BGA Repair and Board-Level Fault Finding
BGA repair on a board with a ball grid array package is less about the package than about the circuit around it. Most faults arrive as a symptom rather than as a diagnosis: a unit that sometimes starts, a supply that will not come up, a rail that is shorted. The techniques below are the ones that find the cause quickly, and they are mostly about knowing which component type fails in which way.
Capacitors Fail First
Capacitor failure is the most common cause of electronic faults, and electrolytic capacitors are the most common of all. The failure modes are a loss of capacitance, a complete loss of capacitance, leakage, and a short circuit, and the symptom depends on the role the capacitor plays. In a digital board most capacitors are supply filters, so a failed electrolytic capacitor on a switching supply may prevent it from starting at all, or may leave the output poorly filtered so that the logic behaves erratically and the unit works only sometimes.
Heat is the governing factor. Capacitor life falls as ambient temperature rises, and the rule applies to every type, not only to electrolytics. That is why a fault-finding pass should start with the capacitors closest to heat sources: beside a heatsink, next to a power device, or anywhere the airflow is poor. A capacitor that has been cooked often shows a bulged case or leaked electrolyte, but a capacitor that has simply lost capacitance can look normal, which is why measuring it is more reliable than looking at it.
What Resistor Failures Look Like
Resistors are the most numerous component and not the most failure-prone, so time spent removing them is usually wasted. When one does fail, an open circuit is by far the most common outcome, a rise in resistance is less common, and a fall in resistance is rare. The pattern also depends on value: low-resistance parts below about 100 ohms and high-resistance parts above about 100 kilohms fail more often than mid-range values.
Appearance helps in one case and not the other. A failed low-value resistor is usually burnt and blackened, so it is visible immediately. A failed high-value resistor often shows no mark at all. Wirewound types, used for current limiting, may show blackening, cracked coating, or a break, and fusible resistors may lose a piece of their coating or show nothing. In practice the resistor is most usefully checked with a meter, comparing the reading with the marked value.
Checking an Operational Amplifier In Circuit
An operational amplifier used linearly is always in a closed loop with negative feedback, and that gives a way to test it without removing it. In a working linear stage the two input terminals sit at almost the same voltage, typically within a few millivolts; in a high-impedance circuit the meter itself loads the node slightly, but the difference should still stay well below a fraction of a volt. A difference of half a volt or more means the device has failed.
The rule changes when the same device is used as a comparator. There the inputs are deliberately unequal: if the non-inverting input is above the inverting input, the output sits near its positive limit; if it is below, the output sits near zero or the negative limit, depending on whether the supply is single or split. A measurement that does not follow that relationship means the device is faulty. Testing this way avoids both removing the part and the substitution method.
Probing Small Surface Mount Parts
Fine surface mount components are awkward to probe: ordinary meter tips are wide enough to bridge two pads, and an insulating coating prevents contact with the terminals. A simple aid is to bind a fine sewing needle to each meter probe with a strand of fine copper wire and fix it with solder. The needle point reaches the component without shorting its neighbours, and it penetrates a conformal coating instead of requiring it to be scraped away. Sharper tips also make it practical to measure a part in circuit without disturbing it.
Finding a Shorted Supply Rail
A shorted supply rail is difficult because every device connected to that rail is a suspect. On a board with few components, checking them one at a time eventually finds the culprit; on a dense board the search becomes a matter of luck. A more effective method uses a variable supply with both voltage and current control, set to the rail voltage with the current limit turned down to its minimum. Apply the supply to the rail and raise the current gradually, then feel the components. The one that becomes noticeably warm is drawing the fault current and is almost certainly the damaged part.
Two precautions govern the technique. The applied voltage must never exceed the working voltage of the rail, and the polarity must be correct, or a healthy device will be destroyed by the test itself. Within those limits, the method finds the fault in a fraction of the time that removing and replacing parts would take.
Contact Problems and Gold Fingers
Boards that plug into a backplane through an edge connector are vulnerable to their environment. Dust, humidity, and corrosive gases in an industrial installation all attack the contact surfaces, and the result is an intermittent connection rather than a clean failure. Replacing the card solves it and is expensive, particularly for imported equipment.
Before that, clean the gold fingers. An ordinary soft eraser rubbed along the contact area removes the film of contamination that has built up, and the card often works again. It is a crude remedy, and it is worth trying before any part is replaced, because the fault lives at the contact interface rather than inside the circuit. Where the problem recurs, the longer-term answer is a protective coating or a change of contact material, which is a design decision rather than a repair; the same reasoning is covered in conformal coating for board protection.
Diagnosing Intermittent Faults
A fault that comes and goes usually has one of five causes, and they can be told apart. The first is a poor contact: a card that does not seat properly in its slot, a broken conductor inside a cable, a terminal that does not grip, or a cold solder joint. These respond to movement and flexing, which makes them findable.
The second is interference. A digital circuit may only fail under the conditions that couple enough noise into it to push a marginal node past its threshold, and the circuit itself may have drifted toward that margin rather than having been designed there. The third is thermal instability, and among components the electrolytic capacitor leads the list, followed by other capacitors, transistors, diodes, integrated circuits, and resistors. The fourth is moisture and dust on the board, which are conductive, present a resistance that changes with temperature, and act in parallel with the circuit, altering its behavior. The fifth is the firmware: a parameter given too little margin reports a fault only when the operating condition reaches the threshold the software uses to declare one.
Working through those five in order is faster than swapping parts, and each leaves a different signature. A fault that appears when the board is tapped points at the first, one that appears at a particular temperature points at the third, and one that appears in a particular environment points at the fourth. Related assembly defects and their causes are described in SMT component shift causes and lead-free versus leaded solder.
FAQ
Why check capacitors before anything else? Because capacitor failure is the most common cause of electronic faults, and electrolytic capacitors fail more often than any other type. Heat accelerates the failure, so parts near a heatsink are the first candidates.
Can an operational amplifier be tested without removing it? Yes. In a linear stage the two inputs should sit within millivolts of each other. A difference of half a volt or more indicates a failed device.
How do I find a shorted rail on a dense board? Apply the rail voltage through a current-limited supply and raise the current slowly, then look for the component that becomes warm. It is drawing the fault current.
Which intermittent faults are hardest to find? Those caused by moisture and dust, because the leakage resistance changes with temperature and humidity and produces a fault that appears and disappears with the environment.
