ADC resolution (Bit number)

Hi Takumi,

I will try to give some very basic explanations.  There are many papers and articles that can be found on the internet by using a web search tool.  You can think of no missing codes like applying an input voltage sweep to the inputs and getting a digital code transitioning at each LSB transition.  No codes are skipped or missing.

ENOB is a little more difficult to explain as this is normally considered as an AC characteristic as opposed to DC.  In the DC case the easiest way to consider this value is by plotting a series of data.  Generally this is done with shorted inputs (or 0 volts).  If you plot the data as a histogram, you would see a Gaussian type distribution where the noise of one standard deviation relates to ENOB.  The number of codes relating to the standard deviation would reduce the overall resolution of the measurement.  Taking the same histogram plot you would see the overall noise +/- of the mean.  This is peak to peak noise and relates to the noise-free resolution.  If you subtract the number of noise bits from the total resolution of the converter you will have the number of noise-free bits.  Generally both ENOB and NFB is given as a value 2^N, so you would have to convert your results accordingly.  NFB is also called flicker-free bits, and is usually the value used to determine the quality and resolution of the measurement as it relates to a stable noise-free result.

As you can see, noise is the determining factor in the measurement.  If you increase the data rate, there is less decimation and the noise increases.  An easy way to think about this is averaging.  By averaging several samples you can 'average out' some of the noise.  It is a little more complicated than that in actual reality, but that is the general idea.  When operating at 80sps you have less averaging than at 10sps and thus increased noise.  The ADS1232 is a delta-sigma ADC where a constant sampling takes place at a specified modulator rate.  The delta-sigma converter is often called an oversampling converter as the sampling occurs much faster than the data rate.  There are many articles that can be found on the internet explaining the delta-sigma concept.

When you apply gain, any external noise will be gained as well and as a result your resolution will lower.  However, there is an interesting effect whereby the PGA actually reduces the ADC input referred noise, so even though the overall resolution decreases the resolution of the measurement will actually increase.  In the end there is an advantage of using the internal PGA over using an externally applied gain source.

Best regards,

Bob B