How-to
How to create a SCADA PLC communication alarm for link loss
Short answer
Create the SCADA-to-PLC communication alarm from the data server's diagnostic item, not from a PLC bit. Browse to IsPresent under the PLC shortcut's Diagnostic Items folder, create a digital alarm in FactoryTalk Alarms and Events with IsPresent = 0 as the active condition, and it activates about 20 seconds after the link drops; IsPresent(10) shortens that to about 10 seconds. A heartbeat bit cannot do this because it travels over the link it monitors and goes unreadable exactly when the link fails.
On this page
Build the SCADA-to-PLC communication alarm from the data server's own diagnostic item, not from a bit inside the controller. In FactoryTalk View Site Edition, browse to IsPresent under the affected PLC shortcut's Diagnostic Items folder and create a digital alarm in FactoryTalk Alarms and Events whose active condition is IsPresent = 0. With the default timeout the alarm activates about 20 seconds after the link drops; the qualified form IsPresent(10) cuts that to about 10 seconds. The heartbeat bit cannot do this job: it travels over the connection it monitors, so it goes unreadable at the exact moment you need it. The steps below cover the configuration, the test timings with the arithmetic shown, and the decision for each failure case.
Decide which failure you are alarming #
Lost communication names three different faults, and each needs its own alarm source:
- The SCADA server lost the controller — cable severed, switch down, or controller off. The shortcut diagnostic item IsPresent detects exactly this, and it is the subject of this article.
- The controller lost a downstream device — a drive, remote I/O drop, or valve bank stopped answering. The SCADA shortcut can stay healthy while this happens, so IsPresent stays at 1. Detect this in the controller's own communications logic.
- A client lost the application services — no display updates and no alarms reach any banner. A PLC communication alarm cannot display on a client that cannot receive alarm data, so fix the client and services first.
If every PLC tag on the affected displays goes unavailable at once, you are in the first case.
Why the heartbeat bit cannot raise this alarm #
A heartbeat bit lives on the connection it monitors. Sever the link and the HMI can no longer read the bit's next state: the last value sits frozen on screen while the tag's quality turns bad and displays show question marks — the failure mode behind why HMI values show question marks during PLC comms timeouts. A digital alarm on that bit now has nothing to evaluate, because its input disappears during the very condition it is supposed to announce. Extra transition logic in the PLC does not repair that dependency.
Keep a heartbeat only if you also need confirmation that the controller program is running and writing outputs, since link reachability does not prove program health. Then the stopped-heartbeat comparison must run outside the controller, and it still rides the same link, making it a second, slower route to the answer IsPresent already gives you. For the other legitimate heartbeat — an operator terminal detecting power loss — configuring a PanelView Plus heartbeat to detect power loss walks through that setup.
Read IsPresent under the PLC shortcut #
The diagnostic item belongs to the communication shortcut, not to the controller program. The FactoryTalk Linx data server publishes these items under every shortcut, and you browse them like any tag:
- Open the HMI tag browser from the alarm's input field.
- Expand the shortcut the affected displays use for this controller. Three folders appear: Diagnostic Items, Offline, and Online.
- Open Diagnostic Items and select IsPresent.
- With the link healthy, read the item: it holds 1 with good quality. Record that baseline.
- During an authorized interruption, watch it transition to 0. The default timeout is 20 seconds of lost communication; writing IsPresent(10) applies a 10-second timeout instead.
Two boundaries matter. IsPresent reports reachability through that one shortcut: a 1 says nothing about the controller's program, its process, or the devices behind it. And if the item never transitions to 0 during your test, you interrupted a different path than the shortcut you selected — recheck which shortcut the displays actually use.
Create the digital alarm in FactoryTalk Alarms and Events #
Configure this as a digital alarm. The trigger is a Boolean diagnostic, so the high-high, high, low, and low-low limit fields of an analog alarm definition have nothing to do here:
- In FactoryTalk Alarms and Events, create a new digital alarm definition.
- Browse the trigger to Diagnostic Items > IsPresent under the affected PLC shortcut, or to the IsPresent(10) form you tested.
- Set the active condition to 0. Activating on 1 would announce healthy communication as a fault.
- Write the message so it names the controller and path, for example: "Loss of SCADA communication to the packaging line controller."
- Enable the alarm. An activation delay, if you set one, adds to the diagnostic timeout; it does not replace it.
- Open the alarm banner's configuration and confirm its subscription and filters include this alarm's source, so the banner actually displays it.
If the diagnostic item changes to 0 but the alarm never goes active, the fault is in the definition: check the trigger, the active state, and the enabled setting before touching network gear.
Worked example: from cable pull to banner #
The following timings come from the documented timeouts; treat them as design arithmetic, not as a measurement of your network.
Default configuration, link severed at t = 0 s:
- IsPresent transitions to 0 at t = 20 s (the default timeout).
- No activation delay, so the alarm is active at t = 20 s and reaches the banner then.
Faster configuration, IsPresent(10) plus a 5 s activation delay, link severed at t = 0 s:
- IsPresent transitions to 0 at t = 10 s.
- The delay adds to the timeout: 10 s + 5 s = 15 s from cable pull to an active alarm.
Restore the link at t = 45 s and the item returns to 1 as soon as communication re-establishes, so the alarm condition goes inactive within one update of the restored connection. Set the timeout by the response you expect: 20 s default for an informational alarm, IsPresent(10) when operators are expected to act.
Test it before the real outage #
Test on the bench or in a planned window. One safety rule: interrupt the link only with the machine in a safe state under an approved test plan, never by pulling the network cable of a running line whose motion depends on it.
- Record the healthy baseline: IsPresent = 1, good quality, no active alarm.
- Open the authorized test point, for example the controller's switch port, never a safety circuit.
- Record when the alarm goes active and compare it with the arithmetic in the worked example above.
- Confirm the banner shows the alarm with your message. Active in the alarm system but missing from the banner means the banner subscription or filter is wrong.
- Restore the link and confirm the item returns to 1 and the alarm condition goes inactive.
- If you adopted IsPresent(10), repeat the test with it installed, so the tested and running configurations match.
The decision for each case #
- SCADA server cannot reach the controller: digital alarm on IsPresent = 0 under that PLC shortcut, 20 s default timeout, or IsPresent(10) when operators must respond. This is your communication alarm.
- Controller cannot reach a drive, remote I/O drop, or valve bank: a different fault with a different source. Alarm it from the controller's communications logic; IsPresent stays at 1 and is correct to do so.
- Heartbeat stopped, quality good: logic problem in the producer, not communication. Fix the producer; no alarm on the bit will help.
- Heartbeat stopped, quality bad: the same link failure IsPresent reports — use the diagnostic item, not the bit.
- No client shows data or alarms anywhere: the failure sits upstream in the client or application services, and no PLC communication alarm will display until that is restored.
Questions asked
What timeout should I set for the diagnostic item?
Keep the 20 second default when the alarm is informational and operators only need it for shift records. Set the qualified form IsPresent(10) when the alarm starts an operator response, because ten seconds still filters momentary interruptions while cutting the announcement time in half. Avoid anything shorter until you have measured your network's recovery behavior, since a timeout below the reconnect time turns every brief switch hiccup into a nuisance alarm.
Does this alarm also catch a controller that lost power?
Yes. IsPresent reports whether the controller is reachable through the shortcut, so a powered-down controller, a severed cable, and a failed switch all drive the item to 0 and raise the same alarm. That is the behavior you want from a link-loss alarm: any path failure between the SCADA server and the controller is announced the same way. Write the alarm message so it names the controller rather than guessing at the cause.
Do I need one alarm per controller?
Yes, one digital alarm per PLC shortcut. Each alarm browses to the IsPresent item under its own shortcut, so three controllers need three alarm definitions. Give each message a different controller name so the banner tells operators which link dropped. One alarm input references one shortcut, and a single combined alarm would also hide which controller actually failed.
Why does the alarm stay on the banner after the link recovers?
Recovery returns the alarm to its inactive state, but the banner normally keeps the entry on screen until an operator acknowledges it, per the alarm definition's acknowledgment settings. Treat that as a feature: the outage leaves a record even after the link returns. Confirm the behavior during your test by restoring the link and watching whether the banner entry clears on its own or waits for acknowledgment, then set the definition to match your plant's practice.
What if no client anywhere shows tag data or alarms?
Then the fault sits above the PLC link. FactoryTalk View clients receive tag data and alarms through the same services, so a client cut off from those services can neither display this alarm nor evaluate the diagnostic item. One documented example from the vendor's support site: FactoryTalk Directory client and server communication loss caused by high memory usage, patched for FactoryTalk Services Platform 6.40/6.50. Restore the services first; the controller link may be perfectly healthy.