A screen that shows its boot screen after an interruption lost power and started again: treat the event as a supply dip and check the power path first. Measure the 24 VDC at the panel terminals with min/max capture, audit everything drawn from that supply, and log reset times against the shared AC feed, the shift schedule and radio use. One panel resetting alone points at its own DC path; both together point at what they share.
An HMI that shows its boot screen after an interruption has lost power and started again. That initializing screen decides the whole investigation: treat the event as a supply dip or interruption and start in the power path. The causes that turn up most often are an overloaded or marginal 24 VDC supply, a load wired in after commissioning, a disturbance on the shared AC feed, and a handheld UHF radio keyed close to the panel. Work them in the order below, starting with proof that the event is a restart at all.
Confirm it is a restart
Get the exact sequence from the operator before touching anything. A panel that displays a boot or initializing screen after an interruption has restarted: its logic lost supply and came back. A screen that goes dark and comes back with the process values still current and no boot sequence has stayed powered and lost its display path. That second fault lives in the backlight, the video feed or the cable, and HMI screen goes blank and freezes: check power first walks through it.
Start a log at the same time: time to the minute, which panel, whether a boot screen appeared, and what else was running. Radio traffic, a motor starting, welding down the aisle, storm weather, scheduled maintenance. Five resets in one morning and then a quiet week is a pattern, and the log is what shows it.
What each symptom points to
| What you find | Where to work |
|---|---|
| Boot screen after every event | Power loss. The DC supply, its wiring and everything drawn from it |
| Both panels reset within seconds | What they share: the AC branch, its protection and wiring, the bonding, or a disturbance that reached both |
| One panel resets, the other stands | That panel's supply, connector, leads and any load added to them |
| Resets cluster at a set time of day | Something on a schedule: a machine starting, radio traffic at shift change, HVAC, maintenance |
| Dark and back, no boot screen | The panel stayed powered. Backlight, video or cable |
| A reset follows a radio key-up | RF pickup into the leads or connectors. Fix the wiring geometry |
Measure the 24 VDC at the panel terminals
Measure at the terminals on the back of the panel, with the process in its normal running state. The pass mark is the input range printed in the panel's manual. Most 24 VDC control equipment works in the band IEC 61131-2 gives for that class of gear: 20.4 to 28.8 V, which is 24 V from minus 15 percent to plus 20 percent. A healthy reading at idle settles nothing, so take the minimum while the panel and everything on its supply are running.
A handheld meter refreshes its display a few times a second, so a dip of a couple of mains cycles never reaches the digits. Use the meter's min/max capture, leave the leads on for a full shift, and read the minimum after the next reset. A minimum that holds above 20.4 V through a reset shifts the search to a fast transient, the shared path or the radio; dips quicker than the meter's capture window need a recording meter or a logger.
If the DC reads low, walk back toward the supply: measure at its output, at any intermediate terminal blocks, and at its AC input. Reseat the DC connector and the terminal screws with the circuit isolated under the site's electrical safety procedure, and look for discoloration that marks a resistive joint. A supply on the edge often announces itself first as a dimming screen; HMI screen dim on PLC power: check the supply drop works through that measurement.
Audit what is drawn from that supply
A supply labeled dedicated stays dedicated only until somebody taps a new device into it. Trace the conductors from the supply output and list every load on it. Hold the connected load at or below about 80 percent of the supply's rating, the sizing practice that leaves regulation with margin, and judge your list against that number. Resets from an added load follow a sag the panel cannot ride through, and the timing matches the added device's operation.
A load that starts while the panel is running pulls the bus down the same way; why a shared DC supply with solenoid coils freezes an HMI takes that case apart. If the audit turns up an added device, move it to a supply of its own or upsize the existing one; sizing a separate 24 VDC supply for HMI inrush current covers the inrush margin the new supply needs.
Example: an added load drags one panel under its floor
Scenario: a control-room panel resets two or three times a week, each time showing the boot screen. Its supply is labeled dedicated and reads a healthy 24.1 V at idle.
The loads:
- The panel draws 15 W at 24 V: 15 / 24 = 0.63 A.
- The supply is rated 24 V, 1.0 A.
- A sensor hub drawing 0.8 A was wired to the supply's output terminals six months after commissioning, so the run to the panel now carries everything.
Total draw: 0.63 + 0.80 = 1.43 A, which is 43 percent over the 1.0 A rating.
The cable adds its share. The supply sits 5 m away on 22 AWG copper (about 54 mΩ/m), so the 10 m loop measures 10 × 0.054 = 0.54 Ω. At 1.43 A the drop along the loop is 0.54 × 1.43 = 0.77 V, call it 0.8 V.
Say the strained supply holds 21.0 V at its output while both loads run; your meter replaces that illustrative figure with the real one. The panel sees 21.0 - 0.8 = 20.2 V, under the 20.4 V floor of the normal band, and it restarts. That restart is the boot screen the operators report.
The decision: move the 0.8 A hub onto a supply of its own, or fit one supply sized for the full 1.43 A at the 80 percent rule: 1.43 / 0.8 = 1.79 A, so the next standard size up, a 2 A unit. After the change, min/max at the panel terminals should hold inside 20.4 to 28.8 V across a week of running, with no boot screens in the log.
Both panels drop together: work the shared path
Two separate DC supplies failing in the same second is the unlikely story. The likely one is something both panels share: the AC branch feeding both supplies, its breaker and wiring, or the bonding that ties the system's references together.
Compare the AC feeds first: same panelboard, same branch, same conduit. Ask what else stopped at the reset minute: a light, a PC, a network switch, a door controller. Then ask what work runs on a schedule: welding, motor starting, utility switching, storm weather. Have the bonding path inspected under the site's electrical procedure, looking for a joint that has loosened or changed since commissioning.
A fault that clears with a power cycle and returns on a pattern usually ends in the supply or ground path; HMI panel needs a periodic power cycle: check supply and ground follows that same trail. Storm damage and utility switching land on that path too; HMI blank screen after a storm: check supply and grounding first starts from that event.
Test the radio theory
If the site runs handheld UHF radios, log radio traffic beside the reset times. Land-mobile handhelds transmit in the 400 to 470 MHz band, and RF at those frequencies gets into a panel through its DC leads, long cable runs and connector seams. Installs with keyed radios beside the panels have produced exactly this fault: healthy wiring, healthy supply, reset on key-up.
The test takes minutes. With someone watching the screen, key a handheld within 30 cm of the DC terminals for ten seconds, then repeat along the cable run and at the display face. A reset that follows a key-up settles the cause.
The fix is physical: shorten the DC leads and run them as a pair, route them away from where radios get used, and clip snap-on ferrite cores over the leads at the panel end.
The fixes, in order
- Reseat the DC connector and terminal screws with the circuit isolated. It is the quickest check and a loose joint is a common finding.
- Audit the loads against the rating and move or upsize per the example.
- Clear the shared path: a clean AC branch for the control-room panels and a bonding inspection.
- For radio pickup, fix the lead geometry and add ferrites.
- Keep the log running for a week after each change, so the next reset, if one comes, arrives with evidence attached.
If this job leaves you wanting a notice board beside the controls for shift messages and plant notices, that is a different screen from an HMI. Our shop-floor displays are built for it: offline playback from a card or USB drive.
Recommendation: run the event log and min/max capture for one week before replacing any hardware. If one panel resets alone, audit its supply for loads added after commissioning and move them off or upsize so the connected load stays at 80 percent of rating or less. If both panels reset together, put the effort into the shared AC branch and the bonding. If resets line up with radio key-ups, fix the lead geometry and add ferrites before buying anything.
Questions
How long can a 24 VDC supply ride through a mains dip?
Industrial switching supplies typically specify a hold-up time of about 16 to 20 ms at nominal input, roughly one cycle of 50 or 60 Hz mains. A dip shorter than that hold-up never reaches the DC side. A sag lasting several cycles passes straight through, and the panel sees the undervoltage. That is why a brownout of a second or two resets a panel even when the DC bus reads 24 V at every visit.
Does a battery backup on the 24 VDC side stop the resets?
It stops the resets coming from the AC side: a backup-capable DC supply carries the panel through upstream sags and short outages. A backed-up panel that sits quiet while its neighbor resets points the search at the shared AC feed. Backup changes nothing on the DC side: an overloaded supply, a loose terminal or RF pickup from a radio still pulls the panel down.
Why would a handheld radio restart a panel?
A land-mobile handheld typically transmits 1 to 5 W in the 400 to 470 MHz UHF band. RF at those frequencies couples into unfiltered DC leads, long cable runs and connector seams, and the induced voltage lands on the same supply lines the controller watches. Keying a radio 30 cm from the terminals and reproducing the reset settles the question on the spot.
Should I update the panel's firmware first?
A boot screen after each event marks a power interruption, so a firmware update rarely changes it. Update firmware once the DC measurements read clean across a week and the resets continue; at that point the evidence sits inside the panel, and the next step is a support case with the panel's maker. Keep the log through the update so the timeline stays honest.
Planning information, not legal or engineering advice. Rules differ by place: check codes and permits with your local authority and your licensed trades.