I was wondering how CMOS-based capacitive electrodes might be designed (e.g. in this article I stumbled on). It seems to me, that it would make sense to capacitively couple the input of the transistor to the signal you were trying to measure.
But how would one go from one bit of information (one transistor conducting or not conducting as a result of coupling) to a more robust representation of the signal with an amplitude described by multiple bits (i.e. some quantized representation of the signal in, say, 16 bits). Is it possible that the transistors are each designed with different threshold values to represent the various bits that are the amplitude of the signal at that point in time? Is that even practical, or is there some other method that might be employed to achieve that affect?