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-Q1: Application on TCA9517-Q1

Part Number: TCA9517-Q1
Other Parts Discussed in Thread: TCA9548A-Q1, , TCA9548A

Hi team,

     Support need!

     Our customer wants host device to communicate with 16 pcs slave devices via the I2C bus with the rate at 400KHz. They worry about the capacitor of I2C device getting too large which could effect the signal (like delay).By the way this project is used in vehicle.

      After talking with them, we get two solutions with Ti device.

     1.use I2C mux to change the channel. For this solution, I want to recommend TCA9548A-Q1 which has 8 channel.

     2.use I2C buffer/repeater to improve signal strength. For this solution, I want to recommend TCA9517-Q1

     So here are my questions.

     1.For TCA9548A-Q1, how many devices it can support in each channel  

     2.For TTCA9517-Q1, how much capacitor it can support in I2C bus with speed at 400KHz

     3. Could you please add some your comments for these solution and the device that i want to recommend?

YOURS 

NAN

  • There is no fixed limit on the number of devices. The only limit is the total bus capacitance, which increases with both the number of devices (see their datasheets) and with the trace length (about 1−3 pF/in).

    A mux would be most useful if the bus topology is star-shaped. On a linear bus, you would still have a large capacitance when you go from the master to the last device in the line.

    A buffer can be placed anywhere on the bus, and creates two bus segments that each have their own 400 pF limit. Buffers have a propagation delay, so you cannot use too many at the maximum data rate.

    In general, you can improve realiability slightly by reducing the data rate. So do not use a higher speed than you actually need.

  • Hi Nan,

    As Clemens pointed out the device limit is a fixed value. I2C limits bus cap to 400pF for standard (100kHz) and fast mode (400kHz). 

    The MUX approach you suggest sounds like it would be the cheaper route since you would likely only need 1 unit. You could place 2 devices on each channel assuming they have unique addresses. This would help to separate the bus cap into 8 segments and should lessen the cap loading when only when channel is enabled at a time. 

    If you use the I2C buffer option you would need to be strategic in where you place it, it would also introduce some prop delays but you wouldn't need to spend bandwidth to communicate with it like you would with the TCA9548A.

    If you need to redrive the signal (the bus load is large or long) then using the I2C buffer option would probably be best. If you think you can segment the bus well using an I2C switch/mux then using that solution would be better in terms of prop delays but will require I2C writes to the I2C mux/switch to enable/disable channels.

    -Bobby