This preview shows pages 1–2. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: ECE 327: Electronic Devices and Circuits Laboratory I Notes for Lab 5 (Analog-to-Digital Converter (ADC) Lab) 1. This is not a lab on analog-to-digital conversion For a signal to be digit al , it must have both a discrete (i.e., countable ) domain (e.g., time) and range (e.g., values). Additionally, its range must be finite . Recall that you can use the digits of your hands (i.e., your fingers ) to count a finite number of items. Discrete-time analog signals can be generated by sampling continuous-time analog signals. * Each sample can have a continuous range of values. * For slow signals, sampling is error free . A digital signal can be generated by quantizing (or quantumizing ) each sample of a discrete- time analog signal. * A quantizer (or quantumizer ) can only produce a finite number of outputs (e.g., 0255). The outputs are called codes . So a quantizer encodes data. The resolution of the quantizer is the number of possible output values. * An analog-to-digital converter (ADC) is a quantizer. When designing ADCs, for a single cost , there is a speedresolution tradeoff. * A digital signal is an estimate of an analog signal, so it is subject to estimation error . NyquistShannon sampling theorem An analog signal that is sampled at constant intervals can be accurately reconstructed so long as the range of frequencies in the signal is small enough. Sampling frequency must be twice the bandwidth of the signal. To reconstruct original signal, sinc interpolation is used to fill-in time between samples. This classical result has been used by analog engineers for years to do lots of useful things. You should NOT associate Nyquist and digital in your head. Quantization noise floor Quantization has two negative effects (that are really the same single effect). (i) It limits the dynamic range (i.e., the maximum / minimum signal ratio) of signals. * To encode a signal that can be large, small variations (even if they are slow) will be lost completely . (ii) It introduces quantization (or quantumization) noise and distortion . * It gives a continuous range of inputs a single identity ( round-off error ). So quantization creates a noise floor . Characteristics smaller than 1 LSB are lost. * LSB = Least Significant Bit = | input range | / resolution * Increase resolution to decrease LSB Quantization noise spectrum can be shaped using dither . * When quantizing a value between two outputs, randomly choose to round up or down based on how close the value is to the nearest output. * For example, for outputs 0 and 1, encode 0 . 75 as 1 75% of the time and 0 25% of the time....
View Full Document
This note was uploaded on 08/05/2011 for the course ECE 209 taught by Professor Staff during the Fall '08 term at Ohio State.
- Fall '08