Hi,
I’m building a PWM fan controller using Arduino Nano and an N-channel MOSFET. The fan current is only 1.2 A, but the MOSFET gets hotter than expected.
Gate is driven directly from the MCU pin at 5V, PWM frequency is 31 kHz.
Questions:
Could switching losses be the main reason for heating?
Would a gate driver help even at this current?
On the PCB, how much copper area should I keep around the MOSFET drain/source for thermal dissipation?
Yes … switching losses could be the main reason for heating. Even at 1.2 A, a MOSFET can heat up if it’s not fully enhanced or switching slowly. At 31 kHz PWM, switching losses start to matter because:
Arduino pin has weak drive strength … results in slow gate rise/fall times
MOSFET spends more time in the linear region during transitions
This can dominate losses more than conduction loss at low current
Yes … gate driver help at this current. Even though 1.2 A is small … but a gate driver helps because:
Faster switching: reduced time in linear region
Lower switching losses at 31 kHz
Cleaner PWM edges: less heat and less EMI
A simple fix … helped me in one of my projects … a small totem-pole driver or dedicated low-side MOSFET driver IC will improve efficiency more than expected. Read it out: https://www.ti.com/lit/ab/slua877/slua877.pdf
PCB copper area for thermal dissipation … For 1–2 A PWM fan loads: Use a large copper area on drain side (especially if low-side switching) … read this guide on thermal dissipation: https://www.aivon.com/blog/pcb-knowledge/embedded-copper-structures-for-extreme-pcb-thermal-management/
Extend copper pour around MOSFET to act as a heatsink
Connect top and bottom copper with thermal vias under the pad … although its abt LED PCB but heat flow calculaitons are helpful: Thermal Resistance and Heat Flow in LED PCB Design: Understanding LED PCB Layout for Better Heat Dissipation - PCB Design & Layout - PCBway
Keep source return path short and wide
Avoid narrow traces between MOSFET and ground
In your case, the heating is more likely switching … gate drive limitation than load current. A proper logic-level MOSFET + stronger gate drive + decent copper pour usually eliminates the issue completely.
In addition of what has being said; 31kHz is not terribly high (for a 328P/328U MCU). But who knows if you are even driving the MOSFET fully. Is the MOSFET a logic level FET? That would go a long way to ensuring the MOSFET is actually turning fully on and off.