test data collection system optimized for storage of high resolution measurements and for highly parallel testing

Affordable Infrastructure

One of the ways to keep test costs under control is the increase of the number of devices tested in parallel. An extreme example is the Test During Burn In process where several dozens of thousands of devices are tested in the same time with the same equipment. On the other hand, for taking the right decision, a significant amount of measurements, generating large size data, needs to be stored. Thus, storing all measurements of a testing floor requires infrastructure which is simply not affordable for most of the companies.

BarnieDAC is the right option for this issue. It embeds a database technology which:

• Contains all information common for all tests in traditional, table format.
• Stores test dependent measurement data with double compression, limiting both the database space and the number of records. Specific tables are used for datatype description and for fast data access.
• Offers the possibility to store large measurement data (like bitmaps) in external files limiting further the required database space.

Easy Correlation between Multiple Test Platforms

BarnieDAC is able to import datalogs from multiple test platforms, even if the datalog format is different. This feature allows a uniform view on the data, and fast correlation between platforms.

• An abstract test data model has been defined and drivers have been implemented for standard datalog formats (STDF, XML).
• For each data source (single or multiple testers), a loader daemon is activated, which scans the availability of test data, parses the datalog and loads content into the database
• The processed datalog is moved in a folder (separated if the load had success or failed) in a compressed format.
  

Configurable Data Retrieval for Post-Processing

When data is needed for post-processing, the following options are available.

• When the measurement information is stored in the format which can be the direct input of a post-processing (like bitmaps and cell lists for memory array analysis), the information can be directly taken out and passed to the processing tool.
• A built-in Table Extractor tool is able to pick up selected measurement values from the datalogs/database and organize them in a table. The generation of the table is configurable and pre-filtering options can be applied. The extraction configuration can be created run-time from the extraction script, input in the GUI or read from a configuration file. The resulting table can be stored in CSV format or passed directly to BarnieMAT or BarnieCOR.

Straightforward Consultation of Content

The content of the datalogs, independently if they are in the native datalog files or loaded in the database, can be browsed by a dedicated graphical user interface. There are several view options:

• Tree View where the measured values are accessed in a hierarchical way, based on job and test selection.
• Wafer Map View where the access is driven by the wafer lot, number and die coordinates.
• Test Board View where the access is by test board position.
• Reached the final measurement, the values are displayed in the viewer (color code option using the Wafer Map View or the Test Board View).

Full Integration in the BARNIE Environment

• Graphical framework shared with the analysis and correlation tools.
• Complex data collection flows can be defined using a powerful scripting language. The same script can contain data collection and analysis flows.
• Basic data collection functions are implemented as “stored procedures” and contained by Windows DLLs. User can add his/her own functions by implementing the code in MFC C++ language, supported by an extensive library.