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.

PCA9515B: PCA9515BDGKR and PCA9546APW

Part Number: PCA9515B
Other Parts Discussed in Thread: PCA9515A, TCA9517A, TCA9517, TCA9546A, TCA9803, TXS0102

I would like to connect my bus master from a daughter card to PCA9546APW I2C switch on a motherboard. Is it okay to then connect PCA9515B on all or some paths of the I2C switch output to drive longer distances? The switch has 4 paths and 2 are driving across back planes. There are not more than 8 devices per path but I wan to be cautious of the PCB trace lengths. Assume 2 feet of distance.

I will add pull ups between then switch and the repeater ICs. Both operate at 3.3V in my design.

We have PCA9515A in our inventory and I assume PCA9515B is a newer and better recommended part to use?

Thanks,

Divakar

  • And one more,

    Also, down the chain is it okay to connect 3.3V to 1.8V I2C buffers for access to 1.8V device

    Master --> Switch --> PCA9515BDGKR --> to some devices at 3.3V and I2C level translator --> 1.8V devices

    Output of PCA9515BDGKR will contain a few 3.3V devices and the level translator.

    It looks like i can't use PCA9515BDGKR for 3.3V to 1.8V and I need another iC?

  • Hey Divakar,

    "I would like to connect my bus master from a daughter card to PCA9546APW I2C switch on a motherboard. Is it okay to then connect PCA9515B on all or some paths of the I2C switch output to drive longer distances?"

    PCA9515B uses a static voltage offset on both sides which means it cannot connect to another buffer on a static voltage offset side. Further detail on this here: http://www.ti.com/lit/pdf/scpa054

    "We have PCA9515A in our inventory and I assume PCA9515B is a newer and better recommended part to use?"

    The difference between these two devices is PCA9515B provides clock stretching support.

    ----------------------------

    2 feet isn't too far. I think if you were to do this TCA9517A may be a better choice if you plan on stacking buffers in series. This device has B side with a static voltage offset, so you can connect multiple in series/parallel and redrive the signal as long as you sit A side to A side or A side and B side together just not B side to B side.

    You may not need to redrive the signal either if the load capacitance isn't too large (I2C spec states 400pF max for standard and fast mode).

    Also if you are redesigning, maybe switch from PCA9546 to TCA9546. Our TCA line up is pin to pin compatible with the PCA and typically more cost competitive (newer process technology). TCA also typically supports lower Vcc rails compared to PCA.

     

    Thanks,

    -Bobby

  • Thank you,

    I was actually just looking at TCA9517. I can change the switch to TCA9546. So I will use TCA9517 as a redriver or level translator and TCA9517 as a switch.

    Divakar

  • Can you send me the datasheet of  TCA9546? I couldn't find it in your website.

    Divakar

  • Ramgopal Divakar said:

    And one more,

    Also, down the chain is it okay to connect 3.3V to 1.8V I2C buffers for access to 1.8V device

    Master --> Switch --> I2C Buffer --> to some devices at 3.3V ---> I2C level translator --> 1.8V devices

    ^this could work; don't forget that an I2C switch can also act as a level translator as well.

    Output of PCA9515BDGKR will contain a few 3.3V devices and the level translator.

    It looks like i can't use PCA9515BDGKR for 3.3V to 1.8V and I need another iC?

    From a functional stand point, this should work as long as you can drive the I2C pins below 0.4V (ViLc of the device). I would just set Vcc to 3.3V since it does not support anything lower than 2.3V. I think the down side to using 1.8V on either side of the device for level translation is some increased leakage current through the input stage since it is not ever fully biased to 3.3V.... You would be very close to 3.3/2V which is where I expect the highest leakage current to occur. Once again though, I would prefer using a newer I2C buffer like TCA9517A or TCA980x <-- (This one you cannot connect B side to the I2C switch though!!!)

    -Bobby

  • Sorry, I forgot the A at the end....

    datasheet is here: http://www.ti.com/lit/gpn/tca9546a

    The other plus with the switch is the PoR requirements are much less strict compared to PCA.

    -Bobby

  • I couldn't access it. It looks like it is removed from your website?

    I can connect my bus master --> TCA9546A --> A side of TCA9517

    If needed I can add more TCA9517 with the limitations you discussed above

    Thanks,

    Divakar

  • Hey Divakar,

    For some reason I can't open it either... I recently tried to push out a revision of our datasheet and I was told it would be up today. I'm wondering if its because the system is uploading the latest copy....

    scps205b.pdf

    EDIT: Here is TCA9546A document revision B (Nov 26th 2019 edition)

    "If needed I can add more TCA9517 with the limitations you discussed above"

    Yes, though there is a voltage rule on VccB. So you can't level translate down to 1.8V on VccB side. Thats why I suggested TCA9803 as a possible alternative since that device should work down to 1.8V on B side.

    -Bobby

  • Bobby,

    thanks for the I2C buffer reference document - http://www.ti.com/lit/an/scpa054/scpa054.pdf. It was quite good. I also have another circuit where I need to connect my microprocessor at 3.3V to a device at 1.8V. It is a 2 wire interface which can be I2C or MDC/MDIO. I saw the reference schematic use TXS0102. I am assuming, they are using I2C because the pull ups are higher. For MDC/MDIO the recommended pull ups are smaller in the slave datasheet.

    In any case, is this chip designed for MDC/MDIO operation?

    Can I use TCA9803 instead on the 1.8V side without the pull ups to accomplish the same?

    If not, should I use TXS0102 or is there a newer IC?

    Thanks,

    Divakar

  • Hello Divakar,

    "In any case, is this chip designed for MDC/MDIO operation?"

    MDC/MDIO is a different standard which operates faster than I2C (normally). The TXS0102 utilizes rise time accelerators to temporarily supply a much faster rising edge in order to meet the MDC/MDIO requirement for rise time. That device looks like it would be a better match than our TCA9803 because it will support the MDC/MDIO clock frequency.I don't think the device was designed specifically for MDC/MDIO.

    The only two downsides I can think of for TXS0102 is that it does not actually buffer the signals and that if you have a lot of inductance/crosstalk/noise then you could possibly trigger the rise time accelerators. For this you may want to include a small dampening resistor.

    "Can I use TCA9803 instead on the 1.8V side without the pull ups to accomplish the same?"

    You don't want to use the TCA980x if you want to support MDC/MDIO.

    To my knowledge TXS0102 does not have a newer IC but I do not directly support that device.

    Thanks,

    -Bobby

  • Hi!

    It wasn't 100% clear. If I am not stacking buffers, is it okay to connect PCA9515BDGKR after TCA9546APWR. Or do I need to consider TCA9517A or TCA980x

    Thanks,

    Divakar

  • Hey Divakar,

    If you will NOT have another PCA9515B connected to another PCA9515B then this is okay. Please note that if you have PCA9515Bs on multiple channels of the TCA9546A then this will count as them being connected when you enable multiple channels.

    "If I am not stacking buffers, is it okay to connect PCA9515BDGKR after TCA9546APWR."

    MCU---TCA9546A---PCA9515B (on a single channel)----slaves

    ^This is okay, pull up resistors may need to be modified to ensure VoL is lower than ViL of PCA9515B before and after TCA9546A.

    -Bobby

  • I want to run by my schematic. Can you add me as a friend?

    Thanks,

    Divakar

  • Hey Divakar,

    I can do that. Though I don't think it notifies me if I receive a private message so you may need to post on here when you sent he message.

    Alternatively, you can reach out to me at duynguyen@ti.com. I assume this way would be easier for us to add attachments.

    -Bobby

  • Hey Bobby,

    thanks. I sent you a schematic.

    Regards,

    Divakar

  • Hey Divakar,

    I've replied to your email. I assume we will continue discussions offline so I will close this thread (internally).

    Thanks,

    -Bobby