Announcement

Collapse
No announcement yet.

Ti AM335x -- LCD very strange things happen!

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Ti AM335x -- LCD very strange things happen!

    Dear all,

    I'm writing here my first post because I have some really strange thing happening. I hope to find a experience pcb designer who can point me in the correct direction,

    First a short intro of my project:
    I'm building an embedded system with a TI AM335x system on module(SoM). For this SoM I builded a carrierboard. The embedded system will be used in my tractor. Here I want to read can data from the bus into the system.
    My first prototype was a 4 layer pcb. Here I had no problems with the LCD (LVDS 1024x600). Now due to other imperfections I made changes to my pcb.
    Also due to costs I reduced the amount of PCB layer to 2. Which I think should not be a problem.

    Now Last week I was testing everything. The canbus works, the outputs work... Then I found out that my LCD screen doesn't work correctly. The colors are incorrect. When I push my 50 pin connector it gets better. So I thought that it was probably the connector which wasn't soldered very well. So I desoldered the connector, cleaned the pcb and resoldered a new conenctor. After I inspected the connector under a microscope and everything looks very nice!

    So I tested the LCD again. I still have the same problem. When I push the ribbon cable between the LCD and connector, the colors changes. At this point I started to stress!

    After I contacted the vendor and asked him if he has an idea of whats going wrong. He told me that it was probably that my transmission line impedance isn't 50 ohm (Which he never told me it should be!) and that I need to refabricate my pcb and ask to match my LCD data lines to 50 ohm.

    Now I'm looking for somebody who could tell me this is possible and give me some support on how to test this.

    I'm very sure that the problem is within my pcb and that it is not software related. On the previous board there where no issues with the screen.
    Attached links are images of the schematics of the screen and a printscreen of the pcb.

    So, can somebody confirm what the vendor says?
    Is it possible to test this changes? Or should I try other things?
    Maybe changing the 33ohm resistors on the pcb to an other value?

    Hopefully somebody can give me some hope.
    Kind regards,
    TMJJ

  • #2
    The problem can be connect to the the change of your stackup - going from 4 layers down to 2 layers. When you touch something, you are changing a lot of paramaters, it doesnt neccessary mean, that the connector is wrong. You may be touching contacts, pins, components - all these may have influence if your PCB layout is too sensitive.

    Especially when I see all the tracks routed very close too each other for a long distance, on 2 layer PCB this is something where a lot of crosstalk may happen - that may damage your signal quality

    Comment


    • #3
      Hi,

      Thanks for the reply.
      So then there is no other way then redesign my pcb and take this into account? Or is there a way to try to make it more stable by for example playing with the resistance which is places in serie with the LCD data lines?

      The reason I ask this is for the following. I have an LVDS LCD and not a RGB. When I compare the schematics from Texas instruments, they make a difference between those two schematics, where the RGB has 33 ohm resistors and LVDS not. So I would say I don't need the resistors (I have no clue that it will have an influence). the vendor of my board says that it doesn't matter.



      It's a pitty. Like I already said, I don't have the big budget. So I will need to have a look how I should solve this if I need to produce a new board.



      Kind regards
      Attached Files
      Last edited by TMJJ; 04-26-2018, 07:05 AM.

      Comment


      • #4
        So then there is no other way then redesign my pcb and take this into account?
        - I would redesign it. When I was starting with HW design, I had similar problems with 2 layer boards. Since then, I almost never use 2 layer PCBs (especially if there is a bus).

        You can play with resistance, however be aware, that even if you make one board somehow work, other boards may still be randomly failing (and you can also see weird behavior between different batches of boards - one batch will work fine, other will not ... not saying you may get complains from different customers as you do not know how and where people are going to use your board). If you do not use 33OHM resistors, I believe you may get even worse results (when you replace them by 0R then crosstalk between the tracks will be probably even higher) - you can try.

        Comment

        Working...
        X