Detecting Lost PROFIBUS Stations After Siemens Y-Link with SFC59

Overview

On a Siemens H-System (S7-400H) where one or more Y-Link couplers are used to attach PROFIBUS PA segments to a PROFIBUS DP network, the loss of a downstream PA slave (or DP slave attached through the Y-Link's lower segment) is normally invisible to the H-CPU at the application level. The CPU only sees a healthy diagnostic frame from the IM 153-2 (or the older IM 157) and is not automatically informed which individual station on the lower segment has dropped out, exchanged a module, or signaled a channel fault.

The supported mechanism for surfacing these events is the diagnostic buffer that the Y-Link's head module (IM 153-2 in current hardware, IM 157 in legacy hardware) maintains for every diagnostics-capable DP-V1 slave on its lower segment. The H-CPU reads this buffer with SFC 59 "RD_REC" (Read Data Record). The data record layout, the slot/index mapping, and the call parameters are documented in Siemens manual entry ID 1142696 – Bus links DP/PA coupler, DP/PA link, and Y link, section 12.2.2 "Reading out the diagnostics of underlying slaves".

This reference covers the complete procedure: SFC 59 invocation, record-number and I/O-ID semantics, the differences between the old IM 157 and the new IM 153-2, integration into the S7-400H fault-tolerant program, and a working STL sample that exposes lost-station events to the OS or HMI.

Y-Link Position in the Network

The Y-Link acts as a DP-V1 master on the lower (PA/DP) segment and as a DP slave on the upper (DP) segment. The head module is addressed by the H-CPU as a standard PROFIBUS DP slave; every PA/DP device that sits behind it is invisible to the upper master unless diagnostics are explicitly requested.

Prerequisites

• Y-Link head module must be configured as a DP-V1 slave on the upper segment. The DPV1 master attribute is required for the diagnostics buffer in the IM 153-2 to be populated.
• Each downstream slave that is to be monitored must be a DP-V1 diagnostics-capable station. Pure DP-V0 devices do not write extended diagnostic records; the buffer entry will not appear.
• STEP 7 V5.x or PCS 7 V6.x/V7.x/V8.x with the corresponding GSD file for the Y-Link installed in HW Config.
• Available OB1 / OB35 / OB82 / OB86 time slice for the SFC 59 call. The call must not be issued more often than the Y-Link refreshes its buffer (typically 1–2 s, dependent on the configured bus parameters).
• For S7-400H: both H-CPUs must be able to issue RD_REC independently; the read result is the same on both sides because the buffer is owned by the Y-Link itself, not by the H-CPU.

Diagnostic Data Flow

SFC 59 RD_REC Call Parameters

The call to SFC 59 is structured as follows. The combination of LADDR, IOID, and RECNUM is what targets the Y-Link's slave-diagnostic buffer rather than any other record.

The semantics of IOID are critical. Per the standard functions manual for S7-300/400, IOID = B#16#54 denotes a "head module" slot. For the Y-Link this is mandatory; using B#16#54 on a normal DP slave slot returns an error.

Data Record Layout

Each valid data record returned by the Y-Link identifies a single underlying slave. The structure is the standard PROFIBUS DP-V1 diagnostic data record, but the Y-Link attaches the slave's PROFIBUS address in the slot/index field, which is what makes lost-station detection possible. The relevant fields are:

STL Sample: Polling the Y-Link Buffer

The following STL shows a complete polling cycle. The instance DB (DB100) holds the returned record, the station address, and a per-station latch bit that the OS can scan.

// OB1 – call once per scan, on rising edge of OB1_PREV_CYCLE
CALL SFC 59 (
    REQ    := TRUE,                              // trigger every cycle
    IOID   := B#16#54,                           // head module
    LADDR  := W#16#007F,                         // Y-Link at DP addr 7
    RECNUM := B#16#00,                           // first record
    RET_VAL:= MW 200,                            // result code
    BUSY   := M 202.0,                           // job in progress
    RECORD := P#DB100.DBX 0.0 BYTE 64            // 64-byte destination
);

// When RET_VAL = 0 and BUSY = 0, parse the record
A     M 202.0;          // BUSY
JC    POLL;             // still running, skip parse
L     MW 200;           // RET_VAL
L     W#16#0000;
<>I;
JC    ERR;              // non-zero RET_VAL

// Extract the station address from byte 4 of the record
L     DB100.DBB 4;      // low byte of PA/DP station address
T     DB101.DBB 0;      // station_of_interest
L     DB100.DBB 5;
T     DB101.DBB 1;      // high byte (PA addresses fit in low byte only)

// Read the station-lost flag from byte 6
L     DB100.DBB 6;
L     B#16#10;          // bit mask 0x10 = station lost
AB    ;
=     DB101.DBX 2.0;    // station_lost_latch

POLL: NOP 0;
ERR:  NOP 0;

Walking the Full Buffer

A single RD_REC returns only one record. To find all failed stations, iterate RECNUM from 0 upward until RET_VAL = W#16#80B2 ("record not available") is returned. The recommended scan period is 2–5 seconds; faster scans saturate the Y-Link and starve the process data cycle.

Old (IM 157) vs New (IM 153-2) Y-Link

The field report explicitly highlights that the diagnostic mechanism is identical in concept, but the data record contents differ between the legacy IM 157 and the current IM 153-2. The principal differences an integrator must handle in code are:

Integration into S7-400H

In an H-system, RD_REC is called on each H-CPU independently against the Y-Link's diagnostic address. Because the Y-Link is a single physical device on the PROFIBUS, both H-CPUs see the same buffer state. To avoid duplicate alarming on the OS, the recommended pattern is:

1. Run the polling on the master H-CPU only, identified by OB70 / OB72 state flags or the standard block FC 100 "SWR\_AGGR" from the PCS 7 library.
2. Mirror the resulting station-lost bits into a redundant DB that is exchanged via the H-sync mechanism, or send them to the standby CPU via SFC 90 "H_CTRL"-compatible shared data blocks.
3. Generate the operator message (e.g., "&PA field failure, addr N") once, from the master side only.

Troubleshooting Matrix

Verification

After implementing the polling loop, verify with the following checks before going into production:

1. Disconnect a single PA slave's segment. Within one poll cycle (2–5 s) the station-lost bit for that address must be set in DB101.
2. Reconnect the slave. The bit must clear after a configurable confirmation time (recommend 3 consecutive successful reads).
3. Power down the entire Y-Link. The H-CPU must detect the Y-Link's own DP slave failure via OB86; the RD_REC path must return 0x80A2 (timeout) and the program must not hang in BUSY.
4. In the H-system, force a master/standby failover (SWR switch). Polling must resume on the new master without operator action.
5. Trigger OB82 (diagnostic interrupt) and confirm the OS does not duplicate the lost-station alarm from the H-CPU that is now in standby.

Field Commissioning Tips

• Always commission with PCS 7 plant view: the OS will auto-generate the operator message text from the slave's tag, so the manual parse above is only needed if PCS 7 messages are not sufficient.
• Set the PA coupler termination resistors ON at the two physical ends of the segment and OFF on every coupler in between. A common cause of intermittent station-lost events is a missing or doubled terminator.
• Keep the RD_REC polling on OB35 (1000 ms) rather than OB1; OB1 will starve the H-CPU's main cycle if the poll returns BUSY = TRUE for several scans.
• Do not poll the same Y-Link from both H-CPUs simultaneously. Coordinate via the redundancy status word.

FAQ