P82B96: Will the P82B96 work in a daisy-chain configuration?

Hey Tony,

"will this work"

-This will not work. This device generates a VoL of ~800mV and has a ViL of about 600mV. This means VoL will never be lower than ViL so it will not drive the signal through another P82B96 connected on the Sx/Sy side.

One thing I could think of is you can place a bunch of these in parallel but connect them on the Tx/Ty side instead of the Sx/Sy side.

Also, how long of cables are you using and what is your Vcc? We may have another device which could be used in this daisy chain configuration if you really need it to be in series rather than parallel.

Thanks,

-Bobby

100kHz I2C

5v Vcc levels.

This is for an expandable data bus system... PCB's with different equipment will be added to chain. The controller board at the beginning will be able to select which PCB to communicate with. 

What If I used the  P82B96 to drive the cables and the TCA9617B to drive the I2C lines on the PCBs?

The TCA9617B has a Vil of .3xVcc max (~1.5v). That means the P82B96 Vol of 1v  (max)  will be able to drive the low signal right?

And  TCA9617B has a Vol of 580mV (max) would should be enough to drive 600mV Vil threshold of the  P82B96.

Thanks

Tony

Hey Tony,

"What If I used the  P82B96 to drive the cables and the TCA9617B to drive the I2C lines on the PCBs?"
If I understand this correctly, this means you want to put the TCA9617 inbetween the "s" side of the P82B96. This will work but only on a unidirectional line such as SCL (unless you are supporting clock stretching). The problem you have with this configuration is that B side of tca9617B has a parameter called ViLc (input low voltage at contention). Contention on an I2C bus occurs at clock stretching or ACKs between master and slave. The ViLc TCA9617B is 0.4V which P82B715 has a VoL of 800mV. In this configuration you could potentially miss an ACK (at 100kHz you may be able to get away with it).

Because each segment is about 15 feet max and you are running at 100kHz, you can probably get rid of the P82B96 and just use the TCA9517 to drive your cables. Normally I would suggest TCA9803 for this job but at 5V the 9803 can't operate at this voltage.

My suggestion:

TCA9517 is a redriver/buffer which you could use for cable transmission for short distances. My calculations when doing this last time was that TCA9517 could likely support communication across a cat5 cable up to 8 meters (about 26 feet). I assume ribbon cables could be less capacitive so you could potentially go further. A design like this (series connect for redrivers) is limited in how many times you can put these boards in series. Each time you put one in series with another, you generate larger propagation delays. At 100kHz you have much more room with this. I did the calculations once for this kind of situation at 400kHz and found that the limit was around 22 devices in series.

Do you know how many of these boards you would have in series with each other for I2C communication?

Thanks,

-Bobby