IMDTouchKnowledge BaseShop main store

Troubleshooting

Fixing a Cape Display That Shows Only a Bright Vertical Line

Panel PCs and displays Published 6 min read

Short answer

Backlight that adjusts from 0 to 100 with only a bright vertical line means the panel is lit but the pixel path is misprogrammed: BB-BONE-4D7C-01-00A1.dts targets am33xx, and the BeagleBone AI is am572x. Rebuild the overlay from BBAI-4D4C-00A1.dts on the v4.19.x-ti-overlays branch and the splash logo returns; any remaining color or geometry fault is a driver timing or colorspace review item, not dead hardware.

On this page
  1. Why does the cape show only a bright vertical line?
  2. Does a backlight that adjusts from 0 to 100 prove the display works?
  3. Which overlay file should the display be built from?
  4. How do you rebuild and load the corrected overlay?
  5. What if the splash logo appears with wrong colors or geometry?
  6. When is the cape compatible with the BeagleBone AI?
  7. Questions asked

A bright vertical line at the right edge, under a backlight that still dims from 0 to 100, is not a dead panel: the cape is powered and lit but no usable pixel data reaches the glass, because the overlay in force targets the wrong SoC. BB-BONE-4D7C-01-00A1.dts is an am33xx overlay for BeagleBone Black-class boards, and the BeagleBone AI (BBAI) is an am572x part. Rebuild the display overlay on the structure of BBAI-4D4C-00A1.dts, carrying over only the panel-specific definitions from the old file, and the splash logo replaces the line. On the documented baseline, a Buildroot image with U-Boot 2021.04 and kernel 4.19.94-ti-r68, exactly that rebuild moved this setup from the vertical line to a visible logo. Whatever remains after that, wrong colors or geometry, is a colorspace or timing item in the driver and overlay, not a hardware fault.

Why does the cape show only a bright vertical line? #

Because the overlay you loaded, BB-BONE-4D7C-01-00A1.dts, was written for am33xx BeagleBone hardware, and the BeagleBone AI is an am572x part: the display routing it programs does not match your SoC, so no frame reaches the panel.

The overlay is doing its job on the wrong platform. It came from the bb.org-overlays tree under src/arm, the overlay family tied to am33xx BeagleBones, and nothing in it describes the am572x display subsystem of the board you moved to. The panel itself is the least likely suspect: the same GEN4-4DCAPE-70CT-CLB and its 4D Cape adapter ran correctly on the earlier BeagleBone, and the only thing that changed in the move is the SoC and the device tree that has to describe it.

Read the symptom the same way:

What you see What it establishes What it does not establish
Backlight adjustable 0 to 100 through /sys/class/backlight/backlight/brightness The backlight control path works end to end Pixel data, timing, or color handling
One bright vertical line at the right edge, no splash logo The panel is powered and lit That a usable frame is being transported
Splash logo after the rebuilt overlay loads The display interface carries a frame Correct colorspace and timing
Logo present, colors or geometry wrong The structure of the interface is right A finished panel-driver configuration

Does a backlight that adjusts from 0 to 100 prove the display works? #

No: brightness control through /sys/class/backlight/backlight/brightness proves only the backlight chain, and on this fault it responds across its full span while the panel still shows no image.

Check the endpoints: write 100 and confirm full brightness, write 0 and confirm dark. Both ends taking effect means the whole span is live, and from there you stop looking at power and backlight entirely. The fault sits between the SoC's display controller and the panel, which means the overlay and its driver. A panel that is dark rather than lit is a different path; work through troubleshooting a panel PC HMI dark screen after reboot for that case.

Which overlay file should the display be built from? #

BBAI-4D4C-00A1.dts, on the v4.19.x-ti-overlays branch of the BeagleBoard-DeviceTrees repository under src/arm/overlays/, is the am572x cape reference: build on its structure and keep the panel-specific definitions from BB-BONE-4D7C-01-00A1.dts.

Match the overlay branch to your kernel before you copy any code. Take the running kernel string, 4.19.94-ti-r68, and cut it at the second dot to get the major.minor pair: 4.19.94 becomes 4.19. Read the branch name, v4.19.x-ti-overlays, the same way, with x standing for any patch level: its pair is also 4.19. Since 4.19 equals 4.19, this branch is the one whose overlays match your kernel; the .94 patch level and the r68 build do not change that choice, and the -ti in each name marks the same kernel flavor. If you later move the image to a 5.10.x kernel, the same arithmetic sends you to the 5.10.x branch and the ported overlay has to be rechecked there.

How do you rebuild and load the corrected overlay? #

Rebuild it in five steps: write down the baseline, pull both overlay sources, port the panel nodes into the BBAI structure, compile to a .dtbo, and boot the Buildroot image with it applied to see the splash logo.

  1. Record the baseline before touching anything: board, GEN4-4DCAPE-70CT-CLB on its 4D Cape adapter, the Buildroot image with its U-Boot and kernel versions as booted, and the symptom exactly as observed, backlight working across its range, one bright vertical line at the right edge, no splash logo.
  2. Pull the platform reference, BBAI-4D4C-00A1.dts, from BeagleBoard-DeviceTrees on the v4.19.x-ti-overlays branch, under src/arm/overlays/.
  3. Pull the panel reference, BB-BONE-4D7C-01-00A1.dts, from the bb.org-overlays tree under src/arm, and mark which nodes are panel-specific, the panel timing and the driver binding, versus am33xx platform plumbing.
  4. Build the new overlay on the BBAI-4D4C-00A1.dts structure, carrying across only the panel-specific definitions and dropping the am33xx plumbing.
  5. Compile it as an overlay; they live under arch/arm/boot/dts/overlays/ in the kernel tree and build alongside the normal DTBs. Put the .dtbo into the Buildroot image and boot with it applied.

Power the board down before reseating the cape or its ribbon connectors; they are not hot-plug parts. The pass condition is the splash logo across the full panel width in place of the line.

What if the splash logo appears with wrong colors or geometry? #

Treat wrong colors or distorted geometry as a colorspace or timing mismatch between the ported panel definitions and the driver: the interface is now carrying a frame, and that is a review item, not a revert.

That is the documented outcome of this rebuild: the logo rendered and the image still looked wrong, with colorspace and timing as the named suspects. Inspect the color format and channel ordering the overlay requests against what the driver actually applies, and check the panel timing values you carried across from the 4D7C file against the panel's own documentation before changing them. Change one item, recompile, reboot, compare. A logo that renders puts you past the failure class covered in fixing startup splash screen errors on legacy panel PCs and into panel-driver territory. Once the image is clean, touch input is the next bring-up step; getting a touchscreen working under a Linux HMI runtime covers that side.

When is the cape compatible with the BeagleBone AI? #

Call it compatible when the rebuilt am572x overlay holds a stable image with correct colors and geometry on your exact baseline, kernel 4.19.94-ti-r68 with U-Boot 2021.04, and the touch controller enumerates; a splash logo alone is partial bring-up.

  • Backlight: both ends of its range take effect through /sys/class/backlight/backlight/brightness.
  • Splash logo and then the full interface render across the whole panel, with no residual line.
  • Colors and geometry match the source image, not a shifted or discolored copy.
  • Touch reports events from the panel's controller.

Pass all four on this baseline and the cape is validated for that image; write the kernel and overlay versions next to the result, because a kernel branch change reopens the overlay question. Fail only the color and geometry checks and you still keep the am572x structure, since the remaining work is timing and color format, not platform.

Questions asked

Is the GEN4-4DCAPE-70CT-CLB a compatible display for the BeagleBone AI?

Partially, on the documented evidence: with the display overlay rebuilt from BBAI-4D4C-00A1.dts, the panel that had shown only a vertical line rendered the splash logo, so the interface outputs a frame. The same test still produced wrong colors or geometry, so treat the cape as brought up rather than validated until color and timing are corrected on your kernel baseline.

Why did the same overlay work on the earlier BeagleBone?

Because that board is an am33xx part, which is exactly the family BB-BONE-4D7C-01-00A1.dts was written against. Nothing in the overlay went bad; the platform under it changed when you moved to the am572x BeagleBone AI, so the platform half of the overlay had to be rewritten while the panel half stayed.

Do the U-Boot and kernel versions change which overlay you use?

The kernel version does: pick the overlays branch whose major.minor number matches, so 4.19.94-ti-r68 pairs with v4.19.x-ti-overlays. U-Boot 2021.04 only has to load the compiled overlay, and the documented fix ran with exactly that pair, so the bootloader is not the variable to chase.

Can you build a full device tree instead of an overlay?

Yes. For a custom board on this processor family, the AM572x EVM device tree, am57xx-evm-reva3.dts, is the file TI's processor forum names as the one to compile and modify. That approach trades overlay management for a single board file, at the cost of rebuilding the full device tree on every change.

Could the vertical line itself be a panel or cable fault?

Only if it survives a known-good configuration. This panel had already run correctly on the earlier BeagleBone with the same 4D Cape adapter, and the line disappeared after the overlay rebuild alone, which puts the fault in configuration. A line that persists across both boards and both overlays would move suspicion to the panel or its cable.

Keep reading

  • Troubleshooting

    Why HDMI Video Works but USB Touch Does Not on a Panel

    A stable HDMI image with dead touch is a USB-path fault. Power the panel from its own supply or an externally powered hub, confirm the host enumerates a supported touch controller family, set the visualization to TargetVisu, and treat misplaced touches as a coordinate-mapping error rather than a touch failure.

    Panel PCs and displays

  • How-to

    Running One HMI Interface on Multiple Touchscreens From One PC

    One computer can drive four touchscreens without four copies of the interface. Put data and commands in one controller, give every screen its own display instance that subscribes to published state and submits commands to one handler, and verify display outputs and touch mapping per screen before you build any views.

    Panel PCs and displays

  • How-to

    How to Commission USB-VGA Video and Touch for a Native HMI

    Commission in three independent branches: the USB-VGA adapter must enumerate with a matching chipset driver, the touchscreen must report correct corner coordinates at the final orientation, and the GUI runtime must launch from the operating account. A working display proves nothing about touch, and both prove nothing about the application.

    Panel PCs and displays