• State Resolved
  • Locked Locked
  • Replies 9 replies
  • Answers 1 answer
  • Subscribers 33 subscribers
  • Views 1021 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

DS90C387: two missing LVDS outputs for a 10k produts series

Remi Vendeville
Prodigy 50 points
Part Number: DS90C387
Other Parts Discussed in Thread: TFP401,

Hello

I'm working on a prototype for a large series (10k products in the firsts years) of audio-visual products, that includes an HDMI to LVDS Open LDI conversion between the CPU and the monitor. I have issues with video transmission on this product, that are critically blocking the prototyping phase and so the serial industrialization phases.

The solution that was chosen for this was coupling two TI chips, a TFP401 followed by a DS90C387 chip. The schematic was done according to those component's datasheets, typical applications and evaluation boards examples that I found, and I have been even more confident in my design after seeing a board similar already sold on the web (link). The wire between the board and the screen is quite short (about 20cm).

Though, during validation phases I'm facing an issue on the video transmission, and it appears that two LVDS channels aren't working on the output of DS90C387 chip, A0M/P and A1M/P. The voltage on those pins stays at about 2.5V for the A0M & A1M output and 0VS for A0P & A1P. I have looked into each signal on my video connector without having a screen connected and on those 4 tracks I see a quite constant voltage whereas I can see some signals looking like frames on A2P or A4M for example.

After several chips replacements and debug tests, I'm asking for some support because I don't know where to search now, there aren't so many configuration entry and even if I tried to change them slightly none of them seem to be related to this particular issue in chip description.

Thank you very much for your quick answer,

Regards,

Rémi

 

HDMI_LVDS_conversion.pdf

  • 0
    TI__Mastermind 35130 points
    Hi Remi,

    It looks like you have PIXS pulled low on the TFP401. Looking at the rest of your schematic, I believe it should be pulled high for 2-pixel/clock operation.

    Also, does your HDMI include audio? The TFP401 does not support HDMI with embedded audio.

    Regards,
    I.K.
  • 0 in reply to Ikechukwu Anyiam
    Prodigy 50 points
    Hi

    Thanks for your quick answer, unfortunately changing the PIXS command level didn't solve the problem. Also, the board I mentioned as a design example seem to use it the same way (and so do they for the equivalent pin of the DS90C387, DUAL). I would be very surprised if the hazard of a single / dual pixel configuration does affect the entire bits of the two LVDS channels and only those two channels.

    About audio, it should not be sent on HDMI but as I can't see the UI on the screen I can't check it properly. I guess that audio data are just ignored by the TFP401 chip, or may it disturb the transmission?



    Regards
    Rémi
  • 0 in reply to Remi Vendeville
    TI__Mastermind 35130 points
    Hi Remi,

    I'm not familiar with the board you're using as an example since TI did not design it.

    For your application, the PIXS pin on the TFP401 needs to be pulled high and the DUAL pin on the DS90C387 also needs to be pulled high. The way you have it configured right now, the TFP401 is only outputting data on pins QE[23:0]. Please see the datasheet for details. I would also recommend tying SCDT to PDO_N on the TFP401.

    If HDMI audio is present in the TMDS datastream, it will cause the TFP401 to output incorrect sync information, so you should not have any audio at all in the stream.

    Regards,
    I.K.
  • 0 in reply to Ikechukwu Anyiam
    Prodigy 50 points
    Hello

    Thank you for all this, I am still working on this issue but I had some problems to deal with (broken chip and broken screen due to testing and debug stresses...)

    It finally appeared that the board I use as an example do use TFP401's PIXS and DS90C387's DUAL pins as you mentioned it, but it seems that they haven't updated there available schematic yet. We have also found that I had inverted horizontal and vertical synchronization pin wiring between both chips.

    I have been able to see some blinking good pictures on my screen before it stopped working, so I assume that the wiring is alright now but some components are damaged, and/or some configuration pins aren’t well set. I am trying to clean this up, and then I'll let you know whether it works or not.

    Regards
    Rémi
  • 0 in reply to Remi Vendeville
    TI__Mastermind 35130 points
    Hi Remi,

    Understand, please update thread once you have new results.

    Regards,
    I.K.
  • 0 in reply to Ikechukwu Anyiam
    Prodigy 50 points

    Hello

    I am close to have this perfectly operational. The screen now prints some things, but it isn’t clean.

    On the attached picture you can see an example of printed text in a message box and a picture in background. This picture is supposed to be made of a single large white line on a black background, and we can see some noisy thin white and grey lines close to the limits between black and white.

    I am investigating, but I was thinking that maybe it is a common failure case that you have already identified and that you might be able to tell me more about.

    [EDIT] I think I have finally identified the problem, it looks that odd and even pixels are switched somewhere. You can figure this out by having a look at the test pattern and test results that we had (I didn’t take a picture, I just reproduced what I saw).

    Do you know if there I a way to switch odd and even pixels either in TFP401 of in DS90C387?

     

    Thanks for all

    Best regards

    Rémi

    [Test pattern]  : 

    [Test result]  : 

  • 0 in reply to Remi Vendeville
    TI__Mastermind 35130 points
    Hi Remi,

    Are you able to share the datasheet for the display that you are using? We can double check the even/odd parameters with the mapping required by the display.

    Regards,
    I.K.
  • 0 in reply to Ikechukwu Anyiam
    TI__Mastermind 35130 points
    Hi Remi,

    I will go ahead and close this thread for now. Please reply again if you have further questions/concerns.

    Regards,
    I.K.
  • 0 in reply to Ikechukwu Anyiam
    Prodigy 50 points

    Hello

    Sorry we have been quite busy trying to solve this problems those times.

    Finally we solved it by inverting odd and even outputs of the DS90C387, but we haven’t been able to figure out why we had to do that, probably some misunderstood naming convention in entry of our screen.

    Anyway, thanks for all, the TFP401 + DS90C387 solution for HDMI to Open LDI LVDS conversion is now operational for us.

    Regards

    Rémi Vendeville