FISCHER FISCHERSCOPE X-RAY 5000 Series, Operator'S Manual

Page: 98 / 188

Measurement device monitoring for the Fischerscope X-RAY Random and systematic deviations
98 FISCHERSCOPE
®
X-RAY
standard: In such a case, the user need not be surprised by the
deviation.
Statistics teaches: Only when a mean value of a test series of at
least 7 single readings misses the nominal value of a calibration
standard by more than one standard deviation can one assert that
the deviation is a systematic one. Expressed better and more
precisely: The deviation is systematic when the nominal value of a
calibration standard is outside the confidence interval of the mean
value of a test series.
Systematic measurement deviations from the nominal value always
have a particular algebraic sign: they are either always positive or
always negative. A systematic deviation must be made to disappear
through suitable corrective measures in order to measure correctly.
These measures will be treated in greater detail beneath.
We speak of a random deviation of a measurement result from the
nominal value when the nominal value is within the confidence
interval around the mean value. Thus, a random deviation should
not be considered an ”error”; quite to the contrary, it is actually
natural because it is caused by the variations of the instrument and
can, therefore, be explained. The mean value is
correct despite a
random deviation from the nominal value . In other words: The
difference between the nominal value and the mean value cannot
be resolved with the existing precision of the instrument.
Check measurement
(test) of the calibration The following applies to a check measurement (test) of the
calibration: The mean value of a test series on a calibration
standard may at times be smaller and at times bigger than the
nominal value of the standard when repeating the test series,
however, it must remain (in 95% of all cases) within the confidence
interval. Such a registered random deviation does not need to be,
or actually cannot be, corrected. The user can increase the
repeatability precision of the instrument (e.g., through longer
measurement times or more measurements in a test series). He
can then hope that the random deviations become very small and
the systematic deviations become more apparent.
If two parties obtain different measurement results on the same
specimen and even at the same position, the comparison of the
difference of the measurement result with regard to their confidence
interval (such as between the nominal value and the measurement
result on a calibration standard) also becomes helpful in detecting
a systematic measurement deviation. Only when the confidence
interval of one of the two measurement results does not overlap the
confidence interval of the other measurement result can one speak
of a (systematic) difference. Otherwise one must assume that in
truth the mean values differ randomly. A random difference may be