IMDTouchKnowledge BaseShop main store

Troubleshooting

HMI Project Opens With a Repeating Error Popup: Isolate It

HMI and touchscreens Published 7 min read

Short answer

A repeating error popup that returns after you close it is a project-open fault, not a crash. Record the exact file the message names, open the unchanged project on a second PC, then remove only that file from a diagnostic copy. If the copy opens, that file carries the defect; if it still fails, build a new project and copy the file in.

On this page
  1. The popup that returns is not a crash
  2. Read the message before touching the project
  3. Isolate the fault in five steps
  4. Read the result
  5. Worked example: bisecting a localization list
  6. Escalate only with an offline reproduction
  7. Verify the fix
  8. Questions asked

A popup that returns every time you close it is a project-open error, not a crash. The editor loads the project, one file inside it fails to parse or load, and the retry runs into the same fault. That loop has a short isolation path: record the exact file the message names, open the unchanged project on a second PC, then remove only that named file from a diagnostic copy. If the copy opens, the removed file carries the defect. If it still fails, build a new empty project and copy the named file into it, and the result tells you whether the fault follows the file or the project. None of these tests needs the PLC attached.

The popup that returns is not a crash #

A crash ends the editor process on its own: Windows closes it, and the fault belongs to the installation or the environment, where diagnostics run through the event log and a repair of the installation. An error popup is different. The editor stays open, the project load reaches the broken item, the popup reports it, and every retry hits the same item again. The repetition looks dramatic, but it is the editor telling you, each time, exactly where it stops. Read it that way and the fault is already half-located: in the project content, not the machine.

One boundary case sits outside this article: a project that refuses to open at all, with no editor window and no popup loop, has its own recovery and backup path, covered in when an HMI project fails to open: recovery and backup. Everything below applies to the project that opens into the repeating popup.

Read the message before touching the project #

Capture the complete error text and the file path it names, character for character, into your test notes before anything else changes. The message sorts the fault into one of two families, and the two take opposite first moves.

Version mismatch: update before you dissect #

If the project was last saved by a newer editor build, or was built for a panel model your installed editor does not know, it fails before any of your content is parsed. The tell is a message about an undefined model number or a missing definition rather than a named file from your own project. The first move is the opposite of isolation: update the editor through the manufacturer's supported installer to a build the vendor documents as compatible with the project, then reopen it. Dissecting project content cannot fix a version gap.

A parse fault points at one file #

If the message names a file from your project, the editor failed to load that file's content. Vendors document this class with codes of their own; one published example is E_FUNCTION_EXPRESSION_PARSER_ERROR, reported when a function expression cannot be parsed. An error of this family makes the named file the isolation target. A parse fault also sits upstream of communications: changing PLC tags, driver settings or network settings cannot touch it, so leave the controller configuration alone unless a later test implicates it. One entry condition to rule out first: if the project reached this PC through version control and came down incomplete, it fails differently, and fixing an HMI project that fails after a Git clone covers that missing-files case.

Isolate the fault in five steps #

  1. Copy the entire project folder to a diagnostic copy. Every test below runs on the copy; the original stays untouched until the fix is proven.
  2. Record the complete error text, with the named file path exactly as printed, in your test notes.
  3. Open the unchanged project on a second PC running the same editor build. A clean open there ends content isolation: the fault lives in the first PC's installation or environment, and repairing or updating that installation is the fix.
  4. On the diagnostic copy, move the single named file out of the project and open it. Move nothing else; one variable per test.
  5. If the copy still fails, build a new empty project with the same editor build, copy the named file into it, and open it. This separates a defective file from a defective project around it.

Run these tests in the editor on your engineering PC. The one move to avoid is probing the fault by downloading test builds to a panel that operators are using on a live line: these tests never need the runtime, and a half-loaded project at an operating station is its own hazard.

Read the result #

Test result Meaning Next action
Opens on the second PC, fails on yours Local fault in your installation or environment Repair or update the editor on your PC; do not rebuild project content
Fails on both PCs; the copy opens without the named file The named file carries the defect Correct or rebuild that file in a fresh copy
Fails with the named file removed That file is not the only fault Re-read the message for a second named item, then run the new-project test
New project with the file copied in fails The fault follows the file The file content is defective; rebuild it from its source data
New project with the file opens The file is usable; the original project has another fault Re-add what the recent edit changed, one item at a time, on a fresh copy

The row that matters most is the second one: a copy that opens without the named file converts a suspicion into a located defect.

Worked example: bisecting a localization list #

A popup that first appeared after a larger localization variable list was imported is timing evidence, and the step 4 result decides it. Suppose the project imported a list of 2,400 localization variables and the popup began. The diagnostic copy without the list file opens cleanly, so the list file carries the fault; now narrow inside it by re-importing halves onto fresh copies:

  • Entries 1 to 1,200: popup returns, so the fault is inside 1 to 1,200.
  • Entries 1 to 600: opens clean, so the fault is inside 601 to 1,200.
  • Entries 601 to 900: opens clean, so the fault is inside 901 to 1,200.
  • Entries 901 to 1,050: popup returns, so the fault is inside 901 to 1,050.
  • Entries 901 to 975: opens clean, so the fault is inside 976 to 1,050.

Each re-import halves the candidate set: 2,400 becomes 1,200, then 600, 300, 150 and finally 75 entries after five rounds. Reading 75 localization strings for a malformed placeholder, a broken key or a duplicated identifier is minutes of work. Fix or rebuild those entries, re-import the cleaned list into the diagnostic copy, and the defect is closed at its source instead of by deleting the feature.

Escalate only with an offline reproduction #

Escalate when the fault followed the project, meaning it failed on both PCs, and the file tests did not close it. Package the diagnostic copy, the editor name and build number, the complete captured error text, and a reproduction sequence: open the project, the popup appears, close it, it returns. Support can analyze a defect it can reproduce, and the reproduction has to run without any PLC connection, so strip controller dependencies and sensitive tag addresses before the project leaves your network. If the project opened cleanly on the second PC, do not escalate at all: the defect is local to your installation, and the vendor cannot repair your PC from their side.

Verify the fix #

Reopen the repaired copy and repeat the exact sequence that used to trigger the popup: open, close whatever appears, open again. One clean load is not proof, because the loop was the symptom; the fix is proven when that sequence loads clean every time. If the two PCs behaved differently at step 3, run the final verification on both. Keep the untouched original archived with the captured error text. It is the evidence that ties the fix to the fault if this project's history is ever questioned.

Questions asked

How do I tell an error popup from a crash?

A crash ends the editor process itself: Windows closes it, and diagnostics move to the Windows event log and the editor installation instead of project content. When the editor stays open and the same popup returns every time you close it, the project is loading but one item inside it fails to parse or load, and the second-PC and file-removal tests in this article are the tools for that case.

What if the error names no file at all?

Some builds report a module name or a bare code with no file path. Work from timing instead: copy the project, undo the most recent edit in the copy by deleting the newly imported list, and open it. If it opens, re-import the list in parts until the popup returns; the part that triggers it holds the defect. Confirm the editor build covers the project's save version before you conclude the content is bad.

Can a version mismatch cause the repeating popup?

Yes. A project saved by a newer editor build, or created for a panel model the installed editor does not recognize, fails before your content is parsed; the message names an undefined model number or a missing definition. Update the editor through the manufacturer's supported installer to a documented compatible build, reopen, and only then start isolating project files if the popup survives the update.

Why must the reproduction run without a PLC connection?

A fault that appears only with the controller attached cannot be separated from driver behavior, tag traffic or network conditions, so it cannot be confirmed as a software defect. A project that repeats the popup in the editor alone proves the defect sits in the project data or the parser, which can be analyzed and fixed. Adapt the project to run offline and strip sensitive tag addresses before it leaves your network.

How long do I keep the untouched original project?

Until the repaired copy has opened cleanly on both machines that differed: the PC where the fault appeared and the second PC, or twice on one machine if there was only one. Archive the original with the captured error text and the trigger sequence, so the evidence stays with the machine's project history instead of in one person's notes.

Keep reading

  • Troubleshooting

    Scaling encoder counts to inches on an HMI display

    A value like 9E-46 on a scaled encoder readout is a data-type symptom: the HMI tag reads integer counter memory as a 32-bit float. Set the tag to the integer type the counter writes, apply two-point scaling from 0 counts/0 inches to 2,540 counts/1 inch, set four decimal places on the object, and let one layer, HMI or PLC, own the conversion so the count is never scaled twice.

    HMI and touchscreens

  • Troubleshooting

    Why a standalone HMI project will not open in your integrated editor

    A standalone Vijeo Designer .vdz archive usually fails in an integrated editor because the two editors sit at different version levels and accept different project operations, not because a license is missing. Open the copy through the destination editor's documented import or restore route, with the destination editor at the same version level or newer than the editor that saved the archive, then verify target and communication settings.

    HMI and touchscreens

  • Troubleshooting

    Fix the libQt5WebEngine.so.5 error when an HMI editor will not start

    The message means the dynamic loader cannot resolve one shared library the editor binary needs, so startup fails before any editor code runs. Run a dependency report on the exact failing binary, install your distribution's Qt 5 WebEngine runtime package that provides libQt5WebEngine.so.5, and relaunch only when the report shows no unresolved libraries. Do not symlink or rename another library to match the name.

    HMI and touchscreens