A panel that needs its power cycled every few days has an electrical fault first and a software fault second. Check the ground termination, then the feed, then drive noise, in that order. A relay that removes power for 30 seconds every 3 days is a last resort, fitted only after the checks and a support run come up empty.
A graphic panel that runs clean for days or weeks and then needs someone to pull the plug has an electrical cause before it has a software one. Work the checks in order: ground termination first, then what actually feeds the panel, then noise from the drives around it. If the panel still locks after all three and a run through the panel maker's support, a small timer relay that removes power for 30 seconds every 3 days holds most of these installations, and it is a stopgap you log, never a repair you file.
What the symptoms tell you
A recurring lockup leaves a pattern, and each function that survives narrows the fault. Write down what still works at the minute of the next one.
- Power and CPU LEDs lit, screen blank. The panel has power, so the lock sits inside it. A dead outlet or a blown fuse would put those LEDs out too.
- Ports and remote access dead at the same time. One event took the whole interface down. Leave the router configuration alone; a network change cannot reach a panel that is blank on its own screen as well.
- Data logging from the controller keeps running. Part of the panel's firmware is still alive. Treat that as one surviving task and check the display, each port and each remote path on their own.
- Power off for about a minute brings everything back. The restart clears the lock; the cause stays in the wiring or the supply. Log the date and the hours run between lockups, because that interval decides every step after this one.
- The lock repeats every few days. That is the signature that points at the supply path, and it is the case this page works.
A first-time blank screen with no history is a different job; the blank-and-freeze checklist covers that triage. A lock that repeats on a short fixed interval, every 50 seconds for instance, points at a polling or timing loop, and a guide to viewer freezes that repeat on a fixed schedule walks that path.
Check the ground termination first
Plenty of recurring cases end here. The panel's ground terminal has to bond metal to metal at a ground point that carries, and paint, anodizing or corrosion under the lug turns the bond into a resistor.
With the supply off, measure from the panel's ground terminal to the supply's ground bus. A sound bonding jumper reads in milliohms, so anything approaching 1 Ω means a bad termination, a corroded lug or a conductor landed on a painted surface. Re-terminate on stripped bare metal and measure again.
NFPA 79, the US electrical standard for industrial machinery, requires machine electrical equipment to carry an equipment grounding conductor back to the supply grounding point. Give the panel's ground terminal its own conductor landed on bare metal. In the field, moving the ground to a proper point has ended panels that locked 3 to 4 times a month.
Check what feeds the panel
Identify the exact feed before anything else. Two arrangements are common: the panel maker's AC adapter on a 120 V branch, or a dedicated 24 V DC supply. The adapter built for this class of panel carries a line-loss sense circuit: when input power drops, it commands a controlled shutdown so the panel closes its files cleanly. Noise on the sense path or a poor ground makes the adapter announce a power failure that never happened, and the panel shuts down or locks with the line still live.
Write down what feeds the panel and every other load on that feed. A filtered 120 V branch shared with the controller is a common layout, and the controller riding clean through the same disturbances is expected: the two devices trip at different margins, so a healthy controller clears nothing about the panel's feed. Put a recording meter on the branch and read the voltage at the minute of the next lockup.
Closer symptoms back the supply theory: a backlight that flickers when the panel polls sits in the same fault family.
Check nearby drives and noise
Variable frequency drives, solenoid valves and contactor coils are the usual noise sources in a cell that locks a panel. Note what each of them was doing at the minute of a lockup: a fault that lands at spindle start or at valve switching is a noise signature, and a fault with the cell idle sends you back to the ground and feed checks.
Keep the panel's supply wiring at least 300 mm, about 12 in, away from drive output cables, and cross any unavoidable crossing at 90 degrees; those are the standard EMC layout rules for control wiring in a cabinet and on a machine. Fit suppression across solenoid and contactor coils. The same fault one cabinet over, a PLC that restarts when solenoid valves switch, follows the same suppression and grounding steps this job needs.
Test a dedicated 24 V DC supply
If the ground bond is sound and the meter shows a clean branch, take the adapter path out of the circuit. Feed the panel from its own 24 V DC supply, sized to the current printed in the panel's documentation, and change nothing else: same controller wiring, same network, same program, so the supply is the only variable under test.
Run the test through one full old fault interval before judging it. A panel that locked twice in six months runs on a 90-day interval, so a clean month is a partial result and 90 clean days is the shortest honest test. The field record splits both ways, which is why the order matters: some of these panels stopped locking after the ground correction alone, and others kept locking after the supply swap, so keep the ground and noise files open until a full interval passes clean.
When a recycle relay is the right last step
Fit the relay only after three things are true: the ground, feed and noise checks came back clean, the panel maker's support has been through the case, and the panel still locks. Then a programmable relay or a small timer PLC on the panel's DC feed is the arrangement that holds these installations, with two settings that hold in the field: an off time of 30 seconds and a recycle interval of 3 days.
Three rules for the wiring and the schedule:
- The relay interrupts the panel's feed only. Never a safety circuit, never the controller's supply.
- Schedule the recycle for a quiet minute in the shift, because the 30-second blank must not catch an operator mid-task on a running line.
- Do not trigger the recycle from the panel's communications heartbeat. The bit stays on while any exchange succeeds, so a frozen screen with live communications looks healthy to it and never fires the relay.
To see the frozen screen from the controller's side, put each connected device's clock on a panel screen and read those values back into the controller logic, one value per device. A value that stops advancing marks the device whose path froze. Values that keep advancing while the screen stays blank tell you the lock is in the display, and that is the fault the relay exists to clear.
Example: a recycle schedule for a panel that locked twice in six months
Inputs, each from the site's own records or the field schedule above:
- Fault history: 2 lockups in 6 months, each one needing a site visit, so one lockup every 90 days.
- Relay settings: power off 30 s, recycle every 3 days.
- Hand recovery measured at the site: power off for about 1 minute brings the panel back clean.
Rule in words: pick a recycle interval much shorter than the expected gap between lockups, then count the planned downtime the schedule adds and set it against the trips it removes.
Arithmetic:
- Recycles per month: 30 days ÷ 3 days = 10 recycles.
- Planned downtime: 10 recycles × 30 s = 300 s = 5 minutes a month.
- Resets between expected faults: 90 days ÷ 3 days = 30 resets before the next lockup would come due.
Result: 5 minutes of planned blank panel a month, against roughly one unplanned trip every 90 days.
Decision: with this history, the 3-day schedule stands. If a lockup still lands between recycles, the interval is too long for this panel, so halve it and re-check. If the panel ends up needing a recycle every shift, the relay is masking an electrical fault: close the relay file and go back to the ground, feed and noise checks. Keep a log line for every recycle while it runs, because that log is what tells you whether the schedule holds or hides.
The order that decides it
Recommendation: Fix the electrical path before you automate around it. Verify the ground bond to the supply ground bus, list every load sharing the panel's feed, separate its wiring from drive output cables, and run the dedicated 24 V DC test through one full fault interval. If the panel still locks after that and after a run through the panel maker's support, fit the timer relay at the schedule worked above, log every cycle, and shorten the interval only when a lockup lands between recycles.
Questions
Is a 5-second power blip enough to clear the lock?
A 5-second blip has cleared lockups where the communications side stayed alive. The full blank-screen lock in this fault class needed power off for about a minute, long enough for internal supplies to drain and the panel to boot clean. Hold the relay's off time at 30 seconds or more, and time one hand cycle at the site so you know the panel's own boot time before you set a shorter schedule.
Which side of the supply should the relay interrupt?
The panel's own DC feed, after its dedicated supply. A relay on the 120 V branch drops every other load on that branch at the same time, which turns one planned 30-second blank into a whole-cabinet outage and defeats the point of the schedule. Keep the controller, the network gear and any safety circuit outside the relay's contacts.
What do I log while the recycle relay is running?
Date and time of each cycle, whether the panel was running clean before it, and what the drives in the cell were doing at that minute. Add any operator report of odd behavior between cycles. The log is what tells you the schedule holds or hides a fault that is still in the wiring, and it is the record the panel maker's support will ask for.
Planning information, not legal or engineering advice. Rules differ by place: check codes and permits with your local authority and your licensed trades.