Resolving Sinamics G120C Profinet Shared I/O Startup Failure

Sinamics G120C PN drives connected to a Profinet plant network occasionally fail to appear in the STEP 7 / TIA Portal hardware configuration immediately after a cold plant power-up, even though they answer ICMP echo (ping), are listed by the "Edit Ethernet Node" tool, and share the network with hundreds of other Profinet devices that come up cleanly. The diagnostic buffer records "IP address could not be allocated", and recovery typically requires a power cycle of the affected drive. This article consolidates field findings, root-cause analysis, and a structured remediation procedure for the issue.

1. Problem Overview

After energising the entire plant in one operation, a subset of G120C drives is reported as not accessible in the online hardware configuration of the S7-319. The same drives are:

• Reachable via ICMP echo from the engineering station.
• Detected by the "Edit Ethernet Node" / topology editor with the correct MAC and configured Profinet name.
• Properly identified by Starter / Startdrive as a Profinet node.
• Shown as present by the F-configuration of the 319F (Profisafe side is up).

The standard 319 hardware configuration, however, cannot complete AR (Application Relationship) establishment with the drive, and the diagnostic buffer records IP address could not be allocated. Cycling 24 V to the drive makes the device reachable on the next PLC scan. Different drives fail on different power-up events, with no apparent correlation to physical position, line topology, or ring redundancy.

2. Affected Hardware and Firmware

Total plant footprint exceeds 200 drives, which rules out per-drive manual recovery as a permanent fix. The remaining population of S110, S120, and ET 200 stations behaves deterministically, which is the strongest indication that the fault class is specific to G120C PN shared-device handling during cold start, not to the physical layer.

3. Network Topology and Shared I/O Configuration

The two controllers are configured with the STEP 7 / TIA "Shared Device" mechanism: a single Profinet device exposes two or more ARs, and the engineering assigns each AR to a different controller. In the affected plant, the G120C PN drives are shared such that:

• The S7-319F owns the Profisafe slot (typically slot 0 of the F-host).
• The S7-319 owns the standard telegram slots (e.g. telegram 1, telegram 350/352/370/371, free PROFIsafe slots).

Both CPUs publish their own GSD-based hardware configuration referencing the same drive name and IP, but each with its own slot assignment. On a clean power-up the AR establishment order is determined by:

1. Physical link-up of the drive's Profinet port (driven by the drive's 24 V and main power rise times).
2. DCP identification request / response (the device is found by name).
3. AR establishment with the configured controller(s) in the order they boot and complete their own PN interface start-up.
4. Parameter assignment and cyclic data exchange.

When the standard 319F comes up first, the drive accepts the F-AR but the F-controller only configures the safety slots, leaving the standard slots un-claimed. If the standard 319 then attempts its own AR and the device has not yet finished a state transition, the controller logs an IP-allocation-style diagnostic. Because the drive keeps its IP (assigned by the F-AR), the standard controller's DCP request fails to acquire the address and the device stays in "not reachable" from the perspective of the 319's HWCN.

4. Observed Symptoms and Diagnostic Buffer Entries

Different drives fail on different cold starts, which rules out a static cabling or addressing problem and points to a race condition between controller boot order and the drive's internal Profinet stack state machine.

5. Root Cause Analysis

5.1 Profinet DCP and AR lifecycle

PROFINET IO uses DCP (Discovery and Configuration Protocol, based on LLDP) to identify a station by its Profinet name and to assign its IP. After successful DCP, the IO controller opens an AR (Application Relationship) by sending an AR-Request. The device replies with an AR-Response containing the slot / submodule configuration it can offer. Only after the controller commits the expected submodules does the device transition to the Data Exchange state and start cyclic I/O.

The G120C PN firmware implements the AR state machine on its internal Profinet stack. If the stack has already accepted one AR (in this case the F-AR from the 319F) and is mid-transition, a second AR from the standard 319F within the stack's reject window is dropped, and the controller logs an IP-allocation diagnostic. After the device reaches the steady state of the first AR it can still accept additional ARs, but only if the second controller retries within the device's tolerance.

5.2 Why the G120C is more exposed than S110 / S120 / ET 200

The G120C uses a more compact Profinet stack than the modular S120, and the CPU resource available to AR handling is more constrained during the first 10–30 seconds of boot. The S120 line modules and the ET 200 IM stations have larger state machines and longer boot budgets, so even a contested AR is accepted on retry. The G120C's stricter boot window is consistent with the field behaviour: the S110, S120, and ET 200 stations come up on every cold start, while the G120C population shows intermittent visibility loss.

5.3 Power-up sequencing as a hidden variable

Field trials show that altering the relative order in which the standard CPU and the F-CPU boot shifts the failure pattern. A faster-booting standard 319 that completes its PN interface start before the F-CPU reaches AR-eligible state tends to claim the standard AR first, after which the F-CPU still succeeds (it is generally more tolerant because the safety stack waits on a watchdog). A faster-booting F-CPU that grabs the F-AR first is the more failure-prone sequence because it leaves the standard CPU in a window where the G120C is least likely to accept a second AR within tolerance.

5.4 Firmware V4.5 vs V4.6

Upgrading a representative block of 10 drives from V4.5 to V4.6 of the G120C PN firmware did not resolve the problem; the same intermittent visibility loss reappeared on subsequent cold starts with all drives at V4.6. The release notes for V4.6 include Profinet stability improvements, but no entry is known to address the shared-device AR race during cold start. V4.7 SPx and later firmware branches should be re-evaluated against the symptoms in this article, but at the time of the field report, V4.6 was the latest available branch for the part number.

6. Engineering Workarounds

The following workarounds are organised from least to most invasive. Apply in order until the symptoms are eliminated.

6.1 External 24 V supply to the drive

Power the G120C's control electronics (terminals 31, 32) from an external 24 VDC source that is energised before the CPUs and before the main contactor on the line side. This brings the Profinet stack of the G120C to a fully initialised state in advance of the first DCP identify from either controller. When the standard 319F then attempts its AR, the G120C is in a steady state and accepts the second AR on the first try.

• Set the external 24 V to come up at least 5 s before the first CPU's RUN.
• Use a dedicated, fused 24 VDC branch with a load capacity that accommodates the worst-case inrush of all G120C control boards.
• Verify with a scope that 24 V is within the G120C tolerance (typically 20.4 V to 28.8 V) at the terminals during plant power-up.

6.2 Delayed SFC12 in a startup OB

Use SFC12 D_ACT_DP in a time-delayed startup OB to deactivate and then re-activate the affected shared-device AR. The typical pattern is:

1. In OB100 (warm restart) or OB101 (hot restart), wait for the configured delay (e.g. 5–15 s) using a timer or a time-of-day comparison, so the F-CPU has had time to claim its AR first.
2. Call SFC12 with MODE = 0 to deactivate the shared device on the standard CPU's AR.
3. Wait one PROFIBUS / Profinet cycle.
4. Call SFC12 with MODE = 1 to re-activate the AR. The G120C will accept the standard AR cleanly because the device is in steady state.

Skeleton STL snippet (illustrative; adapt to slot / hardware identifier of the project):

// OB100 - delayed AR re-activation for shared G120C drives
      CALL SFC 12                // D_ACT_DP
       MODE   := B#16#0          // 0 = deactivate
       LADDR  := W#16#0          // hardware ID of the shared device AR
       RET_VAL:= MW 100          // result word
       BUSY   := M 110.0
      // optional: wait for F-CPU to stabilise
      CALL SFC 1                 // SFC1 / wait variants
       ...
      CALL SFC 12
       MODE   := B#16#1          // 1 = activate
       LADDR  := W#16#0
       RET_VAL:= MW 102
       BUSY   := M 110.1

Wrap the call in a check that only fires on cold restart, to avoid unnecessary AR churn on hot restarts where the issue has not been observed.

6.3 Power-up sequencing automation

If 24 V pre-energisation is not available, codify the energisation order at the switchgear level so the G120C's 400 V / 480 V main contactor closes only after both CPUs report RUN and the F-CPU has logged "safety mode active". This can be done with a small PLC interlock that monitors the PN interface status bits of both controllers and only enables the drive's main contactor after a 5–10 s grace period.

6.4 Drive-side mitigation: explicit restart

For residual failures after 6.1 and 6.2, a small subset of drives can be wired to a controlled digital output of the standard CPU that toggles a relay cutting 24 V to the affected drive's control electronics. The CPU detects the missing AR with a DPV1 / PN diagnostic interrupt OB and triggers the cycle, then waits a fixed time before allowing normal operation. Limit this technique to small populations because it extends startup time.

7. Firmware Update Considerations

When planning a firmware update across the entire G120C PN population on a 200+ drive site, the following sequence is recommended:

1. Stage the firmware in a single, validated image (e.g. V4.7 SPx or later branch) and store it on the engineering server.
2. Use a campaign-mode update via Starter / Startdrive over the plant network, scheduled for a planned outage window.
3. Verify the GSD / GSDML revision of the drive in the STEP 7 / TIA hardware catalogue matches the firmware level. A mismatch can mask the symptoms or introduce new ones.
4. Perform a controlled cold start after the update, with external 24 V pre-energised and SFC12 sequencing in place, and record the cold-start success rate for at least 10 cycles before declaring the campaign complete.

8. Diagnostic Procedure and Verification

Use the following sequence to confirm a G120C shared-device cold-start issue and to verify a fix.

1. Capture the CPU diagnostic buffer of the standard S7-319 after the cold start. Look for entries containing "IP address could not be allocated", "PROFINET IO failure", and the slot / hardware ID of the missing drive.
2. Capture the device-side diagnostic buffer in Starter / Startdrive for the same drive. Look for PROFINET timeouts and Profisafe diagnostics around the cold-start time stamp.
3. Run ping -t from the engineering station for 60 s after cold start to confirm the drive's IP is still assigned even when the AR is missing.
4. Open "Edit Ethernet Node" and confirm the drive is listed with the configured Profinet name and IP. If yes, the issue is AR-level, not addressing-level.
5. Capture a Profinet frame trace (e.g. with Wireshark and a SPAN port or Profinet TAP) during the first 30 s after cold start. Filter for DCP identify / identify-resp and AR-Request / AR-Response. Confirm whether the second AR is being sent before the drive has finished its first AR state transition.
6. Re-run the cold start with the proposed mitigation (24 V pre-energisation, SFC12 delay, or power-up sequence change) and re-collect the same evidence. The fix is successful when the standard AR is established within the controller's first supervision cycle and no "IP address could not be allocated" entries are produced.

9. Long-Term Architecture Recommendations

For plants where G120C PN drives and a shared S7-319 / 319F pair are likely to remain in this configuration, the following architectural decisions reduce exposure to the issue:

• Avoid mixing the F-AR and standard AR on the same G120C PN device if the application permits. Two physical G120C units - one F-only, one standard - eliminate the shared-device AR race altogether.
• Prefer a 24 V control supply that is independent of the line-side contactor for all PN drives, not only the G120C. The same 24 V pre-energisation also improves the cold-start reliability of S120 and S110 units in some configurations.
• Document the cold-start procedure (pre-energise 24 V, energise CPUs, wait, energise main contactor) and the recovery procedure (cycle 24 V to the affected drive) in the operating manual. With 200+ drives, an undocumented recovery path will be applied inconsistently and will create its own failure mode.
• Schedule a quarterly Profinet health review that includes a controlled cold start with frame capture. The 3 % utilisation figure is not, by itself, evidence of Profinet health; the AR-establishment behaviour is a stronger indicator.

10. Frequently Asked Questions