IMDHMIControls LibraryShop main store

Troubleshooting

Troubleshooting intermittent PROFIBUS dropouts on an S7-400 network

Industrial networks Published 7 min read

Short answer

Save and read the CPU diagnostic buffer before touching anything: the earliest OB86/OB85 entries, mapped against the physical segment drawing, tell you whether the losses share one repeater segment. On a bus that ran seven years and began failing after an encoder tie-in was relocated and re-terminated, one physical-layer fault at that tie-in or its segment's end termination explains all three DP indicators and the self-recovery.

On this page
  1. Why one fault lights all three indicators and raises both OBs
  2. Save the diagnostic buffer before disturbing anything
  3. A worked example: one event cluster on one segment
  4. Map the losses onto the physical segment drawing
  5. Inspect the relocated tie-in and the two terminations
  6. Repeater supplies and interference paths
  7. Confirm the installed network against the STEP 7 project
  8. The decision for this network
  9. Questions asked

All three indicators on the DP module lighting red together and clearing on their own — intF, extF and busF, with OB86 and OB85 entries and no restart — is the signature of one physical-layer fault that broke the bus cycle for a couple of seconds and then healed. Save the CPU diagnostic buffer before anyone touches a connector, then map the earliest entries against the physical segment drawing. On this network, seven years stable with roughly 80 nodes, encoders, drives, remote racks, an OP27 and repeaters, the dropouts began after an encoder node address failure was repaired by replacing the encoder and its termination and relocating the tie-in point. The first suspects are the termination states at that segment's two ends and the relocated connection itself — not the CPU, not the DP module.

Why one fault lights all three indicators and raises both OBs #

PROFIBUS DP is cyclic: the master exchanges input and output data with every configured slave on each bus cycle. An electrical interruption on a segment breaks that exchange for every station on it, so a single cable event can produce an OB86 incoming event for each failed station, an OB85 when the operating system cannot update the process image for their addresses, and matching outgoing events seconds later when the bus recovers. OB86 is the rack or distributed-station failure block on the S7-300 and S7-400; OB85 is the program-processing error block. The indicators clearing by themselves means the cyclic exchange resumed — recovery identifies nothing, it only proves the interruption ended.

The CPU staying in RUN through these events also tells you OB86 exists in the program, because without it the first station loss would take the CPU to STOP. The same diagnosis carried out step by step on ET 200 slaves is walked through in S7-300 PROFIBUS ET200 slave diagnostics with OB86 and OB85.

Save the diagnostic buffer before disturbing anything #

The buffer holds a finite number of entries, and a busy fault period fills it with recovery messages that can push the initiating event out of view. Read it in STEP 7 first and save the complete details of every entry:

  • incoming and outgoing timestamps, to the millisecond
  • station address and interface identification for each event
  • CPU operating-mode changes
  • what the machine was doing at those times, from the operator or process log

Order the entries chronologically and find the earliest communication event. That entry, not the pile of recoveries behind it, is the fault. Map every address in it to a physical device and a segment on the cable drawing, and note whether the losses cluster.

A worked example: one event cluster on one segment #

The timestamps below are illustrative; the method and the arithmetic are what transfer to your buffer. Suppose the buffer shows:

  • OB86 incoming, station 44 — 14:32:07.213
  • OB86 incoming, station 45 — 14:32:07.229
  • OB86 incoming, station 46 — 14:32:07.250
  • OB85 — 14:32:07.262
  • OB86 outgoing, station 44 — 14:32:09.855

The arithmetic: 07.229 − 07.213 = 0.016 s, so stations 44 and 45 failed 16 ms apart. 07.250 − 07.213 = 0.037 s, so all three were gone within 37 ms. 09.855 − 07.213 = 2.642 s, the full outage. Three stations failing inside 37 ms and recovering together is one electrical event, not three failing devices. If the drawing puts those three addresses on the segment behind one repeater, that segment is the inspection zone — and if the relocated encoder is one of those addresses, the tie-in sits inside it.

Map the losses onto the physical segment drawing #

A segment is an electrically continuous cable run between two terminated ends; a repeater regenerates the signal between segments, and every station behind it shares that repeater as a failure point. The RS-485 physical layer carries at most 32 stations per segment, and 80 ÷ 32 = 2.5, so at least 3 segments exist on this bus. The node total describes the installation, never the fault zone.

Draw the cable as installed, not the project's station list: every repeater, both ends of each segment, the drives, the remote racks, the OP27, and the relocated encoder tie-in. Mark what the encoder repair changed — the old route, the new route, any abandoned length, and every termination switch position. A tie-in that keeps the same station address still changes the electrical topology, which is why the fault history matters more here than the seven quiet years before it.

What the buffer shows Where the fault lives Deciding check
Several stations behind one repeater drop together That segment's cable, one end's termination, the repeater port or its supply Confirm every affected address sits behind the same repeater on the drawing
Only the encoder's station drops Encoder connector, local cable, supply or address Compare the encoder's diagnostics with the adjacent station's status
Stations on several segments drop together Common path: master interface, first repeater, shared 24 V supply Find the first connection or supply all dropped stations have in common
Losses track drive run/stop or machine motion Coupled interference or a mechanically disturbed connection Correlate buffer timestamps against drive and motion events

Inspect the relocated tie-in and the two terminations #

Both physical ends of a segment carry the terminating network — three resistors combined with a capacitor, built into the 9-pin Sub-D connectors — and every station between them must have termination off. Moving the tie-in could have turned the old endpoint into an intermediate station and some other device into an end; both connector switches must reflect the new geometry. A glance at the switch is not a check, because the termination switch also determines which cable entry stays connected: at an end device the single cable goes on the entry the documentation marks for it, and the switch opens the other entry.

The terminating network is fed from the end station's own supply, so an endpoint that loses power drops the segment's termination at the same moment, and the segment runs reflection-prone until power returns.

Inspect connectors and termination switches only during an approved outage; with the line running, measure supplies instead of unplugging anything. Then work this list on the affected segment:

  • termination on at the two physical ends, off at every intermediate station including the tie-in
  • the cable on the connector entry that the termination switch leaves connected
  • conductor seating, data-line polarity consistent with the rest of the segment, shield contact, strain relief
  • any abandoned length from the old route disconnected, not left as a stub
  • supply present at both end stations whenever the segment must communicate, including under switching states

Repeater supplies and interference paths #

A brief dip on a repeater's supply removes every station behind it and restores them when it returns — the exact self-recovering picture in the buffer. A multimeter sample will usually miss a dip of tens of milliseconds, so log the repeater's supply with a recording instrument during the fault window.

Stable service does not exclude deterioration: the encoder work disturbed connectors, and loosened shields, a failing supply or a re-routed power cable near the bus all eat margin over years. Compare buffer timestamps against drive run/stop and machine motion; a repeatable correlation points at coupling, independence rules it out. Where two suspects survive, a bus-diagnosis instrument that measures signal distortion across the segment — a Procentec Profitrace2 is one such tester — settles it without trial part-swapping.

Confirm the installed network against the STEP 7 project #

Every physical device must appear in the project with the role its configuration gives it — the OP27 and any other masters included, because presence on the cable does not define a device's role. Check that station addresses are unique across the whole bus, then compare the configured network against the installed one before and after any repair, so the fix is proven against the drawing rather than against memory.

The decision for this network #

Rank by the evidence: the bus ran seven years untouched, and the dropouts began after the encoder was replaced, its termination was re-set and its connection was tied in elsewhere. Commit to that order:

  • Termination geometry on the affected segment: a termination still on at a device that is no longer an end, off at a device that now is one, or an endpoint losing its supply. Inspect both ends of the segment named by the earliest OB86 entries.
  • The relocated tie-in itself: seating, shield, polarity, an abandoned stub.
  • The repeater feeding that segment: a logged supply dip.
  • Losses spanning several segments: only then the master's DP interface connection, the first repeater, or a shared supply.

The buffer read decides this in one afternoon. Earliest events on one segment point at the first three; spanning events point at the last; an encoder-only dropout points at the encoder's local connection and supply. Leave the CPU and the DP module alone until the segment map has cleared the cable side: a fault that lights all three indicators and heals itself is a physical-layer fault until proven otherwise.

Questions asked

What makes the indicators clear without a restart?

The bus retries cyclic exchange continuously. When the interruption ends — the connector reseats, the supply dip passes, the interference stops — the master restores exchange with the slaves and the operating system logs the returns as outgoing events, clearing the indicators. Recovery proves the interruption ended at that moment; it does not identify the marginal component, which is why the buffer read comes before any part swap.

Why does OB85 appear in what is a communication fault?

OB85 is the program-processing error block. When a configured DP slave is gone, the operating system cannot update the process image for that slave's addresses and calls OB85. It is a downstream symptom of the same station loss that raised OB86, and its appearance confirms the fault sits in cyclic data exchange, not in the user program itself.

The diagnostic buffer is full of recovery messages. How do I find the first fault?

Save the whole buffer after each event, before it wraps. Work from the earliest entry with an incoming event and note that station address; the later outgoing entries only confirm the return. For a permanent record, program WR_USMSG in OB86 to write the event as an alarm and forward it to a monitoring device, so the initiating fault survives the buffer rolling over.

Can an unpowered end station break a segment that otherwise looks fine?

Yes. In 9-pin Sub-D connectors the terminating network is fed from the endpoint station's own supply, so when an end device loses power the segment loses termination at the same moment. The bus then runs with reflections and drops intermittently, which is why endpoint supplies belong on the fault checklist, including any cabinet that gets switched off overnight.

How do I tell a repeater supply dip from a cable fault?

Both remove stations and both self-recover. A dip takes every station behind one repeater at once and returns them within seconds, while a cable or termination fault usually starts with single-station retries and noise-dependent losses. Log the repeater's supply with a recording meter during the event; a multimeter sample will normally miss a dip of tens of milliseconds.

Keep reading

  • Troubleshooting

    Troubleshooting ControlNet scanning of legacy 1771 specialty I/O

    When a legacy 1771 specialty module will not scan over ControlNet, separate three layers before touching hardware: the chassis backplane allows two master/slave connections and each expander-class module takes one, the TIC-chip ASIC conflict applies to specific pairs only, and RSLogix 5000 gives you no configuration wizards for 1771 block transfer. Bench-test the exact module-and-chassis combination before go-live.

    Industrial networks

  • How-to

    How to map a 1747-SDN scanner for DeviceNet drive control

    Seeing the drive in an RSNetWorx browse is not an I/O connection. Put the drive in the 1747-SDN scanlist, read the M-file words the scanner mapping assigns, and reference those mapped words in the SLC 5/04 ladder. A node address is not an integer-file offset: the mapping view, not node 10, decides where each command and status word lives.

    Industrial networks

  • Troubleshooting

    Why CODESYS EtherCAT master xError stays false on cable pull

    `EtherCAT_Master.xError` stays FALSE because the master does not latch slave loss into the application variables when restart handling is off; the engineering tool reads device state directly, so it still shows the error. Tick `Automatically restart slaves` on the master and `Enable Diagnosis for Devices` for the extended instance, then pull-test: the loss reaches the program and the slave rejoins when the cable returns.

    Industrial networks