FISCHER FISCHERSCOPE X-RAY XDLM 231, Operator'S Manual

Page: 90 / 236

Measurement Uncertainty and Measuring Ranges Systematic Error
88 FISCHERSCOPE
®
X-RAY
Calibration standard set.
Input of the nominal values of the standards. The uncertainties (errors)
must be entered as well. If the pre-set Zero is not changed then the
program will automatically use 1% of the nominal values for coating
thicknesses and area masses, and 1 mass % for concentrations as the
uncertainty, referenced to k=1 (68% confidence interval).
The example applies to the measurement application Au/AuCuCd/Ni/
CuZnPb.
11.3 Systematic Error
While the random errors described above can be characterized
using statistical means, the systematic portion of the error can be
characterized only by comparison with measurement results from
standard samples. The systematic error could be described
precisely and also corrected if the nominal values of the used
standard samples were precise and it were possible to measure
them without a random portion of the error. Unfortunately, this is not
the case. Both the uncertainty of the nominal value and the
(random) measurement uncertainty of the calibration measurement
determine the calibration error. One must additionally take into
account that for measurement results that are distant from the
standard values, the uncertainty of the calibration (with WinFTM,
this is the correction of the standard-free result value) increases
according to experience. The uncertainty of the correction depends,
of course, on the extent of the correction itself. Thanks to the quality
of the FP-supported evaluation program, the required corrections
are typically small, such that this fact rarely dominates the
measurement error. It is significant that this error can now be
specified and the risk is, therefore, known.
One prerequisite for its computation is the specification of the
uncertainties of the standards. This is done in the calibration