S7-300 PROFIBUS DP Intermittent Slave Disconnects: Field Guide

When a Siemens SIMATIC S7-300 CPU315-2DP master loses contact with one or more PROFIBUS DP slaves for less than one second and then recovers automatically, the symptom itself rarely identifies the cause. What the symptom reliably identifies is a physical-layer disturbance, a bus-parameter mismatch, or a configuration edge case that the master corrects with a retry before the slave watch-dog fires. Two identical plants exhibiting the same intermittent disconnects, after wiring has been verified and replaced on one plant, is a strong indicator that the defect lives in the bus parameters, grounding topology, GSD configuration, or electromagnetic environment rather than in any single cable. This guide isolates each cause with the field-proven diagnostic sequence used by PROFIBUS certified engineers, with concrete parameter values, diagnostic buffer codes, OB86 interpretation, and verification steps specific to the S7-300 / TeSys LUFP7 / WAGO I/O-SYSTEM 750 topology.

Problem Statement and Scope

The reported fault pattern is reproducible on at least two physically separate, identical systems. Each system consists of:

• Siemens SIMATIC S7-300 CPU315-2DP (typical order numbers 6ES7315-2AH14-0AB0 or 6ES7315-2AG10-0AB0) acting as a PROFIBUS DP Class 1 master on the integrated DP port. See the CPU315-2 DP device manual for the official pinout, bus parameter set, and diagnostic buffer layout.
• Schneider Electric TeSys LUFP7 / LUFP7B PROFIBUS DP to Modbus gateway interfacing to one or more TeSys U motor management controllers. Reference the Schneider Electric LUFP7 product page for the GSD file and DIP switch layout.
• WAGO I/O-SYSTEM 750 fieldbus coupler (typical order number 750-303 or 750-333) hosting digital and analogue I/O modules. Reference the WAGO I/O-SYSTEM 750 downloads for the current GSD revision.

Observations recorded in the CPU diagnostic buffer and reported by the operator:

• Duration of slave loss: typically less than one second (a few CPU scan cycles).
• Frequency: random — two events within one hour, then several days gap, then a single event.
• Affected slaves: one slave or both slaves simultaneously.
• No permanent SF or BF LED after recovery.
• Wiring has been verified and replaced on one system without resolving the fault.
• Direct I/O mapping (PIW / PQW) is used in the user program; no SFC14 / SFC15 DPREAD / DPWRDAT calls.

This pattern is characteristic of a transient physical-layer event (reflections, common-mode noise, ground loops) or a bus-parameter mismatch causing a brief token-passing or telegram retry fault that the master corrects transparently without transitioning to stop.

Root Cause Taxonomy for Sub-Second PROFIBUS Disconnects

The following categories account for the vast majority of intermittent, self-healing PROFIBUS DP faults. Each category is mapped to the affected system components above and to the corresponding diagnostic signature. Cross-referencing with the diagnostic buffer entries in Step 7 narrows the cause quickly.

PROFIBUS DP Bus Parameter Analysis

Bus parameters control the timing window in which the master and slave must respond. The default profile selected by Step 7 for the CPU315-2DP integrated DP port is usually 'DP' at 1.5 Mbps. With one LUFP7 and one WAGO coupler on a short segment, this default is normally sufficient. The defaults become problematic when:

• The baud rate has been raised (e.g., to 6 Mbps or 12 Mbps) without shortening the cable and verifying termination.
• A repeater has been inserted, introducing a fixed signal delay the master must accommodate.
• Segment length exceeds the spec (200 m at 1.5 Mbps, 100 m at 12 Mbps).
• Some third-party slaves have unusually long internal processing latency.

Reference Bus Parameters by Baud Rate

Bit time at 1.5 Mbps is 0.667 µs; at 12 Mbps it is 83.3 ns. A slave watch-dog (default 10 ms, i.e., 15,000 bit times at 1.5 Mbps or 120,000 bit times at 12 Mbps) fires when the master has not polled within that window and forces the slave to its safe state — this is the event OB86 reports as 'station failure'. When the symptom is sub-second and self-clearing, the watch-dog has not actually fired; the master's retry counter has detected a single bad telegram and transparently re-issued it. The diagnostic buffer time-stamps tell you which event occurred.

PROFIBUS DP Cycle Time Calculation

For a configuration with N slaves, the bus cycle is approximated by:

T_cycle = Sum_{i=1..N} (T_telegram_i + T_propagation_i + T_slave_proc_i) + T_master_proc

For 2 slaves with 16 bytes of I/O each at 1.5 Mbps on a 100 m segment:

• Per-telegram data = 8 bytes × 11 bit times = 88 bit times
• Telegram overhead (SD, DA, SA, FC, FCS, ED) ≈ 110 bit times
• Propagation round trip = 4 ns/m × 100 m × 2 = 0.8 µs = 1.2 bit times
• Slave processing = ~50 bit times
• Per-slave total = 88 + 110 + 1.2 + 50 ≈ 250 bit times = 167 µs
• Master processing between telegrams ≈ 100 bit times
• T_cycle = 2 × (250 + 100) = 700 bit times = 467 µs

The default 10 ms watch-dog provides a 20x margin — the bus can survive a single failed telegram and still not trigger OB86. This is exactly the regime that produces the reported intermittent symptom.

How to Read and Modify Bus Parameters in Step 7 (SIMATIC Manager)

1. Open the S7 project, expand the SIMATIC 300 station, and open Hardware.
2. Double-click the DP master port (X2) of the CPU315-2DP to open Properties - PROFIBUS interface DP.
3. Click Properties → Network Settings → PROFIBUS(1).
4. Verify the Baud rate and Profile. For mixed-vendor slaves (Schneider LUFP7 plus WAGO 750-303), keep the 'DP' profile at 1.5 Mbps until stability is confirmed.
5. Open the Bus Parameters tab. Defaults match the table above; any deviation should be intentional and documented.
6. If a repeater is present, ensure it is counted as one extra Trep delay in the Tslot calculation: each repeater adds roughly 2 bit times at 1.5 Mbps and 7 bit times at 12 Mbps.

Do not increase the CPU cycle (OB1) time as a first-line fix. The OB1 cycle is independent of the DP cycle; the DP cycle is driven by the master's poll list and the configured slave watch-dog. A longer OB1 only delays the application reaction to the gap but does not cause or cure the disconnect.

Hardware Verification: Termination, Cabling, and Connectors

Intermittent, sub-second disconnects across two plants after cabling was replaced points strongly to termination, shielding, or connector issues rather than to the cable itself. Verify in the order below.

Step 1 — Identify the Segment Ends

PROFIBUS DP is a linear bus. Exactly two nodes on the segment must have termination enabled — one at each physical end of the segment. Any node between the ends must have termination off. For the CPU315-2DP → LUFP7 → WAGO 750-303 topology on a single segment:

• If the CPU is at one physical end and the WAGO coupler at the other, termination must be on at the CPU's DP connector and at the WAGO coupler. The LUFP7 in the middle must have termination off.
• If the CPU is at one end and the LUFP7 is at the other (with the WAGO coupler on a drop from LUFP7), termination is on at the CPU and at the LUFP7's PROFIBUS port. On a drop, the WAGO coupler has termination off.

Step 2 — Check Termination Switch Position

• Siemens PROFIBUS connectors (6ES7972-0BA12-0XA0 or 6GK1500-0FC00) have a slide switch. With the switch toward the cable entry, termination is ON; toward the outgoing PG socket, OFF.
• WAGO 750-303 / 750-333 couplers expose a 3-position slide switch inside the housing: OFF / ON with bus termination / ON with bus termination plus 24 V power to the termination network.
• Schneider LUFP7 has a DIP switch block on the front. The PROFIBUS DP port has built-in termination selectable by switch; verify the switch position against the device label.

Step 3 — Measure Termination Resistance With the Bus Powered Down

1. Disconnect 24 V from the CPU315-2DP, LUFP7, and WAGO coupler.
2. Disconnect any repeaters.
3. Measure resistance between pin 3 (RxD/TxD-P, B-line) and pin 8 (RxD/TxD-N, A-line) at any PROFIBUS connector.

Expected reading: 220 Ω ±10% with both segment ends' termination enabled. With only one end terminated, the reading is high-impedance or ~3 kΩ depending on slave input impedance. With three terminators enabled, the reading is ~110 Ω. The standard PROFIBUS termination network is 220 Ω between A and B, 390 Ω pull-up to VP (+5 V), and 390 Ω pull-down to DGND, applied at both segment ends only.

Step 4 — Check Connectors

The two most common intermittent PROFIBUS faults come from the connector itself:

• Pigtail shield clamp not firmly gripping the cable shield. The shield terminates to ground through this clamp, not through the drain wire of the bus cable.
• Loose incoming cable. The D-sub must be screwed down with both UNC 4-40 screws torqued to ~0.4 Nm.
• Oxidized pins. Look for green or black residue on pin 3 and pin 8. Clean with DeoxIT D5 or replace the connector.
• Swapped A and B lines. Pin 3 (B-line, green wire) and pin 8 (A-line, red wire) must not be crossed. The standard PROFIBUS wire color is red for A (pin 8) and green for B (pin 3).

Step 5 — Shield and Equipotential Bonding

Shielding on PROFIBUS is two-sided: both ends of the cable shield must contact the connector clamp, and both ends of the shield must be bonded to the cabinet grounding bar. The cabinet-to-cabinet equipotential bonding conductor should be ≥10 mm² Cu, and the cabinet grounding bar must be at the same potential as the master and slave cabinet grounds. A ground potential difference of more than ±7 V between cabinets causes shield current to flow through the bus cable, raising the common-mode voltage on the data lines and producing intermittent faults precisely matching the reported symptom. Measure AC and DC voltage between cabinet ground bars under normal operating load; the reading should be below 1 V RMS and below 7 V peak.

PROFIBUS Cable Specification and Topology Rules

Topology rules per IEC 61158-2 / EN 50170:

• Stub length (drop) is limited to 0.3 m at 12 Mbps, 1 m at 1.5 Mbps, 6 m at 500 kbps. Avoid drops entirely at 12 Mbps.
• Minimum 1 m spacing between two repeaters on a segment.
• Maximum total cable length with 4 repeaters at 1.5 Mbps: 4 × 200 m = 800 m.
• PROFIBUS cable must not be routed in parallel with VFD output cables, motor leads, or welding transformers. Maintain ≥200 mm separation, 90° crossing only.

Repeaters are mandatory when segments must be extended beyond 32 nodes or 200 m, but each repeater introduces a fixed propagation delay (~2 bit times at 1.5 Mbps, ~7 bit times at 12 Mbps) that the master's Tslot setting must accommodate. With the default Tslot of 100 bit times at 1.5 Mbps, three cascaded repeaters fit; with the default Tslot at 12 Mbps, only two repeaters fit before the master must be re-parameterized.

Configuration and Software Checks

GSD File Selection

Both the LUFP7 and the WAGO 750-3xx coupler require a matching GSD file in the Step 7 hardware catalog. The order of checking is:

1. In Step 7 HW Config, right-click the DP master system, choose Install GSD or Options → Install GSD File depending on the Step 7 version.
2. For the WAGO coupler, the file is named WAGO_0B0E.GSD or revision-specific equivalent such as WAGO_0B38.GSD, available from the WAGO I/O-SYSTEM 750 downloads page.
3. For the LUFP7, the file is shipped on the Schneider Electric TeSys U PROFIBUS DP gateway CD or downloadable as TELEM_LUF.GSD from the Schneider Electric LUFP7 product support page.

An outdated GSD file produces a 'configuration fault' diagnostic normally visible in HW Config. However, a partial GSD mismatch can also cause intermittent DP-V1 read/write transactions to fail without a permanent fault. Update to the latest revision that matches the firmware version installed on the device. Cross-reference your findings against the PROFIBUS troubleshooting checklist from Delta Motion which lists wrong slave address in master, wrong slave address on the device, and incorrect GSD selection as the three primary configuration faults to eliminate before deeper physical-layer diagnosis.

Slave Address Verification

The most common configuration error is a slave address set in HW Config that does not match the address set on the rotary switches of the physical slave. With addresses 1 and 2 typically taken by other masters or services, the LUFP7 and WAGO coupler are usually set to addresses 3 and 4 in this topology. Verify uniqueness on the bus by:

• Reading the rotary switches (decimal) on each device.
• Cross-checking against the PROFIBUS address set in HW Config for that node.
• With the CPU in stop and a live bus monitor, verifying only the configured addresses are active on the segment.

A duplicated address produces a permanent diagnostic, not an intermittent one — but an address that the slave only sometimes acknowledges (e.g., a noisy address-sensing input on the LUFP7) can produce the reported intermittent symptom. Replace the LUFP7 if its address-sensing is unreliable.

Slot Configuration Consistency

Each module inserted in the WAGO coupler must match the slot configured in HW Config. The WAGO 750-303 coupler accepts up to 64 modules depending on configuration; the slot mapping is set when the GSD is dragged into the HW Config slot table. Mismatches typically produce a diagnostic entry 'Module does not match configuration' (event ID 0xE385 in the CPU diagnostic buffer).

Watch-dog Time on Each Slave

The DP slave watch-dog is set per slave in HW Config under the slave's properties → DP-Slave Properties. The default is 10 ms. If a slave reports a fault within the same scan cycle as the master detects it, the master will resync without OB86 firing. If the fault exceeds the watch-dog, OB86 fires with 'Station failure'. The diagnostic buffer will indicate exactly which event occurred. The CPU315-2 DP manual documents the watch-dog interaction with the S7-300 diagnostic buffer.

Diagnostic Buffer and OB86 Interpretation

The CPU diagnostic buffer is the single most valuable tool for isolating the source. Open it via Step 7 → PLC → Module Information → Diagnostic Buffer or via the Web Server of the CPU315-2DP if enabled. The latest diagnostic buffer entries are listed at the top.

Common Diagnostic Buffer Entries

The pattern of the events — single vs. multiple slaves, frequency, time of day — is itself diagnostic. Two slaves failing simultaneously on the same segment (0xE38C) is a strong physical-layer signature. A single slave failing alone is more likely a configuration or device-level issue.

OB86 Programming for Forensic Logging

OB86 (Rack/Station failure error OB) fires when a DP station goes down or comes back. Add code to OB86 to record the logical base address into a flag word or data block. This gives a time-stamped log of every transient event. The CPU315-2DP provides the local variables described in the S7-300 system and standard functions manual.

// OB86 - DP Station Failure / Return
VAR_TEMP
  info : STRUCT
    event_class     : BYTE;    // B#16#38 (failure) / B#16#39 (return)
    fault_id        : BYTE;    // B#16#01 = station failure
    dummy           : BYTE;
    dummy2          : BYTE;
    logical_address : WORD;    // logical base address of failed DP slave
  END_STRUCT;
END_VAR

IF info.event_class = B#16#38 THEN
   "DB_FaultLog".fail_count       := "DB_FaultLog".fail_count + 1;
   "DB_FaultLog".fail_addr        := info.logical_address;
   "DB_FaultLog".last_fail_time   := DT_DATE_TIME;
ELSIF info.event_class = B#16#39 THEN
   "DB_FaultLog".return_count     := "DB_FaultLog".return_count + 1;
   "DB_FaultLog".return_addr      := info.logical_address;
   "DB_FaultLog".last_return_time := DT_DATE_TIME;
END_IF;

With this log enabled, capture the fault window. A fault at address 256 (logical base of LUFP7) versus a fault at 512 (logical base of WAGO) versus simultaneous faults at both addresses tells you whether the physical-layer event is between two specific nodes or on the shared trunk.

Field Diagnostic Tools and Live Monitoring

Static measurements (resistance, continuity) catch gross problems. To catch a fault that occurs for 200 ms once every several days, live monitoring is essential.

Tooling Options

What to Look For in the Live Trace

When running the live bus monitor during a fault event, the following signatures are diagnostic:

• Frame Error Count / Line Error Count on a single slave rising in clusters → physical-layer noise coupling into that segment.
• Diag Repeats / Retry Limit increasing → master detected N unsuccessful attempts to reach the slave.
• Signal voltage peak-to-peak between 4.5 V and 5.0 V on the data lines is nominal; below 3.5 V indicates termination or cable loss.
• Asymmetric signal rise time between A and B lines → shield contact issue or termination asymmetry.
• Repeats coincident with motor starts, contactor closes, or specific VFD ramp events → EMC coupling.

Step-by-Step Resolution Workflow

1. Capture fault history. Enable OB86 logging and run the plant for at least 24 hours under representative load. Capture the fault rate per slave address and per shift.
2. Verify topology. Document the physical layout of the bus — connector locations, segment ends, repeaters, cable lengths, routing. Compare against the topology entered in HW Config.
3. Inspect connectors. Power down. Unplug each PROFIBUS D-sub connector. Inspect the pigtail shield clamp, the screw retention, and the pin condition. Reseat each connector with the retaining screws torqued to ~0.4 Nm.
4. Measure termination resistance. With power off and repeaters isolated, measure A-B resistance at three points on the segment. Confirm 220 Ω at segment ends, high-Z at mid-segment nodes.
5. Check shield continuity. Use a low-voltage ohmmeter to verify shield continuity end-to-end (typically below 5 Ω). Verify the cabinet grounding bar connections.
6. Measure ground potential difference. With both cabinets powered and operating, measure AC and DC voltage between the two cabinet ground bars. Reading should be below 1 V RMS and below 7 V peak.
7. Update GSD files. Replace the WAGO and Schneider GSD files in HW Config with the latest revisions matching the installed firmware on each slave.
8. Verify address switches. Read each device's address switches under power (LUFP7 rotary / DIP; WAGO coupler rotary / DIP) and reconcile against HW Config.
9. Reduce baud rate temporarily. If higher baud rates are configured, drop to 1.5 Mbps for one week and compare fault rate.
10. Live-monitor the bus. Install a ProfiTrace or similar live monitor between the CPU and the first slave. Capture at least one full fault event and the line-error statistics.
11. Inspect routing for EMC. Move any VFD output cables, contactor coils, and welding transformers at least 200 mm from the PROFIBUS cable, with 90° crossings only.
12. Replace suspect hardware. If a specific slave shows disproportionate retry counts versus the rest, swap that slave with a known-good spare. Confirm whether the fault migrates with the device or stays with the segment.
13. Document and lock. Update the as-built schematic, mark the address switch positions, and lock the configuration in the S7 project backup.

Verification and Long-Term Stability

After the corrective action, the system must be verified to be fault-free for a minimum of 168 hours (one full production week) before the issue is closed. The verification criteria:

• OB86 event count for any DP station equals 0 over the verification window.
• Diagnostic buffer 'Station failure' events equal 0.
• ProfiTrace line-error counters are stable, with no sustained rise.
• Process values from LUFP7 and WAGO are consistent with manual readings (no frozen fallback values from the OB86 handler).

The OB86 freeze-the-I/O behavior described in the source is acceptable as a fallback — it keeps the process safe by holding last-known-good values during a fault — but it should be removed once the underlying cause is fixed. Holding values indefinitely defeats the purpose of the I/O and prevents detection of the next, possibly hardware-damaging, event.

Long-Term Hardening

1. Add a periodic health check in OB1 that reads a defined status word from each DP slave (e.g., LUFP7 status register, WAGO 750-3xx diagnostic register) and raises an HMI alarm on degraded communication quality before a full disconnect occurs.
2. Install surge protection (Phoenix Contact PT-IQ-5-HF or equivalent) on the PROFIBUS cable where it enters the cabinet, especially in panels fed from overhead cable trays or in plants with outdoor cable runs.
3. Schedule a quarterly bus health check: ProfiTrace scan, termination resistance verification, connector torque check, and visual inspection of cable routing. Document results and trend the line-error counters.

Reference Topology