• State Resolved
  • Locked Locked
  • Replies 5 replies
  • Subscribers 21 subscribers
  • Views 388 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.

TXS0108E: Input rising/falling thresholds for one-shots, pull-up 4k/40k flip, one-shot kick-back

ZippZapp
Prodigy 30 points
Part Number: TXS0108E

Tool/software:

Hello!

I'm using the TSX0108E in an I2C application and struggle to understand exactly how the behaviour is.

For my 400kHz I2C bus bus-pull-up calculations etc it is essential to know how the TXS0108E is helping (or counteract) the I2C performance.

Please clarify:

1) The 4K pullup is in place when the signal is "high".

When an I2C-device starts to drive the line low;
==>at which voltage will the 4k pull-up be replaced by the 40k-pullup?

Knowing this, is required to figure out if the I2C-device will ever be able to drive the signal all the way down to 0.15V.

2) When the I2C-device is releasing it's active drain-pull; the bus now has a very sloppy 40k pull-up making the rise time far to long on a 60pF I2C-bus.

The rise-time will be greatly improved the moment the flip from 40k to 4k pullup occurs.

==> At which voltage is the flipping from 40k to 4k pullup occur?


3) When an I2C device starts pulling the line down; at which voltage will the pass-element become active (e.g. <10k or so)?
At this point, the burden for the driving side to pull down the line sufficiently will worsen drastically, as the pull-up(s) on the opposite side of the level-converter (internal 4k or 40k (?) in parallel with any external pullups will start to push current trough the pass-element as well, increasing the drive requirements of the I2C-driver.


4) What is the rising/falling voltage thresholds for one-shots to kick in; relieving the I2C-driver while it pulls the line down the last few hundred millivolts.(momentary removing the burden from the opposite side pull-ups)


5) When the one-shot is triggered by a falling (or slowly rising) edge on one side: Will the resulting full-swing-output on the opposite side of the level-konverter result in a new trigger of the opposite one-shot (on the I2C-driver side in this example)?

This will be very important in order to help the I2C bus to regain a "high" quickly on both sides of the translator and decissions about need for additional external pull-up values on both sides of the translator.

Looking forward to your clarifications soon,

Best regards,
ZippZapp

  • +1
    Guru 317260 points

    As far as I know, both the one-shots and the switched pull-ups are triggered when the beginning of a rising/falling edge is detected, near 30%/70% of VCC. (This threshold is not specified or guaranteed.)

    The pass transistor is active for all voltages between GND and VCCA (the lower supply voltage).

    As far as I know, the one-shots are triggered only on one side. But the connection through the switch means that the one-shot will affect both sides at least partially.

    In general, external pull-ups are not necessary with the TXS.

  • 0
    TI__Mastermind 40976 points

    Hey ZippZapp,

    Additional details to the device working of the TXS-Translators can be found in this app note: www.ti.com/.../scea044.pdf

    Regards,

    Jack

  • 0 in reply to Clemens Ladisch
    Prodigy 30 points

    When used to extend a 1.8V I2C-bus with some 1.8V I2C-devices into a 3.3V I2C-segment housing some other 3.3V I2C-deivces: I am still unsure how the existence of the TSX0108E on the 1.8V bus impacts the performance on the local 1.8V I2C-bus-segment and how to calculate what would be a reasonable pull-up-value on this bus.

  • 0 in reply to Jack Guan
    Prodigy 30 points

    Thanks, I have read it, and provides some general understanding og how the device works, but also lack all the details asked for and needed to understand its impact on an I2C-bus. I could not find the explanation of where the Vbias is comming from. I would guess it is in reality the VCCA? And with a Gate Theshold of about 1V and a 1.8V VccA, I would expect that voltages driven below ~0.8V would turn on the N-Fet pass element, exposing the load from the B-side to the A-side driver (or opposite) before reaching the 30% level at ~0.54V (on A-side, and 0,99V on the 3.3V side if opposite direction?)

  • 0 in reply to ZippZapp
    Guru 317260 points

    The TXS was designed for high-speed signals, which I²C are not. Just using it without external pull-ups should be more than enough.

    The switched pull-ups and the one-shots make everything very nonlinear and hard to calculate. I'd recommend checking the waveforms on the actual board.