Hi,
Looking at the topic below, I had a doubt
The MSPM0G1507 datasheet shows this eq. at the 7.13 topic, using the Settling Error.


The answer in the link above said that the Settling Error is an error parameter, in percentage, and shows an example (ADCTsampleCalculation.xlsx) calculating the Min Sampling Time using the Settling Error in percentage. The amount of percentage makes sense because 0.5LSB = ~0.01%(0.0001) of the full scale range.
But watching the Precision Labs series: Analog-to-digital converters (ADCs), topic: Driving SAR ADC without amplifiers, we see this equation:

Is the same equation of the datasheet, skipping some capacitances and assuming a Settling Error = 0.5, in units os LSB, not in percentage.
2*2^N = (2^N)/Settling Error = (2^N)/0.5
So my question is: Should we use the Settling Error in the datasheet equation in units of LSB (like 0.5LSB... 0.25LSB) or in percentage of the full scale range?
Thanks!