Five ways onto a dashboard, and a decision order for choosing.

What each connectivity method requires, what it delivers and what it cannot do — plus the per-machine decision order that produces the cheapest workable answer.

There are five ways a machine on an Indian shop floor ends up on a monitoring dashboard. Most plants use three of them simultaneously, because machine populations are mixed and the right method is a per-machine decision rather than a platform-wide one.

This article compares them on the criteria that matter when you are choosing: what each requires, what it delivers, and what it cannot do.

The five methods

Compared on what actually decides the choice.

MethodRequiresTypically deliversMain limitation
Network read from the controlA reachable control on the plant networkState, cycle timing, counts, alarm presenceDepends on the control generation and options fitted
OPC UAA device or control exposing an OPC UA serverNominated tags, structured and typedNot present on most older equipment; needs configuration on both sides
MTConnectAn MTConnect adapter or agent for the machineA standardised read-only streamAvailability varies by machine builder; often an added component
Modbus (RTU or TCP)A PLC, meter or drive exposing registersWhatever registers you nominateNo semantics — a register is a number until you map it
Digital and analogue I/OWiring to a gatewayStates, counts, sensor valuesNo controller-level detail; needs an installation slot
How to choose, per machine

A decision order that works.

ASK FIRSTIs the control reachable?If yes, read it over the network. Least hardware, least disruption.
THENIs there a PLC or meter?If yes, read nominated registers or tags over Modbus or OPC UA.
THENIs there a standardised stream?Where MTConnect is available on the machine, it is a clean read-only option.
OTHERWISEWire the signalsContactor, counter, stack light or added sensor into a gateway.

The order matters because each step down adds either hardware or installation time. Working top-down produces the cheapest workable answer for each machine rather than a single method imposed across a mixed floor.

Mixed floors are the normal case

A typical Indian machining unit ends up with networked reads on the newer CNCs, Modbus on the energy meters and any PLC-driven equipment, and gateway I/O on the presses and conventional machines. That is not a compromise; it is the correct outcome of applying the decision order above machine by machine.

What matters is that all five methods land in the same record with the same state vocabulary, so a press and a machining centre can be compared without anyone reconciling two systems. A platform that handles only one or two of these methods will cover part of your floor and leave the rest invisible — which is usually where the unmeasured losses are.

What each method cannot do

Stated plainly, because this is where overclaiming happens.

LIMIT

Network read

Cannot exceed what the control exposes. Options fitted at purchase constrain it years later.

LIMIT

OPC UA and MTConnect

Cannot be added to a machine that has no support for them without a component that provides it. Availability is a fact about your machine, not a preference.

LIMIT

Modbus

Carries no meaning. Register 40001 is a number until somebody maps it, and mapping is where errors enter.

LIMIT

Digital I/O

Gives states and counts, never program names, spindle load or alarm codes. Anyone claiming otherwise is describing a different machine.

Questions

Straight answers.

What are the main ways to connect a CNC machine for monitoring?
Reading the control over the network, OPC UA, MTConnect, Modbus RTU or TCP, and digital or analogue I/O through a gateway. Most plants use three of these at once because machine populations are mixed.
Which connectivity method is best?
There is no single best method — it is a per-machine decision. Work top-down: reachable control first, then PLC or meter, then a standardised stream, then wired signals. Each step down adds hardware or installation time.
Can any machine be connected somehow?
Effectively yes, but what you get differs. A machine with no control gives states and counts through wired signals, not program names or spindle load.
Is Modbus enough for machine monitoring?
It transports values but carries no meaning — a register is a number until it is mapped. The mapping work is where accuracy is won or lost.
Do I need the same method across all machines?
No, and insisting on it usually leaves part of the floor unmonitored. What matters is that every method lands in one record with the same state vocabulary.
How do I know which method applies to my machines?
A half-day inventory marking each machine as network-readable, PLC-readable or nothing-readable answers it, and it is the first step of any competent project.
FREE 2-MACHINE PILOT
Live OEE on two of your machines — this week. ₹0 to start.
Software-first setup in under an hour each · your data stays on-premises · plant-specific ROI model included.