• State
  • +1 person also asked this people also asked this
  • Replies 5 replies
  • Subscribers 104 subscribers
  • Views 81 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

TDA4VP-Q1: LVCMOS RX IBIS Model

kcastille
Guru 50762 points
Part Number: TDA4VP-Q1

Tool/software:

Hello, Please use this thread as a continuation of the email and TI-Internal e2e forum discussion.

  • 0
    Prodigy 30 points

    Hi,
    As discussed previously, I see meaningful ripples when using the TDA as my RGMII Rx.
    Those ripples are PVT and rise time depended. 
    Please see below plots with:
    1. TDA at PVT-max (DS-LVCMOS0_fast)
    2. Tx is an ideal 50 ohm driver, Tr-0.4ns, clk pattern 4ns UI.
    3. For debug - used same setup with RLC circuit from TDA ibis file and compared against 3rd party RGMII Rx RLC.


    Please see below plots for PVT max (where ripples are strongest):


    I suspect the root cause is in frequency domain of RLC - look at return loss plots below:

    Waiting for reply,

    Yoel

  • 0 in reply to yoel biberman
    TI__Intellectual 2910 points

    Hi Yoel,

    Could you please try this IBIS model with updated LVCMOS and SDIO buffers?

    j784s4_v0p2.ibs

    Regards,
    Mark

  • 0 in reply to Mark L.
    Prodigy 30 points

    Hi Mark,
    Looks like the ripples are a bit attenuated. 

    Do you agree with my hypothesis regarding the RLC resonance? 
    If so, do you think it's a modelling issue, or real scenario we might face in lab?




  • 0 in reply to yoel biberman
    TI__Intellectual 2910 points

    Hi Yoel,

    Thank you for the plot of both the 0p1 and 0p2 model— the updated LVCMOS model looks better.

    To rule out any IBIS issues, could you please replace our RLC values with the competitor's RLC values and see if the issue still exists?

    Using HyperLynx, I ran a sim using the 3rd-party TX to the same 3rd-party RX. The results look similar to the 3rd-party TX to TI RX. We suggest working with 3rd party to understand options to improve signal integrity with their TX.

    Parameters:

    • Transmission line - 50 ohm, 0.4 ns delay
    • Stimulus - 125 MHz oscillator, 50% duty cycle
    • PVT - fast-strong
    • 3rd-party ipp_sre[2:0] - 0b110 (I believe this is the setting you were using)

    3rd-party TX to 3rd-party RX

    3rd-party TX to TI RX

    Regards,
    Mark

  • 0 in reply to Mark L.
    Prodigy 30 points

    Hi Mark,
    To clarify, in all recent plots I sent I used an ideal 50 ohm driver with 0.4ns rise time, not 3rd party Tx.
    Below are voltage plots at PVT max of 3rd party, RLC Vs. ibis to validate there is no ibis modelling issue.


    Regarding the plots you sent, not sure exactly what you used, but looks like you have almost 4.5V peak and -1V min. Are you sure of your setup? not reasonable for RGMII, as far as I know...

    To summarize, I have an ideal 50 ohm driver and I test 2 different receivers, using ibis and RLC. 
    With TI I get strong ripples at typ and max PVT (ibis and RLC) , with 3rd party I don't.
    I can see TI RLC circuit has a significant resonance at BW of rise time which explains the ripples.