Integrate OpexMX with Your Machines

How OpexMX connects to industrial machines via the OpexMX Edge Connector and FOCAS Agent.

Overview

OpexMX provides a complete edge-to-cloud machine integration using two open-source components:

• OpexMX Connector — A Rust binary running on a Raspberry Pi that reads Modbus TCP, OPC-UA, and MQTT data from your machines and pushes it to the OpexMX cloud.
• OpexMX FOCAS Agent — A companion Windows agent for Fanuc CNC machines that reads data via the FOCAS2 library and publishes it via MQTT to the connector.

No custom gateway code is needed. Configure YAML, deploy, and your machines are online.

Architecture

Factory Floor                                   Cloud
┌───────────────────────────┐                  ┌──────────────────┐
│  OpexMX Edge Connector    │                  │  OpexMX Platform │
│  (Raspberry Pi)           │   HMAC-signed    │  (Cloudflare)    │
│                           │   HTTP push      │                  │
│  Modbus TCP ◄──── PLCs    │─────────────────►│  REST API        │
│  OPC-UA    ◄──── SCADA    │                  │  CMMS            │
│  MQTT      ◄──── Sensors  │                  │  Dashboards      │
│                           │                  │  Analytics       │
│  SQLite (local buffer)    │                  │  Alerts          │
│  React Dashboard (local)  │                  │                  │
└───────────┬───────────────┘                  └──────────────────┘
            │ MQTT
            │
┌───────────┴───────────────┐
│  Windows PC (optional)    │
│  OpexMX FOCAS Agent       │
│  fwlib30.dll ◄── Fanuc CNC
└───────────────────────────┘

Supported Protocols

Quick Start: Edge Connector

1. Flash Raspberry Pi

# Standard Raspberry Pi OS Lite (64-bit)
# SSH into the Pi

2. Deploy the Connector

# From your development machine (macOS/Linux)
git clone https://github.com/dzulfikar08/opexmx-connector-v2.git
cd opexmx-connector-v2

# Cross-compile for ARM64
rustup target add aarch64-unknown-linux-musl
cargo zigbuild --release --target aarch64-unknown-linux-musl --all-features

# Deploy to Pi
./deploy.sh pi@192.168.1.100

The deploy script copies the binary, dashboard, and config, then installs a systemd service.

3. Configure Machines

Edit /opt/opexmx-connector/config.yaml:

connector:
  name: factory-floor-1
  poll_interval_ms: 5000
  push_endpoint: https://your-opexmx-instance.com/api/machine-parameter-values
  hmac_secret: ${HMAC_SECRET}

adapters:
  modbus:
    enabled: true
    devices:
      - id: plc-line-1
        host: 192.168.1.10
        port: 502
        unit_id: 1
        registers:
          - address: 0
            name: spindle_speed
            type: uint16
            mapping: spindleSpeed
          - address: 1
            name: axis_x_position
            type: float
            mapping: position
            axis_name: X
          - address: 10
            name: machine_state
            type: uint16
            mapping: state

  mqtt:
    enabled: true
    external:
      enabled: true
      url: 0.0.0.0
    devices:
      - id: fanuc-mill-1
        topic: opexmx/focas/fanuc-mill-1
        data_type: json

4. Start

sudo systemctl start opexmx-connector
sudo systemctl enable opexmx-connector  # auto-start on boot

Open http://192.168.1.100:3000 for the local dashboard showing live machine data.

Quick Start: Fanuc CNC (FOCAS Agent)

For Fanuc CNC machines with FOCAS2 support:

1. Build the Windows Agent

# From macOS/Linux
git clone https://github.com/dzulfikar08/opexmx-focas-agent.git
cd opexmx-focas-agent
./build.sh
# Output: target/x86_64-pc-windows-gnu/release/opexmx-focas-agent.exe

2. Deploy to Windows PC

Copy to the Windows PC near the CNC:

1. opexmx-focas-agent.exe
2. config.yaml (from config.example.yaml)
3. fwlib30i64.dll (from Fanuc FOCAS2 SDK)

3. Configure

agent:
  poll_interval_ms: 5000
  dll_path: fwlib30i64.dll

mqtt:
  broker: 192.168.1.100    # RPi connector's IP
  port: 1883
  topic_prefix: opexmx/focas

machines:
  - id: fanuc-mill-1
    name: Fanuc VF-2SS
    host: 192.168.1.10     # CNC IP address
    port: 8193              # FOCAS port
    max_axes: 3

4. Run

opexmx-focas-agent.exe

The agent reads CNC status, axis positions, spindle speed, and alarms, then publishes to MQTT. The RPi connector picks it up automatically.

Data Format

All adapters output the same CanonicalMachineData JSON:

{
  "source": "modbus",
  "timestamp": "2025-05-12T09:30:00.000Z",
  "machine": { "id": "plc-line-1", "name": "Line 1 PLC" },
  "status": { "state": "running" },
  "axes": [
    { "name": "X", "position": 125.450, "unit": "mm" },
    { "name": "Y", "position": 50.200, "unit": "mm" }
  ],
  "spindle": { "speed": 4500 },
  "metadata": {
    "spindle_speed": 4500,
    "machine_state": 1
  }
}

Machine states: running, idle, alarm, offline.

Modbus Register Mapping

Map any Modbus register to a semantic field:

Register types: uint16, int16, uint32, int32, float. Apply scaling with the scale field.

Local Features

The connector works even without cloud connectivity:

• SQLite storage — All data stored locally with 7-day retention
• React dashboard — Live WebSocket view, status, device list, history browser
• REST API — Health checks, device status, history queries, config reload
• Retry buffer — 1000-entry FIFO buffer for failed pushes, auto-flushes on reconnect

Security

• HMAC-SHA256 signatures on all cloud push requests
• Environment variable substitution in config (${SECRET} syntax)
• Local network only — connector runs inside your factory LAN
• No inbound connections required — the connector pushes outbound to your OpexMX instance

Monitoring

Direct API Integration

If you prefer to build your own gateway, you can POST directly to the OpexMX API:

POST /api/machine-parameter-values

{
  "machineParameterId": "SPINDLE_SPEED",
  "value": "1500",
  "createdAt": "2025-12-04T07:24:20.069Z"
}

This is what the connector's pusher calls internally. Use this approach for custom integrations or when the connector doesn't support your protocol yet.

Summary

1. Deploy the OpexMX Connector on a Raspberry Pi in your factory network
2. Configure YAML with your machine protocols (Modbus, OPC-UA, MQTT)
3. For Fanuc CNC, deploy the FOCAS Agent on a nearby Windows PC
4. Data flows automatically from machines → connector → OpexMX cloud
5. Monitor locally via the built-in React dashboard at http://pi-ip:3000

No custom code. No vendor lock-in. Open-source connectors you can modify for your needs.