IMDTouchKnowledge BaseShop main store

Troubleshooting

Why HMI USB Downloads Fail and Touch Stops Responding

HMI and touchscreens Published 7 min read

Short answer

A panel that drops USB downloads, shows question marks on numeric tags and ignores touches is usually not broken: fix the data cable, the selected COM port and the PC software installation first, and treat read-only PLC tags as a button write-action fault. In the reported case a clean reinstall restored downloads and cleared the display.

On this page
  1. Separate the four faults first
  2. Make the USB download repeatable
  3. Working the one-button test on a panel that reads but never writes
  4. Touch that answers late is not touch that answers nothing
  5. Keep the PLC path separate
  6. When the panel really is the problem
  7. Questions asked

Four faults that arrive together — USB downloads that fail with "device not found" and "COM port busy" messages, question marks on numeric fields, touch that answers nothing after a firmware load, and buttons that never write to the PLC — look like one dead panel. They are four separate problems, and none of the evidence in the reported case condemned the hardware: a proper data cable and the right COM port made downloads repeatable, a complete removal and reinstall of the engineering software restored them again and cleared the question marks, and the read direction of the HMI-to-PLC link kept working throughout. Work the four faults in that order before ordering a panel.

Separate the four faults first #

Each symptom has a different owner, so treat each one as its own case:

  • Downloads fail over USB. "Device not found" and "COM port busy" name PC-side conditions: nothing enumerated on the cable, or a process holds the serial port. Decision: fix cable, COM selection and the PC installation before the panel is suspected.
  • Numeric fields show a question mark. The runtime has no live value for the tag. In the reported case the marks appeared with no project change and cleared after a full software reinstall. Decision: treat it as a PC installation fault until a repeatable download proves otherwise.
  • Buttons never write to the PLC, but PLC changes appear on the screen. One direction of the link works, so the channel is up. Decision: the fault is in the button's write action, not in the hardware.
  • Touch answers nothing after a firmware load. This is the only fault that can implicate the panel, and only after the download path and calibration are proven good.

Make the USB download repeatable #

  1. Put a real data cable on it. A cable built only for charging cannot carry a download at all; a marginal data cable enumerates the panel and then drops the connection, which is exactly what intermittent success looks like. The reported HMI downloads became reliable once a phone-style charging cable was replaced with the specified data cable.
  2. Select the COM port the PC actually assigned. A USB download on this class of tool still runs through a serial port: open the download connection settings, set the connection type to serial, and pick the COM port the PC shows for the attached interface — not the port that happened to be right last time.
  3. Free a busy port before blaming it. "COM port busy" means a process holds the handle. Close terminal programs, sync tools and any second instance of the engineering software; a handle left stuck by a crashed download usually comes back after unplugging and replugging the cable.
  4. Treat no COM port at all as a PC fault. If nothing enumerates, move the cable to another USB port, ideally on a different controller, before touching the panel. The supplier's support position in the reported case was that the software release had no bearing on USB communication and the cause sat in the PC — an apparent improvement after an update is not proof of root cause, so demand downloads that repeat across several attempts.
  5. Reinstall the engineering software completely when the rest passes. In the reported case, removing the software entirely and reinstalling it restored downloads and cleared the question marks, after reseating the cable and cycling panel power had produced only intermittent recovery. Treat reseat and power-cycle as diagnostic resets that demonstrate intermittency, not as repairs.

If the panel enumerates and the project transfer still fails on a network path instead of USB, the causes narrow differently — see why an HMI is detected but project download fails.

Working the one-button test on a panel that reads but never writes #

This test decides whether the link or the button is at fault, on the smallest project that can show it.

Inputs. One button object and one lamp object on one screen, both bound to the same single PLC bit — the minimal project from the reported case, with nothing else on the panel that can interfere. The PLC programming tool online, able to force that bit and monitor it. A working download from the section above.

The rule. The two directions share one channel but travel differently: the panel polls its read tags continuously, while a write is an event that leaves only when the button action fires. Test one direction, then the other, and the fault lands on either the channel or the object.

The arithmetic.

  • Test A, read direction: force the bit ON from the PLC online. The lamp on the panel changes. Read = pass.
  • Test B, write direction: monitor the bit online and press the button. The bit never moves. Write = fail.
  • One channel, read pass plus write fail means the channel, the driver and the addressing of the readable tag are all good, and the fault is inside the button object.

Result. What fails is the write action itself: the tag the button writes, its address, its data type, or its write mode — momentary, set, reset or toggle.

Decision. Read pass with write fail means fix the object, not the hardware. Only when both directions fail on the one-button project does the fault move to the channel, and only then is the link torn down. Setting up a button that writes more than one bit has its own mapping steps, covered in how to make one HMI button read and write two PLC bits.

One field-documented trap in the write mode: a bit written "set while pressed" can stay set when the release event is lost, so any bit that must not stick is written on press and cleared by the PLC itself.

Touch that answers late is not touch that answers nothing #

Two different faults share one symptom — a screen that seems to ignore you — and the press count separates them.

Press the same on-screen key three times in quick succession. If three operations then run, one after another with a delay, every press was detected and queued: the touch surface and its controller work, and the lag sits downstream of detection. That is the behavior the reported case showed with calibration verified correct, so recalibration will not remove it, and neither will a new touch surface. Calibration that will not hold at all is a different fault with its own causes — see touch calibration fails or drifts: separate the real causes.

Touch that answers nothing after a firmware load is the one fault that can implicate the panel, and it is checked last. One rule first: a firmware write must run to completion with power stable — interrupting it can leave the panel without valid firmware and unbootable, a worse fault than the one being chased. After the write, download the project again, confirm calibration, then run the three-press test. No queued operations and no response of any kind, with downloads repeatable and calibration valid, leaves the panel itself as the remaining suspect.

Keep the PLC path separate #

A PLC that accepts a program once in ten attempts over USB is its own case, and it inherits no verdict from the HMI in either direction. Apply the same discipline — data cable, assigned COM port, free port — and use the escape the reported case used: program the PLC over its serial port instead of USB, which was reliable where USB was not. A forced download that succeeds occasionally is a workaround, not a repair. Once the HMI side is stable and the PLC still will not take a program, escalate the PLC with its project files; a repeatable HMI download proves nothing about the PLC link, because they are two devices on two ports with two drivers.

When the panel really is the problem #

The panel is condemned only when every item on this list is true and touch still answers nothing:

  • a known-good data cable, not a charge-oriented one;
  • the COM port verified against what the PC assigned, and free;
  • the engineering software removed and reinstalled cleanly;
  • the project downloading repeatably, several times in a row;
  • calibration valid and the three-press test showing no queued input.

Then the unit goes to service, with the HMI and PLC projects attached so the write behavior can be reproduced on the bench. When service stops making sense on an aging terminal, the station can be rebuilt around industrial panel PCs running a PC-based runtime in place of the proprietary panel.

Questions asked

Why does the download say device not found or COM port busy?

No device enumerated on that cable, or another process holds the serial port. Use a full data cable and close terminal programs, sync tools and extra instances of the engineering software. The COM number can also change when the cable moves to a different USB socket on the PC, so re-check the assignment after every move.

Why does the screen show PLC values but button presses change nothing?

Reads are served by continuous polling, so the screen stays alive even when the button's write action is missing or mis-mapped; a write only leaves the panel when that action fires. Force the bit online and watch it while pressing the button — a bit that never moves convicts the write action, not the channel.

Should I recalibrate when touch misbehaves after a firmware update?

Calibrate once, following the tool's own pattern, and then stop: calibration corrects where touches land, not whether they register or how fast they run. Touches that queue up and execute late are not an alignment fault, and repeating calibration will not change that behavior.

What does a question mark on a numeric field actually mean?

The runtime cannot resolve that tag to a live value: the connection is down, the address does not resolve, or the project database on the PC is damaged. It appeared with no project change in the reported case and cleared after the complete reinstall, which is why the PC installation is checked before the panel.

Why do several quick touches run late, one after another?

Each press was detected and queued: three rapid touches produce three delayed operations, which proves the touch surface and its controller register input. The lag lives downstream of detection — in panel processing, communications retries or project logic — so neither recalibration nor a new touch surface removes it.

Keep reading

  • How-to

    How to Make One HMI Button Read and Write Two PLC Bits

    Enable the button's write/read offset address and point it at the HMI internal word 0. With the word at 0 the button reads and writes X0; with the word at 1 the same button reads and writes X1. The offset is resolved inside the HMI before any request reaches the PLC.

    HMI and touchscreens

  • Troubleshooting

    Why an HMI Writes Integers but Not B3 Bits Over DF1

    When an HMI writes T4, C5 and N7 data but B3 bits never change over DF1, the serial link is proven and the failure sits in the HMI software's binary-file translation. Check the engineering software version before baud, cable or PLC settings: a documented case was fixed by upgrading EBplus from 2.10 to 5.07.

    HMI and touchscreens

  • Troubleshooting

    When Only One Touchscreen Icon Responds to Press

    If only one of several icons answers a press, the others are most likely status indicators: they display state but never accept input. Read each object's role in the configuration tool before touching hardware — a healthy screen shows exactly as many responsive icons as it has assigned press actions, and a silent indicator is correct behavior, not a dead touch area.

    HMI and touchscreens