Thermo Scientific Apex Service Manual page 94

AuditCheck™
There is a region between these limits where the probability of metal
detection will vary from 0% to 100%. For example, there will be a
point X2 where if the contaminate is passed 100 times it will only be
detected 50 times, there will also be a point X1 where the contaminate
will be seen 99.9% of times.
This means in practice that when a metal detector is to be tested using
a manual test piece, a size is chosen somewhere on the curve above X1
so that it may reasonably be expected to be seen 100% of times. A
typical point might be at 'A'. Manual testing consists of passing a test
piece if size 'A' through the head and ensuring that it is detected.
Consider the scenario of the metal detector losing sensitivity such that
its sensitivity characteristic is represented by curve Y. Such a loss of
sensitivity would not be detected by the manual testing regime, since
the test piece would still be seen 100% of the time. Indeed the
sensitivity might decrease further, as represented by curve 'Z', and the
test still be acceptable. The reason for this is that the test piece is now
equivalent to point 'Z2' (the 50% probability point) and, therefore,
there is a 50% chance that the test piece will be detected the first time
it is passed through the head.
Human nature being what it is, most operators would pass the test
sphere a second time, should it fail on the first attempt, and so long as
it was detected on the second pass the operator would accept that the
detector was operating correctly. The detector sensitivity may need to
decrease further before the manual test system shows up an
operational problem. One must question why the manual test system
is a poor method of tracking the performance of a metal detector's
sensitivity. It is because a manual test system is a qualitative (go/no
go) method. What is actually required is a quantitative measurement
system?
The ideal sensitivity test system would have the following
characteristics:
a. Fully automatic at a user definable frequency
b. Testing carried out in the presence of product
c. Test piece always traveling along the same path in the detector
field
d. Testing carried out under normal detecting conditions
e. A true quantitative measurement system
7-10
AuditCheck™ Thermo Fisher Scientific
REC 4258 Rev C