Bits, Noise, and Linearity; the Imperfections of ADCs

Choose Subtopic

Now that we've seen how a variety of ADCs work, how can we compare and contrast their behavior? Comparisons can be drawn in many ways:

Number of useful bits and encoding scheme

Just because an ADC is physically wired for some number of bits does not necessarily mean that all those bits will be valid under all circumstances. Suppose we think of a 16 bit successive approximations ADC. 3/4 of the way through doing a conversion, the 12 most significant bits have been set. If we simply read out those bits, we have used a 16 bit device to do a 12 bit conversion. Any successive approximations or Σ-Δ scheme be modified so that one can sacrifice resolution in order to gain speed (or vice versa). Similarly, a V to F converter may have hardware allowing, say, 20 bits resolution for a 1 V full scale measurement lasting 10 s. That same hardware would measure 17 bits in 1.25 s (1/8 time = (1/2)^{3}, so 3 fewer bits).

Exercise: For the V to F system described above, how long would it take to do an 8 bit conversion?

Given the answer to the first part of this exercise, if you only wanted 8 bits resolution, would you choose this V to F system or a flash converter?