Understanding Smart Sensors - Nomads.usp

86 Understanding Smart SensorsVccRsV1R1R2RtoV2_+R3(a)R4R7R5_+R8R9VccR11Rg_+ VoutR10GndR9R1R7R8Gnd+−R3(b)R2R5R6+−R4VoutFigure 4.12 Circuit comparison: (a) original 3–op amp design, and (b) reduced componentcount in the new 2–op amp design.measured and trimmed independently of the others. That results in more accuratetrim capability with tighter manufacturing control.The new device has an on-chip heater that is controlled independentlyfrom the rest of the circuit. When the Tco trim is reached, the heater is turned onand a controlled voltage waveform is applied, so that the die is heated and rapidlyreaches a steady-state temperature. The Tco is measured and compensated, andthen the heater is deactivated. The new method achieves a more uniform temperatureprofile, as well as reduced thermal response time and total Tco trimtime. That level of improvement from one design level to the next is part of thesmarter approach that microelectronics can and must provide to sensing.4.4 Digital ConversionVarious A/D architectures are available for integration and interfacing withsensors. Conversion resolution, conversion accuracy, conversion speed or bandwidth,inherent system noise levels, and power consumption are all ADC tradeoffs.In assessing a converter architecture, it is important to consider all thoseaspects. For example, errors due to temperature, supply voltage, linearity, quantizing,and so on, may reduce the accuracy of an ADC by several bits when allerror sources are considered. Also, bit accuracy alone may not be sufficient,especially if the sampling or conversion rate is incorrect for the sensor responseunder consideration [24]. Table 4.2 shows the quantizing errors and otherparameters for 4- to 16-bit A/D conversion [25]. The quantization error (as apercentage of full-scale range) is ±1/2 ⋅ 1/(2 n − 1) ⋅ 100, which is also ±1/2 LSB.The resolution is the least significant bit percentage of full scale/100. The