Sei sulla pagina 1di 69

CHAPTER 2

GENERAL IZED
PERFORMANCE
CHARACTERISTICS OF
INSTRUMENTS

2.1 INTRODUCTION
The type of instrument to be used is decided on the
characteristics required. A 0.5o C accuracy instrument is
allowable for human body (feeling) while it may be useless
for an instrument in a control system. So for selection the
performance characteristics of measuring instruments must
be known.
Instrument performance characteristics is generally broken
down into two, namely
Static characteristics
Dynamic characteristics,

2.2 STATIC CHARACTERISTICS


The set of performance for applications that involve
the measurement of quantities which are constant or
vary only quite slowly are defined in terms of static
characteristics. These characteristics generally show
up as non-linear and statistical effects.
Since static characteristics affects the dynamic
behavior, the overall performance is then judged by
a semi-quantitative superposition of the static and
dynamic characteristics.

2.2.1 Static Calibration


This consists of determining or checking the systems scale.
This is done by using known standard quantities as inputs
(fixed or variable) and noting the outputs and finally
developing input output relationships. This procedure can
be used as a periodic check for a known instrument or for
scaling a new instrument. The output of the new instrument
can be a different measurable item from the one which is of
interest. As an example for measuring temperature variation
of resistance is registered. Theoretically interfering and
modifying inputs are held constant during calibration.

Calibration procedures involve a comparison of the particular


instrument with either (1) a primary standard (2) a
secondary standard (3) a known input source (weighing a
tank of water in the given interval of time to determine the
flow rate.
2.2.2 Readability
This term indicates the closeness with which the scale of the
instrument may be read. An instrument which sweeps
through 180o will have a higher readability than another
instrument which sweeps through 90o for the same range of
measurement.

2.2.3 Least Count


The smallest difference between two indications that can be
detected on the instrument scale
2.2.4 Sensitivity
Static sensitivity is defined as the ratio of the change in output
to the corresponding change in input under static or steadystate conditions ie. the slope of the calibration curve. The
sensitivity can be linear or non-linear (fig_chp2\fig2.1.pptx
). Sensitivity can be affected by interfering and modifying
inputs. The effects are indicated as zero and sensitivity drift
respectively as shown in fig_chp2\fig2.2.pptx .

Zero drift or bias describes the effect where the zero reading
of an instrument is modified by a change in the ambient
conditions. This causes a constant error that exists over the
full range of measurement of the instrument.
Sensitivity drift or scale factor drift defines the amount by
which an instruments sensitivity of measurement varies as
ambient conditions change. It is quantified by sensitivity
drift coefficient.
Example 2.1

example2.pptx

Both effects can be evaluated by running suitable calibration


tests.
If elements of a system having static sensitivities of K1, K2,
K3,, etc. are connected in series or cascaded as shown in
Fig. 2.3 then the overall system sensitivity K is given by
K = K1 x K2 x K3 x
provided that there is no alteration in the values of K1, K2,
K3,, etc. due to loading.
Sensitivity, gain, and magnification are the terms used to mean
the same.

Fig. 2.3 Overall system sensitivity


9

2.2.5. Linearity
It is normally desirable that the output reading of an
instrument is linearly proportional to the quantity being
measured. Based on this convenience linearity is defined as
the maximum deviation from a linear relationship between
input and output i.e. from a constant sensitivity (least
square fitting) and is usually expressed as a percentage of
full scale reading.
2.2.6 Threshold
If the instrument input is increased very gradually from zero,
there will be some minimum value below which no output
change can be detected.

10

This minimum value defines the threshold of the instrument.


Manufacturers specify it as absolute value or percentage of
full scale reading.
2.2.7 Resolution
This defines the smallest measurable input change that will
permit the detection of change in the output.
2.2.8 Hysteresis
This is an effect of producing different readings when the
measured quantity is approached from above or below. It
may be the result of mechanical friction, magnetic effects,
or thermal effects. 2nd law-irreversibility.

11

Curve B
Variable decreasing
Output
variable

Curve A-Variable increasing


Maximum
output hysteresis

Measured variable

Maximum input
hysteresis

Dead Space

Fig. 2. 4 Instrument characteristic with hysteresis


12

2.2.9 Accuracy
This is the term used to indicate the closeness with which the
indications of an instrument approach the true values of the
quantities measured. Inaccuracy is the extent to which
reading might be wrong and it is usually expressed as a
percentage of full scale reading. A 1% accuracy over a full
scale reading pressure of 100 kPa will be accurate within
1 kPa. If this instrument is used to measure 5 kPa, then the

possible error will be 20% !


Such plus or minus inaccuracies are also termed as
measurement uncertainties.

13

2.2.10 Precision
It indicates the ability of an instrument to reproduce a certain
reading with a given accuracy. If there is no reproducibility,
then the instrument is said to have a drift.
As an example for a true value of 100 V, measured values are
104, 103, 105, 100, 105.
The accuracy is 5 V.
Precision is maximum deviation from mean.
deviation = 1 V
Mean = 104 Max.
Precision is 1%.

14

It may be noted that the instrument could be calibrated


so that it could be used to dependably measure
voltages within 1 V. This conveys the message
that accuracy can be improved but not beyond the
precision of the instrument.

15

Low precision
Low accuracy

High precision
Low accuracy

High precision
High accuracy

Fig.2.5 Comparison of accuracy and precision


16

2.2.11 Errors
There are two types of errors in measurements, namely
systematic and random errors.
Systematic or fixed errors or bias
These types of errors will cause repeated readings to be in
error by the same amount. These are related to calibration
errors and they can be eliminated by correct calibration.
Accuracy is related to such type of errors
Random errors
These are caused by personal fluctuations, random
electronic fluctuations in the instruments, various

17

influences of friction, etc. They usually follow certain


statistical distribution.
Such errors are related to precision.
2.3 DYNAMIC CHARACTERISTICS
The dynamic characteristics of a measuring instrument
describe its behavior when input varies with time such that
the instrument response will have transient and steady state
parts.
The most widely used mathematical model for the study of
measurement-system dynamic response is the ordinary
linear differential equation with constant coefficients of the
form

18

n 1

d qo
d qo
dq o
a n n + a n 1 n 1 + ... + a1
+ a oq o
dt
dt
dt
m 1
m
d qi
d qi
dq i
= b m m + b m 1 m 1 + ... + b1
+ b oq i
dt
dt
dt
(nth order)
If we define the differential operator D d the above
dt
equation can be written as

(a n D + a n 1D
n

n 1

+ ... + a1D + a o )q o

= ( b m D + b m 1D
m

m 1

+ ... + b1D + bo )q i
19

The method of undetermined coefficients or Laplacetransform method can be used to get the solution.
Here the former will be used.
In this method the general solution is given by
qo = qocf + qopi
where
qocf = complementary function part of solution
qopi = particular-integral part of solution
The solution qocf is determined from the algebraic
characteristic equation

a n D + a n 1D
n

n 1

+ ... + a1D + a o = 0

(Refer to handout for solutions)

20

2.3.1 Zero-Order Instrument


The simplest possible special case of the general
dynamic equation is when all the as and bs other
than ao and bo are assumed to be zero. The
differential equation then degenerates into the
simple algebraic equation
aoqo = boqi
Any instrument that closely obeys the above equation
is defined to be a zero-order instrument. The two
constants can be combined to give

bo
q o = q i = Kq i
ao
21

where K=bo/ao = static sensitivity


The above equation shows that no matter how qi might
vary with time, the instrument output follows it
perfectly with no distortion or time lag of any sort.
Thus the zero-order instrument represents ideal or
perfect dynamic performance.
A practical example of a zero-order instrument is the
displacement measuring potentiometer. Referring to
Fig. 2.6(a) and fig_chp2\fig2.6.pptx linear
distribution of resistance along length L, the output
voltage eo can be written as
xi
e o = E b = Kq i
L
Measurement error em = Kqi - eo = o (ideal)
22

Fig.2. 6 (a) Zero-order instrument (linear response)


23

2.3.2 First Order Instruments


All as and bs except a1,ao,bo are taken as zero. The
resulting equation is

dq o
a1
+ a oq o = b oq i
dt
Division by ao reduces the No. of coefficients by one
as

bo
a1 dq o
qi
+ qo =
a o dt
ao

bo
K=
ao

Defining a1/ao
24

The above equation gives


(D + 1) qo = Kqi
is a time constant and always has dimension of time.
As an example of a first order instrument let us apply
this to a thermometer (liquid in a glass)
fig_chp2\fig2.7.pptx
The input is Ti(t) which varies with time and the
output is xo (thermometer liquid level change). If
the liquid level is at xo and using the following
definitions
Ttf=temperature of liquid in bulb (uniform), Ttf = 0 at
xo =0
25

Kex = differential expansion coefficient of


thermometer fluid and bulb glass, (m3/m3 oC)
Vb = Volume of bulb, (m3)
Ac = cross-sectional area of capillary tube, (m2)
Displaced volume and volumetric expansion of liquid
in the bulb are related as
xo Ac = Kex Vb Ttf
This will give

K ex VbTtf
xo =
Ac
To get the differential equation we will use the
conservation of energy over an infinitesimal

26

time dt for the thermometer bulb as follows:


Heat rate in Heat rate out = energy storage rate

dTtf
UA b (Ti Ttf ) 0 = Vb c
dt

U=overall heat transfer coefficient across bulb wall,


(W/m2)
Ab = heat transfer area of bulb wall, (m2)
= mass density of thermometer liquid (kg/m3)
c = specific heat of thermometer liquid (J/kg m2 )
The above equation can be rewritten as

dTtf
Vb c
+ UA bTtf = UA bTi
dt

27

Using the equation that relates Ttf and xo which is

dTtf
A c dx o
=
dt
K ex Vb dt

And substitution in the differential equation gives


A c dx o
x oAc
Vb c
+ UA b
= UA bTi
K ex Vb dt
K ex Vb

cA c dx o
K ex

UA b A c
+
x o = UA bTi
dt
K ex Vb

Defining

a1 =

cA c
K ex

UA b A c
ao =
K ex Vb

bo = UA b
28

This will give

dx o
a1
+ a o x o = boTi
dt
Again defining

a1 cVb
= =
(s )
a o UA b

bo K ex Vb
o
K=
=
( m / C)
Ac
Ao

results in the following ODE

dx o

+ x o = KTi
dt
( D + 1) x o = KTi

or
29

As can be seen clearly, the solution depends on the type of


input. Four types of inputs will be dealt with.

Step Input
Initially the system is in equilibrium with qi=qo=0 and at time
t=0+ the input quantity increases instantly by an amount qis.
This will make the initial condition qi=qis at t=0+.

The differential equation is


(D +1)qo = Kqis
The solutions are
qocf=Ce-t/ and qopf = Kqis
30

This gives the complete solution as


qo = Ce-t/ + Kqis
Applying the initial condition qo=0 at t=0
gives C= -Kqis and the solution becomes
qo = Kqis(1 e-t/)
This is shown in fig_chp2\fig2.8.pptx
Non-dimensionalizing gives

qo
= 1 e
Kq is

e-t/ = 0 makes it an ideal instrument (req. small )


31

The speed of response depends on the value of only


and is faster if is smaller.
Measurement error is given by

Kq is
qo
t /
t /
e m = q i = q is
(1 e ) = q ise
K
K
Non-dimensional form

em
t /
=e
q is

(decays with time)

The non-dimensional output and error are shown in


fig_chp2\fig2.9.pptx
32

There are some dynamic characteristics that are useful


in characterizing the speed of response of any
instrument. These are
Rise Time: time required to achieve a response of
90% of the step input.

qo / K
t /
= 0.9 = 1 e
q is
e

t /

= 0.1 t = 2.303

A response is usually assumed to be complete after


t=5 s since

1 e

t /

= 1 e = 0.993 = 99.3%

33

Settling time: time to reach and stay within a stated


plus-and-minus tolerance band value around its final
value. A 5% settling time is equivalent to t=3
Knowing now that fast response requires a small value
of , application to our thermometer example shows
that may be reduced by
1. Reducing , C, and Vb
2. Increasing U and Ab
For the thermometer fluid use small C product.
Reducing Vb decreases Ab (unwanted effect).
Reduced Vb , moreover reduces the sensitivity K.
Results in trade-off between speed and sensitivity.
34

Also as U is dependent on the flow situation (free or


forced convection), caution has to be taken with
respect to specification of the instrument.

Ramp Response
Initially the system is in equilibrium, with qi=qo=0.
qi = q o = 0
t0
qi = q is t
t0
The differential equation becomes

(D + 1)q o = Kq is t

Solution of complementary and particular function

q ocf = Ce

t /

and

q opf = Kq is ( t )

35

The complete solution will be

q o = Ce

t /

+ Kq is ( t )

Applying the initial condition will give

0 = C Kq is C = Kq is

This will give the final solution as

q o = Kq is (e

t /

+ t )

Define measurement error em by (qi-qo/K). Then

e m = q is t q is (e
= q ise

t /

t /

+ t )
q is

transient error(em,t) steady state error(em,ss)


36

Note that the unsteady state output and error die out
with increasing time.
The non-dimensionalized error is given by
t /

q ise
+ q is
em
t /
=
= 1 e
e m ,ss
q is
The response and the non-dimensionalized
measurement error are shown in
fig_chp2\fig2.10.pptx
The solution shows that there is a steady state error
given by em,ss and there is a fixed time lag of .
37

Frequency Response
Here the input is harmonic (sine or cosine function)
For a sinusoidal input
qi=Ai sin t
A graphical representation of a possible solution is
shown in fig_chp2\fig2.11.pptx
The response shows both the steady and unsteady
parts. Usually the transient part decays with time.
Some of the terms used are shown in the figure.

38

The differential equation will be


(D + 1)qo = Kqi =KAi sin t
The complementary function from previous solutions
is
t /

q ocf = Ce
q opf =

and

KAi
1 + ( )

sin(t + )

(Refer to handout for details or example2.pptx )


39

The complete solution will be

q o = q ocf + q opf = Ce

t /

KAi
1 + ( )

sin(t + )

For steady state ie. large t, the complementary function


decays
The phase shift angle is given by

= tan ( )
1

And the amplitude ratio becomes

qo
1
=
2
KAi
1 + ( )

40

It can also be shown that the time lag is t = /


The amplitude ratio and phase lag are shown in
fig_chp2\fig2.12.pptx
The steady state solution shows that for an ideal
frequency response, the phase angle must approach
zero. This will also make the amplitude ratio to be
one. This occurs if the product approaches zero.
This requires
a. For any , there will be some frequency input
below which measurement is accurate
b. For high , the instrument must have a sufficiently
small
example2.pptx
41

Impulse Response
The impulse (Fig.2.13 a) function of strength (area) A
is defined by the limiting process
Impulse function of strength A lim T0 p(t)

Fig. 2.13 (a) Impulse of strength A


42

Its time duration is infinitesimal. Its peak is infinitely


high and its area is A.
For pulse 0 < t < T, it is the same as a step input
(Fig.2.13b) where the input is qi = A/T = qis and the
differential equation to be solved will be
(D + 1)qo = Kqis = KA/T
And the complete solution

KA
t /
qo =
(1 e )
T
This solution is valid only up to time T. At this time
we have

q o @ t =T

KA
T /
=
(1 e )
T

43

44

For t > T, our differential equation will be


(D + 1)qo = Kqi = 0 (Fig.2.13c)
Which gives the solution as

q o = Ce

t /

And imposing the initial condition at t=T which gives


T /

KA
KA
(
1
e
)
T /
T /
C=
(1 e ) = Ce
T
Te T /
KA(1 e T / )e t /
qo =
Te T /

giving

45

46

As T is made shorter, the first part (t < T) of the


response becomes of negligible consequence
(Fig.2.13d), so that we can get an expression for qo
by taking the limit T0
T /

KA(1 e
lim T 0
T /
Te

) t / 0
=
e
0

In such a case LHospitals rule will be used.


Differentiate the numerator and denominator with
respect to T and take the limit. This will give the
approximate solution as ( fig_chp2\fig2.13e.pptx)

qo =

KA

t /
47

48

2.3.3 Second Order Instruments


It follows the equation
2

d qo
dq o
a 2 2 + a1
+ a oq o = b oq i
dt
dt
The essential parameters can be reduced to three, as
follows:
bo
static sensitivity
K

ao

ao
a2

undamped
natural
frequency,
rad/time

49

And
=

a1
damping ratio, dimensionless
2 a oa 2

With these definitions the equation becomes

D
2D
2 +
+ 1q o = Kq i
n
n

A good example of a second order instrument is the


force-measuring spring scale shown in
fig_chp2\fig2.14.pptx
Assuming frictional effect proportional to velocity we
will use the following values.
50

Ks = spring constant
B = damping coefficient (constant)
Considering xo = 0 when fi = 0, application of
Newtons second law yields

dx o
d2x o
fi B
K s x o = M 2 ( MD 2 + BD + K s ) x o = f i
dt
dt
Division by Ks will give

M 2 B
fi

D +
+ 1 x o =
Ks
Ks

Ks

Comparing with the general 2nd order equation


M
B
1
a2 =
, a1 =
, a o = 1, bo =
Ks
Ks
Ks

51

For the spring system this will give


1
K=
( m / N ), n =
Ks

Ks
2
B
( rad / s),
=
M
n K s

Bn
B
where =
=
2Ks 2 Ks M

This will allow the definition of critical damping as

Cc = 2 K s M

and damping ratio

B
=
Cc
52

With step input size qis, the differential equation


becomes

D 2 2D
2 +
+ 1q o = Kq is
n
n

Initial condition: at t=0 qo=0 and dqo/dt = 0


The characteristic equation for the complementary
function will depend on the roots of

D + 2n D + = 0
2

2
n

Depending on the value of (>1 overdamped


=1 critically damped <1 underdamped)

53

The complete solutions are given as follows:

+ 1 ( +
qo
e
=
2
Kq is
2 1
2

1
2

2 1
2

( 2 1 )n t

qo
n t
= (1 + n t )e + 1
Kq is
n t

2 1 )n t

+ 1 ( > 1)
( = 1)

qo
e
2
sin( 1 n t + ) + 1
=
2
Kq is
1

< 1)
54

And

= sin

The above solutions are plotted as functions of the


product nt in fig_chp2\fig2.15.pptx.
The following can be observed from the solution
n is a direct indication of speed of response.
Doubling n will halve the response time.
An increase in reduces oscillation, but also slows
the response. Many commercial instruments use
=0.6 to =0.7.
55

Ramp Response
The differential equation for this case is

D 2 2D
2 +
+ 1q o = Kq is t
n
n

dq

o
q
=
= 0 at t = 0
With initial conditions o
dt
The solutions are found to be :

Overdamped

2 2 1 2 2 1 (
qo
2q is
e
(1 +
= q is
K
n
4 2 1
+

2 + 1 2 1
2

4 1
2

( + 2 1 )n t

2 1 )n t

)
56

Critically damped

qo
2qis
= qis t
K
n

nt nt
e

+
1
1

Underdamped
n t

qo
2q is
e
2

sin( 1 n t +
= q is t
1
K
n 2 1 2

2 1
tan =
2
2 1

57

For steady state all the above will give

qo
2q is
= q is t
K
n
Steady state time lag can be shown to be 2/n.
Measurement error, em can be determined from
qo
qo
= q is t
em = qi
K
K
Steady state error will be
qo
q o 2q is
= q is t
=
ess = q i
K
K
n
fig_chp2\fig2.16.pptx and fig_chp2\fig2.17.pptx
58

respectively show ramp response and non-dimensional


ramp response error.
Frequency Response
For a cosine input, the equation to be solved will be

D
2D
2 +
+ 1q o = KFo cos t
n
n

The solutions can easily be determined (for example


for the underdamped case) as

qocf = e nt (C1 cos 1 2 nt + C2 sin 1 2 nt )


which decays with time

59

And the particular integral function for all the cases


will be

q opf

2 KFo
sin t
=
2 2 2
2
2
4 n + (n )
3
n

( ) KFo

cos
t
+ 2
2
2
4 + (n )
2
n

2
n
2 2
n

The above will be the steady state solution.

60

Trigonometric manipulation will give

qo / K
=
Fo

where

cos(t )

4 + 1
n n
2

tan =

=0.6 to 0.7 is the practical choice for instruments.


fig_chp2\fig2.18.pptx shows the amplitude ratio and
phase lag fig_chp2\fig2.19.pptx .
61

Impulse Response
For impulse strength of A solutions are found to be
Overdamped:

qo
1
(e( +
=
KAn 2 2 1
e

2 1 )n t

( 2 1 )n t

Critically damped

qo
= n te n t
KAn
62

Underdamped:

qo
1
n t
2
=
e
sin( 1 n t )
KAn
1 2
The results are shown graphically in
fig_chp2\fig2.20.pptx

63

2.4 IMPEDANCE MATCHING


The introduction of any measuring instrument into a
measured medium always results in the extraction
of some energy from the medium, thereby changing
the value of the measured quantity from its
undisturbed state thus making perfect
measurements theoretically impossible.
Looking at fig_chp2\fig2.21.pptx the effort to
measure the unknown voltage Eo, is made
theoretically impossible since the circuit is changed
as soon as the voltmeter with its resistance Rm is
connected.
64

Since the loading effect is going to change the output


variable, the concept of input impedance will be
used to characterize this effect.
Let qi1 be the input variable of primary interest as far
as the instrument is concerned. Let there be an
associated variable qi2 such that the product qi1 qi2
has the dimensions of power and represents the
instantaneous rate of energy withdrawal from the
device. Then the generalized input impedance Zgi is
defined as

q i1
Zgi
qi2
65

This gives the power, P, drained by the voltmeter as


2
q i1
P=
Zgi
which shows the requirement of large input
impedance to keep the power drain small.
For the voltmeter the input variable qi1=Em. An
associated variable will be qi2=im. This will give
Zgi=Em/im=Rm, the meter resistance.
For a general approach Thvenins theorem (without
proof) will be used. If the input is a general load
with input impedance Zl as shown in
fig_chp2\fig2.22.pptx, the open circuit voltage Eo
66

is the one that is measured when the load is not


connected and it is also possible to determine the
impedance ZAB. Thvenins theorem states : If the
load Zl is connected as shown in Fig.2. 22b, a
current il will flow. This current will be the same as
the current that flows in the fictitious equivalent
circuit of Fig. 2. 22c. Thus the network contains
only the output impedance ZAB and the single
voltage source Eo.
Applying Thvenins theorem to the voltmeter
connection of Fig.2.21a converted to Fig.2.21b

Rm
Eo
Em =
R ab + R m

67

If the voltmeter is to indicate the true value Eo, then


we must have Rm >> Rab.
For the generalized input, Zgi, and output impedance,
Zgo, and for electrical and nonelectrical systems

q i1m =

Z gi
Z go + Z gi

q i1u

1
q i1u
=
Zgo / Zgi + 1

Where qi1m=measured value of effort variable


qi1u=undisturbed value of effort variable
High value of Zgi is required.
Similar analysis for an ammeter shows the meter
resistance must be sufficiently low.

68

If the device is a power supply to the external load


with Rm
2

E
1 Rm
E Rm

P=
E o =
=
R m R m R ab + R m
R m R ab + R m
2
m

2
o

For maximizing power (constant Eo, Rab) with respect


to Rm (after differentiating) will require Rm=Rab.
Example 2.3
example2.pptx
69

Potrebbero piacerti anche