• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Subscribers 34 subscribers
  • Views 684 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.

FPD-Link III in a noisy environment

Randy Luck
Prodigy 240 points

We have a system transmitting digital video data from an imager via FPD-Link III using a DS90UB913Q to a DS90UB914Q.  We are transmitting this video data in the 70 or so MHz clock rate range.

The link uses a very fine 32 AWG 100 ohm twin ax cable.  The system physical requirements force us to use this very small cable.  This set up works fine most of the time.

However, in bursty situations, the environment is VERY electrically noisy, especially in the band from about 20 kHZ to 1 or 2 MHz or so.  This noise is mostly coupled in common mode.   Things like ferrite beads are useless at these low frequencies.

Is the FPD-Link III with the '913/'914 set the best silicon choice in a very noisy environment?

Is there a common mode limit on the '914 inputs?  Is this limit just basically power and ground?

Would TVS type devices help?

Can I use a high pass filter on the FPD-Link III inputs?  i.e. can I use a much smaller capacitor than the 0.1uF that all the data sheets show?  Or maybe an LC filter with a high pass starting at about 4 or 5 MHz?

Thanks!

  • 0
    TI__Mastermind 20820 points

    Hi Randy,


    Someone is looking into this for you and will respond shortly.

    Regards,

    Michael Peffers

    High Speed Interface Applications

  • 0
    TI__Genius 13475 points

    Randy

    The DS90UB913/914 communicate with one another on two bands - the forward channel contains your data which, with a 70MHz clock will be at about 980Mbps (there is some overhead), and with our encoding, this will be in a band between about 50MHz and 500MHz.   The backchannel which is used to allow the '914 to communicate with the '913 has a carrier at about 2MHz.   My guess is that the upper portion of your noise band is interfering with the lower portion of our backchannel band, and that this is causing your issues.   I would agree that it is very difficult to filter the entire band of your noise with much of it being at very low frequencies, but you may get significant relief by just trying to attenuate the 1-2MHz portion of that band.

    The DS90UB913A and the DS90UB914A are pin compatible versions of the parts that you are using, but they are able to support coax cable rather than the differential signals that you are using now.  For a very thin cable/connector environment, you may be able to get a better shielding environment with coax rather than twinax.

    Mark Sauerwald

     

  • 0 in reply to Mark Sauerwald88474
    Prodigy 240 points

    Mark-

    Thanks for the info.  The noise we have to deal with is worst generally below 100 kHz.

    On one of our '914Q Eval boards, I removed C4 and C5 and substituted each cap with a tacked in a 3 pole high pass LC filter designed for a 100 ohm impedance.  Using Tina this filter is flat down to 300kHz, about -1.5 dB at 100 kHz, -17 dB at 30 kHz, and -46 dB at 10 kHz.  In our real-world situation, this LC filtering actually does help some and makes the link more reliable.  (For reference I find that the standard 0.1 uF circuit is flat to 100 kHz, -0.4 dB at 30 kHz and about -2.3 dB at 10 kHz.)

    We will take your advice and try the '913A/'914A chipset with 50 ohm coax.  In addition to the reasons you listed, we should also see an additional benefit because the characteristic impedance will be cut in half making it less susceptible to external RFI.

    -Randy