TDA 2.1.5

TDA

English | 简体中文

TDA is a .NET 10 library for reading, writing and analyzing semiconductor test data (CP/FT) stored in the STDF (Standard Test Data Format) format.

IO comes in two layers: RecordReader / RecordWriter work at record level (individual STDF records); DataPrism (formerly StdfReader) works at whole-Lot file level — loading, parsing, correlating multiple sublots and retest rounds, statistical analysis and export. DataPrism is the core of the library: after parsing STDF it organizes everything the files carry into one object model — lot / program / tester header information, per-part details with Hard/Soft Bins, test-item definitions with limits, raw per-part per-item test values, plus the computed statistics and yield.

Callers can access that data from any angle (by lot, by test item, by part, by site, by bin, by test round, or by DeviceID to follow one chip across rounds), and every angle offers both query and export-to-file APIs.

DataPrism takes either STDF files or a DataPrismSetup object as input. A DataPrismSetup is the organization of all STDF files of one Lot: the file paths of every TestRound of every Sublot, plus the retest relationships between rounds (which TestRound is the retest of which HardBins of which TestRound).

The TDA suite is published on NuGet as three packages:

Install

dotnet add package TDA
dotnet add package TDA.GUI
dotnet add package TDA.Report

Platform Requirements

• Target framework: net10.0-windows10.0.19041 (Windows only).
• The package enables <UseWindowsForms>true</UseWindowsForms>; consuming projects must target net10.0-windows* on Windows and also enable UseWindowsForms. Linux / macOS TFMs are not supported.

Features

• STDF record-level engine — 32 record types supported (FAR, MIR, SDR, WIR, WRR, PIR, PRR, PTR, MPR, FTR, HBR, SBR, TSR, …), with RecordReader / RecordWriter for full read → write round trips.
• Compressed STDF — reads .stdf, .std, .gz, .zip, .tar.gz/.tgz; two decompression modes: Buffered (fast) and Streaming (~70 KB memory even for multi-GB files).
• DataPrism analysis engine — one instance = one Lot: multi-sublot, first-test + multi-round retest correlation with per-round HardBin targeting, DeviceID tracking, CP wafer X/Y auto-detection.
• Parallel statistics — DoStatisticAnalysis() runs via Parallel.ForEach and computes, per test item: Mean / Median / Min / Max / Sigma, Cp / Cpk / Cpl / Cpu, 13 percentiles (P0.1 … P99.9), per-site statistics, and site-gap indices (SGM/SGI).
• Import / Export — export raw data (NEDA / RPV layout), parameter statistics, lot summary and TestID sets to CSV; import NEDA CSV back into a DataPrism.
• JSON-configurable input — DataPrismSetup (Sublot → TestRound hierarchy) serializes to/deserializes from JSON with relative-path normalization; TestID rename profiles from CSV/XLSX.

Quick Start

Load STDF data and analyze it

using TDA;

// 1. Load one or more STDF files (.stdf/.std/.gz/.zip/.tar.gz, or a setup .json)
using var dp = DataPrism.FromFile(@"C:\data\lot001_sl00_ft1.stdf.gz");
dp.Load();
dp.DoStatisticAnalysis();   // mean/sigma, percentiles, Cp/Cpk, per-site, site-gap

// 2. Query lot-level info
Console.WriteLine($"LotID: {dp.DataSourcesInfoList[0].Mir.LOT_ID}");
Console.WriteLine($"Parts: {dp.PartsDescription.Length}, Test items: {dp.TestItems.Count}");

// 3. Export parameter statistics to CSV
DataPrism.ExportStatistics(dp, outfile: @"C:\output\param_stats.csv");

// 4. Export lot overview (yield, HardBin, SoftBin — all with per-site columns)
DataPrism.ExportSummary(dp, outfile: @"C:\output\lot_summary.csv");   // or ExportYield / ExportHBin / ExportSBin

Multi-round retest configuration

using TDA;

var sublot = new SublotSetup();
sublot.AddFirstTestRoundDataFiles("Init1.stdf", "Init2.stdf");
sublot.AddRetestRoundDataFiles(new[] { "Rt1.stdf" }, new ushort[] { 2, 3 }); // retest HardBin 2,3 only

var setup = new DataPrismSetup();
setup.Sublots.Add(sublot);
DataPrismSetup.SerializeToJsonFile(setup, "lot001.setup.json");   // reusable config

using var dp = new DataPrism(setup);
dp.Load();
dp.DoStatisticAnalysis();

Export raw data (NEDA layout) for a TestID subset

using TDA;
using TDA.TypeDef;

TestIDSet ids = dp.ExportTestIDSet(@"C:\output\testids.csv"); // full TestID list as template
DataPrism.ExportRawDatas(dp, ids, isFinalDataOnly: true,
    layout: RawDataLayout.NEDA, outfile: @"C:\output\raw_neda.csv");

Key Types

Documentation

• Examples & learning material — the tda-learn repository: Examples/Ex101.UseDataPrism, Examples/Ex102.ExportParamsStatistics, Examples/Ex103.ExportRawData, Examples/Ex105.ExportSummary, Examples/Ex204.CfgFileForDataPrismSetup, Examples/Ex301.RecordReaderBasics, Examples/Ex501.UseRecordWriter.
• Command-line usage — the CLI/ folder of the same repository holds the dp manual and per-command reference.
• Design notes (architecture, compressed-STDF internals, …) are maintained in the private source repository and are not published.
• API documentation is embedded as XML doc comments — hover in your IDE.

Links

• GitHub: https://github.com/GeminiDraig/tda-learn
• GitCode: https://gitcode.com/GeminiDraig/tda-learn
• NuGet: https://www.nuget.org/packages/TDA

License

© TDA Project Contributors