ADC Conversion Errors

Quantization Error: A digital error in an A/D converter is based on the resolution of the digital system. In A/D conversion, a continuous analog voltage is represented by an equivalent set of digital numbers. When the digital numbers are converted back to an analog voltage by a D/A converter, the output is a staircase waveform, which is a discontinuous signal composed of a number of discrete steps.

The smallest digital step is due to the LSB and it can be made smaller only by increasing the number of bits in the digital representation. This error is called quantization error or digitizing error.

Analog Error: In an A/D converter, an analog error is mainly due to variations in the DC switching point of the comparator. The variations in switching are mainly due to offset, gain, and linearity error of the operational amplifier used in the comparator. The other sources of analog error are the resistors in the A/D converter, the reference voltage source, and the ripple and noise introduced by the circuit components.

Linearity Error: This is an important measure of A/D converter performance. It is defined as a measure of the variation in voltage step size. This indicates the difference between the transitions for a minimum step of input voltage change. This is normally specified as a fraction of 1 LSB.

Differential Non-linearity (DNL) Error: The analog input levels that trigger any two successive output codes should differ by 1 LSB (DNL = 0) for an A/D converter. Any deviation from 1 LSB value is defined as a DNL error. This counter type and continuous type A/D converters normally have better differential linearity than successive approximation type A/D converters.

Integral Non-linearity (INL) Error: This figure shows that an actual A/D converter characteristics with a missing code. The dotted curve represents the locus of the midpoints of the actual input step voltage ranges. This line is called the code centerline. The maximum deviation of the code centerline from the straight line passing through the endpoints of the ideal characteristics after nulling the offset and gain errors is called Integral Non-linearity Error (INL).