TCA9803: TCA9803 and constant current drive in long cables / high capacitance

2000 pF × 3.3 V / 3 mA = 2.2 µs, so 100 kHz should work.

There are no examples for this situation. You will have to test this yourself.

Hi Eee,

I am not sure what Clemens is calculating here. Looks like RC, where 3.3V / 3 mA = equivalent of a 1.1kohm PU resistor strength. Multiplying by worst case capacitance of the cable = 2000pF gives 2.2 us for 1 RC. 

Standard mode I2C limits the rise-time to <1000 ns according to the I2C spec. I think we are outside this spec since we need a little over 1 RC time constant to reach the required VIH = 0.70 x VCC for the I2C spec. 

I don't think 3 mA of PU current is enough for a 2000 pF cap load at 100 kHz. We need to add stronger PU resistors to increase PU current, but also have sufficient pull-down strength from each device on the bus in order to sink the PU current. 

I would recommend the TCA9617B as this is a fast mode+ capable buffer with IOL's up to 30 mA. I think the question becomes, can the unbuffered target devices on the end of the cable generate enough PD strength in order to sink the PU current from a strong PU resistor? 

Eee Mil said:

This buffer should provide very fast rise time, but is there something I'm missing? The datasheet does not show any examples driving over 400pF capacitances? Is there something I should know when using the B side of repeater from controller PCB to drive long CAT6A cable? Noise, overshoot for CC pull up or other concerns? Propagation delay through the buffer + rise time?

We don't provide examples for TCA9803 driving a cap load over 400pF because it was not designed for I2C modes faster than fast mode. 

Regards,

Tyler