Troubleshooting
WinCC receives no data from an S7 PLC: trace the path
Short answer
There is no fixed interval at which an S7 PLC sends data to WinCC: WinCC polls each external tag at the acquisition cycle configured on that tag, and the CPU only answers. When no data arrives, trace the driver, the connection and the tag mapping in that order, beginning with the running project, and fix the connection before touching tag addresses.
On this page
An S7 PLC never sends WinCC anything on a schedule of its own. WinCC asks: the communication driver polls each external tag at the acquisition cycle configured on that tag, and the CPU only answers. So "no data from the S7" is never a missing broadcast to hunt for in a network diagram — it is a fault in one of three layers: the driver, the connection, or the tag mapping. Trace those layers in that order, beginning with the project that is actually running, and settle the connection before you touch a single tag address. This article gives you the trace for a CPU of the S7-400 class such as the S7-414H, a worked example with the decision rule that separates a connection fault from a tag fault, and the server and client roles that decide how WinCC and OPC exchange data.
Why there is no fixed S7-to-WinCC interval #
Cyclic HMI communication is request-driven. The driver in WinCC issues the read, the CPU answers it, and the pace of those reads is the acquisition cycle set on each external tag — not one interval the controller broadcasts to every HMI. Asked for "the interval at which the S7 sends data to WinCC", the honest answer is that no such number exists to look up: the interval lives in the WinCC project, tag by tag, and two tags on the same connection can run at different cycles.
The same model explains slow screens. The time from a value changing in the CPU to it changing on the display is the CPU scan that produces the value, plus the tag's acquisition cycle, plus communication processing on both ends, plus the screen refresh. A value trailing the process by about one cycle is behaving as configured; every tag slowing at once points at the connection, which the next section finds.
Trace the driver and the connection first #
Work in the project that is actually running on the machine; an offline copy with different connection settings has sent many hours of diagnosis down a network that carries nothing. Then trace outward:
- Open WinCC Explorer and right-click Tag Management → Add new driver to see the installed channels. For direct S7 communication the driver is SIMATIC S7 PROTOCOL SUITE.CHN; if it is not in the list, no S7 connection can exist at all.
- Open an affected tag and read which channel and named connection it belongs to. That pair is the path you are about to test: MPI, PROFIBUS or Ethernet, whichever the connection was built for. The network diagram worth opening is the one carrying that path from the WinCC station's interface to the CPU interface the connection names.
- Compare endpoint settings: the station address on MPI or PROFIBUS, the IP address plus rack and slot on Ethernet, and the computer interface selected in the connection against the network card actually cabled to the PLC network. Addresses and interface must agree with the CPU's interface settings in the PLC hardware configuration.
- Open WinCC Explorer → Tools → Status of Driver Connections and read the state of every configured connection; in a PCS 7 server project this display reports the server-AS connection status. What you want is an established connection with no recurring error count.
- On Ethernet, treat a successful ping as a first filter only. A reply proves the CPU answers an IP reachability test; it proves neither that the driver holds a session nor that it can read your tags.
A connection that drops mid-shift announces itself as dead screens long before anyone opens diagnostics, and making that failure visible on the display is its own setup — showing only a PLC connection alarm in WinCC Flexible walks through it. One more point for an H-system plant: the H in S7-414H says the CPUs are redundant, not that your WinCC link is. Connection redundancy comes from the configured communication routes, and the running configuration documents which path WinCC actually uses.
Then trace the tag: address, type and cycle #
An established connection with a dead value moves the fault to the tag. Check four things in this order:
- External, not internal. An internal tag lives only inside WinCC and never touches the controller, and a value that refuses to move is sometimes bound to one by mistake. Confirm the tag sits under a channel and a named connection.
- Address. Compare the tag's configured address with the real location in the CPU: the same data block, start byte and bit, or the same memory area. An address pointing nowhere reads nothing.
- Data type and length. The WinCC type must match the controller's representation, including signed or unsigned treatment; a mismatch shows rejected reads or garbage values rather than healthy data.
- Quality. Every acquired value carries a quality status, good while the connection is healthy. Judge the link by value, quality and connection status together — a number on the screen alone proves nothing.
Then run the neighbor test: a connection fault drags every tag on it down at once, while a mapping fault hits one tag and leaves its neighbors healthy. Bad quality on more than one tag of a shared connection is a connection problem, however tempting the tag editor looks. When one screen object stays dead while its tag tests fine, the fault sits between the object and the tag — work the same object-to-PLC chain as in tracing an HMI value from its screen object to the PLC.
If you prove the path by writing from WinCC, write to a spare data word that drives nothing; a test write into a live output can start a machine.
The acquisition cycle is set per external tag: in classic WinCC it is a field on the tag in Tag Management, in the panel-style tag tables it is a column of the tag table you can switch on in the work area, and in WinCC Unified it sits in the Inspector under Properties > Properties > Settings. Read the cycle before you blame the network, and never shorten cycles as a repair — it multiplies bus load without serving one bad address or one broken connection.
Tracing a dead value on the plant's S7-414H at a 1 s cycle #
Take the case behind the question: an S7-414H, a WinCC station on Ethernet, screens showing nothing. Every number below is an illustrative example input — read the real ones from your own project before acting on the decision.
The inputs, with their sources: the controller and path are the plant's S7-414H reached over Ethernet, one of the three paths (MPI, PROFIBUS, Ethernet) such a connection can use; the driver and connection are SIMATIC S7 PROTOCOL SUITE.CHN under Tag Management with one named Ethernet connection to the CPU, as set up above; the tag's acquisition cycle is 1 s, illustrative, read from the tag's properties; the observation window is 60 s with the affected screen open; and the observation is that Tools → Status of Driver Connections shows the connection down while the tag's quality stays bad.
The rule in words: WinCC issues one read request per external tag per acquisition cycle, so the polls a window should contain equal the window divided by the cycle, and comparing that count with the connection status and the tag quality names the failed layer.
The arithmetic: at a 1 s cycle, 60 s ÷ 1 s per poll = 60 polls of that tag in the window; at a 5 s cycle the same window holds 60 s ÷ 5 s = 12 polls; and cutting the cycle from 1 s to 500 ms raises the demand from 60 to 120 polls per minute.
The result: with the connection shown down, none of the 60 expected polls can be served, and every external tag on that connection goes bad together — exactly what the observation shows.
The decision: a connection shown down, or bad quality on more than one tag sharing that connection, is a connection fault — repair the driver, endpoint settings and path first. One bad tag beside healthy neighbors on a connection shown up is a mapping fault — repair that tag's address or data type. Here the status reads down, so the fix is at the connection, and no tag edit and no shorter cycle serves anything until it comes up.
How WinCC and OPC exchange data #
OPC is a software interface, not a second network: an OPC server exposes process data, an OPC client reads or subscribes to it, and the direction decides which side you configure. The second part of the question — how WinCC and OPC communicate — resolves by that direction.
Outward, WinCC is the server: another application such as a reporting tool, a historian or a second SCADA client reads WinCC's tags through the OPC interface the installed WinCC provides. Inward, WinCC is the client: where the project has no native driver for a data source, you add an OPC channel under Tag Management instead, and OPC.SimaticNET is the channel that reaches S7 data through OPC rather than through the S7 protocol suite.
Diagnose the two links separately. A handshake between client and server proves that link only; it does not prove the server's own connection to the controller is alive or that the values it serves are current, so check the server's controller connection from the server side with the same status tools as above. When the consumer is another PC application, configure WinCC's server side and test from that application; when WinCC itself must read an outside source, configure its client side and test the tag's cycle exactly as in the worked example.
Questions asked
Can two WinCC stations read the same S7-414H at the same time?
Yes. Each station has its own driver connection and its own tags, and each polls on its own acquisition cycles. The CPU serves the connections independently, so a fault on one station's connection leaves the other reading normally — a quick way to separate a station-side fault from a CPU-side one. The number of simultaneous connections a CPU accepts is finite, so count every HMI, engineering station and OPC client that holds one before adding another.
Why does a dead connection still show a value instead of an error?
A displayed zero or a frozen number is not proof of communication. It can be a valid process value, the tag's initial value, or the substitute value a screen object shows for bad data. Judge the link by tag quality and the driver-connection status, never by the presence of a number on the screen.
Does an OPC subscription change how often the PLC is read?
No value can arrive faster than the slowest link in its chain. A WinCC tag fed from an OPC server is still read at its own acquisition cycle, and the OPC server refreshes its data at the update rate it agrees with your client. The controller is read no faster than the server's own connection to it, so shortening the WinCC cycle below the server's rate only adds local processing, not fresher data.
Is a ping reply enough to clear the network?
No. Ping proves one IP endpoint answered one reachability test at that moment. It says nothing about the driver session, the CPU's free connection resources, or whether the reply came from the device you meant — another station holding a duplicate IP answers just as happily. Treat ping as a first filter for an Ethernet path, then confirm the session in the driver connection status display.