Siemens S7-300 MPI Redundant CPU Network with OP270 Configuration Reference
This reference describes how to design, configure, and troubleshoot a Siemens S7-300 MPI (Multi-Point Interface) network that carries redundant CPU pairs (master/reserve) and multiple OP270/OP370 operator panels. It is written for commissioning engineers who must validate that a four-CPU redundant cluster, two HMI panels, and an engineering PC coexist on a single 187.5 kbps MPI segment without losing tag updates.
1. Reference Topology and Operating Conditions
The canonical problem layout documented in field service reports is:
PC ---- CPU315-2DP (Master A) ---- CPU315-2DP (Reserve A) ---- OP370 ---- OP370 ---- CPU315-2DP (Master B) ---- CPU315-2DP (Reserve B)
Operating parameters assumed by the troubleshooting flow:
| Parameter | Value | Notes |
|---|---|---|
| Physical layer | RS-485, two-wire, shielded twisted pair | PROFIBUS cable (purple), 9.6 kbps to 12 Mbps supported |
| Active baud rate | 187.5 kbps | Maximum MPI profile on legacy S7-300 |
| Bus topology | Linear, terminated at both ends | No spurs; stubs only with repeaters |
| Number of S7-300 nodes | 4 (two redundant pairs) | CPU 315-2DP (6ES7 315-2AFxx / 6ES7 315-2AGxx) |
| HMI nodes | 2 (OP270 / OP370 / MP370) | OP/TP/MP product line, WinCC flexible 2004/2005/2007 |
| Engineering PC | 1 (STEP 7 V5.x) | PG/PC interface = PC Adapter (MPI) or CP 5611 |
| Default MPI address | 2 (CPU), 0/1 (HMI), 0 (PG) | See MPI network configuration (S7-300, S7-400, PC) |
2. MPI Physical Layer and Cable Rules
MPI is a shielded two-wire RS-485 variant that shares the PROFIBUS-DP cable part numbers. For redundant CPU applications, the cable plant must satisfy three constraints simultaneously:
- Termination: Exactly two terminations must be enabled network-wide. The standard 9-pin D-sub PROFIBUS connector ships with an on-board switch; leave it OFF on every intermediate node and ON only at the two physical ends of the segment.
- Shield bonding: Bond the braided shield to functional earth at both cable ends (large surface area clamp) to dissipate high-frequency noise from the variable-frequency drives that typically share the cabinet.
- Baud rate versus segment length: 187.5 kbps supports up to 1,000 m without repeaters. Lower rates (19.2 kbps, 187.5 kbps) tolerate the same total length, but the cycle time grows linearly with the number of slaves that must be polled.
| Baud rate | Max segment length (no repeater) | Typical use |
|---|---|---|
| 19.2 kbps | 1,200 m | Legacy HMI-only networks |
| 187.5 kbps | 1,000 m | Default MPI / redundant S7-300 |
| 1.5 Mbps | 200 m | CPU-to-CPU global data, fast HMI |
| 12 Mbps | 100 m | PROFIBUS-DP only (not standard MPI) |
3. CPU 315-2DP Hardware Profile Relevant to MPI
The CPU 315-2DP used in the reference design is the workhorse of redundant S7-300 cells. The two part families most commonly encountered in service are:
| MLFB (Order number) | Work memory | MPI/DP ports | FW compatibility |
|---|---|---|---|
| 6ES7 315-2AFxx-0AB0 | 64 KB code / 64 KB data | MPI + DP master | FW V2.0 and later |
| 6ES7 315-2AG10-0AB0 | 128 KB code / 128 KB data | MPI + DP master | FW V2.1 and later |
| 6ES7 315-2AH14-0AB0 | 192 KB code / 192 KB data | MPI + DP master | FW V3.x |
The CPU 315-2DP exposes a single PROFIBUS-DP physical port that doubles as the MPI interface. Configuring the port is performed in STEP 7 / HW Config:
- Interface: MPI/DP
- Type: MPI
- Address: 2 (default), 3..31 for any additional CPU
- Highest MPI address: 31 (default; can be raised to 126)
- Transmission rate: 187.5 kbps (must match every other node)
Default MPI address is 2 for any S7-300/400 CPU out of the box, per the official TIA Portal reference: MPI network configuration (S7-300, S7-400, PC). Renumber before placing a second CPU on the segment or both will answer to address 2 and behave as a single device to the HMI.
4. OP270 and OP370 HMI Connection Limits
OP270 and OP370 panels are configured with WinCC flexible 2004 SP1 or later (later replaced by TIA Portal / WinCC Comfort for the newer TP/Comfort line). The maximum number of simultaneous S7-300/400 connections supported by these operator panels is:
| Panel family | Max S7 connections | Notes |
|---|---|---|
| OP270 (6-inch monochrome) | 6 | 4 connections in many field examples; 6 is the documented ceiling |
| OP370 (10-inch monochrome) | 6 | Same WinCC flexible runtime engine as OP270 |
| MP370 (10-inch color) | 6 | Identical connection limit; color display only |
| TP270 / TP370 | 6 | Same generation |
Therefore a single OP270 can be configured to poll up to six CPU 315-2DP nodes. The reference topology (4 CPUs + 2 HMIs + 1 PG) is well within the HMI connection budget, but the connection budget is per-HMI, not shared.
5. Redundant CPU Topology on a Single MPI Segment
Two distinct redundancy models are compatible with the 187.5 kbps MPI segment described here:
- Software redundancy (SWR): The most common model on S7-300. Two CPU 315-2DP units share I/O via DP slave ET 200M. One runs as master, the other as reserve. SWR V2.x supports up to four redundant pairs in a single rack pair, with the redundancy link carried over MPI or PROFIBUS.
- H-system (1+1): Implemented with a CPU 417H in the S7-400 family. The H-system synchronizes over redundant fiber or PROFIBUS links and is not used with CPU 315-2DP. It is included here only to prevent mis-application.
For the S7-300 / CPU 315-2DP case, the redundancy data exchange (link status, swap-over handshake) is transported by the same MPI bus that also carries the HMI traffic. This is the root cause of the latency symptom reported in the field and is the focus of the troubleshooting flow below.
6. Configuring the "Change Connection" Function in WinCC flexible
The OP270/OP370 runtime supports a built-in Change Connection function. The function allows one or more configured S7 connections to be defined as standby; the HMI will use the active connection for tag polling and switch to the standby if the master fails to respond within the configured timeout.
Configure it as follows in WinCC flexible:
- Open the project, select the panel, open Connections editor.
- Create one connection per master CPU: Connection_1 = Master A, Connection_2 = Master B.
- Right-click each connection and select Properties > Change Connection and assign the role of Backup to the connection that targets the redundant partner.
- For the master connection, leave Cyclic Operation = Enabled (default).
- For the backup connection, set Only one Master (Disabled) and Cyclic Operation = Enabled so that the panel warms the link in the background and re-routes the moment the master stops responding.
- Distribute tags across both connections. Do not place all tags on Connection_1 only; the HMI will not see the change-of-master event for tags that are bound solely to the disabled master.
- Compile, download to the panel, and restart it so that the runtime table is reloaded.
Two parallel Change Connection functions are required when both HMI panels are present. Each HMI must own its own pair of connections; the connections are local to the runtime, not global to the segment.
7. Why the HMI Loses Tags on a 4-CPU Segment
The 187.5 kbps MPI segment is a token-passing bus with a fixed token rotation time that depends on the number of active masters and the number of slaves each master services. The 4-CPU + 2-HMI + 1-PG configuration forces every device to wait for its turn, and the SWR redundancy handshake consumes additional token slots because the reserve CPU is also participating in the link-up exchange.
The dominant failure modes are:
| Symptom | Likely root cause | Diagnostic |
|---|---|---|
| HMI stops refreshing on any of the 4 CPUs | Token rotation exceeded the WinCC flexible watch-dog (default 5 s) | STEP 7 > PLC > Monitor/Modify > Cycle Time / Communication Load |
| HMI refreshes only after one CPU pair is removed | Bus is no longer overloaded; token returns in time | Use split-and-add method (see Section 10) |
| Specific tags are stale, others fresh | Tags were bound to a single connection whose master lost a token slot | WinCC flexible > Tag diagnostics; check the connection status of each tag |
| All HMI communication is lost on the full segment | Duplicate MPI address, missing termination, or a passive tap that shorts the bus | Disconnect nodes one at a time until the segment recovers |
8. STEP 7 Configuration: MPI Addresses, Highest Station, and Profile
Use the following parameter set as the starting baseline. Adjust Highest MPI address only if you need to bring segment-scanning tools online.
| Node | MPI address | Role | Notes |
|---|---|---|---|
| CPU 315-2DP Master A | 2 | SWR Master, even pair | Default address retained |
| CPU 315-2DP Reserve A | 3 | SWR Reserve, even pair | Reserve must be one address above its master |
| OP270 / OP370 (left) | 4 | HMI | Default HMI address is 1; renumber to 4 to avoid clashing with PG |
| OP270 / OP370 (right) | 5 | HMI | Second HMI |
| CPU 315-2DP Master B | 6 | SWR Master, odd pair | Address 6, not 2 |
| CPU 315-2DP Reserve B | 7 | SWR Reserve, odd pair | Address 7 |
| STEP 7 PG | 0 | Programming device | Default PG address is 0 |
STEP 7 procedure:
- Open SIMATIC Manager > Project > Open > yourproject.
- HW Config > select the CPU 315-2DP rack > double-click the MPI/DP interface > Properties.
- Set Address, Transmission rate = 187.5 kbps, and Highest MPI address = 31.
- Save & Compile and download to the CPU.
- Repeat for every CPU, ensuring no two nodes share the same MPI address.
- Open Set PG/PC Interface on the engineering station and bind to PC Adapter (MPI) or CP 5611 (MPI).
9. Bus Termination Rules
The MPI segment is terminated by a resistor network integrated in the PROFIBUS connector. The bus is electrically correct only when the two physical ends of the segment are terminated and every other connector is left unterminated.
- Enable termination on the connector at the first node (left end of the segment) and on the connector at the last node (right end of the segment).
- Disable termination on all intermediate connectors; this is the most common field error.
- If a repeater (6ES7 972-0AA01-0XA0) is used, terminate the two segments individually; the repeater provides its own bias but not its own termination.
- Measure with an ohmmeter across pins 3 and 8 of any connector with the bus unpowered: the value must be approximately 220 Ω only at the two ends, infinite (open) everywhere else.
10. Step-by-Step Commissioning Procedure
- Inventory the hardware. Verify part numbers, firmware versions, and that the two CPUs of each redundant pair are the same MLFB. SWR will not pair a 6ES7 315-2AF03 with a 6ES7 315-2AG10 without rejecting the reserve.
- Pre-stage the bus. Lay the cable, terminate both ends, connect shield to earth at both ends.
- Bring up the engineering PG. Connect the PG with address 0 to the segment; use Accessible Nodes from STEP 7 to confirm a clean 1-node segment.
- Add the first redundant pair. Connect Master A (address 2) and Reserve A (address 3) one at a time. Verify both are visible from the PG after each insertion.
- Add the second redundant pair. Connect Master B (address 6) and Reserve B (address 7) the same way.
- Add the first HMI. Use the OP270's service menu to assign MPI address 4. After restart, verify it is visible from the PG.
- Add the second HMI. Assign address 5. Verify visibility.
- Load the WinCC flexible project. On each HMI, transfer the compiled runtime. The runtime will start to poll all four connections.
- Verify the token round-trip. Use Accessible Nodes from the PG. The four CPUs and the two HMIs must all appear without timeouts.
- Stress test the redundancy. Stop the Master A CPU. The Reserve A CPU must take over within the SWR configured switch-over time. The HMI must continue to refresh tags through Connection_1 (Master A > Reserve A path) within the WinCC flexible watch-dog.
- Capture the bus statistics. Record cycle time, communication load, and token rotation time on every CPU. These are the baselines used by the troubleshooting matrix below.
11. Diagnostic and Verification Checklist
| Check | Method | Pass criterion |
|---|---|---|
| No duplicate MPI address | STEP 7 > Accessible Nodes | Exactly one node per configured address |
| Termination correct | Ohmmeter across pins 3-8 | 220 Ω at two ends, open everywhere else |
| Shield bonded | Visual + continuity to PE | Low impedance to PE at both ends |
| Cycle time on every CPU | STEP 7 > PLC > Diagnostic | Below the OB1 maximum set in CPU Properties |
| Communication load | CPU Properties > Communication | Below 50 percent on all four CPUs |
| HMI tag update | WinCC flexible > Tag diagnostics | All four connections report update within watch-dog |
| SWR switch-over time | Force stop on master, watch reserve | Switch-over completes within 100 ms (SWR V2.x default) |
| Token rotation | STEP 7 > NCM Diagnostics | Token rotation < watch-dog / 2 |
12. Troubleshooting Matrix
| Symptom | Priority check | Action |
|---|---|---|
| HMI shows "connection failed" on one of the 4 CPUs | 1. Address conflict 2. SWR partner address 3. Termination | Re-scan with Accessible Nodes. Re-number the offending CPU. Verify its SWR partner is the correct one. |
| HMI shows "connection failed" on all 4 CPUs | 1. Cable break 2. Termination 3. Baud rate mismatch | Power down the segment, measure resistance end-to-end. Re-enable termination only on the two physical ends. Match every node to 187.5 kbps. |
| HMI refreshes only when one pair is removed | Communication load | Reduce tag count per connection. Split the two HMI connections across two physical segments using a repeater. |
| Tags go stale for 5-10 seconds then recover | Watch-dog vs token rotation | Increase the WinCC flexible watch-dog to 10 s only as a temporary measure; do not leave in production. Root-cause is bus load. |
| One CPU drops off the bus intermittently | Connector seating, shield | Re-seat the D-sub. Confirm the connector screws are torqued. Re-bond the shield. |
| SWR switch-over does not occur | SWR partner address, SWR license | Confirm the master is at address N and the reserve is at address N+1. Confirm the SWR option is licensed on both CPUs. |
13. Migration Path: When MPI Is No Longer Sufficient
When the redundant pair, the HMI, and the engineering PG all share a single 187.5 kbps segment, the bus can become the bottleneck for any process that requires sub-second tag updates. The recommended escalation paths are:
- PROFIBUS-DP at 1.5 Mbps for the CPU-to-CPU SWR link. Re-purpose the MPI/DP port as PROFIBUS-DP on each CPU, set address 2..7, and configure SWR over DP. The HMI traffic can stay on MPI or migrate to DP at the same time.
- Industrial Ethernet (CP 343-1 Lean) on each CPU for both SWR and HMI traffic. Move the HMI connections to PROFINET. Use CP 343-1 (6GK7 343-1EX11) on each S7-300. Baud-rate limitations vanish; the only constraint is CP 343-1 cycle time.
- Add a second CP 343-1 to each CPU to physically separate SWR and HMI traffic across two Ethernet subnets. This is the standard solution for high-availability S7-300 cells that exceed the original 187.5 kbps design.
14. Field-Proven Caveats
- OP270 versus OP370. The OP270 is a 6-inch monochrome panel, OP370 is a 10-inch monochrome panel. Both share the same WinCC flexible runtime engine. Always confirm the exact model before assuming the connection budget.
- Default address collisions. Four CPU 315-2DP shipped from the factory all default to address 2. Always renumber at the bench before installing the second CPU in the network. This is the single most common cause of "only two of the four CPUs respond."
- Change Connection tag distribution. When Change Connection is enabled, every tag must be bound to a connection that is part of the active group. Tags bound only to a disabled connection will be reported as "connection failed" even though the bus itself is healthy.
- Cyclic Operation on the disabled connection. Leave it Enabled so that the HMI warms the link; otherwise the first swap-over will stall for the connection-establishment time.
- Token rotation vs watch-dog. The WinCC flexible default watch-dog is 5 s. A 187.5 kbps segment with 8 active devices can exceed 1 s of token rotation; plan for the worst case and shorten the HMI tag list before extending the watch-dog.
15. Quick Reference: Parameter Defaults for a 4-CPU + 2-HMI + 1-PG MPI Segment
| Setting | Value |
|---|---|
| Default MPI address (CPU) | 2 |
| Recommended CPU addresses | 2, 3 (Pair A); 6, 7 (Pair B) |
| Recommended HMI addresses | 4, 5 |
| PG address | 0 |
| Highest MPI address | 31 (default) up to 126 |
| Transmission rate | 187.5 kbps |
| Max active nodes per segment | 8 (use 6ES7 972-0AA01-0XA0 repeater above 8) |
| Max addressable nodes per subnet | 126 |
| Termination | ON at the two physical ends only |
| WinCC flexible connections per OP270/OP370 | 6 (use Change Connection for swap-over) |
| SWR redundancy link | MPI (this design) or PROFIBUS-DP |
For the official Siemens reference on the defaults, address counts, and the active-node limit, consult the TIA Portal / STEP 7 documentation: MPI network configuration (S7-300, S7-400, PC).
How many CPU 315-2DP nodes can one OP270 poll over MPI?
The OP270 and OP370 both support up to 6 S7 connections in WinCC flexible. The four-CPU redundant layout is therefore well within the connection budget, but tag refresh will degrade if the 187.5 kbps segment is heavily loaded by SWR redundancy traffic.
What is the default MPI address of a CPU 315-2DP and why does it matter?
Every S7-300/400 CPU ships with MPI address 2. If two or more CPUs are connected to the same segment without renumbering, both will respond to address 2 and the HMI will see only one device. Renumber each CPU to a unique address (2, 3, 6, 7 in the reference design) before placing it on the segment.
Why does the OP270 lose tags on a 4-CPU MPI segment and recover when two CPUs are removed?
Token rotation time grows with the number of active MPI peers and with the SWR redundancy handshake. With four CPUs plus two HMIs plus a PG, the token round-trip can exceed the WinCC flexible watch-dog (default 5 s). Removing one CPU pair shortens the round-trip below the watch-dog and the HMI resumes refreshing.
How is the Change Connection function configured for a redundant CPU pair on OP270?
Open the Connections editor in WinCC flexible, create one connection per master CPU, and mark each partner connection as "Backup" under the Change Connection property. Distribute tags across both connections, leave Cyclic Operation = Enabled on the backup, and download the project to the panel.
When should I migrate the MPI segment to PROFIBUS or Ethernet?
Migrate when the HMI tag list grows above roughly 200 tags per panel, when the cycle time on any CPU exceeds the OB1 budget, or when SWR switch-over delays are observed. The minimal change is to switch the CPU MPI/DP port to PROFIBUS-DP at 1.5 Mbps; the more robust change is to add a CP 343-1 to each CPU and move both SWR and HMI traffic to PROFINET.