An HMI that locks up overnight on its AC adapter but runs clean on a DIN-rail 24 VDC supply points at the adapter's power path, not the panel. Keep the panel on the 24 VDC supply, measure voltage at both ends under load, bond the supply's DC common, and separate panel wiring from 120 VAC and drive wiring before you replace the adapter.
A panel that dies every night on its AC adapter and runs clean for a week on a DIN-rail 24 VDC supply has told you where the fault lives: in the adapter's power path, not in the panel. Keep the DIN-rail supply feeding the panel while you work, and repair the adapter path in this order: measure voltage at both ends of the run under load, cut out and remake any joint that drops more than a tenth of a volt, bond the supply's DC common to ground the way its instructions show, and separate the panel's wiring from 120 VAC and drive wiring. Judge the adapter itself only after those checks come back clean, because in the field this pattern almost always ends at a joint, a missing bond, or coupled noise, and only rarely at the adapter.
Symptoms and what each points to
- Locked every morning on the AC adapter, healthy on the DIN-rail supply. The fault rides the adapter path: the adapter model, its wiring and joints, its grounding, or noise coupled into it. Work through the checks below in order.
- Locked no matter which supply feeds it. Power is cleared as the cause. Record what the screen last showed and what the machine was doing before each lockup, and take that record to the panel maker's support channel.
- Backlight flickers or dims when the panel polls. That points at supply sag or an overloaded shared feed. HMI backlight flicker points at the power supply handles that fault on its own.
- The lockup lands at about the same time each night. Look for a scheduled load that switches then: a compressor, a pump, a drive dropping to idle, a lighting circuit. A fixed-time fault has a fixed-time cause nearby.
- A power cycle brings it back for a full day. The reset is recovery, not a test. The window that exposes the fault, normally the overnight one, is the window a fix has to pass.
Measure voltage at both ends of the power path
Put the panel back on the adapter and let it run its normal application, not an idle menu screen. Measure DC voltage twice with the panel under that load: once at the adapter's output terminals, once at the panel's power terminals. Then compare.
A supply that holds a healthy voltage at its own terminals while the panel sees several volts less has a failed connection or a dying joint somewhere in the run. A reading that sags at the adapter terminals themselves, under the same load, is a failing or undersized supply. The extreme version shows up as 21 V at the supply terminals with 10 V at the panel terminals: a connection has opened partway, or the supply is collapsing under load.
Sound terminations measure in milliohms, so at the half amp a small panel draws, a good joint drops millivolts. A gap of 0.1 V or more between the two meter points is a joint to cut out. Work from the panel end back toward the supply, measuring across each terminal in turn, and the meter points at the joint that owns the drop.
Example: a 2.5 volt drop that found a 5 Ω joint
An 8-inch HMI has locked every morning for weeks when fed by its 120 VAC plug-in adapter, and has run clean for a week on the DIN-rail supply. The panel goes back on the adapter for the test, running its normal screens.
The meter reads 24.1 V at the adapter output and 21.6 V at the panel terminals. The missing voltage is 24.1 V - 21.6 V = 2.5 V. The panel's label lists 0.5 A at 24 VDC, so the resistance doing the dropping is 2.5 V ÷ 0.5 A = 5 Ω.
The wire cannot own that. The feed is 10 ft of 18 AWG from supply to panel, two conductors, 20 ft of copper. 18 AWG copper measures 6.4 Ω per 1000 ft, so the run is 20 ft × 6.4 mΩ/ft = 0.128 Ω, dropping 0.128 Ω × 0.5 A = 0.06 V at that load. That leaves about 2.4 V, close to 4.9 Ω, sitting in a terminal, crimp or splice.
Both terminations at the panel end get cut out and remade. The panel terminal then reads 24.0 V on the same load, and the panel holds through the next two overnight windows on the adapter. The joint was the fault. The adapter never was.
Ground the DC common
The controls industry's noise-control design guides are blunt about this: grounding a control power supply's DC common knocks common-mode noise down by a wide margin. Many DIN-rail supplies carry a DC-common bond point or jumper for exactly this, and the installation instructions show where it goes. Fit that bond with the supply powered down, as the instructions for the supply and the panel show it.
If the signal cable between panel and controller is shielded, terminate the shield the way the equipment instructions show. A shield cut back and taped under the jacket is just extra jacket.
One boundary on this work: bonding touches the site's grounding design. Make the bond the instructions show, and leave changes to the building's grounding arrangement to the drawings that own it.
Separate panel wiring from drives and 120 VAC
Drive wiring application notes put hard numbers on separation. Where signal or low-voltage control wiring has to run parallel with power wiring, hold 2 inches between the two at minimum, and 2.5 to 3 inches is preferred on parallel runs carrying 120 VAC. Keep panel power and signal wiring out of the same conduit or wire trough as drive input and output cables entirely, and cross unavoidable crossings at right angles.
Distance from the drive matters as much as spacing in the trough. A plug-in adapter sitting beside a drive, its leads run alongside the motor cable, sits inside the drive's field. Move the adapter and its wiring clear of drive cable runs, and cut excess lead length instead of coiling it. A coiled lead is a coupling coil.
Check the firmware revision and the adapter model
Two paperwork checks come out of the panel's documentation. First, check the adapter is the model the documentation lists for that panel; a generic adapter with the right voltage on the label is not automatically the same part electrically. Second, read the installed firmware revision, where the panel prints it, and compare it with the revision the documentation requires for adapter operation. If the panel is behind, update through the maker's own documented procedure; checking the HMI firmware revision before a project transfer walks that comparison.
Set expectations while you are in there. A firmware update clears logic faults, and if the release notes list a watchdog it adds recovery after a hang. It cannot strip noise out of a power path. The wiring fixes do that work; the update is housekeeping beside them.
The fix, in order
- Feed the panel from the DIN-rail 24 VDC supply that has held, and label the adapter path out of service so nobody restores it mid-repair. If the panel joins a supply that already feeds relays or lights, size it for the panel's inrush first; sizing a separate 24 VDC supply for HMI inrush current shows the arithmetic.
- Check the adapter model and the required firmware revision against the documentation, and update through the maker's procedure if the panel is behind.
- Measure at both ends under load and cut out every joint that shows a drop, per the section above.
- Bond the DC common as the supply's instructions show.
- Re-route per the section above: no shared conduit with drive cables, right-angle crossings, adapter and its leads clear of motor cable runs.
- Return the adapter to service and run three consecutive overnight windows, the window that produced the fault.
- Verify the link, not just that the screen lit up in the morning. A PLC heartbeat counter that increments through the whole window is the verification; verify HMI communications with a PLC heartbeat counter sets the counter up.
The decision for each case
- It locks again on the adapter, and the DIN-rail supply holds. The adapter path has failed the comparison with joints, bond and spacing all corrected. Treat the adapter as defective. The DIN-rail feed becomes permanent, and a replacement adapter, the model the documentation lists, re-earns service only the same way: three clean overnight windows.
- It holds three overnight windows after the wiring fixes. Fixed. Write down what changed: the two meter readings before and after, the bond, the routing. The next panel on the floor inherits the pattern.
- It locked on both supplies from the start. Power is out of the picture. Take the record of panel state, timing and machine activity to the panel maker's support channel.
- The documentation lists no adapter for this panel. Then there is nothing to repair. Run it on the 24 VDC supply and close the job.
Recommendation: Keep the panel on the DIN-rail 24 VDC supply for the whole job. Run the two-end measurement first, because a joint explains more of these lockups than an adapter does, then bond the DC common and set the spacing the drive wiring application notes call for. Return the adapter to service only after a lockup has been reproduced with all of that corrected; at that point call the adapter defective, make the DIN-rail feed permanent, and fit the adapter model the documentation lists when you want the spare.
Questions
Why does the HMI lock up overnight instead of during the day?
Overnight faults usually ride on a scheduled event, not on darkness itself: a compressor or pump starting on a timer, a drive dropping to idle, a lighting circuit switching, a heater cycling. Note the clock time of each lockup and what starts or stops within a few minutes of it. A fault that lands inside the same quarter hour every night is handing you its cause's schedule, and the fix is to clean up that load's wiring or move the panel's wiring away from it.
Can the HMI stay on the DIN-rail 24 VDC supply permanently?
Yes, and many installs end the job that way. If the panel's documentation lists a 24 VDC input, a supply that carries the panel's running current plus its inrush, with headroom for anything else on the same supply, is a permanent feed, not a stopgap. Keep the old adapter as a spare only if it later passes the same overnight test after the wiring repairs. If the supply also feeds other loads, recheck its total budget before adding anything.
What should go in the record if the panel locks up again?
Enough for the next person, or for the panel maker's support channel, to reproduce it: which supply fed the panel, the voltage measured at the supply and at the panel terminals under load, the installed firmware revision, the date and clock time of each lockup, what the screen last showed, and what the machine was doing. Add the wiring changes already made, with dates. One change at a time is what makes the record readable; three changes in one night explain nothing.
Planning information, not legal or engineering advice. Rules differ by place: check codes and permits with your local authority and your licensed trades.