What Data for This Category Looks Like
Data for semiconductor packaging and testing comes from wafer test maps, MES systems, bonding and molding equipment, ATE test machines, X-ray/SAM inspection equipment, and customer specification documents. Update frequencies cover per batch of materials, per individual chip test, and daily operation and maintenance nodes. Daily documents include packaging process cards, bonding and molding process parameter tables, wafer incoming material maps, test program version lists, yield reports, and failure analysis reports. Each document includes fields such as process parameters, batch numbers, test bin values, and failure locations. Units include micrometers, volts, batch numbers, and others.
What Constraints These Characteristics Impose on Workflow Orchestration
Semiconductor packaging and testing data is multi-source, heterogeneous, and has varied update frequencies. Workflow orchestration must adapt to three update rhythms: per batch, per single chip test, and daily. Each rhythm triggers corresponding data pulling and verification steps. Daily documents contain many professional fields. The workflow must configure precise variables mapping rules. These rules accurately associate location data from wafer incoming material maps and bin distribution data from ATE tests with quotation items in bid documents. The workflow must also ensure compatibility between data sources and the output formats of MES and ATE equipment. This avoids parameter misalignment caused by format mismatches. A customer specification verification node must be embedded in the process to prevent incorrect use of process programs or batches.
Configuration Settings
The following configurations are based on FastGPT v4.17.0. They meet workflow orchestration needs for bid quotation preparation in semiconductor packaging and testing.
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
| Trigger Node | Three modes: per batch trigger, per single chip test trigger, daily scheduled trigger | Matches the update frequencies of per batch, per chip test, and daily for packaging and testing, ensuring real-time updates of corresponding parameters |
variables Mapping Rules | One-to-one binding of location IDs from wafer incoming material maps, bin values from ATE tests, qualified counts from yield reports, and quotation fields in bid documents | Ensures accurate mapping of professional data to quotation items, avoiding information misalignment |
| Single Data Pull Limit | 500 entries | Adapts to the scale of test data for a single batch, avoiding interface overload |
| Visibility Configuration | Disable viewing original text for sensitive process parameters, enable viewing source for general quotation information | Balances information security and traceability requirements for bid documents |
| Code Execution Node | Enable Python script execution permission, set execution timeout to 300 seconds | Adapts to custom calculation logic such as bin distribution analysis and batch alignment for packaging and testing |
nodeId Verification | Bind unique node identifiers corresponding to each device | Ensures data pulled from MES and ATE test machines is correctly attributed, avoiding cross-source errors |
Three Common Configuration Errors
- Phenomenon: A permission error pops up when running a Python script in a code execution node of the workflow, or the script runs without output. Cause: Python code execution permission configuration is not enabled on the FastGPT platform, or the script references unauthorized dependency libraries.
- Phenomenon: Configured visibility rules do not take effect, and sensitive process parameters can still be viewed in original text by unauthorized accounts. Cause: Configuration switches are not distinguished by content attributes, and permissions for sensitive documents such as packaging process cards are mistakenly set to open.
- Phenomenon: After importing a workflow template, the interface shows import success, but no nodes or configurations appear on the canvas. Cause: The
nodeIdconfiguration in the template file does not match the current platform's node identifier rules, or data truncation occurs during the import process.
How to Verify Proper Configuration
- Trigger a node matching an update rhythm. Check that pulled data matches real-time output from MES and ATE equipment.
- Test the
variablesmapping rules. Input simulated wafer incoming material map data. Confirm that quotation items in the bid document generate correctly. - Verify visibility configuration. Log in to the platform with an unauthorized account. Check that sensitive process parameters cannot be viewed in their original text.
- Run the code execution node. Confirm that the Python script runs normally and outputs expected bin distribution analysis results.
The values provided here are common starting points, and should be evaluated against the reader's own samples.