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 Transmitter Issue with Centaurus

Other Parts Discussed in Thread: TPD12S521
Hi Feroz,

We are using Centaurus SOC(DM8148) in our system and we are not able to retrieve color patterns when we loaded "out file" of Centaurus EVM using cvs.

We have followed all the required constraints of HDMI except that of providing 4 vias instead of 2 Vias as Centaurus and HDMI Connector are in separate boards. Also there is a Board to Board connector (BSH/BTH Series -Samtec) between two modules

Please let me know whether this deviation of SI could cause problem with our functionality. In the current condition we are not able to receive even HDMI clk from the SOC.


Regards
Kinjan Patel
eInfochips Pvt Ltd. 
M: 8758644002
  • Remove the 40.2E resistors. It is not allowed to add any resistance here. (R277, R272, R273 & R274)

    Please use the EVM schematics as a reference.

    Additionally the connector MUST be a high quality impedance matched connector since the signals going across this connector are up to 1.65GBits/s and have very strict integrity requirements.

    BR,

    Steve

  • Hi Steve,

    Thanks for your suggestion

    I have removed the 40.2E resistor and Ferrite bead also but still i am not able to attain my functionality.

    I can predict the SI issue but still i am not able to understand that because of SI will i not receive even a clock? There is no sign of any signals in the receiver side.

    Also when i read the system status register (SYS_STAT) which shows 00000083 when receiver is connected. This reads in RSEN field that "No receiver connected". what can be the cause for this?

    WR,

    Kinjan

  • Are you loading the drivers, enabling the display and putting anything into the frame buffers?

    What do you mean by "SI will not receive even a clock"?

    Do you mean that you are not seeing the HDMI clk signals wiggling?

    The status registers only report the HPDET status and are not necessary to enable the display, but MAY be used by the window manager to change the output resolution depending on what window manager you are using (if any)

    What software are you using? What steps, exactly, are you following in order to turn on and enable the HDMI output.

    BR,

    Steve

  • SI means "Signal Integrity". As you noted earlier that B2B Connector has to be impedance matched. But our connector is not impedance matched. Is this the reason that there is not HDMI clock or data wiggling?

    We are using CCS-V5
    Using the Gel scripts (IO Expansion_test_2.x) DDR3 Controller Initialization and HDMI Clock Initialization was carried out
    On executing the GEL script, a color bar pattern should be displayed in the HDMI LCD TV.

    But we are not receiving color bars in our LCD TV even after loading the HDMI bin file (TI814x_A8_HDMI_Sample.out). Please Check attachment.

    1462.HDMI CCS Fles.zip

    WR,

    Kinjan

  • Bad news.

    If you look at your schematic, you've isolated your TMDS inputs to the TPD12S521 from their corresponding outputs by using different signal names. I think they need to be the same nets. The TPD12S521 isn't "flow-through", at least in the DM8148 EVM reference schematics, which show input connected to output using same net names. Page 2 and 3 of the TPD12S521 seem to agree with me. Note that the TMDS lines aren't drawn on the board schematic, and the layout schematic shows traces routing TMDSI pins to TMDSO pins as board traces under the part. The EVM schematic also has a note that "DIFFERENTIAL SIGNALS ROUTE THROUGH ESD DEVICE" which could be taken either way, but if the part passes through the signals, why would the note only say "DIFFERENTIAL SIGNALS"? I'd check the layout for you, but our engineering server crashed yesterday, and I can't get to it easily.

    If you have a loose TPD12S521 laying around, you can check the part's continuity too, e.g. is pin 4 connected to pin 35?

    You really want a 100 Ohm connection through the connector. The connector's in the middle of the line, so it's extra critical. Those RL filters should go away (as suggested above)along with their traces. The signals really aren't that touchy, but its what folks like us do, and you want to be a member of the club, right? If you do the job correctly, when a customer tries driving 30 ft. over a cheap cable and complains, you won't feel guilty when you tell him it must be the cable.

    If I'm right, you should be able to run the differential signals 30awg over the top of the TPD12S521, making the required connection.

    Note - I'm not a TI guy.

     

  • Herb,

    Excellent catch !!!

    I didn't notice the net names were different.

    If you reference the TPD datasheet it shows exactly how the signals should be routed on page 3.

    Regarding the signal's not being 'touchy', this is not quite true for HDMI. These signals are VERY sensitive to impedance matching since they are running at 1.6GHz. Trace impedance matching is critical for eye diagram compliance, and we have seen some customers have issues with the eye diagrams when impedance matching has not been considered as much as necessary. Usually this has not affected operasion, but it has affected compliance testing.

    Thanks for looking at this, your observation is a critical one.

    BR,

    Steve

  • Hi Steve / Herb,

    I have done the changes in my board as required and now HDMI is working fine. Thanks for your review and observation.

    WR,

    Kinjan