• State TI Thinks Resolved
  • Locked Locked
  • Replies 10 replies
  • Answers 2 answers
  • Subscribers 36 subscribers
  • Views 799 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.

TCA39306-Q1: Can this device support 8M MDIO?

Frank QIN
Genius 9810 points

Part Number: TCA39306-Q1
Other Parts Discussed in Thread: TCA39306, TXS0102-Q1, TXS0102

Hello Team,

I only found there is a description about TCA39306 can support 12.5MHz on i3c. Can I confirm if it can support such high frequency on MDIO?

If not, do we have any other solution can support MDIO over 8MHz? 

BR

Frank

  • 0
    TI__Guru* 79221 points

    I believe MDIO is push pull architecture correct? It will likely depend on the Vcc levels you are using. A wide seperation between the Vcc levels means you may not hit certain rise times since the high level on the lower Vcc side of TCA39306 will likely clip at Vcc1 when side 2 is driving high. If the clipping occurs above the ViH level, I suspect it should. You may need to use a device with a rise time accelerator instead of the TCA39306. 
    What voltages are you trying to translate from and to?

    -Bobby

  • 0 in reply to BOBBY
    TI__Genius 9810 points

    Hello Bobby,

    Thanks for your reply. We applied some samples did a test.

    The schematic is as followed. It is 1.8V MDC and 3.3V MDIO. Now the test condition is 8.3MHz, TP1405(blue waveform) is normal 1.8V voltage. After the chip tested as TP1406, the waveform clipping as only 2.48V(expected 3.3V). 

    We suppose the issue is the pull up capacity of the resistance. But we tried reduce the 1.5k pull up resistor, the LOW voltage turns to over 300mV.

    We changed the R1409 and T1410 to 0R to shrug off the effect of the resistance in series.  

    BR

    Frank

  • 0 in reply to Frank QIN
    Guru 321920 points

    Such slow rising edges are indeed caused by a high capacitance.

    Do you have long traces, connectors, or cables?

  • 0 in reply to Clemens Ladisch
    TI__Genius 9810 points

    Hello Clemens,

    Yes, the traces is indeed long. We are trying to do some test with cutting the trace.

    Do you have any suggestion on the trace/cable length? Or the suggesting range of the capacitance? We only have limited chance to do the test on the PCB so some reference of the length or range would be helpful.

    BR

    Frank

  • 0 in reply to Frank QIN
    Guru 321920 points

    This device has no hard limit. The capacitance and the pull-up resistors form a low-pass filter; a higher capacitance requires stronger pull-ups (and thus more drive strength by the devices on both sides).

    Estimating capacitance is hard. You could assume about 5 pF per device (so ×3), and traces 1−3 pF per inch.

    The TXS0102-Q1 works similarly, but with integrated pull-up resistors and edge accelerators. The edge accelerators will result in faster rising edges, but if the capacitance is too high, you still need additional pull-ups.

  • 0 in reply to Frank QIN
    TI__Guru* 79221 points

    Hi Frank,

    The scopeshot shown is has CH1 going up to 2.5V which isn't the Vcc on that side (3.3V). It seems like you're going to need a device with a rise time accelerator to help instead of using external pull up resistors. 

    I like Clemen's suggestion with the TXS0102-Q1 device. This device should provide a much cleaner rising edge on your higher logic voltage.

    -Bobby

  • 0 in reply to BOBBY
    TI__Genius 9810 points

    Hello Bobby and Clemens,

    Can I confirm if TXS0102-Q1 can drive MDIO signal?

    We changed the MDC output trace to 8mm, and jump wire for the pull up. The results shows it getting better, but still have clipping. The yellow is MDC output, blue is 1.8V input. It can only reach 3.06V. So it looks like we can only choose another device?

    Also, I downloaded the IBIS module of TCA39306, but the 6pin is input pin. Can we provide a IBIS that all the SCL and SDA relate pins are IO port?

    BR

    Frank

  • 0 in reply to Frank QIN
    Guru 321920 points

    There is no IBIS model for the TCA39306's I/O pins because they are not buffered (the TCA39306 is a passive switch).

    The TXS is much faster than the TCA. I do not know if it is fast enough.

  • 0 in reply to Frank QIN
    TI__Guru* 79221 points

    Frank,

    Since TXS-0102 can redrive the high level signal , I suspect it would be able to help in your application (TCA39306 does not redrive signals as Clemens has pointed out).

    I'm reassigning the thread from myself (transceivers) to our subject matter expert that supports the TXS0102-Q1 to provide comment. 

    -Bobby

  • 0 in reply to BOBBY
    TI__Guru 52131 points

    Hey Frank,

    MDC is uni-directional and push-pull. MDIO is bi-directional and open-drain. The TXS0102 can support both of these interfaces, though only up to 21Mbps (10.5MHz) with good signal integrity. You can reach 12.5MHz if you keep your capacitive loads relatively light.

    Best,

    Jack