THE MAGAZINE FOR COMPUTER APPLICATIONS
Circuit Cellar Online offers articles illustrating creative solutions
and unique applications through complete projects, practical
tutorials, and useful design techniques.

ANALOG SYSTEM DESIGN

by George Martin

Start ý Considerations ý Transducer Interfaces ý Thermocouples ý ADCs ý Almost Thereý ý Sources and PDF

ADCs

Letýs now look at ADCs. Two of the key parameters are the number of bits and the input voltage ranges. The number of bits is the precision of the device, but let me tell you that this one parameter can open a can of worms bigger than youýve ever seen. You would expect that an 8-bit converter would have 256 unique output values for 256 unique input voltages. Then you would expect that increasing input voltages would produce increasing output values (monotonic or monoticity). Then you would expect that span of the input voltage for each output value would be the same. With a 10-V input range and 10 output values, each output value would span one input volt (not a very high precision device). If some of the output values only spanned 0.1 V on the input side, then others would span more than 1 V. This converter is monotonic, but not linear. And of course, no missing code would be a good deal.

The other ADC parameter is input range. Associated with input range is the reference voltage for the converter. The input range is just what it appears, 0 to 10 V, or ý10 to 10 V. Many of the new converters have programmable input ranges, even programmable per channel. The reference voltage, which is sometimes generated by the ADC, is a requirement as an input. There are precision IC voltage references that are available if an external reference is required.

The ADC gives you the ratio of the input value to the reference voltage, so a 12-bit converter has 4096 output values. These range from 0 to 0xFFF, or 4095. And, an A/D reading of 4095 is equal to 4095/4096 of the reference voltage. Therefore, the following is true:

I used to use references of 10.24 and 2.048 V to avoid the math of converting to decimal units and the associated rounding errors. But, Iým not so sure thatýs necessary anymore.

PREVIOUS • NEXT