ADS1299 LSB

Hey Vladimir,

The formulae are very similar. Note that 2/2^24 is very close to 1/(2^23-1). This difference is made even more trivial by the fact that the noise level on the ADS1299 is from thermal noise rather than quanitzation noise - that is to say, you'd never be able to measure the difference between 0x000001 and 0x000002 because there would be too much thermal noise anyway.

If you want to go by the textbook, I tend to think 2*Vref/2^24 is the more correct formula. That basically says there is a full scale range of 2*Vref (-Vref to +Vref) and there are 24 binary codes making 2^24 possible combinations therefore 2*Vref/2^24 seems correct. However, if you just look at the positive codes, there are only 2^23 - 1 since the highest positive code is 7FFFFF since the word is 2's complement (this is easy to think about with a 2 bit word where b'10 corresponds to -2, but there is no way to make +2 with 2's complement 2 bits). That way of thinking about it seems to indicate that 2^23-1 is the more correct denominator.

Long story short, there are different schools of thought, and in this case, it really doesn't matter that much because you will never obtain 24-bit ENOB with this device. But thank you for pointing out that inconsistency - it can certainly be confusing for folks.

Regards,
Brian Pisani