Redundant Ethernet Connection from a WinCC Flexible Runtime PC Station to SIMATIC S7-400H

This technical reference documents how to engineer a redundant Ethernet path between a WinCC Flexible 2008 Runtime PC Station and a SIMATIC S7-400H fault-tolerant controller. It is written for system integrators and commissioning engineers who need to match the controller-side H-system redundancy with an HMI-side path that survives a single NIC, switch port, or cable failure. The article addresses the documented HWConfig restriction that prevents placing two IE General components inside a single PC station, evaluates Media Redundancy Protocol (MRP) ring topologies using SCALANCE X-200, and clarifies why the IE Softnet RNA package (6GK1711-1EW08-1AA0) cannot be used to expose a virtual dual-NIC adapter for SIMATIC connections.

1. Problem Statement and Engineering Goal

The customer requirement is to deploy a WinCC Flexible 2008 Runtime on an industrial PC that is wired to an S7-400H system with two independent CP 443-1 communications processors. The H-system already provides a switched, redundant backplane (CPU 410-5H or 412/414/417-4H with synchronization module) and two physically separate CP 443-1 modules. The new requirement is to ensure that:

• Loss of one Ethernet cable, one NIC, or one switch port does not interrupt the HMI connection.
• Loss of one CP 443-1 on the S7-400H side does not interrupt the HMI connection.
• The PC station still presents itself to the operator as a single, continuous HMI image with no tag-area gap during failover.

The HWConfig editor of the SIMATIC Manager (Step 7 V5.5 / V5.6) only allows a single IE General instance inside a WinCC Flexible PC station. Attempting to drag a second IE General into the PC rack produces the error Insertion of this object is not permitted because a similar object already exists. Because WinCC Flexible 2008 RT cannot bind to two physical adapters at the OS level through the same S7-connection block, software redundancy must be engineered around the network layer, not the HMI configuration layer.

2. S7-400H Redundancy Architecture Recap

An S7-400H system consists of two redundant CPUs (designated CPU 0 and CPU 1) connected by fiber-optic synchronization links and (optionally) a backup Ethernet sync link. Each CPU owns a CP 443-1 that participates in the H-system's switched S7 connection pool. From the perspective of any single S7 client, the H-system appears as two logical endpoints, but the S7-400H firmware implements a redundant link so that only one logical connection is maintained by the client while the second path is hot-standby.

When the S7-400H detects a failure of the active CP 443-1, the partner CP 443-1 transparently takes over the connection in the H-system's failure-switchover time, which is typically 100 ms to 300 ms for S7 connections. WinCC Flexible RT does not implement S7 H-system client logic in the classical sense; it relies on a single, unidirectional S7 HMI connection, so the failure-switchover time is whatever the H-system and the underlying network deliver.

3. WinCC Flexible 2008 RT Connection Model

WinCC Flexible 2008 Runtime for PC uses the WinCC Flexible Runtime communication driver stack. The relevant connection types are:

For an S7-400H target, the SIMATIC S7 300/400 connection is the only native driver. It is configured in the WinCC Flexible project under Communication > Connections and stores:

• The HMI station name (must match the PC station's Station Configuration Editor name)
• The PLC address (single IP, not a redundant pair)
• The connection resource (TSAP for ISO-on-TCP, port 102 for TCP, slot/rack of the partner CPU)
• The update cycle and timeout behavior

Only one PLC address can be assigned per connection. There is no concept of a primary / secondary IP in this driver; therefore, true NIC-level redundancy cannot be expressed inside the WinCC Flexible connection definition. The redundancy must be implemented at the layer beneath the driver.

4. Why the HWConfig Restriction Exists

The PC station rack in HWConfig represents the SIMATIC Station Manager view of the industrial PC. It is a logical model that the Station Configuration Editor (SCE) and the SIMATIC NET Configuration Console bind to real Windows network adapters. The restriction on a second IE General exists for two reasons:

1. WinCC Flexible 2008 RT does not have a connection engine capable of binding to two different S7-connection resources in parallel. The S7 HMI driver is single-instance per connection.
2. STEP 7 V5.5/V5.6 PC station type "WinCC Flexible RT" is provisioned with one IE General and one Application slot, sized for a single OPC server / HMI runtime instance.

Attempts to use a custom PC station type with two IE General slots and route WinCC Flexible to the second slot fail because the WinCC Flexible project only references the default HMI application slot, and the SCE only allows one active binding per application slot per IP family.

5. The Recommended Topology: MRP Ring with SCALANCE X-200

The accepted Siemens solution for redundant PC-to-PLC Ethernet is a Media Redundancy Protocol ring (MRP, IEC 62439-2). A SCALANCE X208 (or any X-200/XB-200/XC-200 series that supports MRP) is configured as the MRP Manager, and the two SCALANCE switches at the S7-400H side, the two NICs of the PC station, and the rest of the ring act as MRP Clients. The ring closes on itself, so any single break in the ring is detected and the MRP Manager re-routes traffic on the alternate path within < 200 ms (default MRP reconvergence).

5.1 Required hardware and software

5.2 Logical ring layout

The physical ring order is not important, but the logical role assignment is. Choose one SCALANCE X208 as the MRP Manager; the remaining SCALANCE devices (and the two managed NIC ports of the PC) are configured as MRP Clients. The MRP Manager blocks one of its two ring ports to break the loop during normal operation; that blocked port is automatically unblocked when the ring is interrupted.

A common placement is:

1. MRP Manager SCALANCE X208 sits in the control cabinet next to the PC station.
2. Ring port 1 of the Manager connects to NIC 1 of the PC station.
3. Ring port 2 of the Manager connects to SCALANCE X-200 "A" at the H-rack.
4. SCALANCE X-200 "A" connects to CP 443-1 of CPU 0.
5. SCALANCE X-200 "B" (sitting at the H-rack) connects to CP 443-1 of CPU 1.
6. SCALANCE X-200 "B" closes the ring back to NIC 2 of the PC station, which in turn connects back to ring port 1 of the Manager through an additional segment, completing the loop.

6. Step-by-Step Configuration

6.1 Configure the SCALANCE X208 as MRP Manager

1. Connect a service PC directly to the SCALANCE X208 with a CAT 6 patch cable, set the service PC NIC to 192.168.0.10/24.
2. Browse to https://192.168.0.1 (default SCALANCE X208 IP). Default credentials are admin / admin; change immediately per the SCALANCE X-200 Web Configuration Manual.
3. Navigate to Layer 2 > MRP.
4. Set MRP Role = Manager.
5. Set Ring Port 1 and Ring Port 2 to the two ports that form the ring (e.g., port 1 and port 2).
6. Set Reaction on link change = Default (50 ms test frame interval).
7. Click Set Values and verify that the Status field reports Active.

6.2 Configure SCALANCE X-200 ring clients

1. Repeat for each SCALANCE on the ring. Set MRP Role = Client and assign the two ring ports.
2. Verify that the Web UI on each Client shows Ring state = Open when the Manager is reachable and the ring is healthy.

6.3 Configure the PC station NICs

The PC station's two physical NICs participate in the ring as MRP Clients only at the network layer; they are plain Windows network adapters. Configure their TCP/IP settings as follows:

Two NICs in the same IP subnet is supported on Windows when the application uses the bound interface. However, Windows' automatic metric feature will route all outbound traffic through the lower-metric NIC, which is exactly what is wanted: in normal operation, traffic flows through NIC 1; on a NIC 1 link loss, the SCALANCE MRP Manager unblocks the alternate ring path within < 200 ms and the kernel continues to attempt routing through the remaining NIC.

6.4 Configure HWConfig on the PC station

1. Open SIMATIC Manager and the PC station project that mirrors the WinCC Flexible project.
2. In HWConfig, insert exactly one PC Station > User Application > WinCC Flexible RT and one PC Station > CP > IE General.
3. Bind the IE General to the primary NIC (192.168.0.10) using the Station Configuration Editor on the PC. Compile and download the PC station to the local PC.
4. Insert the S7-400H rack, both CPUs, and the two CP 443-1 modules. Configure each CP's Ethernet interface and IP (e.g., 192.168.0.20 for CPU 0 CP, 192.168.0.21 for CPU 1 CP). Configure S7 connection properties for the H-system.
5. Save, compile, and download to both CPUs.

6.5 Configure the WinCC Flexible connection

1. In the WinCC Flexible 2008 project, open Communication > Connections and add a connection of type SIMATIC S7 300/400.
2. Set the HMI station to the PC station name created in step 6.4.
3. Set the PLC address to the active IP of the H-system. The S7-400H presents a single H-system address (192.168.0.20 in the example); the redundant partner is managed by the H-system firmware.
4. Set the TSAP to 03.02 for the H-system's slot 2 default or to the configured PG/OP slot. Consult the S7-400H System Manual for slot mapping.
5. Set the update cycle to a value appropriate for the process; 1 s is typical for tag polling on a large HMI project. Lower values (250 ms) increase CPU and network load.

7. Verification

After commissioning, run the following checks in order. Each step is non-destructive and can be performed while the HMI is in service.

1. Ring health. On the SCALANCE MRP Manager Web UI, confirm Ring state = Closed and that the manager's blocked port is the expected one.
2. Both CP 443-1 reachable. From the PC, ping 192.168.0.20 and ping 192.168.0.21 should both succeed.
3. WinCC Flexible RT status. Open the WinCC Flexible RT, navigate to Tools > Status, and confirm the connection shows Connected and the cycle time matches the configured value.
4. Link-loss drill 1 - cable pull. Disconnect the patch cable between NIC 1 and the SCALANCE Manager. Within 200 ms the Manager should unblock the alternate port. WinCC Flexible RT should not lose connection (the H-system re-establishes via the partner CP 443-1 within the configured S7 timeout, typically 5 s default).
5. Link-loss drill 2 - CP failure. Power down the CP 443-1 of CPU 0. The H-system switches the active connection to CPU 1's CP 443-1. WinCC Flexible RT should not log a connection loss if the S7 timeout is set to > 300 ms.
6. Link-loss drill 3 - switch failure. Disconnect power to SCALANCE X-200 "A". The MRP Manager detects the ring break and unblocks. The PC station's traffic now flows through NIC 2 to the surviving switch, on to CP 443-1 of CPU 1, and into the H-system.

8. Alternative: Two Independent WinCC Flexible Connections (Soft Redundancy)

If a SCALANCE-based ring is not available, you can implement soft redundancy at the application layer. Define two SIMATIC S7 300/400 connections in the WinCC Flexible project: one targeting the H-system's primary IP, one targeting the partner IP. Tag polling is then performed over both connections, and the WinCC Flexible "change connection" function on the Connections dialog can be used to swap the active connection manually.

The drawback is that this is not seamless failover. The "change connection" function is a user action, not an automatic event, and WinCC Flexible will report a brief loss of connection in the RT log during the swap. For most operator-HMI use cases this is acceptable; for closed-loop control or safety-related visualization it is not.

9. Why IE Softnet RNA Is Not Released for SIMATIC

The IE Softnet RNA package (catalog number 6GK1711-1EW08-1AA0, successor to 6GK1711-1EW62-3AA0) installs a virtual NDIS miniport that bonds two physical NICs into a single logical interface with failover. While the product is functional, Siemens does not release IE Softnet RNA for use with SIMATIC S7-400H communications, and it is not supported with WinCC Flexible RT or WinCC RT. Reasons cited in the release notes are:

• The RNA virtual adapter does not implement ISO-on-TCP TSAP-based connection binding the way a physical CP does, breaking the S7 connection lookup inside the H-system.
• RNA's failover logic can change the source MAC and IP within a connection's lifetime, which is not tolerated by the H-system's connection tracking.
• RNA's interaction with the SCE and SIMATIC NET Configuration Console is undefined and can lead to a stuck "Connection not established" status on the WinCC Flexible RT.

Therefore, RNA must not be used. The MRP ring path or the two-connection soft-redundancy path are the supported alternatives.

10. PC Station Network Adapter Configuration Considerations

Modern Windows versions handle multi-NIC, same-subnet configurations more gracefully than Windows XP, but the engineer should still verify the following settings on each adapter participating in the ring:

• Disable TCP/Offload.ipv4 checksums and large send offload (LSO) on the SIMATIC traffic NICs. LSO can fragment S7 PDUs in a way that the S7-400H interprets as protocol errors under high tag counts.
• Disable Energy Efficient Ethernet (EEE, IEEE 802.3az) and Green Ethernet in the NIC advanced properties. EEE can suspend the link during idle periods, which the S7 keep-alive interprets as a link loss and triggers spurious reconnections.
• Disable Windows Firewall on the SIMATIC network adapters, or open the required ports: TCP/102 (ISO-on-TCP/RFC1006), UDP/34964 (PN-DCP), and any OPC DA DCOM range used.
• Verify that the SIMATIC Station Configuration Editor has the Station Name and the IE General binding correct. The SCE must be closed during HWConfig compile / download or the SCE will not pick up the new PC station name.

For background on the Windows multi-NIC behavior, see the Microsoft Q&A thread 2 separate ethernet connections into 1 desktop, which explains the automatic metric, binding order, and route-table implications of having two same-subnet adapters on one Windows host.

11. Troubleshooting Matrix

12. S7 H-System Error Codes Relevant to Redundancy

The H-system diagnostic buffer emits a small set of well-known event IDs when the redundancy state changes. Engineers commissioning the link should know them:

13. Verification Checklist Before Sign-Off

• MRP Manager SCALANCE X208 reports Ring state = Closed and shows the correct blocked port.
• All SCALANCE Clients report Ring state = Open in steady state.
• Ping to both CP 443-1 IPs succeeds with < 1 ms latency on the management VLAN.
• WinCC Flexible RT shows connection Established with no faults in the last 24 hours of operation.
• Three link-loss drills (cable pull, CP power-down, switch power-down) all recover within the documented S7 timeout.
• Event IDs 0x73A1, 0x73A2, 0x73B1, 0x73B2 have been observed in the H-system diagnostic buffer and are documented in the log book.
• PC station's two NICs are bound to the right Windows metric and have EEE / LSO disabled.

14. Long-Term Operational Notes

The MRP ring solution has been in service at scale since the SCALANCE X-200 firmware 4.x days and is well understood. Firmware 6.x added HRP (High-speed Redundancy Protocol) and PRP (Parallel Redundancy Protocol, IEC 62439-3), neither of which is required for an S7-400H integration. Do not upgrade SCALANCE X-200 firmware to a version newer than the one certified for your STEP 7 / WinCC Flexible version without re-running the full verification matrix. PRP requires Layer 3 capable switches and LRE (LAN Redundancy Entity) firmware on the PC side, which WinCC Flexible 2008 does not support.

Finally, document the IP plan, the SCALANCE role assignments, and the failover scenarios in the cabinet's wiring diagram. The next engineer to service the panel will thank you.