CNC Machine Control Mobile App Development

Developing a Mobile App for CNC Machine Control

A CNC machine is neither a 3D printer nor an IoT sensor. Control systems like Fanuc, Siemens Sinumerik, Heidenhain TNC, LinuxCNC are industrial equipment with strict reliability, safety, and precision control requirements. A mobile app in this context doesn't replace the control panel—it complements it: program monitoring, NC file upload, spindle and axis telemetry, emergency stop alerts.

Industrial CNC Interfaces

Each CNC manufacturer provides its own API. Openness varies:

MTConnect—an open standard (ANSI/MTC1.4) for reading machine state. MTConnect Agent adapts proprietary data to an XML stream. Most major CNC manufacturers support MTConnect as an optional interface.

MTConnect: Reading Machine Data

MTConnect Agent provides HTTP API. Streaming via current (snapshot) and sample (history with buffer):

class MTConnectClient(private val agentUrl: String) {

    suspend fun getCurrent(): MTConnectDocument {
        val response = httpClient.get("$agentUrl/current") {
            accept(ContentType.Application.Xml)
        }
        return parseMTConnectXml(response.bodyAsText())
    }

    // Long-polling stream of changes from sequence number
    fun streamSamples(from: Long? = null): Flow<MTConnectEvent> = flow {
        var nextSequence = from
        while (true) {
            val url = if (nextSequence != null)
                "$agentUrl/sample?from=$nextSequence&count=100"
            else
                "$agentUrl/sample?count=100"

            val response = httpClient.get(url)
            val doc = parseMTConnectXml(response.bodyAsText())

            doc.streams.forEach { stream ->
                stream.events.forEach { event -> emit(event) }
            }

            nextSequence = doc.header.nextSequence
            if (doc.streams.isEmpty()) delay(500)
        }
    }
}

Typical DataItems from MTConnect stream: execution (ACTIVE/STOPPED/INTERRUPTED), program (current program name), line (G-code line number), Xact/Yact/Zact (actual axis positions), Sspeed (spindle RPM), load (spindle load %).

OPC UA: Siemens Sinumerik

For Sinumerik 840D sl—OPC UA Server is built into NCU. Siemens namespace (urn:Siemens:SINUMERIK:NC) contains nodes for all CNC parameters: /NC/Channel/State/actSystemTime, /NC/Spindle/actSpeed, /NC/Axis/actPos.

Connection via Eclipse Milo (Java/Kotlin):

class SinumerikOpcUaClient(private val endpoint: String) {
    private lateinit var client: OpcUaClient

    suspend fun connect() {
        val endpoints = DiscoveryClient.getEndpoints(endpoint).await()
        val ep = endpoints.first { it.securityPolicyUri == SecurityPolicy.None.uri }

        client = OpcUaClient.create(ep.endpointUrl,
            endpointFilter = { it == ep },
            configurer = { config ->
                config.setIdentityProvider(UsernameProvider("OpcUaClient", "password"))
            })
        client.connect().await()
    }

    suspend fun readSpindleSpeed(): Double {
        val nodeId = NodeId.parse("ns=2;s=/NC/Spindle[u1,1]/actSpeed")
        val value = client.readValue(0.0, TimestampsToReturn.Both, nodeId).await()
        return (value.value.value as Number).toDouble()
    }

    fun subscribeToAxisPositions(callback: (AxisPositions) -> Unit) {
        val subscription = client.subscriptionManager.createSubscription(500.0).await()
        val nodes = listOf("Xact", "Yact", "Zact").map { axis ->
            NodeId.parse("ns=2;s=/NC/Channel[u1,1]/MachineAxis[$axis]/actPos")
        }
        // MonitoredItem for each axis with 100 ms interval
        subscription.createMonitoredItems(
            TimestampsToReturn.Both,
            nodes.map { MonitoredItemCreateRequest(ReadValueId(it, AttributeId.Value.uid(), null, null),
                MonitoringMode.Reporting, MonitoringParameters(0.0, 100.0, null, 10, true)) },
        ) { item, _ ->
            item.setValueConsumer { _, value ->
                // update axis position
            }
        }
    }
}

Safety of Control Commands

Writing commands to CNC (program change, cycle start) is a potentially dangerous operation. Requirements:

1. Authentication and authorization: only authorized operators send commands. JWT with short TTL, MFA for critical operations.
2. Network zone: CNC in isolated production network, API gateway is the single entry point with command logging.
3. Locks: program start is blocked if machine door is open (data from safety PLC). Mobile app doesn't bypass physical interlocks.
4. Audit: every command logged with timestamp, user ID, source IP.

suspend fun startProgram(programName: String) {
    // Pre-check state
    val state = getMachineState()
    require(state.doorsClosed) { "Machine door is open" }
    require(state.execution == Execution.STOPPED) { "Machine is not stopped" }
    require(state.emergencyStop == EmergencyStop.ARMED) { "E-Stop not armed" }

    auditLogger.log(Action.START_PROGRAM, programName, currentUser)
    mtConnectAdapter.sendCommand(StartProgramCommand(programName))
}

Uploading NC Programs

G-code programs (.nc, .mpf for Siemens, .cnc for Fanuc) are uploaded via FTP (LinuxCNC, most industrial CNCs support FTP server) or proprietary DNC interface. In Flutter:

Future<void> uploadNcProgram(File program, String remotePath) async {
  final ftp = FtpConnect(
    host: machineIp,
    user: ftpUser,
    pass: ftpPassword,
    timeout: 30,
  );

  try {
    await ftp.connect();
    await ftp.changeDirectory(remotePath);
    final uploaded = await ftp.uploadFile(program);
    if (!uploaded) throw Exception('FTP upload failed');
  } finally {
    await ftp.disconnect();
  }
}

Developing a mobile app for CNC monitoring and control via MTConnect or OPC UA: 10–16 weeks. Complexity determined by CNC type and scope of control functions. Cost individually quoted after interface audit of specific equipment.