HMI alarm stops beeping but keeps flashing after reset

Two states inside one alarm

An alarm in an HMI carries more than one state. The trigger state is the tag condition: your C bit at 1 means the alarm condition, at 0 means the condition is gone. The acknowledgment state records that an operator saw the alarm. Alarm functions react to transitions of the tag, the same habit that makes alarm logging fire only on the state change, and the banner clears only when the trigger is inactive and the acknowledgment has been given. Many alarm systems also let you assign acknowledgment bits or a dedicated acknowledgment tag next to the trigger, which hands the acknowledgment back to the PLC on its own address.

The audible side gets its own path. A buzzer can follow the trigger, follow a silence command, or follow its own register, which is why a buzzer register is not an alarm trigger. A press that reaches the buzzer's path and misses the banner's acknowledgment leaves you with a silent buzzer and a flashing banner. The pattern shows up across platforms: an alarm popup that stays open after the alarm clears is the same two states at work.

Changing screens changes pixels. The banner belongs to the alarm function, so it sits on top of every screen until its own clear conditions are met.

Symptoms and what each points to

• The buzzer stops, the banner keeps flashing. The press reached the buzzer's path and missed the banner's. This points to a missing acknowledge action on the button.
• The C bit reads 0 in the status monitor. The PLC write works. What is left is the alarm definition, its trigger address and active level, and the acknowledgment state.
• The C bit returns to 1 within one scan of the press. The fault condition is still true, or another rung re-sets the bit as your write lands. The banner is honest; go clear the condition.
• The message survives the screen change. Expected. The banner belongs to the alarm function, so navigation moves everything except the alarm.

Checks in order

1. Open the PLC status monitor on the C bit. Press reset and watch the bit for 10 consecutive scans. A rung that re-sets the bit fires within one scan, and a small PLC usually finishes a scan in a few milliseconds, so the watch takes well under a second. Note the result: holds 0, or back to 1.
2. Open the alarm definition in the HMI project. Read the trigger address and its active level. It must be the same C bit, active at 1, the way your logic drives it.
3. Read the reset button's configured actions and list them. In your case that list holds the PLC write and the screen change. If no acknowledge or clear alarm action appears, that is the fault. Judge the button by its configuration, never by its label.
4. Open the rung that drives the C bit. Note whether the bit is latched, set by the fault and held until a reset condition, or driven straight by the live condition.

The fix: one press does both jobs

Keep the PLC write on the button and add the alarm acknowledge action to the same press, or put a separate acknowledge button beside it. If your alarm system supports acknowledgment bits or a dedicated acknowledgment tag, use them, and map the alarm acknowledge to a PLC bit so the controller knows the operator saw the fault.

If the alarm definition's trigger points at a different address, or its active level is inverted, correct it to match the bit you watched in the monitor. An active level of 0 makes the alarm live at the exact moment you write the bit to 0.

For a latched fault, keep the latch in the PLC logic: set it when the fault occurs, clear it only when the fault is gone and the reset is accepted. The acknowledgment belongs to the operator; the bit reset belongs to the machine. Two actions, two reasons, and one button is fine as long as both actions are configured.

If the cell also needs a fault notice readable from the aisle, that job belongs on a separate shop-floor status screen, not on the alarm banner.

Example: a fault banner that stops beeping but keeps flashing

Inputs, all from the machine as built: the alarm trigger is the C bit, active at 1, set by the machine logic. The reset button writes the C bit to 0 and changes screens, and its configuration lists no acknowledge action. After the press, the buzzer is silent, the banner still flashes, and the bit reads 0 in the status monitor. Take the operator as standing 10 ft from the screen for the text check at the end.

The rules in words: a rung that re-sets the bit fires within one scan, so 10 consecutive scans at 0 settle the PLC side. The banner clears when its trigger is inactive and its acknowledgment is given, and the acknowledgment comes from a configured action. Alarm text read in a hurry wants 20 ft of viewing distance per inch of capital height, with line spacing 1.7 times the capital height.

The arithmetic:

• Scan watch: the bit reads 0 on scans 1 through 10, so the logic is holding it off and the fault lives in the HMI configuration.
• Acknowledgment: the button's action list holds the PLC write and the screen change and nothing else, so the missing piece is the acknowledge action. Add it and download the project.
• Banner text: 10 ft ÷ 20 ft per inch = 0.5 in of capital height, and 0.5 in × 1.7 = 0.85 in line spacing. The FHWA planning figure, 30 ft per inch, would allow 10 ÷ 30 = 0.33 in, but a flashing alarm is read fast, so hold the 0.5 in.

The result: with the acknowledge action added, one press writes the bit, acknowledges the alarm, returns the screen, and the banner clears on that same press. The decision rules that come out of it: bit at 0 for 10 scans with the banner still flashing, the acknowledgment was the whole problem. Bit back to 1 inside a scan, stop editing the screen and clear the fault. Banner capitals at 0.5 in or better for a 10 ft read.

Test both transitions before you call it fixed

Make the condition true: the bit goes to 1, the banner appears, the buzzer sounds. Clear the condition and press reset: the bit holds 0 and the banner clears on the press. Trigger it once more and watch it come back. Two transitions, one in each direction, are the whole test, and they take minutes. Run them with the machine in a safe state, and never delete the alarm or force the banner off to keep a line moving. If the banner still flashes while the bit reads 0 after these checks, go back to the alarm definition and read the trigger address and active level again; a mismatch there keeps a dead alarm on the screen.

Recommendation: keep the PLC write on the reset button and add the HMI alarm acknowledge action to the same press. Watch the C bit for 10 consecutive scans after reset: 0 every scan with the banner still flashing means the acknowledgment was the whole problem, while a return to 1 inside a scan means the fault is still live and the screen is doing its job. Size the banner at 0.5 in capitals for a 10 ft read, and close the job out with two transitions.

Questions