Gerber, BOM and CPL Consistency Checks

A Gerber BOM CPL consistency check confirms that board manufacturing data, the bill of materials (BOM), and the component placement list (CPL) describe one approved assembly configuration. Match release identity, populated reference designators, quantities, footprints, and board sides before manufacturing release. Compare files only after declaring the assembly variant and export scope. Use zero unresolved release-blocking discrepancies as the release criterion: every difference must be corrected or supported by an approved, traceable exception. Matching filenames or equal totals alone do not establish consistency.

The files serve different roles. Gerber layer data describes board features and can carry attributes. The BOM links component identities to references; the CPL defines each reference’s position, rotation and board side. An assembly drawing adds visual intent and instructions. Compare the information each file actually carries, not just its filename.

Establish One Release Identity Across Gerber, BOM and CPL

Gerber, BOM, and CPL files belong together when an approved release record explicitly associates them with the same assembly configuration. Their individual revision labels do not have to be identical. A bare-board revision can remain unchanged while the BOM or assembly variant changes; the release must document that relationship rather than disguise it with matching filenames.

Create a manifest identifying the assembly number, release ID, bare-board revision, assembly variant, source design snapshot, export configuration, and exact delivered files. Include assembly drawings, drill data, and the panel definition where applicable. Record file digests against the approved package so later recipients can check whether files changed. A digest comparison checks file identity, not engineering correctness or approval.

For example, an illustrative release M24-REL07 can approve PCB revision C, BOM revision 05, CPL export EX12, and drawing revision D for variant STANDARD. Those identifiers differ, but their compatibility is explicit. A directory called “final” provides no equivalent evidence.

Use this cross-file review matrix to organize the release decision. Board-data evidence means the released CAD or component export where available, supported by the actual Gerber and drill geometry—not an assumed component list extracted from artwork.

Match Reference Designators Between BOM and Placement Data

BOM-to-CPL matching must compare physical component instances for one board and one variant. Expand grouped references before comparing them, and inspect duplicates before converting lists to sets. Otherwise, a repeated reference can disappear during deduplication and make an incorrect file appear consistent.

For a declared inclusive range, R1–R4 expands to R1, R2, R3, and R4. A BOM line claiming quantity five still needs investigation. Do not silently convert R01 into R1, remove channel suffixes, or change case unless the project’s annotation rules authorize that mapping. Retain the original strings alongside normalized comparison keys.

Classify every raw row before filtering. Retain a ledger of DNP entries, alignment features, and out-of-scope items with their supporting rule; an unclassified row remains open rather than disappearing from the comparison.

Use three sets for the comparison: B contains populated, on-board physical references from the selected BOM; C contains populated references in the declared CPL scope; M contains approved fitted references intentionally handled outside that scope. The expected relationship is B = C ∪ M, with C ∩ M empty. Check the combined population against the variant-resolved design inventory as well. Agreement between two stale exports is not enough.

Investigate B − (C ∪ M) as unexplained missing placements and (C ∪ M) − B as unexplained additional instances. Compare part identity and footprint after the references match; set equality does not prove that the correct part is assigned to each location.

A mismatch is a symptom, not a diagnosis. A BOM-only J1 could be a documented hand-inserted connector, an accidental export exclusion, or a stale BOM entry. Follow the reference to the approved design and assembly instruction before deciding which file to change.

Resolve multi-unit schematic symbols to physical packages before counting. Keep panel instances such as PCB1:R1 and PCB2:R1 distinct under a declared mapping. Alternate footprints can share a logical designator; select the instance fitted in the released variant before checking physical uniqueness.

Reconcile Populated Counts and DNP Exceptions

Populated counts must describe fitted physical components per finished board, not BOM rows, purchased pieces, or every footprint visible in CAD. State the quantity basis first, then separate do-not-populate (DNP) positions, non-component features, and parts assembled outside the CPL scope.

DNP handling is an export agreement, not a universal omission rule. The version 9.0 exporter example provides options to exclude DNP footprints, select only SMD footprints, and exclude footprints containing through-hole pads. It also supports footprint-level exclusion from placement files. Confirm which settings produced the delivered data before calling a missing reference an error.

Worked count reconciliation

Consider this illustrative single-board configuration. Its released inventory contains 120 candidate physical component positions, including eight DNP positions. Four of the fitted components are through-hole connectors intentionally excluded from an SMT-only CPL. Three additional fiducials are non-component features and are not part of the 120 positions.

If a receiving system accepts all 120 component positions in a raw CPL, eight require an explicit non-placement disposition and four require the agreed separate assembly route. Do not assume a CSV’s DNP field is understood merely because it exists. Confirm the imported fit status and programming scope with the assembler.

Reconcile top and bottom populations separately before combining them. For the same example, 80 top-side and 28 bottom-side SMT placements total 108. If the intended split is 80/28 but the imported split is 79/29, the matching total conceals a side discrepancy.

For a four-board panel with the same variant in every position, the expected SMT population is 4 × 108 = 432 placements. Add no panel fiducials to that total. Mixed variants require a separate sum for each instance. Keep feeder allowances, spare components, and purchasing overage outside all these fitted-count calculations.

Verify Footprints, Board Sides and Drawing Intent

Footprint and side checks must establish that each selected component fits the released location and that all assembly instructions describe the same mounting intent. Matching reference names and quantities can leave an incompatible package or a wrong-side component undetected.

Compare the BOM’s exact part selection with the released footprint and the applicable component package drawing. Check pin count, pitch, body outline, exposed pad, and relevant mounting holes. A matching library name is useful evidence, but it is not a geometric comparison. Treat changes affecting the land pattern as design issues, not spreadsheet naming corrections.

For board sides, map the export’s side tokens to the project’s top and bottom definitions before comparing them. Inspect the assembly drawing’s viewing convention and the corresponding CAD layer. The referenced supplier’s CPL guide explicitly identifies mounting side as placement data; its importer specification is not a universal column convention.

Do not resolve contradictory side instructions by automatically choosing the newest file or always preferring the drawing. The design owner must establish the intended configuration from the approved design and change record, then correct every affected output. Put unresolved conflicts on hold even when a viewer can display them plausibly.

Check connector mating direction, pin-one intent, polarized components, and mechanical notes where they reveal disagreement between files. This is a consistency review, not a substitute for detailed rotation or land-pattern validation. Route a coordinate-origin or angle-convention problem into the placement-file review without changing geometry simply to make an overlay look aligned.

Detect Stale Exports and Mixed-Revision File Sets

Stale exports are identified by inconsistencies with the approved source configuration, not by modification time alone. A newer Gerber file can be a harmless re-export, an unapproved board change, or part of a valid release whose unchanged BOM was deliberately retained. Establish which explanation applies before ordering boards.

Compare the release manifest with the source snapshot and export settings. Then compare changed references, selected parts, footprints, sides, board outline, and relevant drill features with the approved change description. If R17 moved from top to bottom but the CPL still places it on top, that is specific evidence requiring resolution. A timestamp difference only tells you where to investigate.

Use content and geometry comparisons rather than relying exclusively on byte differences. Two exports can have different headers yet describe equivalent board features. Conversely, identical filenames can contain different geometry. Record the verified relationship between retained and regenerated files.

When a mismatch is confirmed, correct the controlled source and regenerate affected outputs into a clean release directory. Reissue the complete manifest, identify superseded files, and have the assembler acknowledge the replacement package. If import preparation creates a derived file, retain its connection to the approved source and recheck imported references, fit status, and side totals. Avoid email instructions that require manufacturing to combine an old archive, a replacement spreadsheet, and an informal correction into its own guessed configuration.

Record File-Consistency Exceptions Before Manufacturing Release

A file-consistency exception must state what differs, why it is intentional, who approves it, and how manufacturing will handle it. An explanation such as “connector missing from CPL” is not closure until the connector has an approved installation route and the package no longer gives contradictory instructions.

Record the exception ID, release and variant, affected references, file-specific evidence, approved disposition, accountable owner, approval record, replacement-file identity where needed, and recheck result. These fields make the record usable without requiring a reader to reconstruct a conversation.

An illustrative completed exception reads: “EX-004; M24-REL07; STANDARD; J1–J4. BOM 05 specifies four fitted connectors; CPL EX12 intentionally contains SMT positions only. Drawing D, note 6, assigns J1–J4 to separate insertion. Design engineering approved the fit decision; manufacturing engineering accepted the route under approval record APR-018. No replacement files required. Recheck: 112 fitted components = 108 SMT placements + four separately inserted connectors. Closed for this release only.” This is an example, not a production record.

Use this release sequence:

Freeze configuration → verify file identity → reconcile references and counts → verify package and side intent → close discrepancies → approve package → obtain manufacturing acknowledgment.

Keep the release on hold while any release-blocking discrepancy remains unresolved. Treat an unexplained missing or extra fitted reference, ambiguous package or side, and unassociated file revision as blockers; a high overall matching percentage does not cancel them. An approved exception can explain a deliberate scope difference; it cannot authorize contradictory instructions to remain active. Recheck exceptions after a variant, component, or assembly-route change instead of copying their closed status into the next order.

Export Conventions and Applicable Specifications

Export conventions determine how a valid design appears in a manufacturing package. Agree the receiving interface before interpreting differences. The following comparison separates representation choices from evidence of a genuine design conflict.

The exporter example supports both combined and separate side outputs and the three placement formats listed here. These are documented exporter options, not a claim that every assembler accepts them.

Gerber X2 attributes can identify layer and feature functions, while the X3 extension carries component information. Use supported metadata when present, but do not infer a complete fitted-parts list from bare-board copper images alone. Pair image-only data with a released component inventory or assembly drawing that resolves the references.

Assembly variants may share a bare board, whereas fabrication variants change board output. A shared PCB revision therefore does not prove a shared populated configuration. Apply the relevant format documentation and agreed receiving specification; a file-format label alone cannot establish that the BOM, placement data and fabrication package belong together.

Five File-Consistency Scenarios

File-consistency reviews should follow the change that can invalidate a release. These illustrative scenarios show how to select the evidence and close the discrepancy without treating every difference as an export error.

Prototype revision after a connector move

The latest board data relocates J3, while the placement file retains its former location. Compare the change record, physical reference, and released board geometry. Regenerate the affected placement and drawing outputs from the approved source. Renaming the old CPL does not resolve its contents.

Two product variants sharing one bare board

A reduced-feature variant omits a communication device that the full-feature variant fits. Use the variant-specific BOM and fit decisions to define the expected population. Keep the unchanged bare-board data only when its association with both assembly configurations is approved; do not merge the two placement populations.

Repeat order with a changed part selection

Purchasing proposes an alternate component while reference names and quantities remain unchanged. Check the selected part, footprint, and assembly instructions against the approved configuration. Equal counts cannot approve an alternate package or different pin assignment. Record the approved substitution and recheck affected files before release.

Panelized production after single-board prototypes

Manufacturing receives a four-board panel but a single-board placement list. Agree whether the assembler will replicate the approved single-board data or requires an expanded panel file. Check instance mapping and the expected population under that agreement. Neither representation is inherently wrong; an undocumented mixture is unacceptable.

Double-sided assembly with a missing file

The BOM covers components on both sides, but only the top-side CPL reaches manufacturing. Reconcile each side separately and identify the absent bottom-side population. Restore the approved bottom-side file or regenerate the agreed combined output, then recheck duplicates before acknowledging the replacement package.

FAQ

Gerber, BOM, and CPL questions should be resolved against one approved configuration. Establish the variant and export scope before using a numerical difference as evidence of a manufacturing-file error.

Next Step: Submit One Reconciled Assembly Package

Submit the Gerber and drill package, variant-specific BOM, CPL, assembly drawing, release manifest, and closed exception record together. State the board quantity, panel basis, and separately installed components. Ask the assembler to acknowledge the exact release before fabrication, sourcing, or placement programming proceeds.

For JASPER PCB assembly services, provide that coordinated package with sourcing responsibilities and the required inspection or testing scope. Make the purchasing handoff identify one approved configuration—not a collection of files that manufacturing must reconcile by assumption.