A 4-pin PWM header controls fan speed by modulating a 25 kHz pulse signal while maintaining constant 12 V power, which gives better torque and smoother low-RPM operation than voltage reduction. The four pins carry ground, 12 V supply, tachometer feedback, and the PWM control signal. Below I'll walk through the pinout, signal behavior, and the mistakes I've seen trip up both first-time builders and experienced engineers.
Key Takeaways
- PWM fans run on constant 12 V and change speed via duty cycle, not supply voltage—this prevents the stalling you get with voltage-starved 3-pin fans below ~40% speed.
- The PWM pin is open-collector: the motherboard pulls it low, the fan pulls it high internally, so you cannot drive it with a standard push-pull 5 V GPIO without a level-shift stage.
- TACH outputs two pulses per revolution; missing or erratic pulses are your only real-time indicator of bearing wear or blockage before the fan stops entirely.
- Plugging a 3-pin fan into a 4-pin header usually works because the first three pins match, but the header must be set to DC mode or the fan will run full speed.
- Current limits per header are often 1–1.5 A; daisy-chaining fans through a Y-splitter works until you hit that ceiling, then the trace or header fails—I've seen melted connectors on builds that ignored this.
4-Pin PWM Header Pinout
The physical connector is a 1×4 housing with a keying tab that prevents backwards insertion.
- Pin 1: GND – Common ground return for power and signals.
- Pin 2: +12 V – Continuous fan supply, typically regulated to ±5% (11.4–12.6 V).
- Pin 3: TACH – Open-collector speed sensor output, usually two pulses per revolution.
- Pin 4: PWM – 25 kHz ±10% control input, open-drain/open-collector logic.
Pin 1 is marked by a small arrow or chamfer on the housing. Reversing the connector sends 12 V into the TACH or PWM pin and will damage the fan's internal circuitry instantly.
How the PWM Control Signal Works
PWM frequency sits around 25 kHz because it's above audible range and fast enough that the fan's inductive load smooths the pulses into nearly continuous torque. The motherboard or controller sinks the PWM line to ground during the "on" portion of each cycle; the fan has an internal pull-up (typically to 5 V, sometimes 3.3 V) that defines the high state.
Duty cycle maps to speed, but the relationship is not linear. Most fans have a minimum duty cycle—often 20–30%—below which they cannot maintain rotation due to insufficient average power to overcome bearing friction and back-EMF. At 100% duty cycle the line stays high and the fan runs at its rated maximum RPM.
From a practical standpoint, if you're generating the PWM signal with a microcontroller, you need an open-drain output or a simple NPN/N-channel stage. A standard 3.3 V or 5 V push-pull GPIO will fight the fan's internal pull-up and either clip the high level or source current back into your MCU pin—both bad outcomes.
TACH Feedback and Fault Detection
The TACH pin is your only real-time window into fan health. It outputs a square wave with two rising edges per mechanical revolution, so a fan spinning at 1200 RPM produces 2400 pulses per minute, or 40 Hz. The motherboard counts edges over a known interval and calculates RPM.
If the TACH signal drops out or becomes irregular, the BIOS triggers a fan-fail alarm. In my experience, intermittent TACH readings are the first sign of bearing wear—long before you hear grinding. Ignoring TACH and relying on "it sounds fine" has led more than one server deployment to an unplanned thermal shutdown during a summer heatwave.
The signal is open-collector, so it requires a pull-up resistor on the motherboard (typically 4.7–10 kΩ to 5 V or 3.3 V). If you're reading TACH on a custom board, add the pull-up externally or enable the internal pull-up on your MCU pin.
3-Pin vs 4-Pin Fans on a 4-Pin Header
A 3-pin fan uses only GND, +12 V, and TACH. When you plug it into a 4-pin header, it physically mates because the first three pins are identical. Speed control depends on the header's mode:
- DC/Voltage mode: The motherboard varies the voltage on pin 2 between roughly 5 V and 12 V. The fan slows as voltage drops, but below ~7 V most fans stall because there isn't enough torque to overcome startup friction.
- PWM mode: The header keeps pin 2 at 12 V and modulates pin 4. A 3-pin fan ignores pin 4, so it runs at full speed continuously.
Conversely, a 4-pin PWM fan on a 3-pin header loses the PWM signal and runs at whatever voltage the header supplies—usually 12 V, so full speed.
| Fan Type | Header Type | Control Method | Low-Speed Stability |
|---|---|---|---|
| 3-pin | 3-pin | Voltage modulation | Poor below ~40% |
| 3-pin | 4-pin (DC mode) | Voltage modulation | Poor below ~40% |
| 3-pin | 4-pin (PWM mode) | None (runs full speed) | N/A |
| 4-pin PWM | 4-pin (PWM mode) | Duty cycle modulation | Good down to ~20% |
| 4-pin PWM | 3-pin | Voltage modulation (if variable) | Moderate |
The takeaway: use 4-pin PWM fans when you need stable low-RPM control, especially in noise-sensitive environments. Use 3-pin fans only when cost is the priority and you can tolerate higher minimum speeds.
BIOS and Software Fan Control
Modern UEFI firmware exposes per-header configuration:
- Mode selection: PWM or DC/Voltage. Set this to match your fan type or the fan will misbehave.
- Fan curve: A lookup table mapping sensor temperature (CPU, chipset, or custom thermistor) to duty cycle or voltage. Most boards offer preset profiles (Silent, Standard, Performance) and a custom curve editor.
- Zero-RPM / Fan Stop: Some boards allow the fan to stop completely below a threshold temperature, typically 40–50 °C. This only works if the fan supports true zero-RPM operation—many budget fans have a minimum speed floor and will judder or stall if commanded to stop.
- TACH monitoring: Enable or disable fan-fail alarms. Disabling is useful during testing but risky in production.
In my projects, I always validate the fan curve under load before calling it done. A curve that looks reasonable on paper can still let the CPU hit thermal throttling if the ramp rate is too slow or the setpoint is too optimistic.
OS-level tools (SpeedFan, iCUE, Argus Monitor) can override BIOS settings, but they rely on chipset support and may not persist across sleep or reboot.
Electrical Specifications
| Parameter | Typical Value | Notes |
|---|---|---|
| Fan supply voltage | 12 V ±5% | 11.4–12.6 V measured at the header |
| Max current per header | 1–1.5 A | Check your motherboard manual; server boards may allow more |
| PWM frequency | 25 kHz ±10% | 22.5–27.5 kHz is within spec |
| PWM logic | Open-collector/open-drain | Fan provides internal pull-up to ~5 V |
| TACH output | Open-collector, 2 pulses/rev | Requires external or internal MCU pull-up |
| TACH sink current | ~3 mA typical | Check fan datasheet for exact value |
The 1–1.5 A limit is per header, not per fan. If you connect two 0.8 A fans via a Y-splitter, you're at 1.6 A and likely exceeding the header's rating. The trace might handle it, or it might not—depends on copper weight and via count. I've seen headers survive 2 A for months and others fail at 1.3 A after a few thermal cycles. Don't gamble; use a powered hub or separate 12 V rail if your total fan current exceeds the spec.
Using 4-Pin Headers in Custom Builds
You can drive a 4-pin PWM fan from a microcontroller or embedded SBC by replicating the header's electrical behavior:
- 12 V rail: Use a buck converter or separate 12 V PSU. Tie its ground to your logic ground.
- PWM generation: Configure a timer/PWM peripheral for 25 kHz. Route it through an NPN transistor (2N2222, 2N7002) or open-drain buffer (74LVC1G07) so the output pulls low and the fan's internal pull-up defines high.
- TACH reading: Connect TACH to a GPIO with interrupt capability. Enable the internal pull-up (or add a 4.7 kΩ resistor to 3.3 V). Count rising edges over a 1-second window and multiply by 30 to get RPM (assuming 2 pulses/rev).
I've used this approach in a laboratory chiller controller where the fan had to track coolant temperature with ±50 RPM precision. Closed-loop control (read TACH, adjust PWM, repeat) converged in under 2 seconds and held steady once tuned.
One gotcha: if your MCU outputs a 3.3 V logic-high PWM directly to the fan's PWM pin, the fan sees a lower pull-up voltage than expected (3.3 V instead of its internal 5 V) and may interpret duty cycle incorrectly or not respond at all. Always level-shift to open-drain.
Achieving Quiet Cooling
PWM control is the reason modern desktops can idle silently. A well-tuned fan curve keeps duty cycle low during light loads (20–40%) and ramps smoothly as temperature rises. Because the fan motor always has 12 V available, it maintains full torque even at low average speed, so there's no cogging or startup hesitation when the curve steps up.
Contrast this with voltage control: dropping from 12 V to 6 V cuts both speed and torque. The fan may spin slowly, but it can't accelerate quickly when load spikes, and it may stop entirely if dust increases bearing friction.
When designing for quiet operation:
- Set the minimum duty cycle just above the fan's stall threshold—usually 25–30%.
- Use a slow ramp rate (e.g., 5% duty change per °C) so speed changes are gradual and inaudible.
- Place the idle setpoint 10–15 °C below the CPU's thermal throttle point to leave headroom for transient spikes.
- Test with sustained workloads (Prime95, Cinebench loop) to verify the curve doesn't oscillate or undershoot.
I've seen fan curves that looked perfect in BIOS testing but caused audible ramping during web browsing because the curve reacted to every momentary 2 °C temperature wiggle. A small hysteresis band (±2 °C deadzone) eliminates that.
Common Setup Mistakes
Reversed connector orientation: The keying tab prevents this if you align it correctly. Forcing it in backwards applies 12 V to TACH or PWM and usually kills the fan's controller IC. I've diagnosed this failure three times; each time the user swore the connector "only fit one way," but they'd bent the plastic tab.
Assuming 0% PWM stops the fan: Most fans have a minimum duty threshold (20–30%) below which they either run at minimum speed or stall unpredictably. Only fans explicitly marketed as "zero-RPM" or "fan-stop" will halt at 0% duty.
Overloading a single header with a splitter: Daisy-chaining fans is fine until you exceed the header's current limit. Two 0.5 A fans = no problem. Three 0.8 A fans = 2.4 A, which is beyond spec and will eventually melt the connector or lift a trace. Use a powered hub or fan controller for high-current setups.
Mixing 3-pin and 4-pin fans on one header (via splitter): If the header is in PWM mode, the 3-pin fan runs full speed while the 4-pin fan follows the curve. If the header is in DC mode, the 4-pin fan loses PWM control and you're back to voltage modulation for both—defeating the purpose. Keep fan types matched per header.
Wrong BIOS control mode: A 4-pin fan on a header set to DC mode will run at full speed because it ignores voltage changes and expects a PWM signal that isn't being sent. A 3-pin fan on a header set to PWM mode also runs full speed because it ignores the PWM pin. Always verify mode matches fan type after swapping hardware.
Ignoring TACH warnings: BIOS fan-fail alerts are annoying until they're not. A flaky TACH signal means the fan is on its way out. I've ignored these warnings twice and regretted it both times—once with a GPU thermal shutdown mid-render, once with a server hitting 95 °C before I noticed.
Design and Integration Checklist
Pin order and polarity: Wire GND to pin 1, +12 V to pin 2, TACH to pin 3, PWM to pin 4. Mark pin 1 on the PCB and cable with a stripe or arrow. Double-check continuity before applying power.
PWM signal conditioning: Use open-drain or open-collector drive at 25 kHz. If your MCU has only push-pull PWM, add an NPN or N-channel FET stage. Confirm the fan's pull-up voltage (usually 5 V) is compatible with your logic levels.
Current budget per header: Sum the maximum current of all fans on one header. If the total exceeds the header's rating (typically 1–1.5 A), use a powered splitter or separate 12 V supply with only the PWM and TACH signals connected through the header.
Fan type and control mode alignment: Document which headers use PWM mode and which use DC mode. Label fan cables to prevent accidental mismatches during assembly or maintenance.
BIOS/firmware configuration: After initial setup, export or photograph the fan curve and mode settings. Re-verify these settings after BIOS updates, battery changes, or hardware swaps—they sometimes revert to defaults.
PWM range validation on custom builds: Sweep duty cycle from 10% to 100% in 10% steps and record RPM via TACH at each point. Identify the minimum duty cycle that maintains stable rotation and set your control law's floor 5% above that.
Documentation: Include the 4-pin pinout, PWM frequency, TACH pulse count, and current limits in your build log or schematic. Future you (or the next technician) will thank you when troubleshooting at 2 AM.
Conclusion
The 4-pin PWM header's combination of constant 12 V power and 25 kHz duty-cycle control delivers stable, quiet fan operation across a wider speed range than voltage modulation alone. Correct pinout, open-collector signal drive, TACH monitoring, and current-limit awareness are non-negotiable for reliable long-term operation. Match your BIOS control mode to your fan type, validate your fan curve under real load, and treat TACH warnings as predictive maintenance alerts rather than nuisances.
The next time you're chasing a thermal issue or unexplained fan behavior, check the basics first: pin alignment, control mode, and current budget. I've spent hours debugging "broken" fans that turned out to be plugged into a header set to the wrong mode.
Frequently Asked Questions
Will a 4-pin PWM fan run at full speed if the PWM signal is missing?
Yes. The fan interprets a missing or floating PWM signal as "no control input" and defaults to maximum speed for safety.
Why does my PWM fan still spin at 0% PWM?
Most PWM fans have a minimum speed floor, typically 20–30% duty cycle, below which they either maintain minimum RPM or stall. Only fans explicitly rated for zero-RPM operation will stop completely.
Can I use a 4-pin PWM fan on a PUMP_FAN header?
Yes. PUMP_FAN headers are electrically identical to other 4-pin headers but often default to 100% duty cycle or a higher minimum speed in BIOS. You may need to manually configure the header for normal fan control.
Can I connect two fans to one 4-pin PWM header using a splitter?
Yes, provided the combined maximum current does not exceed the header's rating (typically 1–1.5 A). Both fans will receive the same PWM signal, but only one TACH signal (usually from the first fan) will be reported to the motherboard.
Does PWM control reduce fan lifespan?
No. PWM switching is a standard operating mode and does not accelerate bearing wear or reduce lifespan compared to continuous operation at the same average speed.
How do I know my fan is actually following PWM control?
Monitor RPM via BIOS or software while changing the duty cycle. RPM should increase as duty cycle increases. If RPM stays constant at maximum, the fan is either not receiving a PWM signal, the header is in the wrong mode, or the fan does not support PWM.
