To achieve an acceptable degree of accuracy at high frequencies in some standardized methods for primary calibration of laboratory standard (LS) microphones, the front cavity depth lfc of each microphone must be known. This dimension must be measured using non-contact methods to prevent damage to the microphone diaphragm. The basic capabilities of an optical depth-measuring microscope were demonstrated by the agreement of its measurements within 0.7 μm of the known values of reference gage blocks. Using this microscope, two basic methods were applied to measure lfc. One (D) uses direct measurements at the microphone front surface annulus and conventional data reduction techniques. The other (GB) uses measurements at the surface of a gage block placed on the annulus, and plane-fitting data reduction techniques intended to reduce the effects of the slightly imperfect geometries of the microphones. The GB method was developed to provide a smoother surface of measurement than the relatively rough surface of the annulus, and to simulate the contact that occurs between the annulus and the smooth, plane surface of an acoustic coupler during microphone calibration. Using these methods, full data sets were obtained at 33 measurement positions (D), or 25 positions (GB). In addition, D and GB subsampling methods were applied by using subsamples of either the D or the GB full data sets. All these methods were applied to six LS microphones, three each of two different types. The GB subsampling methods are preferred for several reasons. The measurement results for lfc obtained by these methods agree well with those obtained by the GB method using the full data set. The expanded uncertainties of results from the GB subsampling methods are not very different from the expanded uncertainty of results from the GB method using the full data set, and are smaller than the expanded uncertainties of results from the D subsampling methods. Measurements of lfc using the GB subsampling method with only nine measurement positions exhibit expanded uncertainties (with coverage factor k = 2) within 4 μm, and can improve the uncertainty of microphone calibrations by an order of magnitude over the result from use of generic standardized microphone type nominal lfc values and tolerance limits.
展开▼
