How Do You Rework and Repair a PCB Assembly After a Defect Is Found?

Detailed Explanation

Inspection and test (see PCB assembly inspection and testing) exists to find defects — but finding a defect is only half the problem. IPC-7711/7721 is the standard most contract manufacturers cite for what happens next, yet the rework and repair process itself has no dedicated coverage here. This page covers that process directly: what distinguishes rework from repair, the component-class-dependent techniques used for each, and the decision point where reworking a board stops being worth it.

Rework vs Repair: Two Different Activities

IPC-7711/7721 draws a specific distinction the industry uses consistently:

• Rework restores a board to its original, fully compliant design intent — replacing a wrong-value component with the correct one, or re-attaching a component that failed inspection, using the same materials and footprint the design specifies.
• Repair restores functionality using a method that deviates from the original design intent — a wire jumper bridging a lifted pad, a patched trace, or a replacement component in a non-standard orientation because the original footprint is damaged. Repair is a workmanship compromise, acceptable for prototypes or field repair but generally not for volume production units held to IPC-A-610 Class 2/3 acceptability.

Hand Rework: Leaded and Larger-Pitch Components

Components with accessible leads and generous pitch — through-hole parts, SOICs, connectors, 0603 and larger passives — can typically be reworked with standard bench tools: a temperature-controlled soldering iron, desoldering braid or a solder sucker, and flux. The process is straightforward: apply flux, reflow the existing joints while lifting the component clear, clean the pads, then place and solder the replacement. The main risks are pad lifting from excessive dwell time at the iron tip and thermal damage to adjacent components from prolonged heat exposure — both avoidable with a correctly set iron temperature and working quickly once the joint is molten.

Hot-Air and Hot-Bar Rework: Fine-Pitch and Area-Array Packages

QFN, LGA, and BGA packages cannot be reworked with a soldering iron — their pads or balls are inaccessible from the side, hidden entirely beneath the package body. These require a dedicated rework station:

• Hot-air rework stations direct a controlled stream of heated air, localised to the target component through a nozzle sized to its footprint, following a programmed preheat-and-reflow profile similar in principle to the reflow oven profile used in original assembly (see reflow profile design) but scaled to heat only the one component and its immediate board area, protecting neighbouring parts with thermal shielding or Kapton tape.
• Hot-bar (pulse-heat) rework uses a heated bar that presses directly onto a component's leads — common for fine-pitch QFP and connector rework where a defined, repeatable contact-heating profile is preferred over convective hot air.

BGA and QFN Reballing

Removing a BGA or QFN leaves the PCB pads coated in residual solder rather than a clean, even surface — placing a new component directly onto this residue produces unreliable joints (see the FAQ above). Reballing addresses this: residual solder is removed from the pads (typically with desoldering braid, a vacuum extractor, or a dedicated cleaning process), then new solder is deposited evenly — either a stencilled paste-dot array matching the ball grid pitch, or discrete pre-formed solder balls placed with a ball-attach jig — before the replacement part is placed and reflowed with the hot-air or hot-bar station.

The Rework-vs-Scrap Decision

Rework is not always the right call. Before reworking a board, weigh:

• Board value and volume stage. A one-off prototype or low-volume pilot board is almost always worth reworking; a low-value, high-volume production board may be cheaper to scrap than to pay per-unit rework labour on.
• Defect isolation. A single clearly-identified component failure is a good rework candidate. A defect with an unclear or systemic root cause (e.g. a design error affecting every unit) is not fixed by reworking one board.
• Pad and substrate condition after removal. If removal is expected to lift pads, damage plated-through holes, or stress a high-density interconnect (HDI) via structure, repair risk rises sharply — see the FAQ above on lifted pads.
• Rework cycle count at that location. Each rework cycle degrades the pad and surrounding laminate somewhat; a location already reworked once carries materially more risk on a second attempt (see the FAQ above).
• Conformal coating or potting on the board. A conformally coated board needs the coating locally removed and reapplied around the rework site — extra steps but still practical. A potted assembly is generally not reworkable at all once the potting compound has cured — see conformal coating and potting for why potting is effectively a one-way decision against future rework.

Design Considerations

• Leave rework clearance around fine-pitch and area-array packages at the layout stage. Adjacent tall components or insufficient keepout around a BGA/QFN site can make hot-air rework impractical without risking damage to neighbouring parts — a design-for-rework consideration alongside design-for-assembly.
• Specify a surface finish and solder mask that tolerate the expected rework cycle count. Some fabrication and finish combinations hold up to repeated reflow better than others; discuss expected rework rates with the assembly house during process qualification if the product line is expected to see field or production rework.
• Budget rework into the production test strategy, not as an afterthought. A board that fails ICT or functional test (see PCB assembly inspection and testing) needs a defined rework path — which components are rework-eligible, what tooling is available, and at what point a failed board is scrapped instead.

Zeus Design's electronics engineering team can diagnose and rework defective prototype and low-volume production boards as part of rapid prototyping and hardware validation.

Common Mistakes

• Reflowing a removed BGA's original solder instead of reballing. Covered in the FAQ above — residual solder left on the pads after removal is uneven and contaminated, and skipping reballing is one of the most common causes of BGA rework failure.
• Using a fixed-temperature iron or heat gun on a fine-pitch or area-array package. Uncontrolled, non-localised heat risks damaging neighbouring components and rarely reaches the hidden joints of a QFN or BGA reliably — a dedicated hot-air or hot-bar rework station with a controlled profile is required for these package types.
• Reworking a location that has already failed a previous rework cycle. Each cycle degrades the pad and surrounding laminate; treating a location as infinitely reworkable eventually produces a lifted pad or an unreliable joint — see the FAQ above.
• Treating a repair (a workmanship compromise like a wire jumper) as equivalent to a rework. A repair may restore function but does not meet the same acceptability criteria as a true rework to original design intent — appropriate for a one-off prototype fix, not for units built to IPC-A-610 Class 2/3.

Zeus Design's electronics engineering team designs boards with rework and manufacturability in mind — see PCB design for manufacturability for the layout-stage practices that keep future rework practical rather than accidentally impossible.