PCB Bring-Up Checklist: First Power-On for a New Board

Detailed Explanation

The first power-on of a new PCB is the moment where design errors, assembly defects, and manufacturing tolerances all manifest at once. A systematic bring-up process catches problems before they destroy components — and identifies whether a fault is in the hardware, the firmware, or the schematic.

Bring-up is not the same as production testing. It is the process a hardware engineer goes through on a prototype or first-article board to validate that the design is correct. Production testing (AOI, ICT, functional test) is separate — see PCB assembly testing.

Before You Apply Power: Visual Inspection

1. Inspect for solder bridges. Use magnification (10–30× loupe or USB microscope) to inspect all fine-pitch ICs, BGAs (where accessible), and power component pads. A bridge between adjacent power supply pins can destroy components instantly on power-on.

2. Check component orientation. Verify polarised components: electrolytic and tantalum capacitors (check the stripe/plus-mark against the schematic and silkscreen), diodes, LEDs, and any ICs with asymmetric pinouts. An electrolytic capacitor backwards on a power rail will conduct, heat, and potentially rupture.

3. Check for missing or wrong components. Compare the assembled board against the schematic. Check value markings on passives if readable (many 0402 capacitors are unmarked). Pay special attention to voltage regulators, protection ICs, and crystal oscillators — wrong values here prevent the board from functioning and may be hard to diagnose by measurement alone.

4. Check for cold solder joints. Under magnification, a cold joint has a dull, grainy appearance and may show a gap between the lead and the pad fillet. Press gently with a probe tip — a cold joint may be physically loose.

Pre-Power Electrical Checks

5. Continuity test: supply rails to GND. Before any power is applied, use a multimeter in continuity (beep) mode to check resistance from each power supply rail to GND. A short (beep or very low resistance reading) is a hard short circuit — a reversed capacitor, solder bridge, or component in the wrong location. Do not apply power until shorts are resolved.

Note: Some designs intentionally have low resistance from a supply rail to GND through bulk capacitors or ferrite beads — know your schematic. The relevant test is: is the resistance significantly lower than expected (indicating a hard short)?

6. Continuity test: key net connectivity. Spot-check a few critical nets: verify the MCU reset line connects to where it should, that a key SPI CS connects to the expected MCU pin, and that the crystal oscillator connects to the MCU's OSC pins. This catches wiring errors that a visual inspection misses.

First Power-On: Current-Limited Supply

7. Use a bench supply in constant-current mode. Set the voltage to the correct supply voltage (e.g. 3.3 V or 5 V) and set the current limit to 10–20% of expected operating current (or 50–100 mA if the expected current is unknown). Connect the supply to the board's primary power input.

8. Observe the supply display. When the supply is turned on:

• Voltage holds at setpoint, current below limit: Power rails coming up normally — proceed.
• Voltage collapses, current at the limit: A significant short circuit is loading the supply. Power off immediately and investigate.
• Voltage holds, current slightly elevated: May be acceptable (charging bulk capacitors, components with higher than expected quiescent current) — let it settle for 1–2 seconds and reassess.

9. Check for hot components immediately. After 5–10 seconds at limited current, touch larger components (regulators, MOSFETs, processor) carefully with a fingertip. Abnormal warmth in the first seconds of power-on indicates a problem — power should not be dramatically dissipated in a quiescent unprogrammed board.

Power Rail Verification

10. Measure all power supply rails. With the board powered and stable, measure every supply rail with a multimeter:

• Is the 3.3 V rail actually 3.3 V (within ~2%)?
• Is the 5 V rail correct?
• Is any reference voltage (ADC VREF, sensor supply) at the expected level?

A rail at the wrong voltage is a high-priority problem — it usually means a feedback resistor value error, a wrong regulator part, or a resistor shorts that shifts the output.

11. Measure power supply ripple with an oscilloscope. On switching regulators, verify ripple is within the datasheet specification. On LDOs, measure the output noise level. See oscilloscope basics for measurement setup. Significant ripple at unexpected frequencies can indicate oscillation or an incorrect output capacitor value.

12. Verify power sequencing (if applicable). On boards with multiple supply rails, check that rails come up in the correct order and within the required timing window. Some processors have strict sequencing requirements for core voltage and I/O voltage — violating sequencing can cause latch-up or permanent damage.

Firmware Loading

13. Connect the debug probe. Connect JTAG/SWD to the debug header. Verify the probe is recognised by the host software (STM32CubeIDE, Ozone, OpenOCD, etc.).

14. Confirm debug connectivity. Most debug tools can read a MCU device ID before loading firmware — confirm the correct device is identified. If the probe cannot connect, check: SWD signal integrity (oscilloscope on SWDIO and SWCLK), MCU supply voltage, clock source availability (HSI is usually available at reset without external crystal).

15. Load and run minimal firmware. Start with the simplest possible firmware — a GPIO blink on an LED, or a UART "hello world" — not the full application. This confirms that: the MCU runs, the clock tree is functional, and at least one peripheral works. Only add application complexity once this baseline is confirmed.

Peripheral Validation (Subsystem by Subsystem)

16. Validate one peripheral at a time. Don't load the full firmware application and hope everything works. Write or download simple test firmware that exercises each peripheral independently:

• UART: Print a counter over UART at 1 Hz. Verify the correct baud rate on a logic analyser or USB-UART adapter.
• SPI/I2C: Read a device ID register from an external component. Confirm the logic analyser shows the correct transaction and the expected response byte.
• ADC: Measure a known voltage (e.g. connect a voltage divider to a fixed 3.3 V rail) and confirm the ADC reading matches the expected value.
• PWM: Output a 1 kHz PWM signal and verify the frequency and duty cycle on an oscilloscope.
• GPIO interrupts: Toggle an input pin externally and confirm the ISR fires with the expected latency.

17. Confirm crystal / oscillator. If the design uses an external crystal oscillator, verify it is running at the correct frequency using an oscilloscope on the MCU's OSC_IN or XO pin. A crystal that is not starting, starting at the wrong frequency, or oscillating erratically is a common first-spin issue — incorrect load capacitor values, PCB trace length, or a wrong crystal type being the typical causes.

18. Check thermal performance under load. Once all peripherals are validated and the application firmware is running, operate the board at representative load for 10–15 minutes and check temperatures. Use a thermal camera if available, or a temperature probe on key components (regulators, power MOSFETs, processors). A component running significantly hotter than its datasheet's expected junction temperature at the actual power dissipation indicates a cooling problem.

Bring-Up Notes and First-Article Report

19. Document every finding. Keep a log of what was tested, what worked, what was marginal, and what required correction. This brings-up report becomes the basis for the board revision, guides the production test specification, and is the institutional memory for the next revision.

For Zeus Design customers, rapid prototyping services include first-article bring-up and a written test report covering all steps above — shortcutting the time from "boards arrived" to "confirmed working."

Design Considerations

• Design for bring-up: A debug header (SWD minimum), UART test point, accessible power rail test points, and a power LED dramatically reduce bring-up time. A board with none of these requires destructive probing to access basic debug signals.
• Place current sense options on power rails: A zero-ohm resistor footprint in series with each primary supply rail allows a multimeter to be inserted for current measurement during bring-up without modification, then the zero-ohm is loaded for production.

Common Mistakes

• Applying full supply voltage to a board that has not been inspected for shorts — a reversed electrolytic or solder bridge at first power-on can destroy the regulator, MCU, or both before the fuse responds.
• Loading the full application firmware before validating individual peripherals — when something doesn't work, it's much harder to isolate whether the fault is in the peripheral configuration, the application logic, or the hardware.
• Skipping thermal validation — a component that is only marginally over-dissipated (running at 75°C when the rating is 70°C) may work reliably for weeks in the lab and fail in the field after a year of slightly higher ambient temperature.
• Not documenting what was found and corrected — the next engineer to work on the board (or the same engineer in six months) loses valuable bring-up insights that would have sped up the next revision.