other
Blog
Can You Add Edge Analytics Without Rewriting the PLC? A PACSystems RX3i Brownfield Guide Oct 30, 2026
PACSystems RX3i · Brownfield Edge Lab

Can You Add Edge Analytics Without Rewriting the PLC?

A practical decision guide for plants that want better data, dashboards and analytics while protecting deterministic control, approved logic and production continuity.

Control firstData by designCompatibility proven
Deterministic controlLogic, I/O, interlocks and machine response
Approved data pathTags, timestamps, context and read/write boundaries
Analytics environmentStorage, visualization, calculations and applications
The question is not“Does RX3i support Ethernet?”
The real question is“Which data may cross which boundary?”
The proof isPerformance, security and rollback evidence
Why this topic matters now

AI interest is rising faster than brownfield change tolerance

Many plants already have useful production data inside a PACSystems RX3i application, but the running controller was designed to operate equipment—not to become an unrestricted data source for every new dashboard, cloud connector or analytics experiment. That creates a familiar conflict: management wants more visibility, maintenance wants earlier warning and engineering wants evidence, while operations wants the stable control system left alone.

The sensible answer is neither “connect everything” nor “replace everything.” The first step is to separate deterministic control from the data and application layers. Official PACSystems material describes RX3i as a scalable control architecture with industrial Ethernet, open protocols and application portability. Official PACEdge material adds data collection, storage, processing, visualization and edge applications. Those capabilities can support a phased brownfield strategy, but they do not remove the need for model-specific engineering.

An existing GE PACSystems RX3i installation may contain a CPE305, CPE310, CPE330, CPU315, CPU320, communication modules, PROFINET controllers or other combinations. The available ports, protocols, performance, firmware and redundancy features vary. A statement that is true for a CPE400 edge controller must not be applied automatically to every installed RX3i CPU.

Interactive brownfield decision compass

Point the compass at your biggest constraint

Select the pressure that is shaping the project. The compass does not choose a product; it exposes the first architecture route, evidence pack and technical boundary the team should review.

Four pressures · Four starting routes
RX3i EDGEDECISION CORE
Compass result · protect the shutdown window

Start with the existing approved interface

When outage exposure dominates the decision, first determine whether the installed CPU or Ethernet interface already supports a suitable, controlled data path. A limited read-oriented tag set can reduce control-layer disturbance, but the exact model, firmware, protocol, traffic and project baseline still require verification.

First routeExisting supported interface
Evidence to openCPU, firmware, ports, traffic baseline
Do not crossDo not hide logic or firmware changes inside the analytics scope
Compass result · define the data contract

Fix context before adding more tags

Analytics quality depends on more than access. Define tag ownership, engineering units, valid range, quality status, timestamp source, sampling need and communication-loss behavior. Keep advisory analytics separate from any command or setpoint path unless a distinct controls project authorizes it.

First routeSmall, explicit data contract
Evidence to openTag map, units, time source, loss behavior
Do not crossA large tag list is not proof of useful or trustworthy data
Compass result · separate change cycles

Place analytics outside the deterministic control contract

A separate edge environment can let dashboards, storage and analytics evolve without treating every application update as a PLC change. This route adds ownership for the operating system, users, applications, certificates, backups and network boundary, so those responsibilities must be assigned before deployment.

First routeSeparate supported edge platform
Evidence to openPower, cabinet, network, owner and rollback plan
Do not cross“At the edge” does not mean “inside the control decision”
Compass result · design for repeatability

Compare a shared edge layer with an integrated controller route

For several assets, compare a separate edge platform that aggregates approved interfaces with supported integrated RX3i edge controllers such as CPE400 or CPL410. Official material describes concurrent real-time control and PACEdge capability on these models; it does not establish universal drop-in compatibility.

First routeShared platform versus CPE400/CPL410 study
Evidence to openAsset count, protocols, loads, I/O and acceptance needs
Do not crossDo not treat a family name as migration proof
Architecture choices

Three ways to reach the data—without pretending they are equivalent

Choose by installed evidence

Use the existing approved interface

Keep the present controller and expose a limited set of data through a protocol and interface the exact installed hardware supports.

  • Lowest control-layer disturbance when already available
  • Requires protocol, performance and security verification
  • May need an existing Ethernet module or configured CPU port
  • Best when the data requirement is modest and clear

Add a separate edge computer

Leave control execution on the RX3i and place data storage, visualization or analytics on a separate supported edge platform.

  • Separates application lifecycle from PLC lifecycle
  • Creates a clear network and ownership boundary
  • Requires cabinet, power, environmental and maintenance planning
  • Useful when several assets feed one analytics layer

Move to an integrated edge controller

Evaluate CPE400 or CPL410 when the project requires high-performance control and a supported PACEdge environment in one device.

  • Official architecture combines real-time control and PACEdge
  • CPE400 pairs PACEdge with Movicon Connext; CPL410 with WebHMI
  • Requires application, I/O, network and acceptance review
  • Never describe it as a universal drop-in replacement
Reality check

What “no PLC rewrite” can—and cannot—mean

A careful project can avoid rewriting the control algorithm while still requiring configuration, testing and documentation work.

Work item May remain unchanged Still requires engineering
Control logic Existing sequences, interlocks and calculations may remain untouched when the chosen data path does not require logic changes. Prove that performance, scan behavior and communications remain within the approved baseline.
Tag access Existing variables may be reused if they already provide the required meaning and quality. Map names, units, ranges, access rights, sample rates and loss-of-data behavior.
Controller hardware The installed CPU may stay in service when it supports the approved interface and capacity. Verify exact model, suffix, firmware, available ports, protocol support and resource margin.
Network An existing plant network may carry the approved traffic when architecture and capacity permit. Define zones, routes, firewall rules, switch capacity, redundancy and monitoring.
Analytics Models and dashboards can evolve outside deterministic logic. Control versions, users, updates, backups, data retention and rollback.
Operator action Analytics may begin as advisory information only. Define who may act, how advice is validated and when a separate control change is required.
Brownfield failure patterns

Six ways an apparently simple data project becomes a control-system problem

Most failed pilots do not begin with a dramatic hardware fault. They begin with a boundary that was never written down.

Preventable ambiguity
The model is shortened

“RX3i CPU” is entered in the project notes without the full catalog number, suffix or hardware revision. The team then assumes a port, protocol, memory value or redundancy feature that belongs to a different controller. Keep the complete identity from the first survey through quotation and receiving.

Every available tag is requested

A broad extraction looks future-proof, but it increases traffic, engineering effort, storage and interpretation risk. Start from a decision and request only the signals, quality and timing that support it. Add more data under change control when a justified use case appears.

Network reachability is treated as approval

A successful ping or open port proves only that a path exists. It does not prove protocol suitability, authentication, authorization, acceptable traffic, redundancy behavior or ownership. Architecture and security approval must precede production connection.

Analytics becomes an unofficial control loop

A dashboard recommendation gradually becomes an operator instruction or automated write without a formal hazard, interlock and failure review. Define advisory status clearly. Any path that influences setpoints or commands belongs in a separately engineered control scope.

The pilot has no recovery owner

The integrator can demonstrate the application, but the site cannot restore users, certificates, containers, databases or dashboards after a failure. Handover must include backups, versions, access ownership, update rules, storage limits and a tested recovery path.

Performance is checked only after complaints

Without a baseline, teams cannot distinguish a new communications load from an existing network or controller problem. Record controller, task, communication and network behavior before connection, during peak collection and under loss or restart conditions.

Official capability boundary

CPE400/CPL410: powerful, but still model-specific

Official product data describes a 1.2 GHz quad-core platform with 64 MB PACSystems user memory, high-speed Ethernet, embedded PROFINET capability, secure boot and a Trusted Platform Module. The edge-controller architecture uses real-time hypervisor technology so deterministic PACSystems control and PACEdge software can run concurrently.

Those are platform capabilities. They do not prove that a legacy rack, fieldbus, project, HMI, safety function, third-party driver or plant network can be transferred without design work. Confirm the exact hardware and software combination, then build a migration and acceptance plan around the installed evidence.

Do not collapse these questions

Control compatibility: Can the application, I/O and communications be supported?

Data capability: Can the required tags be collected at the required rate and quality?

Cybersecurity: Are identities, certificates, updates, network paths and access rights governed?

Operations: Can the site diagnose, back up and recover both environments?

Commercial scope: Which licenses, software packages, accessories and support are required?

Interactive readiness gates

Click the evidence you already have

These checks do not calculate suitability; they reveal which questions remain unanswered before a quotation or design review.

Six evidence gates

Unchecked items are not automatic project failures. They are information gaps that should be closed before hardware selection or implementation.

From idea to controlled pilot

A six-workstream deployment plan

Use case and value

Choose one operational decision: energy deviation, cycle loss, condition trend, quality correlation or fleet comparison. Define the user, response and measurable outcome before choosing technology.

  • Avoid “collect every tag” as a project objective.
  • Separate the pilot success measure from future ambitions.
  • State what the analytics will not control.

Installed-base evidence

Capture complete controller, communication, I/O, power and rack identities. Export the approved project record and document firmware, software, protocol and network state.

  • Photograph labels and surrounding connections.
  • Record free ports and existing traffic.
  • Identify third-party interfaces and unsupported assumptions.

Data and security design

Create the smallest useful tag contract. Define read/write direction, identity, authentication, encryption, certificate ownership, network route, time source and loss behavior.

  • Apply least privilege.
  • Keep management and application access accountable.
  • Plan updates, backups and credential recovery.

Test and handover

Measure controller and network behavior before and during the pilot. Test communication loss, bad quality, restart, time drift, storage limits, user access and rollback.

  • Retain a baseline for comparison.
  • Document deviations and acceptance.
  • Assign long-term owners for control, network and applications.

Technical boundary: This guide explains planning decisions using manufacturer-published PACSystems and PACEdge capabilities. It does not replace the exact product manuals, license terms, secure-deployment guidance, project documentation, site cybersecurity procedures, safety analysis or qualified controls engineering for the installed system.

Nordwel compatibility desk

Send the installed evidence before choosing the edge path

For a useful RX3i review, include complete controller and communication-module labels, suffixes, firmware and software revisions, current network purpose, required data points, quantity, destination and the reason for the project.

Related RX3i references on Nordwel

Use exact-model pages as identification starting points

These internal pages support model and suffix review. They do not establish replacement or migration compatibility by themselves.

Nordwel links only
Decision questions

Short answers to the questions buyers ask first

Does edge analytics require a CPE400?

Not automatically. An existing controller may provide an approved data interface, or a separate edge platform may be used. CPE400/CPL410 should be evaluated when integrated control and PACEdge capability fits the wider architecture.

Can the control program stay unchanged?

It may, if the chosen data path uses existing variables and supported communications without requiring logic changes. Configuration, security and performance testing are still required.

Is OPC UA support enough to approve a connection?

No. Confirm that the exact controller and revision support the required function, then define access, certificates, tags, rates, network path and failure behavior.

Does secure boot make the whole system secure?

No. Secure boot and TPM are platform security capabilities. System security also depends on architecture, identities, configuration, updates, networks, procedures and recovery.

Can CPE400 replace any RX3i CPU?

No universal replacement claim should be made. Application conversion, I/O, rack, communications, firmware, licenses, performance and acceptance must be reviewed for the installed project.

What should be sent with an inquiry?

Send full model codes and suffixes, clear label photographs, firmware and software revisions, rack and network context, required data, quantity, destination and the project objective.

Should analytics write values back to the PLC?

Begin with an explicitly read-only boundary where possible. Any command, setpoint or automated write path requires a separate controls and safety review with clear ownership and testing.

How should a pilot be accepted?

Compare controller and network performance to the baseline, test loss and restart behavior, verify data quality and access controls, prove rollback and obtain named operational acceptance.

PDF
Nordwel site-hosted manufacturer file

PACSystems RX3i Product Datasheet

Use this official family document as an installed-base reference. It does not prove that a specific CPU, firmware, I/O configuration or application can be replaced without an exact-system review.

Open PDF
Leave A Message
If you are interested in our products and want to know more details,please leave a message here,we will reply you as soon as we can.
Submit

Leave A Message

Leave A Message
If you are interested in our products and want to know more details,please leave a message here,we will reply you as soon as we can.

Home

Products

contact

whatsapp