siglent-sds-mcp

siglent-sds-mcp

Enables AI tools to remotely control a SIGLENT SDS800X HD series oscilloscope, configure acquisition, capture waveforms, and analyze signal quality safely through high-level SCPI tools.

Category
访问服务器

README

siglent-sds-mcp

MCP server for controlling a SIGLENT SDS824X HD oscilloscope via SCPI over raw TCP.

Project status: SDS824X HD hardware-tested alpha. Core measurement-driven auto setup is functional on real hardware.

Quick start

# Install
python -m venv .venv && source .venv/bin/activate
pip install -e '.[dev]'

# Test
pytest -q

# Run MCP server (stdio, for AI client config)
python -m siglent_sds_mcp.server

# Quick connectivity probe
python examples/tcp_idn_test.py <scope-ip>

Auto setup — one-command screen setup

The main feature: point at an unknown signal and let the scope measurement engine find usable VDIV / OFST / TDIV settings.

# Backwards-compatible CLI name; internally uses measurement-driven auto setup
python examples/auto_find_waveform_tcp.py <scope-ip> --channels C1 C2 C3 C4

# With signal-type hint for better trigger slope policy
python examples/auto_find_waveform_tcp.py <scope-ip> --signal-hint uart

# Direct TTL wiring / 1X probe
python examples/auto_find_waveform_tcp.py <scope-ip> \
    --channels C1 \
    --signal-hint clock \
    --probe 1

# Restart acquisition after capture only when explicitly requested
python examples/auto_find_waveform_tcp.py <scope-ip> --restart-after-capture

Default behavior: leave_stopped=true. The tool intentionally leaves the scope stopped on the final visible frame. The return object includes screen_hold, final_panel_state, measurements, final_settings, and probe_steps.

Architecture

MCP client (AI)
  │ MCP tool calls (stdio)
  ▼
server.py — FastMCP tools, auto-reconnect
  │
  ▼
sds_tcp_adapter.py — SDS800X HD command adapter
  │  channel / acquisition / trigger / measure / screenshot /
  │  waveform capture (WAVEDESC adaptive decode + envelope decimation) /
  │  measurement-driven auto_setup
  ▼
tcp_transport.py — RawTcpTransport
  │  socket-level SCPI, IEEE 488.2 binary block parser,
  │  thread-safe (RLock), pre-query socket flush
  ▼
auto_setup.py — compatibility wrapper for historical auto_find_waveform API

SIGLENT SDS824X HD oscilloscope (LAN port 5025)

Command verification pipeline

candidate → official-doc → tested → implemented → safe-tool

Tracked in docs/sds824x-hd-command-matrix.md. Do NOT expose an untested command as a default MCP tool.

Key design decisions

Measurement-driven auto setup

SDS800XHDTcpAdapter.auto_setup() uses scope measurements (PKPK, MEAN, FREQ, PER, MAX, MIN) to select display settings. This avoids relying on a separate offline CSV analyzer for first-pass screen setup.

WAVEDESC adaptive decode

WF? DAT2 returns 8-bit signed bytes. Voltage decode queries WF? DESC for the WAVEDESC descriptor and uses the descriptor-derived codes_per_div with current panel VDIV? / OFST? for decoding.

Min/max envelope decimation

When max_points < raw sample count, each bucket outputs min + max voltages instead of naive stride-sampling. Preserves glitches/spikes stride would miss.

ARM/STOP behaviour

get_waveform() must ensure waveform memory reflects the current panel settings before WF? DAT2. This path is still hardware-sensitive; keep validation notes in docs/sds824x-hd-command-matrix.md up to date when changing ARM/SINGLE/AUTO behaviour.

Trigger level policy

C?:TRLV <level> is a known issue on SDS824X HD firmware 4.8.12.1.1.6.5. Display-oriented auto setup should not depend on this command by default.

Project structure

src/siglent_sds_mcp/
  server.py              — MCP tools, auto-reconnect, FastMCP
  sds_tcp_adapter.py     — Command adapter, WAVEDESC decode, envelope, auto_setup
  tcp_transport.py       — Raw TCP socket, lock, binary block parser
  auto_setup.py          — Compatibility wrapper for auto_find_waveform API
  uart_analyzer.py       — Offline UART CSV analyzer
  rs485_analyzer.py      — RS485 differential pair analyzer
  modbus_timing.py       — Modbus RTU timing calculator
  report.py              — Markdown field report generator
  artifacts.py           — Timestamped artifact paths, JSON writer
  transport.py           — PyVISA fallback (legacy, not wired into MCP)
  scope_driver.py        — SiglentSDSDriver with safety gate (legacy)

docs/
  architecture.md                  — Layered design, UART capture reference
  sds824x-hd-command-matrix.md     — Per-command verification status
  sds824x-hd-knowledge-base.md     — Instrument-specific knowledge
  verification-workflow.md         — Hardware verification procedure

tests/   — pytest, parser tests, TCP transport tests, auto setup helper tests
examples/ — auto_find_waveform_tcp, TCP IDN probe, waveform/RS485 capture

Safety model

Allowed Blocked
*IDN?, run/stop/single *RST, factory reset
Channel/timebase/trigger setup Firmware update
Measurement query Network config changes
Screenshot/waveform fetch File deletion/formatting
Offline waveform analysis Arbitrary SCPI writes

Raw SCPI writes are NOT exposed as MCP tools. safe_scpi_query_tcp only accepts ?-suffixed commands.

Test coverage

pytest -q

Key test areas:

  • test_unit_parsing.py_parse_voltage, _parse_time, _parse_sample_rate
  • test_wavedesc.py / test_wavedesc_parser.py — synthetic WAVEDESC decode, ASCII prefix handling
  • test_tcp_binary_prefix.pyquery_binary IEEE 488.2 / BMP prefix skipping
  • test_auto_setup.py_pick_vdiv, _pick_tdiv, measurement parser and SCPI number formatting
  • test_tcp_transport_parser.py — socketpair binary block parsing

Target device

  • SIGLENT SDS824X HD / SDS800X HD family
  • SCPI over raw TCP, port 5025
  • Firmware verified: 4.8.12.1.1.6.5

Reference

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选