HMI screen goes blank and freezes: check power first

Why does an HMI screen go blank and freeze?

Start with the supply. This class of panel wants a dedicated feed rated at least 1.5 A with nothing else attached. A screen that blanks seconds into startup is usually a rail that sags before the project finishes loading.

Inside the panel a supervisor circuit watches the rail. When voltage falls below the level the processor needs for reliable operation, the circuit holds it in reset. That partial, temporary dip is a brown-out: the screen blanks, the graphics freeze, the boot starts over. Startup makes the dip likelier. A panel draws more current while the display and memory come up than it does sitting on a loaded project screen. A supply that holds its voltage with the panel idle can still sag at power-on. That is why the maker's floor for this panel class is 1.5 A on a private feed, sized for the startup transient. Your panel's documentation prints its own minimum; if that figure is higher, it wins.

Which symptom points to which cause?

Each symptom narrows the list. Blanking a few seconds into startup points at the supply. A freeze with the link LEDs still blinking points at the panel's boot path. Repeated memory or reload messages point at boots that keep getting interrupted.

• Blanks a few seconds after power-up, then runs on a retry: startup sag. The rail dips while the display and memory come up. Supply check first.
• Freezes mid-session while the link LEDs keep blinking: the panel's boot or display path stalled while the network side stayed alive. Terminals, then board seating.
• Messages about erasing memory, reloading and rebooting: the boot was interrupted before the project finished loading. Text that repeats means the interruption is still happening. A panel that loses power mid-boot can come back without its project at all; the article on an HMI losing its project after a power cycle covers getting it back.
• Status indicator green on one boot, red on the next, same wiring: the boot is not completing the same way twice. Borderline supply, or a connector that only sometimes makes contact.
• A quarter of the screen dark from the very first power-up: panel hardware, and returning that unit was the right call. A second unit failing in the same wiring moves the suspicion to what the two units share: the supply, the wiring and the board.

The room is cool and the Ethernet lights blink, so why does it still fail?

Because both observations sit in the wrong place on the list. A bench at 24 °C is far below the 0 to 50 °C ambient ratings common on industrial touch panels. Blinking link LEDs only report that the Ethernet link is up.

Heat earns its place in the diagnosis when the panel sits in a hot enclosure. In an air-conditioned room it belongs at the bottom of the list. The LEDs say more than first appears: they live on the expansion communication board, so traffic flowing while the graphics freeze localizes the trouble. The board's network side is alive while the panel's boot and display path stalls. That is the pattern a damaged or poorly seated board connector produces. Two habits waste afternoons here. Rebooting until the screen comes back restores the session and leaves the cause in place, so write down the exact message text before you cycle power again. And putting a third panel on an unchecked supply and connector puts another unit on the same fault.

How do I check the 24 V supply at the panel terminals?

Measure at the panel's own terminals, during power-up and during the failure. Keep the reading inside the input range printed in the panel's documentation for every second of the sequence. A supply checked only at idle has not been checked.

1. Read the supply nameplate for its output voltage and current rating, and confirm the panel is alone on it. The maker's floor for this class of panel is 1.5 A dedicated; your panel's documentation is the authority if it prints more.
2. Check the wiring to the panel: tight terminals, undamaged conductors, polarity per the panel's terminal markings.
3. Measure across the panel's terminals first, then across the supply's output terminals. A gap between the two readings is voltage lost in the wiring, and the gap is itself a finding.
4. Power up and keep the meter on the panel terminals through the moment the screen blanks, then through a freeze.
5. Decide from the readings. A dip that lands as the screen blanks is the fault, so correct the supply or the wiring and retest before reconnecting anything else. A reading that holds inside the documented range through three failing boots clears the supply and moves the board test to the front.

Can other loads share the panel's supply?

No. The maker's floor is a dedicated 1.5 A feed. On a 2 A supply that leaves 0.5 A of headroom (2.0 A - 1.5 A = 0.5 A). One 24 V coil drawing 0.8 A takes the draw to 2.3 A, over the rating.

That 0.5 A is the startup reserve, already spoken for. Even when the running total fits on paper, a coil's inrush at switch-on and the spikes it drops on the rail land while the panel is pulling its own startup current. The documented result on a shared feed is a panel that freezes when the coil kicks in; our article on a shared DC supply with solenoid coils freezing an HMI walks that failure and its wiring fix. Decide it once and move on: the panel gets its own supply. If the cabinet's single 24 V rail makes that impossible, the feed still has to carry the panel's full minimum with every coil's running and inrush current accounted for. Coil switching stays near the top of the suspect list the next time the screen misbehaves.

Worked example: the supply reads steady, so what comes next?

The bench panel runs alone on a dedicated 24 V DC, 2 A supply in a 24 °C room. The meter holds 24.0 V at the terminals through three failing startups. The supply is cleared and the expansion communication board is next.

Here is the chain with its numbers. A 15-inch class touch panel carries a maker's minimum of 1.5 A on a private feed. The supply is rated 2 A, so the headroom is 2.0 A - 1.5 A = 0.5 A, a quarter of the supply's rating. Nothing else is plugged into it. At idle the terminals read 24.0 V. Through three consecutive boots that end in a blank or frozen screen, the meter never moves off 24.0 V: the rail never left its level. Supply and wiring are cleared, and the application is still untouched.

Power down. Pull the expansion communication board, following the maker's installation instructions. Power up bare: the panel reaches its startup screen and holds through the same minutes that used to fail, three runs in a row. One condition changed, one result changed, so the board or its connector carries the fault. Under good light the inspection finds a single pin bent out of line. A displaced pin lands on a contact it was never meant to meet, and connector-pin faults like this are their own documented failure mode in electronics work. The behavior that follows ranges from blanking to freezing to messages that name memory and reloading.

Close it out the same way. Reinstall the board only if every pin is straight and the connector seats with normal force. Rerun the failing sequence with the board in, and the fault is closed when the full sequence runs clean start to finish. If the panel had failed bare, the next stop would be the terminals again, then support with the evidence.

Can a bent pin on the expansion board blank the screen?

Yes. One bent pin on an expansion communication board produced blanking, freezing and reload messages on a panel that had already been reseated several times. The strange behavior stopped once the damaged connection was found and corrected.

Reseating without inspecting can miss it, and each reseat works the pin further out of shape. The test is cheap and settles the question:

1. Power the panel down before you touch the board. That is the safety step; a board pulled live can damage more than a pin.
2. Remove the board per the maker's installation instructions.
3. Inspect the connector and every pin under good light, magnifier if you have one: bent, misaligned, or riding outside its mate.
4. Power the panel with the board out and run the failing sequence: startup, the minutes where it blanked, the messages.
5. Change nothing else. Same supply, same application, same wiring; board in and board out is the only variable, or the comparison is worth little.
6. If the panel runs clean bare, reinstall the board only when the inspection passed and it seats with normal force, then bring the PLC links back one at a time. If one link drops on its own schedule after that, that is a separate fault, and the article on an intermittent PLC communication timeout picks it up from there.

Board in fails, board out runs clean, pins inspected: the board or its connector is the part. Bent pins are replaced or repaired the way the maker documents it, never with an improvised bend at the bench.

When do I stop testing and send the panel in?

Stop when terminal voltage holds inside the documented input range through failing boots, the panel boots clean with the board out, and the fault returns when the inspected board goes back in. That combination points at the board or the panel, and support should see the evidence.

Also stop on physical findings: pins you will not repair in the field, a connector that no longer grips, a display section dark from the first power-up. Keep the unit and the board powered down once inspection shows damage, because more bench cycles can only make it worse. Support can separate a board fault from a panel fault in minutes with the measurements you already hold. Without them, the first question you get is the one you just answered, and the clock runs while you go measure it.

Recommendation: Order the checks supply, then board, then application. Measure at the panel's terminals through a failing startup on a dedicated feed sized to the maker's minimum. If the reading sags as the screen blanks, fix the feed and retest. If the reading holds, power the panel with the expansion communication board removed. A clean bare boot puts the fault in the board or its connector, and an inspected board that fails again goes back for replacement under the maker's documented process. Leave the application untouched until a bare-board boot reproduces the fault, because a project reload can cost you the build while the real cause stays in the hardware.

Once the cabinet work is finished and a separate floor notice board is on the list, our shop-floor screens are built for that job, and the andon board guide covers where a small screen earns its place on the floor.

Questions