Home › Guides › PCB Troubleshooting

PCB Troubleshooting: A Systematic Method That Finds the Fault

A board that does not work is not a mystery to be solved by luck. It is a chain of stages, and somewhere in that chain something is broken, missing, shorted or wrong. Troubleshooting is the discipline of narrowing that chain down, one test at a time, until only the fault is left. This guide gives you that method: what to check, in what order, and what the common faults actually look like when you find them.

The golden rule: divide and conquer

Every troubleshooting method that works follows the same idea. You split the circuit in half and test which half the fault is in. Then you split that half again. A board with a hundred components becomes ten tests instead of a hundred guesses, because each test eliminates a whole section at once. The practical version of this is signal tracing: feed a known signal in at the input, and follow it through the circuit stage by stage until it disappears or goes wrong. The stage where the signal breaks is the stage with the fault, and everything before it is proven good.

The opposite approach — replacing parts at random, resoldering everything, or turning up the supply until something happens — is how boards get damaged beyond repair. Pads lift, neighbouring parts get cooked, and the original fault gets buried under new damage. Resist the urge. The method below is slower for the first five minutes and far faster for the remaining hour, and it leaves the board in a state you can still fix.

Safety first, even on low-voltage boards

Most boards you will troubleshoot run on low voltages, but that does not make them harmless. Mains-powered boards, boards with large capacitors, and boards connected to motors or batteries can all store enough energy to hurt you or destroy test equipment. Before you touch anything, disconnect the power, discharge large capacitors through a resistor rather than a screwdriver, and check that the supply you are about to connect is set to the right voltage and current limit. A current-limited bench supply is your best friend here: set the current limit low, and if the board pulls the full limit immediately, you have a short and the supply has just saved your board from burning. Never defeat a current limit to force a board to run.

Electrostatic discharge is the other quiet killer. CMOS parts, MOSFETs and many modern integrated circuits can be damaged by a static shock you cannot even feel. Work on an earthed mat with a wrist strap when handling bare boards, keep boards in antistatic bags when they are not on the bench, and avoid touching component pins directly. A board damaged by static often fails intermittently or weeks later, which makes it one of the hardest faults to trace back to its cause.

Step 1: visual inspection

Start with your eyes, before any instrument. Look at the board under good light, ideally with magnification, and work systematically across it. You are looking for the faults that testing often misses or misleads you about. Solder bridges — a thin smear of solder joining two adjacent pins or pads — are the single most common assembly fault on fine-pitch parts, and they are often visible before any meter confirms them. Cold joints look dull, grainy, cracked or balled up instead of smooth, shiny and concave; they may not wet one side of the joint at all. Missing components are surprisingly common on hand-assembled boards, as are components fitted in the wrong orientation: electrolytic capacitors, diodes and integrated circuits all have polarity marks, and a reversed part is a fault even if it looks soldered perfectly.

Also look for physical damage: cracked ceramic capacitors, chipped component packages, lifted pads, scratched traces, and burn marks or discolouration that point to an overheated part. Check connectors for bent pins and for solder that has wicked up into a connector where it should not be. Compare the board against the bill of materials and the silkscreen, part by part, if you can. Many faults are found at this stage alone, and even when they are not, the inspection gives you a mental map of the board that makes every later test faster.

Step 2: power rails, unpowered

Before applying power, check the power rails with a multimeter on resistance or continuity mode. Measure between each supply rail and ground. A healthy rail reads some resistance — often hundreds of ohms or more, depending on the circuit — and that reading usually rises as capacitors charge from the meter. A reading near zero ohms is a short, and it means something on that rail is failed, bridged or fitted wrong. Do not power the board until you understand that short, because applying a supply to a shorted rail drives large currents through the fault and can turn one failed component into several.

If you find a short, narrow it down before removing anything. Look for solder bridges on the rail, check the polarity of every polarised component on it, and consider the usual suspects: a reversed tantalum capacitor fails short very reliably, and a damaged voltage regulator can short input to output or output to ground. A useful technique is to measure the resistance from several points along the rail; the short is usually closest to the point of lowest resistance. Another is to apply a very low, current-limited voltage and feel — carefully — for the component that warms up, or use a thermal camera if one is available. The warm part is dissipating the fault current, and it is usually the culprit or very close to it.

Step 3: power rails, powered

Once the unpowered checks pass, apply power through a current-limited supply and watch the current. A board that draws roughly its expected current is behaving. A board that draws nothing may have an open input: a blown fuse, a broken trace, a faulty connector or a switch that is off. A board that draws far too much has a partial short or a failed part that the unpowered check did not catch, because some faults only appear under voltage.

Now measure the voltages, in order, from the input to the output of the power chain. Check the input voltage at the board connector, then at each regulator's input and output, then at the supply pins of the main integrated circuits. The fault is at the first point where the voltage is wrong, and everything upstream of it is good. A regulator with the right input but no output is either failed, shut down by an enable pin, or loaded by a short downstream — and you can tell those apart by checking the enable signal and by lifting the load, if the design allows it. Write the voltages down as you go. A written record turns a confusing board into a map, and it is what lets you pick up the job again after a break without starting over.

Step 4: continuity and component checks

With power off again, use continuity mode to verify the connections that matter. Check that traces are continuous where they should be and open where they should not, paying attention to vias and to any area that has been reworked, since rework is where traces get cut and pads get lifted. Check connectors pin by pin against the schematic, because a swapped or miswired connector is invisible to every other test until you do this.

Test individual components where you can, remembering that in-circuit measurements are affected by everything connected to the part. A resistor measured in circuit reads lower than its marked value if other paths run in parallel with it, so in-circuit readings are a guide, not a verdict. Diodes and transistors can be checked with the diode mode of a multimeter, which shows the forward voltage drop; a short in both directions or an open in both directions is a strong sign of failure. Capacitors are harder in circuit: a shorted capacitor shows up on the resistance check, but a capacitor that has lost capacitance or gone high in series resistance usually needs to be removed for a proper test. When a reading is ambiguous, removing one leg of the part — or the whole part, if you can replace it — is the honest way to settle it.

Step 5: signal tracing

Now follow the signal. For an analogue board, inject a known signal at the input — a sine wave from a signal generator, or simply the board's normal input — and measure it with an oscilloscope at each stage: input, first amplifier, filter, second amplifier, output. The signal should change in the way each stage is designed to change it: amplified, filtered, shifted, or converted. Where the signal disappears, distorts unexpectedly, or stops changing as it should, you have found the faulty stage. Then work inside that stage: check its supply, its biasing, its components and its solder joints, in that order.

For a digital board, the same idea applies with logic levels and timing. Check that clocks are present and at the right frequency, that reset signals release properly, that enable lines are in the right state, and that data moves between the parts that should be talking. A logic analyser or even a multimeter on a toggling pin can tell you whether a microcontroller is running at all: a dead microcontroller often shows no activity on any pin, while a running one shows toggling on clocks and data lines. Communication buses deserve special attention, because a single stuck line — a data line held low by a failed part or a solder bridge — can silence a whole bus and make every device on it look dead.

The common faults, and what causes them

Most boards you troubleshoot will fail for one of a short list of reasons, and recognising them saves enormous time. Solder bridges, especially on fine-pitch and surface-mount parts, short adjacent pins and usually come from too much paste, a misaligned stencil, or a part placed slightly off. Cold joints, dull and grainy, come from insufficient heat, a contaminated pad, or movement while the joint was cooling; they cause intermittent faults that change with temperature and vibration. Wrong parts — a resistor of the wrong value, a capacitor with the wrong dielectric, an integrated circuit from the wrong family — come from picking errors or from a bill of materials that was not updated, and they produce boards that are wrong in a consistent, repeatable way rather than intermittent.

Reversed parts are their own category: diodes, LEDs, electrolytic and tantalum capacitors, and many integrated circuits have a polarity, and fitting one backwards usually destroys it and sometimes its neighbours. Damaged parts come from handling, from overvoltage during testing, or from a previous fault that cascaded. And design faults, though rarer, do happen: a trace routed too close to a noisy source, a decoupling capacitor omitted, a pin left floating that should have been tied off. When you have proven every component and joint good and the board still misbehaves, the schematic itself becomes the suspect, and that is a different conversation from a repair.

Intermittent faults: the hard ones

A board that fails sometimes is harder than a board that fails always, because every test you run may pass while the fault is hiding. The key is to find what makes it fail and then hold it in that state. Temperature is the most useful lever: a hot-air gun or a can of freeze spray, applied to one area at a time, will often push a marginal joint or a temperature-sensitive part over the edge. Mechanical flex is the second lever: gently flexing the board, or tapping suspect areas, reveals cracked joints and broken traces that only open under movement. Vibration, supply variation and time all work the same way. Once you can make the fault appear on demand, it becomes an ordinary fault, and the ordinary method finds it. Until then, be patient and systematic, and resist the temptation to declare the board fixed because it passed once.

Repair, retest, record

When you find the fault, repair it properly rather than quickly. Remove the failed part cleanly, clean the pads, fit the replacement with the right orientation, and solder it with a joint you would be happy to inspect. Then retest the whole chain, not just the repaired stage: a repair can introduce its own faults, and the board is only fixed when it passes the same tests it failed before, plus a functional check of everything around it. Finally, record what you found: the symptom, the tests, the faulty stage and the cause. That record is what turns one repair into knowledge, because the same fault on the next board will be found in minutes instead of hours. Troubleshooting is a skill, and like every skill it compounds — but only if you write down what you learned.

Frequently asked questions

For how boards are made in the first place, see PCB manufacturing process and PCB design with AI. For the wider picture of AI in engineering work, see AI predictive maintenance and AI in power systems.