TXS0102: VIH of the TXS0120

So, is the CC2650 also running on 3.3V? Or are you supplying 3.0V to that, and 3.3V to the logic device?

I don't think there are any issues, I just want to be clear -- if both devices are running off the same supply, you should expect the translation to work without any problems. If you start adjusting supplies individually, there might be issues.

Hey sanghoon,

The TXS family of translators have a pretty unique internal architecture.  They work on the same principles as the LSF series (explained here: The Logic Minute), but have some added 'bells and whistles' to improve signal integrity.

The CC2650 is a standard CMOS device and thus will output _very_ close to the rails as long as you are outputting to a relatively high-impedance input. You can expect to see the output of the CC2650 at ~3.3V if your supply is 3.3V, and thus the input to the TXS device will also be ~3.3V.  If you have additional loads on that line, then this can change, but from the sounds of your post, the TXS is the only thing connected to the UART pins of the CC2650.  There's nothing to worry about on this end.

To further clarify, you said that "VOH of the CC2650 is typicaly 2.68[V]@3.0VDDS" -- note that the supply is different, and there's no current rating listed. I looked in the datasheet, and the 2.68V rating is at a load of 8mA -- that's a ridiculous amount of current for the input of the TXS. I would expect to see maybe 1uA of current into the TXS... thus your voltage will _not_ be 2.68V, and it will be about 3.3V (likely 3.299999V... most equipment can't detect that accurately, so I say 3.3V).

On the output of the TXS, the internal pull-up resistor (10kohm) will be forcing the output to 5V.  If your load is similarly light (~1uA) you can expect to see about 4.99V at the output.  The only way this will change is if you drive a significant current (ie connect a 1k resistor to ground or other heavy load), then there might be problems.

To summarize - you have nothing to worry about.