• State TI Thinks Resolved
  • Locked Locked
  • Replies 11 replies
  • Answers 4 answers
  • Subscribers 16 subscribers
  • Views 735 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

ISO1541: Output voltage level too high at active low with ISO1541

Feng bin Su
Expert 7540 points
Part Number: ISO1541
Other Parts Discussed in Thread: TCA9517A, TCA9800, ISO1640, TXS0102, TCA9406, TCA9517

Hi team,

My customer use our isolation IC  ISO1541 at POE application to have such problems:

1. SCL is a one-way signal, and the test high and low levels meet our requirements;

2. SDA is a two-way signal, and the SDA signal from the CPU to the PSE IC is all It can meet our requirements;

3. The SDA signal from PSE IC to ISO1541 chip is good and can meet our high and low level requirements, but from ISO1541 to Level shift IC (in the direction shown by the red dotted line) this path reaches low level 800mV, as shown in the figure, and the low level required by level shift is only 150mV, so it does not meet our requirements, which indirectly causes the low level of the level shift IC to the CPU to be very high.


Have you ever encountered a situation where customers use this chip to feedback a relatively high low level?

Would you have any suggestion to reduce the low level here to about 150mV to meet the low level requirement of the level shift IC? Thank you.

  • 0
    TI__Mastermind 49445 points

    Hi Feng,

    800mV of LOW output voltage on SDA1 of ISO1541 is expect according to the datasheet and is designed so to be able to achieve bidirectional function, please refer the below FAQ for more details. To resolve this issue, please use a different level translator that is compatible with ISO1541 logic levels. Some of the other level translators include TCA9517A and TCA9800.

    ISO1541 is an older generation I2C device, we now have ISO164x family that has better overall performance including slightly lower VOL spec (710mV). We recommend customer to use ISO1640. Thanks.

    Regards,
    Koteshwar Rao

  • 0
    Guru 243120 points

    Which level shift IC are you using? The TXS0102‌/‌TCA9406 does not actually have a 150 mV requirement, and does not require external pull-up resistors, which would make the low level unnecessarily high.

  • 0 in reply to Clemens Ladisch
    TI__Expert 7540 points

    Hi Clemens,

    Customer originally used TXS0102. Would you mean the VIL of TXS0102 will be higher than 0.15v? Can TXS0102 be used with ISO1541? Thanks.

  • 0 in reply to Feng bin Su
    Guru 243120 points

    The TXS is a passive switch. Its recommended operating conditions say:

    (2) The maximum VIL value is provided to ensure that a valid VOL is maintained. The VOL value is VIL plus the voltage drop across the pass-gate transistor.

    The TXS does not really have a VIL. You need to keep VIL low only if you want to achieve a certain VOL.

    The voltage drop across the pass-gate transistor depends on the current through the switch, which depends on the pull-up resistors. If you have external pull-up resistors (as shown in the schematic above), the current and the voltage drop are unnecessarily high.

    The TXS VOL is always larger than the VOL of the driving device. If the 800 mV (plus the switch voltage drop) of the ISO1541 is too large for the CPU, then you must use a buffer like the TCA9800.

  • 0 in reply to Clemens Ladisch
    TI__Genius 17100 points

    Hi Feng, Clemens,

    What is the V_IL threshold for the CPU SDA line? If using TXS0102 was not enough for the CPU to register low signal levels, we do recommend using a level-translator with a lower V_OL spec, which in this case could be TCA9800.

    The max V_OL spec for TXS0102 is 0.4V while for TCA9800 it is 0.26V on side B and 0.13V on side A. Please let us know if our customer is willing to try this I2C level translator, and if it is sufficiently low to meet the CPU's V_IL threshold limits.


    Respectfully,
    Manuel Chavez

  • 0 in reply to Manuel Chavez
    TI__Expert 7540 points

    The V_IL threshold for the CPU SDA line is around 0.4 v. Customer has tried to use  TCA9517  to replace TXS0102 in this design but it seems doesn't work well.

    In the picture below, the blue signal on the ISO1541 side has a small spike, indicating that there is an ack, but the green signal shows that there is no small spike on the CPU side after entering TCA9517, so that the CPU cannot receive the ack, and the PSE chip reads failed.
    At the same time, there is a second problem that the SCL waveform is raised to the 780mv at low level, and the original TXS0102 does not have this problem.

         WithTXS0102

  • 0 in reply to Feng bin Su
    Guru 243120 points

    The two images are identical.

    The TCA9517 uses an voltage offset; the TCA9800 does not.

  • 0 in reply to Clemens Ladisch
    TI__Expert 7540 points

    Hi Clemens,

    Sorry. Reattached the waveform of TCA9517. 

    It seems TCA9517 doesn't detect the low level of the output of ISO1541.

  • 0 in reply to Feng bin Su
    TI__Genius 17100 points

    Thanks Clemens.

    Hi Feng,

    TCA9517 has a minimum low-level output voltage, V_OL, spec of 0.45V with a maximum of 0.6V for this spec on its "B" side. These levels are likely not low enough for the CPU SDA line to register as LOW signals.

    We recommend using a level translator with lower V_OL specs, like TCA9800. Please encourage our customer to try this device instead of TCA9157, and let us know if it is sufficient to meet the CPU's V_IL threshold limits.


    Thank you,
    Manuel Chavez

  • 0 in reply to Feng bin Su
    Guru 243120 points

    Two I/Os with an offset voltage cannot be connected together. This happens here with the TCA9517 and the ISO1541.

    It is not possible to swap the A/B sides of the TCA9517 because VCCB would be too low. This is why I've recommended the TCA9800.

  • 0 in reply to Clemens Ladisch
    TI__Genius 17100 points

    Thanks Clemens!

    Agreed, we also recommend using TCA9800 in this circuit.


    Respectfully,
    Manuel Chavez