WinCC V6 Redundancy: Fix 'u' Values and Time Drift in Tag Logging

Problem Overview

A WinCC V6.0 SP4 installation with two redundant servers and three WinCC Web Navigator clients exhibited two correlated anomalies in the Online Table Control (OTC) of the runtime archive:

1. A literal letter "u" appeared after several archived tag values for specific days (e.g., 12-09-07 and 13-09-07) on a printout generated on 25-09-07. A printout of the same view generated on 21-09-07 showed no "u" suffix for those same days.
2. Subsequent values for the same flowmeter and the same dates differed between the 21-09-07 printout and the 25-09-07 printout, even though the underlying process variable was identical.
3. A time-format mismatch existed between server (displayed as hh:mm, e.g., 23:50) and Web Navigator client (displayed as hh:mm:ss.ms, e.g., 23:50:00.078).

All three symptoms originate from the same root mechanism: how WinCC V6 evaluates, substitutes, and synchronizes tag values when one server in a redundant pair loses contact with the PLC, and how the second server rebuilds the archive segment from its own PLC connection after a switchover. The remainder of this article maps each symptom to its underlying cause, then walks through the configuration, verification, and mitigation steps a commissioning engineer should apply.

Symptom Matrix

Root Cause Analysis

1. What the letter "u" means

The "u" suffix in a WinCC V6 tag logging archive row is the substitution-value marker. It is appended to a value that was not obtained from the PLC at the requested time, but was instead generated by the WinCC Data Manager because the connection to the AS (automation system) was unavailable at that moment.

According to Siemens Support Entry 109762744, the Online Table Control (and Online Trend Control) can render dynamic parameters with current WinCC tag values. When the tag value requested for the column cannot be retrieved, the control renders the configured substitution value (default: the value at the time of last successful read, or a user-defined default such as 0). The "u" character is the visible status flag the archive driver adds in the printable export to indicate "substituted / unreliable".

The presence of "u" on 12-09-07 and 13-09-07 means that on those two calendar days the connection between WinCC and the PLC was interrupted long enough for the Data Manager to insert substitute values. By the time the 21-09-07 printout was generated, either:

• the original archive segment on the server that printed on 21-09-07 still contained the bad rows and the operator had not refreshed the view, or
• the 25-09-07 printout was generated from a server that had synchronized the affected segment from its partner after a redundancy switchover, bringing in the rows flagged with "u" that the first server's view did not show.

This is the first direct link between the "u" symptom and the value-drift symptom: both are governed by the redundant tag logging synchronization cycle.

2. Why the numeric values drift

When the WinCC/Redundancy option is licensed, both servers in the pair are active. Each server maintains its own connection to the PLC. Tag logging runs independently on both servers, producing two parallel archive streams. The redundancy option does not unify the two streams into a single value stream at the field level; instead, it periodically synchronizes the database segments between the two servers so that whichever server is currently the master (and therefore the one answering Web Navigator queries) holds a complete view of the recent history.

The synchronization model is described in Siemens Support Entry 22273427 and operates on whole segments, not on individual rows. Segments are bounded by configurable size or time criteria. When a segment is closed on server A and is missing on server B (because server B was offline, or because server B's PLC connection produced no data for that window), the closed segment is copied wholesale from A to B.

Two consequences follow:

1. Per-row minor differences are normal. Because each server timestamps the value at the moment its own Data Manager polls the PLC, the two servers will not produce identical time stamps for the same physical event. The drift is typically one or two polling cycles (often 250 ms to 2 s depending on the configured acquisition cycle and the AS scan time). The redundancy layer accepts this and does not attempt to reconcile individual rows.
2. Segment-level differences are decisive. The value displayed in a WinCC Web Navigator OTC is always the value from the current master's segment. If the current master rebuilt the segment from its own PLC stream, the visible value will be the master's value, not the value the standby had. After a switchover, the OTC is re-rendered from the new master's segment, which is why the two printouts of the same view can differ.

There is no configuration in WinCC V6 that eliminates per-row timestamp and value drift between the two redundant servers, because the two parallel PLC connections are by design. The mitigation is to control which server's segment is considered authoritative and to ensure that segment roll-over events do not occur during a critical view.

3. Why the time format differs between server and client

The WinCC archive stores every value with a 100 ms time stamp internally; the displayed format is purely a property of the column control. The WinCC server's Online Table Control, when run in the WinCC Explorer / RT context, defaults to the column format defined in the project. The WinCC Web Navigator client runs the same control in a browser-hosted WinCC Runtime and, if no explicit column format was assigned in the project, falls back to a more verbose default that includes seconds and milliseconds.

The result is that two viewers of the same archive data show different text strings, even though the underlying time stamp is identical. This is a presentation issue, not a data-integrity issue, but it is the issue that most often confuses operators who compare a server printout with a client printout.

Technical Details: WinCC V6 Redundancy Architecture

The two redundant servers communicate over a dedicated redundancy channel (typically the terminal bus / plant bus). The redundancy service on each server polls the partner at a fixed interval. When the partner does not respond for the configured number of attempts, the local server declares itself master and the Web Navigator clients re-route to it. Archive segments accumulated while the partner was down are copied to the partner in the background after the partner returns.

Configuration: Redundancy Editor

Open the WinCC Explorer on the master server and edit the project's Redundancy editor (Computer → Properties → Redundancy). The relevant fields are summarized below.

Configuration: Tag Logging Acquisition and Substitute Value

For each archive-relevant tag, open Tag Logging → Properties → Archive. The settings that influence the "u" behavior are:

• Acquisition type: Acquisition cycle (polled) or on change. A polled cycle at e.g. 1 s and an archive cycle at e.g. 10 s means that one out of ten polled values is archived; a connection drop of 1-2 polls is invisible, a drop of 30 polls produces a substituted segment.
• Substitute value: Initial value, Last valid value, or Substitute value (user-defined). The choice controls what number is written into the archive, but it does not suppress the "u" flag in the printable export.
• Archiving mode: Smoothed vs. on change. Smoothed tags interpolate, so a brief connection drop is hidden; on-change tags will show "u" more readily.

Step-by-Step: Resolving the Customer's Three Issues

Step 1 — Confirm the "u" rows correspond to a real connection event

1. Open the WinCC Explorer on the server that generated the 25-09-07 printout.
2. Open Tag Logging → Archives → [Archive name] → Properties → Time Range and filter to 12-09-07 00:00:00 to 13-09-07 23:59:59.
3. Identify the rows marked with "u" and note the time stamps.
4. Cross-check the GSC diagnostic file (Computer → Properties → GSC Diagnostics) and the Windows Event Viewer (Application log, source S7 or SIMATIC NET) for connection-loss events at the same time stamps.
5. If the diagnostic files do not show a connection loss, the "u" rows were produced locally by a transient on this server only; if the partner server's diagnostics show a connection loss, the rows were synchronized in from the partner.

Step 2 — Identify which server produced the value in each printout

1. From the Web Navigator client, note the server name in the title bar of the OTC (configured under Computer → Properties → Web Navigator → Server).
2. Re-print the same view on each server individually (Server A's WinCC Runtime, then Server B's WinCC Runtime).
3. Compare the values row by row. The rows that differ are the rows where the current master and the partner disagree, which is the expected behavior described in FAQ 22273427.
4. Document which server is the authoritative source for the report. The drift cannot be eliminated; it can only be assigned.

Step 3 — Align the Web client time format to the server

1. In WinCC Explorer, open the Graphics Designer picture that hosts the Online Table Control.
2. Double-click the control to open its Configuration dialog.
3. Open the Columns tab and select the time-stamp column.
4. Set the Time format property to hh:mm to match the server view. Optionally set the Date format to dd.MM.yy for full parity.
5. Save the picture, redeploy the runtime project, and restart the Web Navigator service on both servers.
6. Refresh the Web Navigator client (Ctrl+F5) and verify the time column now reads hh:mm without the trailing ss.ms.

Step 4 — Stabilize the segment boundaries

1. Open Tag Logging → Archives → [Archive name] → Properties → Archive parameters.
2. Set the segment size to a fixed disk quota (e.g., 32 MB) and a fixed time window (e.g., one week), in that order of priority.
3. Schedule the report-generation job so it runs at least one full segment roll-over after the period it reports on, not in the middle of an open segment.
4. Increase the number of segments kept online so segment copy events from the partner do not force a roll-over on the local server mid-report.

Verification

Troubleshooting Matrix

Field-Commissioning Notes for WinCC V6 Redundancy

• Document the authoritative master. Pick one server (e.g., the one with the higher-numbered IP on the terminal bus) as the source of record for archive reports, and pin the Web Navigator clients to it. The drift is then bounded to a single, documented source, not a per-report surprise.
• Align polling and archive cycles. Set the Data Manager acquisition cycle to a divisor of the archive cycle (e.g., 1 s acquisition, 10 s archive). The smaller the divisor, the smaller the per-row timestamp drift between the two servers for the same physical event.
• Watch the segment copy direction. After a switchover, segments are copied from the new master to the former master. During the copy window, the former master may show stale data; suppress report jobs during this window using the Redundancy event log as a trigger.
• Avoid mixing the OTC column format between designs. The column format is stored in the Graphics Designer picture (.pdl) and in the deployed runtime project. Copying a picture from a third-party project that uses hh:mm:ss.ms will silently override the local format. Lock the format with a project-wide template.
• Distinguish "u" from a real bad value. A "u" row is a documented connection event. A row with an obviously wrong numeric value but no "u" is most often a unit-scaling or OPC quality-code issue, not a connection issue. The two are diagnosed differently.

Differences vs. WinCC V7 and TIA Portal RT Professional

For engineers who will migrate this site to a newer WinCC, the following differences are worth noting. The behavior described in RT Professional online configuration of the table view moves the segment-rollover question to a per-view configuration: in RT Professional, the table view can be configured to retain or discard online changes on a screen change or at the end of runtime, which gives finer control than the V6 project-wide segment model. The redundancy model itself is simplified in V7.x to a "master + standby, copy on switchover" pattern; the underlying per-row timestamp drift is unchanged.

FAQ