• State Resolved
  • Locked Locked
  • Replies 13 replies
  • Subscribers 34 subscribers
  • Views 433 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.

DP83867IS: Regarding Compliance Testing

Ryu Yamashita
Genius 4565 points

Part Number: DP83867IS

Hi team,

Regarding the “1000 Base-T, Point A Peak Output Voltage” test, the measurement result is approximately 680mV, while the Pass Limits are “670.0 mV to 820.0 mV”.
Although it passes, the value is small and close to the lower limit.
Is the IC output intentionally targeting the lower limit of the range?
Or is the value attenuated due to PCB design or the influence of other devices?

Best Regards,
Ryu.

  • +1
    TI__Mastermind 26771 points

    Hi Yamashita-san, 

    The IC aims to target the lower limit to ensure minimal power consumption. 

    Best,
    J

  • 0 in reply to J
    Genius 4565 points

    Hi J-san,

    Thank you for your reply.
    I understand.
    One more thing—overshoot occurs during testing.
    Since I don't know the cause, would it be possible for you to review the schematic and layout?
    I'd like to discuss this via private message.

    Best Regards,
    Ryu.

  • +1 in reply to Ryu Yamashita
    TI__Mastermind 26771 points

    Hi Yamashita-san, 

    Below are my comments:
    1. The magnetics customer is using for the board does not fit our datasheet recommendation. We recommend 350uH open circuit inductance for magnetics as our PHY has been validated magnetics with such specification. 
    2. It looks like the RJ45 connector GND is not isolated on L2. Could the customer ensure the grounds are isolated?


    How bad is the overshoot?

    Best,
    J

  • 0 in reply to J
    Genius 4565 points

    Hi J-san,

    Thank you for your comments.
    I'd like to confirm one point: the mismatched inductance values are for T7-T10, correct?
    Does this discrepancy only affect the overshoot?
    Are there any other potential issues?

    Best Regards,
    Ryu.

  • 0 in reply to Ryu Yamashita
    TI__Mastermind 26771 points

    Hi Yamashita-san, 

    Ryu Yamashita said:
    I'd like to confirm one point: the mismatched inductance values are for T7-T10, correct?

    Yes, and this will cause more noise to go through the Ethernet signal affecting the signal integrity. 

    In the case of overshoot, I think this can be due to the ground isolation issue as I previously pointed out. Ground isolation is very important on Ethernet design. 

    Best,
    J

  • 0 in reply to J
    Genius 4565 points

    Hi J-san,

    It seems that overshoot is significantly affected by GND separation, but what mechanism causes GND separation to have such a large impact?
    Also, since changing the PCB layout now is difficult, is using damping resistors effective for preventing overshoot?

    Best Regards,
    Ryu.

  • +1 in reply to Ryu Yamashita
    TI__Mastermind 26771 points

    Hi Yamashita-san, 

    GND separation is important since it protects against noise coupling and ensures electrical isolation. This way, a lot of noise is filtered out and ensures a good EMI performance since EMI performance is heavily affected when grounds are not isolated. The series resistor on the MDI line may help with overshoot, but it will affect the signal integrity which may cause errors on the ethernet packets. 

    Best,
    J 

  • 0 in reply to J
    Genius 4565 points

    Hi J-san,

    When we performed the same test on the EVM, overshoot was observed.
    Are there any possible causes?
    I will send the overshoot waveform and block diagram via private message.

    Best Regards,
    Ryu.

  • 0 in reply to Ryu Yamashita
    TI__Mastermind 26771 points

    Hi Yamashita-san, 

    I received your message. I performed a similar measurement on our end. We connected the DP83867ERGZEVM to Keysight compliance board using 50cm compliance cable and probed the signal with 100 ohm differential signal from the compliance number 2 section using 1169A probe system with 2678A probe head. 


    Below are the waveforms: First one is zoomed in and the second is zoomed out. 
    m:


    As you can see, I did not observe any overshoot on the EVM. I tried to mimic the customer measurement setup as much as possible, too. 
    Could this be an issue with the customer setup?

    Please let me know their thoughts. 

    Best,
    J

  • 0 in reply to J
    Genius 4565 points

    Hi J-san,

    Thank you for verifying on the EVM.
    What could be the possible cause?
    Please tell me all possible causes.

    Best Regards,
    Ryu.

  • 0 in reply to Ryu Yamashita
    TI__Mastermind 26771 points

    Hi Yamashita-san,

    What is the board that the customer is connecting our EVM to? How is the customer generating packets on our EVM? I used the 100Base-TX compliance script to generate traffic on the EVM. What is the probe that the customer is using? Because this is happening on the EVM also and I do not see it, it must be the measurement setup issue.

    Best,

    J 

  • 0 in reply to J
    Genius 4565 points

    Hi J-san,

    The measurement waveform should be similar.
    The following register settings generate the signal.
           Reg 0x1F = 0x8000 //Reset PHY
    　　Reg 0x0 = 0x2100 // Programs DUT to 100Base-TX Mode
    　　Reg 0x10 = 0x5008 // Programs DUT to Forced MDI or MDIX mode (0x5028)
    　　Reg 0x1F = 0x4000 // Restart PHY

    The measurement environment has been sent via private message.

    Best Regards,
    Ryu.

  • +1 in reply to Ryu Yamashita
    TI__Mastermind 26771 points

    Hi Yamashita-san, 

    Thank you for the images. It looks like the measurement setup could be causing the overshoot. The customer is correctly terminating the signal to 50 ohm to ground for measurement. However, the signal is 100 ohm differential to each other past the transformer which is the customer's measurement point. Also, the customer soldered resistors onto RJ45 connector. This can cause reflection on the signal and it makes sense with the overshoot on the MLT-3 waveform the customer is seeing as the waveform has both overshoot and undershoot along with ripples afterwards. This is the sign of lower signal quality. Could the customer measure using the signal using compliance board? If not, I would suggest customer not to worry too much on the reflection/overshoot. 

    Please let me know. 

    Best,
    J