We Open-Sourced Our SECS/GEM Python Driver for Semiconductor Equipment
核心结论
Open-source SECS/GEM driver: 3 lines of code to connect equipment. Replaces $100K+ commercial licenses. Full SECS-I/HSMS/GEM support on GitHub.
In semiconductor manufacturing, the SECS/GEM protocol (SEMI E4/E5/E30/E37) is the standard interface for equipment-to-host communication. Every major equipment maker — Applied Materials, Lam Research, TEL, KLA — uses HSMS/SECS-II to exchange data with MES, FDC, and R2R systems.
Yet the available options for SECS/GEM communication are either expensive commercial packages (Brooks, Cimetrix) or outdated C/C++ libraries. The Python ecosystem has lacked a production-ready, configuration-driven SECS/GEM driver — until now.
Today we are open-sourcing secsgem-driver — an async Python SECS/GEM driver that powers the communication layer of our NeuroBox edge AI platform, battle-tested in real semiconductor fabs.
Connect to Equipment in 3 Lines
from secsgem import SecsGemDriver
driver = SecsGemDriver("configs/amat_centura.yaml")
await driver.connect()
response = await driver.send("S1F1") # Are You There
No hardcoded message structures. Equipment messages, variables, events, and commands are all defined in YAML configuration files. Switching equipment? Just swap the YAML file.
Key Features
- HSMS Protocol (SEMI E37) — Active/Passive modes, Select/Deselect/Linktest, heartbeat
- SECS-II Codec (SEMI E5) — Full encode/decode for all data types
- Configuration-Driven — YAML equipment profiles, zero code changes for new tools
- Async/Await — Built on Python asyncio for non-blocking I/O
- Auto-Reconnect — Configurable reconnection with backoff
- Event System — Decorator-based subscription to equipment events (S6F11) and alarms (S5F1)
- Type-Safe — Pydantic-validated configuration
Why Open Source
The SECS/GEM protocol is a published SEMI standard. The communication layer should not be a barrier to innovation. Our competitive advantage lies in the AI algorithms above it — physics-informed virtual metrology, trust-region R2R optimization, uncertainty-quantified fault diagnosis — not in implementing a well-documented protocol.
By open-sourcing the driver, we help engineers get equipment data flowing faster, which accelerates AI adoption across the industry.
How It Relates to NeuroBox
secsgem-driver is the communication foundation of the NeuroBox E series edge AI platform. NeuroBox adds on top:
- Physics-Informed Neural Network (PINN) virtual metrology
- Trust-region constrained R2R optimization
- Uncertainty quantification for fault diagnosis
- TensorRT-accelerated inference (<50ms latency on Jetson)
Get Started
pip install secsgem-driver
GitHub: github.com/shensi8312/secsgem-driver
Stars, issues, and contributions are welcome. If you are working on semiconductor equipment communication and have questions, feel free to reach out.
