• State TI Thinks Resolved
  • Locked Locked
  • Replies 16 replies
  • Answers 4 answers
  • Subscribers 22 subscribers
  • Views 846 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.

LSF0108: Output Bn unstable

Hailiang Zhang
Genius 11200 points
Part Number: LSF0108
Other Parts Discussed in Thread: TXU0104, TXS0104E, TXB0108, TXU0304, SN74AXC8T245

Hi Team，

In the customer's switch project, the 1.2V MDC and MDIO signals of the Switch chip are converted to 3.3V through the LSF0108 device, and then connected to the FPGA terminal. After disconnecting resistor R1629, VEN=2V, 1.2V at terminal A1 of the measuring chip is normal (blue signal), but the MDC voltage at terminal B1 fluctuates (yellow signal).

Please help to review the schematic diagram and analyze this issue.

Regards,
Hailiang

  • 0
    TI__Guru 52696 points

    Hi Hailang,

    Please attach the schematics to help clarify, thanks

    Best Regards,

    Michael.

  • 0 in reply to Michael Ikwuyum
    TI__Genius 11200 points

    Hi Michael，

    Pls see the schematic. Thanks

    Regards,
    Hailiang

  • 0 in reply to Hailiang Zhang
    Guru 317480 points

    Are there pull-down resistors anywhere? The LSF works only with (and requires) pull-up resistors.

    The yellow trace has rising edges that are much too slow. The speed of the rising edges is limited by the low-pass filter formed by the trace/pin capacitance and the pull-up resistor. You probably need pull-ups much smaller than 4.7 kΩ, and you should try to reduce the capacitance as much as possible.

    Consider using a unidirectional translator (e.g., TXU0104) for the MDC signals and a translator with edge accelerators (e.g., TXS0104E) for the MDIO signals.

  • 0 in reply to Clemens Ladisch
    TI__Genius 11200 points

    Hi Clemens,

    There is no pull down resistor.

    Do you think a large poll up resistor may cause this issue? The engineer is trying a smaller one, and will up date later.

    It seems LSF0108PWR is ok to use in this application. Is there any other suggeations?

    Regards,
    Hailiang

  • 0 in reply to Hailiang Zhang
    TI__Guru 52696 points

    Hi Hailang,

    Any feedback on the pull up sizing? Also if I understand correctly, the issue only occurs when R1629 is disconnected?

    While disconnected, how is the EN configured? Please see Using the EN Pin with the LSF family to further help clarify, thanks.

    Best Regards,

    Michael.

  • 0 in reply to Michael Ikwuyum
    Prodigy 10 points

    Hi Ikwuyum

    EN configure is as shown below.

    Change the pull-up resistor to 2.2kohm, the MDC signal voltage will increase to 1.85V.
    Change the pull-up resistor to 1kohm, the MDC signal voltage will increase to 2.5V.
    But this still does not reach the 3.3V level, and the waveform is like a triangle wave.
    The waveform of 2.2komh pull up resistor is as follow:
    The waveform of 1komh pull up resistor is as follow:
  • 0 in reply to Kane Chen
    TI__Guru 52696 points

    Hi Kane,

    Thanks for clarifying.

    In summary, it seems the pull-up used is not providing enough drive strength for your lumped capacitive loading with the data rate being operated with.

    If you know the data rate speed and the capacitive loading, the appropriate pull-up can be estimated similar to the below. If not, I would recommend keep lowering the pull-up <1K, as well as the data rate if feasible; to help further estimate, thanks. 

     

    Best Regards,

    Michael.

  • 0 in reply to Michael Ikwuyum
    TI__Genius 11200 points

    Hi Michael，

    Sufficiently small pull-up resistors have been tried, but this does not fix the problem. Any other suggestions? Additionally, the customer tried XX. R2019 is changed to 0 ohm, and the following waveform is obtained. Yellow is the waveform after passing TXB0108PWR, and blue is the input waveform. It can be seen that the back-end waveform of TXB0108PWR cannot reach 33M. Please help analyze this case?

    Regards,
    Hailiang

  • 0 in reply to Hailiang Zhang
    TI__Guru 52696 points

    Hi Hailang,

    Note that the feedback provided was for LSF and not TXB. For TXB, pull-ups <50Kohms are not recommended.

    Hailiang Zhang said:
    It can be seen that the back-end waveform of TXB0108PWR cannot reach 33M. Please help analyze this case?

    Note that the TXB0108 datasheet is spec'd for 10 MHz as shown below.

    Please help clarify if your issue is with LSF or TXB as the schematic shows LSF. I also do not see any R2019. Are these two separate issues?

    Best Regards,

    Michael.

  • 0 in reply to Michael Ikwuyum
    TI__Genius 11200 points

    Hi Michael,

    It's the same issue. The customer initially used LSF0108, but encountered the abnormal waveform we discussed at the beginning. Recently the engineer tried to use the TXB0108, but there is the problem I reported yesterday. The customer uses 1.8V, so the Data rate seems to be able to reach 50M

    Regarding using the LSF0108, reducing the pull-up resistors did not solve the problem. Any other suggestions?

    Sorry, please refer to this schematic, you can find R2019 here. The pull-up resistor on the B side is not placed.

    Regards,
    Hailiang

  • 0 in reply to Hailiang Zhang
    TI__Guru 52696 points

    Hi Hailiang,

    Hailiang Zhang said:
    It's the same issue. The customer initially used LSF0108, but encountered the abnormal waveform we discussed at the beginning.
    Hailiang Zhang said:
    Regarding using the LSF0108, reducing the pull-up resistors did not solve the problem. Any other suggestions?

    For the LSF, did you lower the pull-ups estimating with the lumped capacitive load and targeted frequency similar to the below? Did it work with lower frequency?

    Michael Ikwuyum said:

    If you know the data rate speed and the capacitive loading, the appropriate pull-up can be estimated similar to the below. If not, I would recommend keep lowering the pull-up <1K, as well as the data rate if feasible; to help further estimate, thanks. 

     

    For the TXB, pull-ups and pull-downs are recommended for >50Kohms and lumped capacitive loading recommended for <70pF. Is this the case?

    Hailiang Zhang said:
    The customer uses 1.8V, so the Data rate seems to be able to reach 50M

    I am guessing this is now for 1.8V to 3.3V and the device is specified for 60 Mbps or 30 MHz.

    Hailiang Zhang said:
    Sorry, please refer to this schematic, you can find R2019 here. The pull-up resistor on the B side is not placed.

    Please note that the schematic seems to not be attached, thanks.

    Best Regards,

    Michael.

  • 0 in reply to Michael Ikwuyum
    TI__Genius 11200 points

    Hi Michael,

    For the LSF0108, the engineer lowered the pull-up resistors, but that didn't work. So try TXB0108.
    Sorry for missing the schematic. please check.

    Regards，
    Hailiang

  • 0 in reply to Hailiang Zhang
    TI__Genius 11200 points

    Hi Michael，

    Michael Ikwuyum said:
    For the TXB, pull-ups and pull-downs are recommended for >50Kohms and lumped capacitive loading recommended for <70pF. Is this the case?

    It works when setting the frequency to 800K HZ, but normally, the SPI frequency should be 33Mhz. And lower the pull-up is helpless for 33MHZ.

    Michael Ikwuyum said:
    For the TXB, pull-ups and pull-downs are recommended for >50Kohms and lumped capacitive loading recommended for <70pF. Is this the case?

    Yes, we have reworked the pull up resistor to 50Kohm, but the wave is strange provided by the previous email. And the lumped capacitive loading <70pF.

    Do you have any suggestions？

    Regards，
    Hailiang

  • 0 in reply to Hailiang Zhang
    TI__Guru 52696 points

    Hi Hailang,

    The LSF requires the 200K bias resistor for the device to operate. Please use as recommended and try again.

    Could you help confirm where 33 MHz is recommended for 1.8V to 3.3V with the TXB0108 device? The TXB device is not recommended with lower pull-ups. Lowering lumped capacitance (even lower) - i.e minimal parasitic, minimal trace length / cables, no connectors etc, is what will help in this case, especially since above the datasheet's max recommendation, thanks.

    Best Regards,

    Michael.

  • 0 in reply to Michael Ikwuyum
    TI__Genius 11200 points

    Hi Michael,

    The customer calculated the lumped capacitive load of the BIOS SPI signal of the Artemis project, and the table is as follows. Which IC do you recommend customers to use? Thanks.

    Regards,
    Hailiang

  • 0 in reply to Hailiang Zhang
    TI__Guru 52696 points

    Hi Hailang,

    The TXU (TXU0304) or the AXC (SN74AXC8T245) family, thanks.

    Best Regards,

    Michael.