The TXS0102 does seem like a promising chip (I just noticed that the voltage levels on either side of the shifter/buffer can be equal)\
I was just wondering the tolerance of the chip to voltage "imperfections"-
e.g. If the chip could still handle the VCCb ("lower side") input could be slightly lower (3.25V) than the VCCa ("higher side") input (3.35v), in worst case scenario, if some of the regulators are running a bit low/high respectively...
That may be the area in which the PCA9512 would be slightly (but not much) better- it can interface different voltages, and the "high side" and "low side" don't matter
That's a good question about going between two 3.3V circuits, using the TXS chip, when the B side may end up slightly lower at 3.3V nominal than the A side due to regulator tolerances. I don't see leeway in the data sheet. I'd expect there would be some on the order of 0.1V, but I'd like to know for sure before counting on it. That is not an issue with the NXP chip, because it does not care which side is lower or higher.
Comments
I was just wondering the tolerance of the chip to voltage "imperfections"-
e.g. If the chip could still handle the VCCb ("lower side") input could be slightly lower (3.25V) than the VCCa ("higher side") input (3.35v), in worst case scenario, if some of the regulators are running a bit low/high respectively...
That may be the area in which the PCA9512 would be slightly (but not much) better- it can interface different voltages, and the "high side" and "low side" don't matter
Sorry for my lack of eloquence in explaining
BuildAMachine
As Phil pointed out in post #14, the Prop's diodes will pull the line back down to 3.3V.
Prior to Phil pointing this out, I also would have thought pullups to 5V would have worked.
Actually to 3.3V + Diode forward voltage ~ 3.3+0.6 = 3.9V. Still a bit low for a 4V high level.
Andy
That's a good question about going between two 3.3V circuits, using the TXS chip, when the B side may end up slightly lower at 3.3V nominal than the A side due to regulator tolerances. I don't see leeway in the data sheet. I'd expect there would be some on the order of 0.1V, but I'd like to know for sure before counting on it. That is not an issue with the NXP chip, because it does not care which side is lower or higher.