A WellerPCB manufacturing case study on why correct working Gerber data does not always guarantee a defect-free multilayer PCB
The central lesson is clear: if a PCB-based product is intended for commercial sale, supplier selection should be based on manufacturing capability, process control, traceability, and quality assurance—not order quantity or price alone.
Whether the order contains ten boards or ten thousand, it is generally safer to work with an established PCB manufacturer that provides transparent engineering review and controlled production processes. Extra caution is advisable when using online instant-quotation suppliers, as aggressive pricing and highly automated order handling may come with hidden risks in panelization, CAM data processing, electrical testing, and accountability.
For PCBA intended only for early prototyping or proof-of-concept validation, a low-cost online PCB supplier may still be a practical choice. However, buyers should understand that latent manufacturing defects can produce misleading test results, consume engineering resources, and significantly extend the product-validation cycle.
The Failure Appeared Only After Assembly
One week ago, a customer asked a friend of ours to supply a small commercial product built around a four-layer PCB. The bare boards had been ordered from an online PCB manufacturer, assembled, and then powered for functional testing. During testing, one circuit was found to be shorted to ground between two designated points. The fault caused an IC to fail.
The end customer returned all 101 assembled boards and requested compensation of approximately RMB 12,000. At that point, the loss was no longer limited to the price of the PCBs. It included components, assembly labor, inspection time, failure analysis, logistics, schedule disruption, and customer confidence.
The supplier-generated Gerber production files were reviewed first. No abnormal connection could be identified in the relevant area. The original design also appeared correct, which initially shifted suspicion toward a defective component, an assembly problem, or an overlooked design condition. Without the returned PCBA, however, none of those theories could be verified.
Bench Testing Confirmed a Real Short Circuit
Once the returned PCBA arrived, the investigation started with a basic continuity check. A multimeter was placed across the two points identified in the circuit review. The measurement confirmed a direct short to ground. The failure was therefore physical and repeatable; it was not simply a software symptom or an intermittent component issue.
This confirmation changed the direction of the investigation. If the visible copper, assembly workmanship, component orientation, and available Gerber data did not explain the short, then the defect could be hidden inside the multilayer structure. On a four-layer board, an unintended connection on L2 or L3 may be impossible to see after lamination and solder-mask application. The board can look normal while containing an electrically destructive defect.
Why the Customer Gerber Review Did Not Reveal the Problem
A Gerber review is essential, but it answers only one question: does the reviewed dataset represent the intended design? It does not automatically prove that exactly the same geometry was used at every downstream manufacturing step.
PCB factories normally convert customer data into internal tooling data. This may include layer mapping, drill optimization, copper balancing, etching compensation, solder-mask adjustment, panelization, coupons, tooling holes, fiducials, and factory-specific process features. These operations are legitimate and necessary. The risk appears when a modification is incorrect, applied to the wrong feature, or released without an effective verification step.
In this case, the files reviewed by the customer showed no short. The defect was introduced later in the factory’s production-panel data. Reviewing only the incoming Gerber files therefore could not expose it.
Root Cause Excessive Compensation in the Production Panel
The defective assembly was returned to the PCB manufacturer for analysis. One day later, the manufacturer reported that the large production panel had been compensated incorrectly. Excessive compensation altered copper features on the L2 and L3 inner layers and created unintended connections to ground.
Further inquiry indicated that several different customers’ four-layer PCB designs had been combined on a single large manufacturing panel. Mixed-job panelization can improve laminate utilization and lower the cost per design, especially for prototypes and small orders. It is not inherently unacceptable. However, it increases the need for disciplined data control because different designs, copper distributions, clearances, and process requirements are being managed together.
The manufacturer also reported using internally developed software to apply compensation automatically rather than relying on a manual engineering adjustment. Automation can improve consistency and speed, but an automated output still requires validation. If the compensation algorithm, layer recognition, polarity interpretation, clearance rule, or job mapping is wrong, the system can reproduce an error efficiently across an entire panel.
Manufacturing Risk
Excessive panel compensation changed the L2/L3 geometry and created hidden connections to ground, even though the customer-approved Gerber data was correct.
How Panel Compensation Can Create an Internal Short
Compensation is used because the image transferred to copper does not always reproduce at exactly the nominal design dimension after imaging and etching. A factory may enlarge or reduce selected features so that the finished conductor width, spacing, annular ring, or clearance remains within specification after processing.
The operation becomes dangerous when copper expansion reduces a required isolation gap. On a negative or plane layer, a polarity or clearance mistake can be especially difficult to recognize. A small geometric change around a pad, via, thermal relief, or antipad may connect a signal feature to a ground plane. After the inner layers are laminated into the stack-up, the evidence is buried.
The failure mechanism in this case can be summarized as follows:
1
Approved design: The customer Gerber data contained the correct isolation.
2
Panelization: Multiple customer jobs were combined on one production panel.
3
Automatic compensation: The factory modified the manufacturing dataset.
4
Clearance loss: Compensated L2 and L3 geometry lost the required isolation.
5
Hidden short: The inner-layer connection remained inside the finished bare PCB.
6
Assembly failure: Power application damaged an IC and triggered the batch return.
Which Quality Controls Should Have Detected the Defect
A robust PCB manufacturing process should not depend on a single review. Several controls can prevent or detect this type of failure before the board reaches assembly.
| Quality Control | What It Should Verify |
|---|---|
| Post-compensation CAM verification | Compare final tooling data with the approved customer data and check spacing, connectivity, netlists, and polarity. |
| Netlist-based electrical test | Check opens and shorts against an independent IPC-D-356 or equivalent trusted netlist. |
| Independent source data | Keep the electrical-test reference separate from the production CAM dataset. |
| First article review | Confirm electrical performance, dimensions, and stack-up before full-order assembly. |
| CAM change traceability | Record data changes, applied rules, software versions, approvals, and released revisions. |
| Mixed-panel risk control | Identify and verify each customer job using rules appropriate to its own design. |
What Buyers Should Ask a PCB Supplier
Price matters, but buyers should understand what makes a price possible. High material utilization, mixed-job panels, automated CAM processing, reduced engineering review, and limited testing can all lower cost. The important question is whether the controls remain appropriate for the product risk.
Before placing a commercial multilayer PCB order, ask the supplier:
01
Will this design be placed on a mixed-customer production panel?
02
How is etching or feature compensation calculated and approved?
03
Is final manufacturing data compared with the customer-approved dataset?
04
Is 100% electrical testing performed, and what independent netlist is used?
05
How are CAM revisions, software outputs, approvals, and test records retained?
06
What failure-analysis and liability process applies if a defect reaches assembly?
For prototypes, a very low price may be acceptable when the boards will be thoroughly validated before use. For products that will be assembled, certified, sold, or installed in the field, the cost of failure is much higher. The sourcing decision should therefore reflect total risk, not only the unit price of the bare PCB.
Supplier Selection Tip
For commercial multilayer PCBs, compare suppliers by data control, electrical testing, and traceability—not only by bare-board price.
The True Cost Was Not the Bare PCB
101
assembled boards returned
≈ RMB 12,000
compensation claim
L2 / L3
inner layers involved
IC Failure
after power was applied
This is why an apparently inexpensive PCB can become the most expensive part of a product. A small saving at the bare-board stage is quickly erased when a latent defect is discovered only after components, labor, firmware, testing, packaging, and customer commitments have been added.
Conclusion
Correct Gerber files are necessary, but they are not the final guarantee of a correct PCB. Manufacturing introduces additional datasets and process decisions between customer approval and the finished board. Production panelization, compensation, polarity handling, CAM automation, and electrical-test preparation must all be controlled.
The practical lesson is not to reject automation or panel sharing. Both can be effective when supported by independent verification, traceable engineering approval, and reliable electrical testing. The lesson is to select a PCB supplier whose quality controls match the commercial consequence of failure.
At WellerPCB, we believe PCB sourcing should be based on manufacturability, process transparency, appropriate testing, and total product risk—not price alone. When a multilayer board will become part of a finished commercial product, careful engineering review before production is far less expensive than failure analysis after assembly.
