Failure Analysis · Diagnostics

The 5 most common reasons
electronics fail at board level

May 02, 2026 Level: Beginner Failure Modes

Understanding root cause failure

PCB failures follow predictable patterns. Identifying these patterns accelerates diagnosis and repair. Failure modes are not random—they cluster around five dominant mechanisms that account for over 80% of field failures across consumer electronics, industrial control boards, and mobile devices.

This guide maps each failure mode to real symptoms, test methodology, and corrective action. Each mechanism has distinct signature characteristics: electrical signatures on rails, visual thermal patterns, component degradation markers, and measurable resistance shifts.

Understanding the difference between failure mechanism (how it fails) and failure mode (what you observe) is critical for efficient diagnosis. A dead PPBUS rail can result from five different mechanisms—each requiring different troubleshooting paths.

Failure 1: Thermal stress and thermal cycling

Thermal cycling kills more boards than any other single failure mode. Temperature differentials create mechanical stress at solder joints, decoupling capacitor leads, and ball grid array (BGA) interconnects. Coefficient of thermal expansion (CTE) mismatch between copper (17 ppm/°C), silicon (2–4 ppm/°C), and FR-4 substrate (15–16 ppm/°C in XY plane, 60–80 ppm/°C in Z) generates shear stress at interfaces.

Signature characteristics:

  • Intermittent cold-start failures; board operates after warm-up
  • Resistance reading on power rail test point creeps upward over time (e.g., 0.45Ω at power-on, 0.89Ω after 30 seconds)
  • Voltage rail droops appear only on second/third cycle, not first boot
  • Micro-fractures at solder fillet bases, visible under magnification as concentric stress lines

Thermal stress concentrates near high-current switching nodes. ISL6259 buck regulators and TPS51125 VRM controllers generate 3.5–5.2V of heat during transients. The solder interface between CPU power connector and PCB substrate is a primary failure site because thermal mass is unbalanced: the processor dumps heat into the BGA while the connector receives minimal thermal coupling to the board.

Micro-fractures do not show up on resistance measurements until mechanical stress is reapplied. A board can read 0.02Ω continuity cold, then measure 3.5Ω under thermal load. Always test under operating temperature, not room temperature alone.

Repair approach:

Reflow the affected node. Cold solder joints and stress fractures at BGAs and high-current connectors require heat gun reflow (250–280°C) or rework station treatment. Clean flux residue before final inspection.

Failure 2: Power delivery failures and rail collapse

Dead or sagging rails cause 25–30% of board failures. Failure sources include shorted output capacitors, failed PWM controllers, shorted Schottky diodes in synchronous buck stages, and MOSFET gate failures. The distinction matters: a shorted output capacitor fails instantly and reads near-zero impedance; a failed gate driver fails progressively as switching losses accumulate.

Signature characteristics:

  • Rail reads 0V–0.2V under load despite PWM IC reporting correct switching frequency
  • Inductor current saturates; inductor temperature exceeds 60°C within 10 seconds
  • Low-side MOSFET gate-source voltage measures 2.1–3.8V instead of nominal 8–12V
  • Output capacitor ESR swings erratically (e.g., 18 mΩ, 890 mΩ, 12 mΩ on successive measurements) indicating internal delamination

Capacitor aging is predictable: electrolytic and hybrid polymer capacitors lose capacitance and gain ESR over 5–8 years at typical operating temperature. A 100µF/10V capacitor specified at 85°C will reach end-of-life (20% capacitance loss, 3× ESR increase) by 8 years. On mobile devices running at sustained 45–55°C, this timeline compresses to 2–4 years.

Measure capacitor impedance (ESR + XC) at the actual operating frequency using an ESR meter, not a multimeter. Multimeters cannot detect capacitor aging because they test at DC. A capacitor can read "OK" on continuity while displaying 2.2Ω ESR at 100 kHz switching frequency.

Repair approach:

Replace tandem output capacitors on buck converter outputs. Capacitors age together; replacing one without the second will result in rapid re-failure. Test PWM IC switching frequency and duty cycle on the GATE signal (scope check: 0V–5V square wave at expected frequency, typically 300 kHz–2 MHz). If PWM output is missing, replace the controller IC.

Failure 3: Corrosion, contamination, and electrochemical migration

Moisture + flux residue + applied voltage = electrochemical migration. Conductive filaments grow between solder pads at 3.3V–5V bias. Migration occurs fastest on high-density BGA pads (0.8 mm pitch or finer) where interstitial spacing is under 0.2 mm. Under humid conditions (>85% relative humidity) with inadequate solder mask coverage, migration initiates within weeks.

Signature characteristics:

  • Intermittent short between adjacent power planes or signal nets
  • Resistance between two pads drops from >10 MΩ to 50–500 Ω over days or weeks
  • Conductive filament visible under 20× magnification as a whisker or dendritic structure
  • Flux residue visible between pads; rosin flux appears amber/brown; no-clean flux appears translucent with crystalline deposits

Manufacturing process violations accelerate migration: inadequate reflow profile (peak temperature too low, dwell time <10 seconds), post-assembly contamination, and improper storage (>50% RH without desiccant). Wave solder machines generate more residue than reflow, increasing migration risk by 3–5×.

Conformal coating provides only partial protection. It slows but does not prevent electrochemical migration because ions dissolve into the coating during humid operation. Migration under coating is often invisible until failure occurs.

Repair approach:

Clean affected BGA with isopropyl alcohol (99%+ purity) and a soft brush. Work under magnification to dissolve flux residue without shorting pins. If migration is extensive (multiple dendritic growths across large area), replace the BGA or affected connector. Prevention: maintain assembly area humidity <60% RH and ensure solder mask coverage on high-density areas.

Failure 4: Component degradation and age-induced drift

Passive components degrade predictably. Electrolytic capacitors lose capacitance at ~1% per year at 70°C, accelerating to ~2% per year at 85°C. Tantalum capacitors fail catastrophically (usually short) after 10–15 years. Film capacitors are stable but ceramic X5R/X7R capacitors drift capacitance by ±10% over rated temperature range and can exhibit aging drift (capacitance loss unrelated to temperature, ≈0.5–3% per year).

Signature characteristics:

  • Output voltage on regulated rail drifts high (e.g., 3.35V on nominal 3.3V rail) due to feedback network capacitor drift
  • Decoupling capacitors on high-current rails no longer absorb transient current; noise on rail spikes to 200–400 mV during load changes
  • Timing or frequency errors accumulate: clock oscillator frequency drifts by >500 ppm
  • Resistor dividers shift value by 2–5% due to metal film resistor tolerance creep in high-temperature environments

Critical-path failures occur when multiple components age simultaneously. A voltage regulator with drifted feedback resistors (drift +2%) and aged output capacitor (drift +3%) together produce output voltage drift of +5%, pushing the rail outside valid operating range.

Capacitor datasheet aging curves assume constant temperature. Boards experiencing day/night thermal cycling (e.g., outdoor equipment) age faster than boards in climate-controlled environments. Cycling accelerates migration rate by 2–3× because each thermal cycle reactivates ion mobility at stress boundaries.

Repair approach:

Replace all tandem capacitors in feedback networks and decoupling zones on old boards. Replace tantalum capacitors with ceramic alternatives (100 µF+ in 1210 case, rated ≥20V). Verify output rail voltage post-replacement with 10-minute load soak test to confirm stabilization.

Failure 5: Design defects and marginal specifications

Design margins are often inadequate. A regulator IC specified for operation at 3.0–3.6V output may exhibit instability near the limits if loop compensation is not properly tuned. Inadequate decoupling, under-sized heat sinks, and non-optimal PCB layout create early-onset failures that appear after 100–2000 hours of use, not in the first week.

Signature characteristics:

  • Failure rate peaks between 30–500 operating hours (infant mortality curve), not at hour 1
  • Failures correlate with specific operating conditions: high ambient temperature, maximum load, or continuous duty cycle
  • Multiple identical boards fail with identical error signature
  • PWM loop oscillation visible on scope: low-frequency ripple (10–50 kHz) superimposed on high-frequency switching noise

Common design failures: insufficient input filtering for TPS51125 and ISL6259 controllers (input filter inductor too small, allowing >200 mV input ripple), inadequate output impedance (ESR target not met by capacitor selection), and layout errors (noisy ground return path, >2 cm trace length on gate signals).

Do not ignore design defects in production boards. Early-life failures compound warranty costs and reputation damage. Request engineering review if failure rate exceeds 0.5% in field. Common fixes: add series input inductor, swap capacitor to lower-ESR variant, or implement PCB respin with improved layout.

Diagnosis and repair:

Scope the PWM feedback loop and output voltage during transient load changes (apply 50–100% load step). If loop is unstable (ringing >20% of nominal voltage, settling time >1 ms), confirm component values match schematic. If component values are correct, issue is design-related: request engineering revision. If field retrofit is possible, add compensation capacitor to feedback network or increase input filtering.

Failure mode detection matrix

Use this table to narrow diagnosis in field failures:

Failure Mode First Symptom Test Point Expected vs Actual
Thermal cycling Intermittent cold boot Solder joint resistance <0.05Ω cold → 1.2Ω after 2 min
Power delivery failure Rail dead or sagging VOUT test point 3.3V nominal → 0.6V under load
Contamination Intermittent short, high current draw Voltage between adjacent pads >1 MΩ nominal → 100 Ω within weeks
Component degradation Voltage drift or noise spike Rail voltage, frequency measurement ±0.05V drift per year typical
Design defect Fails under sustained load Feedback loop oscillation <5% ripple nominal → 25% ripple at full load

Verification before release

After repair, execute these checks to prevent re-failure:

  1. Thermal soak test: Operate board under full load for 15 minutes. Measure rail voltage every 2 minutes. Voltage should stabilize within ±2% by minute 5. If drift continues, suspect capacitor aging or design defect.
  2. Impedance sweep: Measure rail impedance at operating frequency using a power integrity analyzer or ESR meter. Target: <10 mΩ impedance peak. High impedance peaks indicate inadequate decoupling.
  3. Thermal inspection: Use thermal imaging to identify hot spots >10°C above ambient on passive components. Hot spots indicate high resistance (failed solder joint) or excessive current (short).
  4. Visual inspection under magnification: Inspect all BGAs, connectors, and high-current joints for micro-fractures, dendritic growth, or solder voids. Use 10–20× magnification with ring light.
A properly repaired board should pass all four checks without exception. If even one fails, the root cause has not been addressed. Do not release the board for customer return without full diagnosis.