• State TI Thinks Resolved
  • Locked Locked
  • Replies 6 replies
  • Answers 2 answers
  • Subscribers 20 subscribers
  • Views 1490 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.

TXB0104: Significant apparent crosstalk on output

Steve Kleiser
Prodigy 50 points
Part Number: TXB0104
Other Parts Discussed in Thread: SN74LV4T125

  When using TXB0104 to level shift 2.5MHz SPI from 3.3V to 5V, the 5V (See attached) output appears to have significant clock-to-data crosstalk, effectively corrupting the output, w/ only an unterminated 21" "zip" cable as a load. (Also see attached.) No improvement with the real load (SPI-controlled LEDs) connected.

  The 3.3V inputs are from a Microchip DSPIC33EP128MC506, and are clean.

  This was originally posted to tech support as case CS0034489, & they referred me here.

  • 0
    TI__Mastermind 45725 points

    Hey Steve,

    I'll be happy to help with this issue. Could you provide a scope shot of both signals (SDO and CLK). Also are you able to remove the cable all the way to completely isolate the device from anything downstream?

  • 0
    Guru 316660 points

    This looks as if noise from the clock line affects the data line, and that the resulting voltage spikes are large enough to trigger the TXB's edge accelerators. I also suspect that the capacitance of the cable (and its connectors) is larger than the allowed 70 pF, which can lead to ringing.

    It might be a better idea to use a unidirectional translator with low-impedance outputs, e.g., SN74LV4T125.

  • 0 in reply to Dylan Hubbard
    Prodigy 50 points

    Dylan,

      A scope capture of both clock & data should appear below, followed by another showing cleaner signals when the cable is disconnected. There's an intermediate connection in the cable at 7.5" from the PCB edge, & when that's disconnected the signal improves somewhat. It's a rough lab-only cable (also shown below).

  • 0 in reply to Clemens Ladisch
    Prodigy 50 points

    Clemens,

      Thanks for the reply. The equipment we have on hand may not be capable of measuring down to 70pF, but I can investigate the SN74LV4T125 for a future board spin.

      Steve K.

  • 0 in reply to Steve Kleiser
    Prodigy 50 points

    Clemens & Dtlan,

      I found an LCR meter capable of measuring low capacitance, & it shows the capacitance of the 21" cable at about 20pF at 10KHz (the highest frequency of which the meter is capable).

      Steve K.

  • 0 in reply to Steve Kleiser
    Guru 316660 points

    I estimate the time constant of those round curves to be about 500 ns; with the TXB's output impedance of 4 kΩ, the capacitance would be about 125 pF.

    In any case, the the problem is not the capacitance, but the crosstalk. On a PCB, I would recommend a ground trace between the two signals. A cable is more difficult; try two twisted pairs, clock+ground and data+power.