Troubleshooting
When Aveva Edge cannot browse S7-1500 OPC UA tags
Short answer
AVEVA Edge connecting but showing no S7-1500 tags is a browse fault, not a comms fault: UAExpert reading the same server proves the tags are exposed. Match Edge's endpoint and security settings to the working client, verify the CPU's X1 Ethernet addresses, put a virtual machine on the PLC subnet, and check the browse timeout before rebuilding the datasheet.
On this page
- Why does AVEVA Edge connect but show no S7-1500 tags?
- How do you test each hop from Edge to the tag?
- Do AVEVA Edge's endpoint and security settings have to match UAExpert?
- Which TIA Portal interface settings does the browse path depend on?
- Does running Edge on a virtual machine change the fault?
- Could a browse timeout leave the session alive with no tags?
- How do you prove the fix before rebuilding the datasheet?
- Questions asked
When AVEVA Edge connects to an S7-1500 but shows no data-block tags, the OPC UA session is not the fault — the browse is. UAExpert browsing and reading the same server proves the S7-1500 is publishing the tags, so the work sits on the Edge side: match the endpoint and security settings to the working client, verify the CPU's X1 Ethernet addresses, put a virtual machine on the PLC's subnet, and check the browse timeout. This article walks each hop in order and ends with the single-tag test that proves a fix before you rebuild the tag datasheet.
Why does AVEVA Edge connect but show no S7-1500 tags? #
Edge has opened an OPC UA session but not completed the browse: session establishment and namespace browsing are separate services under the OPC UA specification (IEC 62541), so a connected client can still return no data-block tags.
That split decides where to work. UAExpert browses and reads the same server, so the CPU is publishing the data-block variables and the network path to the CPU is sound; what fails is Edge's discovery of them. The reported installation matches that shape: Edge lists the CPU with entries named DataBlockInstance and ServerInterface whose values read 0, and the browser offers no data-block variables to select.
The S7-1200 habit of adding tags to a server interface is the wrong trail here. An S7-1200 publishes only what its server interface lists, but this S7-1500 already shows its tags to UAExpert, so server-side exposure is proven — and adding a server interface in the S7-1200 style changed nothing in the reported case. Treat the empty browser as a discovery fault and test the hops in order.
How do you test each hop from Edge to the tag? #
Test four operations in order and record each result — endpoint session, namespace browse, single-node selection, single-value read — because each result either passes the hop or names the one you must fix.
- Session. Connect Edge's OPC UA client to the CPU's IPv4 address using the opc.tcp:// scheme on port 4840. A connected state proves the endpoint, the network path and the accepted security policy — and nothing about the tags.
- Browse. Browse from the Objects folder and count what returns. Two server-owned entries with no data-block children means the browse completed with the wrong scope or came back truncated.
- Select. Pick one known data-block tag by browsing to it rather than typing an identifier from memory; a typed reference can name a node this server does not publish.
- Read. Force a non-zero value into the PLC tag and watch Edge's copy; a value that tracks the forced change closes the whole path end to end.
Change one hop at a time and retest before moving on — the same discipline that clears a Modbus troubleshooting case where each communication hop has to match.
| What Edge shows | What it proves | Where the fault sits |
|---|---|---|
| No session | Endpoint or path rejected | Endpoint address, security policy, network path |
| Session open, browse returns only DataBlockInstance and ServerInterface | Session accepted, discovery incomplete | Browse scope, operation timeout, address path |
| UAExpert browses the tags and Edge does not | The server publishes them | Edge-side settings only |
| A selected tag stuck at 0 | The datasheet names a non-variable node | Node selection, not the PLC |
Do AVEVA Edge's endpoint and security settings have to match UAExpert? #
Yes — point Edge at the endpoint UAExpert uses, the CPU's IPv4 address on port 4840 with the opc.tcp:// scheme, and set the same security policy and message mode; a mismatch can leave a session that looks healthy while the browse returns nothing.
Read what UAExpert actually used before copying anything by eye: endpoint address, port, security policy and message-security mode. UAExpert ships permissive defaults and accepts the server certificate on its first connection, which is why it looks like it needs no configuration; an Edge project carries explicit values that must be set to the same ones. Where the policy is signed or encrypted, Edge must also trust the server's certificate — the same trust step as trusting the Citect OPC UA client certificate on a B&R PLC.
The reported installation isolated the fault with security disabled on an isolated bench network, which is acceptable as a controlled test; restore the security policy and certificate trust before the machine runs.
Which TIA Portal interface settings does the browse path depend on? #
Check the whole set under Device properties > PROFINET interface [X1] > Ethernet addresses — subnet PN/IE_1, the IPv4 address mode, and the PROFINET device name — because the reported fix changed this set together with a restart and never isolated a single property.
Before blaming the interface, confirm the server is switched on at all: select the CPU, open its properties, and activate the server under OPC UA > Server. UAExpert browsing the tags already proves that step done in this fault. Then work the three settings as a set:
- Subnet: PN/IE_1 must be the subnet the Edge PC actually sits on — with a VM, the adapter's subnet, as the next section works out.
- IPv4: set the address in the project, because that project address is the one the endpoint must contain; the working configuration used the project-set address rather than one held only in the device.
- PROFINET device name: keep the automatically generated project name so the CPU answers on the name the project expects.
Download the corrected hardware configuration so the CPU runs the project addresses, then restart the computer and all services — in the reported case the tags appeared only after that restart, so record the full set instead of crediting any single item.
Does running Edge on a virtual machine change the fault? #
It can: a VM on a NAT adapter sits on a different subnet than the CPU, so Edge takes a path the project never intended, and the fix is to bridge the VM adapter onto the PLC's subnet before touching any tag definitions.
Worked example, with illustrative addresses — substitute your project's values:
- Project subnet PN/IE_1 is 192.168.0.0 with mask 255.255.255.0, and the CPU's X1 IPv4, set in the project, is 192.168.0.1.
- Edge runs on a VM. With a bridged adapter the VM holds 192.168.0.50 with the same mask; with a NAT adapter it typically holds an address such as 10.0.2.15.
- Mask each address against 255.255.255.0: 192.168.0.50 AND 255.255.255.0 = 192.168.0.0, and 192.168.0.1 AND 255.255.255.0 = 192.168.0.0. Equal networks, so the endpoint opc.tcp://192.168.0.1:4840 is reachable with no router.
- NAT case: 10.0.2.15 AND 255.255.255.0 = 10.0.2.0, which differs from 192.168.0.0. The endpoint is then off the VM's network, and Edge's session and browse take a path the project never intended — the difference from UAExpert running on the host.
Decision: bridge the VM adapter onto the PLC's subnet, or route to it deliberately, and make the two masked results equal before you edit a single tag in Edge.
Could a browse timeout leave the session alive with no tags? #
Yes — browsing is expensive work on the server and a separate operation from the session, so Edge can hold a healthy connection while its browse operation exceeds its own timeout and returns a partial or empty tag list.
The browse runs on the client's operation timeout, not the session's. Open the OPC UA client settings in the Edge project and read the configured value; another client's default proves nothing about Edge's, and Edge's value is what decides.
Separate timing from scope by the shape of the result. If a retry, or a longer configured window, brings in more of the namespace, timing is the fault: scope the browse to the specific data-block instance instead of the whole Objects tree, or raise the client's operation timeout until it covers the namespace with margin. If the same truncated list returns on every attempt, the browse is completing rather than timing out — go back to scope and the address path. Do not raise a timeout blindly: first establish whether the browse is still progressing or repeatedly returning the same incomplete set.
How do you prove the fix before rebuilding the datasheet? #
Prove it with one client, one tag and one screen object: add a single tag from the known-good node, display it, and confirm the value tracks a forced PLC value before you rebuild the full datasheet.
Keep UAExpert as the control client and record four facts for both tools: endpoint address and port, security policy and mode, browse scope, and operation timeout. Apply one change at a time and retest after each, so the change that worked stays identifiable. Then run the single-tag test: one OPC UA client in the Edge project, one tag browsed from the known-good node, one screen object bound to it, a non-zero value forced into the PLC tag, and Edge tracking it live. Only then rebuild the full datasheet — rebuilding first hides which change mattered.
The decision, case by case: no session means the endpoint address, security policy or network path; a session with a browse stuck at DataBlockInstance and ServerInterface means browse scope, operation timeout, or the address path through the X1 settings and the VM adapter; UAExpert working while Edge does not keeps the fault on the Edge side; a selected tag stuck at 0 means the datasheet names a non-variable node. Prove each fix with the single-tag test before you call the machine ready.
Questions asked
Does enabling the OPC UA server on the S7-1500 disturb the existing PROFINET communication?
No. The OPC UA server is one service on the CPU's X1 interface and shares that interface with PROFINET, HMI connections and the web server. OPC UA clients connect over TCP port 4840 while PROFINET IO runs as its own real-time traffic, so activating the server under OPC UA > Server adds a service without displacing the existing communication.
What are the DataBlockInstance and ServerInterface entries Edge shows with a value of zero?
They are server-side objects the S7-1500 OPC UA server publishes to organize its address space, not process values, so a datasheet pointed at them reads 0 because no variable sits behind the node. That 0 says nothing about the PLC data. Browse into the children of the data-block instance and select the individual variable nodes instead.
Is it safe to leave the S7-1500 OPC UA server without security after a bench test?
No. Disabling security is a bench-only troubleshooting step on an isolated network. Before the machine runs, re-enable the security policy the project requires and make Edge trust the server certificate, because without them any client that reaches the CPU can read process data and write values into the PLC. Record the working endpoint, policy and trust settings in the project notes so nobody troubleshoots the next fault by turning security off again.
Why did the tags appear only after restarting the computer and all services?
OPC UA clients and runtimes cache endpoint data, certificates and open sessions from the address path they were started with. Changing the CPU's X1 Ethernet addresses or the VM's adapter does not reach a session that is already running, so every service keeps the stale path until it is restarted. Restart the machine and the Edge services after any address change so each client rebuilds its session from the new path.