Siemens 840D PCU50 HMI Operate Production Data Acquisition

System Architecture Overview

The SINUMERIK 840D sl with PCU50 (Panel Control Unit) and HMI Operate is a PC-based CNC platform where the HMI runs on Windows (typically Windows 7 Embedded / Windows 10 IoT depending on hardware generation). The NCK (NC kernel), PLC (SINUMERIK Integrated PLC, S7-compatible), and HMI exchange data through internal coupling mechanisms that an integrator can leverage for shift-log reporting.

Key subsystems for production data capture:

• NCK – exposes channel status, axis data, tool data, feedrate/spindle override, program state via NC variables (read with PI service PI _N_NVRD or via OPC UA NC server).
• PLC – exposes user alarms, handshakes, RFID reader data, station interlocks via DB/M/I/Q areas accessible to the HMI.
• PCU50 HMI Operate – hosts the operator interface and can run user scripts, log files, batch files, or third-party collectors that pull from NC/PLC.
• Network – X120/X130/X127 industrial Ethernet ports expose the PLC to higher-level systems via S7 communication, OPC UA, or TCP socket.

For machines with 50+ axes and multi-channel configuration (common in transfer lines and large gantries), the HMI Operate is the natural collector because it has visibility into all channels and the PLC at once.

Production Data Points and Source Mapping

The shift log requires time-stamped events. The table below maps each required field to its origin and the recommended read mechanism.

Built-In HMI Operate Logging Capabilities

HMI Operate writes several internal logs that can be redirected, parsed, or post-processed:

• oem_alm_log.txt – operator-visible alarms (NCK + PLC), path C:\Program Files (x86)\Siemens\Automation\HMI-Operate\log or its Windows-10 IoT equivalent.
• crash_alm_log.txt – crash / system-stop alarms.
• Program list and tool list under NC:\mpf and NC:\tools on the local CFast/SSD card (HMI Operate exposes the NC file system to Windows as a mapped drive on supported builds).
• Operator login audit through Windows event log or the Operate-specific user directory.

For most commissioning data needs, the alarm log alone does not meet the requirements because it does not include feedrate-override changes, tool-change timing, or RFID codes. A custom collector is normally required.

Method 1: MT-Connect Integration

MTConnect is an open, royalty-free manufacturing data standard. The standard defines an XML/HTTP protocol that the agent polls from an adapter running on or near the machine. SINUMERIK 840D sl supports MTConnect via:

1. Running the MTConnect adapter as a Windows service on the PCU50 itself (requires the SINUMERIK Integrate Client or a custom adapter built on libmtconnect).
2. Reading NC and PLC data through a bridge: the adapter polls NC variables (via the SINUMERIK HTTP server or PI _N_NVRD) and PLC tags (via S7 communication or LibNoDave / Snap7).
3. Publishing to a local MTConnect agent which serves /current and /sample endpoints over HTTP.

Field experience: MTConnect delivers raw data; the integrator must build the time-series database, event correlation, and report layer. Plan 4-8 weeks of effort for a single-machine deployment that includes alarm correlation, override-event capture, and RFID stitching.

Method 2: OPC UA NC Server

For new machines or those upgraded to SINUMERIK Operate V4.x or later, the integrated OPC UA NC Server publishes the same address space browsed in HMI Operate to any OPC UA client. This is the most scalable path for multi-channel 840D sl machines because one client can subscribe to all channels.

Recommended client pattern:

1. Subscribe to the data-change nodes corresponding to progState, feedRateOvr, tool/tNo, and a custom RFID node exposed by the PLC.
2. On the client side, persist only the deltas with a local timestamp.
3. Push deltas as JSON lines (JSONL) to a database, queue, or MES.

Method 3: Direct S7 / PLC Polling

All custom events that do not exist as NC variables (RFID scans, operator-ID handshakes, station interlocks, material load/unload) are typically in the PLC. The PCU50 HMI is a full Windows host, so any of the following works:

• Snap7 open-source S7 client (works against 840D sl PLC, port 102).
• Siemens S7.NET or libnodave (legacy).
• A commercial OPC DA/UA to SQL bridge.

Sample polling loop (C# pseudocode):

// Snap7 example: read RFID string and operator ID every 250 ms
var client = new S7Client();
client.ConnectTo("192.168.214.1", 0, 1, 102);   // PCU50 PLC
byte[] buf = new byte[64];
int res = client.DBRead(100, 0, buf.Length, buf);
string rfid = System.Text.Encoding.ASCII.GetString(buf, 0, 32).Trim();
string operatorId = System.Text.Encoding.ASCII.GetString(buf, 32, 16).Trim();
if (rfid != lastRfid) { log("RFID", rfid); lastRfid = rfid; }

Method 4: Third-Party DNC / Production Reporting Software

Off-the-shelf tools commonly deployed against 840D sl include:

For a custom 50+ axis multi-channel machine, generic "out of the box" OEE packages typically require a services engagement to model the channel layout, station interlocks, and RFID handshakes. Plan a scoping phase before purchasing seats.

Cell Controller Reporting Reference Architectures

For context, fully integrated cell controllers from Toyoda, Fastems, Mori Seiki, and Makino log all the data the original post asks for: program name, tool changes, override, alarms, RFID, RGV movement, and produce scheduled reports. These are vertical solutions: the cell controller is a separate PC, not HMI Operate, and it polls each machine through MTConnect or vendor-proprietary protocols.

If the customer eventually scales to a flexible manufacturing system, consider running the 840D sl machine as an MTConnect data source and consuming it from a cell controller, rather than building two parallel reporting systems.

Recommended Architecture for a Custom 50+ Axis 840D sl Machine

1. Enable the OPC UA NC Server on the PCU50 (license permitting) and verify namespace exposure with a generic UA client (UA Expert).
2. Add a Windows service on the PCU50 called DataCollector.exe that: - Subscribes via OPC UA to NC variables on all configured channels.
- Polls the PLC DBs (RFID, operator ID, station state) every 250 ms via Snap7. - Maintains a circular buffer of last-known values and emits a JSONL line only on state change. - Writes to a local SQLite (no external DB dependency on the machine) plus forwards lines to a network share / MQTT broker.
3. Run a separate reporting service (off-machine) that ingests the JSONL stream and produces the shift report (CSV + HTML at end of shift).
4. Sync PCU50 time via NTP to a domain controller. Without NTP, the timestamp is unreliable across reboots and DST transitions.

NC Variable Polling Example (Windows Shell + SINUMERIK HTTP Server)

curl http://192.168.214.1:80/sinumerik/nck/data/.../progState
# or use the SINUMERIK Integrate HTTP REST endpoint to read a single variable
# (endpoint paths depend on firmware; consult the Commissioning Manual for your SW version)

Commissioning Checklist

Troubleshooting Matrix

Performance and Sizing Notes

For a 50+ axis machine with 4-6 channels and 1-second event resolution:

• OPC UA subscription throughput is not a bottleneck; 1-second scan rate handles 100+ monitored nodes.
• Snap7 polling at 250 ms across 4 DBs (RFID, operator, station, override) is well within PCU50 CPU budget.
• JSONL line volume: 50-200 events per shift is typical; well under 1 MB per shift even verbose.
• Disk: reserve 5 GB for the rolling JSONL archive; SSD or industrial CFast is sufficient.

FAQ