A-Series SSD: Ceramic Capacitor Damage from Underfill Calibration Drift

Overview

Fourteen A-Series LTS SSD units failed LI testing with 4523 (LINK PM FAIL). Hardware confirmation found shorts and visible damage in the multilayer ceramic capacitors (MLCCs), establishing that the pattern was a true process-induced defect rather than normal test variation or a simple false fail.

The investigation traced the damage upstream to SMT underfill calibration. Residual calibration liquid on the sensor surface created a mismatch between the laser sensing reference and the nozzle contact reference, producing an abnormal nozzle-height condition capable of damaging the MLCC.

My contribution covered LI failure review, hardware-evidence analysis, process-window tracing, underfill calibration review, laser/nozzle reference comparison, corrective-action coordination, monitoring-requirement definition, and follow-up defect-rate review.

Problem

LI reported a repeated 4523 (LINK PM FAIL) pattern across 14 units. The symptom could initially have resulted from test-program behavior, contact instability, interface or power-management variation, a false fail, or permanent component damage.

After hardware confirmation identified MLCC shorts and visible damage, the investigation needed to locate the upstream process mechanism, distinguish an isolated component defect from a repeatable calibration-control issue, and establish an earlier control point before damaged units reached LI.

Data Used

• LI-stage 4523 (LINK PM FAIL) failure records and 14 affected units
• Affected-unit distribution and hardware-confirmation results
• MLCC short and visible-damage observations
• SMT underfill process and calibration review
• Laser sensing and nozzle contact reference-height comparison
• Sensor-surface residue inspection findings
• Post-corrective-action defect-rate monitoring

Review Scope

• LI was the detection point, but the failure code did not identify the upstream component-damage mechanism.
• MLCC short and visible damage established a true defect but did not independently identify the process condition that created it.
• The calibration mismatch depended on the relationship between the laser sensing reference and the physical nozzle contact reference.
• Immediate recurrence control required practical cleaning and inspection steps that production teams could execute consistently.
• The visualization and alarm concept was coordinated as a process-control requirement; the documented defect-rate recovery followed SOP reinforcement and shift-level residue inspection.

Approach

Followed the failure chain backward from LI detection through hardware confirmation, MLCC damage analysis, SMT process tracing, underfill calibration review, and laser/nozzle reference comparison. The investigation then separated immediate corrective controls from the longer-term monitoring requirement and tracked the defect rate after the immediate controls were implemented.

Investigation Focus

• Confirmed a repeated LI 4523 pattern across 14 affected SSD units.
• Used hardware confirmation to identify MLCC shorts and visible component damage.
• Traced the damage mechanism upstream to the SMT underfill calibration process.
• Identified residual calibration liquid as the source of the laser/nozzle reference-height mismatch.
• Supported calibration-cleaning SOP reinforcement and shift-level residue inspection.
• Coordinated a laser/nozzle height-difference visualization and alarm requirement for earlier detection.

Key Investigation Choices

Tools & Methods

• LI Failure Records
• Hardware Failure Confirmation
• Process Traceability
• SMT Underfill Calibration Review
• Laser/Nozzle Height Comparison
• Defect-Rate Monitoring

Result & Impact

• 14 units
  Affected population
• Reference-height mismatch
  Root-cause mechanism
• Returned to normal
  Post-action status
• Requirement defined
  Upstream monitoring

The defect rate returned to the normal level after calibration-cleaning SOP reinforcement and shift-level residue inspection. The investigation also converted the hidden reference-height mismatch into a defined laser/nozzle height-difference visualization and alarm requirement, establishing the intended upstream control point for future SMT calibration monitoring.

Notes

• A downstream test failure can be the first visible signal of an upstream process-control problem.
• Hardware confirmation is critical when separating test false-fails from true process-induced defects.
• SOP reinforcement is useful, but hidden calibration drift needs visual or alarm-based control.
• The most valuable quality improvements often move detection upstream, before the defect reaches final test.
• Manufacturing digitalization is most useful when it turns hard-to-see process variation into visible control points.