Because of the holidays, TI E2E™ design support forum responses will be delayed from Dec. 25 through Jan. 2. Thank you for your patience.

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.

TXS0108E: Short pulse in input cause output wrong

Part Number: TXS0108E
Other Parts Discussed in Thread: TXB0302, TXB0304, TXB0106, TXB0108, , LSF0108

Hi TI team,

We need to translate between 1V8<->3V3 for QSPI application. But we have the issue is that when input has a short pulse, the output can not go high back. 

It look oneshot circuit can not work normally with short pulse like this. So when input go high back but output go high by pull up resistor not by oneshot

Is it correct? 

If yes, Any constraint for input to avoid this issue?

Note: Above 2 pictures are not in the same scale

Thanks,

  • Hey Duong,

    Thanks for the sending the waveforms. Generally, we do not recommend TXS family devices for QSPI applications. A better solution here is the TXB0106 if it meets your data rate or the TXB0304 + TXB0302 combination. What is the data rate and could you also share the full schematic so I can further debug the issue? 

    Its also important to keep in mind the loading conditions when operating with this device. See the below for more information about the one-shot timing out if the load is too heavy:

    Regards,

    Jack 

  • Thank Jack for your response

    Below is the diagram. In the waveform, input is measured at BMC, the output is measured at Flash

    In the old design, we use TXB0108 but we have a problem with ISSI flash because IO2 and IO3 must be pulled high at the Flash side.

    We also use a weak pull-up 50kOhm but it doesn't work. So in the new design, we need to change to TXS0108

    I know about capacitive load but it work with TXB0108 (except using ISSI Flash), so I don't think it cause issue on TXS0108

    Thanks,

  • Hey Duong,

    TXS0108E differs slightly in the architecture (it has internal pull up resistors populated internally already) when compared to the TXB0108. These pull up resistors range from 4k (when driving high) and 40k (when driving low). Could you remove the pullup resistor on the output and see if the issue is gone? 

    For more information, here is an app note on the effects of pull up resistors used with TXB and TXS devices.

    Regards,

    Jack

  • Hi Jack,

    I've read the effects of pull-up resistor if using TXS but I see it only effect VOL. Not the same issue with us

    I share more visual waveform as below

    C4 is input and C2 is output. I guess the pulse length is too short so one shot circuit can detect the falling edge and can not detect the rising edge. So when input go high, output go high by pull up resistor not by one shot circuit. I cause the slow rising edge at output. Is this correct?

    Do you have any other ideas?

    Thanks,

  • Hey Duong, 

    Yes, your understanding is correct- if the pulse duration is too short, the rising edge one shot circuitry can still be timed out from the previous I/O high. Note that there are two independent one shot systems to help with the rising edge and falling edge. Can you confirm that your pulse duration is meeting this requirement?

    As for the 4.7k pull up resistor, this is recommended to be removed as it impacts the VOL levels. However, if you are satisfied with the output, then it is ok to leave as is. 

    Regards,

    Jack

  • Thank Jack for your response

    Now the pulse width is around 6ns. Maybe it causes output to look weird

    But I see the same specs "pulse duration" in TXB0108 datasheet. Why does the behavior differ when using TXB and TXS?

    Thanks,

  • The TXB and TXS have similar edge accelerators, but their basic architecture is different (TXS uses passive switches; TXB uses weak buffers).

    In theory, it might be possible to use the LSF0108, a translator without edge accelerators. However, to reach high speeds, it needs extremly strong pull-resistors, which increases the VOL of all connected devices to a level that might be too high.

  • Yes, I totally agree with you

    LSF can not meet high frequency specs in my application

  • Hey Duong,

    As Clemens' mentioned, LSF is a possible solution as long as as pull ups/ small external capacitances are taken into consideration . Provided below is the estimated max data rate that the LSF0108 can handle. See if it meets your system requirements.

    Regards,

    Jack