How-to
How to connect PROFINET IO to a CompactLogix through a gateway
Short answer
A CompactLogix cannot act as a PROFINET IO controller, so an EtherNet/IP-to-PROFINET gateway does the job: it is an EtherNet/IP adapter to the controller's scanner and the PROFINET IO controller to the field devices. The one reported gotcha on a 1769-L18ER-BB1B with a PLX82-EIP-PNC was byte swapping on both the input and output maps, fixed in the gateway rather than in ladder.
On this page
- Can a CompactLogix controller speak PROFINET IO on its own?
- Which gateway should sit between the two networks?
- Should the gateway be a scanner or an adapter on the EtherNet/IP side?
- How do I size the data map before opening either software?
- What are the configuration steps from ProSoft Configuration Builder to Studio 5000?
- Why do my values arrive byte-swapped, and where do I fix it?
- How fast does I/O cross the gateway?
- How do I prove the link before writing control logic?
- Questions asked
A CompactLogix controller cannot act as a PROFINET IO controller, so a gateway terminates both protocols: on one side it is an EtherNet/IP adapter serving the controller's scanner, on the other it is the PROFINET IO controller polling your field devices. The reference installation behind this procedure ran a 1769-L18ER-BB1B with PROFINET RT remote I/O through a ProSoft PLX82-EIP-PNC, and the one commissioning correction it needed was byte swapping on both the input and output maps. The steps below assign the roles first, build one data map that matches in three places, and prove the raw link before control logic exists.
Can a CompactLogix controller speak PROFINET IO on its own? #
No — a CompactLogix controller has no native PROFINET IO controller role, so a gateway terminates both protocols: it is an EtherNet/IP adapter to the controller's scanner and the PROFINET IO controller to the field devices, copying one data image between the two sides. There is no stack to enable and no tag or message instruction that turns the controller's Ethernet port into an IO controller, so reworking ladder logic will never bring those devices online. If the onboard I/O behaves while the PROFINET side sits stale, the fault is at the protocol boundary, not in the code. Controller-to-controller status exchange through a gateway proves routing only; fast cyclic I/O fails on conditions a slow status exchange tolerates. Bridging with a spare controller, as in using a PLC as a Modbus RTU to Profibus gateway, is a different design: there one processor terminates both protocols, while here the Logix controller owns one and the gateway owns the other.
Which gateway should sit between the two networks? #
Pick a stand-alone gateway that terminates both protocols in their I/O roles — an EtherNet/IP adapter on one side and a PROFINET IO controller on the other — because a unit that only passes general Ethernet messaging will never carry cyclic I/O. The unit documented here, the ProSoft PLX82-EIP-PNC, is a stand-alone DIN-rail device with two Ethernet ports used for communications, remote configuration, and diagnostics. The same project shortlisted the AnyBus AB7649 as the alternative; the procedure below is documented only on the ProSoft unit, so treat it as the reference configuration rather than a comparison verdict. Match the PROFINET class as well as the role: the field devices in the reference job were PROFINET RT, and the gateway carried that RT cyclic traffic successfully — a candidate that cannot run your devices' class as IO controller is disqualified at the selection stage. Put the devices on their own network and addressing plan; the gateway is a data boundary, not a merger you administer.
Should the gateway be a scanner or an adapter on the EtherNet/IP side? #
Adapter — the CompactLogix always initiates the EtherNet/IP connection, so the controller is the scanner and the gateway is the adapter, while on the PROFINET side the roles flip and the gateway runs as the IO controller with every field unit as an IO Device. Assign roles from connection ownership, never from the direction a signal travels, because inputs and outputs cross the same connections both ways:
- The EtherNet/IP scanner, or originator, opens and times the connection: the CompactLogix.
- The EtherNet/IP adapter, or target, serves the data image: the gateway.
- The PROFINET IO controller parameterizes the devices and exchanges cyclic data: the gateway again.
- The PROFINET IO Devices expose their input and output modules: your field units.
This is the same split you already run when you start and stop a PowerFlex 700 over EtherNet/IP from CompactLogix — controller as scanner, drive as adapter — with the gateway standing in as the adapter and holding the second protocol behind it.
How do I size the data map before opening either software? #
Total the mapped bytes in each direction and make each total appear identically in three places — the PROFINET device layout, the gateway maps, and the Logix I/O connection — because a mismatch of one byte rejects or truncates the connection. Write the map down before touching either editor: for every item record device, direction, byte offset, and length. Two devices make the arithmetic visible. A remote I/O block contributing 16 input bytes and 8 output bytes, plus a valve terminal contributing 8 input bytes and 4 output bytes, map like this:
| Device | Direction | Length | Offset range |
|---|---|---|---|
| Remote I/O block | Input | 16 bytes | 0–15 |
| Valve terminal | Input | 8 bytes | 16–23 |
| Remote I/O block | Output | 8 bytes | 0–7 |
| Valve terminal | Output | 4 bytes | 8–11 |
Input image: 16 + 8 = 24 bytes. Output image: 8 + 4 = 12 bytes. Those two totals are what the Logix I/O tree entry must carry; if an editor there counts 32-bit words instead of bytes, convert on paper first: 24 ÷ 4 = 6 words in, 12 ÷ 4 = 3 words out. On the gateway side the ranges are roomy — the start byte offset of a mapped block is accepted up to 19998 per the ProSoft Configuration Builder release notes — so a 24-byte image sits nowhere near a limit. Size the connection from the mapped byte ranges, including reserved positions the device layout forces on you, never from the number of tags someone plans to display.
What are the configuration steps from ProSoft Configuration Builder to Studio 5000? #
Start in ProSoft Configuration Builder, build the PROFINET side first, then add the Logix I/O tree entry with matching lengths; ten steps in this order produce a working raw exchange before application logic touches it.
- Inventory the PROFINET devices: for each unit record identity, module order, input byte count, output byte count, and the device name or address its GSD engineering data requires.
- Create the gateway configuration in ProSoft Configuration Builder and set the addresses for both networks.
- Add the PROFINET IO Devices and their modules in physical order, with the gateway acting as IO controller.
- Allocate each device a unique input range and a unique output range in the gateway map; reject overlaps and undocumented gaps, and keep the signed map beside the controller tag layout.
- Add the gateway to the Logix I/O tree as a generic EtherNet/IP node and enter the input and output sizes that match the map totals — 24 and 12 bytes in the example above.
- Map Logix tags onto that image, keeping raw gateway data separate from application commands and status so the byte-order correction stays visible during commissioning.
- Set byte order from the test in the next section, not from assumption.
- Download the gateway, bring the PROFINET devices to data exchange, then download the controller and open the EtherNet/IP connection.
- Prove the raw exchange with the four-point trace below before any logic consumes it.
- Only then wrap the interface in an AOI if you want one; the reference installation added one to simplify connecting and controlling multiple devices.
Why do my values arrive byte-swapped, and where do I fix it? #
Swap the bytes inside each 16-bit word in the gateway map — on both the input and the output side — because the reported working installation on a 1769-L18ER-BB1B needed exactly that correction in both directions, and a one-sided swap leaves half the interface wrong. A gateway can hold both connections healthy while delivering unusable numbers: a correct byte count says nothing about byte order. Prove it with one value whose bytes are easy to tell apart. Drive a mapped 16-bit input with 4660, which is 0x1234: 1×4096 + 2×256 + 3×16 + 4 = 4660. If the Logix tag reads 13330, it holds 0x3412: 3×4096 + 4×256 + 1×16 + 2 = 13330 — the two bytes inside the word have exchanged places. Enable the gateway's byte-swap option on that map, which is precisely the correction the reference installation applied in the gateway instead of writing conversion code, then repeat the test in the output direction with a value the device can display. Apply the correction in exactly one place, the gateway, and never a second time inside an AOI, or you build a double conversion whose result depends on execution order.
How fast does I/O cross the gateway? #
Budget the sum, not one number: worst-case one-way latency is the PROFINET cycle plus the EtherNet/IP requested packet interval (RPI) plus one scan at each end, so with an illustrative 2 ms cycle, 10 ms RPI, and 5 ms scans you plan on the order of 22 ms. Those numbers are illustrative, chosen to show the arithmetic: 2 + 10 + 5 + 5 = 22 ms one way, before the gateway's internal copy time, which the published material does not number — measure it. Toggle one controller output bit onto a spare device input and timestamp both ends; half the round trip bounds the one-way figure.
The decision falls out of the arithmetic: if the machine needs response better than about 20 ms, tighten the PROFINET cycle and the RPI before you build, because no ladder pattern recovers time lost at the boundary; if 50 ms-class response is acceptable, settings in that range leave margin. The cyclic exchange is the same mechanism you use to control a PROFINET servo drive with cyclic telegram data: the telegram layout sets the bytes, the configured cycle sets the cadence.
How do I prove the link before writing control logic? #
Toggle one known value at each end and watch it cross four points — the producing device, the gateway source map, the gateway destination map, and the Logix tag — before any AOI or control code exists, so every fault you find is a network or mapping fault rather than logic. Commission against raw tags only, one item at a time. When something misbehaves, the symptom names the layer:
- Nothing moves in either direction: wrong role assignment, devices short of data exchange, or the EtherNet/IP connection never opened.
- The connection is rejected although the gateway answers pings: an input or output length disagrees with the gateway image in one of the three places.
- Only part of the image changes: truncated length, overlapping ranges, or module order that no longer matches the device.
- Values land in wrong positions: byte order, as above.
- One direction works alone: that direction is unmapped, swapped wrongly, or refused by the device.
Keep actuators disconnected or unpowered until the raw exchange reads correctly in both directions, because a byte-swapped output word is still a plausible number to whatever it commands.
Questions asked
Can I copy the finished configuration to another line?
Yes, with one check. The gateway file and its maps are controller-independent, so they carry to the next line; what changes is the controller side. A reported copy to a third line failed because that processor's add-on Ethernet interface passed messaging but would not carry I/O, so confirm every interface in the path supports cyclic I/O before you promise a schedule.
Do I need an AOI to control the PROFINET devices?
No. The raw exchange runs on plain tags mapped to the gateway image. The reference installation added an AOI afterwards to simplify connecting and controlling multiple devices; if you write one, keep raw gateway data, converted data, and application commands as separate tag groups so the byte-order correction stays visible during commissioning.
Will every installation need byte swapping?
Not by nature. It depends on how each side represents multibyte values, so test rather than assume. It is a property of the pairing: once proven for a device type and corrected in the gateway, the correction holds for identical devices. The reference CompactLogix installation needed it on both the input and output maps.
Does the byte swap change where discrete bits land?
It can. The correction rearranges bytes, so packed Boolean data sharing a word with numeric data can shift position once the swap is enabled. Registers come out clean; mixed images do not. After enabling it, re-read every mapped item in both directions before releasing the interface to application code.