IMDHMIControls LibraryShop main store

Selection guide

SCADA and PLC platform selection for plant revamps

SCADA and data logging Published 8 min read

Short answer

Judge the whole chain — engineering tool, exact controller, fieldbus role, OPC UA endpoint, SCADA runtime and client — not the IDE alone. Freeze pass-or-fail gates on the license boundary, target hardware, fieldbus, OPC UA and documentation, then pilot one representative plant unit on the real controller. Adopt only after deployment, protocol behavior, diagnostics and failure recovery all pass there; until then the incumbent platform stays the baseline.

On this page
  1. The chain you are actually selecting
  2. Freeze the gates before you compare tools
  3. The license gate
  4. The target hardware and fieldbus gate
  5. The OPC UA gate
  6. Size the pilot to one plant unit
  7. Run the failure-mode tests
  8. The decision for each case
  9. Questions asked

A revamp platform decision is not a choice of IDE. You are selecting a whole chain: the engineering tool, the exact controller hardware, the fieldbus interface, the OPC UA endpoint, the SCADA runtime and the client on the wall. Commit to a platform only after a pilot on that real controller has passed deployment, protocol behavior, diagnostics and failure recovery. Freeze the pass-or-fail gates before you install anything: the license boundary, the target hardware and its fieldbus role, the OPC UA role, and the documentation. A candidate that cannot name its controller, its PROFINET role and the license terms for every component that runs is eliminated before piloting. Until one candidate clears every gate on real hardware, your incumbent platform stays the baseline for live revamp work. Below are the gates, a pilot sized with worked numbers, the failure-mode tests, and the decision for each outcome.

The chain you are actually selecting #

Six layers make up the platform, and they fail separately. The engineering tool — editor, compiler, simulator — is layer one. The controller hardware and its firmware are layer two. The fieldbus interface is layer three: in a plant revamp, usually the controller acting as a PROFINET scanner for remote I/O. Layer four is the information path, an OPC UA server in the controller or in a separate component beside it. Layer five is the SCADA runtime that owns tags, alarms and history; layer six is the client on the wall.

Evaluations go wrong when the IDE is judged as the product while the runtime and the communications stay untested. A strong SCADA pick cannot compensate for a PLC ecosystem that has no deployable hardware, lacks the required bus or ships unusable diagnostics. Where each of these layers sits in the working discipline is laid out in the overview of the PLC engineer role across disciplines and platforms.

Freeze the gates before you compare tools #

Write the gates down as pass-or-fail before opening any vendor page. A gate is binary: it either names a documented fact — a datasheet line, a license term, a manual — or the candidate fails it. This table is the whole article in one view:

Gate Verify on Pass condition
License Engineering tool, simulator, compiler, SCADA runtime, protocol options, clients Every component the pilot needs is licensed for your scope, with no per-runtime, per-tag or per-client cap your sizing exceeds
Target hardware Exact controller order designation and firmware The datasheet and manuals name the fieldbus interface, the I/O support and the diagnostics for that exact target
Fieldbus role The controller PROFINET interface The target supports the role you need — scanner for remote I/O, device for coupling — on that order number
OPC UA role The endpoint the SCADA connects to Namespace browses from the SCADA, subscriptions deliver at the configured rate, writes are validated in the controller, quality goes bad on failure
Documentation Hardware, fieldbus, OPC UA, diagnostics and deployment manuals Each manual exists for this target and version before implementation starts
Pilot One representative plant unit, on real hardware Deployment, protocol behavior, diagnostics and every failure-mode test pass, with results recorded

The license gate #

A free download is not a free deployment right. Check separately: the engineering tool, the compiler, the simulator, the protocol configuration, the right to download to hardware, the SCADA runtime license, and the client license. Runtime licensing commonly counts installations, tags or clients, and a protocol option can carry its own entitlement even when the engineering tool itself is a no-charge download.

List every component that will run in the pilot, then apply one rule: if any of them is licensed per installation, per tag or per client, and that piece is not free for your scope, the platform fails the licence-free requirement as stated. If the real constraint is only on the PLC engineering tool, say so and re-run the SCADA candidates under the relaxed rule — a gate applied to the wrong layer silently rejects good platforms and passes bad ones.

The target hardware and fieldbus gate #

Fieldbus support is a property of the exact controller, not of the software family. At least one soft-PLC ecosystem lists PROFINET support as an optional, separately ordered component tied to specific targets — which is exactly why capability claims are read against the target datasheet and never transferred from the IDE feature list to every device in the family.

For the order designation you intend to buy, establish from its own documentation: that the fieldbus interface exists on that hardware, which role it supports (scanner to drive remote I/O, or device to couple into an existing network), which I/O device classes it accepts, what diagnostics objects it exposes, and whether any of it needs an added component or entitlement. The decision rule for the whole gate: if the documentation cannot name the interface, the role and the I/O support for that order number, the candidate fails, whatever the IDE looks like. And if its controllers reach remote I/O only over a bus the revamp has excluded — EtherCAT, for instance — this gate fails and the conversation ends.

The OPC UA gate #

PROFINET and OPC UA do different jobs, so both can run: PROFINET carries deterministic I/O at the plant unit, OPC UA carries information up to the SCADA. Test the path as it will run. Browse the namespace from the SCADA and confirm the names are stable across a controller download. Subscribe to values and check they arrive at the configured publishing interval. Issue a command and confirm the controller, not the screen, validates it. Break the link and confirm quality reports bad rather than leaving a stale value that looks alive.

Define ownership before the pilot starts: the controller owns equipment states, permissives, interlocks and alarms; the SCADA owns presentation, history and operator workflow. Safety note: keep every permissive and interlock in the controller, so a lost link, a dead screen or a stale tag can never be the thing that starts equipment or clears a permissive.

Size the pilot to one plant unit #

Pick the largest or most typical single unit in the plant, not a demo. The unit must include one sequence, one permissive chain, one alarm class, one analog measurement with engineering-unit scaling, one operator command with feedback, one status echo, and one communications-failure state — a motor demo hides everything plant engineering actually does. The pilot station itself runs at the line, so run the SCADA client and the data logging on an industrial panel PC in the cabinet door rather than an office desktop.

Worked example, illustrative numbers. Give each of the unit's 48 digital I/O points a status tag, a command tag and a fault tag: 48 × 3 = 144 tags. Give each of its 12 analog loops a scaled value, a raw value, a quality flag, a high-alarm bit, a low-alarm bit, a setpoint and an output: 12 × 8 = 96 tags. Add 8 unit-level tags for running, ready, permissive, communications health and the like: 144 + 96 + 8 = 248 tags. If the candidate licenses the runtime by tag count and the tier under consideration caps at 200 tags, this one unit exceeds it — the architecture needs a higher tier or a split into two runtimes, and you learned that in one pilot instead of one contract.

Run the failure-mode tests #

Execute the list on the real controller and record what the operator sees every time:

  1. Break the controller-to-SCADA link during operation: quality must show bad, values must blank or visibly freeze, and reconnection must restore names unchanged.
  2. Restart the SCADA runtime while the controller keeps running: permissives and outputs must hold, and nothing may move because a screen came back.
  3. Download or restart the controller: the SCADA must show bad quality, not the last values, and the namespace must survive or every break point must be documented.
  4. Refuse a permissive and issue the command from the client: the command must be rejected at the controller, and the screen must not display it as accepted.
  5. Force an analog fault, such as an open input: bad quality must reach the client and the alarm list, and every control using that value must take its configured fault action.
  6. Power-cycle a fieldbus device: the failure must be readable in diagnostics at both the controller and the SCADA without opening a cabinet.

Then budget the alarm path. Worked example, illustrative numbers: controller scan 10 ms + OPC UA publishing interval 500 ms + SCADA runtime update 250 ms + client draw 250 ms = 1010 ms, about 1.0 s worst case from condition to screen. If the requirement you are designing to is alarms visible within 2 s, the chain passes with margin. The same chain with a 5000 ms publishing interval totals 10 + 5000 + 250 + 250 = 5510 ms, about 5.5 s, and fails — and the fix is the publishing interval or the alarm class, not a different SCADA platform.

The decision for each case #

  • Every gate passes on the real controller: adopt. Run the first live revamp on the new chain with an incumbent project open beside it as the reference for sequence logic, alarms and diagnostics.
  • Target hardware or fieldbus gate fails: eliminate the candidate outright. IDE quality cannot rescue a controller that lacks the interface, the role or the diagnostics.
  • License gate fails: eliminate for this scope. If the licence-free requirement applies only to the PLC engineering tool, re-run the SCADA candidates under the relaxed rule and keep the gate intact on the PLC side.
  • Pilot passes the logic but fails the failure-mode tests: fail this release, record the firmware and SCADA versions that failed, and retest after the next update.
  • The incumbent passes every gate: the revamp proceeds on the incumbent, and the new platform stays a structured learning exercise until it clears a gate the incumbent fails.

The pilot doubles as your syllabus, because the gaps it exposes are exactly the ones the from-scratch PLC, SCADA and DAQ roadmap sequences. The platform you adopt also becomes a career asset: the SCADA engineering career path and vendor platform roadmap maps platform skills to the roles they lead to.

Questions asked

Can you pick the SCADA platform before the PLC platform is settled?

Shortlist SCADA candidates at any time, but do not commit either choice alone. The license boundary, the tag count and the OPC UA behavior cross both layers, and a SCADA runtime that cannot browse the controller namespace or propagate bad quality fails the pair, not just one side. Test them together in one pilot on the real controller and record the result as one decision.

Is a simulator pilot enough to adopt a platform?

No. Simulation proves your logic compiles and runs; it proves nothing about the target controller fieldbus role, the OPC UA server behavior, the diagnostics or the license enforcement. Deploy to the exact controller you intend to buy. Keep the simulator afterwards as the logic debugger, because it lets you develop sequences without stopping a live unit.

How many tags should the pilot unit carry?

Enough to stress licensing and update behavior, which a small demo never does. A unit with 48 digital points and 12 analog loops produced 248 tags in the worked example above, and that number exposed a 200-tag license tier immediately. Scale the pilot unit to the largest single unit in your plant and repeat the tag-count arithmetic before any license tier is chosen.

When is the pilot finished?

When the list is done, not when a date arrives. The pilot ends when deployment to the real controller, every failure-mode test, one controller download with the SCADA reconnecting, and one full alarm path from input to screen have each been executed once with results recorded. If an item cannot be completed because documentation is missing, that is the finding and the gate fails.

What eliminates an unsuitable platform fastest?

The target hardware and fieldbus gate. Ask for the exact controller order designation and its firmware, then require the datasheet line that names the fieldbus interface, the role it supports and the diagnostics objects. A candidate that answers with the IDE feature list instead of the controller datasheet is eliminated on the spot, before any pilot time is spent.

Keep reading

  • How-to

    How to learn PLC, SCADA and DAQ from scratch

    Learn the three as one project, not three courses: pick one route, wire one real input, program a three-state sequence, exchange one Modbus register, bind SCADA tags and store a timestamped value. Pass the five checkpoint tests on the way; simulation alone never exercises terminals, polarity or reconnection.

    SCADA and data logging

  • Troubleshooting

    Why HMI values show question marks: PLC comms timeouts

    Yes — a running PLC program can cause intermittent HMI timeouts by consuming nearly all controller execution time, and clean Ethernet counters do not clear it: utilization measures bytes, not the cost of servicing requests. Work controller load first: Task Monitor above 90% on the dropping units against 7–15% on healthy ones, then message pacing, packet-rebuild address forms, read grouping, and power last.

    PLC programming