Professional Documents
Culture Documents
THEORETICAL BASICS
1. Theoretical Basics
Preliminary it was already noted that a measurement is always defective. Therefore,
it should be generally examined first, how to describe measuring errors, how they
must be considered in the measuring result and how they can be explained by the
type of measurement. In accordance with The German Institute for Standardization
(DIN 1319‐1) you do not speak any more from a measuring error but from a
measuring derivation. Nevertheless let us talk about errors.
1.1 Measuring error
The absolute error fa is defined as the difference between the actual value xA which
bears the error, and the desired value xW which should be genuine.
f a x A xW (1.1.1)
The absolute error is a part of the measured value and therefore must be indicated as
a measure with its dimension unit.
Also a relative error fr may be defined which represents an absolute error related to
the desired value. Therefore it is dimensionless.
x A xW
fr (1.1.2)
xW
While electrical engineers mostly use an indication error. In doing so the absolute
error is related to the final value in the measuring range xMB and given in percent.
One names this error then the indication error fAr.
x A xW
f Ar 100 % (1.1.3)
xMB
Wherefrom the errors result, it was not mentioned yet. Following DIN 1319 there are
two different kinds of errors, namely the systematic and the random ones.
Systematic errors are a type of errors whose sign and value either are already known
or, however, are to be calculated anyhow, therefore a correction of the measured
value is possible. Beneath others you will find most of the errors resulting of
environmental effects, as for example a subject to the ambient air temperature.
Random or stochastically errors permit no statement about their sign and their value.
Beneath this you find, e.g., errors which appear while reading the display. Therefore,
they can be estimated only with the methods of the theory of probabilities.
Fachhochschule Frankfurt am Main ‐ University of Applied Sciences
Electrical Engineering
QUIRDER ELECTRICAL METROLOGY 1 ‐ 2
THEORETICAL BASICS
1.1.1 Average Value, Standard Deviation, Variance
A statistical error will be determined by calculating the average value x of n
measurements under the same conditions and with the same equipment
x1 x2 xn
1
x (1.1.4)
n
1 n
x xi (1.1.5)
n i 1
furthermore the standard deviation s as the root mean square deviation
s
1
n 1
x1 x x
2
2 x 2
xn x 2
(1.1.6)
n
1
xi x
2
s (1.1.7)
n 1 i 1
or with an algorithm which is used by your pocket calculator,
1 n 2 1 n
2
s xi xi (1.1.8)
n 1 i 1 n i 1
The transformation of Gl. 1.1.7 to the Gl. 1.1.8 you find in the appendix.
For a sufficiently large number n goes s to the limit , the standard deviation of the
population. This also applies if for an actually insufficiently large number of values
all values are included in the calculation.
The square of this limit is called variance of the population
1
n
xi x
2
2 im s 2 im (1.1.9)
n n
n 1 i 1
The so calculated average value is indicated frequently as the result of the
measuring. But this must not represent the true value of the measured parameter,
however. Therefore, one defines the so‐called confidence limit v within which the
true value with the statistic confidence S is to be expected.
Fachhochschule Frankfurt am Main ‐ University of Applied Sciences
Electrical Engineering
QUIRDER ELECTRICAL METROLOGY 1 ‐ 3
THEORETICAL BASICS
1.1.2 Normal Distribution
For a better understanding of the confidence limit some terms of the theory of
probabilities should be explained first.
In the metrology statistical, independent and in advance unknowable errors lead to
random variables as measured values which are subject to a so‐called normal or
Gaussian distribution. Besides, it is a continuous distribution whose probability
density can be calculated with the function f (x).
x x 2
1
fx e 2 2
(1.1.10)
2 2
x; x; 2 x
A measured value x, a random variable, where such a density function may be
applied, is called normal‐distributed. The expected value x , that means the average
value, and the deviation 2 are the parameters of this distribution. If these both
parameters are known, the probability density is uniquely determined.
The bell‐shaped curve of the normal distribution is symmetrically to the maximum,
the average value. Where the curve shows its inflexion points, there the values of the
standard deviation lie. Then the curve progression asymptotically approaches the
abscissa. Therefore the more the curve is rampant, the less the deviation 2 will be.
The distribution function F (x) of the normal distribution indicates the probability
that a value x is located in the examined area of the normal distribution.
x
F x z; x; 2 dz
(1.1.11)
z x 2
1 x
F x 2
dz
2
e
2 2
x; x; 2
Fachhochschule Frankfurt am Main ‐ University of Applied Sciences
Electrical Engineering
QUIRDER ELECTRICAL METROLOGY 1 ‐ 4
THEORETICAL BASICS
Fig. 1.1.1: Gaussian Bell‐Shaped Curve
If you enlarge the upper integration limit to x , the distribution function will
enclose all values and the probability will be one.
Fig. 1.1.2: The Distribution Function
F x z; x; 2 dz 1
(1.1.12)
The distribution makes use of the shape shown in figure 1.1.2.
For easier handling very often the function will be used as a normalized or
standardized normal distribution. Doing so the function values may be given in
tabular form. With x 0 and 2 1 the functions will have the following form.
x; 0;1 x (1.1.13)
Fachhochschule Frankfurt am Main ‐ University of Applied Sciences
Electrical Engineering
QUIRDER ELECTRICAL METROLOGY 1 ‐ 5
THEORETICAL BASICS
x2
1
e 2 x
2
as density of the standardized normal distribution and
x; 0;1 x (1.1.14)
z2
1 x
e 2 dz
2
as the distribution function of the standardized normal distribution.
Then the associated curves change to the shape as shown in figure 1.1.3.
The function values are present in the relevant literature as tables mostly for x 0 .
The existing symmetry causes the following:
x x (1.1.15)
and
x 1 x (1.1.16)
Fig. 1.1.3: Curve of the Density‐ and the Distribution‐Function of the Standardized Normal
Distribution
If the table is calculated with the lower integration limit x 0 , the distribution
function follows as:
x 0 x
1
(1.1.17)
2
and
Fachhochschule Frankfurt am Main ‐ University of Applied Sciences
Electrical Engineering
QUIRDER ELECTRICAL METROLOGY 1 ‐ 6
THEORETICAL BASICS
x 0 x
1
(1.1.18)
2
1.1.3 Confidence Region
The confidence region marks out the limits, within those and with the specification of
the statistical confidence a measured value may belong to a certain normal
distribution following the distribution function x . To define the confidence
region you may use tabular values as they were calculated for the t test, also
studentʹs test called, (s. Appendix). Then the confidence limit v becomes:
t
v s (1.1.19)
n
The genuine value of the measurement lies with a statistical confidence S given in
percent within the confidence region
xv x xv (1.1.20)
1.1.4 Uncertainty of Measurement
By adding the systematic error xF – estimated or calculated ‐, the measuring
uncertainty u will be determined:
t
u s x F (1.1.21)
n
Therefore after the correction of the registered systematic errors the measuring result
follows from the average value and the measuring uncertainty u:
xe x u (1.1.22)
1.2. Error Propagation
As far as measured values are linked together by a mathematic algorithm their errors
have to be calculated too. Again you differentiate between the systematic and the
random errors.
Fachhochschule Frankfurt am Main ‐ University of Applied Sciences
Electrical Engineering
QUIRDER ELECTRICAL METROLOGY 1 ‐ 7
THEORETICAL BASICS
The measuring values with their systematic mistakes are examined, as follows:
a) Addition:
x1 x1F x2 x2 F x1 x2 x1F x2 F (1.2.1)
b) Subtraction:
x1 x1F x2 x2 F x1 x2 x1F x2 F (1.2.2)
c) Multiplication:
x1 x1F x2 x2 F x1 1 x1F x2 1 x2 F (1.2.3)
x1 x2
x1 x1F x2 x2 F x1 x2 1 x1F
x2 F
(1.2.4)
x1 x2
d) Division:
x1F
x1 1
x1 x1F x1
(1.2.5)
x2 x2 F x2 F
x2 1
x2
If you consider, e.g., only the positive sign and further on you extend the
denominator, so this applies:
x1F x1F x2 F
x1 1 1 1
x1 x1
x 1 x 2
x2 F x2 x2 F x2 F (1.2.6)
x2 1 1 1
x2 x 2 x 2
x1F x2 F x1F x2 F
1
x1 x1 x2 x1 x2
2
(1.2.7)
x2 x
1 2 F
x2
2
x x x
Then with 1F 2 F 1 and 2 F 1 it applies to both signs of the relative errors:
x1 x2 x2
Fachhochschule Frankfurt am Main ‐ University of Applied Sciences
Electrical Engineering
QUIRDER ELECTRICAL METROLOGY 1 ‐ 8
THEORETICAL BASICS
x1 x1F x x x
1 1 1F 2 F (1.2.8)
x2 x2 F x2 x1 x2
e) Raising to a Higher Power:
x1 x1F m x1m 1 m x1F (1.2.9)
x1
Generally it applies to the linkage of systematic errors:
If you add or subtract measured values, you calculate the sum of the absolute errors.
If you multiply or divide measured values, you calculate the sum of the relative
errors.
If the measuring result is calculated by statistically processed measuring values, you
get
y f x1 , x ,, xm (1.2.10)
If you square this equation and sort it according to the purely square and mixed
terms, you get:
2
my m
y y
dyi 2
dx ji 2 dx ji dxki (1.2.12)
x j
j 1 x j xk
j k 1
Because of the statistically distributed signs in the second term, this can be neglected.
For the sum of all square deviations you get:
n n y 2
y
2
y
2
And therefore the standard deviation will become:
2 2 2
y y y
s y s12 s 22 s m2 (1.2.14)
1x
2x
mx
Fachhochschule Frankfurt am Main ‐ University of Applied Sciences
Electrical Engineering
QUIRDER ELECTRICAL METROLOGY 1 ‐ 9
THEORETICAL BASICS
1.3. Dynamic Measuring Errors
Beside the static measuring errors which are to be considered if the whole system to
be measured is in steady state condition, there are additional errors which result
from the measurement of time variant values, and then you speak of dynamic errors.
Now the dependence between the measured values xa and the result xe can be
described with a differential equation.
d n xa d n1 xa d xa
Tn n
T n 1 n 1 T1 xa B0 xe (1.3.1)
dt dt dt
The dynamic behaviour of measuring systems is described like time dependent
systems by their character functions and values in the time and frequency domain.
Dynamic measuring errors are not treated for their complexity.
1.4. Reporting Measured Values
Measuring you may follow three different aims:
* Investigation of the dependence between two or several measured values as a
function of an independently changeable parameter (e.g., current/voltage
characteristic curve of an electric light bulb).
* Investigation of continuous stochastic measured values (e.g., voltage with overlaid
noise voltage measured continuously).
* Investigation of stochastic measured values by single measurements (e.g., repeated
measurement of a resistance).
1.4.1 Pairs of Measured Values
A typical report of pairs of measured values, which are a function of an independent
changeable parameter, is a chart. The uncertainty u may be drawn as a vertical dash,
so that you can draw a curves adaptation within the framework of the measuring
uncertainty with estimation by sight. If you do it with a computer, you can arrange
the curves adaptation mathematically by a so‐called curves‐fit, i.e. you calculate the
coefficients of the equation so you get the minimum square deviation to all measured
pairs.
Fachhochschule Frankfurt am Main ‐ University of Applied Sciences
Electrical Engineering
QUIRDER ELECTRICAL METROLOGY 1 ‐ 10
THEORETICAL BASICS
Fig. 1.4.1: Presentation of Measured Value Pairs
For n measured pairs of values xi and yi you may get a linear function
y a0 a1 x (1.4.1)
where the sum of the square deviation in each measuring pair becomes a minimum.
This kind of the curves adaptation is called linear regression and the function is a
regression straight line. The coefficients of the regression straight lines are:
n n n
n x i y i xi y i
a1 i 1 i 1 i 1
2
(1.4.2)
n
n
n xi xi
2
i 1 i 1
1 n n
a0 yi a1 xi (1.4.3)
n i 1 i 1
A further analysis determines the correlation coefficient r,
n n n
n x i y i xi y i
r i 1 i 1 i 1
(1.4.4)
n
n 2
n
n
2
n xi x i
n y i y i
2 2
i 1 i 1 i 1 i 1
Fachhochschule Frankfurt am Main ‐ University of Applied Sciences
Electrical Engineering
QUIRDER ELECTRICAL METROLOGY 1 ‐ 11
THEORETICAL BASICS
1.4.2 Continuous Stochastic Measured Values
The evaluation and graphic presentation of continuous, stochastic measuring
parameters result in an average value in examined section.
Fig. 1.4.2: Continuous Stochastic Measured Values
mathematically seen you apply an integration:
1 T
T 0
x xd t (1.4.5)
Besides the correlation analysis finds its application on such measured values. You
distinguish two kinds of correlation.
There we have the autocorrelation, which enables us to find periodic signals in a
strongly jammed signal. By shifting a sequence of the signal as far as the product of it
and the measured values becomes a maximum (mathematical: convolution) (radio
astronomy).
1 T
T 0
11( ) u1t u1t d t (1.4.6)
Fig. 1.4.3: Block Diagram of a Correlator (Autocorrelator: Switch Position S1)
Fachhochschule Frankfurt am Main ‐ University of Applied Sciences
Electrical Engineering
QUIRDER ELECTRICAL METROLOGY 1 ‐ 12
THEORETICAL BASICS
The other type is the cross correlation which you apply to measured values to find
out the interrelation between two stochastic signals by a convolution (controlled
systems).
1 T
T 0
12 ( ) u1t u 2t d t (1.4.7)
Fig. 1.4.4: Example of an Autocorrelation
1.4.3 Stochastic Measured Values
For the presentation of stochastic measured values you arrange the statistic
distribution of each measured values around a desired value, a histogram, a bell
curve or a curve of the cumulative frequency.
Moreover you divide the area determined by the measured values into k equal
classes and assign the measured values to the classes. Class width m should be
chosen smaller than 1/3 of the standard deviation s (with 9 < k < 25).
In the cumulative frequency curve the cumulative frequency values H will be
inserted every time on the lower border of a class. They are formed by the addition
the frequency values H of the separate classes, beginning with the highest class.
Moreover, the frequency values are applied mostly as per cent values. This qualified
presentation gives you the chance of a straight comparison of different
measurements.
Fachhochschule Frankfurt am Main ‐ University of Applied Sciences
Electrical Engineering
QUIRDER ELECTRICAL METROLOGY 1 ‐ 13
THEORETICAL BASICS
The average value is located at 50% of the cumulative frequency curve. The value of
the standard deviation you find as the difference between 50% and 84,13% or
between 50% and 15,87% of the cumulative frequency.
For the presentation of the cumulative frequency curve you find special paper whose
ordinate is divided in such a way so that you get a straight line by a normal
distribution.
Fig. 1.4.5: Cumulative Frequency Curve on Special Paper
If one applies the class frequency H of the measured values in the middle of each
class, one can draw a histogram or a bell curve too. The maximum of the curve
represents the average value. The difference between the average value and the
60,6% value indicates the standard deviation.
Fachhochschule Frankfurt am Main ‐ University of Applied Sciences
Electrical Engineering
QUIRDER ELECTRICAL METROLOGY 1 ‐ 14
THEORETICAL BASICS
Fig. 1.4.6: Histogram, Bell Curve and Cumulative Frequency
1.5. Measuring Parameters and Unities
Measured values consist, how already mentioned, of the unit and the absolute
measure. Therefore the units have to be defined and as far as possibly to be
internationalized. The units have been named as the so‐called „international unity
system“ (SI) in 1960 by the
11th General conference for measurement and weight of the international meter con‐
vention.
Finally in 1971 it was extended to altogether seven basic unities and two supplement
unities.
Fachhochschule Frankfurt am Main ‐ University of Applied Sciences
Electrical Engineering
QUIRDER ELECTRICAL METROLOGY 1 ‐ 15
THEORETICAL BASICS
Base item Formular symbol Basic unit Short symbol
1)
1 meter are 1.650.763,73 vacuum wavelengths of the radiation which corresponds to the crossing between the
levels 2p10 and 5d5 of the atom krypton 86.
2)
1 kilogram is the mass of the international kilogram prototype.
3)
1 second is the duration of 9.192.631.770 periods of the radiation which corresponds to the crossing between
both hyperfine structure levels of the initial state of the atom caesium 133.
4)
1 ampere is the strength of the time invariant current through two straight‐line, parallel, infinitely long
conductors of the relative permeability 1 and from negligible cross section which have the distance 1 m and
between those the force caused electro‐dynamically by the current for each 1 m of the double line amounts
2∙10‐7 N in free space.
5)
1 Kelvin is the 273,16th part of the (thermodynamic) temperature of the tripe point of water.
6)
1 candela is the light intensity which a black body of the surface 1/600.000 m2 radiates vertically to its surface at
the solidification temperature of the platinum and the pressure of 101.325 Pa.
7)
1 Mol is the material amount of a system of a certain compounding which exists of the same number of
particles as atoms are contained in 12/1000 kg of the atom carbon 12.
[ Avogadro constant: 12 g 12C contain (6,02252 ± 0,00028) ∙ 1023 atoms]
8)
1 Radiant is the angle which is shaped by two rays going out of the centre of a circle with the radius 1 m and
including a segment of the circle with the length 1 m. (1 rad = 57°17 ʹ 44,8 ʺ)
9)
1 Steradiant is the solid angle which is shaped by a band of rays going out of the centre of a ball with the radius
1 m and including the surface of 1 m2 on the surface of this ball.
Fachhochschule Frankfurt am Main ‐ University of Applied Sciences
Electrical Engineering
QUIRDER ELECTRICAL METROLOGY 1 ‐ 16
THEORETICAL BASICS
In a lot of applications these unities are either too big or, however too small.
Therefore, headers are internationally defined in order to form decimal multiple and
parts.
SI‐ abbreviaton factor SI‐ abbreviation factor
header header
exa E 1018 deci d 10‐1
1.6. Principle of Indication
Referring to the principle of indication one selected measuring instruments into
analogue and digital ones, today the difference is not so easy to explain. Measuring
instruments constructed in analogue technology may have a digital display and
those in digital technology may have an analogue scale reading (wrist watch!).
1.7. Principle of Measuring
One can measure in two ways:
One method is to set as much units against the measuring parameter till the system is
balanced. The advantage of this method is that you need no calibration of the
measuring system, because only the equality is stated by the measuring parameter
and the measure. The units required for the balance represent the measuring result
(beam balance).
Fachhochschule Frankfurt am Main ‐ University of Applied Sciences
Electrical Engineering
QUIRDER ELECTRICAL METROLOGY 1 ‐ 17
THEORETICAL BASICS
The other way is to use a pointer which will be deflected in relation to the measuring
parameter. The result may be read out on a scale. Moreover the scale must be
calibrated, i.e. it must have been compared with a normal of the measuring
parameter previously and scaled.
1.8. Measuring Instruments in the Measuring Circuit
If you bring a measuring instrument in a system to measure it, the system is always
affected and the measuring result is falsified generally.
In the electric metrology this particularly appears measuring current and voltage
simultaneously, due to the fact that the distribution of the current and the voltage in
the measuring circle are disturbed. Because this is, however, a systematic error, a
correction is possible.
Measuring simultaneously the current and the voltage you may distinguish between
two systematic errors, i.e. the ʺcurrent‐correctʺ or the ʺvoltage‐correctʺ circuit of the
measuring instruments; one names both variants also ʺvoltage‐errorʺ or ʺcurrent‐
errorʺ.
In the ʺcurrent‐correctʺ circuit the current measuring instrument – ampere meter ‐
measures the correct current I I A . The voltage measuring instrument – voltmeter ‐
measures the total voltage minus the voltage drop across the ampere meter
U U V Ria I .
Fig. 1.8.1: ʺcurrent‐correctʺ or ʺvoltage‐errorʺ Measurement
Fachhochschule Frankfurt am Main ‐ University of Applied Sciences
Electrical Engineering
QUIRDER ELECTRICAL METROLOGY 1 ‐ 18
THEORETICAL BASICS
In the ʺvoltage‐correctʺ circuit the voltmeter measures the correct voltage U U V .
The ampere meter measures the total current minus the current needed in the
U
voltmeter I I A .
Riv
Fig. 1.8.2: ʺvoltage‐correctʺ or ʺcurrent‐errorʺ Measurement
Therefore the application of the measuring circuit is corresponding to the value of
the load resistor Rx:
voltage‐error method: R x 100 * R iA
current‐error method: R x 0,01 * R iV
With a measuring resistor in the range of the internal resistor Ria of the ampere meter
the voltage drop across the ampere meter can not be neglected, i.e. it must be
measured voltage‐correct. With a measuring resistor in the range of the internal
resistance Riv of the voltmeter, however, the current flow across the voltmeter can not
be neglected, i.e. it must be measured current‐correct.
1.9. Limits of the Measurability
As a result of the thermal agitation of the charge carriers stochastic fluctuations are
superposed to the currents and the voltages. This limits the measurability and the
measuring accuracy of small parameters. In addition the mechanically indicating
measuring instruments are influenced by the Brownian motion of the mobile
measuring element.
Fachhochschule Frankfurt am Main ‐ University of Applied Sciences
Electrical Engineering
QUIRDER ELECTRICAL METROLOGY 1 ‐ 19
THEORETICAL BASICS
For a critically damped galvanometer, e.g., with the closing resistor Rgr 4 k , the
time of oscillation T0 10 s , the temperature of the measuring circuit 290 K , the
Boltzmann constants k 1,38 10 23 J K 1 and the efficiency factor of the
galvanometer 0,2 the average fluctuation squares of the current i 2 and the
voltage u 2 become
T0 1
i 2 Rgr T0 u 2 k
Rgr 2
Now we solve the equation for i :
1
i k
2 Rgr T0
And we get:
i 0,5 pA and u Rgr i 2 nV
This is a theoretical value which can not be reached in practice.
Fachhochschule Frankfurt am Main ‐ University of Applied Sciences
Electrical Engineering
QUIRDER ELECTRICAL METROLOGY 1 ‐ 20
THEORETICAL BASICS
Appendix:
Calculation to form the algorithm of the standard deviation:
n
1
xi x
2
s
n 1 i 1
and
1 n 2 1 n
2
s xi xi
n 1 i1 n i1
should be same, or:
x
2
n n
1 n
x xi
2
i x 2
i
i 1 i 1 n i 1
First the bracket on the left side of the equation is dissolved:
x x x
n
i 1
i
2
i 1
n
2
i 2 xi x x
2
i 1
n
x 2 xi x x
2
i
n
i 1
2
Now it is to be proved that this applies:
2
n
1 n
2 xi x x xi
2
i 1 n i 1
If one forms the sum of all 2 x x x , the term x
2 2
i exists n‐times in this sum, i.e. it
is follows:
2 x x x 2 x x n x
n n
2 2
i i
i 1 i 1
If you now use the equation for the average value (see Gl.1.5), so you find:
2
n
1 n 1 n
n
2 xi x n x 2 x i xi n xi
2
i 1 i 1 n i 1 n i 1
1 n n
n 2
2 xi xi xi
n i 1 i 1 i 1
2
1 n
xi q.e.d .
n i 1
Fachhochschule Frankfurt am Main ‐ University of Applied Sciences
Electrical Engineering
QUIRDER ELECTRICAL METROLOGY 1 ‐ 21
THEORETICAL BASICS
Table 1: Distribution Function:
z2
x
x; 0;1 x
1
e 2 dz
2
x x x x x x x x
Fachhochschule Frankfurt am Main ‐ University of Applied Sciences
Electrical Engineering
QUIRDER ELECTRICAL METROLOGY 1 ‐ 22
THEORETICAL BASICS
Table 2: t – Distribution
‐ values in per cent
Bilateral problem
m\ 50 25 10 5 2 1 0,2 0,1
1 1,00 2,41 6,31 12,7 31,82 63,7 318,3 637,0
2 .816 1,60 2,92 4,30 6,97 9,92 22,33 31,6
3 .765 1,42 2,35 3,18 4,54 5,84 10,22 12,9
4 .741 1,34 2,13 2,78 3,75 4,60 7,17 8,61
5 .727 1,30 2,01 2,57 3,37 4,03 5,89 6,86
6 .718 1,27 1,94 2,45 3,14 3,71 5,21 5,96
7 .711 1,25 1,89 2,36 3,00 3,50 4,79 5,40
8 .706 1,24 1,86 2,31 2,90 3,36 4,50 5,04
9 .703 1,23 1,83 2,26 2,82 3,25 4,30 4,78
10 .700 1,22 1,81 2,23 2,76 3,17 4,14 4,59
11 .697 1,21 1,80 2,20 2,72 3,11 4,03 4,44
12 .695 1,21 1,78 2,18 2,68 3,05 3,93 4,32
13 .694 1,20 1,77 2,16 2,65 3,01 3,85 4,22
14 .692 1,20 1,76 2,14 2,62 2,98 3,79 4,14
15 .691 1,20 1,75 2,13 2,60 2,95 3,73 4,07
16 .690 1,19 1,75 2,12 2,58 2,92 3,69 4,01
17 .689 1,19 1,74 2,11 2,57 2,90 3,65 3,96
18 .688 1,19 1,73 2,10 2,55 2,88 3,61 3,92
19 .688 1,19 1,73 2,09 2,54 2,86 3,58 3,88
20 .687 1,18 1,73 2,09 2,53 2,85 3,55 3,85
25 .684 1,18 1,71 2,06 2,49 2,79 3,45 3,72
30 .683 1,17 1,70 2,04 2,46 2,75 3,39 3,65
40 .681 1,17 1,68 2,02 2,42 2,70 3,31 3,55
60 .679 1,16 1,67 2,00 2,39 2,66 3,23 3,46
120 .677 1,16 1,66 1,98 2,36 2,62 3,17 3,37
.674 1,15 1,64 1,96 2,33 2,58 3,09 3,29
m/ 25 12,5 5 2,5 1 0,5 0,1 0,05
Unilateral problem
Fachhochschule Frankfurt am Main ‐ University of Applied Sciences
Electrical Engineering
QUIRDER ELECTRICAL METROLOGY 1 ‐ 23
THEORETICAL BASICS
Example 1.1
A resistor measurement under the same conditions shows the following 20 values of
Ri in ohm:
680 684 684 672 664 692 685 681 676 668
693 681 676 696 689 672 688 682 677 680
Please calculate the average value, the standard deviation, the confidence region for a
statistic safety S = 95% and the measuring result. The estimated systematic error
amounts to 0,1%.
Example 1.2
the measuring result is the function of the measuring parameters:
A) y a x1 b x2 c x3
x1 x2
B) y
x3
Calculate in each case formally the standard deviation.
Example 1.4.1
Between the current I and the voltage U of an electric light bulb exists a dependence
which can be described with the power function:
a
U
I k I 0
U0
With the measured values U and I
U in V 30 50 70 90 110 130 150 170 190 210 230
I in mA 340 421 496 563 623 677 726 772 817 860 903
are to be calculated the exponent a, the factor k and the correlation coefficient r.
Fachhochschule Frankfurt am Main ‐ University of Applied Sciences
Electrical Engineering
QUIRDER ELECTRICAL METROLOGY 1 ‐ 24
THEORETICAL BASICS
Example 1.4.2
A resistor measurement under the same conditions shows the following 20 values of
Ri in ohm:
680 684 684 672 664 692 685 681 676 668
693 681 676 696 689 672 688 682 677 680
Outline the histogram, the bell curve and the cumulative frequency curve.
Fachhochschule Frankfurt am Main ‐ University of Applied Sciences
Electrical Engineering