Embedded designers should pay attention to the following circuit and firmware concerns with regard to analog sensor signal conditioning.
Sensor inputs need to be protected against Electrostatic Discharge (ESD), overvoltage, and overcurrent events.
Analog filters are recommended to improve Analog-to-Digital Converter (ADC) performance. When properly designed, they help prevent interference from ADC aliasing and can reduce sample frequency requirements. In many cases, single-pole RC passive filter is sufficient although active filters can be implemented for higher filtering performance.
Microchip FilterLab® is an innovative software tool that simplifies active filter design, which consists of resistors, capacitors and op-amps. It generates the frequency response and the circuit as well as the SPICE model of the designed filter. It is available at no cost at Microchip's FilterLab® website.
- Designs low-pass filters up to an 8th order filter with Chebychev, Bessel or Butterworth responses from frequencies of 0.1 Hz to 1 MHz.
- Designs band-pass and high-pass filters with Chebychev and Butterworth responses
- Low-pass filters use either the Sallen Key or Multiple Feedback topologies
- Band-pass filters use the Multiple Feedback topology
- High-pass filters use the Sallen Key topology
- Capacitor values can be manually selected
- Generates a SPICE model of the designed filter
- Design Wizards to simplify user inputs
- Design Filter Wizard
- Design Circuit Wizard
- Filter Selection Wizard
- Anti-aliasing Wizard
In most sensor conditioning circuits, the conditioned sensor outputs are converted to digital format by an ADC. Most sensor outputs are ratiometric so that variations in power supply are corrected by the ADC, e.g., Wheatstone bridge.
Sensor errors can be corrected by calibrating each system. The can be accomplished in hardware, e.g., digital potentiometer or firmware calibration in non-volatile memory.
Correction for other environmental parameters may be needed. For example, a capacitive humidity sensor may need correction for temperature. Non-linear sensors need additional corrections. Polynomials or other mathematical functions in the MCU can be used to produce the best estimate of the correction parameter. Linear interpolation table in firmware can also be used to correct sensor error.