A high-side N-channel MOSFET in a buck converter needs a gate voltage about 10 V above its source, and a ground-referenced PWM source cannot supply that. A bootstrap circuit solves it with one diode, one capacitor and a small gate-driver stage. This article starts from a plain 15 V to 12 V buck converter that cannot work, then adds the bootstrap circuit step by step. It ends with a High-Side N-Channel MOSFET Buck Converter design with Bootstrap Circuit where the gate swings to about 29 V while the switch node swings between 0 V and 15 V.
The original circuit and why it fails
The starting point is a 15 V to 12 V buck converter with an IRF540 as the high-side switch. A pulse generator (0 V to 15 V) drives the gate directly, with a 10 kΩ resistor from gate to ground.
Ref | Value | Role |
|---|---|---|
BAT1 | 15 V | Input supply |
Q1 | IRF540 | High-side N-channel switch |
D1 | 1N5819 | Freewheel diode |
L1 | 470 µH | Output inductor |
C1 | 2200 µF | Output capacitor |
LOAD | 12 V, 24 Ω | Load |
R1 | 10 kΩ | Gate to ground |
Vpwm | 0 V to 15 V pulse | Gate drive |
The problem is the gate-to-source voltage. A MOSFET turns on according to Vgs = Vgate − Vsource, and Q1's source is the switch node, which rises toward 15 V when Q1 conducts. The gate can never exceed 15 V, so Vgs shrinks toward zero as the source climbs. The IRF540's threshold is 2–4 V, so the source stalls at roughly 10 V with Q1 only partly on and its resistance rising. A full 12 V output is out of reach. Fully turning the MOSFET on needs about 10 V of Vgs, which means a gate near 15 V + 10 V = 25 V.
How a bootstrap circuit works
The bootstrap capacitor acts as a small battery that floats on the switch node. It is charged when the switch node is low and used to lift the gate when the switch node is high.
- Charge phase (Q1 off). D1 pulls the switch node to about −0.4 V. Current flows from the 15 V rail through the bootstrap diode D2 into the bootstrap capacitor C2, charging it to about 14 V.
- Boost phase (Q1 on). The gate driver connects Q1's gate to the top of C2. As Q1 turns on, its source rises to 15 V and lifts the bottom of C2 with it, so the top of C2 reaches about 15 V + 14 V = 29 V.
- Hold. D2 is now reverse-biased, so C2 cannot discharge back into the supply. The gate sits at about 29 V against a 15 V source, giving Vgs of about 14 V.
The gate must be switched between the top of C2 and the switch node, so the driver stage has to run from that floating supply. This design builds it from discrete parts:
- Q3 (NPN) connects the gate to the top of C2 to turn Q1 on.
- Q4 (PNP) connects the gate to the switch node to turn Q1 off.
- Q2 (NPN) translates the ground-referenced PWM signal into the floating stage.
This stage inverts the signal: PWM high turns Q1 off, and PWM low turns Q1 on.
Two limits apply. C2 only recharges while the switch node is low, so duty cycle must stay below about 90–95%. And C2 must be large enough that the gate charge barely dents its voltage: a 1 µF capacitor against a 50–70 nC gate charge drops only about 50–70 mV per cycle.
Step by step: from the original circuit to the bootstrap version
The power stage (BAT1, Q1, D1, L1, C1, load) stays as it is. Everything below is added around Q1's gate.
- Add the bootstrap diode D2 (UF4001) from the 15 V to the new bootstrap rail.
- Add the bootstrap capacitor C2 (1 µF) between the switch node and the bootstrap rail. This is the key step: C2 must float on the switch node, not sit on ground.
- Add the totem-pole. Q3 (2N2222A, NPN) has its collector on bootstrap voltage rail and its emitter on gate. Q4 (2N2907, PNP) has its collector on switch node and its emitter on gate. Tie both bases together.
- Add the pull-up R2 (2.2 kΩ) from base of Q3 to bootstrap voltage rail, so the bases are pulled high when Q2 is off.
- Add the level shifter Q2 (2N2222A) with its collector on junction of Q3 and Q4 bases and its emitter on ground. Drive its bases from the PWM source through R1 (1 kΩ).
- Connect the driver output to the gate through R3 (47 Ω).
- Gate pulldown resistor. add R4 (22 kΩ) from the Q1 gate to the switch node, so the gate is held at source level if the driver is off.
- Move the PWM source. It now drives Q2's base, not the gate directly. Because the stage inverts, the PWM high time is Q1's off time.
- Add input capacitors C3 (470 µF) and C4 (0.1 µF) from Q1 drain to ground, close to Q1 and D1.
High-Side N-Channel MOSFET Buck Converter with Bootstrap Circuit
The finished design uses an IRFZ44N as the switch and realistic practical values for the inductor and capacitors.
Ref | Value | Role |
|---|---|---|
BAT1 | 15 V | Input supply |
Q1 | IRFZ44N | High-side N-channel switch |
D1 | 1N5819 (1N5822 for 3 A) | Freewheel diode |
L1 | 220 µH | Output inductor |
C1 | 470 µF | Output capacitor |
C3 | 470 µF | Input bulk capacitor |
C4 | 0.1 µF | Input ceramic capacitor |
D2 | UF4001 | Bootstrap diode |
C2 | 1 µF | Bootstrap capacitor |
Q3 | 2N2222A | Totem-pole, turns Q1 on |
Q4 | 2N2907 | Totem-pole, turns Q1 off |
Q2 | 2N2222A | Level shifter |
R1 | 1 kΩ | Q2 base resistor |
R2 | 2.2 kΩ | Totem-pole pull-up |
R3 | 22 Ω | Gate resistor |
R4 | 22 kΩ | Gate to switch node |
LOAD | 12 V, 24 Ω | Load |
V1 | PWM source | Drives Q2 |
PWM settings and simulation results
The simulated gate waveform shows the bootstrap working: Q1's gate reaches about 29 V while the switch node swings between about −0.4 V and 15 V. That gives Q1 a Vgs of about 14 V, so it is fully on.
Set the V1 pulse source as follows for 50 kHz at about 82% on-time. Because the driver inverts, V1 is high while Q1 is off.
Field | Value |
|---|---|
Initial Value | 0 |
Pulse Value | 15 |
Rise / Fall Time | 10n |
Pulse Width | 3.6u |
Period | 20u |
The output follows Vout ≈ D × Vin, so Q1's on-time sets the voltage. A longer pulse width shortens Q1's on-time and lowers the output; a shorter pulse raises it. Adjust the width in small steps until the load probe settles near 12 V.
Three notes on reading the plots:
- Wave shape: at 20 µs per period, the zoomed trace shows the gate and switch node pulsing together with the diode drop (about −0.4 V) between pulses, as expected.
- Start-up: in the first 20–25 ms the load trace rings with very large swings that decay. A 15 V source cannot physically produce those voltages, so treat this as simulator start-up behavior, and judge the converter from the settled waveforms.
- Run time: with 220 µH and 470 µF the output filter is lightly damped, so run the simulation to at least 0.2 s before judging the final output voltage.
Video Animation of High-Side N-Channel MOSFET Buck Converter with Bootstrap Circuit
Real hardware: values, MOSFET choice and common mistakes
Values for a 15 V to 12 V converter at up to 1–2 A, running at 50 kHz:
Part | Typical value |
|---|---|
L1 | 220 µH, saturation current ≥ 3 A, low DCR |
C1 (output) | 470 µF, 25 V, low-ESR, plus 0.1 µF ceramic |
Input capacitors | 470–1000 µF, 25 V, plus 0.1 µF at the MOSFET drain |
D1 | 1N5822, SS34 or SS54 (3 A Schottky) |
C2 | 0.47–1 µF, 25 V, X7R ceramic |
R3 (gate) | 33–47 Ω for real hardware, since 22 Ω puts the BJTs near their current limit |
R4 (gate to source) | 10–47 kΩ |
MOSFET choice. The IRFZ44N (55 V, 49 A, about 17.5 mΩ) is the best fit and runs cool without a heatsink at this current. The IRF540N (100 V, about 44 mΩ) also works and suits a later move to 24 V input. The IRF740 (400 V, about 0.55 Ω) works but wastes power: roughly 0.55 W at 1 A, so it is not recommended at 15 V. All three have a gate charge around 60–70 nC, so the same gate drive suits each.
Common mistakes:
- Connecting C2 from the bootstrap rail to ground instead of to the switch node.
- Leaving out the pull-up resistor R2, so the totem-pole bases have nothing pulling them high.
- Driving a transistor base straight from the PWM source with no series resistor.
- Reading the BJT pin order as C, B, E in a Proteus netlist; it is B, C, E.
- Forgetting that the stage inverts, so PWM high turns the MOSFET off.
- Asking for 100% duty: C2 needs off-time to recharge.
A real converter also needs feedback to regulate against load and supply changes; a TL494, SG3525 or a microcontroller with ADC feedback can provide it. For a finished design, a driver IC such as the IR2104 is simpler and more robust than the discrete stage, and the discrete version is best treated as a learning build.
Related blog posts:
- Inverter Design with TL494 PWM controller
- Push Pull TL494 Inverter to Modified Sine Wave Inverter - How to modify?
- How to Convert TL494 Inverter to Modified Sine Wave Inverter
- Measurement & Waveform Analysis of TL494 Modified Sine Wave Inverter
- Improving the TL494 Modified Sine Wave Inverter Design
- DIY 12V DC to 220V AC Inverter using TL494 PWM
- TL494 Vs SG3525, which one to use?
