Troubleshooting
Troubleshooting a PLC-5 ControlNet connection path from a driver
Short answer
The route for a driver to a PLC-5 over ControlNet ends at the processor, not at the network: 1,2,2,4 becomes 1,2,2,4,1,0 once the processor slot is confirmed. Test the terminal route and the short route separately with the build recorded, and treat a working RSLinx read as proof of reachability, not proof the driver accepts the same string.
On this page
- Why does my PLC-5 driver fail on ControlNet when RSLinx reads fine?
- What do the numbers in the connection path mean?
- Does the driver need 1,0 added to the end of the ControlNet path?
- Is the CNBR slot the same as its ControlNet node number?
- Why does the status cycle between connecting and disconnected?
- Does the 7.1.7 path fix cover my driver build?
- Can I copy the working DH+ route to ControlNet?
- How do I test both ControlNet routes cleanly?
- When do I stop testing routes and send the case to the vendor?
- Questions asked
A driver that reads a PLC-5 through RSLinx but cycles between connecting and disconnected on its own route has a client-side path problem, not a dead network. The fix starts in the string: a route for a driver ends at the processor, so 1,2,2,4 becomes 1,2,2,4,1,0 once the processor slot is confirmed. Test that terminal route and the short route separately, record the driver version and build before changing anything, and treat a working RSLinx read as proof of reachability only. The case behind this page: a 1756-ENBT/A in slot 9, a 1756-CNBR in slot 2 at ControlNet node 17, a PLC-5 at ControlNet node 4, and a third-party OPC UA driver that would not hold the connection.
Why does my PLC-5 driver fail on ControlNet when RSLinx reads fine? #
A working RSLinx topic proves the PLC-5 is reachable through that client's route; it proves nothing about whether your driver accepts the same string, so test the short route 1,2,2,4 and the terminal route 1,2,2,4,1,0 separately, one variable at a time.
In the reported case the driver pointed at the 1756-ENBT/A IP address in slot 9, crossed the 1756-CNBR in slot 2 — that module held ControlNet node 17 — and targeted the PLC-5 at ControlNet node 4 using 1,2,2,4. An RSLinx DDE/OPC topic built on the same ControlNet path read the processor without trouble. That split is diagnostic: the bridge, the cable and the node are alive, because another client just read through them. What stays unproven is the driver's handling of the string itself, so the checks below move through the hop meanings, the terminal entry, the slot-versus-node distinction, the build, and then the test procedure.
What do the numbers in the connection path mean? #
Each entry moves the request one hop: 1 steps onto the local backplane, 2 steps off at the bridge in slot 2, the second 2 crosses the bridge's ControlNet port, and 4 is the remote node, so 1,2,2,4 is four hops that stop at the network.
The driver documentation for these bridges defines a routing path as a series of backplane hops whose last hop points to the destination backplane — the sentence that predicts this failure, because 1,2,2,4 ends at a node, not at a backplane. The same reading covers the DH+ route in the other half of the case: 1 is the backplane, 8 is the 1756-DHRIO/B in slot 8, 2 is its DH+ channel, 10 is the PLC-5's DH+ node. The hop logic is what you enter when you Configure ControlLogix MSG reads and writes to a PLC-5; a slot-versus-node mix-up fails a message the same way it fails a driver.
Does the driver need 1,0 added to the end of the ControlNet path? #
Yes for this driver: the route must continue past the network to the processor, so the full string is 1,2,2,4,1,0 — backplane, bridge slot, ControlNet port, node 4, backplane again, processor in slot 0 — and the last number must be the processor's real slot.
Build the terminal route hop by hop from the reported rack: 1 (enter the backplane behind the ENBT in slot 9), 2 (exit at the CNBR in slot 2), 2 (cross its ControlNet port), 4 (arrive at the PLC-5's node 4), 1 (enter the PLC-5's own chassis backplane), 0 (address the processor in slot 0). Concatenate and you get 1,2,2,4,1,0 — six entries for six hops. Drop the last pair and you have four entries that stop at the network; that four-versus-six count is the whole difference between the two test routes. The vendor's guidance for this driver was to tack the terminal pair on, and in the reported case the corrected route still cycled, which is why the build check below belongs to the same test, not to an afterthought.
Is the CNBR slot the same as its ControlNet node number? #
No — slot 2 is the CNBR's chassis position in the local rack, node 17 is its address on the ControlNet cable, and the path uses the slot; the remote 4 is the PLC-5's node, which never tells you the slot of its processor.
The string never contains node 17; the module's network address plays no part in routing. The number that appears twice is 2: once as the slot you step off at, once as the network port you cross. For the terminal pair, read the last number off the remote chassis. A PLC-5C-series processor carries the ControlNet connection on board, which is why the final backplane-and-slot pair lands directly on the processor — and also why the slot value must come from the chassis and not from habit. The assumption that an on-board port means slot 0 is the kind of entry that turns a correct six-entry route into a dead end; verify the slot once and the terminal route is determined.
Why does the status cycle between connecting and disconnected? #
The alternating status means the driver accepted a path string but never established a stable session to the processor; it tells you the connection is not sustained and it does not identify which hop failed, so treat it as a result to log, not a diagnosis.
In the reported case the status alternated on 1,2,2,4 and again on 1,2,2,4,1,0, so neither route form was cleared or condemned by the cycling alone. Read the status per test with these outcomes:
- Stable connection, reads return: the tested route sustains the session; read several tags across the data table, then keep the route.
- Cycling between connecting and disconnected: no stable session; verify the terminal pair and the processor slot, then confirm the build before another attempt.
- RSLinx reads while the driver cycles: reachability is proven and compatibility is not; compare your string against the RSLinx topic entry by entry instead of re-testing RSLinx.
Does the 7.1.7 path fix cover my driver build? #
Not by itself: a ControlNet connection-path fix was reported for release 7.1.7, and the failing installation still cycled on 7.1.8 (b6247) and on the development build 7.1.9-beta2 (b6361), so a version note is not evidence that your exact build is fixed.
Record what is actually running: the version and the build identifier the installation reports — 7.1.8 reporting b6247 is the level of detail a vendor case needs, not simply the latest. Then hold the discipline from this case: do not combine a route edit with a version change in the same test. The reported installation changed neither between its 1,2,2,4 and 1,2,2,4,1,0 attempts, which is why those two results can be compared at all; preserve that in your own log.
Can I copy the working DH+ route to ControlNet? #
The structure transfers — backplane, bridge slot, channel, remote node — but the DH+ result does not: 1,8,2,10 worked both with and without 1,0 appended in the tested rack, so DH+ behavior decides nothing about the ControlNet string.
The DH+ side of the case ran a 1756-ENET/B with a 1756-DHRIO/B in slot 8 at DH+ node 3, bridging to the PLC-5 at DH+ node 10, and the processor answered on both 1,8,2,10 and 1,8,2,10,1,0. That tells you the driver completes a session either way when the destination network is DH+, and it isolates the ControlNet path as the only difference that matters here. Treat the DH+ rack as the control group for the case, not as a template for the route; the bridging mechanics behind it are the ones laid out in How to route a ControlLogix MSG to a PLC-5 over DH+.
How do I test both ControlNet routes cleanly? #
Run one route per test with nothing else changed: save 1,2,2,4,1,0 with the slot confirmed, restart the driver or connection, watch the status until it settles, and only then move to the next string — never a route edit and a version change in the same test.
Changing a driver route touches only your data-collection connection, and the one safety rule on this network is that reseating or power-cycling a 1756-CNBR to refresh a route is a bridge replacement on a live ControlNet, not a troubleshooting step. Run the sequence:
- Write down the physical truth before touching software: Ethernet module and slot, bridge module and slot, bridge node, PLC-5 node, and the processor's slot read off the remote chassis.
- Record the driver version and build as it runs, for example 7.1.8 (b6247).
- Restart the connection, enter the terminal route with the slot appended — 1,2,2,4,1,0 for slot 0 — and watch the status to a settled state, logging whether reads return.
- If step 3 fails, restart again, enter the short route 1,2,2,4, and watch it the same way.
- Log each test as one line: route string, build, status result, reads returned or not.
- Set the working RSLinx topic beside your string and compare entry by entry before concluding the driver is at fault.
Between steps 3 and 4 the only variable that moves is the route string, and the build stays fixed throughout — that is what makes the two results comparable and the final decision clean.
When do I stop testing routes and send the case to the vendor? #
Stop when both 1,2,2,4 and 1,2,2,4,1,0 fail on a confirmed processor slot with a recorded build while RSLinx still reads the same node, and hand over the route strings, the build, the status behavior, and the contrasting RSLinx result.
While the case is open, leave the RSLinx topic in service for any read that matters and stop toggling the same path, because repeated identical attempts return identical results and bury the evidence you already have. The decision comes from the last settled test. Stable on 1,2,2,4,1,0 with the slot verified: keep that route, re-verify reads after a driver restart, and document the slot and the hop order so a later edit cannot drop the terminal pair. Stable on the short route only: keep it, note that your build sustains the four-entry form, and say so in the vendor case. Cycling on both after the slot and build checks: write the case, attach the log lines, and end the route experiments.
Questions asked
How many entries should a routed path to a PLC-5 have?
Count the hops you intend and enter exactly that many. A path that crosses one bridge, one network and finishes at a processor slot carries six entries: backplane, bridge slot, network port, remote node, remote backplane, processor slot. A path that ends at a network address carries four, like the DH+ route 1,8,2,10 in the working half of this case. A count that disagrees with your design points straight at the missing or extra entry.
Does the ControlNet port sit in the PLC-5 processor itself?
On PLC-5C-series processors the ControlNet connection is on the processor, which is why the route's final backplane-and-slot pair lands directly on it. That placement does not fix the processor to any slot: read the slot from the remote chassis and write it as the last entry. Node 4 identifies the processor on the network; it never identifies the slot, and the two numbers are not interchangeable.
Which ControlNet bridge modules use this routing form?
The Rockwell documentation for ControlNet network configuration covers the 1756-CNB, 1756-CNBR, 1756-CN2 and 1756-CN2R as bridge modules in ControlLogix chassis, and the backplane, slot, port, node structure applies across them. Port and channel details differ by module, so confirm the network-port entry against the module actually installed before reusing a route written for a different bridge.
Can changing the driver path disturb scheduled ControlNet I/O?
No. The driver opens an unscheduled messaging connection, and editing its path changes only that connection; scheduled I/O transfers between the bridge and its nodes keep their established schedule. That is why a route test needs no network outage. The action that does disturb a live network is reseating or power-cycling the bridge module, which is a replacement procedure, not a test.