Characterising and propagating the sources of error and uncertainty in the optical measurement of gears using designed experiments

Abstract In recent years, the growth of optical techniques has introduced the possibility of allowing several alternative methods for non-contact gear measurement to be utilised. Optical methods can offer many advantages over tactile, such as the potential to evaluate delicate surfaces quickly and measure the whole area of the gear tooth flank at the sub-micron level. However, to maximise their potential, the magnitude of error and characterisation of the sources of error and uncertainty need to be understood. By utilising a series of designed experiments with known size gear artefacts, the effects caused by the change of specific key instrument parameters can be evaluated. These measurement trials demonstrate how the results from experimental methodologies can be used to determine the statistical significance of any predetermined instrument variables under study. When correctly planned, designed experiments allow the identification of the sources of error. By applying statistical methods, we can determine if these sources are significant or not. This will allow determination of which parameters need to be defined when optimising the conditions for measurement by comparing the results with those from the UK National Gear Metrology Laboratory. This comparison can also provide guidance on developing measurement uncertainty.