What the data for this category looks like
Data sources for PCB manufacturing inspection and process record data include customer Gerber files, drill files, BOM lists, CAM engineering files, production equipment parameters, bath solution test reports, AOI inspection reports, electrical test reports, and section analysis reports. Data updates occur per production batch, per shift, or through continuous collection, and are generated in real time as production progresses. Daily documents include engineering production instructions, drilling lamination parameter records, chemical analysis ledgers, and others. Fields include coordinate values, resistance values, defect locations, batch numbers, and more. Some documents use mil or mm as length units, while resistance values are measured in ohms.
What constraints these characteristics impose on workflow orchestration
Multiple scattered data sources require the workflow to connect multiple data acquisition nodes, docking separately with different data sources such as CAM systems, AOI equipment, and electrical testing instruments. High-frequency data updates require the workflow to support triggering per production batch or on a scheduled basis, to avoid data lag. Complex document structures and multiple field attributes require the workflow to use variables to uniformly associate identifiers of different documents, preventing data misalignment. Severe error consequences require the workflow to be configured with branch judgment and failure rollback nodes. These nodes terminate execution or send alerts immediately when abnormalities are detected, avoiding full batch scrapping or subsequent claims.
The following configurations are formulated based on FastGPT v4.17.0.
How to set the configurations
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
| Trigger Mode | Triggered by production batch completion events, interval no more than 10 minutes | Matches the data update rhythm per production batch or shift, ensuring data timeliness |
variables | Bind Gerber file ID, AOI report number, and electrical test data batch number | Uniformly pass associated identifiers for multi-source documents, preventing data misalignment across different links |
| Branch Judgment Threshold | Divide abnormal and normal branches based on a 0.5% defect rate | Intercept unqualified batches in advance to avoid the risk of full batch scrapping |
nodeId | Assign a unique identifier to each data source node | Accurately locate the execution status of each link, facilitating failure rollback and log troubleshooting |
| Timeout Setting | Single node timeout of 600 seconds, calibrated based on actual measurements | Covers the time requirements of long processing links such as Gerber parsing and bath solution testing |
| Raw Content Output Switch | Enable raw content output for specified reply nodes | Preserve special formatted output content such as HTML, avoiding format filtering |
Three common mistakes
- Phenomenon: API consumption is not attributed to the specified account when using the login-free workflow link. Cause: No association parameter for the user's API key is passed via
variables, causing the system to use the default quota. - Phenomenon: HTML code output in the workflow is not displayed, and the workflow proceeds directly to the next branch. Cause: The raw content output switch for the specified reply node is not enabled, causing automatic format stripping that triggers the branch judgment.
- Phenomenon: Workflow execution takes too long. Cause: Processing nodes for multi-source data are not split by batch, and loading all historical documents at once causes context bloat, slowing down execution speed.
How to confirm the configuration is complete
- Trigger a test production batch workflow, check if the corresponding batch's Gerber, AOI, and electrical test data are correctly loaded in the node logs.
- Call the login-free workflow link, verify that API consumption matches the preset account quota.
- Insert HTML code to test the specified reply node, confirm that the output content contains complete HTML tags and is not filtered.
- Adjust the data volume processed by a single node, compare execution durations across different configurations, confirm that the speed meets business requirements.
The values given are common starting points and should be measured against the reader's own samples.