What is interbus?

Conceptual illustration of an INTERBUS controller and field devices

What Is INTERBUS? Data Mapping, Wiring and Maintenance

INTERBUS is a fieldbus for exchanging control data between a central controller and distributed equipment. Remote I/O, sensor interfaces and suitably equipped drives participate in a logical data ring. The controller transfers a structured process image through that ring rather than treating it as a general-purpose computer network. It is neither a virtual machine nor another name for an Ethernet connection. [1]

For someone maintaining a machine, the useful questions come next: Which device manages the network? What does each data word mean? What happens when communication stops? Those questions connect fieldbus fundamentals to the actual PLC program. A compatible connector or a successful device scan is not enough to answer them.

About the illustration: The header image is conceptual, not an installation drawing or a product specification. Its external return loop must not be read as an instruction to install an extra cable from the last device to the controller. In the conventional copper arrangement, the outgoing and returning data paths share the same cable. [1]

Start with the installed hardware and the process image

Record the controller, network module, couplers, terminal order and drive option cards before changing anything. A manufacturer family name is not a network configuration. For example, the TSXIBY100 is an INTERBUS module for Modicon Premium, but that does not establish that every Premium installation has this interface. Keep the engineering project and the information needed to open it alongside the hardware record. PLC programming cables provide access to the engineering system; they do not substitute for the fieldbus connection. [10]

Next, separate network identification from application meaning. Beckhoff's BK4000 description explains automatic ID and length information and assignment to the controller's process image. A device can nevertheless have the expected data length while the program interprets its words incorrectly. Signedness, byte order, status bits, scaling and input/output direction all need an explicit definition. [9]

Worked example: build a data inventory before assigning addresses. Assume four functional groups, with no additional status bytes, padding or PCP allocation. This is an illustrative PLC memory layout, not the fixed layout of a particular coupler or a ready-to-load device configuration.

Example host memory requirements; input and output address spaces are independent
Functional groupPLC inputsPLC outputs
16 digital inputs2 bytes: I[0–1]None
16 digital outputsNone2 bytes: Q[0–1]
Four 16-bit analog values8 bytes: I[2–9]None
Illustrative drive with 3 input and 3 output words6 bytes: I[10–15]6 bytes: Q[2–7]

The application needs 16 input bytes and 8 output bytes: 24 bytes of host storage in total. This total is not the INTERBUS frame length. The two memory directions cannot simply be added to determine the length of a circulating telegram. Actual module layouts, identification lengths and controller configuration remain authoritative.

When integrating sensors, establish where conversion takes place. An analog sensor may feed a local terminal; the coupler, not the sensor, then participates in INTERBUS. Likewise, a compatible variable-frequency drive needs its documented communications interface. Calling every connected transducer an “INTERBUS sensor” hides a useful diagnostic boundary. [2] [4]

For a deliberately simple scaling example, suppose a linear position transducer represents 0–100 mm as 0–10 V. An ideal reading of 3.5 V corresponds to 35 mm. The ADC's numerical representation must still be taken from the actual terminal manual; no universal INTERBUS raw-count range is implied. The 0–10 V and 4–20 mA signal guide covers the signal layer, while measurement accuracy and calibration address a different question from communication success.

Carry the engineering unit and validity information into the HMI or SCADA tag, not just the numerical value. A screen showing 35.000 mm does not prove micrometre accuracy. Before using a received measurement in a PID control loop, verify the sign, range and behaviour when the measurement is unavailable. These are application checks, not additional functions supplied automatically by the bus.

Cyclic process data and PCP are different services

A drive may need a small, regularly exchanged process image and a separate way to access parameters. In INTERBUS terminology, PD denotes process data; PCP is the Peripherals Communication Protocol. The distinction matters when allocating a device's communication area. A parameter transaction should not be mistaken for either a cyclic setpoint update or confirmation that the motor has moved. [4]

A concrete example is the SEW-EURODRIVE DFI11B option for MOVIDRIVE MDX61B. Its manual does not extend that support to MDX60B. Within a six-word allocation, the document identifies 4 PD + 2 PCP with ID E0h, and 6 PD without PCP with ID 03h. A proposed 6 PD + 2 PCP allocation exceeds that budget. These are documented configurations for this option, not universal limits for all INTERBUS drives. [4]

Use the MOVIDRIVE B family guide to identify the family and the SEW equipment and documentation tree to navigate related equipment. The MDX61B0005-5A3-4-00 product entry identifies a drive model; it does not prove that a particular unit is supplied with DFI11B. Inspect the installed option and its revision. The wider SEW-EURODRIVE archive should not be treated as a single compatibility list.

For servo applications, retain the motor, feedback and safety configuration as well as the fieldbus settings. During commissioning, verify status and harmless read-only information before enabling motion. This guide intentionally provides no generic drive start command: the correct command source, enabling conditions and safe test arrangement depend on the machine.

Why INTERBUS is described as a logical ring

The conventional system can be understood as a shift register distributed across the participating devices. A central master moves data through that structure, with process data occupying defined positions. Identification and normal data exchange are separate activities. This is the basis for understanding the aggregate transfer commonly described as the summation-frame approach. [1]

The master/slave relationship describes responsibility, but it does not make every master-controlled protocol work in the same way. Reading INTERBUS as a sequence of independent Modbus RTU polls misses the importance of the configured data structure. When replacing equipment, check the device identity and length as well as the meaning of each assigned process-data area.

A logical ring is not automatically a redundant network. The installed cable route may look like a line, and systems with suitable branching equipment can have a more elaborate physical layout. Both directions of the conventional remote-bus data path are carried within the cable. An ordinary Ethernet switch does not create an INTERBUS branch, and a ring-shaped drawing does not establish what will remain operational after a cable fault. [1] [3]

Also distinguish the remote bus connecting participants from the local terminal system behind a coupler. In the Beckhoff examples, the latter is the K-bus. One remote-bus coupler with many local terminals is not the same thing as many separately connected remote-bus devices. This distinction affects both capacity planning and the location of a fault. [1] [2]

Read hardware limits as separate engineering constraints

There is no useful single “maximum size” unless the interface, physical medium and local I/O arrangement are specified. The following figures belong to the Beckhoff BK4020. They illustrate how to read a product specification, not a guarantee covering every INTERBUS installation. [2]

BK4020 example: do not combine unlike limits
SpecificationPublished value and interpretation
Transfer rate500 kbaud. This is not a PLC scan time.
Process-data capacity64 input bytes and 64 output bytes. Check each direction separately.
Local terminals64, or 255 with suitable K-bus extension. These are not 255 separate remote-bus addresses.
Copper distances400 m between participants; the product description gives 13 km for the total ring. The latter is not the length of one cable section.
K-bus supply1750 mA for local electronics. Field-load current is a different budget.

A configuration can remain below the terminal-count limit and still exceed its process-data area or available supply current. Treat those as independent checks. Similarly, a faster replacement interface does not make an existing mixed network faster unless the controller, every relevant interface and the installed connection components support the selected operating conditions.

Illustrative power check: Assume twelve terminals draw 60 mA each and four draw 100 mA each. Their combined electronics demand is 12 × 60 + 4 × 100 = 1120 mA. Against 1750 mA, the arithmetic difference is 630 mA. This is not spare current available for solenoid coils. Use real terminal figures, consider starting conditions and additions, and calculate the power supply and field loads separately.

Cabling: connector appearance is not a wiring specification

On the conventional copper remote bus, differential channels carry outgoing and returning data. The BK4000 manual describes its interface in RS-422 terms; the DFI11B manual describes two RS-485 channels. Preserve those product-specific descriptions rather than assuming every document uses the same electrical label. Neither description means that a generic USB serial adapter implements INTERBUS. The RS-232, RS-422 and RS-485 comparison explains why an electrical interface and a complete protocol are different layers. [1] [4]

Use the drawings for both ends of the actual connection. Check incoming versus outgoing ports, pair assignment, shielding and any connector bridge used to signal continuation. BK4000 documentation places the terminating resistors inside the devices; the continuation bridge has a separate purpose. Do not import an external-termination arrangement from another bus merely because both use differential signalling. A DB9 communications port does not establish RS-232 pin compatibility. [1]

Include cable routing, shield contact, mechanical strain and bonding in the inspection plan. Disconnecting a cable, removing a terminal or altering switches must follow the equipment's stated power-off requirements. An electrical test procedure should distinguish control-electronics checks from work involving hazardous voltages and unexpected movement. Record the existing arrangement before disturbing it. [1] [4]

A worked timing calculation, with its limits stated

A published bit rate does not tell you how soon a sensor change will affect an actuator. Beckhoff's BC4000 manual, section 7.4, supplies a useful model for its stated INTERBUS conditions. It expresses bus reaction time as twice the calculated cycle time. Use the model's units and variables, not an assumed universal millisecond figure. [3]

Manufacturer model:
Tcycle = SW + [13 × (6 + N) + 1.5 × M] × Tbit
TINTERBUS = 2 × Tcycle

Here M is the number of bus couplers and N is the effective data length in bytes under that model. The documented assumptions include SW = 0.2 ms and Tbit = 0.002 ms. For an illustrative M = 4 and N = 40:

Tcycle = 0.2 + [13 × 46 + 1.5 × 4] × 0.002
= 1.408 ms
TINTERBUS = 2 × 1.408 = 2.816 ms.

The 40-byte assumption is independent of the earlier 24-byte host-memory example. Word counts must be converted where the formula requires bytes. Do not silently reuse these constants for different hardware or a different transmission rate. Sensor filtering, local terminal communication, PLC task scheduling, drive response and mechanical motion are outside this result. Error recovery also needs separate consideration.

For acceptance testing, define the observation points: sensor input, controller image, output command and actual response. A data logger timestamp can refer to receipt or recording rather than the original measurement event. Record the slowest relevant update under the required operating load, not just an attractive average. The calculation and the test answer related but different questions.

Troubleshoot the failing function, not just the bus name

First preserve evidence. Capture the controller diagnostic record, observed LEDs, last visible participant and the conditions immediately before the fault. “The bus is down” can describe a missing supply, an interrupted remote-bus section, a configuration mismatch or an application that is ignoring valid data. Resetting everything removes useful distinctions.

Connection detected but no process exchange: Compare the loaded configuration with the identified devices and data lengths. In BC4000 diagnostics, an established incoming connection indicated by RC is not by itself evidence of active transfer; BA is part of the interpretation. RD can indicate a disabled continuation section, including a cable fault or an action by the master. Use the combination documented for that device, not a colour legend borrowed from another manufacturer. [3]

Several downstream devices disappear together: Investigate the boundary between the last reachable device and the first missing one. Check their supplies, the outgoing/incoming connection and the continuation arrangement. Many missing devices do not prove many simultaneous component failures. Plan controlled isolation tests rather than repeatedly inserting equipment into a live machine network.

The network is active but a value is wrong: Inspect raw input data, offsets, direction, data type and scaling before blaming the sensor. A constant value may be a valid constant measurement; conversely, a display retaining its last value does not establish freshness. Use the available validity indication, data-age information or a suitably designed application counter instead of relying only on visible value changes.

The coupler runs but an output does not: Separate electronics power from power at the field contacts, then follow the command through the output channel to the load. BC4000 documentation explicitly distinguishes those two supply indications. On a drive, also check the command source, enabling conditions and drive faults. Do not bridge an interlock merely to demonstrate that communications are working. [3]

Only parameter access fails: Check PCP support, allocation, the requested parameter and the returned error information. Do not merge a rejected parameter service and a complete loss of cyclic exchange into one undifferentiated alarm. For intermittent problems, correlate the timing with switching loads, mechanical movement and supply disturbances. A successful restart is evidence of recovery, not proof of the root cause.

Keeping INTERBUS while changing the upper-level network

Read the role on each gateway port before selecting a migration architecture. HMS specifies the AB7836-F as an EtherNet/IP adapter and an INTERBUS slave. The engineering implication is that this gateway does not, by that role assignment alone, replace the INTERBUS master. The function that manages the INTERBUS side must still exist. [5]

The same product distinguishes 20 bytes of process data per direction from 512 bytes with PCP. The larger figure must not be presented as 512 bytes of directly exchanged cyclic I/O. Our illustrative 16-byte input / 8-byte output inventory is numerically below 20 bytes per direction, but this comparison proves only a preliminary size check—not a valid mapping, update time or operating sequence. [5]

For a replacement based on PROFINET or EtherCAT, preserve a functional specification rather than just an address list. Compare signal meanings, validity flags, parameter access, restart conditions and loss-of-communication behaviour. During a phased changeover, define which controller is allowed to issue the final command. Two working network connections are not a control-authority policy.

Product lifecycle decisions are also specific to the item. When checked on 18 September 2026, Beckhoff's BK4000 page did not recommend it for new projects and identified BK4020 as an alternative. That is not a statement that every INTERBUS product has reached the same lifecycle stage, nor proof of an unexamined drop-in replacement. Recheck the configured data image and electrical requirements. [9]

An IIoT or reporting project may only require a controlled read-only route to selected data. Adding an Ethernet port does not turn the original fieldbus into Ethernet. The same planning discipline is useful for other installed networks such as Fipio: identify the retained controller role, the data contract and the required fault response before choosing the new connection.

Communication health, machine safety and cybersecurity

Define what each output must do when data is lost and how the machine may restart after recovery. Holding the last command can be acceptable for one function and hazardous for another. A normal process-data watchdog is not, by itself, a SIL claim. Verify the required response within the machinery safety design. Where Safe Torque Off or another safety function is used, its own device documentation and validation requirements remain applicable.

Transmission error detection is not encryption or authentication. INTERBUS connectivity alone does not establish that control data is confidential or that a connected engineering system is authorised. Treat the engineering computer, co