A “dead” PCB should not be diagnosed by randomly replacing components. The fastest method is usually to follow the board’s energy path: input → protection → power rails → reset/clock → processor activity → peripheral sections. At each stage, make one measurement that tells you whether the fault is before or after that point.
Safety boundary: this workflow is for low-voltage electronics you are qualified to service. Do not probe live mains, high-voltage power supplies, charged capacitors, or unknown-energy circuits unless you have the training, isolation equipment, PPE, and procedures for that work.
1. Define “dead” before measuring
“Nothing happens” can mean very different faults:
- no input power reaches the board;
- the main rail is shorted;
- a protection device has opened;
- a regulator has no output;
- the MCU is held in reset;
- the MCU is running but the display/communications/output section has failed.
Record the symptom first. Does current draw stay at zero, hit the bench-supply current limit, pulse repeatedly, or look normal? Does any LED flash briefly? Is there heat anywhere? Those observations decide the next test.
2. Inspect the board with power removed
Start visually before applying power. Use good lighting and magnification. Look for:
- burned or cracked components;
- corrosion or conductive contamination;
- lifted pads and broken traces;
- reversed electrolytic capacitors or diodes;
- damaged connectors;
- solder bridges, especially on fine-pitch ICs;
- loose inductors, transformers, relays, and large connectors;
- evidence of liquid ingress or overheating.
If a schematic or boardview exists, mark the input rail, protection components, regulators, major power domains, reset pin, clock source, and programming/debug connector before taking measurements.
3. Check for an obvious short before powering
With the circuit safely de-energized and capacitors discharged, measure resistance from the main input rail to ground and from major regulated rails to ground. A very low reading can indicate a short, but do not use one fixed resistance threshold for every design: processors, radios, FPGAs, and large capacitor networks can legitimately appear low depending on the measurement method.
Compare suspicious rails with known-good documentation or another working board when available. If the resistance changes as capacitors charge from the meter, that behavior is different from a hard short that remains low.
4. Verify the input path
Apply the correct supply using current limiting when appropriate. Measure at the connector first, then move through the protection path:
- input connector voltage;
- fuse or resettable fuse;
- reverse-polarity protection;
- input diode/MOSFET path;
- transient protection area;
- regulator input.
The key technique is to measure on both sides of each series element. If the correct voltage exists before a component but not after it, you have localized the problem far more effectively than replacing parts around the board.
5. Check every power rail in order
Once the regulator input is correct, measure each output rail against its expected value. Typical boards may have 5 V, 3.3 V, 1.8 V, 1.2 V, analog rails, or separate sensor/radio supplies.
| Observation | Likely direction |
|---|---|
| No regulator input | Fault is upstream in connector/protection/power path. |
| Correct input, zero output | Check enable pin, shorted load, regulator fault, startup conditions. |
| Output rises then collapses | Check current limit, short/overload, brownout, unstable enable sequence. |
| Rail is high | Stop; wrong feedback/regulator/assembly can damage downstream ICs. |
| Rail correct at regulator but low at load | Check damaged trace, connector, ferrite bead, series resistor, or excessive load. |
Do not inject voltage blindly into a shorted rail. Current-injection fault finding can be useful in experienced hands, but the safe voltage/current limit depends on the rail and downstream components.
6. Use current draw as a diagnostic signal
A current-limited bench supply can reveal useful patterns:
- Near-zero current: open input path, disabled regulator, disconnected load, or board not actually powered.
- Immediate current limit: possible short, reversed device, failed IC, or incorrect supply voltage.
- Repeated rise/fall: regulator hiccup, brownout-reset loop, or load repeatedly starting and collapsing the rail.
- Normal-looking current but no function: move on to reset, clock, firmware/debug, and interfaces.
Record the current instead of relying on “it seems high.” Comparing a failing board with a known-good board is especially useful.
7. Check reset and enable signals
If the rails are valid, check whether the processor is allowed to run. Measure the MCU reset pin and any power-good or enable signals involved in startup.
A reset line permanently asserted can come from a supervisor IC, external reset circuit, programmer/debugger, low power rail, watchdog behavior, or damaged MCU. Use the schematic to determine the active polarity and expected voltage instead of assuming every reset pin behaves the same way.
8. Check the clock only after power/reset
If the design uses an external oscillator/crystal and it is required at startup, verify it with an appropriate oscilloscope/probing method. Poor probing can disturb crystal circuits, so use the MCU documentation and proper probes.
Some MCUs start from an internal oscillator and switch clocks later, so “no crystal waveform” is not automatically proof of a hardware failure. Confirm the architecture first.
9. Try the programming/debug interface
A debugger can divide the problem quickly:
- Can it identify the target device?
- Can it read the device ID?
- Can it halt the core?
- Is flash readable/programable?
- What is the reset cause?
- Where is the program counter?
If SWD/JTAG works and the CPU is executing, the board is not truly “dead.” The fault may be firmware, boot configuration, peripheral initialization, communications, display, or the external load.
10. Divide the board into functional sections
Once core power and execution are proven, stop troubleshooting the entire PCB at once. Test one section at a time:
- sensor supply and bus activity;
- USB/UART/CAN/Ethernet transceiver;
- display/backlight;
- motor/relay/driver stage;
- analog front end;
- external memory;
- radio module.
For each section, verify power → enable/reset → input signal → output signal. This creates a boundary around the fault.
11. Use thermal clues carefully
An unexpectedly hot component can identify a shorted or overloaded section. Compare temperatures with a known-good board and remember that some regulators, processors, and power devices normally run warm.
Power the board only within safe limits while looking for abnormal heating. Do not keep a faulted board energized long enough to turn one failure into several.
12. Repair only after the fault is localized
Before replacing a component, collect evidence that points to it. A regulator with correct input, correct enable, a healthy load resistance, and no output is a much stronger suspect than a regulator chosen because it “looks related to power.”
After repair:
- repeat resistance checks with power off;
- power with current limiting;
- verify all rails;
- confirm reset/clock/debug behavior;
- run the original failed function;
- run a short thermal/load test;
- record what failed and why.
Fast diagnostic sequence
Visual → resistance/short check → input voltage → protection path → regulator inputs/outputs → current draw → reset → clock → debugger → peripheral sections → final functional test.
This sequence prevents the most common troubleshooting mistake: jumping into the middle of the board without first proving that the earlier stages are healthy.
For a related method, see why replacing random components is the worst PCB repair strategy. If the board is your own design, use the PCB manufacturing review checklist before the next revision.