete turn-off - check for stray gate voltage, insufficient dead time, or dv/dt-induced Miller turn-on.
- Bootstrap validation. Measure the bootstrap capacitor voltage at maximum high-side on-time and confirm it stays above the high-side UVLO threshold with the pull-down current included.
Design Checklist
- Fit a gate-to-source resistor on every MOSFET in the bridge, referenced to that device's own source.
- Start at 47 kΩ when no driver leakage figure is available; reduce toward 10 kΩ for hot, high-leakage, or contaminated-environment designs.
- Verify R_pd × I_leak(max) is a small fraction of VGS(th,min) at maximum junction temperature.
- Never return the pull-down to a positive rail to create a default-ON state.
- Add the pull-down current to the bootstrap capacitor budget for the high-side device.
- Keep the pull-down, the series gate resistor, and any gate-clamp components in a tight loop with the FET package.
FAQ
What value should I use for a MOSFET gate-to-source pull-down resistor?
Compute R_max = V_GS(margin) / I_leak(max) using the driver's low-state output leakage plus gate and board leakage at maximum junction temperature. If the driver does not specify leakage, start at 47 kΩ and verify by measuring VGS with the driver biased and the inputs held off.
Can I omit the gate-source resistor if the driver has internal pull-downs?
No. Internal pull-downs are typically only active when the driver is powered and out of UVLO. The external resistor still protects the FET when the driver bias is absent, during supply ramp, or if a gate connection opens.
Why does a gate pull-up to +15 V cause shoot-through?
The pull-up and the series gate resistor form a divider with the driver's output stage. When the driver pulls the output low, VGS settles at V_supply × R_gate / (R_gate + R_pullup) instead of 0 V. If that residual sits near VGS(th), the device stays partly on while the opposite device turns on.
How do I set a defined half-bridge state during microcontroller boot?
Define the state on the driver's logic inputs, not on the gates - use discrete logic, an RC plus Schmitt buffer, or the driver's shutdown/enable pin, released by an explicit "controller ready" signal. Do not depend on it for safety-critical protection, since a failed controller may drive an output actively.
Does the high-side pull-down resistor affect bootstrap capacitor sizing?
Yes. It draws V_drive/R_pd continuously from the bootstrap capacitor while the high-side FET is on. Add that current to the floating quiescent current and gate charge, then confirm the capacitor droop keeps the bootstrap voltage above the high-side UVLO threshold at maximum on-time.