OPC Server Connected but Tags Frozen? Fix Stale Values on ABB and GE Systems
A green OPC server connection does not prove that every tag is fresh. On ABB 800xA and GE PACSystems installations, follow the value, quality, and timestamps from the controller through the server to the client before replacing network hardware.
Why can tags freeze under a healthy connection?
The server may stay connected while individual item paths, scan groups, subscriptions, or client settings stop delivering useful updates. First check whether the source value really changes. Then compare the server's source and server timestamps, quality or status, and the client's last displayed update. Some legitimate stable process values have no new change event; a frozen display alone does not prove a communication fault.
How do you check the PLC address map?
Compare each affected OPC item identifier with the live controller variable, data type, and array index. On a GE PACSystems RX3i controller such as the IC695CPU315, verify the actual variable mapping used by the installed driver; do not assume all PACSystems projects use the same data-block scheme. An address may still exist but no longer be written by logic. For an ABB system using a PM861A processor, check the installed connectivity component and its diagnostics for unresolved items. Avoid relying on a gateway name that may not exist in that installation.
Can subscriptions and scan groups create stale-looking values?
Yes. Compare the requested and actual update rates and look for dropped updates, revised intervals, timeouts, or overloaded queues in the particular OPC server and client. Group items by real operational need. A fast requested cycle is not necessarily delivered and should not be prescribed by a generic number. Change one group at a time, then verify affected item timestamps and quality.
Could deadband be hiding small changes?
A data-change deadband can suppress notifications when the source changes by less than a configured threshold. Check whether it is applied at the PLC, OPC server, subscription, or historian layer and whether it is absolute or percentage-based. Compare a known changing source value with the received value under an authorized test. Choose a threshold from process resolution, noise, and required visibility; zero or nonzero deadband is a design decision, not a universal rule.
How do you assess controller communication load?
Count direct readers and inspect controller, network adapter, and server diagnostics for connection limits, response time, retries, and timeouts. A GE PACSystems Ethernet module such as the IC695ETM001 is one place to identify the installed hardware during the review; use its specific manual for limits and counters. Consolidate redundant polling only after checking availability and process needs. A healthy server session can coexist with an overloaded controller or a fault in one client.
Will moving from OPC DA to OPC UA fix the problem?
Not automatically. OPC DA uses Windows COM/DCOM; OPC UA has a different communications and security model with certificates and item status or timestamps where supported. A migration can simplify some DCOM-related issues but does not fix wrong addresses, unchanged source values, or poor subscription settings. Verify the particular ABB 800xA and GE connectivity versions, namespace mapping, security trust, and client behavior before planning a migration.
What should the troubleshooting record include?
Save the affected item paths, source values, quality and timestamps at each layer, actual update rates, deadband settings, communication diagnostics, and the change that restored updates. Keep a safely changing test point if the site permits it, but never force a live process value without an approved procedure.
Author: Zhang Weijian is an industrial automation engineer with over 10 years of experience in PLC, DCS, and control systems.
