• State Resolved
  • Locked Locked
  • Replies 12 replies
  • Answers 2 answers
  • Subscribers 33 subscribers
  • Views 424 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.

TCA9517: TCA9517 fails where TCA9617 works

Bob Budd
Prodigy 40 points
Part Number: TCA9517
Other Parts Discussed in Thread: TXS0102, , TCA9617B,

Tool/software:

I have a simple application of TCA9x17 where TCA9517 is not working but TCA9617 is.  The A side is an iMX8 processor on a SoM and the B side is an HDMI I2C port (to determine the unique parameters of the display, etc.).  With the display plugged in SCLB is stuck at a low voltage and SDAB is not switching.  With the display unplugged the SCLB and SDAB signals are switching as expected.  I am replacing TXS0102 that I also had trouble with.  I started with a reference design specifying 1k pullups on both sides but it didn't work until I removed the pullups altogether, and the circuit worked on the reference design with 1k pullups installed.  For TXS0102 the pullups are unnecessary as noted in the datasheet and, even with 1k pullups the waveforms looked fine aside from the absence of an ACK.  Presently the pullups are 1k due to the reference design ultimately will be adjusted per 3mA current limit.

  • 0
    TI__Mastermind 47278 points

    Hi Bob,

    To determine why the TCA9617B is working and the TCA9517 fails, can we see a schematic or a scope capture of the SDA/SCL channels on both sides of the buffer? 

    Is the TCA9517 / TCA9617B being hot-plugged?

    Regards,

    Tyler

  • 0 in reply to Tyler Townsend
    Prodigy 40 points

    The schematic is as shown with wires attached to this PCBA allowing me to plug in one of several simple prototype PCBAs with each translator device.

    The signals on the PCBA side are unremarkable and quite square, i.e. rise time is very small as expected with the 1k pullups.  The cable side does not switch.

    The parts are not being hot plugged, i.e. the PCBA can be swapped as noted above but we are only swapping when power is off.

  • 0 in reply to Bob Budd
    TI__Mastermind 47278 points

    Hi Bob,

    I will respond by noon CST tomorrow. 

    Regards,

    Tyler

  • 0 in reply to Tyler Townsend
    Prodigy 40 points

    I took a look at voltage thresholds across the parts I am testing.

    • TCA9617 limits VCCA to 0.8VCCB wherein I cannot use the part with VCCA=3.3V and VCCB=5V.  Further Vil = 0.3VCCA for A side and Vil=0.4V for B side.  Driver current is 30mA.    Please confirm
    • TCA9517 Vil = 0.3VCCA for A side and Vil=0.4V for B side.  Driver current is 6mA.    Please confirm
    • TXS0102 Vil = 0.15V - VERY LOW - for A side and B side.  Driver current is not spec'd though with pullup accelerator a small pullup is unnecessary.    Please confirm
  • 0 in reply to Bob Budd
    TI__Mastermind 47278 points

    Hi Bob,

    Bob Budd said:
    TCA9617 limits VCCA to 0.8VCCB wherein I cannot use the part with VCCA=3.3V and VCCB=5V.  Further Vil = 0.3VCCA for A side and Vil=0.4V for B side.  Driver current is 30mA.    Please confirm

    The recommended operating conditions limits VCCA to be within the range of 0.8V < VCCA <= VCCB. 

    VCCA = 3.3V and VCCB = 5 V is a valid case. 

    VILA = 0.3 x VCCA (A-side)

    VILB = 0.4 V  (B-side) 

    Driver current IOLA and IOLB is max of 30 mA. 

    Bob Budd said:
    TCA9517 Vil = 0.3VCCA for A side and Vil=0.4V for B side.  Driver current is 6mA.    Please confirm

    Yes this is correct. 

    TCA9517A (A-version) changes the VILB (b-side) to 0.45V typical. 

    Bob Budd said:
    TXS0102 Vil = 0.15V - VERY LOW - for A side and B side.  Driver current is not spec'd though with pullup accelerator a small pullup is unnecessary.    Please confirm

    TXS0102 is different architecture than the TCA9517 and TCA9617B. TXS0102 is not a true buffer. It is a passFET between side 1 and side 2 while a buffer exists between A-side and B-side of the 9517 and 9617B buffers. 

    TXS0102 is a passive device. The "re-drive" aspect comes from the rise-time accelerators (one-shot accelerators) present within the device.  

    VIL = 0.15V is the input low voltage that the device was validated with in the datasheet. In practice, a VIL could vary depending on the low-side voltage supply. 

    The gate voltage of the passFET is set to approximately one threshold voltage (VT) above the low-side voltage. Thus is VCCB = 5V and VCCA = 3.3V, this means that the gate voltage is roughly 3.3V + 0.7V = 4.0V. A-side of the TXS0102 will begin to track B-side once either side falls a VT below the gate voltage. 

    See the architecture of the TXS0102

    For reference, see the architecture of the 9617B / 9517 buffer: 

    Regards,

    Tyler

  • 0 in reply to Tyler Townsend
    Prodigy 40 points

    Tx for the clarification and especially the correction.

  • 0 in reply to Bob Budd
    TI__Mastermind 47278 points

    Hi Bob,

    Absolutely. If you have any more questions/comments/concerns, please feel free to respond back to this thread. 

    Regards,

    Tyler

  • 0 in reply to Tyler Townsend
    Prodigy 40 points

    TCA9517 vs TCA9617 question has not been answered, i.e. there was no B signal when there was A signal.  A pullups are now 10k and signal looks good with Vlo~= 0V.  As yet none of the parts are getting an ACK.  I am now looking at reducing pullups from 1k on B, I'm thinking Vlo is not low enough.

  • 0 in reply to Bob Budd
    TI__Mastermind 47278 points

    Bob,

    The VIL on B-side needs to be < 0.4V for both devices in order for a LOW to be properly transmitted through the device. Reducing pull-up strength would decrease VOL and make it easier to reach the 0.4V VIL requirement on B-side. 1kohm is quite strong on B-side especially if the I2C target/controller device on B-side has a weak driver. 

    Regards,

    Tyler

  • 0 in reply to Tyler Townsend
    Prodigy 40 points

    Tx, I have begun increasing the pullups, at 2k Vlo was still ~0.4V.  Onward...

  • 0 in reply to Bob Budd
    TI__Mastermind 47278 points

    Bob,

    VLO on B-side input needs to be < 0.4V in order to satisfy VILC for the TCA9517. Increasing the PU resistors from 1k to 2k should affect the VOL, unless there is some type of passFET or I2C switch device also connected to B-side which could affect the equivalent PU resistance. 

    Regards,

    Tyler

  • +1 in reply to Tyler Townsend
    Prodigy 40 points

    We found that all candidates are working with 10k pullups, even 4k was preventing many from working.  In the HDMI case the issue does not appear to be Vol but instead how large a pullup is needed to get a usable ACK from the HDMI.  There were cases between 2k and 10k where ACK was pulled to ~1V or something above Vol and the transactions were failing but at 10k ACK pulled to ~0V - well below Vol that still was ~0.4V - and transactions were successful.