• State
  • Locked Locked
  • Replies 7 replies
  • Subscribers 20 subscribers
  • Views 1718 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.

TXB0108: Level translator damages XIlinx Zynq Ultrascale+ FPGA bank

Paulius Klimenko
Prodigy 50 points
Part Number: TXB0108

Hi,

We are using a TXB0108 translator to translate 1V2 xczu3eg FPGA bank push-pull logic to 3V3 push-pull logic on MT9V032 sensor. If we use the TXB, it damages the FPGA irrecoverably on first turn on. During turn on the FPGA is High-Z.

If a TXS0108 translator is used, this behavior is not observed.

The board in question is Ultra96-V2

What could be causing the issue?

  • 0
    Guru 316760 points

    Does this always happen, with every I/O pin? Or is the entire FPGA broken?

    I'd like to see oscilloscope traces of what happens during turn on.

    The most likely reason would be overvoltage, but the TXB should never output a voltage above 1.2 V. Are the FPGA bank and the TXB powered by the same supply?

  • 0 in reply to Clemens Ladisch
    Prodigy 50 points

    We are using an Ultra 96 V2 devboard and the pins themselves seem to function normally after the board/fpga is damaged. However, we are using internal deserialiser and impedance matching (DCI) resources (for LVDS signal) which seem to cease functioning completely after using TXB. These resources are not even connected to the TXB's I/O pins.

    As for the traces, the 1.2V side and 3.3V just ramps up steadily on the VccA and VccB pins, see pdf attached. If I/O pins are tested, the same result is observed. The bank and TXB are not powered by the same supply. The FPGA bank I/Os have their own power supply and the TXB is powered by an LDO. 

    WaveRunner8404 Hardcopy.pdf

  • 0 in reply to Paulius Klimenko
    Guru 316760 points

    When powering up, the TXB can output a 1.2 V signal at all A pins. Is it possible that this happens before the FPGA bank has powered up, and that the FPGA does not allow voltages higher than its own supply?

  • 0 in reply to Clemens Ladisch
    Prodigy 50 points

    The TXB 1.2V is provided by an LDO, which is powered by a PMIC with a 5V DC/DC on the Ultra96 board. Same PMIC also powers the 1V2 FPGA bank with 1V2 DC/DC.  Since the behavior you described happens with TXS0108 as well, I doubt it would be causing the problem with the damaged FPGA as the TXS works perfectly. 

  • 0 in reply to Paulius Klimenko
    Guru 316760 points

    I would expect a linear regulator to power up faster than a switching supply.

    The TXB and TXS have different architectures, so it would be plausible that the TXB drives the I/O high earlier.

    Can you keep the EN pin of the TXB low for a certain time while powering up?

  • 0 in reply to Clemens Ladisch
    Prodigy 50 points

    You're probably right about the start times. However, the part that is damaged inside the FPGA is not even connected to the TXB pins. We can try keeping the pin low, but we are burning quite expensive boards and buying another one just for a test is not feasible for us. 

    We will probably control the EN of the TXS shifter in the future as it did not damage anything even when EN was not controlled. 

  • 0 in reply to Paulius Klimenko
    Guru 316760 points

    I suspect that the high voltage from the TXB goes through the ESD diodes into the FPGA's 1.2 V power supply, and thus affects anything connected to that supply.