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.

TMDS181: Faild in HDMI2.0 Intra-pair skew compliance test (HF1-4)

Guru 19785 points
Part Number: TMDS181

Hi Team,

Our customer failed in HDMI2.0 compliance test HF1-4, Intra-pair skew.

Could you please take a scope shot of the TMDS181 with HDMI2.0 signal which Intra-pair skew could be read ?

It would be helpful if you could take data for D0, D1, D2, and CLK.

Best Regards,

Kawai

  • Hello Kawai,

    Was this this failure on the EVM or a board that you designed yourself? 

    -Michael

  • Hi Michael-san,

    This is based on customer board, so I understand that this is probably application problem.

    I would like to proof that TMDS181EVM doesn't show any problem in the intra-pair skew requirement.

    Since we don't have the high speed scope, it would be helpful if you could help us to convince customer with the intra-pair skew scope shot.

    Best Regards,

    Kawai

  • The TMDS181 has passed all the compliance tests for HDMI 2.0 on the EVM, so we do not see the same issue that your customer is having with the intra-pair skew. Could you send me the layout and schematic file and I can see what is wrong form there?
  • Hi Michael-san,

    First we would like to convince customer by showing the waveform information that the device is innocent. Do you have any data ?

    Best Regards,
    Kawai
  • Here is a passing HDMI 2.0 test report for the TMDS181. I do not have any of the scope shots for the intra-pair skew test.

    TMDS181_compliance.pptx

    Also please look at the following sections of the datasheet to improve intra-pair skew.

    which lanes are failing the intrapair skew test? Also the CLK does not have to pass the intra pair skew test for HDMI but some equipment tests for it and should not.

  • Hi Michael-san,

    Sorry for my delay. Customer tried with PRE_SEL= -2dB. Intra-pair skew had improved, however, V-swing (amplitude) had decreased then got out of the requirement.

    They had increased VSWING through Register 0x0C, however not working well.

    When changed Register 0x0D [0] = 1, intra-pair skew and VSWING passed the requirement, but now fall time exceeded the requirement.

    As described in 9.2.3.1.2.1/2, VSADJ resistor may be needed to be changed. I believe this resistor sets the typical value of the VSWING, then with the REG 0x0C, user could fine tune VSWING by +/- 7%,14%,21%. Am I correct ?

    Best Regards,

    Kawai

  • Kawai-san

    Your understanding is correct. The resistor on Vadj set s the default value of the VSWING, and then you can use register 0x0C to adjust by certain percentage.

    Thanks
    David
  • Hi David-san,

    Thanks for your comment. I am asking customer to tweak VSADJ resistor and VSING regsiters (REG 0x0C).
    Please allow me to ask questions again if they cannot solve the problem.

    Best Regards,
    Kawai
  • Hi David-san,

    They were able to pass the compliance test by tweaking the VSADJ resistor.

    We had feedback from customer.
    Regarding the Compliance test report which Michael-san had attached above, R_VSADJ = 7.06kohm may not be the optimized value. Our customer also had the similar result and when they tested several times, they observed about 20mTbit variation for Intra-pair skew measurement each time. Since the report says Intra-pair skew is 139mTbit, requirement is +/-150mTbit, they think if the test was done for more than 10 times, you may see it fail.

    Best Regards,
    Kawai
  • Kawai-san

    The R_VSADJ value depends on the system design, it needs to be tuned to set the appropriate Vod, pre-emphasis and termination level in order to pass the HDMI compliance. The 7.06kohm is recommended value as this value enables most system to pass the compliance.

    Thanks

    David 

  • Hi David-san,

    We have two additional questions from customer.

    Could you please share us the actual register setting which was used when TMDS181 passed the Compliance test ?

    Customer cannot get the similar result even tested with the EVM. TI compliance test result has much better tr/tf with high VSWING.

    In there further evaluation, they were able to get similar performance by changing the "Slew Rate Control" bar in the EyeScan Tool.

    This seems to be changing the RESERVED register, Page0 Addr 0x0B [7:5] and Page1 Addr 0x13.

    They would like to use this function since this setting improves the rise/fall time to have more margin and which also improves EMI performance.

    Could you please disclose what these registers/bits are doing ?

    There seems to have 25% difference in rise/fall time when Slew Rate Control bar is set to minimum vs maximum.

    Best Regards,

    Kawai

  • Kawai-san

    TMDS181 compliance test was done awhile back, I do not have the setting we used for passing the compliance.

    Page 0 Addr 0x0Bh bit [7:6] changes the slew rate and Page 1 Addr 0x13h updates the change in slew rate. But writing to Page 0 Addr 0x0Ah would also update the change in slew rate.

    The characterization we have done on TMDS181 shows the size of adjustment is very small and doesn’t impact slew rate much, so we do not publish it in the datasheet.

    Thanks
    David
  • Hi David-san,

    Thanks for your continuous support and the register detail information.

    We understand that these registers are reserved and not open to public, however, if customer wants to use this can they use these registers ? Our understanding that Page 0 Addr 0x0B [7:6] setting is the following, are we correct ?

    "00" : Maximum slew rate
    "01" : 1 step down
    "10" : 2 step down
    "11" : Minimum slew rate

    One more question I have is that somehow I cannot change the slew rate control bar in the EyeScan Tool (gray out) which our customer can change without special configuration. Is there any limitation ?
    My EyeScan Tool version is V1.0.0.0, Windows7 64bit OS laptop.


    Best Regards,
    Kawai
  • Kawai,

    If changing the slew_rate improves the customer performance they may do so. Our lab validation showed little impact from slew_ctl settings. Your interpretation of the values is correct. EyeScan is not enabled to impact slew_ctl through the control bar for the TMDS181. Regsiter settings must be used.

    Regards,
    JMMN