Académique Documents
Professionnel Documents
Culture Documents
= (1)
[ ]
2
2
5 . 0
) (
) 1 %, 27 , 68 (
r
m
cal
U
n
t s
n t u +
= (2)
NOTE 1 In case the tolerance T of the reference standard is given (t
r
T) instead of U
r
the respective uncertainty
U
r
(k=2) can be calculated:
3
653 . 1 ) 2 (
T
k U
r
= =
NOTE 2 The calibration uncertainty u
cal
is only valid in a small thickness range around t
r
. In case of a larger thickness
range of interest, the uncertainty u
cal
should be estimated on both sites of the thickness range and a linear interpolation
between both values gives the uncertainty of interest as a function of the thickness.
NOTE 3 Very often the accuracy of the calibration is limited by the given uncertainty of the reference standard, as the
uncertainty of the calibration cannot be smaller than the uncertainty of the reference standard used. In order to improve
the calibration a reference standard with a smaller uncertainty is necessary.
NOTE 4 Usually a normalization or zeroing on an uncoated base metal is recommended by the manufacturer at the
beginning of a measurement. The resulting uncertainty of this normalization is considered to be already included in u
cal
.
NOTE 5 ) 1 %, 27 , 68 ( n t : student factor (degrees of freedom 1 = n f and level of confidence with % 27 . 68 = p ).
Respective values are summarized in Annex E.
8.3 Stochastic errors
General repeated measurements are recommended in order to improve the accuracy of the arithmetic mean
value t of the thickness values measured (see paragraph 7.2.), i.e. to reduce the uncertainty of the thickness
result. In case of n repeated measurements the standard uncertainty u
sto
(k=1) of the arithmetic mean t can
estimated by (Type A):
n
t s
n t u
sto
) (
) 1 %, 27 . 68 ( = (3)
The standard uncertainty u
sto
is a measure of all errors arising from unpredictable or stochastic temporal and
spatial variations of influence quantities.
NOTE 1 The standard uncertainty u
sto
can be reduced by increasing the number of repeated measurements. This can
be important e.g. in case of rough sample surfaces.
NOTE 2 Not all contributions to the uncertainty u
sto
are of random nature (Type A). This depends on the design of
experiment. E.g. the measured thickness of a larger sample with a thickness gradient results in a high uncertainty u
sto
because of the systematic thickness variation. In case of a reduced measurement area u
sto
is reduced and the arithmetic
mean value t gives a better description of the local thickness.
NOTE 3 Care should be taken to the risk that Type B standard uncertainties (see e.g. paragraph 8.4.) which might
contribute to Type A standard uncertainties are counted twice.
ISO/IEC WD 2178
12 ISO/IEC 2012 All rights reserved
8.4 Uncertainties caused by factors summarized in paragraph 5
The influence of the factors summarized in paragraph 5 should be minimized by means of a calibration
whenever this is possible. Very often however these influences can only be estimated and the resulting
uncertainty must be considered as a component of the combined uncertainty of the measurement. Simple
experiments to estimate the uncertainty of some of these factors are given in Annex D. Usually the influence
of these factors and therefore the resulting uncertainties are a function of thickness. Consequently in order to
estimate the uncertainty for a given thickness or at least a small thickness range the experiments must be
carried out with samples with the thickness of interest.
As an example the variation of the magnetic properties of the base metal is considered (permeability
variation). As described in Annex D.5 the expected variation should be estimated for the thickness of interest.
The resulting thickness variation with respect to the selected reference base metal should
be ) ( ) (
max min r r bm
t t abs t t abs t = = . This gives the standard uncertainty caused by the variation of the
base metal properties u
bm
(k=1):
3
bm
bm
t
u
= (4)
The same estimation of the standard uncertainty shall be carried out for all relevant factors listed in paragraph
5. For example in case of an expected variation of the surface curvature resulting in
cs
t with respect to the
procedure D.4 the standard uncertainty can be estimated as u
cs
(k=1):
3
cs
cs
t
u
= (5)
NOTE 1 In case the influence of a factor is minimized by a calibration the remaining uncertainty of this calibration must
be considered.
NOTE 2 Some of these factors influencing the accuracy can be minimized by means of flexible foils as reference
standards, e.g. base metal properties (5.4.) or surface curvature (5.6.), if the calibration is carried out with foils on the base
metal with identical material and curvature properties as the sample of interest shows. In this case only expected
variations of the sample properties must be considered.
8.5 Combined uncertainty, expanded uncertainty and final result
The combined uncertainty summarizes all standard uncertainty components (8.2., 8.3, 8.4 and possibly other).
In the described simplified approach to estimate the uncertainties only for a given thickness or a very small
thickness range the sensitivity coefficients can be considered to be equal to 1 (see Annex B). This results in
the combined uncertainty u
c
:
...
2 2 2 2
+ + + + =
cs
bml
sto
cal
c
u u u u u (6)
As the final result the expanded uncertainty U(k=2) is calculated (2-sigma level, 95.45%):
c
u k U 2 ) 2 ( = = (7)
And the complete result of the measurement with the thickness value t :
) 2 ( = = k U t t (8)
9 Test report
The test report shall include the following information:
ISO/IEC WD 2178
ISO/IEC 2012 All rights reserved 13
e) all information necessary for the identification of the test specimen;
f) a reference to this International Standard, including its year of publication, i.e. ISO 2178:20xx;
g) the sizes of the test areas over with the measurements were made in square millimetres (mm
2
);
NOTE Other units of measurement may be used, with agreement between supplier and client.
h) the location(s) of the test area(s) on each specimen;
i) the number of test specimens measured;
j) an identification of the instrument, probe and standards used for the test, including reference to any
validation certification of the equipment;
k) the results of the test, reported as the measured thicknesses, in micrometres, at each area at which the
test was carried out, including the results of the individual determinations and their mean for each
reported measurement;
l) the name of the operator and testing organisation;
m) any unusual features observed and any circumstances or conditions though likely to affect the results or
their validity;
n) any deviation from the method specified;
o) any unusual features (anomalies) observed during the test;
p) date of the test.
ISO/IEC WD 2178
14 ISO/IEC 2012 All rights reserved
Annex A
(informative)
Basics principle of all measurement methods
Figure A.1 shows a ferrous ring coil with a small gap.
Key
I electric current through a coil
n number of windings in the coil
r radius of the ferrous ring
x length of the air gap
H
C
magnetic field strength in the ferrous ring
H
G
magnetic field strength in the gap
Figure A.1 Set-up with a ferrous ring having a gap
The magnetic flux density of a ferrous ring coil (see Figure A.1) with a small gap is calculated from the
equation
x x r
I n
B
r
r
+
=
2
0
(A.1)
where
B is the magnetic flux density;
n is the number of windings;
I is the path of the closed loop;
r
is the relative magnetic value of the used ferrous material;
0
is the magnetic constant;
r is the radius of the ferrous ring;
x is the length of the air gap.
ISO/IEC WD 2178
ISO/IEC 2012 All rights reserved 15
If the gap length x is zero equation (1) becomes
r
I n
B
r
=
2
0
(A.2)
Equation (2) is identical to the formula describing a pure ferrous ring coil.
Considering the other extreme, the gap length becomes identical to the coil length ( ) r x = 2 and
equation (1) describes an air ring coil
r
I n
B
=
2
0
(A.3)
The result of the comparison of equation (2) to equation (3) is: The magnetic flux density in the air gap
decreases by the factor
r
if the ferrous metal is completely replaced by air.
ISO/IEC WD 2178
16 ISO/IEC 2012 All rights reserved
Annex B
(informative)
Basics of the determination of the uncertainty of a measurement of the used
measurement method corresponding to ISO/IEC Guide 98-3
Coating thicknesses are generally determined as mean value of several single measurements which were
carried out at a fixed section of the layer's surface.
On the basis of these measurements a mean value is allocated to the measurand "coating thickness". That
one is assigned an uncertainty value which informs about reliability of the allocation of the value.
The analysis is carried out progressively and begins by drawing up a model equation which shows the
functional correlation between the indicated output value t and all the relevant influence quantities H
i
.
) ... ,... , , (
2 1 0 n i
H H H H H F t = [B 1]
To every influence quantity belongs a sensitivity coefficient c
i
, which indicates how strong a
modification H
i effects the result t.
When the function F is given as analytic expression the sensitivity coefficients may be calculated by partial
derivation.
i
i
H
t
c
= [B 2]
If there is no knowledge about the kind of the functional correlation, an approximation by means of polynomial
functions is recommended.
In many practical cases this formulation is expressed by a linear dependence i. e. the sensitivity coefficients
become one. There is this situation, for example in sections of limited coating thickness.
As the uncertainties of various fault influences must be summarized appropriately, all single uncertainty
components must refer to a level of confidence of 68,27 %, the so called standard uncertainty.
Referred to the calculation of the uncertainty of a measurement to be carried out there are two types of
uncertainties. (Type A und Type B):
Typ A)
The standard uncertainty of typ A is a measure of all random errors arising from unpredictable or
stochastic temporal and spatial variations of influence quantities.
The standard uncertainty corresponds to the point of confidence of the mean value.
n
t s
n t u
sto
) (
) 1 %, 27 . 68 ( = [B 3]
Whereas s is the empirical standard deviation of the repetition measurement n.
ISO/IEC WD 2178
ISO/IEC 2012 All rights reserved 17
) 1 (
) (
1
2
=
=
n
x v
s
n
j
j
[B.4]
and ) 1 %, 27 , 68 ( n t student factor (degrees of freedom 1 = n f and level of confidence with % 27 . 68 = p .
Respective values are summarized in Annex E.
Typ B)
Many influencing factors or faults are not to be described by Type A, e.g. influencing factors of clause 5.
In order to realize a balanced combination of those fault influences with the faults of Type A, those ad-
hoc probability factors are allocated. In many practical case, the influencing factors treated here, are to
be described by a uniform distribution (rectangle distribution).
If an influence quantity may fluctuate within a section H
i
, the resulting uncertainty may be calculated as
follows:
12
max min
)
t t
u
B
= [B.5]
These fluctuation sections generally must be estimated or determined experimentally (see
Annex D).
Partially, at uncertainty analysis already known uncertainties are used e.g. when it comes to the statement
of the uncertainty of reference standards. In this case it must be taken into consideration that these
statements of uncertainty must be converted into the standard uncertainty e. g. for U(k=2) follows the
standard uncertainty:
2
%) 45 , 95 (
%) 27 , 68 (
U
u = B6]
In order to summarize all investigated uncertainties the so called combined uncertainty is to be calculated.
Therefore the fractions of the standard uncertainty are multiplied by their sensitivity coefficients and added
up squared. In a simplified case the sensitivity coefficients are equally one.
=
i
i i
u e u
2
) ( [B 7]
By the multiplication with a coverage factor k 2 to be indicated, you get the expanded uncertainty to
be calculated which should be indicated in the actual result.
u k U = [B 8]
ISO/IEC WD 2178
18 ISO/IEC 2012 All rights reserved
Annex C
(informative)
Basic performance requirements for coating thickness gauges which are
based on the magnetic method described in this standard
C.1 Technical Specification
The manufacturers technical specification should at least provide the following technical information for
instruments and probes:
a) principle of measurement;
b) measuring range;
c) basic information on measuring uncertainty or permissible error of measurement if measuring is carried
out under conditions specified by the manufacturer;
d) information on how measuring results are influenced by the material, curvature and thickness of the base
metal and by the edge effect (measurements close to an edge);
e) battery operating time;
f) function of an undervoltage monitor and automatic undervoltage switch-off;
g) permissible operating temperature;
h) permissible storage temperature;
i) available methods for calibration and adjustment;
j) contact force of probes with spring loaded guiding sleeves;
k) availability of temperature compensation;
l) measuring rate;
m) data memory (design, capacity, data communication);
n) size and weight of instrument (with batteries) und probes.
C.2 Check / Verification of instruments and probes prior to the supply, after repair
and at regular intervals after use
After instrument and probe have been adjusted according to the manufacturers instructions, the measuring
accuracy has to be checked and verified by using a plane and uncoated base metal and a representative
number of coated calibration standards or calibration foils, whose coating or foil thicknesses should be equally
distributed within the measuring range of the respective probe.
Measurement errors must not exceed the manufacturers technical specification.
C.3 Check / Verification of instruments and probes performed on site
The accuracy of instrument and probe should be verified daily. After the instrument has been adjusted
according to the manufacturer`s instruction, verification shall be made with an appropriate number of coated
ISO/IEC WD 2178
ISO/IEC 2012 All rights reserved 19
calibration standards made from the same base metal as the items to be measured or by means of calibration
foils put onto the base metal to be measured. Their thicknesses should cover the expected coating thickness
range. If curved coated items shall be measured verification needs to be executed on items of the same base
metal, geometry and curvature as the items to be measured.
Measurement errors must not exceed the manufacturers technical specification.
ISO/IEC WD 2178
20 ISO/IEC 2012 All rights reserved
Annex D
(informative)
Experimental estimation of factors affecting the measurement accuracy -
examples
D.1 General
Factors affecting the measurement accuracy are summarized and described in paragraph 5. In practical
measurements it is important to estimate the influence of these factors or the resulting uncertainty. Therefore
some examples of simple experiments are given in order to show how the influence of these factors can be
estimated. These experiments are also the basis to estimate the respective uncertainty.
NOTE The following factors can cause differently pronounced influences in case of an instrument working with
combined measuring principles in one probe. Consequently the factors should be estimated separately for each combined
measuring principle.
D.2 Edge effect
A simple edge effect test, to assess the effect of the proximity of an edge, consists in using a clean, uncoated
and even sample of the base metal as follows. The procedure is illustrated in figure D.1.
Step 1
Place the probe on the sample, sufficiently away from the edge.
Step 2
Adjust the instrument to read zero.
Step 3
Progressively bring the probe towards the edge and note where a change of the instrument reading occurs
with respect to the expected uncertainty or to the given thickness tolerance.
Step 4
Measure the distance, d, from the probe to the edge (see figure D.1).
The instrument may be used without correction provided that the probe is further from the edge than the
distance as measured above. If the probe is used closer to the edge, a special adjustment or calibration
correction is required or the additional resulting uncertainty for the used distance must be considered. If
necessary, refer to the manufacturers instructions.
ISO/IEC WD 2178
ISO/IEC 2012 All rights reserved 21
Key
d distance from the probe to the edge
Figure D.1 Schematic representation of the test for edge effect
D.3 Base metal thickness
A simple test to prove that the metal base thickness t
0
is larger than the critical minimum base metal thickness
t
0
crit
, consists in using two (or more) clean, uncoated and even samples of the base metal with the thickness of
interest. The procedure is illustrated in figure D.2.
Step 1
Place the probe on the first sample. It should be proven that the reading is not affected by the edges of the
sample (see D.2)
Step 2
Adjust the instrument to read zero.
Step 3
Place the second sample beneath the first one, place the probe on top of this stack and check the instrument
reading. In case the instrument reading is still zero with respect to the expected uncertainty the base metal
thickness t
0
is larger than the critical minimum base metal thickness t
0
crit
and no additional uncertainty must be
considered. In case of a changed negative instrument reading with respect to the expected uncertainty t
0
is
smaller than t
0
crit
, i.e. the measurement is affected by the too small base metal thickness.
Step 4
In case t
0
is smaller than the t
0
crit
place a third sample beneath the stack of step 3, place the probe on top of
this stack and check the instrument reading. If the instrument reading is still the same as in step 3 with respect
to the uncertainty the critical minimum base metal thickness lies within t
0
< t
0
crit
<2t
0
. In case the instrument
reading shows a larger negative value than in step 3 than two times of th
0
is still smaller than t
0
crit
. Go ahead
and stack further samples in order to estimate t
0
crit
.
The instrument may be used without correction provided that the base metal thickness th
0
is larger than t
0
crit
. If
t
0
is smaller than t
0
crit
a special calibration correction is required and it must be considered that possible base
metal variations cause an increase of the respective thickness uncertainty.
The experimentally determined critical minimum base metal thickness t
0
crit
can be used to estimate the
resulting uncertainty.
ISO/IEC WD 2178
22 ISO/IEC 2012 All rights reserved
In order to improve the accuracy of the estimation of t
0
crit
samples with smaller thickness than th
0
should be
used.
Figure D.2 Schematic representation of the test for base metal thickness
NOTE 1 The procedure to stack several samples in order to simulate an increase of the base metal thickness allows
only a rough estimation of t
0
crit
because the air gap between the samples causes a change of the magnetic properties of
the sample stack in comparison to the respective homogeneous material. However this simplified procedure can easier be
carried out instead of producing base metals with variable thickness.
NOTE 2 In case that the instrument does not display negative values, it is recommended to use a thin foil (e.g. 10 m)
between probe and base metal to observe the decrease of the thickness.
D.4 Surface curvature
A simple test, to assess the effect of the influence of the sample surface curvature, consists in using a clean
uncoated sample of the base metal with different curvature diameters (e.g. cylinder) as follows. All used
samples should provide the same material properties as the base metal. The procedure is illustrated in figure
D.3 using the example of a convex curvature.
Step 1
Place the probe on an even sample (no curvature). It should be proven that the reading is not affected by the
edges of the sample (see D.2) and that the base metal thickness of the sample is larger than the critical
minimum base metal thickness (see D.3)
Step 2
Adjust the instrument to read zero.
Step 3
Place the probe on each sample starting with the largest available diameter and then continue the test with
decreasing sample diameters and note the diameter where a change of the instrument reading (positive
increase) occurs with respect to the expected uncertainty or to the given thickness tolerance.
The instrument may be used without correction provided that the sample of interest shows a larger diameter
than the noted one. If the diameter is smaller, an adjustment or special calibration correction is required or the
additional resulting uncertainty for the used distance must be considered. If necessary, refer to the
manufacturers instructions.
In practical situations the diameter of the samples of interest varies very often. In such a situation the smallest
and the largest diameter expected should be estimated and the instrument should be adjusted on an uncoated
sample close to the average diameter. In this situation the measured deviation for the smallest and largest
diameter can be estimated from the described procedure and used to estimate the uncertainty which must be
taken into account during the measurement.
ISO/IEC WD 2178
ISO/IEC 2012 All rights reserved 23
In order to improve the accuracy of the estimation of the curvature influence, increase the number of samples
with different diameters.
Bild D.3 Schematic representation of the test for curvature effect
NOTE The same procedure can also be used in case of samples showing a concave curvature, however this
concave curvature results in negative thickness readings. In case that the instrument does not display negative values, it
is recommended to use a thin foil (e.g.10 m) between probe and base metal to observe the decrease of the thickness.
D.5 Magnetic properties of the base metal
In practical situations the magnetic properties of the base metal varies very often. A simplified procedure is
described to reduce this influence and to estimate the resulting uncertainty. This procedure requires several
uncoated, clean and even samples representing approximately the expected variation of the base metal
variation. The procedure is illustrated in figure D.4.
Step 1
Place the probe on one of the samples. It should be proven that the reading is not affected by the edges of the
sample (see D.2), that the base metal thickness of the sample is larger than the critical minimum base metal
thickness (see D.3) and that the sample is even (no curvature, see D.4)
Step 2
Adjust the instrument to read zero.
Step 3
Place the probe on each of the samples and notice the reading. It is recommended to carry out repeated
measurements on each sample and to use the average value in the next steps.
Step 4
Calculate the average of the readings of all samples and select the sample with the smallest deviation from
this average.
Step 5
Use this selected sample as a reference base metal to carry out the zero adjustment for all measurements.
The instrument may be used without correction provided that the deviation of the sample with the smallest
reading (or with the largest reading) from the calculated average value is smaller than the expected
uncertainty or the given thickness tolerance.
ISO/IEC WD 2178
24 ISO/IEC 2012 All rights reserved
In case of larger variations the selected sample should be used as a reference base metal and the estimated
deviation of the readings of the described procedure can be used to estimate the uncertainty which must be
taken into account during the measurements.
Figure D.4 Schematic representation of the test for base metal permeability test
ISO/IEC WD 2178
ISO/IEC 2012 All rights reserved 25
Annex E
(informative)
Table of the student factor
Number of
measurements
n
Fraction p in percent
68,27% 95,45%
2 1,84 13,97
3 1,32 4,53
4 1,20 3,31
5 1,14 2,87
6 1,11 2,65
7 1,09 2,52
8 1,08 2,43
9 1,07 2,37
10 1,06 2,32
11 1,05 2,28
12 1,05 2,25
13 1,04 2,23
14 1,04 2,21
15 1,04 2,20
16 1,03 2,18
17 1,03 2,17
18 1,03 2,16
19 1,03 2,15
20 1,03 2,14
1,00 2,00
ISO/IEC WD 2178
26 ISO/IEC 2012 All rights reserved
Annex F
(informative)
Example of uncertainty estimation (see paragraph 8)
F.1 Sample to be measured
a) paint/steel (part of a car body)
b) expected thickness: around 25 m
c) base metal is not accessible, but possible thickness variations caused by the used car body steel charges
(permeability variations) have been determined by an experiment (see D.5): measurement of uncoated
steel parts from car body production representing the variability of used steel from different suppliers,
charges etc., resulting complete thickness variation range at m t m t
bm
2 . 1 : 25 = =
F.2 Steps to do
1) verification of the probe calibration
10 repeated measurements with a reference foil of m t
r
2 . 25 = on base metal (including zeroing
on base metal)
given tolerance of the reference foil: m T 5 . 0 =
used base metal : selected reference base metal (see D.5)
result m t n 06 . 24 : ) 1 ( = = and m t s 11 . 0 ) ( =
calculation of uncertainty and E (see 8.2.):
standard uncertainty of reference foil : m
m T
u
r
29 . 0
3
5 . 0
3
= = =
standard uncertainty of verification measurement (only stochastic component is
considered): ( )
( )
m
m
n
t s
n t u
sto
04 . 0
10
11 . 0
06 . 1 1 %, 27 . 68 = = =
combined uncertainty: ( ) ( ) m m m u
c
29 . 0 29 . 0 04 . 0
2 2
= + =
expanded uncertainty: m u k U
c cal
58 . 0 2 ) 2 ( = = =
( )
96 . 1
58 . 0
14 . 1
2
= =
=
=
m
m
k U
t t
E
cal
r
calibration is not correct. A significant deviation has been detected, because 1 96 . 1 > = E , i.e.
the difference between the measured value t and the given reference foil value
r
t t is larger
ISO/IEC WD 2178
ISO/IEC 2012 All rights reserved 27
than m k U
cal
58 . 0 ) 2 ( = = ; consequently the calibration accuracy can be improved by means of
this reference foil
2) adjustment of the instrument with the reference foil
3) verification of the improved probe calibration
10 repeated measurements (repeat of step 1)
result m t n 87 . 24 : ) 1 ( = = and m t s 11 . 0 ) ( =
calibration is o.k., because 1 56 . 0 < = E , i.e. the difference
r
t t is smaller than
m k U
cal
58 . 0 ) 2 ( = =
,
no significant deviation can be proven now
4) uncertainty of probe calibration (result of step 3):
( ) ( ) m m m u
c
29 . 0 29 . 0 03 . 0
2 2
= + =
: m u
cal
29 . 0 =
5) measurement of the sample
7 repeated measurements within the given measurement area of the sample
result m t n 8 . 22 : ) 7 ( = = and m t s 76 . 0 ) ( =
6) calculation of all measurement uncertainty components and combined uncertainty
stochastic uncertainty: ( )
( )
m
m
n
t s
n t u
sto
31 . 0
7
76 . 0
09 . 1 1 %, 27 . 68 = = = m u
sto
31 . 0 = (see
8.3)
standard uncertainty caused by possible base metal deviation from calibration (expected
thickness variation range m m t
bm
2 . 1 ) 25 ( = : m u
bm
69 . 0 = (see 8.4)
combined uncertainty: ( ) ( ) ( ) m m m m u u u u
bm sto cal c
81 . 0 69 . 0 31 . 0 29 . 0
2 2 2 2 2 2
= + + = + + =
m u
c
81 . 0 = (see 8.5)
7) calculation of the expanded uncertainty and expression of the result
expanded uncertainty: m u k U
c
6 . 1 2 ) 2 ( = = = (see 8.5)
final result of the measurement: m m t 6 . 1 23 =
NOTE 1 All other possible factors affecting the measurement accuracy are considered to be negligible in this example
(edge effect, base metal thickness, curvature, temperature drift etc.)
NOTE 2 Further conclusions: it is obvious that the resulting uncertainty is limited by the largest uncertainty component,
in this case the possible base metal property variation (permeability variation). Therefore an increase of the number of
repeated measurements would reduce u
sto
, however the combined uncertainty wouldnt be strongly affected in this way