No output on HDMI, bad EDID

Hi,

I have an issue to connect a display to an Ixora V1.1A board (iXM6). It’s an external display with a DVI input, which I connect with an HDMI → DVI cable to the board. The resolution is 1280x800.

My u-boot commandline to set up the display is the following

Apalis iMX6 # env print vidargs
vidargs=mxc_hdmi.only_cea=1 video=mxcfb0:dev=hdmi,1280x800M@60,if=RGB24 video=mxcfb1:off video=mxcfb2:off video=mxcfb3:off fbmem=32M

However, the display stays blank. Also, the dmesg output shows that the EDID information is wrong.

imx-drm display-subsystem: HDMI-A-1: EDID is invalid:
[00] BAD  00 ff ff ff ff ff ff 00 0a 0a e5 e5 d9 d9 11 11
[00] BAD  22 22 01 01 80 80 0e 0e 2a 2a b5 b5 57 57 94 94
[00] BAD  20 20 54 54 00 00 01 01 01 01 01 01 01 01 01 01
[00] BAD  01 01 01 01 01 01 c6 c6 00 00 50 50 17 17 30 30
[00] BAD  36 36 d9 d9 00 00 00 00 00 00 00 00 00 00 26 26
[00] BAD  20 20 50 50 30 30 30 30 20 20 00 00 00 00 00 00
[00] BAD  4b 4b 3c 3c 00 00 20 20 20 20 20 20 00 00 00 00
[00] BAD  00 00 30 30 20 20 30 30 20 20 4e 4e 30 30 00 00

However, when I connect the display to my Laptop (Linux, also HDMI output), then it detects the display correctly.

% edid-decode < /sys/devices/pci0000:00/0000:00:02.0/drm/card0/card0-HDMI-A-2/edid

EDID version: 1.3
Manufacturer: BUR Model 12e5 Serial Number 1091636441
Made in week 34 of 2019
Digital display
Maximum image size: 22 cm x 14 cm
Gamma: 2.10
DPMS levels: Off
RGB color display
First detailed timing is preferred timing
Display x,y Chromaticity:
  Red:   0.5917, 0.3398
  Green: 0.3095, 0.5791
  Blue:  0.1503, 0.1279
  White: 0.3134, 0.3291
Established timings supported:
Standard timings supported:
Detailed mode: Clock 71.100 MHz, 217 mm x 136 mm
               1280 1328 1360 1440 hborder 0
                800  803  809  823 vborder 0
               +hsync -vsync
               VertFreq: 59 Hz, HorFreq: 49375 Hz
Monitor name: B&R AP1000
Monitor ranges (GTF): 50-75Hz V, 40-60kHz H, max dotclock 70MHz
ASCII string: S01 F01 INA01
Checksum: 0x6c (valid)

What could be the reason that the Toradex board cannot identify the display? I appreciate any pointers.

Thanks, Michael

Please note that BSP 3.0b1 was not released for Apalis iMX6 and is therefore not recommended. Have you tried the same with our latest stable BSP 2.8 at all?

Hi @marcel.tx,

Thank you for having a look. I run my own distribution, which is based on 3.0 (thud) but uses the linux-toradex rt kernel from the Torizon distribution (which does support iMX6). It might of course be a problem with my setup, and I understand that you only want to support released versions. However, in this case I believe it is a specific problem where I, unfortunately, don’t have much experience in.

Here are some more experiments, which indicate that it is not a problem with my setup.

  • I tried with BSP 2.8b5, and the display also stays blank.

  • When booting the official TEZI image, the screen also stays blank.

  • I have another monitor with DVI input (with a resolution of 1024x768), which works fine with my own setup as well as BSP 2.8b5.

Interestingly, when I boot my Qt application, it does show up on the screen. It seems even with the correct resolution. However, I don’t see any output during boot, and dmesg gives me the errors shown in the OP.

Again, it is entirely possible that the EDID from the monitor is wrong. I also contacted the monitor manufacturer for this issue. However, what is confusing to me is the fact that my Linux laptop has no problem identifying the monitor.

Does anybody have any idea where else I could look?

Thanks again!

What exact version of that rt kernel? What exact device tree? Any device tree overlays at all (which is BTW how Torizon handles displays)? Please note that BSP 2.8b5 is no longer supported. What exactly do you mean by official Toradex Easy Installer image (e.g. what exact version thereof)? What exact Qt version with what exact Qt backend configuration are you running? Don’t get me wrong. It’s a free world. However, if you expect us to help you, you will need to be a tiny little bit more specific. Especially, as you don’t seem to like using any regular proven configurations.

I re-tested with the following images

  • Torizon Core 1.0b1.9
  • TEZI 1.8 (20181019)
  • 2.8b6.184

The behaviour is always the same, the display stays blank during boot. As I said, all the configurations work with another monitor. I’ll see what information I get from the manufacturer, and report back if I find a solution.

Hi,

Somehow my question seems to hit you the wrong way. That wasn’t my intention. I’ll take your point about being specific, but in this case I think your insistence on versions is not helpful. A couple of points.

I thought that I was at least a tiny little bit specific. I showed the vidargs and the errors from dmesg. I also wrote that the problem is not specific to one configuration, but happens to several. This is why I was hoping for some general pointers to where I could look into.

The problem with the proven configurations from Toradex is the following. The last time I tried, the RT patch didn’t apply to the freescale kernel of BSP 2.8. Torizon is new and follows the master branch for most of its repos. I’m not sure that any of these alternatives is the best choice for a stable production system. And last time I had this discussion in this forum, the answer basically was that Toradex provides some sample images, and customers are supposed to build their own based on these samples. And this is exactly what I’m doing.

I wrote that another monitor works with all the setups, including my own distro.

Please also don’t get me wrong. This forum helps a lot, and the support from Toradex members such as yourself is outstanding. It’s just that in this particular case I feel that the tone of the response is unjustified.

Hi @eglimi

Thanks for your Input.

Actually from Toradex side, we test and validate the regular Bsp images for the standard interfaces of the module with some standard devices. Thus we know that our Bsp images works with this standard configuration.

When a client uses some different devices or display for his application and it does not work, then are happy if he tests this a regular Bsp and share the changes he has done for his application.

So, could you provide us the following information:

  • HW Information (manufacturer, model, …) about your working and non working display?
  • Test with a regular Bsp 2.8b6 with and share the changes to any files/settings?
  • Share the complete bootlog in a file?
  • Output of fbset and xrandr?

Thanks and best regards,
Jaski

Hi @jaski.tx,

Many thanks for your response. I see your point. Also, I didn’t mean to imply that Toradex doesn’t test its releases, I apologize if it came through as such. The issues I currently face with external displays is frustrating. Especially since I thought that HDMI should work fine (at least with well-known manufacturers), and now since I realise that this is not the case, it takes me way too much time. This has nothing to do with Toradex of course.

I’m currently testing various displays with DVI input attached to the Ixora HDMI interface. There is also an open ticket with the manufacturer for the display that doesn’t work. I’ll update this topic with all the information you requested once I have collected them.

Hi @eglimi

You are welcome. Thanks for your Input.
We will wait for you Information about the displays.

Best regards,
Jaski

Here are some more information as requested by @jaski.tx. I tested several images with several monitors. I try to be as clear as possible as to what output belongs to which setup. Please let me know if I should add anything else. Thank you again in advance for having a look at this.

I didn’t change anything from the default installation. The images were installed directly from TEZI (online). I tried to change the vidargs as shown in the original question, but it didn’t change anything.

Here are the attachments.

Monitor 1, no output on HDMI

This is the monitor that I cannot get to work. It always stays blank: No bootup log for 2.8 and no bootup image for Torizon.

Monitor: 10.1", 1280x800, DVI input. Unfortunately, I don’t have exact specifications of this monitor. It is the following model from B&R.

https://www.br-automation.com/de-ch/produkte/visualisieren-und-bedienen/automation-panel-multitouch/panels/5ap1130101e-000/

Monitor 1, BSP 2.8

Image: Toradex Embedded Linux Demo with LXDE, 2.8b6.184 (2019-04-01)

These are the relevant files for this configuration.

  • dmesg output: br_2.8_dmesg
  • fbset: br_2.8_fbset
  • xrandr br_2.8_xrandr

Monitor 1, Torizon

Image: TorizonCore image, 1.0b1.9 (2019-08-24)

These are the relevant files for this configuration.

  • dmesg output: br_torizon_dmesg
  • fbset: br_torizon_fbset

Monitor 2, comparison

This is used for comparison. With BSP 2.8, the monitor works but uses a wrong resolution of 640x480. With Torizon, the bootup image and later the login screen is show with the correct resolution of 1280x800.

Monitor: Ampire 10.1", 1280x800, DVI input. Ordered from Distronik.

Display (PDF):
https://www.distronik.de/media/files/ampire-standard/AM-1280800J3TZQW-00.pdf

Display Controller:
https://www.distronik.de/tft-displays/display-controller-displayport.html

Monitor 2, BSP 2.8

Image: Toradex Embedded Linux Demo with LXDE, 2.8b6.184 (2019-04-01)

These are the relevant files for this configuration.

  • dmesg output: distr_2.8_dmesg
  • fbset: distr_2.8_fbset
  • xrandr distr_2.8_xrandr

Monitor 2, Torizon

Image: TorizonCore image, 1.0b1.9 (2019-08-24)

These are the relevant files for this configuration.

  • dmesg output: distr_torizon_dmesg
  • fbset: distr_torizon_fbset

hi @eglimi

Thanks for the files. We will look into this and come soon back to you.

best regards,
Jaski

Here is one more information I got from the manufacturer. The EDID they expect the display to transfer is shown in the attached picture.

When I attach the display to my Linux laptop (Debian 10, SMP Debian 4.19.37-5+deb10u2), it works as expected and I get the EDID below with the following command.

edid-decode -o hex /sys/devices/pci0000:00/0000:00:02.0/drm/card0/card0-HDMI-A-2/edid out.hex

00 ff ff ff ff ff ff 00 0a b2 e5 12 d9 0c 11 41
22 1d 01 03 80 16 0e 6e 2a 85 b5 97 57 4f 94 26
20 50 54 00 00 00 01 01 01 01 01 01 01 01 01 01
01 01 01 01 01 01 c6 1b 00 a0 50 20 17 30 30 20
36 00 d9 88 00 00 00 1a 00 00 00 fc 00 42 26 52
20 41 50 31 30 30 30 0a 20 20 00 00 00 fd 00 32
4b 28 3c 07 00 0a 20 20 20 20 20 20 00 00 00 fe
00 53 30 31 20 46 30 31 20 49 4e 41 30 31 00 6c

They are both different, and neither match the one I get from the Toradex board. However, unlike the one on the Toradex board, both seems to be valid according to http://www.edidreader.com/

What EDID did you get from Toradex Board?

The EDID from the Toradex board is shown in the original question. I couldn’t find any other EDID information other than what was printed in dmesg.

Please note that I had to return the display to the manufacturer because of these problems. I can therefore not do more experiments with it. I’m sorry about that, but keeping it just for further experimentation was not worth it.

The manufacturer will continue investigating the issue. In case there are some results, I’ll report them back here.

Thank you for looking into this and for your support!

Hi @eglimi

Thanks for the Information.

The EDID from the Toradex board is shown in the original question. I couldn’t find any other EDID information other than what was printed in dmesg.

If you compare the EDID Information provided by your desktop and Apalis iMX6, then you can see that on Apalis iMX6 every second byte is repeated. So that’s why I think the problem lies on the reading data of EDID. So either the problem is i2c connection or the i2c driver.

Please note that I had to return the display to the manufacturer because of these problems. I can therefore not do more experiments with it. I’m sorry about that, but keeping it just for further experimentation was not worth it.

That’s pity. You can ask the manufacturer to check on readout the EDID information using i2c command and check if the Information you get is correct.

Thank you for looking into this and for your support!

You are welcome.

Best regards,
Jaski

Hi @jaski.tx,

Thank you for the interesting findings! With this, it’s really a pity I had to return it… Nevertheless, I’ll contact the manufacturer about this and update the post if anything new comes out of it.

Thanks!

Hi,
you are welcome. Sorry that we did not see this before.
We will be glad to get any news.

Best regards,
Jaski