LVDS suddenly black

  • I have an armStoneA9r2 with quad core. I'm using a 1080p60 LVDS display. The display is operated in split mode.


    Display Datasheet: https://www.data-modul.com/sit…pecification_12024698.pdf


    Display Settings:


    When I start the board, the 4 Tux come, when it switch to the splash screen, the screen goes black and it stays black. Everything else continues to work.

    The clock lines of the two LVDS signals continue to clock.


    If I reduce the clock frequency of the display, so that the refresh rate is also reduced, then everything works as it should. The splash screen comes up and the GUI is displayed. However, sometimes the display goes black at some point.

    For example, if I run /opt/g2d_samples/g2d_overlay_test , it runs through a maximum of one or two times, after which no graphical output works anymore. The display stays black.


    The only interesting thing I find in the dmesg output is:

    Code
    1. dmesg | grep ipu
    2. imx-ipuv3 2400000.ipu: IPU DMFC ONLY-DP HIGH RESOLUTION: 5B(0~3), 5F(4~7)
    3. imx-ipuv3 2800000.ipu: IPU DMFC ONLY-DP HIGH RESOLUTION: 5B(0~3), 5F(4~7)
    4. imx-ipuv3 2800000.ipu: WARNING: adapt panel end blank lines
    5. imx-ipuv3 2800000.ipu: IPU Warning - IPU_INT_STAT_10 = 0x00080000
    6. imx-ipuv3 2800000.ipu: WARNING: adapt panel end blank lines
    7. imx-ipuv3 2800000.ipu: IPU Warning - IPU_INT_STAT_10 = 0x00080000

    However, I am unsure whether the output has anything to do with my problem, as this dmesg output comes when the graphical output is still working.


    Does anyone have any idea where the problem can come from?


    The CPU should be able to make 1080p60 with LVDS?

  • Hello,


    which release are you using? Can you reproduce this with the new fsimx6-B2021.10.1 release?


    The warning IPU Warning - IPU_INT_STAT_10 = 0x00080000 indicates, that there is data starvation, so data that should be sent to display is not available in IPUs internal FIFO.


    You could also try to apply this solution from the NXP forum:

    https://community.nxp.com/t5/i…488/highlight/true#M94889


    by running

    setenv extra dmfc=3

    in U-Boot befor booting.

    This will increase the size of the internal IPU FIFO.


    Your F&S Support Team

  • Thank you for your prompt reply!


    I had already found the link and entered the dmfc = 3. This is where the message comes from:


    Code
    1. IPU DMFC ONLY-DP HIGH RESOLUTION: 5B(0~3), 5F(4~7)

    otherwise this would be shown:

    Code
    1. IPU DMFC NORMAL mode: 1(0~1), 5B(4,5), 5F(6,7)


    The warnings come because I have set the front porch and back porch to 0 in the device tree. I corrected that, now I don't get any more warnings.


    The problem that the display goes black, however, remains. When I start the board, the 4 Tux come, when it switch to the splash screen, the screen goes black and it stays black. If i reduce the clock frequency it gets better.


    Since we use Yocto, I have always tested with fsimx6-Y2020.03 so far.

    I tested with the precompiled SD card image and with a self-compiled image with the settings fus-imx-fb and fus-image-qt5.

    The behavior is the same for both.


    Today I tested it with the precompiled fsimx6-B2021.10.1. Everything seems to be
    working stably here with the normal clock frequency.


    Is there a known bug, why 1080p60 does not work in split with fsimx6-Y2020.03?

    Will there be a new yocto version soon?

    Do you have an idea what I can still test to get the display to work with the current yocto version?








  • Hello,


    unfortunately a fsimx6 Yocto Release isn't scheduled soon.


    I have made some Tests with a different 1080p LVDS Display

    https://www.data-modul.com/sit…pecification_12017304.pdf

    and the fsimx6-Y2020.03 on the armstonea9r2q and could not reproduce the issue.

    I ran the g2d_overlay_test several times and played 1080p videos.

    Please let me know if you have a better way to reproduce it.


    I noticed that the output of fbset -i differs yo yours


    Especially the framebuffer address and size, did you make any changes here?


    Another possibility could be the 3v3 voltage supply. Your Datasheet marked the min. supply voltage of 3.15V red, so maybe there are some problems here? Maybe you could measure if this voltage is stable when the issue occurs?


    Your F&S Support Team

  • The frame buffer addresses and size differ for fus-imx-x11 and fus-imx-fb. However, I tried both and got stuck with both. I have now switched to buildroot. So far everything works without any problems. Anyway, thank you for your help.