• State Resolved
  • Locked Locked
  • Replies 8 replies
  • Subscribers 32 subscribers
  • Views 8235 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

HDMI/DVI to LVDS Bridge 24bit or 18 bit mode

Thomas Lapauw
Prodigy 50 points
Other Parts Discussed in Thread: TFP401, DS90C387, DS90C387A

Hi,

I am trying to build a HDMI to LVDS bridge using this pdf as a guide: www.ti.com/.../slla325a.pdf

Using the TFP401 and the DS90C387 for a dual channel LVDS screen. But now I am lost on how the DS90C387 decides to output 2 channel 6bit data (2*3 pairs + clk) or 2 channel 8 bit data(2*4 pairs + clk). The datasheet states it is capable of both.

Does it infer the mode from the data on the input or are the LSB's pixel data mapped on the link that is used for 8 bit but not for 6 bit?

I am sure I am missing something butI am out of ideas on where to look.

Thanks for the help in advance

Thomas Lapauw

  • 0
    TI__Mastermind 29655 points
    Hi Thomas,

    Thanks for your post and my apologies for the long response delay. Please allow me an opportunity to look over it and get back to you by next week.

    Michael
  • 0
    TI__Mastermind 29655 points

    Hi Thomas,

    I looked over the app note to see where the confusion may have come from.

    If you look at the way that the pins are mapped on Figure 4, you will notice that 2 bits of each RGB color are presented on A3 and A7, the highest LVDS pair:

    The reason for this scheme is for instances where only 18-bit RGB information is available. If only 18-bit information is provided, it is desired that these bits are conveyed on the six MSBs of the RGB data. The designer has a choice of mapping either the two MSB of the RGB values into A3 and A7, or the two LSB of the RGB values instead. It is common to see the LSBs of 24-bit data mapped to A3 and A7. Therefore, if 18-bit RGB data is sent, it will still pass on the MSBs of A0-A2 and A4-A6 normally, while A3 and A7 bits will be unconnected on the DS90C387A inputs.

    So, to answer your question, 18-bit or 24-bit data is not inferred by the input, but rather the LVDS mapping is done in such a way that the LSBs are mapped to the same LVDS pairs, which can be ignored when interfacing 18-bit RGB.

    Are you facing a situation where the system will not know a priori whether the input is 18-bit or 24-bit?

    Thanks,

    Michael

  • 0 in reply to Michael Lu (Santa Clara)
    Prodigy 50 points

    Hi Michael,

    Thanks for the answer already.

    When mapping the LSB 24 bit data to A3 and A7 the output is then conform the JEIDA standard? 

    I am facing a situation where interoperability between an 18bit and 24 bit LVDS output is a plus. So using the jeida format will improve compatibility.

    Thanks in advance

  • 0 in reply to Thomas Lapauw
    TI__Mastermind 29655 points

    Hi Thomas,

    This mapping conforms to the Open LDI standard, which is supported by JEIDA (see Open LDI Press Release (1999)).

    I checked with the JEIDA DISM Standard, and I was able to verify that this matching is compatible with JEIDA.

    For a copy of the JEIDA and Open LDI standard, please see attached:

    JEIDA-59 1999: JEIDA_DISM_Standard.pdf

    Open LDI Specification: openldi.pdf

    In particular, if you reference either of the two standards, you will see the method for mapping 18bpp versus 24bpp, and this matches the mapping scheme suggested by the DS90C387A. See the JEIDA standard p. 19 table below:

    Thanks,

    Michael

  • 0 in reply to Michael Lu (Santa Clara)
    Prodigy 50 points
    Am I correct stating that interoperability between a 18bit and a 24bit input is not possible since 24bit adds 2 LSB's which then won't match the LVDS data mapping.
  • 0 in reply to Thomas Lapauw
    TI__Mastermind 29655 points

    Hi Thomas,

    The 18 bpp and 24 bpp can interface directly.  When using 24 bpp, the 2 LSBs for each RGB value travel on the highest LVDS pair, and this highest LVDS pair remains unused when operated in 18 bpp. When switching to 18 bpp, the 6 bits per color should map to the 24 bpp MSBs. For example, 18 bpp R[5:0] => 24 bpp R[7:2].  This is part of the standard for interfacing 24 bpp and 18 bpp, so there should be no issue between having 18 bpp interfacing into a 24 bpp serializer.

    Here is another resource regarding mapping 24 bpp v. 18 bpp (LVDS Display Interface (LDI) TFT Data Mapping for Interoperability with FPD-Link)

    Overall, the ability to map the bits correctly to the right Tx is dependent on whether the inputs to the DS90C387 comply to the same JEIDA/OpenLDI standard for mapping. If they do, then they should correctly map the 18 bpp information to LVDS pairs 0-2 while leaving LVDS pair 3 unused.

    Thanks,

    Michael

  • 0 in reply to Michael Lu (Santa Clara)
    Prodigy 50 points
    The LVDS output can be made compatible between 18 and 24 bit. This is the answer I was looking for.

    An other question:

    How does the TFP401 handle the difference between 18 and 24 bit input data? Does it map the 6 bit data to [2:7] when only 6 bit/color is available or does it map it [0:5].

    for example Odd Blue in 6 bit to QO[2:7] or to QO[0:5], leaving the other bits 0
  • 0 in reply to Thomas Lapauw
    TI__Mastermind 29655 points
    Hi Thomas,

    Thanks for the clarification. I think I understand what you are looking for. It looks like you are searching for a way that either 24 bpp or 18 bpp data can come into the TFP401, with the 18 bpp data mapping to the MSBs of 24 bpp data locations.

    Unfortunately, the TFP401 and the DS90C387 do not have a built-in automatic way to re-map 18 bpp data into the 24 bpp scheme. The TFP401 takes in the TMDS data it receives and maps to CMOS the same way. Likewise, the DS90C387 takes the CMOS data it receives and maps to the LVDS pairs the same way.

    To move the 6-bit RGB [5:0] values into the 8-bit RGB [7:2] locations, some rewiring or remapping external to the TFP401 and DS90C387 is required. To do this after the TFP401, you may be able to use an intelligent FPGA before the DS90C387A that can remap the RGB 6-bit information to the correct RGB 8-bit mapping locations. Alternatively, if the TMDS transmitter before the TFP401 is able to do a remapping to move the 18 bpp data into the 24 bpp MSB locations, then the 18 bpp can be directly interfaced with with a 24 bpp.

    Thanks,

    Michael