PCB Troubleshooting: Safe Tests That Find Root Causes
PCB Troubleshooting: Find the Fault Before Replacing Parts
Effective PCB troubleshooting narrows a fault with evidence before changing the assembly. Begin with documented symptoms and unpowered checks, proceed to controlled startup only when appropriate, and measure the signals most likely to distinguish competing explanations. Replacing the hottest or most suspicious-looking component first can destroy useful evidence without fixing the cause.
The workflow below concerns low-voltage electronics handled by trained personnel. Mains-connected equipment, high-energy batteries and hazardous stored energy require product-specific safety procedures and qualified service; a current-limited supply is not a substitute for those protections.
Capture a repeatable symptom
Record the board revision, firmware, supply, load, temperature and sequence that produces the failure. Distinguish “does not start” from “starts and resets when the load connects.” Those observations suggest different measurements.
Check the test setup as well as the board. A cable, fixture, supply setting or software mismatch can reproduce a convincing hardware fault. If a known-good assembly is available, compare it under the same conditions and avoid swapping several variables at once.
Preserve photographs and initial readings. A board that works after being cleaned or flexed has changed state, but the change is not yet a root-cause explanation.
Inspect before applying power
Disconnect all relevant sources and follow the approved discharge procedure. Confirm the safe state with appropriate equipment before touching or measuring the circuit. Stored charge can remain after the main connector is removed.
Under suitable lighting and magnification, check polarity, fitted part values, connector orientation, damaged components, solder bridges and disturbed lands. Compare with the schematic and assembly data, not the silkscreen alone.
A dull lead-free joint is not automatically defective. Visual assessment must consider the joint's geometry, wetting and applicable acceptance requirements. Hidden joints and internal board defects may require other inspection methods; absence of a visible flaw does not establish that the assembly is sound.
Keep resistance tests unpowered
Measure resistance, continuity and diode behavior only in the appropriate de-energized, discharged state. Fluke's resistance-measurement guidance calls for removing power and discharging capacitors before resistance testing. Fluke resistance measurement guide.
A continuity beep is not a diagnosis. Low-voltage processor rails can have low resistance, capacitors can change the reading as they charge from the meter, and semiconductor junctions make polarity matter. Parallel paths can also prevent an in-circuit reading from representing one component.
Use the schematic and comparison measurements to interpret the result. An unexpected open connection and an unexpectedly low rail resistance are useful clues, but each needs a circuit-specific explanation. Do not apply an arbitrary resistance threshold to every rail.
Progress to powered measurements only after the safety and unpowered checks support doing so. Preserve the observations from each stage.
Separate controlled startup from signal diagnosis
For a suitable low-voltage assembly, verify polarity, input rating and the intended current-limit setting before energizing it. Monitor supply behavior and stop if the observed condition falls outside the test plan. A low current limit can itself prevent normal startup, so interpret limiting behavior rather than assuming it proves a short.
Check input voltage at the board, regulator inputs and outputs, enables, power-good signals and reset sequencing as relevant. A rail at 0 V can result from a disabled regulator, missing input, protection behavior or a short. That reading alone does not identify which one.
Only after the supplies are understood should you pursue clocks, buses and analog signal paths. Use the device documentation to select meaningful test points and expected behavior. Capture the fault event instead of relying exclusively on a steady-state reading after the reset has already happened.
Make sure the instrument is not creating the fault
An earth-referenced oscilloscope probe ground can short a circuit node if connected incorrectly. Do not defeat the oscilloscope's protective earth. Use appropriately rated differential or isolated measurement equipment when required by the circuit. Tektronix explicitly warns against floating a grounded oscilloscope by disabling its protective grounding. Tektronix measurement-safety guidance.
Probe loading and ground-lead inductance can also distort a waveform without creating an obvious failure. Compare measurement arrangements before blaming the board for every observed spike. A logic analyzer can help interpret digital transactions, but it does not replace an analog view when voltage levels or edge quality are suspect.
Thermal imaging can identify a region for investigation. It does not prove the warmest device is defective: a functioning regulator may be dissipating power because a downstream load is abnormal. Surface finish and emissivity also affect apparent temperature.
Test an explanation that can be disproved
Suppose a hypothetical controller resets when a radio transmits. One explanation is a supply dip; another is a reset-line disturbance. Capture both the local rail and reset behavior around the event with a suitable measurement setup. If the rail remains within its required envelope while reset changes unexpectedly, the first explanation loses support.
Continue with one controlled change at a time. Check whether the same fault occurs across several assemblies, one production lot or one panel position. A common footprint error, an intermittent joint and an incorrect firmware configuration need different corrective actions.
Avoid repeatedly reflowing parts until the symptom disappears. Temperature and mechanical disturbance can temporarily change intermittent faults, making the board harder to diagnose afterward.
Close the loop after the repair
Repeat the original failing condition, verify the intended fix and check functions the intervention may have disturbed. Document the evidence connecting cause to correction. A single successful boot is not equivalent to reproducing the original acceptance test.
For recurring defects, feed the findings into design rules, assembly process controls, inspection or test coverage. There is no universal rule that a board must be redesigned after exactly two repairs. The decision depends on the failure mechanism, product requirements and whether the process can be controlled.
FAQ
Does a continuity beep between power and ground prove a short?
No. Meter thresholds, capacitors, semiconductor paths and normal low-resistance loads affect the result. Interpret the reading with the unpowered circuit's schematic and suitable comparison data.
Should I replace the hottest component first?
Not without further evidence. It may be operating normally or reacting to another fault. Check its operating conditions and the surrounding circuit before changing it.
Use the Findings in Your Next Build
Share your PCB manufacturing or PCBA assembly requirements with icallin, including any revised layout, assembly notes or test requirements from the investigation.
