FLOATING-POINT ANALOG-TO-DIGITAL CONVERTER

The demand on very high resolution A/D converter can be eliminated by using the invented floating-point A/D converter when the resolution is merely used for covering the signal dynamic range rather than the quantization accuracy. This can be achieved by producing m amplified analog signals with amplifications 2.sup.(i-1)k where k=constant and i=1, 2, . . . m. The largest linearly amplified signal will be selected by a logic circuit (after sampling) and converted into an n-bit digital data code by an A/D converter. In the same time, the logic circuit produces an m-bit logic flag code. The n-bit data code (u), the m-bit logic flag code (v) and the constant k are combined to form a final digital output uv.sup.k with n+(m-1)k bits. In this way, the resolution and dynamic range can be designed independently. Unlike the know logarithmic amplifier solution, the floating-point A/D converter give a linear digital code output directly without using any look-up table. For large and small signals, the effective resolutions are kept constant (or quasi-constant to be accurate). It is also very useful to work with a small input range imposed by a low power supply voltage since its virtual input range is much larger than the actual one. As an A/D converter, its accuracy distribution along the signal amplitude is more rational, an advantage similar to that of a floating-point number representation.