Tell a frozen HMI panel from a dropped PLC link

7 min read · Published November 12, 2026

Two tests separate the faults: press a screen-change button, then read a live value on an independent path such as a PLC heartbeat tag. A panel that answers touch but shows stale values has a communications fault; a panel that ignores touch has a power or panel fault. Meter the 24 V DC feed at the terminals during the fault before touching the network.

A frozen panel and a dropped PLC link look identical from across the line: the numbers stop moving. Before you change anything, find out which fault you have. Two tests do it. Press a screen-change button and see whether the display still answers. Then read one live value somewhere the panel is not involved, such as a heartbeat tag in the PLC programming tool. A screen that answers touch but shows stale values has a communications fault, and the work is on the cable, the switch port or the PLC's Ethernet card. A screen that ignores touch has a power or panel fault, and the work starts at the supply feeding the panel. This page gives the decision path, the supply measurements to take during the fault, and an order of checks that keeps every result usable.

Two tests tell the faults apart

An HMI panel draws its own screens and runs on its own processor. The PLC link only feeds it values. That split is why one dead-looking screen can hide two very different faults.

First test, touch. Press a button that changes screens, then a numeric entry that opens a keypad. If the display answers, the panel's processor, touch layer and power feed are alive enough to run the interface.

Second test, values. Read a live value without the panel. The cleanest tool is a heartbeat: a tag the PLC adds 1 to every second, visible both in the programming tool and on the screen. If the number climbs in the programming tool but sits still on the panel, the link dropped and the panel is fine. If it stops in both places at once, suspect the PLC side or a shared feed. The setup is described in verify HMI communications with a PLC heartbeat counter.

The same comparison scales up. SCADA software raises stale-data alarms on frozen PLC tags when one value stops moving while the rest of the system keeps running, and that alarm is the plant-wide version of the test you just did on the panel.

Recommendation: when the screen answers touch but the values are frozen, leave the panel alone. Meter the supply at its terminals during the fault, then work the link checks. When touch is dead, treat it as a power or panel fault and keep the switch configuration out of it until the feed measures clean.

Symptoms and what each one points to

Write the symptom down before the first fix attempt. Each pattern points somewhere specific:

  • The whole screen is static and touch is dead on every control. The panel or its power feed. Start with the supply measurement below; the network is not the first suspect.
  • Touch works and every value sits at the same number. The communications path: the cable, the switch port, the PLC's Ethernet card, or the panel's communication settings. The panel itself is running.
  • Touch works but values arrive seconds late. A slow link or a busy panel. Blinking or animated objects can drag touch and update response down; blinking text that slows HMI button response shows how to measure that before resetting anything.
  • One screen or one function stalls while the rest runs. A reproduction case. Note which screen was open and what it runs, such as trend graphs or alarm logging, and try to bring the freeze back with that screen alone. A restart wipes the setup, so capture it first.

Also record the time, the machine state and the supply reading at the moment of failure. Those three lines turn a recurring annoyance into a comparable event.

Check the 24 V DC feed at the panel terminals

Most HMI panels run from a 24 V DC control supply, and a sagging feed can look like a freeze: far enough down, the panel resets or drops its communications while the screen still shows a picture.

Measure at the panel's DC terminals, under load, with the fault present. A reading taken at the supply on a good morning tells you nothing about the moment of failure. The band to measure against comes from IEC 61131-2, the input-voltage standard for industrial control equipment: a 24 V DC rating covers operation from 20.4 to 28.8 V, and supplies built for control panels print the same band on the label. If the panel manual states its own input range, the manual's numbers are the ones to use.

Take two readings during the fault: one at the supply output, one at the panel terminals. The difference is the drop across the cable, terminals and connectors on the way. Then list everything else that shares the supply. A load that kicks in mid-shift can pull a weak feed below the panel's limit, and a terminal that loosens as it warms does the same job more slowly.

A lockup that follows the power path, such as a panel that locks up on AC power but runs on 24 V DC, is diagnosed the same way: readings at the terminals, under load, with the fault present, and one change per test.

Example: a 24 V DC panel that freezes mid-shift

Inputs.

  • Nominal feed: 24 V DC, one control supply shared by four panels in the cell. This is the common arrangement for machine-mount panels.
  • Operating band: 20.4 to 28.8 V DC, the IEC 61131-2 operating range for a 24 V DC industrial control supply. The panel manual's printed range governs if it is narrower.
  • Meter readings, taken in this scenario while panel 2 was frozen: 24.6 V DC at the supply output and 23.8 V DC at panel 2's terminals.

Rule. During the fault, the voltage at the panel terminals must sit inside the operating band. The difference between the two readings is the drop in the wiring between the supply and the panel.

Arithmetic.

  • Wiring drop: 24.6 - 23.8 = 0.8 V DC.
  • Band check: 20.4 ≤ 23.8 ≤ 28.8, so the terminal voltage is inside the band.
  • Headroom above the low limit: 23.8 - 20.4 = 3.4 V DC.

Result. Panel 2's feed held 23.8 V DC with 3.4 V of headroom while the screen was frozen. Power is not the cause, so the work moves to the link checks.

Decision. Terminal voltage inside 20.4 to 28.8 V DC during the freeze: skip the power work and go to the link. Terminal voltage below 20.4 V DC: repair the power path first and re-run the fault before touching the switch. The contrast case shows why each panel needs its own reading: a fourth panel on the same feed measured 19.1 V DC at its terminals during its own freeze, which is 1.3 V below the limit (20.4 - 19.1 = 1.3). That one is a power fault: check its terminals for looseness, the conductor length back to the supply, and which shared load kicks on with it.

When the panel is alive and the values are frozen, work the path in this order.

  1. Physical layer. Link lights at the panel's port, the PLC Ethernet card and every switch port in between. A dark light ends the guesswork about which hop dropped. Swap in a known-good cable before anything cleverer.
  2. Counters. The panel's Ethernet diagnostics and a managed switch keep error and statistic counters per port. If your panel and PLC talk EtherNet/IP, ODVA's troubleshooting guide lists the link counters to pull when a connection misbehaves. Snapshot them during a freeze and again after recovery, and compare the two.
  3. Traffic. If machine cells and office PCs share one chain of unmanaged switches, every broadcast on the network lands on the panel port. Multicast goes further: when a switch has no multicast filtering configured, traffic meant for one group of devices floods out of every port, and the panel processes traffic it never asked for. A managed industrial switch that filters multicast keeps group traffic off the panel's port. Configure it with the switch manual open, verify the PLC-to-panel path afterward, and change one setting per test.

If several PLCs feed one panel, snapshot each connection's own diagnostics so the freeze points at the link that actually dropped.

The fix and the retest

Make one change per test, and keep the notes with the panel's record.

  • Supply or wiring fault measured: repair the loose terminal, the undersized conductor or the shared-load problem, then re-run the same fault conditions with the meter still connected.
  • Traffic shown by counters: configure multicast filtering or move the machine cells onto their own switch, then confirm the PLC-to-panel path before production leans on it.
  • Clean feed, clean counters, freeze still reproduces: the fault is inside the panel or its firmware. That goes to the panel vendor with your notes: symptom, timestamps, terminal voltages and counter snapshots.

Retest by running what ran before the freeze: same part, same screens, same load on the supply. Watch four things together: touch response, value updates, terminal voltage and the switch counters. The job is done when the panel holds through a full shift with touch answering, values moving, the feed inside the band and no new counter errors.

One thing not to do while the panel is down: reach for a signage display as a stand-in. A shop-floor signage screen plays its own files offline and has no input from your PLC, so it cannot show the values the operator needs. Fix the path; the sign and the panel are different tools.

Questions

Can heavy Ethernet traffic really freeze an HMI panel's screen?

Traffic is a suspect, and the counters decide. The pattern that matters is broadcast and multicast landing on the panel port: when a switch has no multicast filtering configured, group traffic floods out of every port and the panel processes traffic it never asked for. Snapshot the port's error and broadcast counters during a freeze, isolate the traffic with managed-switch filtering, then compare the same counters after. If the counters were clean through the freeze, look elsewhere.

What if the panel runs from an AC adapter instead of a 24 V DC feed?

The check is the same, one step earlier: measure the adapter's DC output at the panel end, under load, during the fault, against the input range printed in the panel manual. An adapter sharing a receptacle with heavier gear, or feeding a long DC cable, can sag under load. A panel that misbehaves on the adapter and runs cleanly on a stable 24 V DC feed has answered the question: the adapter or its wiring is the fault.

A power cycle brings the screen back. Is that a fix?

No. A restart restores the screen and wipes the evidence: the touch state, the frozen values, the counter snapshots and the supply reading at the moment of failure. Take the measurements first, power-cycle only when the machine is in a safe state, then run the same conditions again with the meter still connected. A fault that returns next week with no data behind it is a fault you have to chase from scratch.

Planning information, not legal or engineering advice. Rules differ by place: check codes and permits with your local authority and your licensed trades.

Keep reading

Build your signAsk for a quote