Troubleshooting
Fix PROFINET connect errors between a controller and an S7-1200
Short answer
Separate the engineering port and the PROFINET port onto different IP layers on different physical ports, connect the controller to the S7-1200 directly or through a PROFINET-capable switch, make the device name and address match the project, and run the PROFINET IO task at priority 1 or 0. A scan that finds the S7-1200 proves discovery only; the connect stage is where DB814006 appears.
On this page
- What does DB814006 Connect Parameter Error mean on a PROFINET controller?
- Does a scan that finds the S7-1200 prove the connection works?
- Which IP addresses and masks belong on each Ethernet port?
- Does the S7-1200 device name and IP address have to match the project?
- Which task priority keeps the PROFINET IO connection stable?
- Why does the controller log fill with "Check Firewall..." messages?
- Can the PROFINET link run through an office Wi-Fi router and switch?
- What is the fastest order to fix the connection?
- Questions asked
The fix that clears this fault is architectural: put the engineering connection and the PROFINET connection on different IP layers on different physical ports, wire the PROFINET port straight to the S7-1200 or through a switch that suits PROFINET, make the project's device name and address match the physical CPU, and run the PROFINET IO task at priority 1 or 0. In the reported case a controller scanned the S7-1200, copied it into the project, compiled and downloaded, and still failed every connect with DB814006 - Connect Parameter Error while the CPU blinked red. The scan was never the problem: discovery and the I/O connection are separate stages. This article gives the addresses that worked, the subnet arithmetic that shows why the earlier arrangement failed, and the order to work when your own link sits in the same error.
What does DB814006 Connect Parameter Error mean on a PROFINET controller? #
DB814006 - Connect Parameter Error is the stack reporting that its connect request was rejected or never answered; it is an outcome, not a diagnosis, so treat it as the signal to check four things in order: device name, address assignment, network path, and the IO task priority.
The code does not name the parameter that offended. The pattern from the reported case — clean compile, clean download, the same error at every login while the S7-1200 blinks red — means the device side sees no controller connection at all, so start at the top of the list rather than hunting one setting. Each check has its own section below; the controller log and a packet capture tell you which one is still failing, the error text alone never does.
Does a scan that finds the S7-1200 prove the connection works? #
No — the scan only proves discovery: the controller exchanged the frames that list what is on the wire, while the I/O connection is a separate stage that needs the station name, the address assignment and the task schedule to agree before any cyclic data moves.
The scan, the copy to the project and the GSD install all belong to the engineering stage. The I/O connection opens at runtime, against the station name and the addresses the route table resolves, over the physical path the frames actually take. Engineering access is the classic false positive: a programming session can run happily over one path while the I/O connection dies on another — the same trap as an S7-1200 that shows link lights but TIA Portal cannot connect. Treat every green engineering indicator as evidence about its own path only.
Which IP addresses and masks belong on each Ethernet port? #
Put each traffic path on its own IP layer: in the reported fix, the engineering port carried 192.168.1.133 with mask 255.255.255.0, the PROFINET port carried 192.168.2.144 with 255.255.255.0, and a cable ran from that port straight to the S7-1200 with no switch between them.
An IP layer is one interface address plus its subnet mask, and the mask — not the dots — decides where the network ends. Give the S7-1200 an address in the PROFINET port's layer; 192.168.2.133 with 255.255.255.0 is the example used here. The arithmetic:
- 192.168.2.144 AND 255.255.255.0 = 192.168.2.0 — the PROFINET port's layer.
- 192.168.2.133 AND 255.255.255.0 = 192.168.2.0 — the CPU sits on the same layer, so the peer is directly connected and exactly one interface in the route table owns it.
- 192.168.1.133 AND 255.255.255.0 = 192.168.1.0 — a different layer, so engineering traffic cannot drift onto the PROFINET port or the reverse.
With a 255.255.255.0 mask the first three octets decide; with any other mask, compute the AND, because two addresses can look close while sitting on different layers. The broken intermediate state in the reported case had both controller ports inside 192.168.2.0/24: two interfaces claimed one connected network, the route to the CPU was ambiguous, and frames could leave through the engineering port, where the office unit filtered them. Overlapping layers produce this same class of fault elsewhere, as when a SCADA gateway cycles between Connecting and Disconnected on a subnet mismatch. A controller with a single Ethernet port cannot split layers; run it on one layer and let the switch decision below carry the rest.
The working test, with an example address for the CPU:
| Interface | Role | Address | Mask | Layer |
|---|---|---|---|---|
| Controller port 1 | Engineering access | 192.168.1.133 | 255.255.255.0 | 192.168.1.0/24 |
| Controller port 2 | PROFINET IO | 192.168.2.144 | 255.255.255.0 | 192.168.2.0/24 |
| S7-1200 (example) | PROFINET device | 192.168.2.133 | 255.255.255.0 | 192.168.2.0/24 |
Does the S7-1200 device name and IP address have to match the project? #
Yes — a PROFINET controller opens the connection to the station name written in the project, so the name assigned to the physical S7-1200 and the address it holds must match the project values exactly, letter for letter and octet for octet.
PROFINET addresses stations by name, so the connect request carries the project's station name and the device answers only if its assigned name is identical. Read the CPU's live name and address from its device configuration, compare them with the device properties in the controller project, and correct whichever side is wrong.
The S7-1200 documentation covers the assignment in its own tool: the Online & diagnostics view of the device in the Devices & networks editor assigns the name online, and the Assign PROFINET device name dialog lists every configured name and asks which PG/PC interface faces the wire.
Close the project traps while you are there: install the S7-1200's GSD into the controller's device repository and confirm the scan resolves to that same physical CPU. Then re-scan once at the end — the discovered station must be this unit on the intended port, not merely a similar name.
Which task priority keeps the PROFINET IO connection stable? #
Run the PROFINET IO task at priority 1 or 0 in the runtime's task configuration; with any other priority the cyclic communication will most likely never become stable, because ordinary logic tasks then delay the fixed-deadline processing the protocol expects.
The task configuration table gives every task in the project a priority number, and the lower number wins, so 0 and 1 sit above every ordinary logic task. The task that carries the PROFINET IO is created with the PROFINET controller in the device tree; set it to 1 or 0, download, and confirm in the task monitor that it actually runs at the value you set. Keep the fix in perspective: priority stabilizes scheduling, it does not repair a wrong device name, a wrong address or a filtered path, so apply it after the layers and identity are right, not instead of them.
Why does the controller log fill with "Check Firewall..." messages? #
A log full of "Check Firewall..." entries means the controller's stack reports that its UDP-based frames are not getting through, which in the reported case traced to the office router path and the overlapping subnets rather than to an actual firewall rule on either device.
Treat the message as a lead, not a verdict. Keep the full logger output from startup through the failed connect — repeated messages and their timing show which stage stops — and capture packets to see which traffic class is being dropped, either with Wireshark on the controller or with a TAP or a mirrored switch port in the path. Read the message against the protocol's traffic classes too: not all PROFINET traffic is UDP, and the cyclic real-time frames do not use it, so the log text alone cannot name the class that dies. In the reported case the captures plus the working point-to-point test settled it: the frames stopped at the office unit, not at either endpoint.
Can the PROFINET link run through an office Wi-Fi router and switch? #
Do not run the PROFINET connection through a combined office Wi-Fi router and switch: the reported case connected only after that box was removed and a cable ran port to port, while the same model worked for another engineer only with its own settings corrected.
Discovery frames and real-time frames are Ethernet traffic types a consumer routing unit can filter, bridge badly or drop between its wireless and wired sides. Decision: with two devices, run one cable between the controller's PROFINET port and the S7-1200; with more devices on the link, use a switch you have confirmed suits PROFINET and keep the office unit on the engineering side only. The other engineer's working same-model switch proves the box is configurable, not that yours is — without a capture you cannot verify what a given unit does to PROFINET frames, so default to removing it from the path.
What is the fastest order to fix the connection? #
Work the stages in the order the connection needs them: separate IP layers first, then device name and address match, then a cable or PROFINET-capable switch without the office router, then the IO task priority — each stage either clears DB814006 or hands you the next check.
Stop the machine before you start: re-addressing ports and re-connecting the link drops IO, and outputs can glitch through the transition.
- Draw the two IP layers: engineering port on 192.168.1.0/24, PROFINET port on 192.168.2.0/24, S7-1200 inside the PROFINET layer. Any distinct pair of layers obeys the same arithmetic.
- Match identity: assign the station name to the CPU with the vendor tool, install the correct GSD in the controller's device repository, and make project values equal live values.
- Clear the path: cable port to port, or through a switch that suits PROFINET; no office Wi-Fi router between controller and CPU.
- Set the PROFINET IO task priority to 1 or 0 and download.
- Read the result: connection established means done; DB814006 again means read the logger, and repeated "Check Firewall..." messages point back to steps 1 and 3, so capture at both ends before touching the project again.
Decision, by hardware: a controller with two Ethernet ports takes the split-layer arrangement above and should never share one port between engineering and PROFINET IO; a controller with one port runs a single layer, uses a switch that suits PROFINET, and needs everything else on this list unchanged. Either way, a scan is the start of the test, not the end of it.
Questions asked
Can the S7-1200 receive its PROFINET name and IP from the controller above it?
Yes, when it is configured for that. The S7-1200 settings include an option that lets a higher-level controller set the device name and address; when that option is not enabled, the CPU keeps the values its own project downloads. Either way the two projects must agree, or the connect stage rejects the station.
Does PROFINET real-time data cross a router?
No. Cyclic IO data travels as Ethernet frames carrying the PROFINET Ethertype 0x8892, not as IP packets, so nothing that routes, including the router function of a combined Wi-Fi unit, can forward it. Keep the controller and the S7-1200 on one physical link, switched only by equipment that suits PROFINET.
Can I keep everything on one flat subnet instead of splitting the layers?
Yes, when the controller has a single Ethernet port: put the controller, the S7-1200 and the engineering PC on one network, connect through a switch you have confirmed suits PROFINET, and keep the office Wi-Fi router out of the path. With two ports, the split layers remove the route ambiguity that produced DB814006 in the reported case.
Can a packet capture from one side settle where the frames stop?
No. Capture on both sides of the suspected hop, with Wireshark on the controller and on the engineering PC, or with a TAP or a switch port mirror in the path. The logger gives the controller's view; the captures show whether frames leave one endpoint and reach the other, which identifies the unit that drops them.