Designing for sensor accuracy:
- ask yourself if you need absolute, relative or null measurement
- ask yourself if you need direct or indirect measurement
- ask yourself what precision and accuracy you need
- draw out and define the system
- with those answers in mind choose your sense methodology
- then choose your sensor+detector
- calculate the output equation that calculates the output from the input
- add all the sources of error to that equation
Some things to consider:
- size the detector output swing (output value from min to max) to the full range of the detector input
- understand your noise sources and choose a sensor that is less sensitive to these sources
- bound the error sources in your design, eg control the physical environment (like vibration), power supply voltages, etc to mitigate error (noise)
- use binning (to reduce process error)
- break out components and discreetly address error in each component and implement a more accurate version
- use multiple sensors (e.g. RSS or differential measurement)
- calibrate the remaining error
- pre-filter the source signal
- post-filter the output
- choose your bandwidth (or integration time) carefully
- make sure you utilize the full-scale voltage of each input stage
- perform a tolerance analysis of your design
- perform a sensitivity analysis of your design