• State Resolved
  • Locked Locked
  • Replies 9 replies
  • Answers 3 answers
  • Subscribers 34 subscribers
  • Views 410 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.

DS90UB954-Q1: Backchannel eye diagram when SS oscillator is used

Vicent Climent14
Intellectual 360 points
Part Number: DS90UB954-Q1

Dear TI members, 

We are currently validating the Vod-bc requirement of the back channel.

We have tested under two different conditions.

  1. Oscillator without spread spectrum --> Pass the Vod-bc requirement from your datasheet Vod-bc = 400mV (spec: 380mV to 520mV). See osc caputres below.
  2. Oscillator with spread spectrum --> Fails the the Vod-bc requirement from your datasheet Vod-bc = 280mV (spec: 380mV to 520mV). See osc caputres below.

We have been testing in the past without SS, but to have a pass in EMCs we need to use a SS clock. In our case we are using +-0.1% deviation. In the past a former colleague was already in touch with you to define the SS required in this post --> e2e.ti.com/.../870913

Is it possible to relax the Vod-bc spec if SS is used?

Thx for your support,

Vicent Climent.

WITHOUT SS

WITH SS

  • 0
    TI__Genius 11340 points

    Hi Vincent,

    The SSC should only affect the frequency, not the amplitude of the signal. Could you check your oscillator input?

    Best,

    Jiashow

  • 0 in reply to Jiashow Ho
    Intellectual 360 points

    Hello Jiashow,

    I have just realized that for some reason the pictures are not attached. Find then now attached at the end. 

    The oscillator input amplitud seems to be okay. Vhigh = 3.24V and Vlow = 0.1V

    Let me explain how we are measuring Vod-bc, maybe it is not the correct way. First we plot the backchannel eye diagram by following steps recommended on this thread --> https://e2e.ti.com/support/interface/f/138/t/877979.

    Once we have the eye ploted, we measure the eye height, and consider this to be our Vod-bc. As we are using an spread spectrum clock, there is some jitter that can be observed in the eye diagram and hence it is much closed than the equivalent eye without SS enabled. 

    I was wondering if the eye height is not a good parameter to measure the Vod-bc or if there is some other way to get this measurement when the SS is enabled.

    WITHOUT SS

    WITH SS

  • 0 in reply to Vicent Climent14
    TI__Genius 11340 points

    Hi Vicent,

    In your scopeshot, with SSC the freq is 50MHz but without SSC the freq is 32MHz? The back channel at sync mode is running at 50MHz. This makes me suspect the measurement wasn't done correctly.

    You can actually measure the back channel without a forward channel. The easiest way to do this is to unplug the serializer, turn on the deserializer back channel, and check it on the scope. You can even do single-ended measurement by probing Rin+ and GND. I would suggest probing the AC cap on the side closer to the pin.

    Best,

    Jiashow

  • 0 in reply to Jiashow Ho
    Intellectual 360 points

    Hi Jiashow,

    Thanks for your reply. 

    Actually the freq measurement was wrong as the parameter was not measuring the recovered clock from the oscilloscope SW tool, but making some measurement directly on the raw backchannel signal.  (If you check the mean freq for the measurement with SS and without SS is the same ~42MHz)

    The rest is being done as you proposed, so actually, the eye diagram is the same as the one shown before. When using SS the eye diagram is much more close, which of course is expected. The problem is that when the SS is enabled we are not fulfilling the Vod requirement. 

    Br,

    Vicent

  • 0 in reply to Vicent Climent14
    TI__Genius 11340 points

    Hi Vicent,

    Can you do a comparison of the regular oscillator output and the SSC output?

    Are you getting lock when you use an SSC?

    Jiashow

  • 0 in reply to Jiashow Ho
    Intellectual 360 points

    Hello Jiashow,

    In terms of Vhigh and Vlow the behavior is very similar. When checking the jitter, the one with the SSC is much higher as expected. See pics below.

    Yes, when using SSC we are getting lock. The ECU is completely functional with and without SS activated, but we are not fulfilling the Vod requirement in this case. 

    Oscillator without SS:

    Oscillator with SS:


  • 0 in reply to Vicent Climent14
    TI__Genius 11340 points

    Hi Vicent,

    Both scopeshots show the correct amplitude. Are you referring to the back channel differential output? Can you measure it on the scope?

    Best,

    Jiashow 

  • 0 in reply to Jiashow Ho
    Intellectual 360 points

    Hello Jiashow,

    Exactly, the Vod-bc problem I am referring is on the back channel differential output when we are using SSC. The osc captures were shared on my message from November 12. Find it attached in this msg too. 

    What we are trying to validate is the following parameter of the datasheet:

    So we are correlating Vod-bc with our eye height measurement. Maybe this correlation is wrong, and we should just simply measure Vmax in single ended.

    Below the pics of Vod-bc:

    VOD-BC WITHOUT SS

    VOD-BC WITH SS

  • 0 in reply to Vicent Climent14
    TI__Genius 11340 points

    Hi Vicent,

    I see your point regarding the eye height. We don't have characterization data on this with SSC, so it's difficult for us judge whether eye height can affect the link performance.

    Since you're seeing no issues right now, what I would suggest is instead of looking at the eye height, run the margin analysis program to check the margin with SSC: 

    If your margin is good, then the link should be good.

    Best,

    Jiashow