P82B96: P82B96, TCA9800, TCA9517

Part Number: P82B96
Other Parts Discussed in Thread: TIDA-060013, , TCA9800, TCA9517

Hi Suppot Team,

I have got helps from TI;s support team at 'logic forum" regarding P82B96's VOL issues. Many thanks for this.  After a few readings we decided to go for "differential transmission" of the I2C signals, which is almost the same as your reference design "TIDA-060013, I2C Range Extension", but just with multiple-point slaves.

Now my questions are: 1. in the diagram TIDA-060013 as above: is it OK to replace P82B96 with TCA9800, or TCA9517? Since the long wire is handles by CAN differential transceiver not P82B96, and we need the VOL (Sx/Sy) facing to SALVE's I2C port to be lower than or equal to 0.4V.  We do not see why needs P82B96 here. Please advise.

2 We assume  "mutli-point configuration is allowed", and with termiantion resistors at cable's two-ends,  correct?

Thank you very mcuh!

Tom

  • Hi Team,

    To correct some portion of my question in last post: I should have said "to repalce P82B96" with another part which firstly convert single-end SDA/SCL into differnetial signals (such as  NXP pca9615?),  then by using TCAN1042's differnetial transmission capability to achieve the goal of long cable I2C communication. (please ignroe my saying of TCA9800/TCA9517). Thanks,

    Tom

  • Hi Tom,

    I don't think PCA9615 would be an appropriate replacement for the P82B96 in this set up. The PCA9615 is not creating a true TX and RX output, it is simply creating a differential signal where one signal is the opposite of the other. The P82B96 is creating a true TX and RX output so that when it sends a signal only the TX is active and when it reads a signal it will read it through RX. I believe that the PCA9615 expects to read a differential signal back and the TCAN1042 will not send a differential signal through the TX and RX pin.

    Multipoint configurations are allowed on a CAN bus and as you said it requires 120 ohm termination resistors at both ends of the cable.

    Best,

    Chris

  • Hi Chris,

    Thank you very much . This mostly sovled my issue. The reason why I wanted to repalce P82B96 is its VOL too high "typical ~880mV". Our sensor(the slave)'s I2C port only allows 0.8V as VIL threshold. This is the key issue for us now. If I still have to use P82B96, that will mean I have to add another "voltage level translating" part in order to meet our sensor's 0.8v VIL requirement, which is a big cost for mutiple salves in this long cable application. Would you able to recommend a TI's part which can replace P82B96 but with Sx/Sy VOL no bigger than 0.5~0.6V? I am attaching our circuit block diagram as following (we have multiple slaves(( I2C sensor), the block diagram only shown one  slave).

  • To add a bit to last post: we will still use TCAN1042 as differential transmission, the same as TIDA-060013, but wiht multiple slaves.

  • Hi Tom,

    You can actually change the VOL of the device by changing the pull up resistance. In the datasheet you can see that when the current through the Sx and Sy ports decreases the VOL also decreases.

     .

    By changing your pull up resistance and limiting the current to 0.2 mA the VOL becomes 0.73V typical with a max of 0.79V. Obviously there is a limit to how much you can increase the resistance. The upper limit to a pull resistor is calculated as follows:

    Where tr is your maximum allowed rise time and Cb is your bus capacitance. Since you won't be using the long cables for the I2C side of communication your bus capacitance should be relatively small. This setup was also done at 100 kHz so the maximum rise time should be 1000ns. This is all to say you should have a pretty high ceiling to how large your pull up resistors can be.

    Now, I can go looking for another part but my only concern is that down the line we might find some detail that we skimmed over that makes it hard to interface with the TCAN1042. Since this setup is nicely documented it would be much easier to make it work with the P82B96. However, if you would still like me to find an alternative part let me know.

    Best,

    Chris

  • Hi Chris, 

    I do understand if decrease pull up at Sx/Sy to I2C interface will likely slightly decrease its VOL, however I believe it still has limit not far from 0.8v(maybe within 0.6~0.73V). We believe to need more margins for reliability than just ~100mV. We have done many tests using P82B96: the communication result is unstable (sometimes work sometimes doesn't). We had previously tested the MASTER-SlAVE with direct I2C connection on PCB it worked every time. So that is why we are suspecting the VOL 0.8V caused by P82B96 ( in our long cable application) is likely the problem.

    Although we much appreciate if you could find alternative part replacing P82B96 +TCAN1042  for this application, you do not have to do so if it takes you a lot effort. We do not have much time to wait to make decision.

    Thank you very mcuh for all youe attention and efforts. You have well answered our questions.

    Tom