Why Digitize Data?
Before we plunge into how digitization works, it's a good idea to remind ourselves why digitization has become ubiquitious.
Physical quantities are noisy. Light is made up of photons, whose numbers typically vary according to a Poisson Distribution (uncertainty in the number of photons = square root of that number). Volumes, masses, currents, pressures, and temperatures all fluctuate. Every time a signal is transduced from one form to another, noise is added -- until the signal is converted to a number. While there is a small probability that a digital computer will fail or a memory chip wipe out, digitzed data typically is fixed for all time. If the number doesn't change, there's no additional noise. Rule of thumb for the digital age: digitize the signal as close (in space, time, and circuitry) to the point of measurement as possible.
- Flexibility of Digital Processing.
Once we have data recorded as numbers, we can process it in many ways. Signal averaging? Lock-in amplification? Fourier transformation? Filtering? Taking logarithms or exponentiating? Picking every 17th point? Any of these are simple in a computer, and the various data processing algorithms can all be used on exactly the same numbers. In continuous or quantized physical systems, each data stream can be processed only as it occurs and only in one way. Even if several electronic modules are connected to the same node in instrument, the noise in each module will be slightly different.
Results can be delivered in a form most convenient for the user rather than in a form most closely linked to the result of signal transduction.