How to show full test duration on a live trend chart

Match the symptom to the likely cause

Watch both the left edge of the trace and the horizontal axis labels. Their behavior helps separate a chart setting from a storage limit.

• The chart always shows the same duration, with both ends moving forward: a rolling window or follow-latest setting is likely active.
• The axis starts at zero, but the beginning of the trace disappears: the live buffer, history query, or plotted series may have a sample limit.
• The axis shows dates or clock times: it is displaying timestamps rather than elapsed duration. That does not necessarily mean any data is missing.
• Older points move or become negative when another test starts: the start reference may have changed while earlier samples remain in the series.
• The axis expands to an unexpectedly large range: check for an unset start reference, mixed timestamp units, or samples from a different run.
• The newest point stops advancing: investigate acquisition or refresh behavior before changing the time axis. The checks for factory screen update delay address that separate symptom.

More than one limit can apply. Removing a rolling window may reveal that the underlying buffer is also too short.

Check what controls the visible time range

Open the chart configuration and inspect three separate settings: the horizontal axis type, its range behavior, and the series data source. Names vary between applications, so use the software's documentation to identify the equivalent controls.

A date/time axis places samples according to their timestamps. A rolling-window option then restricts the view to a recent interval. Autoscaling the measured-value axis will not change that interval.

For a full-run view, the horizontal range needs to extend from the run's beginning to its latest sample. Some chart tools support this directly with a relative-time axis. Others need a numeric elapsed-time expression on a linear axis.

Also check for a manually fixed maximum, a zoom selection, or a query that requests only recent records. A chart can have correct axis settings while receiving only the last part of a test.

Before editing anything, record the current settings and inspect the oldest and newest timestamps actually available to the series. This gives you a baseline for deciding whether the missing beginning is hidden or absent.

Configure elapsed time from a fixed run start

Use a controlled test or recorded dataset while changing the chart. Keep these changes separate from machine control and protective functions.

1. Choose the event that defines zero. If the requirement is time since application startup, capture that event. If the application stays open between tests, capture the start of each test instead. Application uptime and test duration are not interchangeable.
2. Capture the start reference once. Store run_start before collecting that run's first sample. Do not assign it again during each chart refresh or acquisition cycle. Associate it with the run so a later restart does not give old data a new zero point.
3. Calculate elapsed time for each sample. Use elapsed = sample_timestamp - run_start, with both values expressed in compatible units and taken from a consistent time basis. For timestamps measured in seconds, divide the result by 3600 to display hours.
4. Plot matching X and Y values. Pair each measured value with the elapsed time calculated from that same sample's timestamp. Avoid combining values from one channel with timestamps from another unless they have been explicitly aligned.
5. Select an appropriate horizontal axis. Use a numeric linear axis for numeric elapsed values, or the application's documented relative-time mode. Remove the rolling-width restriction. Set the lower bound to zero and configure the upper bound to follow the latest elapsed value.
6. Label the units. Use a clear label such as Elapsed time (h). Check the actual conversion rather than changing only the label.

The start event may occur slightly before the first measurement. In that case, the first point should appear slightly to the right of zero. Do not shift it artificially unless the intended definition is time since the first recorded sample.

If the application provides a documented elapsed-time timer, it may be preferable for a live run because wall-clock corrections can distort timestamp subtraction. Keep the timer and sample records aligned, and preserve the run reference needed to interpret saved data.

Check that the available history covers the run

A full-duration axis does not imply full-duration data. Follow the records through each stage:

• Acquisition storage: does the source retain the earliest samples, or overwrite them in a finite buffer?
• History retrieval: does the query request the whole run, or only a recent period or limited record count?
• Chart series: does the component impose its own maximum number of points?
• Display reduction: is the chart drawing a reduced representation of a larger retained dataset?

For fixed-rate sampling, estimate the required sample count per channel as:

sample count ≈ run duration in seconds × samples per second

For example, a four-hour run sampled once per second produces approximately 14,400 samples per channel. Allow additional capacity for endpoint handling and runs that exceed their planned duration. Variable-rate or event-based logging needs an estimate based on the actual record rate instead.

A history limit measured in records may be shared across channels, so check how the software counts it. Increasing a chart's point limit will not increase a separate acquisition buffer.

For long runs, archived history may be more suitable than holding every sample in the live chart. A chart can retrieve the archive and reduce the number of points drawn while retaining the original measurements elsewhere. Choose reduction carefully: simple averaging can hide short peaks, while minimum-and-maximum summaries can preserve more of the visible extremes.

Do not lower the acquisition rate merely to make the chart fit unless the test's measurement requirements permit it. If early records were overwritten and no archive exists, axis changes cannot recover them.

Confirm the full-run view stays correct

Use a short, controlled run that lasts longer than the chart's previous window. Check the following before relying on a longer test:

• The stored start reference remains unchanged throughout the run.
• The horizontal axis begins at zero and its upper limit grows as samples arrive.
• The earliest retained sample remains visible after the old window duration has passed.
• The latest plotted elapsed value agrees with its sample timestamp minus the stored run start.
• A known measured value appears at the correct elapsed position, confirming X and Y alignment.
• The oldest available source timestamp still belongs to the beginning of the run.

Then check a second run. It should get its own start reference without shifting the previous run's history. If closing and reopening the application is part of normal use, confirm that the saved run reference and history are restored together.

Finally, distinguish a chart configuration problem from a display choice. A screen presents the data supplied by the application; it does not determine the application's history retention. The distinction between a live factory data display and a dashboard matters when deciding where acquisition, calculations, and chart rendering should happen.

Questions