Troubleshooting
HMI display fault with PLC outputs off and I/O LED dark
Short answer
An HMI showing no live data, every digital-output LED dark and no I/O LED activity on the controller has one likely shared cause: the processor is stopped or faulted and no logic is executing. Read the controller mode and fault code first, prove the network with a ping test, then check the output module; download a backup only after the program itself is proven damaged.
On this page
- What the three symptoms share
- Work the checks in this order
- Read the controller mode and fault status first
- Worked example: ping test on a 10.0.0.x cabinet network with a dark controller
- Check the output module before rewriting logic
- Restore the communication path before touching the project
- When the backup download is actually the right fix
- If the terminal itself is the failed unit
- Prove the fix before returning to Run
- Questions asked
Three faults at once — an HMI that will not show live data, every digital-output LED dark, and no blinking on the controller's I/O LED — usually have one cause, and it is not the panel. A stopped or faulted processor stops scanning logic, so nothing commands the outputs and no live values exist to display; the terminal just sits on a screen with dead or stale data. Prove that first, before reloading anything. The order below reads the controller mode and fault status, splits network faults from device faults with a ping test, checks the output module, then restores the communication path. On the question of fixing it without the backup: yes, most machines with this symptom set come back without a download, because the program was never damaged — the controller only stopped running it.
What the three symptoms share #
Read the symptoms as evidence, not as a verdict. A screen that is lit proves terminal power and a working display; output LEDs that are all dark prove either no command or no module power; a quiet I/O LED proves no I/O data exchange at the controller.
| What you see | What it proves | Where the fault can sit |
|---|---|---|
| HMI screen lit but no live values | Terminal power and display are good; only the data source is dead | Controller mode, network path, driver target, tag bindings |
| Every digital-output LED dark | No output is being commanded, or the module has no power | Controller mode and logic, module power, I/O connection |
| Controller I/O LED never blinks | No I/O data exchange at the controller | Stopped or faulted controller, I/O connection, module, network |
Taken together, the three point hardest at a controller that is not executing logic, because that one condition produces all three at once. They do not prove a damaged program and they do not prove a damaged HMI project. A fully dark screen is a different branch — the terminal then has its own power or startup fault, which PanelView 1000 no display or dark screen troubleshooting walks as a single-symptom case.
Work the checks in this order #
- Read the controller's mode and fault status in the programming software, and record the fault code before changing anything.
- Ping the terminal and the controller from a laptop on the same switch, to split network faults from device faults.
- Compare the commanded output tags with the module LEDs, and read the module's own status.
- Check the HMI driver target and the tag bindings against the live controller.
- Download the backup only if a step above proved the program or project is the fault.
- Verify the whole chain at the machine before returning to production.
Read the controller mode and fault status first #
Connect with the programming software and read the actual mode — Run, Program or faulted — rather than judging from the front of the cabinet. Rockwell's Logix fault documentation describes the decisive case: a major fault halts logic execution, the controller switches to faulted mode and the OK LED flashes red, while the controller status display names the fault type, including I/O faults. A stopped or faulted controller still holds its program; stopped does not mean lost.
Decide from what you read:
- Faulted: the fault code names the subsystem. An I/O fault points at a module or its connection; other fault types point at logic or configuration. Repair the named cause, then clear the fault and return to Run.
- Program or otherwise stopped: logic is not executing, which on its own explains the dark outputs and the missing live data. Look for what stopped it — a power interruption, a stop command, a person.
- Unreachable: if the software cannot connect at all, check the controller's power and its network port before anything else.
The quiet I/O LED is part of the same picture, not a second fault: with logic execution halted there is no scan driving I/O activity, so the LED follows the mode. If the fault the controller reported is an I/O fault, the fault code already names the subsystem to inspect.
Worked example: ping test on a 10.0.0.x cabinet network with a dark controller #
Inputs — the addresses are illustrative, so substitute your own, read from the terminal's communications settings and the controller's network configuration:
- Terminal: 10.0.0.51
- Controller: 10.0.0.10
- Laptop plugged into the same cabinet switch: 10.0.0.99
- Subnet mask 255.255.255.0 on all three devices
- Test: four ping requests to each device from the laptop
Rule in words: a device answers a ping only if it is powered, its Ethernet port, patch lead and switch port are good, and the requester sits on the same subnet. The same-subnet test is a mask AND on each address; equal results mean one subnet with no router in between.
Arithmetic:
- 10.0.0.99 AND 255.255.255.0 = 10.0.0.0
- 10.0.0.51 AND 255.255.255.0 = 10.0.0.0
- 10.0.0.10 AND 255.255.255.0 = 10.0.0.0
All three addresses sit in the 10.0.0.0/24 subnet, so the laptop can reach both devices directly if they are powered and cabled.
Result of the illustrative run: the terminal answers 4 of 4 requests at 0% loss with round trips of 1 to 2 ms; the controller answers 0 of 4 — 100% loss.
Decision, by outcome:
- Both answer: the path is proven, so the fault sits above IP — go to controller mode, driver target and tags.
- Terminal answers and the controller stays silent, as in this run: the terminal's replies prove the laptop, switch and cabling good, so work the controller side — its power, its port LEDs and the lead into it.
- Both silent: re-check the laptop's own address and mask first, then the switch.
The threshold that makes these decisions safe: on a quiet switched cabinet link the healthy number is zero loss, and any lost request is a power, cable or device fault, not congestion. A reply proves reachability only — read the controller's mode in the programming software before touching the program.
Check the output module before rewriting logic #
With the controller mode known, go online and compare each commanded output tag with its physical LED. Three patterns cover the field:
- Tags off and LEDs dark: nothing is commanding the outputs, which is exactly what a stopped controller produces. Trace the logic conditions and permissives only after the controller runs again.
- Tag on but LED dark: the problem moved from logic to hardware. Read the module's status and connection diagnostics — an I/O connection fault points at module configuration, the network path to the module or module availability, while a healthy module points at its power or wiring.
- LED on but the load dead: the fault is field-side power, wiring or the load itself, and it has nothing to do with the HMI.
Do not force outputs as a diagnostic shortcut; command one output deliberately with the machine in a safe state and watch the module respond.
Restore the communication path before touching the project #
If the controller runs and answers, the remaining HMI symptom is a data path problem, and it divides into two different faults. A path fault means the terminal cannot establish the connection at all, so every object on every screen stays dead regardless of how it is tagged. A binding fault means the connection works but a specific screen object points at a tag or address that does not exist in the live controller.
On a PanelView Plus CE terminal, the network settings and configured devices live in the terminal's Communications control panel application, and the multi-homed device list there is part of the same check. Work one layer at a time — reachability first, then the driver target, then the individual screen objects — so a network fault is not mistaken for a project fault. When the screens come back but values display as question marks instead of numbers, that is the same path fault seen from the screen side, and PanelView Plus logic module shows question marks on live values walks that symptom.
When the backup download is actually the right fix #
The download is the answer only when a check above proved the files are the fault: the controller demonstrably lost its program, the project on the terminal is confirmed damaged, or a targeted correction cannot restore the intended configuration. In this symptom set that is the minority case — a stopped or faulted controller recovers with a mode change and a fault fix, and a lost path recovers with settings, not a reload.
If you do download, verify first that the backup matches the installed controller and hardware configuration, because a mismatched program stacks new faults on the old ones, and follow your site's change-control procedure. Restoring an archive is not always the end of the story either — Why an MER download shows no live data on the HMI covers the case where the screens come back but the data path still needs work.
If the terminal itself is the failed unit #
If pings to the terminal fail while the controller answers, the terminal's supply voltage checks correct and the terminal never reaches its runtime, the terminal — not the controller — is the dead component. Where the evidence points at the terminal and the repair is not worth it, the same operator role moves to industrial panel PCs running the HMI runtime against the same controller.
Prove the fix before returning to Run #
Confirm the whole chain before handing the machine back: live values update on the HMI, commanded outputs follow at the module LEDs, the module status reads clean and the controller shows no active fault.
One safety note before restarting: a controller that has been stopped resumes logic the instant it returns to Run, and any commanded output turns on at that moment. Confirm the machine is in a safe state and personnel are clear, then return to Run and watch the first cycle.
Questions asked
Does a ping reply prove the controller is running?
No. A ping only proves the Ethernet stack, IP address and cable path are alive; the network interface runs independently of logic execution, so a faulted controller can still answer. Use the reply to prove reachability, then read the controller's mode and fault status in the programming software before deciding anything about the program.
The controller address changed and the HMI stopped communicating - why?
A new controller address invalidates the shortcut the HMI uses to reach it, so the terminal keeps calling the old address and reports communication failure even though the project is intact. Correct the target in the terminal's communications settings rather than rebuilding screens, and ping the new address from a laptop first so you are not chasing two faults at once.
Should I clear the major fault as soon as I see it?
Read it first. The fault code and type tell you whether the stop came from an I/O connection loss, a fault in the logic or something else, and each points at a different repair. Clearing the fault erases that evidence, and if the underlying cause is still present the controller faults again the next time the condition appears. Fix the named cause, then clear and return to Run.
I restored the backup and the HMI still shows nothing - what now?
A download replaces the program and the screens, but it cannot repair a wiring fault, a dead module, a wrong network path or a power problem. If the three symptoms survive a confirmed-good download, stop treating the files as the fault and re-check the physical layers: module power, network cabling, switch ports and the terminal's own power supply.
Can an operator's button press cause trouble when communications drop?
Yes - if an operator was holding a momentary HMI button when the link dropped, the release write never reaches the controller and the bit can stay on after the connection returns. Compare every HMI-initiated bit with its screen state, clear any bit that stayed on, and check whether the logic that consumed it already acted on it.