New: the AI firewall watches the whole platform and explains each incident in plain words. See how it works

Dashboard last updated time: is that value live?

A frozen value looks exactly like a steady process. How to show data age on a card, spot a stuck reading, and pick a stale limit per signal.

A dashboard last updated time shows how old the value on a card is. Without it, a reading from a sensor that stopped reporting an hour ago looks the same as a live reading from a steady process. Print the age beside the value, give each signal a stale limit, and change how a value looks once it passes that limit.

Many devices send a reading only when it moves past a deadband, or once per interval. A tank level that holds still and a logger with a flat battery leave the same number on the screen, and the operator cannot tell which one it is.

How do you show data age on a dashboard?

Show a relative age next to the value, such as “updated 4 min ago”, taken from the reading’s own timestamp. It needs no arithmetic and no time zone.

Use the time the device measured the value, or the time it arrived if the device clock cannot be trusted, never the time the page loaded. Keep the age small and grey while the value is fresh, so it does not compete with the alarms (more on why most of an operator screen should be grey), and give it colour only when the value goes stale.

How can you tell a frozen value from a steady one?

From the value alone you cannot. You need a second clue that keeps moving while the process is quiet:

  • A known rhythm. A sensor that sends every 60 seconds and has been silent for ten minutes is frozen, whatever the number says.
  • Noise. A raw analog reading sent at a fixed interval usually jitters in its last digit, so an hour of perfectly flat trace points at a stuck input.
  • A related signal. If a pump reports running and the flow behind it has not changed by a single digit in an hour, one of the two is wrong.

Industrial protocols often carry a quality flag with each value. OPC UA status codes have a severity of Good, Uncertain or Bad, and the specification says results with a bad status “shall never be used” and uncertain ones “shall be used with care”.

When should a dashboard value be marked stale?

Mark it stale once its age passes a limit set from how often that signal should report. One limit per board fails, because a meter sending every minute and a level sensor sending every 15 minutes go quiet on different clocks.

How the signal reportsStale limit I would start withExample
Fixed intervalTwo to three missed reportsEvery 5 minutes: stale after 15
Only on changeTwo heartbeat intervals; with no heartbeat, no limit is safeHourly heartbeat: stale after 2 hours
Daily uploadThe schedule plus a marginDue at 06:00: stale after 25 hours

A stale value should look different. Fade it, keep the number with its age, and turn a state such as “Running” into “Unknown”, because a confident green word is the worst thing to show for a pump nobody has heard from. Controls bound to it should stop taking input.

Hermetry connects field devices to live dashboards, rules and alerts: IoT telemetry and SCADA in one platform. Every dashboard card has a Data is old after (seconds) setting, from 0 (never, the default) up to 604,800, and shows old data with an “Old data” badge by default, or faded, or as the no-data message. Among the stat, gauge and state cards, a stat card can print “Updated … ago”, a gauge fades on old data by default, and a state card shows “Unknown”. Switches, sliders, set points and forms are disabled by default when their reading is old.

Whether someone gets called when a sensor goes quiet is a separate setting, covered in alerting on missing data. What a chart should draw across the hole is in a gap, a zero or the last value.

Frequently asked questions

Does refreshing a dashboard make its values live?

No. A refresh asks again for the newest stored reading of each value. If the device has sent nothing since, the page shows the same old reading with a new load time. Only the reading’s own timestamp says how old the value is.

Should the age use the device clock or the arrival time?

The device’s measurement time is better when its clock is synchronised, because readings buffered during an outage keep their real age. If the device clock drifts or resets after a power cut, use the arrival time, or the label can show a value as older or newer than it is.

Should a stale value be hidden from the screen?

I would keep it, faded and with its age. The last known value tells the person driving to the site what to expect. The exception is a state a decision hangs on, such as running or stopped: show it as unknown, because an old “Running” invites the wrong call.

Share

See your field data live this week

Start a trial in your region with your first devices, or book a demo and bring your own use case.