Troubleshooting
Why WinCC Flexible Shows a Blank Station Search Over MPI
Short answer
A blank station field in WinCC flexible means the S7ONLINE programming path reaches the CPU but the HMI connection has no defined route. Give every MPI node a unique address — PG 0, panel 1, CPU 2, a second PC 3 — confirm the USB adapter supports OP as well as PG connections, and expect STEP 7 symbols only through an integrated HMI station.
On this page
A blank station field in WinCC flexible's connection editor is not a broken search: the editor lists only the stations reachable through the path selected for the HMI connection, and when the PC talks to the CPU through a programming adapter, that path has never been defined. The S7ONLINE access point that carries your STEP 7 online session is the programming route, not the HMI-to-PLC route, so a healthy PG connection and an empty station list can coexist on the same desk. The fix is a committed plan: give every MPI node a unique number — programming PC 0, panel 1, CPU 2, a second PC 3 — confirm the adapter carries operator (OP) traffic as well as programming (PG) traffic, and integrate the HMI station into the STEP 7 project if the symbol table matters.
Two paths run through one adapter #
STEP 7 goes online through the access point S7ONLINE: engineering PC, adapter, MPI cable, CPU port. That session proves the PC reaches the bus. The HMI connection is configured separately — an HMI station created in SIMATIC Manager owns a connection entry with a local endpoint, the panel or the PC running WinCC flexible RT, and a PLC endpoint, the CPU. That entry is what connects a PC runtime or an operator panel to the controller, and a station search is a view of it and nothing else. It cannot borrow results from the programming session.
When the HMI connection borrows the programming adapter and its S7ONLINE access point, the editor has no route of its own to walk, so nothing is listed while STEP 7 still finds the CPU. That is the exact combination in this failure: online access works, the field stays blank.
Watch the two physical networks too. The adapter enters the CPU's MPI port; a panel wired to the CPU's DP port sits on the PROFIBUS DP subnet, which the MPI adapter does not see at all, no matter how tidy the cables at the CPU look.
The address rules the plan rests on #
MPI hands out factory defaults: the programming device takes 0, the operator panel takes 1, the CPU takes 2; a fourth master, such as a second PC running WinCC flexible RT, takes 3. The default number range is 0 to 31, and every node on the subnet shares the bus settings — transmission rate, 187.5 kbit/s by default, and highest address, 31. A PC joining an existing network has to adopt that network's settings instead of its own, or it will not find a single station.
A number typed into a project never changes a device. The CPU carries its address in its interface configuration, the adapter's node is the one entered in Set PG/PC interface, and a runtime transmits through the node of the physical interface it actually uses — on the programming PC that is the adapter's own number, never a second one. One interface, one number.
Take the machine out of automatic before downloading an interface change to the CPU: the new address takes effect immediately, and every connected station drops until the HMI side is updated to match.
Worked example: numbering the bus #
The bus here carries four stations: the programming PC with its USB adapter, WinCC flexible RT running on that same PC, a panel to be connected after commissioning, and the CPU. A second PC is planned but not yet on the cable. The table assigns each station and names where its number is actually held.
| Station on the subnet | Node number | Where the number is held |
|---|---|---|
| Programming PC with USB adapter | 0 | Set PG/PC interface |
| WinCC flexible RT on that same PC | 0, shared with the adapter | the runtime's interface settings |
| Operator panel, connected after commissioning | 1 | the panel's own communication settings |
| CPU, MPI interface | 2 | the CPU's interface configuration |
| Second PC, planned | 3 | that PC's Set PG/PC interface |
Uniqueness check: the numbers in force are 0, 1, 2 and 3 — the runtime adds no number because it transmits through the PC adapter — so 4 distinct numbers serve 4 stations with no duplicates. The default range offers 32 numbers, 0 through 31, leaving 32 − 4 = 28 free for future nodes. With every interface holding a different number, a search through a defined HMI route should list the CPU at 2; if it still shows nothing, the fault sits in the route definition or the adapter's roles, not in arithmetic.
Confirm what the adapter can carry #
The older serial PC adapters were sold in separate PG and OP variants — one for programming sessions, one for operator-panel traffic — so going online in STEP 7 has never been evidence that a unit carries HMI communication. The PC Adapter USB is documented for attachment to MPI, PPI and PROFIBUS (DP) networks, but the connection roles of the specific variant in hand still need confirming in its own documentation before node numbers get blamed.
If the runtime path through the adapter turns out not to work, the targeted correction is a bus interface in the runtime PC — a communication processor of the CP5611 class — not a further round of address edits.
Why the symbol list stays empty #
Communication and symbol availability fail separately. WinCC flexible can exchange values with the CPU while offering none of the STEP 7 symbols, because a connection running through the programming adapter gives the project no integrated link between the HMI station and the S7 program for the symbol table to flow through. The pragmatic path under time pressure is to type the tags by hand; the runtime runs fine on manual tags.
To get the symbols back, create the HMI station inside the STEP 7 project, connect it to the S7 program, and put the CP5611-class bus interface in the runtime PC in place of the programming adapter — the integrated station is what the symbol browser reads. This integration behavior belongs to the classic WinCC flexible chain; the hosting and remote-access choices around WinCC Unified are a different subject, covered in Where WinCC Unified Runtime Can Run and Remote HMI Options.
Decision path from blank field to working runtime #
Work the branches in this order and stop at the first that fails:
- STEP 7 cannot go online. The fault is the bus or the PC interface — adapter seating, cable, CPU MPI port, wrong entry in Set PG/PC interface. Chasing the station field before this works wastes the afternoon.
- STEP 7 goes online, search still blank. Confirm an HMI station exists inside the STEP 7 project and its connection entry names both endpoints: the local runtime or panel, and the CPU at its MPI address. No defined route means nothing for the search to list.
- Runtime holds a number of its own. On the programming PC it transmits through the adapter's node; a separate PC takes the next free number with its own interface.
- Bus settings disagree between nodes. Match transmission rate and highest address across the subnet; a PC joining an existing network adopts that network's settings.
- Adapter roles unconfirmed. Check OP support in the adapter's documentation; if the variant cannot carry it, plan the CP5611-class interface.
- Symbols required. Integrate the HMI station with the S7 program and use the bus interface; otherwise stay with manual tags.
- Before the panel goes on the wall, exercise the configured connection against the live CPU the way Testing HMI Simulation Against a Live PLC sets out, so the first download is not the first test.
Questions asked
What does the only-master option on an MPI connection mean?
MPI is a token-passing bus where the PG, the HMI and the CPU are all masters that pass permission to transmit between them. The checkbox, labelled along the lines of 'Address is the only Master', tells your station that no other master exists, so it never waits for the token. On a bus that also carries a PG, panel or CPU it must stay unchecked; with it set, the station transmits without taking its turn and the station search or reads fail.
Can the same adapter transfer the finished project to the panel?
Project transfer over MPI or PROFIBUS to a Windows-based panel is a documented use of that path, with one condition: when the configuration PC joins an existing network it has to take that subnet's bus settings, the transmission rate and the highest node address, instead of its own. A PC left on its standalone bus parameters will not find the panel even though the cable is seated.
Why does the panel on the DP port never appear in the station search?
The adapter plugs into the CPU's MPI interface; a panel hanging off the DP port lives on the PROFIBUS DP subnet, a separate network with its own addresses and bus parameters. Two cables meeting at one CPU do not merge the networks. To reach that panel from the PC, the PC interface has to join the DP subnet at its transmission rate, or the station has to be reached through the CPU by an explicitly configured route.
Can more than one HMI share the MPI subnet with the PG?
MPI permits several masters, so a PG, more than one panel and the CPUs can share one subnet; each station still needs its own number in the default range of 0 to 31. A workable sequence is PG 0, first panel 1, CPUs 2 and upward, further panels or PCs 3 and upward — the exact numbers matter less than uniqueness. Every added master joins the token rotation, so access times grow with each station on the bus.