How to Connect Older Machines to Production Tracking

Decide what a machine signal needs to explain

An older machine does not need to become a fully connected asset before it can contribute useful production information. The first question is what someone needs to decide. A planner may need to know whether an order has started. Maintenance may need the time a recurring stop began. Quality may need a count linked to the right batch. Each question asks for different evidence.

Start with the order and reporting process. Find where staff currently write quantities, record stops and correct mistakes. Watch a representative job with an operator, including setup and interruptions. Establish whether the missing information would actually change a decision. Collecting every available signal can leave the factory with more data and the same unanswered questions.

The accompanying connection survey helps define one useful event and assess how it might be acquired. It deliberately separates signal availability from signal meaning. An electrical pulse may be easy to capture and difficult to interpret. A manual confirmation may remain necessary where only a person can establish that the work is acceptable or belongs to a particular order.

Excellence industrielle's 2023 report describes a demonstration micro-factory connecting order programming to final assembly, alongside work on data management, cybersecurity and training. It illustrates why machine connectivity belongs within a wider operating discussion. It does not establish that every older controller can be connected or prescribe the technical approach for your plant. [1]

Inventory access without disturbing control

Identify the machine, controller, relevant software versions and available documentation. Ask the manufacturer or a qualified controls specialist which supported interfaces exist, what they expose and whether additional licensing or hardware is required. Confirm support and warranty implications before changing anything. Older machines vary considerably; age alone does not establish that a connection is possible or impossible.

Where supported, a read-only interface can provide useful events without issuing commands to the machine. Verify that the proposed access really is read-only and that polling or communication will not impair the controller. Keep the monitoring scope separate from control changes. Production tracking is not a safety system and must not become an unintended route for remote machine operation.

If no suitable interface exists, assess an independent sensing approach with qualified people. It may be possible to observe a cycle or state without modifying the control logic. The survey should record the engineering assessment, installation responsibilities and any limits on interpretation. It is not a wiring guide. Never bypass guarding, interlocks or safe isolation arrangements to obtain a convenient signal.

Define the event before designing the dashboard

Consider a hypothetical stamping operation. A cycle signal may count a test stroke, a rejected piece or a cycle with no acceptable finished part. Displaying that signal as good output would give planning a precise but wrong answer. The team needs to decide whether it is measuring cycles, total pieces, accepted pieces or completion of an order, and which additional confirmation each definition requires.

Give each event a machine identifier, timestamp and event definition. Where a count is involved, record its unit and whether the source provides an increment or a cumulative counter. Define how resets, rollover and machine restarts are recognized. Keep the raw observation distinguishable from any inferred state so an unexplained result can be investigated later.

Agree how jobs are associated with events. An operator scan or selection may be a sensible first step, provided it is validated against the current order and revision. Define what happens when nobody selects a job or when a changeover overlaps the selection. Do not quietly attribute unassigned production to the last order because it makes the chart look complete.

Make missing data visible

A lost network connection should not look like a machine standing idle. Include an indication of data freshness and distinguish unknown, offline and stopped states where the evidence permits. Establish how an interruption is detected, who investigates it and how the period is represented in reports. If the data cannot establish what happened, preserve the gap.

Decide whether a local gateway needs to buffer events during an outage. Specify the required outage coverage, available storage, clock handling and recovery behaviour. On reconnection, late events should retain their original timestamps and should not be counted twice. Test this behaviour with the implementation team before relying on the totals for scheduling or performance discussions.

Keep corrections traceable. Operators may identify an incorrect job selection or a count that includes setup work. Record the original event, the correction, its reason and the responsible person according to the site's access policy. A system that makes correction impossible encourages unofficial parallel records; a system that silently overwrites history makes disagreements harder to resolve.

Put a boundary around the connection

Bring the site's IT and operational-technology owners into the design. Agree network placement, permitted traffic, authentication, credentials and logging, rather than treating the gateway as an ordinary office device. Review supplier remote access separately, with explicit authorization and a way to disable it. The ability to read production data does not justify unrestricted access to controls.

Assign an owner for updates, backups and configuration changes. Document which events depend on which controller settings, sensors or software versions. A maintenance change that preserves machine function can still change the meaning of a monitoring signal. The connection needs a review trigger when equipment or production methods change, not only a commissioning sign-off.

Estimate the ongoing work honestly. Gateways, subscriptions, installation, troubleshooting and replacement parts all belong in the decision. Some machines may remain on structured manual reporting because a safe, reliable connection is not worth its cost. The goal is a better production decision, not a requirement to connect every asset regardless of value.

Compare the record with the work

Begin with one machine and a defined question after the operating need and access boundaries are understood. Compare the collected events with a separate observation record over representative production, changeovers, stops and restarts. Check timestamps, job attribution, counts and unknown states. Agree acceptable differences for the intended use rather than applying an unexplained accuracy percentage.

Ask the intended user to act on the information. Can the planner distinguish a late order from a stale connection? Can a supervisor investigate a count discrepancy without returning to handwritten notes? If the information does not support those actions, refine its definitions before expanding the installation. A successful data transfer is only one part of acceptance.

Quindart's approach connects operational information to the decisions it should support. On an older machine, a modest, well-defined event can be more useful than an elaborate dashboard built on ambiguous signals. Keep the people who understand the machine involved, preserve safe operating boundaries and expand only when the first connection has earned its place in the production process.