Because of the Thanksgiving holiday in the U.S., TI E2E™ design support forum responses may be delayed the week of November 23.

TCA9517: VCC configuration for TCA9517DGKR

Part Number: TCA9517

Hello,

 

I am attching the representative schematic in which both VCCA and VCCB are connected to 3.3V, the SCLA/SDAA is connected to 3.3V through 4.7K. The SCLB/SDAB is connected to 5V through 2K. 

The VCCB was supposed to be connected to 5V but it got left out and is connected to 3.3V. Everything seems to function as intended. Is there any downside. We dont want to respin the board if it not a big concern.

Thank you.

 

20261023_133300.jpg

 

  • Hi Gaurav,

    I am looking into this and will respond by EOD 10/26.

    Regards,

    Jason

  • Hi Gaurav,

    The configuration should not damage the device, but I would like to investigate this further. Do you have any waveforms of scla, sclb, sdaa, and sdab? Any information about the target/MCU connected?

    Regards,
    Jason

  • this is a failed case in which there is a glitch on the clock pulse which messes the i2c read.

    D0/A0 --  UHP_SDA, D1/A1 -- UHP_SCL, D2/A2 -- ESP32_SDA, D3/A3 -- ESP32_SCL. Dx/Ax are the Digital and Analog channels respectively.

    the next transaction fails as well....teh clock gets pulled low. The transaction after the below is OK.


    example of a good read....my i2c clock is set to 50Khz.

    these plots are with VCCB set to 3.3V and VCCB set to 5V. With VCCB set to 3.3V, I don't see a single i2c read failure.

    Thanks,

  • Hi Gaurav, 

    Is this the complete block diagram, or are there other devices connected to the i2c bus on A-side and on B-side? 

    The buffer seems to be releasing to VCCB, the A-side stops holding low and crosses the VILA threshold. 

    Either some device is actively driving high, or there is something in the firmware for the I2C timing. 

    Regards,

    Tyler

  • Hi

    the picture that I had provided earlier was the modified version in which the i2c read does not fail (both VCCA and VCCB connected to 3.3V on the Master side). When the VCCB is connected to 5V (5V power supply is on the master side but the 2k pullups are inside the slave pulled up to a 5V supply inside the slave) I can see the clock glitch which I had posted earlier. below is the original configuration that fails all plots that were provided so far as with this configuration. on the ESP32 side which is the VCCA side, the i2c bus is not shared with any other devices. on the slave side (SCLB, SDAB), as far as I know the bus is not shared either.

    Can i use the configuration in which both the VCCA and VCCB are tied to 3.3V as this works for me and I havent seen any i2c read failures as I have seen when the VCCA is at 3.3V and VCCB is at 5V.

    Thanks.

  • Hi Gaurav,

    Thanks for the updated information and schematic. I will need some time to look into this and will respond as soon as I can.

    Regards,

    Jason

  • Hi Gaurav,

    I am still looking into this and will respond by EOD tomorrow, 10/29.

    Regards,

    Jason

  • Hi Gaurav,

    The configuration works for VIH above VCCB. 

    VCCA=VCCB=3.3V while input is 5V is okay

    VCCA = VCCB = 5.0V while input is 5V is also okay. 

    Did you find anything about the releasing the bus ? 

    Does the ESP32 have the ability to control the timing parameters of the I2C state machine, or is there some spec from the  UHP-2500 in terms of PMBus timing that I am unaware of. 

    Regards,

    Tyler