• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Subscribers 36 subscribers
  • Views 1042 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.

question about LVDS eye diagrams

Mike Valvo
Intellectual 470 points


hello, I'm setting up an LVDS test bed where I'll be hooking up my oscilloscope at the receivers to generate eye diagrams.  My question: if I had several inches of wire from my PCB to the scope, is that going to introduce a "stub" and possible degrade the LVDS signals?  My PCB will be contained in a metal box where the leads for the eye diagram will go to a panel mount connector, so the several inches of wire may not be terminated all the time.  In the past I've experienced degradation if I've had wires on the bus that are not terminated.  any thoughts or suggestions are appreciated. thanks

  • 0
    TI__Guru* 93570 points
    Hi Mike,

    Based on the data rate, this several inches of wire could act like stub and or antenna and you may see signs of reflection(based on the edge rate). The best case is to use a very capacitance high impedance probe, like Tek P7330, and attached to the receiving pins. If this cannot be achieved, please be mindful of your setup and take the side effects into your analyses.

    Regards,,nasser
  • 0 in reply to Nasser Mohammadi
    Intellectual 470 points
    we have the Tek TDP1000 differential probes, which have an input impedance of 1MEG. I think what I'll do is have a PCB connector as close to the receiver as possible and mate/connect wires to the panel mount connector only when I'll do eye diagram tests. when not connected, hopefully the stub will be short enough so it won't interfere. given the high impedance of the differential probe, should I expect that this connection will not cause reflections/degradation? the data rate is slow, 2Mhz

    is this what you would do?
  • 0 in reply to Mike Valvo
    Prodigy 100 points

    Let me start by saying that for the data rate you are working on, with reasonable length(2-3 meters) impedance matched cable from TX to RX, the eye will be ridiculously wide open to justify an eye diagram verification. The transmission is solid by design, and you can fairly casually connect PCB to the panel mount connector you want to use.

    The following is for data rate at around 100Mhz range.

    To achieve optimal performance you could layout two SMA/MCX/MMCX connector to the differential pair, and layout a panel mount PCB, which has a connector to your Scope/TDP1000 probe and another pair of SMA/MCX/MMCX connectors. Then you can use a pair of RF coaxial cables to connector the PCB test connectors to the panel mount PCB. The stub caused by the RF connectors on your PCB is relatively benign for the frequency you are verifying (100Mhz with roughly 1-1.5G edge rate). Be sure to layout them as close as possible to the pairs you want to verify.

    If the real estate on your DUT PCB can not accommodate two RF connectors, then just simply have two sets of test points on the differential pair and  solder two RF cables, center each to the differential pair, and shielding to PCB ground. This approach is not as reflection free as the permanent test connectors approach, but for your working range, I believer you will still do fine.

    When you split and solder the RF cables to the PCB, be sure to watch for two things:

    1. the split should be kept as short as possible.

    2. There need to be some sort of removable strain relief between your PCB and the RF cables to ensure reliable connection.

    When you layout the panel mount PCB, obviously the traces from the RF connectors to TDP testpoints should also be impedance matched.

     

    Good luck.