3

Basic Electrical Measurements

3.1 Electromechanical measuring devices current is delivered to the coil by two springs – these
springs are also used as the mechanisms generating
There are several advantages of traditional returning torque for the pointer.
electromechanical instruments: simplicity, reliability,
low price. The most important advantage is that the
majority of such instruments can work without any
additional power supply. Since people’s eyes are
sensitive to movement also this psycho-physiological
aspect of analogue indicating instruments (with moving
pointer) is appreciated.
On the other hand, there are several drawbacks
associated with electromechanical analogue indicating
instruments. First of all, they do not provide output
signal, thus there is a need for operator’s activity
during the measurement (at least for the reading of an
indicated value). Another drawback is that such
instruments generally use moving mechanical parts,
which are sensitive to shocks, aging or wearing out.
Relatively low price of moving pointer instruments FIGURE 3.1
today is not as advantageous as earlier, because on the The example of moving coil indicating instrument (1- moving coil, 2 –
market there are available also very cheap digital permanent magnet, 3 – axle, 4 – pointer, 5 – bearings, 6 – spring, 7 –
correction of zero).
measuring devices with virtual pointer.
Regrettably, it can be stated that most of the The moving coil is placed into the gap between the
electromechanical analogue instruments are rather of magnet poles and soft iron core, shaped in such a way
poor quality. In most cases these instruments are not as to produce uniform magnetic field. The movement
able to measure with uncertainty better than 0.5%. The of the coil is caused by the interaction between the
accuracy is also affected by so-called parallax error, in magnetic field of the magnet and the magnetic field
which the reading result depends on the position of the generated by the coil. The rotation of the coil (and the
user’s eye. The measurement is often invasive, because pointer attached to it) is due to the torque M, which
such mechanisms may need relatively large power depends on the flux density B of the magnet, on
consumption to cause the movement. Thus, dimensions d and l of the coil, on number of turns z of
electromechanical voltmeters exhibit insufficiently the coil and of course on the measured current I:
large resistance, while the resistance of
electromechanical ammeters is not sufficiently small. M  Bzdl  I (3.1)
There is no doubt that the future is for automatic,
computer supported measuring systems. But The angle of rotation  results from the balance
electromechanical instruments are still present in our between the torque and the returning torque of the
lives (for example the attempts to substitute such springs Mz = k  (k is the constant of the elasticity of
instruments in cars finished with not a success). the spring). Thus from the condition M = Mz we find
The moving coil instrument is the most popular that the rotation is
indicating electromechanical device. An example of
such an instrument is presented in Figure 3.1. Bzdl
A rectangular coil with the pointer fixed to its axle is  I  cI (3.2)
k
used as the moving part in such instruments. The conic
ends of axles are pressed against the bearings. The
 Handbook of Electrical Measurements 39

The angle of rotation is proportional to the measured where T is the time constant, T0 is the period of
current I, which is advantageous, because it means that oscillations of the moving element, b is the degree of
the scale is linear. The larger is the constant c in damping and P is the damping coefficient.
Equation (3.2) the more sensitive (thus better) is the
measuring device, because less current is required to
cause the movement of the coil. The best way to
improve the sensitivity of the device is to use large
magnetic flux density B. The increasing of the number
of turns or the dimensions of the coil is not very
effective, because at the same time the weight and the
resistance of the coils increase. Currently, it is possible
to manufacture the moving coil device with the power
consumption not larger than several W (and current
not larger than several A) for the full deflection of the FIGURE 3.3
pointer. The movement of the pointer after connection of the device to the
measured current.
.
Thus the character of the movement depends on the
a) b) ratio between the mass, the elasticity of the springs and
N S the damping. In the case of other instruments a special
air damper is used in order to obtain correct damping of
the movement. But in the case of a moving coil device
the aluminum frame of the coil can work as the damper
– the eddy currents induced in this frame interact with
the magnetic field of the magnet slowing down the
N velocity of the movements.
S

FIGURE 3.2
The moving coil mechanism: a) the symbol of instrument, b) the
principle of operation.

The elasticity of the spring plays important role

because it influences the character of the pointer
movement. It is convenient if this movement is with FIGURE 3.4
small oscillation (Figure 3.3). In the case of pure The enlargement of the length of the pointer – the light indicator.
inertial movement without an overshoot the observer is
not sure if the pointer reaches final position. It is The sensitivity of the device can be improvement by
important to obtain the oscillatory movement with a increasing of the length of the pointer. The best
short period and with reasonable damping of solution is to substitute the mechanical pointer by the
oscillation. Ideally, only one oscillation period should light indicator (Figure 3.4) – the length can be
be visible – the next one should be damped. additionally increased by multiple reflection of the light
The parameters of the movement depends on the beam. Sometimes also beari8ng are substituted by
mass m of the moving part and on the elasticity ribbons. This ribbons act as the current supplying wires
coefficient, k and also as the springing parts.
The moving coil device can be used directly as the
T0 m P microammeter without any additional elements (Fig.
T , T0  2 , b (3.3) 3.5a). If in series with the moving coil device an
1  b2 k 2 mk additional resistor Rd (series resistor) is connected, then
we obtain the millivoltmeter or voltmeter (Fig.3.5b)
(because the current I in the device is then proportional
40 Basic Electrical Measurements

to the voltage U). When the millivoltmeter is connected Thus the error caused by the change in temperature
in parallel with another resistor Rb, called a shunt depends on the ratio Rd /r. It is easy to calculate that if
resistor we obtain the ammeter (Fig.3.5c), because the millivoltmeter is designed for measurements with
voltage Ub is proportional to the measured current Ix uncertainty better than 0.5% then it is necessary to use
(the resistance of millivoltmeter is much larger than the resistors with values Rd = 7·r. This means
resistance of the shunt resistor Rb thus we can assume deterioration of the sensitivity of the millivoltmeter.
that Ub  Ix Rb). Let us consider a case of a moving coil device with
resistance 10  and nominal current 1 mA.
a) b) Theoretically, such device could be used to design a
millivoltmeter with a minimal range Unom=Ir = 10 mV.
I= Ix r r But if we are planning to design a millivoltmeter of the
I=Ux/(r+Rd) Rd
class of accuracy 0.5% it is necessary to use additional
resistance Rd = 70 , which limits the minimal range
Ux of such millivoltmeter to 80 mV. For the voltmeters, the
problem of temperature errors correction is usually
r easy to solve, because it is necessary to use the series
c) I Rd resistor. For example, in order to design a 10 V range
voltmeter with a device described above it is necessary
Ix Rb to connect a resistor of about 10 k, much larger than
is required for the temperature error correction.
Ub The ammeter instrument can be designed similarly to
I=Ub/(r+Rd) Ub = IxRb the millivoltmeter – by measuring voltage drop on the
shunt resistor Rb (Fig. 3.5c). For example, if we use the
moving coil device with the parameters described
FIGURE 3.5
The design of microammeter (a), voltmeter (b) and ammeter (c)
above and we would like to design an ammeter with a
range 1 A and the accuracy class 0.5% then it is
The temperature influences the flux density B of the necessary to use a shunt resistor which would result in
permanent magnet and the elasticity of the springs k. voltage drop larger than 80 mV (thus Rb = 80 m).
Fortunately, both of these influences act in opposite Of course as better is instrument (more sensitive) as
changes of the . Therefore, their influences are smaller shunt resistor is necessary. In the case of
negligible when the device is used as the voltmeter we require resistance as large as possible.
microammeter (Fig. 3.5a). Reversely is in the case of ammeter – in this case we
The case of the millivoltmeter (Fig. 3.5b), and also expect that the resistance should be as small as
indirectly of ammeter (Fig. 3.5c), is more complicated. possible.
The change of temperature causes change or the
I r Rd
resistance r of the coil (the changes of resistance of the
other resistors Rd and Rb are negligible, because they R1
are prepared from manganin – special temperature R2
independent alloy). Thus the current I, in the device Rn-1 I/V
changes with the temperature for fixed value of the
measured voltage U, according to relation Rn
I = U / (r + Rd). This change is significant, because + In I4 I3 I2 I1, U1 U2 U3 U4 U5
copper wire of the coil exhibits change of the resistance
of about 4%/10C. The temperature error of the FIGURE 3.6
millivoltmeter circuit presented in Fig. 3.5b we can The design of universal multi-range voltammeter
describe as follows:
Figure 3.6 presents the design of universal ammeter
U U and voltmeter (voltammeter) with selectable ranges. To

r  Rd r  r  Rd 1 r 4% obtain the multi-range ammeter the special design of
T    universal shunt resistor is very useful. The universal
U R r R
1 d 1 d shunt resistor is designed to obtain the same current I
r  Rd r r for various input currents. Thus it should be:
(3.4)
 Handbook of Electrical Measurements 41

I n  I Rn  I r  Rd   R1  Rn  

1 dL 2
I (3.7)
 (3.5)
I n 1  I Rn 1  I r  Rd   R1  Rn 1 
2k d
Although the deflection is a nonlinear function of
After simple calculations we obtain the condition of measured current it is possible to design the device (the
universal shunt resistor in form: component dL/d) in such a way that the expression
I n R1 (dL/d) I2 is close to linear. Because the response of
 (3.6) the device depends on the squared value of the current
I1 Rn it is possible to obtain the meter of rms value. Due to
the error caused by magnetic hysteresis (when DC
The moving coils measuring instruments are current is measured) these devices are used almost
usually manufactured as the panel meters (with class of exclusively for AC measurements.
uncertainty typically 1, 1.5 or 2.5%) and as the
laboratory meters with class of uncertainty typically
0.5%. Fig. 3.7 presents an example of analogue
indicating meter. a) b)

FIGURE 3.8
The moving iron meter: a) the symbol of such instrument, bc) the
principle of operation

FIGURE 3.7
The moving iron meter exhibits several advantages:
The examples of analogue panel meter (permission of EraGost) simplicity of the design – no need to supply the moving
element, easy change of the range by selecting the
The main disadvantage of the moving coil meters is number of the turns in the coil. The drawbacks of
that they indicate only DC values of the signals. In the moving iron devices are relatively large power
past, these devices were also used for measurements of consumption (0.1 – 1VA) and small sensitivity (in
AC values with the aid of rectifiers. Although such comparison with moving coil device). The smallest
devices measure the average value it is possible to scale obtainable range of moving iron milliammeter is
it in rms values, knowing that Xrms /XAV = 1.11. But this several mA. Also, the frequency bandwidth is limited to
dependence is valid only for pure sinusoidal signals. about 150 Hz.
Thus the rectifying AC measuring devices can be used
only for the measurements of poor accuracy.
For AC measurement can be used the moving iron a) b)
meter. The main advantage of moving iron meter is that
such instrument measures the rms value of the signal.
The design of moving iron meter is presented in Figure
3.8.
The measured current is connected to the stationary
coil and the magnetic field generated by this coil
interacts with the moving iron element. The iron part is
simply attracted by the coil acting as electromagnet
(Figure 3.8b). The angular deflection  depends on the
FIGURE 3.9
measured current I and the change of the inductance dL The electrodynamic meter: a) the symbol of such instrument, b) the
caused by this deflection: principle of operation.
42 Basic Electrical Measurements

The electrodynamic meters (Figure 3.9) operate person in order to account the energy used (i.e. there is
directly according SI definition of the ampere (see page no output signal which could be read automatically).
30) – attraction between current carrying wires. Moreover, these meters are electromechanical with
Therefore these meters were formerly used as the most quite complex system of error correction. Thus, in the
accurate indicating instrument. Today for accurate future the mechanical energy meters will be substituted
measurements these instruments are substituted by the by electronic ones. This process is slow due to the
digital devices. range of problems – it is necessary to replace millions
The electrodynamic device design is based on two of devices.
coils: a stationary and a moving one. The currents
flowing through these coils induce a force, which
causes rotation of the movable coil. The torque M 23 5 7
resulting from the interaction between two coils U
depends on currents: I1 in stationary coil, I2 in movable
one and the phase shift  between these currents: u
M  cI1I 2 cos  (3.8) I

Thus if one coil is connected to the current and the i

second to the voltage we can directly measure the h
power, because P=UIcos . Fig. 3.10 presents typical
connection of electrodynamic meter as the wattmeter.
The wattmeter has two pairs of terminals – the current
and the voltage terminals. In the voltage circuit there is
usually introduced a series resistor Rd. Thus the torque FIGURE 3.11
The principle of operation of the induction watt-hour meter.
can be calculated from the following equation:
Figure 3.11 presents the principle of operation of the
1
M c IU cos   k  P (3.9) induction watt-hour meter (Ferrari’s system). Two
Rd independent cores are supplied by the currents
proportional to the current and the voltage. These
+V currents generate magnetic fluxes i and u , which
flow through a rotating aluminum disc, in which eddy
I currents are induced.
+I -I The rotating torque Mr is due to the interaction
between the eddy currents and the fluxes. The torque
-V depends on the values of the currents in the cores and
V Rd Ro the phase angles between them

V/Rd M r  c I1I 2 sinI1 , I 2  (3.20)

The first current is proportional to the measured

FIGURE 3.10 current I1 = I, while the second current is proportional
The connection of the wattmeter for the measurement of electric to the voltage I2 = kU. Due to large inductivity of the
power.
voltage core the current I2 is shifted in phase by almost
The electrodynamic meters can be used for current 90 with respect to the supplied voltage and sin(I1,I2) 
and voltage measurement (in such cases both coils are cos(U,I) = cos . Thus the torque is dependent on the
connected in series). But the main drawback of measured power
electrodynamic devices is large power consumption
(several VA) and therefore pure sensitivity. Therefore M r  ck  IU cos  (3.21)
nowadays they are practically used only as wattmeters,
where several VA power consumption is negligible. Additionally, the induction meter is equipped with
The induction watt-hour meters (energy meters) are the braking magnet. Interaction between the magnetic
still present in our houses, although they exhibit serious field of the permanent magnet and the eddy currents
drawbacks. First of all the reading must be taken by a induced by this field causes a braking torque
 Handbook of Electrical Measurements 43

proportional to the angular speed of the disk. Under the Thus the condition of the balance Uout = 0 of the
influence of both torques the watt-hour meter acts as bridge circuit is
the asynchronous motor with the speed of the disk
proportional to the power supplied to the load. As a R1 R4  R2 R3 or R1 R4  R2 R3  0 (3.24)
result, the number of revolutions n in the time period t
(angular speed) is the measure of power The condition (3.24) is a universal condition for all
bridge circuits, and can be described as: the bridge
n circuit is in the balance state when the products of the
 KUI cos  (3.22) opposite impedances are the same.
t
The mechanical register counts the number of
revolutions and hence indicates the energy consumed R1 R2
by the load.
The principle of operation described above is
significantly simplified. In the real instruments the Uout
phase shift in voltage coil is not exactly 90 thus
additional phase correction winding is necessary. The
braking torque is caused not only by the magnet, but R3 R4
also by the two cores and additional magnetic shunt is
necessary for correction of this effect. Also additional
correction is necessary to compensate for the effect of U0 or Io
friction in the aluminum disc bearings. The total error
of the induction meter is various for various measured FIGURE 3.12
power and it is described by the error characteristic. All The Wheatstone bridge circuit.
corrections should be precisely set to ensure that the
characteristic of errors does not exceed required limits. The bridge circuits are used in two main modes of
The main weakness of the induction watt-hour meters operation: as balanced (null type) circuit (Warsza
is that these corrections, hence generally the 2005a) or as unbalanced (deflection type) circuit
performance of the meter, changes with the aging (Warsza 2005b). The null type bridge circuit is
process resulting in the risk that consumer or energy balanced by the setting of one or more impedances to
distributor are deceived. obtain the state Uout = 0 and then the measured value of
resistance Rx = R1 is determined from the equation
3.2 The bridge circuits R3
The bridge circuits were used as the most accurate Rx  R2 (3.25)
R4
devices for the measurements of resistance (and
generally impedance). Nowadays, the bridge circuits In the deflection type of bridge circuit we first
are not as important, because now, more effective balance the bridge circuit and then we can determine
direct methods of impedance measurement are the change of resistance from the output signal as
developed (based on the Ohm’s law). But the bridge
circuits are commonly used as the resistance Z x
(impedance) to voltage converters. U out  S U0 (3.26)
Two main bridge circuits: supplied by the voltage Zx
source or the current sources are presented in Figure
3.12. Thus the unbalanced bridge circuit operates as the
For the bridge circuits presented in Figure 3.12 the transducer of the change of impedance to the voltage (S
dependence of the output voltage Uout on the circuit is the sensitivity coefficient of the unbalanced bridge
parameters are as follows: circuit).
R1 R4  R2 R3
U out  U (3.23a)
 R1  R2  R3  R4  0 3.2a Balanced bridge circuits
In the balance mode one or more elements are
R1 R4  R2 R3
U out  I0 (3.23b) changed to obtain balance condition. For example in
R1  R2  R3  R4 the bridge presented in Figure 3.12 the balance is
obtained by changing resistance R2 (Figure 3.13a). But
44 Basic Electrical Measurements

such method is inconvenient because such change is

usually realized manually. Because bridge circuit is
composed from two voltage dividers (see Figure 2.9)
BFD
instead of changing resistance we can introduce change
of voltage in voltage divider. Figure 3.13b presents the
method of balancing the bridge by introducing
additional voltage drop on resistor R4”. This way we Bx Iout
can remote balance the bridge.

a) b)
Rx R2 Rx R2 FIGURE 3.15
Auto balanced bridge circuit by external feedback

The bridge circuit with four resistors (as in Figure

R'4 3.12) is known as Wheatstone bridge. Instead of
R3 R4 R3 R"4 resistors it is possible to connect impedance
I
Z  Z e j . If we supply the bridge by AC voltage the
balance condition 3.24 is now:
FIGURE 3.13
Balancing the bridge circuit by change of resistance (a) or by change

 Z1 Z 4  Z 2 Z 3
 (3.27)
1   4   2   3

of current (b)

If we are able to balance the bridge by current it is

Thus to obtain the balance of AC bridge circuit two
easy to introduce feedback to auto-balance the bridge,
conditions should be fulfilled: magnitude and phase
as it is presented in Figure 3.14.
(3.37). This means that in order to balance such bridge
circuit two independent adjusting elements are
Rx necessary. The process of balancing is therefore more
Iout complicated than in the DC bridge circuit.

a)
Lx L2
Ro Rx C R2
Rw
A NI I1 B
I2

FIGURE 3.14 R3 R4
Auto balanced bridge circuit by internal feedback
D B
Figure 3.15 presents other method of auto-balancing. b) R 4
D' I 2
In the first arm of bridge is connected sensor known as D"
I1jLx

magnetoresistor – resistor changing resistance with

magnetic field. If we connect the coil generating I 2R
3
magnetic field on opposite direction the sensor is now
I1jL2

detector of balance of the magnetic field. The feedback 0 I1(R2+r) C" C' I1Rx
current is changing to obtain balance of the bridge.
Some producer of magnetoresitive sensors design A
special planar coil to introduce feedback magnetic
field.
FIGURE 3.16
An example of AC bridge and simplified diagram illustrating the
process of balancing
 Handbook of Electrical Measurements 45

Let us consider the vector diagram presented in Fig. (the switch of null indicator in position 1), and then the
3.16b. The bridge is balanced by the successive bridge Z1Z2Z3Z4 (the switch of null indicator in position
approximation. Assume that for balancing we use the 2). (Sometimes it can be necessary to balance both
elements R2 and R3 and the unbalanced voltage is bridges many times – approaching the equilibrium state
represented by the line C’D’. By changing the R3 we in a stepwise manner). In the state of balance of both
move the point D’ to position D”. Note that it is not bridges the potentials of points a,b and e are the same
possible to obtain the zero value of the C – D distance and equal to the potential of earth. Therefore the
because in this step of balancing the C’ – D” distance capacitances Cae and Cbe do not influence the
is the local minimum of the output voltage. If we now distribution of currents. The capacitances Cde and Cce
change the R2 value we move the point C’ to position are connected to the Wagner elements and also do not
C” (this time the distance C” – D’’ is the local influence the balance condition of the main bridge
minimum of the output voltage). We can see that to circuit.
obtain the balance it is necessary to perform many steps
of approximation. a) Cx b)
C2
By appropriate design of bridge circuit it is possible R2
to improve the balancing process – even down to two Rx
Rx R2
steps. It is also possible to use two null indicating
devices with a 90 phase shift between them, which NI NI R4
enable practically mutually independent balancing of
both components. But generally the time of
measurement using the AC bridge circuit is limited due R3 R4 R3
to the complex problem of searching for the balanced C4
state conditions.
U0 U0
c
FIGURE 3.18
Z1 Cce The Wien bridge circuit (a) and Maxwell-Wien bridge circuit (b)
Z5
Z2
Cae A huge number of various AC bridge circuits were
e designed and developed: Maxwell, Wien, Schering,
U0 a b Hay, Owen, Anderson, de Sauty, Heaviside etc.
Cbe
Moreover, all these bridges exist in various mutations
Z3 Z4 and modifications. [Hague 1971, ].Historically the
Cde Z6 oldest and most known are the Wien bridge (Fig. 3.18a)
and Maxwell bridge (Fig. 3.18b) circuits.
d The conditions of the balance state of Wien bridge
2
NI circuit are as follows:
1

C2 R4 R (1   C2 R2 ) 2 2 2
Cx  , Rx  3
FIGURE 3.17 R3 (1   C2 R2 )
2 2 2
 2 R2 R4C22
The methods of reduction of the influence of parasitic capacitances by
including the Wagner earth additional elements
(3.28)

In the case of the AC bridge circuits another The conditions of the balance are frequency
problem appears – it is difficult to eliminate influence dependent. Therefore the Wien bridge is rather seldom
of the stray and to earth capacitances (Figure 3.17). For used for capacitance measurement, but it is frequently
that reason, it is necessary to shield all the elements in used as the frequency-dependent part of the oscillator,
the AC bridge circuits. Shielding does not eliminate the according to the dependence:
capacitive coupling but enables investigators to
1
establish their level during the balancing. More 2  (3.29)
effective is to use the Wagner earth (Wagner ground) RxC x R2C2
with additional elements Z5, Z6 connected as presented
in Fig. 3.17. The conditions of the balance state of the Maxwel-
The bridge circuit with the Wagner elements consists Wien bridge (called also often as the Maxwell bridge)
of two bridges. First, the bridge Z1Z3Z5Z6 is balanced are as follows:
46 Basic Electrical Measurements

R additional resistor r is used. This resistor is connected

Lx  R2 R3C4 , Rx  R2 3 (3.30) to Lx or L2 element (this connection is chosen
R4
experimentally – only in one position is possible to
balance the bridge).
a) b) The inductance bridge circuit is useful for
Cx C2 Cx
R2 measurements of inductance LxRx as well the Q factor
Rx
R2
Rx
Lx/Rx. It is also possible to measure the mutual
inductance Mx. For determination of the Mx value the
NI NI R4 measurements are performed two times – with the coils
connected in the same directions of the flux L’ and with
R3 R4 C3 the coils connected in opposite directions L”
C4
L  L1  L2  2M and L  L1  L2  2M (3.35)
U0
Then, the mutual inductance can be calculated as
U0
L   L 
FIGURE 3.19 M (3.36)
Two examples of the bridge circuits for capacitance measurements: de 4
Sauty-Wien bridge (a) and Schering bridge (b)
Zx
Fig. 3.19a presents the AC bridge circuit for
capacitance measurements (the de Sauty-Wien bridge).
The conditions of the balance state can be described as: n1 m1

R4 R
C x  C2 , R x  R2 3 (3.31)
R3 R4 n2 m2
Zw
The tgx = CxRx can be calculated as:

tg x  C2 R2 (3.32)

FIGURE 3.20
An example of the transformer bridge
Fig. 3.29b presents special kind of the bridge circuit
– the Schering bridge designed for high voltage and Figure 3.20 presents the example of the transformer
cable testing. The main part of supply high voltage is bridge circuit with two transformers. The output
on the capacitances Cx and C3, and adjustable elements transformer acts in this circuit as the current
R4, C4 are additionally grounded. The measured comparator – the null indicator points to zero, when the
parameters can be determined from the equations resultant flux in the transformer is also equal to zero.
The condition of the balance of this circuit is
R4 C4
C x  C3 ; R x  R2 , tg x  C4 R4 (3.33)
R2 C3 Z x n1 m1
 (3.36)
Z w n 2 m2
Also simple inductance bridge circuit presented in
Figure 3.16 can be used for inductance measurement. Thus the state of balance can be obtained not by
Assuming that the impedances of the arms are as changing the values of impedance but by change of
follows: Zx=Rx+jLx, Z2=R2+jL2, Z3=R3, Z4=R4 after number of turns. This is very convenient, because the
simple calculations we obtain the balance conditions in number of turns can be precisely adjusted. Especially in
form the case of the digital bridge circuits it is much easier to
connect the windings than to change the resistors or
R3 R
L x  L2 , R x  R2 3  r (3.34) capacitors. Figure 3.21 presents the example of
R4 R4 transformer bridge designed for capacity measurement.
To obtain the balance of this bridge circuit the Q The conditions of the balance of the circuit presented
factor (Q=L/R) of the inductances measured Lx and in Fig. 3.21 are as follows:
standard one L2 should be the same. For that reason, an
 Handbook of Electrical Measurements 47

n2 n4 nn n5
C x  Cw ; R x  Rw 1 3 ; tg x 
n1 n3 n 2 n5 n 4Rw C w R3 R4
(3.37) NI
Cx
I1

Rx R'3 R'4
n1 n3
Cw r
I0 Rx=R1 R2
n2 I2C n4

I2R n5 FIGURE 3.22

Rw
The methods of reduction of the influence of the connecting wires in
the Kelvin bridge circuit
FIGURE 3.21
An example of the transformer bridge circuit designed for capacitance First of all, the resistance of connection wire r
measurement should be small – therefore such wire is prepared as a
short and large diameter wire. The second term in the
The transformer bridge circuits exhibit several equation (3.38) as negligible if the following condition
important advantages in comparison with impedance is fulfilled
bridge circuits. As was mentioned earlier, the balancing
is possible by the change of the number of turns. In R3 R4
transformer bridges the parasitic capacitances shunt the R3 R4'  R3' R4 or  (3.39)
R3' R4'
transformer turns and practically do not influence the
conditions of the balance. Also the sensitivity of the
The condition (3.39) is relatively easy to achieve by
transformer bridges is significantly better than in the
mechanical coupling of the resistors R3 /R3’ and R4/R4’.
case of impedance bridges. In order to make use of
In such case, the condition for balance of the Kelvin
these advantages it is necessary to construct the
bridge is the same as for the Wheatstone bridge. The
transformers very precisely, with minimal stray fields.
Kelvin bridge enables measurement of the resistances
Therefore the transformer bridges are usually more
in the range 0.0001  – 10 .
expensive than classic circuits without transformer
coupling.
Returning to the Wheatstone bridge circuit it should 3.2b Unbalanced bridge circuits
be noted that this bridge exhibits limitations when very The unbalanced bridge circuits are used as the
small resistances are measured. In the case of the transducers converting the change of the resistance
measurements of very small (less than 1) the result (and generally impedance) into the output voltage:
can be influenced by the contact resistances,
thermoelectric voltages and most of all the resistances r  Rx
U out  SU o  SU o (3.40)
of the wires connecting the resistance to the bridge. Rxo
The influence of the thermoelectric voltages can be
reduced by performing the measurement procedure in where S is the sensitivity of the transducer and  is
two steps – for positive and negative polarization of the the relative change of the resistance
supply voltage, and then by calculation of the average
value from these two measurements.   Rx 
For very small resistance, very useful is the Rx  Rxo   Rx  Rxo  1    Rxo 1    (3.41)
modification of the Wheatstone bridge in the form  Rxo 
presented in Fig. 3.22 (the Kelvin bridge). The
condition of the balance for this bridge is as follows where Rxo is usually the resistance in the balance
state and  = Rx/Rxo.
R3 R3 R4'  R3' R4 Of course described in previous auto-balanced circuit
Rx  R2 r (3.38)
R4 R4  R3'  R4'  Rp  with feedback also operates as transducer of relative
change of resistance:
48 Basic Electrical Measurements

 Rx We can see that these circuits are nonlinear. The

I out  S  S (3.42) nonlinearity depends on the design of the bridge – for
Rxo
circuit b) it does not depend on the n value, but in the
circuit a) it depend on the m value. Figure 3.24 presents
There are a lot of sensors where the output signal is
the example of the dependences Uout=f(Rx/Rxo).
proportional to Z/Z or R/R, for example the
The nonlinearity of the bridge transducer is not
temperature sensor RT=RT0(1+T). The unbalanced
always a drawback – in some circumstances this bridge
bridge circuits are usually designed with symmetry in
nonlinearity can be used to correct the nonlinearity of
respect to the output diagonal (Fig. 3.23a) or to the
the sensor. Let us consider the example presented in
supply diagonal (Fig. 3.23b).
Figure 3.25. We use a thermistor sensor with very
nonlinear characteristic R=f(T) in order to measure the
a) b) temperature. If the bridge characteristic would be
Rx mRxo Rx Rxo linear, then the resultant characteristic of the transducer
Uout = f(T) would also be nonlinear – curve 1 in Figure
3.25. By appropriate choice of the bridge configuration
Uout Uout (bridge nonlinearity), in our case by applying m=0.3
we obtain almost linear processing of the temperature
into the output voltage – curve 3 in Figure 3.25.
I0 Rxo mRxo I0 nRxo nRxo
Uout/Uin
U0 U0 R[k] 1.0

80
0.8
FIGURE 3.23 1 2
Two kinds of symmetry of the unbalanced bridge circuit
0.6
3
Substituting the relation (3.41) into the equations
(3.23) after simple calculations we can derive the 0.4
dependencies of the transfer characteristics of
unbalanced bridge circuits: 0.2
for the circuit a)
0 25 T [oC] 10 20 30 40 T [oC]
m
U out / U 0   (3.43)
1  m2  1  m FIGURE 3.25
The transfer characteristic of the typical thermistor sensor and the
resultant characteristics of the bridge circuit with thermistor sensor (1
and for the circuit b) – calculated under assumption that the bridge circuit is linear; 2 –
1 calculated for m = 1; 3 – calculated for m = 0.3).
U out / U 0   (3.44)
2 2   
R R2
R Uout=U0 2R 1+
Uout/Uin [%] Uout=-U0 2R R1
0.2 m=0.
5
U0
15 m= U0

0 m=1 R R
R R
-15 Uout
- +
-
-30 + +-
R2
 [%] R R+ R+
R
-40 0 40 R R R1
FIGURE 3.24
The example of the transfer characteristics of unbalanced bridge FIGURE 3.26
circuit Two examples of the bridge circuit with feedback (Kester 1999)
 Handbook of Electrical Measurements 49

There are various methods of linearization of the It is very convenient to use two differential sensors
unbalanced bridge circuit. The best is applying a instead of just one sensor. In the differential sensors the
feedback because in such case only small linear part of changes of the resistances are in the opposite direction:
transfer characteristic is used (see Figure 2.21). Indeed
the transfer characteristic of the circuit presented in Rx1  R xo 1   
 (3.47)
Rx 2  Rxo 1   
Figure 3.14 is practically linear. Figure 3.26 presents
similar circuit with feedback [Kester 1999].
Also in this case we can connect the sensors in two
Rx=R0(1+)
kinds of symmetry as it is shown in Figure 3.28.
Substituting the relation (3.49) into the equations
U'out
x S (3.23) after simple calculations we can derive the
y
dependencies of the transfer characteristics of
Uout
unbalanced bridge circuits:
for the circuit a)

2m
U out / U 0   (3.48)
1  m2   2
FIGURE 3.27
The linearization of the bridge circuit by applying the multiplier
device
and for the circuit b)
Other method of linearization is applying of the
multiplier (Figure 3.27) [Tran 1987]. Taking into 1
account Eq. 3.42 we can assume that the change of U out / U 0   (3.49)
2
output signal of the bridge circuit supplied by the
voltage U = 0.5 Uo is: Thus the bridge circuit with differential sensors in
symmetry b) is linear.

U out  U (3.45) Let us consider the sensitivity of the unbalanced
2    bridge circuit. Neglecting the nonlinearity (calculating
the S factor for  = 0) from Equations (3.43) or (3.48)
After connection of the multiplier circuit as it is we obtain:
presented in Fig. 3.27 the output voltage is: for the single sensor:
 m
U U S (3.50)
 2   1  m2
U out 1
  U out
U out  out out 
K 1  U out 1   2
2 and for two differential sensors:
(3.46)
2m
Thus the transfer characteristic of the bridge circuit S (3.51)
with multiplier is linear. 1  m2
a) b) The bridge circuit with differential sensors is two
Rx1 Rx1 times more sensitive than the bridge with one sensor. If
mRxo Rx2
the bridge circuit is an AC bridge then the sensitivity is
a complex value and:
Uout Uout
Z2 Z 2 e j 2 Z 2 j  2 1 
m   e  m e j (3.52)
I0 Rx2 mRxo I0 nRxo Z1 j Z1
nRxo Z1 e 1
U0 U0
Thus the sensitivity S for the differential sensors is

FIGURE 3.28 2m
Two kinds of symmetry of the unbalanced bridge circuit with S (3.53)
differential sensor. 1  2 m cos   m 2
50 Basic Electrical Measurements

measuring system [Pallas-Areny et al 2001]. It can be

S simply an amplifier, but sometimes it can fulfill other
functions as: linearization of the sensor, error
reduction, analog - to digital conversion, saving in
memory and even interfacing to net or computer.
The conditioning circuit is indispensable in the case
100 of parametric sensors – sensors where the measured
value causes change of parameter: resistance, capacity
 or inductance. We cannot send these output and we
1 should convert it into signal – voltage or current. This
is a function of condition circuit.
0.01
-180o +180o a) b)
m
Io
Rx
FIGURE 3.29
The dependence of the sensitivity S of the bridge circuit on the circuit
Rx Uout Uo
configuration. Rs Uout
Fig. 3.29 presents a graphical representation of the
dependence (3.53). From this figure we can conclude Uout
that: c)
- the sensitivity is largest when the ratio m is equal to 1,
- the sensitivity can be larger, when the phase
difference between impedances Z1 and Z2 is larger.
Uouto
The general dependence Uout=f(R/R) of the Rx
unbalanced bridge circuit with four sensors is Rxo

1   R1  R2  R3  R4  FIGURE 3.30
U out      U o (3.54) The typical converters of the resistance into the voltage: voltage drop
4  R1 R2 R3 R4  (a), voltage divider (b) and their transfer characteristic (c)

Thus for one sensor the sensitivity is S=1/4, for two Figure 3.30 presents two methods of conversion of
sensors it is S=1/2, while for four sensors we obtain the resistance to the voltage. The first one seems to be
four times larger sensitivity in comparison with the the most obvious – it utilizes the Ohm’s law – the
one-sensor case resistance is supplied by the stabilized current Io and
the voltage drop Uout is proportional to the sensed
U out   U o (3.55) resistance

Unbalanced bridge circuit as the converter U out  I o Rx (3.56)

Uout=f(R/R) exhibits two important advantages:
- the zero component is removed by balancing the Thus we have linear conversion of the resistance into
bridge and we convert only signal; proportional to the voltage signal. But the dependence (3.56) is valid only
change of resistance R/R. It is important because often if the resistance connected to the output of transducer is
we detect only small change of the large resistance; infinitively large. Consider finite resistance of the
- according to Equation (3.54) in unbalanced bridge we output load Rout . In such case we obtain a nonlinear
can realize the differential operation - rejection of characteristic Uout = f(Rx) and this nonlinearity depends
common component, for example temperature zero on the ratio Rx/Rout. The Ohm’s law is also used in the
drift (see Figure 2.31). converter presented in Figure 3.30b. The measured
resistance is connected in the circuit of the voltage
3.3 The conditioning circuits divider supplied by the voltage source Uo. The output
Often to the sensor is connected the conditioning signal Uout is described by the equation
circuit mediated between sensor and the rest of the
 Handbook of Electrical Measurements 51

U out  U o
Rs
 Uo
1
(3.57) U out 
1
 x  T  T  T  T   1  x (3.61)
Rs  Rx R 4 4
1 x
Rs It is much better to use the differential type of the
sensor (1 = +x, 2 = - x and R1 = Rxo(1+), R2 = Rxo
The conversion is nonlinear, which is not always a
drawback, because in certain cases it can be used for (1-)) connected to the adjacent arms of the bridge
linearization of a non-linear sensor. The main circuit. In this case we obtain elimination of
disadvantage of the circuits presented in Figure 3.30 is interference effects with two times larger output signal.
And of course the best case is to use four differentially
that the dependence Uout = f(Rx) does not start from
connected sensors, because in such circuits we obtain
zero because the resistance of the sensor usually also
four times larger output signal.
does not start from the zero value but from the certain
Rx0 value a) b)
Iw
 Rx  R1=Rx
  Rxo 1    (3.58)
R2
Rx0+Rx
Rx  Rxo  Rx  Rxo 1  
 Rxo  Uout +
-
R3 R4 Rx0-Rx
Similarly the output signal of the converters
presented on the Figure 3.30 includes the constant
component Uouto, because Uo

U out  U outo 1    (3.59) c) Uout

This offset component is disadvantageous, because

more convenient is the case when the output signal of
the transducer is zero for starting point of the range of
the sensor. If this condition is fulfilled then we can Rx0 Rx
connect the typical voltmeter as the measuring
instrument. Moreover, large offset component can FIGURE 3.31
cause the saturation of the amplifier (if any amplifier is Converters of the resistance into the voltage with elimination of the
offset component: unbalanced bridge (a), with differential amplifier
used). (b) and their transfer characteristic (c)
Therefore better is to use the circuit with common
mode rejection – Figure 3.31. The most frequently is The unbalanced bridge circuit usually exhibits non-
used the unbalanced bridge circuit described in linear transfer characteristic. But as it was described in
previous Section (deflection type bridge). If this bridge previous Section it is possible to correct non-linearity –
circuit is in the balance state for the starting point of the the best recommended method is to use feedback.
sensor resistance then the output signal is offset-free, Similar performances as the unbalanced bridge
because circuits exhibits the circuit with differential amplifier
(Figure 3.31b). The differential amplifier converts the
Rx
U out  S  S  (3.60) difference of the input signals (Uout = Ku (U1-U2)). Thus
Rxo if one of the input resistors is active (a measuring
sensor) and the second one is the same passive resistor
The deflection type bridge circuit exhibits several we obtain elimination of the offset voltage and also
other important advantages. First of all such a circuit is elimination of the interferences. Such principle is used
immune to the variation of the external interferences, in Anderson loop – described in Chapter 2.3.
for example changes of the ambient temperature. If all If we use four active sensors in a bridge circuit we do
resistors in the bridge circuit are the same and only one not have free resistor to balance the bridge.
of them is the sensor (1 = x), then with the change of Theoretically, we can connect parallel balancing
the ambient temperature all resistors change their resistor to one of the arms. But in this case one of the
resistances: 2 = 3 = 4 = T while 1 = x+T. Thus resistors exhibits different performances than the other
according to Eq. (3.54) T components are eliminated three resistors, which can cause incomplete elimination
and only useful x component remains at the output of of the external influences, for example temperature
the bridge error. Figure 3.32 present two examples of the methods
enabling to balance such bridge circuits.
52 Basic Electrical Measurements

much easier than voltage and frequency output we can

consider as simple and cheap digital output.
For to frequency conversion classical oscillators can
be used. Figure 3.33 presents two such circuits –
+ Hartley and Colpitts oscillators. In these circuits, the
- nonlinear dependence of the frequency on the measured
parameter is inconvenient, because usually frequency f
is f 1/X, where X is C or L.

FIGURE 3.32 b)
Methods of balancing of four sensor bridges
a)
C R C
The methods of conversion of the resistance to the - R -
voltage signal described above are also suitable for the
conversion of the capacitance, inductance or generally + +
impedance. In such case the bridge circuit should be
supplied by the AC voltage. The Wheatstone bridge R2 R
C
can be substituted by the special AC bridge circuit R1
(Maxwell, Wien or other). But in this case we have to
eliminate parasitic capacitances and inductances. It is
also necessary to consider the influence of the cable’s FIGURE 3.34
capacitance. The AC circuits require balancing of two The conversion of the capacitance C or resistance R to the AC signal
components – magnitude and phase. with frequency dependent on the measured parameter: a) multivibrator
oscillator, b) Wien bridge oscillator
+
a)
Out It also is possible to adapt the oscillator utilizing the
operational amplifier presented in Figure 3.35a. The
frequency depends on the capacitance C (or resistance
C R) as follows

1
f  (3.62)
+ 2 RC ln1  2 R1 / R2 
b)
Out
Figure 3.34a present also RC oscillator with classical
Wien bridge with the resonance frequency:
L

1
f  (3.63)
2 RC
We can also convert measured value into voltage and
use one of voltage to frequency converters. Of course
FIGURE 3.33
The conversion of the capacitance C or inductance L to the AC signal such converters are ready to use in the case of voltage
with frequency dependent on the measured parameter: a) Hartley output sensors.
oscillator, b) Colpitts oscillator Figure 3.35 presents a typical design of a V/f
converter with an integrator circuit charging a capacitor
Even if we balance both components of the
C at a rate proportional to the amplitude of input
unbalanced AC bridge circuit we do not solve all
voltage Uin. Each time when the output voltage of
problems. The output signal contains two components
integrator circuit reaches certain value equal to the
– one in phase with the supply voltage, and the other
reference voltage Uref the comparator switches into a
one shifted by 90 degrees. The shifted component can
reset mode and discharges the capacitor (because diode
be effectively eliminated by phase-sensitive rectifier
D change its biasing). The frequency of the output
(more details will be given in Section 3.6).
signal depends on the amplitude of input voltage
Especially in the case of capacitance or inductance
sensors instead of voltage converters better is to use U in
converters to frequency. Frequency we can transmit f  (3.64)
RCU ref
 Handbook of Electrical Measurements 53

R
C
-
-
IN
+ Comp OUT
+ Uout
Uref DC

Uin1
Uref

Uout
FIGURE 3.37
Uin2<Uin1 An example of the temperature VCO device (voltage controlled
Uref oscillator)

Uout 3.4 The main sensors of physical values

Sometimes we can meet mismatch in usage of such
FIGURE 3.35
terms as: sensor, transducer and detector. According to
The voltage to frequency converter [Tran Tien Lang 1987]
VIM the sensor is an element of a measuring system
Figure 3.36 presents the hybrid voltage to frequency that is directly affected by a phenomenon, body, or
converter of Analog Devices – model AD-537. This substance carrying a quantity to be measured. The
converter AD537 enables the conversion of the input transducer is a device, used in measurement, which
voltage to the frequency up to 100 kHz with provides an output quantity having a specified relation
nonlinearity error less than 0.05%. The conversion to the input quantity. Thus the sensor is some kind of
factor K=Uin /f can be set by connecting appropriate transducer. But a new generation of the sensors, known
external R and C elements to the device. This figure as intelligent sensors, has included sometimes quite
presents the example of application of this transducer complex electronic circuit - operating as transducers.
to the temperature conversion with the conversion The detector is some kind of sensor that indicates the
factor 10 Hz/C (for chromel-constantan thermocouple presence of a phenomenon, body, or substance when a
sensor). threshold value of an associated quantity is exceeded.
Thus the sensor is a first element of the measuring
system directly contacting with investigated value.
The sensors can be used to directly measurement of
7 6 5 4 3 2 1
investigated physical value. But often they can be used
Tx - AD537
Uref
+ as intermediate element. For example the temperature
sensor can be used to measure of temperature, but also
Tref to measure humidity in psychrometer device, air flow
I/f driver in anemometer device or CO2 contents in
10Hz/ oC thermoconductive analyzer. The Hall sensor of
8 9 10 11 12 13 14 magnetic field is more often used to measure
+5V
mechanical values and electrical current than to
measure the magnetic field intensity.
Recently a huge number of various sensors have
FIGURE 3.36
An example of the temperature transducer utilizing AD537 voltage-to- been developed [Fraden 2003]. In this Sections only
frequency converter of Analog Devices eight selected, most important primary sensors are
described to present typical problems of their
In measuring systems very often is used device applications and signal conditioning.
known as voltage controlled oscillator VCO presented
in Figure 3.38. For control the frequency the varicap 3.4.a. Resistance temperature detector - RTD
diode DC can be used. Varicap is a diode with the One of the most frequently used sensors is the
capacity depending on the voltage. temperature sensor. Various thermal effects can be
used to detect the temperature – thermal expansion,
54 Basic Electrical Measurements

thermoelectric effect, thermoresistive effect, thermal operating temperature (-270 – 850 C), , immunity to
radiation, change in quartz frequency, change of corrosion and very stable (it is estimated that it is better
semiconductor junction properties etc. The change of than 0.05C per year). Figure 3.39 presents typical
resistance is widely used to measure the temperature designs of platinum sensors of temperature.
because sensor is relatively simply and in case of Usually thin platinum wire is wound on ceramic or
platinum sensor with high accuracy. insulated metal core, often in bifilar mode to avoid
inductivity (see Figure 2.64). Better accuracy exhibits
R [ ] design with helical wire inserted in the bores – this way
PTC
the influence of thermal stress is eliminated (Figure
3.39b). For cheaper design is also used the thin film
technology (Figure 3.39c).
RTD
r
100 
NTC R2
Rx=R1

r
0Co temperature o
80 C
r R3 R4
FIGURE 3.38
The example of transfer characteristics of the main thermoresistive
sensors (after Fraden 2003)

Figure 3.38 presents transfer characteristics of the U0

main thermoresistive sensors: metal platinum resistor
RTD, thermistor NTC and thermistor with positive FIGURE 3.40
Three wire connection of the sensor
temperature coefficient PTC. Thermistors have much
larger sensitivity but they are rather nonlinear and with The platinum sensor is usually connected to the
modest accuracy. The metal thermoresistors are close bridge circuit (see Figure 3.12). When the sensor is
to linear and with high accuracy (resistance of platinum connected to the bridge circuit with relative long wires,
sensor are described by international standard IEC the temperature changes of the resistance of these wires
751). can cause significant error. In such case three wires
connections is recommended (Figure 3.40).
a) b) c) If all three wires exhibit the same resistance (the
same length) we can write that:

Rx  r R4  R2 R3  r  (3.33) (3.65)

and

Rx R4  R2 R3  r R2  R4  (3.66)

If additionally the condition R2 = R4 is fulfilled then

the influence of the resistance r of the connecting wire
is negligible.
Better accuracy is possible to obtain in four wires
FIGURE 3.39 connection (Figure 3.41). In this connection there are
Typical design of platinum temperature sensors
two pairs of terminals – for current delivery and for
Among various metal platinum is the best choice voltage sensing. The current wires are outside the
because it exhibits: very linear dependence on the measuring circuit therefore their resistance ri does not
temperature, relative large temperature coefficient ( = influence the measuring result (especially if we supply
the sensor with current stable source). The resistance of
0.385 %/ C), high resistivity and good plasticity
sensing wires rv also does not influence the result
enabling preparation of thin wire, wide range of
because voltmeter has usually very high resistance thus
 Handbook of Electrical Measurements 55

the current is very small. Generally the four wires R T   Ro 1  AT  BT 2  (3.69)
connection is obligatory for very small (less than 1 )
measured resistance.
for T > 0C;

where: Ro is the resistance in 0C, A = 3.908310-3

C-1, B = -5.77510-7 C-2; C = -4.18310-12 C-4.

RT V It can be easy to calculate that for temperature 600

C the error of nonlinearity is about 6%. Knowing
exact dependence of R(T) it is possible to introduce the
ri rV correction. But usually it is necessary to perform
certain mathematical operation. One of such solution is
described in Application Note AN-709 of Analog
FIGURE 3.41 Device [King et al 2004]. The correction is performed
Four wire connection of the sensor in circuit presented in Figure 3.43.
There are many other measuring circuits for
thermoresistive sensors - Figure 3.42 presents simple
circuit with operational amplifier. If RT = R the output RT ADC C
signal is equal to zero I other cases is:

Vo
VT    RT  R  (3.67) Rref
2R

RT
FIGURE 3.43
R The measuring circuit with microcontroller for linearization [King et
Vo al 2004]
VT
In the circuit presented in Figure 3.43 the resistance
R is calculated as URT/Uref – note that in this way change
R
of supplying current is negligible. In the presented Note
authors consider three methods of linearization:
- direct mathematical method,
FIGURE 3.42 - single linear approximation,
The measuring device with operational amplifier
- piecewise linear approximation.
Platinum RTDs typically are available in two classes In the first method following relation was calculated
of accuracy - class A and Class B. Sensors of class A using the microcontroller:
have in ice point tolerance of  0.06 ohms. Class B is
standard accuracy and has an ice point tolerance of  Z1  Z 2  Z 3  RT
T  RT   (3.70)
0.12 ohms. The error of RTDs increases with Z4
temperature – at 600 C tolerance is  0.43 ohms (1.45
C) for class A and  1.06 ohms (3.3 C) for class B. where: Z1 = -A, Z2 = A2-4B, Z3 = 4B/Ro, Z4 = 2B.
Specially prepared from 99.999% pure platinum
resistors are used as Standard Platinum Resistance The relation (3.70) is relatively simple but it is valid
Thermometers (SPRT) and exhibit temperature only for T > 0C, for T < 0C it is much more complex.
coefficient of 0.3926 %/C). This method of linearization is accurate but requires
Unfortunately the platinum sensor is not fully linear. math library. Simple linear approximation was very
Its transfer function is described as: fast, did not require math library and very small code
space was required. But the accuracy was poor
R T   Ro 1  AT  BT 2  C T  100  T 3  especially for wide temperature range.
(3.68) The best result was reported for piecewise linear
approximation. This method was sufficiently accurate
for T  0C and fast and the math library did not require. In
56 Basic Electrical Measurements

comparison with single linear approximation greater junction of two different metal wires (thermocouple)
code size was necessary to use. Figure 3.44 presents generates voltage dependent on the temperature of the
other measuring circuit applying 4 wire connection and junction1. Figure 3.45 presents the principle of
microconverter ADuC834. Note that this operation of the thermocouple sensor.
microconverter includes excitation, gain stage, ADC
circuit and microcontroller). To third metal
junction
ADuC834 Tx
IEXC1
UT
AIN1
RTD
AIN2 UART hot
SPI junction reference
REFIN+ I2C junction
Etc.
REFIN-
FIGURE 3.45
The principle of operation of thermocouple sensor

FIGURE 3.44 At the end of two different metal wires connecting as

The measuring circuit with microconverter [King et al 2004]
“hot junction” appears the voltage:
Besides the problem of linearization in design of the
measuring device with RTD two other important UT  S12 Tx  To  or UT  S1Tx  S2To (3.72)
problem should be solved:
- self heating where S12 is the Seeback constant (or S1, S2 – Seback
- response time. constants related to common reference temperature,
As the sensor is wound with very fine wire the usually 0C).
current is strongly limited (usually it does not exceed 1
Thus we see that the thermocouple sensor does not
mA). Moreover self-heating effect can destroy
measure the absolute temperature but it measures the
distribution of measured temperature. Usually we
difference of temperatures.
determine acceptable current as:
It would be not reasonable to extend thermocouple
wires to voltmeter connection because some wires, for
T
I (3.71) example platinum one, are expensive. We profit the
T(T )kw “low of third metal”: The algebraic sum of
thermoelectric voltages in a circuit composed of any
where T is the increase of temperature and kw is number of dissimilar materials is zero if all are at
dissipation factor. The dissipation factor depends on uniform temperature. Thus we can use connection of
the design of the sensor and environment conditions for example copper wires and no additional voltage is
and roughly can be estimated as kw = 3 -5 mW/K for generated if the ends if this third materials have the
still air and kw = 10 - 20 mW/K for air 1 m/s. The best same temperature.
solution is to determine experimental the dissipation The thermocouple sensors are commonly used to
factor by applying stepwise increase of current. measure the temperature although they have many
The time constant of typical temperature sensor is serious drawbacks:
several seconds and depends on the external medium - The sensitivity is poor (see Figure 3.46). For
and its flow velocity. But usually conditions of the comparison: assuming current 1 mA we can expect the
measurement are influenced by housing of the sensor. output voltage of the RTD sensor of about 40 mV for
In the sensor with housing the time constant can temperature change of 100 C. Meanwhile the best
increase to dozen seconds and dynamics is closer to thermocouple sensor generates the voltage of about 6
second order (thus two time constants are present). The mV for similar change of temperature. That is why in
problem of temperature exchange between medium and thermocouple measurements the high quality, low
sensor is quite complex and strongly influences the noises, low zero drift amplifiers are necessary. We can
measurement result. increase the sensitivity by connecting several
thermocouples in series.
3.4.b. Thermocouple sensors
The thermoelectric effect was discovered in 1822 by 1
In original experiment of the Seebeck the current of the junction
Estonian physician Thomas Seebeck. He stated that the copper and bismuth was observed by applying of the compass.
 Handbook of Electrical Measurements 57

- Simple construction. It is sufficient to solder, to weld

Output signal [V]
or even to twist two wires to obtain thermocouple
E J K sensor.
40 - Wide range of temperature, from – 200 C to more
than 2000 C.
- Small dimensions enabling point testing of
temperature distribution.
20 T - Excitation in not necessary, thus also problem of self-
R heating is omitted.
Table 3.1 present typical thermocouple sensors.
TABLE 3.1
Typical thermocouples and their performances.
0 200 400 600 800 1000 T [0C]
Type Junction Range S (25 oC)
FIGURE 3.46
C V/oC
The transfer characteristics of typical sensors B PtRh30/PtRh6 100 - 1800 0.3
E NiCr/CuNi -270 - 700 61
- The transfer characteristics are nonlinear – the J Fe/CuNi -210 - 750 52
K NiCr/NiAl -270 - 1000 41
Seeback coefficient changes with the temperature (see N NiCrSi/NiSi -270 - 1000 27
Figure 3.47). Fortunately most of the transfer R PtyRh13/Pt -50 - 1600 6
characteristics are tabularized and are available in S PtRh10/Pt -50 - 1600 6
Internet (for example NIST tables available at: T Cu/CuNi -270 - 350 41
http://srdata.nist.gov/its90/main). The transfer
characteristic can be described in polynomial form: The largest sensitivity exhibits the thermocouple
chromel-constantan (type E), moreover it is
T  a1UT  a2UT2  a3UT3  a4UT4  ... (3.73) nonmagnetic. The most versatile is the thermocouple
chromel-alumel (type K). Thermocouples based on
The polynomial coefficients are also available at platinum measure the largest temperature, are most
NIST Internet page and next can be used in numerical stable and accurate although with low sensitivity. The
linearization [Malik 2010]. Note that for thermocouple thermocouples copper-constantan and iron-constantan
type K the Seeback coefficient is constant in wide (type T or J) are easy to prepare (these materials are
range of temperature. commonly available), have high sensitivity and limited
temperature ranges. The thermocouple nitrosil-nisil
Seeback coefficient [mV/oC] (type N) is stable and resist to oxidation. The
thermocouple type B has very low Seebeck coefficient
E
80
for temperatures below 100 C and therefore can be
J used without compensation of the reference junction.
T

K
40 Tx

0
0 400 800 T [0C]
To
FIGURE 3.47 FIGURE 3.48
The dependence of the Seeback coefficient on temperature The compensation of the reference temperature by unbalance bridge
and second temperature sensor
- It is necessary to guarantee that the reference
temperature is constant. Thus the reference junction The thermostat for stabilization of the reference
should be inserted into thermostat. More often the temperature is non-convenient. More often is used the
electronic circuit for reference temperature temperature dependent source of the voltage. An
compensation is used. example is presented in Figure 3.48. As such source is
But the thermocouples have also important used the unbalanced bridge circuit with additional
advantages: temperature sensor.
58 Basic Electrical Measurements

dR dl 2dr d 
E      1  2      K  (3.75)
60.9 V/oC R l r 
J
51.7 V/oC
where:  = dl/l - strain (deformation),  - Poisson’s
K,T ratio (dependence between longitudinal and transversal
buffer 40.6 V/oC deformation),  - piezoresisitvity constant (for metals
o
10 mV/ C equal to C(1-2), C – Bridgman constant, K – strain
temperature R,S
sensor 6 V/oC gauge factor.
For metals piezoresistivity constant is negligible
small and strain gauge factor is:

FIGURE 3.49 K  1  2 (3.76)

The thermocouple reference junction compensator – model LT1025 of
Linear Technology [Williams 1988]

Figure 3.49 presents reference temperature

compensator developed by Linear Technology. The
various sources of compensating voltage arte delivered
depending on type of the thermocouple. Figure 3.50
present the full circuit designed for temperature
measurement with thermocouples.
J
51.7 V/oC FIGURE 3.51
Typical design of the strain gauge sensor.

Tx Figure 3.51 presents typical design of strain gauge

sensor. The resistor in form of meander is etched from
the foil and bonded to plastic backing. The transversal
LT 1025 LT 1001 10 mV/oC
parts are wider to avoid sensitivity to transversal
component of strain. The whole strain gauge is usually
To attached by the glue to tested material – thus the
deformation of materials is transmitted to the sensor.
FIGURE 3.50
The strain gauge is the sensor of strain (directly) or
The circuit for temperature measurement with thermocouple reference mechanical stress  – indirectly according to the
junction compensator LT1025 proposed by Linear Technology Hooke’s law:
[Williams 1988]
l
Also other producers proposed ready to use  E  E (3.77)
amplifiers specially designed for thermocouples. For l
example Analog Devices developed amplifier AD8497
with included cold (reference) junction compensator where E is the Young’s modulus.
[Duff et al 2010]. As the material for strain gauge commonly is used
constantan – alloy of copper and nickel %%Cu45Ni. It
3.4.c. Strain gauge sensors has high resistivity, good plasticity and low
The stain gauge (or strain gage) is one of the main temperature influence. It has limitation for high
sensors of mechanical values. The principle of temperature (above 65 C) and then special alloy
operation is simple – resistance of the metal wire known as karma (NiCrFeAl) is used. Karma is
depends on its geometry length l and cross section A especially recommended for long time static
(or radius r): measurements. Both constantan and karma have
Poissons’s ratio close to 0.5 and therefore the strain
l l
R  2 (3.74) gauge factor is usually equal to 1.8 – 2.4.
A r Both materials enable to prepare so called STC (Self
Temperature Compensated) strain gauges. The
and the change of this resistance is:
 Handbook of Electrical Measurements 59

influence of temperature can be described by following drift and interferences. This bridge circuit was supplied
relation: by AC voltage next modulated by sensor signal. The
principle of operation of such bridge is presented in
R
T     K     T (3.78)
Figure 3.53.
R
where  is temperature coefficient of resistance of
strain gauge material,  is thermal expansion x
coefficient of substrate (tested material) while  is PSD Filter
U1 K uU 1 U2 Uout
thermal expansion coefficient of strain gauge material.
The  coefficient is relative small (40ppm/C). It is U0 U0
possible to select material in such a way to obtain OSC
condition  -  = 0 for chosen tested materials. These
materials are known as A-alloys.
FIGURE 3.53
The measured deformation are usually very small Strain gauge bridge instrument with carrier amplifier
and expressed in microstrain ( = 10-6 for 1 strain). If
we measure small deformation of several strains it If the measured strain  varies periodically with the
means that the output signal of the bridge circuit with frequency  then the variation of the resistance of the
strain gauge is very small – in V range). It is the main sensor with the constant K is
measurement problem in application of strain gauges. It
means that even if we have STC gauges the changes of R x
temperature can influence the result (we cannot  R x  K x  K m sin t (3.79)
Rx
distinguish which part of the change of resistance
comes from temperature and which from strain). That The output voltage of the bridge circuit of the
is why usually it is recommended to use second sensitivity factor S supplied by the voltage signal
“temperature compensation” sensor as it is presented in
U0=Uomsin t is
Figure 3.52.
U1  S RxU 0  SK mU om sin t sin t (3.80)
a) +
After simple calculations we obtain
-
U1  U1m cos     t  cos     t  (3.81)
+ where U1m=SKmU0m.

b) 0 From the equation (3.81) we can see that the bridge
circuit supplied by the AC voltage works as a
modulator device – the magnitude of the output signal
FIGURE 3.52 of the frequency the same as the supply voltage is
Connecting of the sensors to a bridge circuit to eliminate temperature
error: a) two active sensors, b) active and passive sensors
modulated according to the variation of resistance (and
stress). The spectral characteristics of these signals are
Of course the best is the case when we can organize presented in Fig. 3.54.
two differential sensors – as it is presented in Figure
3.52. If we do not have differential sensors accepted is U1 U2
to connect the second dummy (passive) sensor to x U0
Ku
adjacent arm of the bridge circuit. In the case presented filter

in Figure 3.52b the dummy sensor is in transverse

direction to be insensitive to the longitudinal measured
stress.   f    2 2 f
f
As it was mentioned above the main problem of
instrumentation is very small signal of the sensor.
FIGURE 3.54
Therefore formerly as strain bridge instruments have The spectral characteristics of the signals of the circuit presented in
been used special carrier amplifiers to decrease zero figure 3.53
60 Basic Electrical Measurements

The transfer characteristic of phase sensitive detector 3.4.d. Linear Variable Differential Transformer
PSD is The linear variable differential transformer LVDT
sensor is the most frequently used sensor of
U 2  K uU1 cos  t (3.82) displacement.
x
or as a series
L1
2  2 2 
U 2  K uU 1 1  cos  t  cos 4 t  ... (3.83)
  3 15  L1 L2 L1-L2
x
thus L2

 1 
 sin  t  sin  2    t  
2 3
U 2  KuU 1m   (3.84)
  sin  2    t  ...
1  FIGURE 3.55
 3  Inductive sensors and their transfer characteristics

The phase sensitive detector acts as a selective filter. Let us start with simple inductive sensor with
Now it is only necessary to connect the low-pass filter reluctance variation presented in Figure 3.55. The
to correctly recover the measured signal inductivity of single sensor is:

2 n2 n2
U out  KuU1m sin t (3.85) L  (3.86)
 R lo l
 Fe
o Ao  Fe AFe
The carrier amplifier is rather complex, the AC
supplying require sometimes additional phase balance
where l, , A are length, permeability and cross section
of the bridge circuit and the useful frequency
of air and iron part, while n is number of turns and R is
bandwidth is limited to the 20% of carrier frequency.
Meantime on the market appeared low noise, low zero the reluctance. For Fe >> o we can write:
drift monolithic amplifiers. Therefore recently as strain
1
amplifier the DC supplied bridge instruments are L  n 2 o Ao (3.87)
commonly used. Nevertheless for very small lo
deformations and strong interferences the AC supplied
bridge instruments can be recommended. Table 3.2 If we assume movement of internal iron part then l =
presents comparison of two instruments of the same 2x + 2lo and reluctance is:
company Showa Measuring Instruments Inc. - one with
2lo 2x
AC 5 kHz bridge (carrier amplifier) model 5683 and R   (3.88)
DC bridge model 5693. o Ao o Ao
TABLE 3.2
Comparison of AC and DC strain gauge bridges instruments. and the inductance is

1 2 o Ao 1  A
Parameter AC bridge DC bridge
Sensitivity [V/strain] 10/200 10/1000 Lo  n where Lo  n 2 o o (3.89)
Smallest range [strain] 200 1000 2 lo 2 lo
Noise level [strain] 0.6 20
Zero drift [strain/oC] 0.1 1 Thus the change of inductivity (and output signal of a
Zero drift [strain/24h] 0.5 5 sensor) (L-Lo)/Lo is:
Frequency bandwidth DC – 2 kHz DC – 500kHz
Nonlinearity 0.1% FS 0.01% FS
Stability 0.2%/24h 0.05%/24h x
2 3
T Cu/CuNi -270 - 350  L1 x x x
lo
         .. (3.90)
L1 x lo  lo   lo 
From this comparison we see that AC bridge exhibits 1
much better resolution while DC bridge exhibits much lo
better frequency bandwidth and linearity.
 Handbook of Electrical Measurements 61

The output signal of single inductive sensor is x

nonlinear (see Figure 3.55). We can significantly
improve the linearity by connecting into bridge circuit U'2
U2
two differential sensors because:
U1
3 5
 L1  L2 x x x U”2
        .. (3.91)
L1 L2 lo  lo   lo 

FIGURE 3.57
U1 The solenoid type linear variable differential transformer LVDT
x
In transformer sensor presented in Figure 3.57 if the
moving high permeability ferromagnetic part is in the
U’1 U”1 central position both secondary voltages are the same.
U’2 U”2 Because output voltage is the difference of the
secondary voltages the output signal for central
position of the moving part is zero. Movement of the
U2
ferromagnetic part causes that one of the secondary
voltages (and its phase shift) is dominating and output
FIGURE 3.56 signal increases as it is presented in Figure 3.58.
Differential transformer sensor
U2 
We can significantly improve properties of the same 180
sensor connecting it in transformer circuit (Figure 3.56)
instead of bridge circuit. The input voltages are: x

L1 L2
U 1  U 1 and U 1  U1 (3.92)
L1  L2 L1  L2
0
x
while the output voltages are:

n2 n
U 1 and U 2  2 U 1
FIGURE 3.58
U 2  (3.93) The transfer characteristics of typical LVDT sensor – output voltage
n1 n1 and phase shift (related to input voltage)

The output signal is the difference of the secondary In the real LVDT sensor transfer characteristic is
voltages thus: usually slightly nonlinear – mainly in x = 0 point. The
reason is not completely compensation of two AC
n2 L  L1 voltages – due to possible asymmetry. The solution is
U2  U1 2 (3.94)
n1 L2  L1 to use the conditioning circuit with synchronous
detector (Figure 3.59).
Assuming that air gaps are: 2lo-2x and 2lo+2x and
taking into account Eq. (3.87) we obtain: x

U'out
n x U"out
U 2  2 U1 (3.95) U'out
n1 lo
SD Uo
The transformer sensor is linear. Usually instead of
x
choke sensor presented in Figure 3.56 the other Uref
U"out
solenoid sensor (presented in Figure 3.57) is recently PS
used. It is more convenient in use (non-limed distance
of moving part) and to design (simply three coils on the
plastic frame.
FIGURE 3.59
The conditioning circuit of LVDT sensor – with synchronous detector
62 Basic Electrical Measurements

Synchronous detector removes the orthogonal S1

component and the transfer characteristic becomes
linear. And moreover what is important DC output
signal enables to detect direction of the movement. But 
if we apply the synchronous detector it is necessary to
add the phase shifter because the phase shift between S2
excitation reference voltage and output signal is other
than 0/180. Moreover this phase shift changes with
temperature. That is why sometimes simple rectifiers
are used as it is presented in Figure 3.60.

FIGURE 3.62
The RVDT sensor (resolver) for measurement of the angle [after
Padmanabhan 2000].

Voltages induced in two stators depends on the

rotation angle of the rotor:

U S1  A1  Em sin  and U S 2  A2  Em cos  (3.96)

FIGURE 3.60 The angle can be calculated as:

The conditioning circuit of LVDT sensor – with diode rectifiers and
differential amplifier [after Nyce 2004]
A1
  sin1 (3.97)
In the circuit presented in Figure 3.60 diodes load the A  A22
2
1
sensor, voltage drops on the diodes decrease output
signal, parameter of diodes should be the same. More 3.4.e. Glass sensor for pH measurement
sophisticated rectifying circuit includes in their Measurements of the pH are important not only in
conditioning circuit AD 598 Analog Devices. chemical industry but also in many other areas, as
Additional advantage of this solution is independence medicine, food industry, agriculture etc. For pH
of the output signal on the value of exciting voltage. measurement most commonly the glass sensor is used.
x The pH is the measure of acidity (less than 7) or
basicity (alkalinity) (more than 7) (Figure 3.63).
AD598
distilled water

ABS
1n CH3COOH

UA
orange juice

black coffee

soapy water
lemon juice

filter
gastric acid
0.1n H2SO4

1n NaOH
sea water

value
ammonia

A-B
bleach

filter
A+B
ABS filter
value 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
acid base
UB
Osc
FIGURE 3.63
Typical values of pH of various substances.

In distilled water the co9ncentration of hydrogen ions

FIGURE 3.61 is the same as concentration of hydroxide ion:
The conditioning circuit of LVDT sensor developed by Analog
Devices.
 H    OH    10 14  10 7 (3.98)
The LVDT sensor enables measurement of the
displacement with resolution of m. Beside the linear And as pH unit it was assumed concentration of
sensors often are used also Rotating Variable hydrogen:
Differential Transducer RVDT used to the angle
measurement. Figure 3.62 present an example of such pH   log  H   (3.99)
sensor known as resolver.
 Handbook of Electrical Measurements 63

E [mV]
500
Hg2Cl2

pH
KCl KCl
3 6 9

-500
FIGURE 3.64
Two electrodes for pH measurement (glass and reference calomel one)
or combined electrode.
FIGURE 3.65
For measurement of concentration of hydrogen ion Typical E=f(pH) dependence of the glass electrode.
(pH value) commonly is used glass electrode. On the
The slope of E = f(pH) dependence is about 58
thin glass bulb (thickness of 50 – 200 m) are
mV/pH and increase of about 2 mV/10C. The
concentrated charges depending on the pH value. As a
common point for all temperatures (point of rotation) is
glass specially doped by lithium ions is used. As result
called isopotential point. For the designers of the glass
on the glass electrode appears the potential described
electrodes it is recommended if this point is close to pH
by the Nernst equation:
= 7.
2.303RT
E  Eo  pH (3.100) UAB
F
where Eo is standard electrode potential, R is universal
gas constant R = 8.314 472(15) J K−1 mol−1, T is
absolute temperature, while F is Faraday constant F =
9.648 533 99(24)×104 C mol−1. RT
After substitution of constants we obtain that the
electrode generates potential FIGURE 3.66
Typical circuit of the pehameter instrument.
E  0.19847  T  pH  mV  (3.101)
There are two main problems in pH measurements.
or The first one is that the resistance of electrodes (source
of the voltage) is usually very large – 500 M or more.
E  58.13  pH mV  (3.102)
Thus the input resistance should be much larger (see
Figure 2.39). Typical voltmeter cannot be used for pH
The glass bulb is filled by KCl solution and to obtain measurement because its input resistance is not so large
the potential the reference electrode is included. As the – the special electronic circuits with feedback should
reference electrode silver-chloride electrode or calomel be designed.
electrode is used. Because on the glass electrode we The second problem is that the transfer
have only the potential to obtain voltage it is necessary characteristics of the electrodes (presented in Figure
to use the second reference electrode with potential 3.65) can be different for various electrodes. Therefore
independent on pH. Usually it is the same electrode as for accurate pH measurements it is recommended to
used in the glass electrode. scale the instrument by using special standard buffer
The calomel reference electrode is presented in solutions. The point of zero voltage of electrodes
Figure 3.64. The glass tube ended with porous seal and should correspond with pH = 7 and to correct this point
filled with KCl solution. The inner electrode is filled by the adjustable voltage source UAB can be used. The
calomel Hg2Cl2 and mercury. Recently often instead of change of slope can be corrected by resistor R T – this
two electrodes one combined electrode is used. The resistor is sometimes the temperature sensor to auto
example of such electrode is presented in Figure 3.64. temperature correction. As the transfer characteristic E
Figure 3.65 present typical dependence of E = f(pH) = f (pH) is close to straight line it is sufficient to use
of glass electrode. two buffer solutions to calibrate the pehameter.
64 Basic Electrical Measurements

3.5 AC to DC conversion feedback the threshold voltage Up’ decreases to

3.5a. The diode rectifying devices negligible small value Up’  Up/Ku. (Ku – gain of the
It is relatively easy to convert the value of AC amplifier). In such case the converter behaves as a
signal to the DC signal using diode rectifiers. A typical practically linear transducer of input voltage uin
rectifying diode conducts the current for only one
R2
direction of applied voltage. An example of the transfer U out   U in (3.103)
characteristic of a silicon rectifying diode is presented R1
in Figure 3.67.
2R2 2R2
I I
a) I b)
R1 R2
Iav
t U1 C
R1
- -
uin A1 A2
+ + Uout
U

FIGURE 3.67 FIGURE 3.69

The principle of diode rectifying. The full-wave converter of average value of AC voltage to the DC
voltage (Tran Tien Lang 1978)
A typical transfer characteristic of the rectifying
diode is presented in Figure 3.67a. In the conducting Figure 3.69 presents another converter of average
direction the characteristic is nonlinear and the value of AC voltage to DC voltage. The amplifier A1
conductance starts from a certain non-zero threshold works a half-wave rectifier according to the following
voltage (Up  0.5 V). Moreover, a small inverse current conditions
will be conducted in the “non-conducting” direction.
For that reason, the simple rectifying diode circuits can  u for u  0
U1   (3.104)
be used only in non-accurate devices – for for u  0
0
measurement purposes it is necessary to support the
diode with a more sophisticated electronic circuit, for The second amplifier (A2) adds two voltages
example containing an operational amplifier.
U out  uin  2U1  (3.105)
R2
thus, the full-wave rectification is achieved:
Uout u for u0
R1 U out   (3.106)
 u for u0
-
Uin 3.5b. The peak value converters
+ Figure 3.70 presents the principle of the conversion
of the peak value of the AC signal. The rectified
voltage is connected to the capacitor C. If the time
FIGURE 3.68
constants of the RC circuits (circuits of charge and
AC/DC converter with rectifying diodes and the operational amplifier. discharge respectively) are designed in such a way that
process of charging is fast and discharging is slow then
Figure 3.68 presents the typical AC/DC converter the voltage across the capacitor is equal to the peak
with rectifying diodes and operational amplifier. The value of the supplying voltage. By connecting two
amplifier operates as a current transducer – rectified converters it is possible to detect peak-to-peak Upp
current flows in the feedback circuit. For a strong value as it is presented in Figure 3.71.
 Handbook of Electrical Measurements 65

error [%]
U
UC converter rms
0
u=Umsint UC
C con
R0 -20 ver
t ter
av

FIGURE 3.70
The principle of operation of the peak detector. 1 2 4 6 CF

FIGURE 3.73
The error of conversion depending on the crest factor (converter rms
and rectifying (averaging)converter).
Uin
-
TABLE 3.3
Crest factor of different signals.
+ Uout Waveform rms value CF Error [%]
Sine 0.707 1.41 0
Square 1 1 11
Triangle 0.577 1.73 -3.8
Noise 0.333 3
FIGURE 3.71 Pulse 0.1 10 -44
The principle of operation of the peak-to-peak detector.
If the rectifying measuring device is calibrated for
In the circuit presented in Figure 3.72 the capacitor sinusoidal waveform the measuring error for square
C1 is charged to the maximal (peak) value. The waveform is larger than 10%, and for crest factor equal
feedback circuit (elements R1 and D1) protects the to 10 error is very large.
amplifier from the saturation when Uin < Uc. Periodical The conversion of the rms value of the AC signal to
connection of the resistor Rp enables the capacitor to the DC voltage is more difficult. According to the
discharge and repeat the conversion. equation:

1 2
u  t  dt
T
R1 U rms  (3.107)
D1 -
-
Uin + we should perform the following operations: square,
+
Uout
mean value calculation and root operation (root-mean-
square). These mathematical operations are possible to
Rp perform using two multipliers, as is schematically
C1
shown in Figure 3.74.

squarer integrator root operation

FIGURE 3.72
The precise peak detector circuit. x
Uin y
3.5c. The true rms value converters x
y
The converters presented in Section 3.5.a allow
determining averaged rectified value of AC signal. On -
many digital instruments is added notice “true rms”. -
Indeed old instruments used simple rectifying device +
+ Uout
and they were scaled for sinusoidal shape. But if the
signal is distorted the error of rectifying converters can
be large. Figure 3.73 presents error of conversion
versus crest factor CF = Umax/Urms. FIGURE 3.74
The true rms converter using two multipliers.
66 Basic Electrical Measurements

In the circuit presented in Figure 3.74 the first

fullwave
multiplier is squaring the input voltage, then this log divider
rectifier antilog
voltage is averaged in the integrator circuit and next
rooting is performed. The root operation is achieved by U2in
Uin logU2in log U2in
second multiplier in the feedback circuit of operational Uin Uout
(2logUin) Uout
amplifier. (2logUin-logUout)

filter
Squarer integrator
U2in
and divider logUout
Uout
Uin
x2 Uout
y

- FIGURE 3.77
- Uout The principle of operation of the AD637 rms converter of Analog
+ Devices (Kitchin 1986)
+
Figure 3.77 presents the principle of operation of
another true rms converter - utilizing the log-antilog
circuit (logarithmic and exponential amplifier –
FIGURE 3.75 described in next Section). This converterperforms the
The true rms converter with indirect calculation of rms value. following operations:
The converter presented in Figure 3.74 is realized 1) 2 logU in 2
the same as: logU in
 
seldom due to its poor dynamics. More often is used
circuit presented in Figure 3.75 with indirect 2) 2 logU in  logU out the same as: log U in
2
/ U out
calculation of rms value. Indeed in this circuit we have
fulfilled following relation: 3) 2
anti log(logU in / U out )  U in
2
/ U out
2
4) U in / U out
U in2
U out  (3.108)
U out 5) U out  U in
2
/ U out

Thus we have Thus, as the result of these operations we obtain:

U out  U in2 (3.109) U out  U in2 which is exactly the definition of the
rms value.

I2in squarer
absolute value
Iin divider I4
Iout and
Uout
2 Iout U/I converter I1
Uin absolute value Urms
squarer current A4
and U/I filter Iout
divider mirror
converter
R0
A2 I3
Uin
A1
A3

FIGURE 3.76
The principle of operation of the AD636 rms converter of Analog
Devices (Kitchin 1986) FIGURE 3.78
The simplified circuit of the AD637 rms converter of Analog Devices
There are two strategies of true rms calculation. In
the first one presented in Figure 3.76 the multiplier is In the circuit presented in Figure 3.78 realized is
used. Instead of multiplier the logarithmic amplifier operation
can be used as it is presented in Figure 3.77.
I 4  I12 / I 3  I12 / I 4 (3.110)
 Handbook of Electrical Measurements 67

Instead of utilizing the relationship (3.107), it is also

possible to perform AC/DC conversion by applying the
rectifier physical definition of the effective value: “the effective
(rms) value of the AC current is equal to the DC
Current
current value, which in the same resistance R in the
mirror
time of one period causes emission of the same value of
the heat”. Such idea was realized by the rms converter
Ix Iz of Linear Technology (model LT1088)2 presented in
Q1 Q2 Figure 3.80.
Q4 The voltage generated by the thermocouple is a
measure of the heat emission on the resistor and is
Q3 Iy compared with the voltage generated by the second
Iz-Iy
thermocouple sensor heated by the AC current. Thus,
we compare the heat effects of both currents. Such a
transducer enabled AC/DC conversion in very large
FIGURE 3.79 bandwidth (up to 300 MHz) with a large crest factor.
Structure of translinear rms-dc converter [Wassenar et al 1988]
Recently is available only in laboratory scale
Very often as the rms core the translinear circuit
+5V
introduced by Gilbert is used [Gilbert 1975, Wassenar
et al 1988]. For the loop Q3 – Q1 – Q2 – Q4 according Uin

to Kirchhoff’s law we can write: 1 rectifier

14

2 13
U BE3  U BE1  U BE2  U BE4 (3.111) squarer
3 divider 12

Assuming exponential relationship between the base- 4 11

emitter voltage and collector current: current
mirror 10
5
kT3 I c3 kT1 I c1 kT2 I c2 kT4 I c4 out
ln  ln  ln  ln (3.112) 6 9
q Is q Is q Is q Is In bufor
7 8
Iout
Assuming the same temperature T and the same
saturation current Is we obtain translinear principle:
Ux
35 7
I c3 I c1  I c2 I c4 (3.113)

In the circuit presented in Figure 3.79 Ic3 = Ic1 = Ix

and Ic4 = Iy and Ic2 = Iz. Thus we obtain:

I x2  I z I y and I z  I x2 (3.114)

FIGURE 3.81
The design of the true rms voltmeter design – the proposal of Analog
Devices
Uin
Uout
Fig. 3.81 presents an example of the application of
the rms converter – as the true rms voltmeter (proposed
T1 T2 by the Analog Devices).
Table 3.4 presents performances of different rms
converters. Beside the log/antilog converters Linear
Technology designed converters basing on sigma-delta
FIGURE 3.80 conversion.
The thermal true rms converter

2
LT1088 is not currently manufactured by LT.
68 Basic Electrical Measurements

TABLE 3.4 both input signals. This principle of operation is

Performances of market available rms converters.
presented in Figures 3.82 and 3.83.
IC Bandwidth Accuracy CF
1% error
AD637 200 kHz/1% 1 mV 10 Ux
AD737 33 kHz/1% 0.2 mV 5
AD8436 65 kHz/1% 0.01 mV 10
ADL5502 450 MHz-6GHz 0.25 dB 10
LMH2120 50 MHz-6GHz 0.5 dB
LTC1966 6 kHz/1% 0.1% 4
Uref
LTC1968 500 kHz/1% 0.1% 3
LT1088 300MHz 0.01 Fs 40
Note1: AD – Analog Devices, LT – Linear Technology, LMH – Texas
Instruments = 0o 0o << 90o
Note2 – LT1088 (finished production - only for comparison)

The translinear circuits enable also conversion in

high frequency range – up to 6 GHz [Nash et al 1999].
Ux
As an example we can consider ADL5502 converter
designed for microwave power measurements. This
circuit converts both – rms value and envelope value
and this way the measurement of CF is possible.
Uref
Of course the rms value computation is possible by
digital signal processing.
 = 90o  = 180o
3.5d. Phase sensitive detectors
Synchronous detectors (known also as synchronous
detector or phase-sensitive demodulators) play a very UxUrefcos
important role in the measurements and
instrumentation. It helps in the separation of two signal
components – in phase and 90 degrees out of phase, as FIGURE 3.83
The principle of the operation of phase-sensitive rectifier.
it was presented in the example earlier. But the main
their feature is excellent sensitivity enabling noise The phase-sensitive rectifier works according to the
rejection and small signal amplification as lock-ion following equation
amplifier (next Section)..
U out  U ref U x cos  (3.115)

pass when simple where cos is the cosine of the phase shift between Ux
Uin > 0 rectifier and Uref.

a) b)

pass when phase sensitive Uref

R1
Uref > 0 rectifier Uref Uout
U1
R2 Ux

Uref

Ux U2
FIGURE 3.82
The difference between simple and phase-sensitive rectifier.

FIGURE 3.84
The phase-sensitive rectifier converts the measured The simplest diode phase-sensitive rectifier.
voltage Ux with respect to additional reference voltage
Uref. Thus, it is equipped at least with two inputs and A typical diode polarized as conducting and non-
one output. The reference voltage switches the conducting element can be used as the switch in phase
rectifying elements in such a way that the mean value sensitive rectifier. An example of such a rectifier is
of the output signal depends on the phase shift between presented in Figure 3.84.
 Handbook of Electrical Measurements 69

In the phase-sensitive rectifier circuit presented in

Figure 3.84 the voltage signal Ux is added to the 12
reference voltage Uref in the first half-period of this 1
2 + -
signal and is subtracted from this voltage in the second AMP A 13
half of the period. According to the phasor diagram 20 - + Uout
presented in Figure 3.84b the output signal Uout is 17 -U
18 + 14
AMP B 15
U1  2
U ref  2U ref U xcos   U x2 19 - 16
Uref 7
9 +
U2  2
U ref  2U ref U xcos   U x2 (3.116) COMP
10 -
and
8
U out  U 1  U 2 
FIGURE 3.87
 U2  (3.117) The functional block diagram of phase-sensitive detector model
2U x cos   1  2x sin 2    2U x cos  AD630 of Analog Devices
 U 
 ref 
There are also available ready-to-use monolithic
For correct operation of this rectifier it is required phase-sensitive rectifiers. As an example a functional
that Uref > Ux. The full-wave version of diode phase- block of the AD630 circuit developed by the Analog
sensitive rectifiers is presented in Figure 3.85. Devices is presented in Figure 3.87. This IC includes
two precise input amplifiers and a third amplifier
switched by the reference voltage connected to the
input 9.

Uref oscillator phase

Ux shifter
Uref +15V
Uin 16 9 11
+ 17 Uout
x 1000 15 AD630 13
19 12
- 20 1 10 14 8
R0
instr. amp. -15V
FIGURE 3.85
The full-wave version of diode phase-sensitive rectifiers.
FIGURE 3.88
Figure 3.86 present the phase sensitive detector basing The application of the phase-sensitive rectifier as the output circuit of
on the operational amplifier. the unbalanced type AC bridge

Phase sensitive detectors are commonly used as:

- to reject noises in lock-in amplifiers;
- to separate real and imaginary components, for
- example in impedance analysis;
Ux - as output device of alternating bridge circuit do
+
Uout improve balancing and detect the direction of the
change of impedance.
Uref The lock-in amplifiers are described in the next
Section. Figure 3.88 present the application of AD630
circuit to alternating bridge. By using the synchronous
detector we are able to detect increasing or decreasing
of the tested impedance – for example direction of
FIGURE 3.86
The phase-sensitive rectifier with the operational amplifier. movement of LVDT sensor (Figure 3.59) or strain
gauge sensor (Figure 3.53). Figure 3.89 presents the
application of synchronous detector to impedance
analysis.
70 Basic Electrical Measurements

should be sufficient. Also the gain Ku should be the

phase shifter same for all frequencies in the bandwidth. There is also
90o
cos amplitude distortion caused by non-linear transfer
characteristics and introduction to the signal additional
Rs PSD1 Im(Z) harmonics. The non-linear distortion appears when the
osc input signal is large. Therefore it is recommended to
Zx PSD2 Re(Z) limit input signal of the amplifier by applying the
sin negative feedback. If the change of the phase with
frequency in not linear also the phase distortion can
decrease of output signal quality.
FIGURE 3.89 Usually, it is assumed that the frequency bandwidth
The application of the phase-sensitive rectifier to the impedance
analysis. is the frequency range for which the Ku factor does not
roll off more than 3dB. And another important factor –
By using of the synchronous detector in lock-in the dynamics of the amplifier (ratio between the
amplifiers we are able to measure small alternating smallest detectable signal and the largest – limited by
voltages even in the nV range. In measurement small the supply voltage signal). Dynamics (the smallest
resistance (Figure 3.90b) or small change of resistance signal) depends mostly on the level of noises and zero
(Figure 3.90a) the output signal is very small. By drifts.
supplyin the measuring circuit by alternating voltage The real revolution in the amplification technique
and using this voltage as reference we are able to was the development of the integrated operational
measure this very small signal. Figure 3.90b presents amplifiers, and later the instrumentation amplifiers. No
the application of phase sensitive detector to measure wonder that operational amplifier is one of the most
of  resistors. frequently described electronic circuits (Coughlin
2000, Franco 2001, Jung 2004, Mancini 2002, Stanley
a) b) 2001, Zumbahlen 2008).
The operational amplifier (OpAmp) is a special kind
of the integrated amplifier with the following
PSD performances: very large amplification factor – larger
Rs than 105, very large input resistance and small output
osc
resistance. Additionally, small temperature errors
Rx PSD coefficient, small level of noises, small zero drifts and
large CMR (common mode rejection) factor are
required.

3 5
+Uin offset
FIGURE 3.90
The application of the phase-sensitive rectifier to measure of small
7
changes of resistance or small resistances. +U

3.6 The measuring amplifiers 6

3.6.1. Operational amplifiers Uout
The amplifiers are generally used for the
amplification of the voltage signal. However, the -U
amplification process enables also improvement of the 4
2
signal quality – mostly the signal-to-noise ratio. A good -Uin
8 7 6 5

amplifier should exhibit sufficiently large and steady -

741
amplification factor (gain) Ku=Uout/Uin, large input 1 offset +
resistance and small output resistance.
1 2 3 4
It is also required to perform the amplification
without the distortion. The amplifier can process the
signal with frequency distortion caused by the unequal FIGURE 3.91
gain of all frequencies comprising the signal. To obtain An example of the design of the operational amplifier – historical
amplifier 741.
small linear distortion the bandwidth of frequency
 Handbook of Electrical Measurements 71

Figure 3.91 presents the typical design of the use is also the parameter CMR – Common Mode
operational amplifier. The operational amplifier Rejection defined as
practically does not work intrinsically – with such large
amplification it would be immediately saturated. CMR  20 log CMRR (3.121)
Therefore the operational amplifier works always with
suitable feedback. We can say that the operational Taking into account this parameter the output voltage is
amplifier is some kind of semi-finished product
 U 
U out  U1  U 2 K  1 
designed for the construction of various electronic 1
 (3.122)
circuits. By connecting an appropriate feedback circuit  CMRR U1  U 2  
we can obtain the required device: multiplier, adder,
integrator, generator, filter and of course amplifier. The second component in the square brackets of the
equation (3.122) describes the error caused by the
+U
presence of the common component.

U1 Ku= 1 + R2/R1 Ku=-R2/R1

+
Uout +
U1 U2 In Out
- R1 R2
Uout
- -
U2 R2 In

+ Out
a) b)
R1
-U

FIGURE 3.92
The differential amplifier and its symbolic representation. FIGURE 3.93
Non-inverting (a) and inverting (b) configuration of the amplifier.
Modern operational amplifiers can be quite complex
but in all the main input part constitutes the differential The main idea of operational amplifiers consists in
amplifier (Figure 3.92). The important advantage of application of feedback. If gain of the amplifier is high
such an amplifier is the possibility of suppression of the the performances of the amplifier with feedback
parasitic signals. The input signal is processed as the depends only on the feedback loop. Thus by design
difference of two inputs signals different feedback we can obtain amplifier with desired
performances. Figure 3.93 presents the two main
U out  Ku U1  U 2  (3.118) configurations of the amplifier. By joining both these
configuration we obtain the differential amplifier - as it
The parasitic interference signals U are the same on is presented in Figure 3.94.
both inputs. Therefore the output signal is Ku=R2/R1 R2/R1= R4/R3
U out  Ku U1  U   U 2  U   Ku U1  U 2 
(3.119) R1 R2
+
Thus it is possible to amplify the voltage difference Out
In
with the large common signal U in the background.
-
The possibility of the rejection of the common
parasitic component is described by the coefficient R3
CMRR – Common Mode Rejection Ratio defined as
R4

K
CMRR  (3.120)
K FIGURE 3.94
Configuration of differential amplifier.
where K- is the amplification of the voltage difference
and K+ is the amplification of the common signal. In
72 Basic Electrical Measurements

frequency. Figure 3.96 present typical Bode plot of

a) R b) Ro Iout operational amplifier. The gain usually is presented in
decibels3 – most often the fall-off (the slope) of the
Iin R gain versus frequency is 20 dB/dec starting from
- - certain cutoff frequency (Figure 3.96). Because product
Uin of gain and bandwidth is constant often this product is
+ + presented as GBW factor.
Uout From Bode plot presented in Figure 3.96 results that
we can increase the gain at the cost of the frequency
bandwidth - the largest is for K = 1). From the Bode
FIGURE 3.95
Converters: current-voltage (a) and voltage-current (b). plot we can also consider the problem of stability.
Indeed if the gain K of the closed loop is:
The application of appropriate feedback enables to
design various performances of the amplifier. Figure Ku
K (3.126)
3.95 presents examples of two main transducers: 1   Ku
current to voltage converter (high resistance input/low
resistance output) and voltage to current converter (low (where Ku is the gain of open loop and the  is
resistance input/high resistance output). feedback coefficient) it is possible that the denominator
In the current to voltage converter (Figure 3.95a) the can be equal to zero and the whole circuit is then
output signal Uout is proportional to the input current unstable.
according to the following equation It is unacceptable if the phase shift is larger than
180 for the gain larger than 1 because it means that the
U out   RI in (3.123)
negative feedback is changed to positive. Therefore
usually it is determined the phase margin PM (Figure
while in the voltage-to-current converter (Figure 3.95b)
3.96) as the margin to unstable state.
the output current is:

U in 100 100
I out  (3.124)

phase margin - degree

open loop gain [dB]

R
The conversion factor depends on the value of
resistance R, which we can set very precisely.

20 log(K)
open loop gain
0 0
amplitude

closed loop gain

20dB/dec -20
1kHz 1MHz frequency
log(f)
0 dB
FIGURE 3.97
phase

Bode plot of AD847 opamp.

PM 3
Decibel dB is logarithmic unit of ratio of power P or voltage U:
180o
P2 U
(out/in) dB  10 log  20 log 2 (3.125)
P1 U1
dB P2/P1 U2/U1
FIGURE 3.96 -40 0.0001 0.01
Bode plot of typical opamp. -20 0.01 0.1
-3 0.5 0.707
The frequency bandwidth of the operational 0 1 1
amplifier depends on the gain – it decreases as the gain 1 0 0
20 100 10
is increasing. The relationship between gain and 40 10 000 100
bandwidth often is presented as the Bode plot – 100 10 000 000 000 100 000
dependence of logarithm of gain versus logarithm of
 Handbook of Electrical Measurements 73

Fortunately as it is presented in Figure 3.97 the The offset voltage is not a great problem because we
opamps producers design such devices with sufficient are able to compensate it – internally or by external
phase margin. But is certain work conditions, for resistors. But danger is offset temperature drift because
example capacity load this margin can be at risk of the we are not able to separate it from the input signal with
loss of stability. In such case appropriate compensating DC component – as in the case of infrared sensors,
element should be connected to the circuit. strain gauge or thermocouple sensors. Recently the best
There are many factors influencing the quality of method to decrease this drift to the level of 20 nV/10C
signal conversion by the opamp circuit, as noises, is auto-zero principle joined with chopper technique.
offset, zero drift, bandwidth etc. It is not possible to But due to improvement of technology it is possible to
design the circuit with the best of all these factors - obtain similar small temperature drift without auto-zero
therefore available are specialized amplifiers, as for what is important because auto-zero operation limits
example: low noise, low zero drift, high speed or the frequency bandwidth.
energy saving amplifiers. The noises level is especially important if we are
The datasheets usually described many various interested in large dynamics and low distortion. For
properties of the amplifier, as: open loop gain, offset example presented in table 3.6 low noise amplifier
voltage, offset voltage drift, noise level, 1/f noise, input AD797 enables to obtain the total harmonic distortion
bias current, input offset current, total harmonic THD4 as high as 130 dB. We separately consider the
distortion, input resistance, output resistance, 0.1 dB noise density in nV/rtHz and low frequency 1/f noise in
bandwidth or GBW, slew rate, settling time, overshot the frequency range 01 - 10 Hz as peak-to-peak
of pulse response, phase margin, CMR ratio, supply voltage.
voltage, quiescent current, power consumption. Almost
all these parameters depend on temperature and
frequency – therefore often they are presented in form overshoot
of a graph. Table 3.5 presents the main parameters of
output voltage

three popular operational amplifiers and Table 3.6

presents the best performances of AD opamps.

U
TABLE 3.5
Main parameters of typical opamps. slew rate – U/t
LM358 TL071 LM833 t
GBW 0.7 MHz 3 MHz 16 MHz time
Slew rate 0.3 V/s 13 V/s 7 V/s settling time
Bias current 250 nA 200 pA 700 nA
Offset 7 mV 10 mV 2 mV
Offset drift 7V/C 18 V/C 2 V/C FIGURE 3.98
Noise 1 kHz 40 nV/rtHz 18 nV/rtHz 5 nV/rtHz Answer to step change of the input signal.
CMRmin 65dB 70 dB 80 dB
TABLE 3.6 The speed of the amplifier can be evaluated by
The best performances of opamps. testing the answer to the step change of the input signal
AD549 AD8630 AD797 – as it is presented in Figure 3.98. Modern high speed
Low bias I Low drift Low noise amplifiers have by 2 GHz bandwidth the slew rate as
Bandwidth 1 MHz 2.5 MHz 8 MHz high as 10 kV/s and settling time less than 2 ns.
Slew rate 3 V/s 1 V/s 20 V/s
Bias current 0.04 pA 30 pA 250 nA
3.6.2. Instrumentation amplifiers
Offset 0.3 mV 1 V 25 V
Offset drift 5V/C 2 nV/C 0.2 V/C
The operational amplifier should exhibit large gain
Noise 1 kHz 35 nV/rtHz 22 nV/rtHz 0.9 nV/rtHz but it is not absolutely indispensable to have steady and
Noise < 10Hz 4 Vpp 0.5 Vpp 0.05 Vpp precise value of this factor – it depends on the feedback
elements. For accurate signal processing it is
The small bias current is especially important when
reasonable to use special kind if IC amplifier, known as
the source of signal exhibits large impedance
the instrumentation amplifier. Such amplifier is
(electrometer mode) – for example the piezoelectric
designed for measurement purposes, with precise laser
sensor, photodiode or pH electrode. Usually in low bias
trimmed resistors.
current opamps in the input circuit the FET transistors
are used. It is possible to obtain the bias current as
small as 40 fT what correspond with input impedance 4
The THD is usually calculated as ratio of rms value of the first nine
as large as 1015 . harmonics to rms value of total voltage.
74 Basic Electrical Measurements

The instrumentation amplifier is not a special kind of TABLE 3.9

The best parameters of instrumentationamplifiers.
operational amplifier. Operational amplifier is used as a
basis of various devices: amplifiers, filters, oscillators AD8429 AD8230 AD8220
G =100 G =100 G=100
etc. The instrumentation amplifier is almost exclusively
Bandwidth 1.2 MHz 2 kHz 120 kHz
designed for amplification. It should exhibits following Offset drift 0.1 V/C 50 nV/C 5 V/C
feature: differential input with large input resistance Bias current 150 nA 0.15 nA 10 pA
> 109  and large CMR enabling to process of small Gain error 0.3% 0.01% 0.2%
signal in presence of large common voltage; small error Gain(T) 5 ppm/C 14 ppm/C 3 ppm/C
or nonlinearity, small error of gain, small noises and Noise 1 kHz 1 nV/rtHz 240 nV/rtHz 14 nV/rtHz
Noise 1/f 0.1 Vpp 3 Vpp 0.8 Vpp
temperature zero drift. Table 3.7 presents performances
Nonlinearity 2 ppm 20 ppm 30 ppm
of two typical instrumentation amplifiers. CMRR 90 dB 120 dB 100 dB
TABLE 3.7
Rin 1.5 G 104 G
Main parameters of instrumentation amplifiers. Slew rate 22 V/s 2 V/s 2 V/s
AD620 AD620 AD8228
G=1 G =100 G=100 The differential amplifier presented in Figure 3.94
Bandwidth 1 MHz 120 kHz 110 kHz requires very precise selection of the resistors. Even
Offset drift 0.1 V/C 0.1 V/C 0.5 V/C small changes or differences significantly deteriorate
Bias current 0.5 nA 0.5 nA 0.4 nA the CMR ratio. Also if the resistance of the source is
Gain error 0.01% 0.15% 0.05% not fully symmetrical (what can be for example in case
Gain(T) 10 ppm/C 10 ppm/C 1 ppm/C of bridge circuit) the CMR ratio is also decreased.
Noise 1 kHz 9 nV/rtHz 9 nV/rtHz 8 nV/rtHz
Therefore for professional applications the differential
Noise 1/f 3 Vpp 0.3 Vpp 0.3 Vpp
Nonlinearity 10 ppm 10 ppm 5 ppm amplifier is constructed from two or three amplifiers
CMRR 90 dB 130 dB 120 dB [Zumbahlen 2008, Kitchn and Counts 2006], as it is
Rin 10 G 10 G 100 G presented in Figures 3.99 and 3.100.
Slew rate 1.2 V/s 1.2 V/s 2.5 V/s

All parameters influence the accuracy in different

R1
way. Table 3.8 presents the error budget analysis of Uout=(1+2 )(U1-U2))
U1 +
RF
general purpose IA given by a manufacturer Analog
Devices. -
R1 R2 R2
TABLE 3.8
Error budget of AD 620 IA for input values Rs = 350 , Uin = 100
mV, .gain 100, temp.; 25C [after MT-066] RF -
R1 R2
Error +
[ppm]
Offset Uos 55 V 550
Offset Ios 350   0.5 nA 1.8 - R2
Gain error 0.15% 1500 U2 +
Nonlinearity 40 ppm 40
CMRR error 120 dB, 1 ppm  5 V 50
1/f noise 280 nV 2.8
Total unadjusted error 2145
Resolution 42.8 FIGURE 3.99
Configuration of typical instrumentation amplifier.
Because gain, CMR error can be decreased by
calibration the remaining errors are considered as In the amplifier presented in Figure 3.99 only one
resolution. The dominating error is caused by gain error external resistor is used to adjust the gain. Such
– unfortunately this error increases with gain and for solution is recently often substituted by internal,
the same IA and gain G = 1000 it is 0.35%. Presented precise laser trimmed resistor adjusted by pin-strap.
in the table error 0.05% for G = 100 is the smallest The external resistor can have other properties than the
available. rest of internal resistors what can cause additional gain
Similarly as in the opamps case there are available error. Available are also amplifiers with fixed one/two
variety different amplifiers of specialized gains. The best solution is to use programmable gain
performances. Table 3.9 presents the data for very amplifiers PGA [Zumbahlen 2008, Kitchn and Counts
small zero drift (auto-zero function) and very s mall 2006].
bias current (FET input).
 Handbook of Electrical Measurements 75

R1 R1 RSET FILTER
Uout=(1+2 + ) (U1-U2)
RF R2 VCC
R1 GND
R2 R2 R1
OUT
RF -
VREF
+ AD8290
U2
-
U1 + FIGURE 3.102
Instrumentation amplifier designed to bridge circuit.

Figure 3.103 present other instrumentation amplifier

FIGURE 3.100 designed to cooperation with shunt resistor – for
Two opamps version of differential amplifier. current measurement. Due to high CMRR coefficient it
is enable to measure small shunt voltage drop in
Sometimes it is necessary to have differential output presence of high up to 80 V voltage. The amplifier
– for example to connect it to AS device. Figure 3.101 works in temperature range -40 - +125C with fixed
presents an example of the full input/output differential gain G = 20, small offset drift 0.1 V/C and 5 ppm/C
amplifier. gain drift. The circuit is protected before overvoltage
by the internal low dropout regulator LDO.

+
Uout-
AD8218
- IX
Uin+
+
+
- -
4-80 V

Uin- -
LDO
+
Uout+
FIGURE 3.103
FIGURE 3.101 Instrumentation amplifier designed to shunt current measurement.
Full input/output differential amplifier [Zumbahlen 2008].

On the market there are many various specialized 21.1k 380k

instrumentation amplifiers ready to use as measuring
IX
devices, as for example strain gauge bridge, 380k
thermocouple temperature meter, shunt current
measurement etc. They are often equipped with useful
tools as for example EMI filters. Figure 3.102 present 380k
270 V

the CMOS instrumentation amplifier AD8290 designed

to cooperation with sensors and bridge circuit. It is
20k AD629
equipped with precise excitation current source to
supply the bridge circuit or RTD temperature sensor. It
has fixed gain G = 50 with gain error 0.5%, gain
nonlinearity 0.0075%, low noise level and low FIGURE 3.104
Instrumentation amplifier designed to shunt current measurement in
temperature drift and high CMRR coefficient.
presence of high voltage.
76 Basic Electrical Measurements

Often the shunt voltage is measured in presence of when these both parts are supplied from different
much higher common voltage. In such case solution is sources. The difference between the ground potential of
to use separating transformer but it introduces both circuits can be as large as hundreds of volts.
additional errors. Other solution is to use more Without galvanic separation such difference can cause
expensive insulating amplifiers – described in next destruction of the amplifier (and of course could be
Section. Relative simple solution is to use voltage dangerous to the servicing personnel). Therefore in
dividers. Figure 3.104 presents the instrumental medical equipment it is indispensable to use the
amplifier known as difference or subtraction amplifiers isolation amplifiers. Galvanic isolation helps also in
with high CMR ration. The internal voltage dividers rejection of various electromagnetic interferences.
enable to measure the shunt voltage drops in presence
the common voltage as high as 270 V – with 
protection to 500 V. 50mV 300V

Rs Uout Rb
Uin
380V load Ro
REF
Iout
Uout

LOAD
FIGURE 3.106
Iout = Uout/Rs The application of the isolation amplifier for separation supply source
and receiver when the potential difference of grounding can be large.

FIGURE 3.105 In the isolation amplifier the connection between

Instrumentation amplifier with current output.
input terminals and output terminals does not exist (and
the same applies for input ground and output ground).
In industrial practice the output signal can be
Sometimes, also the supply sources are physically
transmitted and in this case the current output is
separated. The rejection of the isolation voltage UIM
preferred (as eliminating the increase of resistance
change of transmission wires). In the market there are (voltage between common parts of input and output
special instrumentation amplifiers known as 4-20 mA circuits) is described by IMRR – Isolation Mode
drivers (or current out drivers). Figure 3.105 presents Rejection Ratio
simple method to obtain current output in typical
 U  U IM
instrumental amplifier PGA [Zumbahlen 2008, Kitchn U out  K u U in   (3.127)
and Counts 2006].  CMRR  IMRR

3.6.3. Isolation amplifiers There are different tools of galvanic separation – the
There are circumstances, when it is necessary to use of transformer, capacitor, light or magnetic field. In
ensure the galvanic separation of the input and output all these methods relatively easy is to separate AC
circuit. For example, when we measure very small signals – but from isolation amplifiers is required to
differential signal superimposed on large common separate both AC and DC components.
signal. Such a case is presented in Figure 3.106 where It is convenient to use a small transformer as the
the relatively low voltage (with the level of tens mV) isolation device. Because transformer does not
across the shunt resistor Rb is measured in the presence transform DC signals it is needed to use the modulator
of relatively high voltage (several hundred V), which is (signal is transformed as the magnitude varying carrier
used for supplying the load Ro. Between the ground of AC signal) and next the demodulator to recover again
the indicating instrument connected to the amplifier the DC signal. Such idea is realized in the isolation
and the ground of the supply source there could be a amplifiers of Analog Devices (Figure 3.107). In the
voltage difference dangerous for the servicing AD215 model presented in Figure 3.107 (two-port
personnel (and also the amplifier could be damaged isolator) the power section is not isolated from the
due to this large potential difference). output. In other model AD215 (three port isolator) both
Also dangerous is the situation when the input circuit input and output are isolated form power section by
is distant from the rest of the circuit and even more additional transformer.
 Handbook of Electrical Measurements 77

AD215 3652 Uout=Uin (R2/R1)

IN OUT
I2
- buffer
+ I1 R2
modulator demodulator filter
- D1 D3 D2
+ Uin R1 Uout
- I3 -
A1 A2
+15V Iin + +
isolated -15V
power
DC oscillator
supply 430kHz

FIGURE 3.109
FIGURE 3.107 Isolation amplifier with light separation – model 3652 of Burr-Brown.
Isolation amplifier with transformer separation – model AD215 of
Analog Devices. The light separation could be an excellent tool for
isolation amplifiers. Unfortunately the conversion is
Presented in Figure 3.107 isolation amplifier with large nonlinearity error. Therefore two strategies
provides 1500 V common mode voltage protection. are used: one with feedback to eliminate this
With fixed gain G = 1 it converts voltages in bandwidth nonlinearity, the second with conversion to digital form
120 kHz with low nonlinearity error 0.005% and low where influence of nonlinearity is negligible.
harmonic distortion 80 dB by 1 kHz. The first principle is utilized in the isolation
amplifier developed by Burr-Brown – model 3652
ISO124 (Figure 3.109). The LED diode D3 illuminates two
identical photodiodes D2 in the output circuit
(converter current/voltage) and D1 in the input circuit
(converter voltage/current). The diode D1 is in
IN OUT feedback circuit, which enables to minimize the
- -
A1 A2 nonlinearity errors. Since both diodes are illuminated
+ + identically therefore I1 = I2 = Iin.
S/H
Similar principle isolation amplifiers with two
OSC
infrared phototransistors developed Clare Inc.
Company. The LOC110 linear optocoupler enables to
convert voltage with nonlinearity error less than 0.01%,
FIGURE 3.108 THD 87dB and bandwidth 200 kHz.
Isolation amplifier with capacitor separation – model ISO124 of Texas
Instruments.
input isolation output
boundary
Another principle of galvanic separation is applied in
the isolation amplifier of Texas Instruments – for voltage S
filter
regulator modulator
example in model ISO124 presented in Figure 3.108. In
this case a capacitor with capacitance 2 pF is used as clock decoder voltage
encoder
the isolation device. The input signal is converted to generator and D/A regulator
AC signal using integrating analogue to digital circuit
A1. Next, this signal transmitted by the capacitance is LED drive detector
circuit circuit
again converted to the DC signal using integrating
amplifier A2. Additional sample-and-hold circuit is
used to obtain the output signal without the ripple FIGURE 3.110
voltage inherent the demodulation process. Isolation amplifier with light separation – model HCPL-7850 of
Presented isolation amplifier enables conversion of Avago Technologies.
the voltage with gain fixed to G = 1 in the presence of
the common mode voltage 1500 V (IMR coefficient is Figure 3.110 presents the isolation amplifier based
equal to 140 dB). The nonlinearity error is better than on the second principle. The input signal id converted
0.01%, zero drift better than 200 V/C and frequency to digital form by sigma-delta converter. Next digital
bandwidth is 50 kHz. signal is transmitted by the optocoupler and the analog
signal is used by applying digital-to-analog converter.
78 Basic Electrical Measurements

To this noise budget also noise generated by the

circuit resistors should be added. For resistor R this
shield NiFe noise in temperature T for the bandwidth B is:

U R  4kTBR (3.129)
OUT
where k is the Boltzmann constant K = 1.38 10 -23 J/K.
planar
coil At room temperature this noise approximately is
GMR 4nV / Hz  R where R is in k. Figure 3.112
sensor IN presents the noise of one of the best low noise
amplifiers on the market.

noise [nV/rtHz]
FIGURE 3.111
Isolation amplifier with magnetic separation – model model developed AD8429
by Nonvolatile Electronics Inc. 100
G=1

Figure 3.111 presents the magnetic galvanic 10 G=10

separation IC device developed by Nonvolatile
G=100
Electronics Inc. The main advantage of magnetic 1
separation is that it correctly converts both DC and AC G=1000

signals (transformer and capacitor convert only AC 0.1

signals). The current in the planar coil generates the 1 100 10k f [Hz]
magnetic field detected by the GMR (giant
FIGURE 3.112
magnetoresistance) sensor. Signal of the sensor is the Voltage noise of low-noise instrumentation amplifier AD8429.
source of the current generating the balancing feedback
magnetic field. The GMR sensor only detects the The noise consists of unlimited number of signals
presence of magnetic field and thus its nonlinearity with various frequencies distributed randomly (white
does not influence the linearity of the device. This noise) and sometimes the noise itself is larger than the
device can operate in analog or in digital mode with useful input signal. If we know the frequency of this
transmission speed even above 1 GB/s. input signal we can separate this signal from noises
using a selective amplifier or filter. The use of the
3.6.4. Amplifiers of very small signals
selective amplifiers (tuned to certain known frequency)
From the data presented in Table 3.7 we see that the
is limited because it is difficult to ensure stable
resolution of general purpose AD620 instrumentation
amplification. Small deviation from the resonance
amplifier is limited by following factors:
frequency causes large variation of the amplification.
- voltage input noise: UNi = 10 nV/rtHz,
For that reason recently the best tool for the
- voltage output noise: Uno = 70 nV/rtHz
amplification of small AC signals is the lock-in
- G = 100 1/f voltage noise: 0.3 V,
amplifiers (Scofield 1994, Blair and Sydenham1975).
- current input noise: INi = 1 nA/rtHz,
In the lock-in amplifiers a phase sensitive detector is
-1/f current input noise: 10 pA,
used as the selective element. This detector selects
- offset drift of about U = 1 V/10C. from the input signal only these components that have
Thus in the case of AC+DC small signals the the same frequency as the reference voltage. When the
temperature zero drift and 1/f noise are predominating ordinary selective amplifier exhibits the Q factor of
and in this case the resolution is on the level of V. In about 50 - 100, the lock-in amplifier can exhibit this
the case of only AC signals it is possible to eliminate factor as large as 100 000 (it is possible to select the
the influence of zero drift and only noises influence the signals of 10 kHz in the frequency bandwidth 0.01 Hz).
resolution. In this case the resolution is no the level of In this way it is possible to select from the noises
tens of nV. signals of the level below nV.
The total noise when the input source has Resistance R s The synchronous detector can be realized as special
is: device (as presented in Section 3.5.d) but the same
effect we can obtain using simply a multiplier.
UN  U Ni   U No / G    Rs I Ni 
2 2 2
(3.128)
 Handbook of Electrical Measurements 79

bandpas lowpass In
filter PSD filter
+
Uin A ABcos Uout -
-
noise +
B generator

phase shifter 100 dB +

Uref attenuation - Out
trigger AD630

oscillator

FIGURE 3.113 In
The operation principle of the lock-in amplifier.

Figure 3.113 presents the operation principle of the

lock-in amplifier. After preliminary amplification the
Out
input signal is connected to the synchronous detector,
where it is rectified with respect to the frequency and
phase of the reference signal. Next, this signal is
filtered by low-pass filter. FIGURE 3.115
An example of the lock-in amplifier based on the AD630 device of
Analog Devices. Output signal when to the sinusoidal input signal are
fin = fref fin = 2fref fin = 3fref added noises 100 dB larger. (From Data Sheet of AD630).

In applying of lock-in amplifier it is necessary to

have the reference signal. Often this signal is in natural
case available in all carrier amplifiers where the high
frequency signal is modulated by low frequency small
Tref signal. The carrier signal can be used as the reference
f signal.
The amplitude modulation is realized by multiplying
high and low frequency signals:
FIGURE 3.114
The signal in the “window” of period of rectangular reference signal. U mod  U in sin t U ref sin  t 
(3.130)
In Figure 3.14 is presented the frequency effect of U inU ref cos      t  cos      t 
synchronous rectifying. If input signal has the same
frequency as the reference signal its value is the largest. Such multiplication is realized by every bridge
For second harmonics (1 octave) this signal is zero. For circuit as it is presented in the example of strain gauge
comparison the first order RC filter decreases the signal bridge (Section 3.4.c) or in the Figure 3.90 because:
by 6dB/octave). This selectivity effect is amplified by
R
the phase relations because as it is presented in Figure U out  U o sin  t  sin t (3.131)
3.83 the “in phase” signal is the largest but for the R
phase shift by 90 the signal is zero. It is assumed that
by using the lock-in amplifier it is possible to recover Sometimes for obtain the modulation effect the
signal buried in 60 dB larger noises. multiplier device can be used. Figure 3.116 presents the
The making impression results of experiment are example when the reference modulated signal is added
presented in Figure 3.115. As synchronous rectifier was artificially. Because output signal of DC SQUID device
used monolithic circuit AD630 of Analog Devices. To is small and noisy usually the supporting magnetizing
the signal was added the noise larger of 100 dB. It was winding is added supplied by the oscillator. This
possible to recover the measured signal in presence of additional signal is next modulated by the measured
noises 100 000 times greater. SQUID signal.
80 Basic Electrical Measurements

signals (measured and reference) are converted to the

SQUID DC synchronous digital signals and as synchronous detection the digital
detector multiplication is used.

Uin filters

oscillator
Yout
90o

FIGURE 3.116
Application of the lock-in amplifier in DC SQUID magnetometer.

The professional lock-in amplifiers are equipped

Xout
with internal precise PLL oscillator and it is possible to Uref
tune this oscillator to find the main frequency phase
PLL OSC
component of amplified signal. shift
To demodulation of the carrier modulated signal
usually the multiplier is used. Indeed according to the FIGURE 3.117
relationship (3.115) the phase sensitive rectifier also Functional block diagram of typical lock-in amplifier.
performs multiplication of two signals.
The main problem in the amplification of small DC
(3.132) signals is the temperature zero drift of the amplifier,
U dem  AU in sin t  U ref sin  t U ref sin  t  because it is very difficult to separate this zero drift
signal and the useful DC signal. The temperature zero
A 2
U inU ref  sin t  sin  2    t  sin  2    t  drift is caused by non-ideal technology – even small
2 differences in various parts of the circuit can be the
source of DC offset signal. This effect is minimized by
It is relatively east to remove high 2 frequency laser trimming to the level of 1 – 50 V/ C, but even
components by using the low-pass filter and to recover such small offset is still limiting the amplification of
input low frequency signal. To correct perform carrier small DC signals.
modulation it is recommended if  > 5. Looking at To decrease the zero drift influence there are two
the Figure 3.112 by using the amplifier near carrier main techniques. The first one is based on the
frequency we escape from the large 1/f noises band. conversion of the DC signal to the AC signal. Then it is
Figure 3.117 presents typical functional block possible to separate DC zero drift and AC useful signal.
diagram of lock-in amplifier. Usually two multipliers The AC signal can be amplified and converted back to
are used as synchronous detectors for two component the DC one. This technique is called a chopper
of measured signal: in-phase and orthogonal one. For amplifier - the conversion to the AC signal is achieved
measurement of orthogonal component the 90 phase by very fast connecting and disconnecting the signal to
shifter is used. Lock-in amplifiers often are equipped the input.
with internal oscillator to use in the case when The second technique employs the auto-zero
reference signal is not available. principle. The amplifier is periodically disconnected
Typical lock-in amplifier, for example model 850 of from the input, then the input is short-circuited to
Stanford Research System has following parameters: ground and the zero component is detected. The auto-
sensitivity: 2 nV – 1 V; input noise: 6 nV/rtHz for 1 zero technique enables to eliminate the zero drift down
kHz, gain accuracy 1 %, frequency bandwidth 1 mHz – to the level of 5nV/ C.
100 kHz, phase resolution 0.001, dynamics 100 dB, The principle of operation of the chopper amplifier
demodulator time constant 10 s – 30 ks. is illustrated in Figure 3.118. The signal is switched
The analogue lock-in amplifier exhibits several with frequency of up to tens of kHz. To the amplifier
drawbacks: it is necessary to use an analog low-pass are delivered pulses with amplitude modulated by the
filters, the dynamics is rather poor, and the bandwidth input signal. After amplification this signal is
is limited. Therefore, recently on the market are demodulated by the synchronous switching (or using
available digital lock-in amplifiers. In such device both the phase-sensitive detector).
 Handbook of Electrical Measurements 81

chop1 chop2
Uin
R2

Uin e1 R1
e2 e3 e4
Uout Uout

C
CF
oscillator chop3

FIGURE 3.118
The principle of operation of the chopper amplifier. FIGURE 3.119
The chopper amplifier with autocorrection feedback.
If the input signal is uin(t) and the commutation
frequency is o, then the signal at the input of amplifier The chopper amplifier exhibits several drawbacks:
is the limitation of frequency bandwidth (it is limited by
the frequency of the carrier signal), necessity of the
1 2  signal filtration. For that reason the chopper technique
e1  uin t   cos ot 
2 2
cos 3ot  cos 5ot...
2  3 5  is often substituted by auto-zero technique. But last
(3.133) time it can be observed renaissance of this technique
{Wong 2011, Moghimi 2011]. Analog Devices
If we describe the zero component as Uo and the zero developed chopper amplifier of next generation -
drift as Uo and the analysis is limited to the first presented in Figure 3.119. This amplifier has high
harmonics (we can eliminate higher harmonics using chopping frequency of 200 kHz and special correction
the filter at the output) the output voltage of the feedback loop helping in remove of ripples. As result
amplifier is the amplifier ASA4528 has offset drift 2 nV/C and
noise 5.6 nV/rtHz – the best parameters currently
R2 1 2  available.
e2  t    uin  t    cos o t  
R1  2  
(3.134) Ux K1
 R2   R2 
 1   U o   1   U o
 R1   R1  K2
-
The DC components are blocked by the capacitor C,
thus the e3 signal is +

R 2 
e3 t   uin t  2  cos o t  (3.135) K3 C0
R1   
FIGURE 3.120
After the second commutation the signal is The principle of operation of the auto-zero amplifier.

1 2 
e4 t   e3 t   cos o t 
2 The operation principle of auto-zero method is
cos 3o t... (3.136)
 2  3  presented in Figure 3.120. The amplifier works in two
phases. In the first phase the input is shortened, the
Thus after the filtration we obtain the output signal amplifier is amplifying its own zero drift voltage and
the result is stored in capacitor Co. In the second phase
uout t   uin t  2
R 4 the input voltage is connected to amplifier and the
 U o2 (3.137) voltage stored on capacitor is subtracted from input
R1  2
voltage.
The Uo2 component is the zero drift of the last The real auto-zero amplifier is more complex to
amplifier and it is negligible because it is added to the avoid break in the amplifying process and to improve
large output signal. quality of amplifier (Figure 3.121). The amplifier is
82 Basic Electrical Measurements

now composed of two amplifiers – wideband A and connected to third input of amplifier A. To third inputs
nulling B. These amplifiers heave three inputs and of both amplifiers are connected voltages stored on
working as: capacitors CM1 and CM2.
As result of switching between two phases the output
U out  A U1  U 2   U N (3.138) voltage is (AD 8551):

Uout  AUin  A UOSA  UOSB  (3.139)

U1
+
UOSA A Thus the input voltage is amplified by a factor of
U2 A, while the zero drifts are amplified by the factor  -
- - Uout
times smaller. The resultant zero drift component UOS
B UOSB B CM2 is

A
+ U OSA  U OSB
B U OS  (3.140)
-
CM1 
-
A
Because  is large the resultant zero component is
U1 reduced to the nV level. Thus these two steps of the
+ auto-zero operation significantly reduce the zero drift
UOSA A components without limitation of the frequency
U2
- - bandwidth and with the possibility of the differential
mode of amplification.
B UOSB B CM2 Both methods of decrease of the zero drift –
+
chopping and auto-zero have pros and cons [Moghimi
A B 2011]. Auto-zero exhibits low zero drift at the expense
CM1
- -
of higher low frequency noise due to aliasing the
A wideband noise into the low frequency band. The
chopper amplifier has limited bandwidth and ripple of
the chopping frequency. Therefore it was developed the
FIGURE 3.121
The principle of operation of the auto-zero amplifier – model AD8551 new low zero drift amplifier AD 8628 combining these
of Analog Devices. both technique.
TABLE 3.10
The real auto-zero amplifier is more complex to The performances of low zero drift amplifiers.
avoid break in the amplifying process and to improve
AD4051 AD8551 AD8628
quality of amplifier (Figure 3.121). The amplifier is
Bandwidth 125 kHz 1.5 MHz 2.5 MHz
now composed of two amplifiers – wideband A and Offset drift 20 nV/C 50 nV/C 2 nV/C
nulling B. These amplifiers heave three inputs and Offset 2 V 1 V 1 V
working as: Noise 1 kHz 95 nV/rtHz 42 nV/rtHz 22 nV/rtHz
Noise 1/f 2 Vpp 1 Vpp 0.16 Vpp
U out  A U1  U 2   U N (3.138) Current noise 100 fA/rtHz 2 fA/rtHz 5 fA/rtHz
AD4051 – chopper, AD 8551 – auto-zero, AD8628 – chopper +auto-
zero
Two amplifiers are commuted by switches A and
B. When switches A are closed realized is phase A It should be noted that modern amplifiers due to
– auto-zero phase. When switches B are closed the improvement of technology have now very low zero
second phase is realized – the output phase. voltage drift even without more complex auto-zero or
In phase A the amplifier A is connected to input chopper techniques. For example ultra-precision opamp
voltage while the amplifier B has shortened input. Thus OP 177 of Analog Devices exhibits noise 11 nV/rtHz
amplifier B simplifies its own zero voltage U0SB and and zero drift 100 nV/C.
amplified voltage is stored on capacitor CM1. To the
third input of amplifier A is connected voltage 3.6.5. Charge amplifiers (electrometers)
previously stored on the capacitor CM2. The operational and instrumentation amplifiers
In phase B the input voltage is connected to both exhibit usually very large input resistance. For example
amplifiers. The output voltage of amplifier B is general purpose IA AD620 has input bias current 1 nA
 Handbook of Electrical Measurements 83

and input resistance 10 G. One of the commonly used aA (1a (atto) = 10-18). To put that into perspective
methods to increase resistance is to use in the input consider that the voltage of about 10 V causes in a
FET transistors. For example general purpose opamp typical insulator a current of several pA.
TL071 of Texas Instruments has bias current 65 pA and We can obtain very large input resistance by special
input resistance 1012 . technology (special kind of insulation and connections)
The special designed amplifiers with FET transistors and also by using special electrical circuits. As the
can have input resistance as large as 1013 . For amplifier with large input resistance we can use
example opamp AD549 of Analog Devices has bias current-voltage transducers presented in Figure 4.123.
current only 60 fA and input resistance 1013 . Figure This circuit is also known as transimpedance amplifier.
3.122 presents the circuit of FET transistors For this circuit we can write that
instrumentation amplifier AD8220 with bias current
600 fA and input resistance 1013 .
U out   I in R (3.141)

In the circuit presented in Figure 3.123 the small

capacitor C is sometimes added to prevent gain peaking
[Pallas-Areny, Webster 1999]. The time constant
depends on RC and for large R it can be as large as
OUT
several seconds. Therefore this circuit exhibits low-
IN+ IN- pass filtering characteristics with relatively small cut-
off frequency. The application of very large T
resistance R can be inconvenient and therefore the
modified amplifier circuit presented in Figure 3.124
can be recommended [Pallas-Areny, Webster 1999].

FIGURE 3.122
Simplified circuit of the FET transistors input instrumentation R
amplifier AD8220.
R1
Iin R2
Even such large input resistance in certain -
applications can be insufficiently large. For example
some piezoelectric sensors, photoelements, pH sensors + Uout
exhibit very large input resistance close to the insulator
state and conventional amplifiers are in this case short-
circuit of the sensors output. FIGURE 3.124
An example of transimpedance amplifier with extended input
C resistance.

For this circuit presented in Figure 3.124 we can write

R the following dependence
Iin
-   R   R 
U out   I in  R2  R1  2    I in R1  2  (3.142)
  R1   R1 
Cp + Uout
Thus we obtain multiplication of the input resistance
by the ratio R2/R1. The transimpedance amplifier with
the input signal source in form of capacitor is
FIGURE 3.123 sometimes called the charge amplifier (Figure 3.125).
An example of transimpedance amplifier (Cp – capacity of the cable).
In this case in the feedback circuit resistance R is
Special kinds of amplifiers or instruments with substituted by capacity C (parallel resistance R F can be
extremely large input resistance, for potential added to correct frequency characteristic). For this
measurement are designed. They are called circuit we can write that
electrometers. For example, Keithley Company
Q Q
developed the electrometers with the input resistance of U out   (3.143)
10 P (1 P (penta) = 1015) and input current 400 C F  C  C F / K u C F
84 Basic Electrical Measurements

We obtain the transducer of the charge with Figure 3.126 presents the circuit of the amplifier
negligible influence of the cable capacitance C (when realizing the integration of the input signal – the
the gain of amplifier Ku is sufficiently large). integrator. For ideal integrator circuit we can write the
following dependence
RF t0 T
uout t     uin t dt  U 0
1
R R1 CF (3.144)
R1C
- t0

C Thus the circuit realizes integration with time

+ Uout
sensor constant R1C and U0 is the voltage across the capacitor
C before start of the integration.
The presence of the zero voltage and associated zero
U
drift is quite significant problem of the operation of
Q
20log integrator circuit. It depends on quality of the amplifier
(low bias current, low zero drift) but also on the quality
CF of capacitor [Stata 1967, Scholes 1970]. Especially this
problem is troublesome in when input signal contains
log DC components. In such case usually it is necessary to
perform discharge of the capacitor (by shortening it)
1/RFCF 1/R1C and correct zero. To obtain low frequency operation the
capacitor leakage resistance R4 should bet very large. In
FIGURE 3.125 the case of AC signal the frequency bandwidth and
An example of the charge amplifier and its frequency characteristic. problems with zero drift can be limited by inserting
additional resistor R4 (Figure 3.127).
Figure 3.125 present the typical frequency
characteristic of the charge amplifier. The time R4
constant of this transducer depends on the RF CF. For Ku
example for RF = 10 G and CF = 100 pF the low cut-
C R3 Kuo (R4)
off frequency is 0.16Hz. The high cut-off frequency
depends on R1 C and for example for R1 = 50  and C R1
= 200 pF it is 1.6 MHz. - (R3)
3.6.6. The mathematical amplifiers
+
By appropriate application of the feedback we can log
create various transfer characteristics of amplifier and R2
in this way we can realize various mathematical
functions as: integration, differentiation, logarithm,
exponential dependence, etc.
FIGURE 3.127
The practical circuit of the integrator.
C
We can set low and high cut-off frequencies by
R1 appropriate choice of resistance R3 and R4. Figure 3.127
presents the transfer characteristic without (dashed
- line) and with correction resistors.
Uin The integrator is a very important circuit in the
+ Uout measurements. Figure 3.128 presents the main
application of this circuit. In magnetic measurements it
can substitute old-fashioned moving coil fluxmeter
device for measurement of the change of magnetic flux
 (it practically measures charge – see Figure 3.125).
FIGURE 3.126
The circuit of the ideal integrator. If it is connected to the coil with n turns the output
voltage is:
 Handbook of Electrical Measurements 85

n The ideal circuit presented in Figure 3.129 exhibits

e  (3.145) inherently instability because gain is increasing with
RC
frequency (Figure 3.130). Therefore usually additional
In the case of AC magnetic circuit the voltage elements are added to limit the frequency.
induced in the secondary winding is proportional to Ku
dB/dt. Thus to recover the flux density value it is R4 R1

9
necessary to integrate this voltage (Figure 3.128a). Kuo

12
3.
g.
Fi
-

in
dB/dt

as
a) R3 (R4) (R3)
+
B
R2 log
1

b) FIGURE 3.130
The practical circuit of the differentiator.

IN OUT The risk of instability and increase of high frequency

noise can be prevented by inserting additional resistors
B R3 and R4 (Figure 3.130).Thus this circuit operates as
differentiator only to the frequency 1.
FIGURE 3.128
The main applications of the integrator.
V
Integrating circuit is the main part of two analog to
potential [mV]

digital converters – dual slope and sigma delta. In these

converters is utilized feature of integrating circuit that
for stepwise change of input signal the output signal is
increased linearly versus time. dV/dt

R1
C acid/base

- FIGURE 3.131
Uin The derivative as the method of detection of the deflection point.

+ Uout The differentiator is not as commonly used as

integrator. It can be used as the high-pass filter. Figure
3.131 present the possible application of the
differentiator to detect the deflection point in the
potentiometric titration. Similarly it can be used to
FIGURE 3.129
The circuit of the ideal differentiator. detection the peak value of the signal.
Between the collector current Ic and base-emitter
By simply interchanging positions of resistor and voltage UBE of silicon bipolar transistor is exponential
capacitor we obtain the differentiator circuit. For an dependence:
ideal differentiator circuit (Figure 3.129) there is:
I c  I s eU BE / UT (3.147)
du t 
uout t    R1C in (3.146) where k is the Boltzman constant, q is the electron
dt
charge, T is temperature, Is is the collector reverse
saturation current and UT = kT/q.
86 Basic Electrical Measurements

Thus by appropriate connection of such transistor It is possible to correct the temperature characteristic
into the feedback circuit of operational amplifier we of the whole amplifier by appropriate choice of the
can obtain logarithmic log amp or exponential temperature characteristic of the RT element.
amplifier antylog amp (Figure 3.132).
R
T
R CI
- -

+ lnUin + eUin Iin

Iref

FIGURE 3.132 out

The principle of operation of logarithmic and exponential amplifiers. 10V Uout=-Alog(Iin/Iref)
For the logarithmic amplifier we can write the in
following dependence:
10nA 1A 100A
I U
U out  UT ln in  UT ln in (3.148)
Is RI s

U out  I s R  eUin / UT (3.149) FIGURE 3.134

Simplified functional diagram of 4127 logarithmic amplifier of Texas
Instruments.
where UT = kT/q is thermal voltage - in ambient
temperature UT  25 mV. Figure 3.134 presents the circuit of logarithmic
amplifier model 4127. This amplifier converts
log(Iin/Iref) to output voltage. Depending in the direction
R of the input current it can be connected directly to the
- input or by current inverter CI. The voltage input signal
Uin is connected with serial resistor. The amplifier is
+ Uout
R2 equipped with internal source of reference current. By
appropriate connection it can be transformed to antilog
R amplifier. The accuracy of the amplifier is 0.5% and
- RT
frequency bandwidth to 250 kHz.
Uref + 10dB 10dB 10dB 10dB 10dB

In limiter
FIGURE 3.133
log out
An example of the logarithmic amplifiers.

The transfer characteristic Uout = f(Uin) depends on

the temperature (also current Is depends on the FIGURE 3.135
Successive detection log amplifier.
temperature). Therefore, the practical circuit of the
logarithmic amplifier is more extended, as it is
presented in Figure 3.133 (Tran Tien Lang 1978). With Logarithmic amplifiers are commonly used as light
correction the transfer characteristic is described as diodes amplifier in fiber-optic data transmission for
compress and demodulate RF signal. In such case the
RT  R2 U amplifier presented in Figure 3.134 has to small
U out   UT ln in (3.150) bandwidth. Much higher bandwidth is possible to
RT U ref obtain in the successive detection logarithmic amplifier
 Handbook of Electrical Measurements 87

presented in Figure 3.135. Basing on this principle such circuits two techniques of multiplying are applied:
AD641 demodulating logarithmic amplifier of Analog Gilbert transconductance multiplier or log/antilog
Devices operates in bandwidth from DC to 250 MHz. operation.
The logarithmic amplifiers can be used for
compression of the signals or linearization of the I1 I2
transfer characteristic (when the characteristic of the
sensor is exponential). Using the logarithmic amplifier Ic3 Ic4 Ic5 Ic6
it is easy to perform many mathematical operations
because:
Ux
X
log X  log Y  log XY ; log X  log Y  log
Y (3.151) Ic1 Ic2
2log X  log X ; 0.5log X  log X
2

After operation it is sufficient to perform antilog Uy

operation to realize multiplication or squaring. Basic on
this principle operates true rms converters (see Figure I0
3.77) or multipliers (later in this Section).

FIGURE 3.137
The transconductance Gilbert cell.
R
- Designed by Gilbert in 1968 (Gilbert 1968, Gilbert
Uin + 1972) the multiplier circuit is still (with modifications)
Uout used in analogue semiconductor devices. Its main
advantage is that it is easy to implement such a device
R2 Uref R into the integrated circuit (the same semiconductors as
- the rest of the circuit), large frequency bandwidth and
what is important it can be implemented as four
+ quadrant multiplier. For the Gilbert multiplier
RT presented in Figure 3.137 we can describe currents as:

I0 I0
I c1  ; Ic2 
1  exp  U y / U T  1  exp U y / U T 
FIGURE 3.136
An example of the antylog amplifiers. I0
I c3 
Figure 3.136 present the example of exponential 1  exp  U x / U T   1  exp  U y / U T  
amplifier. For this antilog circuit the transfer I0 (3.153)
I c4 
characteristic is described as: 1  exp U x / U T   1  exp  U y / U T  
U ref  q RT  
I0
U out  R1 exp  U in  I c5
1  exp U x / U T   1  exp U y / U T  
(3.152)
R  kT RT  R2 
I0
I c6 
1  exp  U x / U T   1  exp U y / U T  
One of the most important devices in analogue signal
processing is the multiplier, enabling to perform
various operations: multiplication, division, square,
root, trigonometric functions, rms calculation, electrical and
power calculation, phase sensitive modulation or
demodulation. I1  I 2  I c3  I c5   I c4  I c6    I c3  I c6    I c4  I c5 
Recently, there are available various monolithic Ux Uy
multipliers with a multiplying error not larger than  I0 tanh tanh (3.154)
0.1% and frequency bandwidth up to tens of MHz. In
2UT 2U T
88 Basic Electrical Measurements

For small value of Ux,Uy < UT we can assume tanh(x) proposed by Analog Device (Data Sheet of AD534) are
= x+x3/3+…x and presented in Figure 3.140.

I0
 I  I1  I 2  U xU y  K U xU y (3.155) X1 WY X1 WY
4UT2 X2 (x1-x2)(Y1-Y2) X2 10V(Z1-Z2)
+Y1
10V (X1-X2)
Y1 Z1 Y1 Z1
Y2 Z2 Y2 Z2

multiplier divider

X1 WY
X1 WY A
X2
X2 A2-B2
10V(Z1-Z2) +X2
B 10V
Ux Uy Y1 Z1 Y1 Z1
Y2 Z2 Y2 Z2

square-rooter difference of squares

X1 WY
X X1
X2 WY
X Ysint X2 Y
Ysint 10V
Y1 Z1 1+Y
Y1 Z1
FIGURE 3.138 Y2 Z2 Y2 Z2
The four quadrant multiplier AD534.

Figure 3.138 present the circuit 1 MHz, 0.25% linear AM modulator bridge-linearization
accuracy multiplier AD534. Figure 3.139 presents its
functional diagram. FIGURE 3.140
Various examples of the applications of a typical multiplier device
(proposed by Analog Devices).
AD534

+Uz reference It is possible to design a multiplier as the

voltage
Uref
log/antilog principle presented in Eq. 3-151. An
-Uz and bias example of such multiplier is presented in Figure
X1 + 3.141.
X2 - R2
translinear R
Ux
Y1 multiplier LOG Uout
+ element Uref -
Y2 - + UA EXP
Out + UC
- Uy R
Z1 +
0.75 atten LOG
Uref
Z2 - UB

FIGURE 3.141
FIGURE 3.139 The log/antilog multiplier.
The functional diagram of multiplier AD534.
Ux Uy
The multiplier device enables calculation of the U A  K1 log , U B  K 2 log (3.157)
U ref U ref
following equation

U out 
 X1  X 2  Y1  Y2   Z 
and assuming that K1=K2=K
1  Z2 (3.156)
10V U xU y
U C   U A  U B    K log 2
(3.157)
By appropriate connection of the multiplier it is U ref
possible to realize various operations. Several examples
 Handbook of Electrical Measurements 89

and at the output Figure 3.143 presents the circuit of log-antilog

multiplier AD538. Beside typical multiplier functions it
U xU y is possible to use it as logarithmic amplifier. In
U out  U ref 10 U c / K 3  (3.158) comparison with transconductance Gilbert cell
U ref
multiplier AD534 the multiplier AD538 has narrower
bandwidth to 400 kHz, larger error equal to 1%. As
The multiplier based on the logarithmic function AD534 enables to multiply in 4-quadrant the AD538
usually realizes the following equation operates only in single quadrant.
M The operational amplifier allows performing the
U  operations of addition or subtraction. An example of
U out  U y  z  (3.159)
such device is presented in Figure 3.144.
Ux 

Figure 3.142 presents the operational diagram of a U1

typical log/antilog multiplier. The parameter M is set R2 R0
by connection of suitable resistors. U2
R3
U3 - Uout
Ix lnX
LOG logZ-logX +
Ux
S
Iz
LOG
Uz lnZ
FIGURE 3.144
M(lnZ-lnX) M(lnZ-lnX)+lnY The operational amplifier used for addition of the signals.

The output voltage of adder presented in Figure 3.144

0,2<M<5 S EXP is:

Uz M U U U 
Iy Uout=Uy U out    1  2  3  R0 (3.160)
LOG Ux  R1 R2 R3 
Uy lnY

FIGURE 3.142
If all input resistors are the same we have averager
The operational diagram of typical log/antilog multiplier – model device.
AD538 of Analog Devices.
Uout
Uy
Ux
U

- + - + U1 Uout Uin
+
U2 -
+

U 1> U 2 U 1< U 2
-
- +
- +
Uref
FIGURE 3.145
The operating principle of the comparator.

Uz Uout In the measurements often occurs the necessity of

FIGURE 3.143
comparison of two signals. The devices called
The log/antilog multiplier – model AD538 of Analog Devices. comparators realizes the following function
90 Basic Electrical Measurements

U out  1 for   U   U   0 in the passband frequency characteristics5 and the

 (3.161) transition between the passband and stopband is not
U out  0 for   U   U   0 vertical (there is a finite transition frequency
bandwidth). The width of transition band depends on
where output signal are logical 0 or 1 corresponding the dynamic of the signal and order of the filter).
with  Uout.
Figure 3.145 illustrates the operating principle of the
G(f)
comparator device. Theoretically, every operational passband ripple
amplifier without feedback acts as a comparator. But in
the typical operational amplifier time of switching is
relatively large – several s. Therefore, there are transition
available specially designed devices – comparators band
with switching time of several ns. If it is necessary to passband ideal characteristic
real characteristic
perform the comparison very fast, for example for
detection of zero crossing by measurement of phase stopband
shift, the comparator is very useful. For slow processes, stopband attenuation
for example monitoring of the level of the signal, f
ordinary operational amplifier is sufficient. passband fo fs stopband
Comparators play a very important role in digital signal cutoff frequency cutoff frequency
processing, as main part of analog to digital converters.

FIGURE 3.147
3.7 The analog filters The transfer characteristic of the real lowpass filter.
The analog filters are used mainly to pass only
selected part of the frequency bandwidth. By using the The performance of the filter is described by the
lowpass filters LP we can reject the high frequency amplitude and phase frequency characteristics (Bode
interferences (for example radio frequency characteristics) – Figure 3.148. The ideal amplitude
interferences). And vice versa by using the highpass characteristic should be flat in the passband till cut-off
filter HP we can eliminate the industrial frequency frequency (attenuation 0 dB). For real characteristic we
interferences (if our signal is in the bandwidth above assume that the passband is for the attenuation 0 – 3dB.
these frequencies). Also, we can separate the useful As the stopband we assume the frequency band where
signal from the interference signals by using the the attenuation is larger than the assumed value (for
bandpass BP filter. And inversely we can use the example 100 dB).
bandreject filters. Exist also all-pass filters – used as a The slope of the characteristic in the transition band
phase shifters. depends on the order of the filter – for the first order
filter it is 20 dB/decade (6 dB/octave6), for the second
order filter it is 40 dB/decade (12 dB/octave) etc.
Uout The phase frequency characteristic is also important
because incorrect phase performances mean that the
filter introduces signal distortions (various phase shift
of harmonics). To obtain correct phase performances it
is required that the phase varies linearly with the
LP BP HP frequency. Usually the phase characteristic is not linear
in the whole bandwidth (see Figure 3.148). Moreover it
f is very difficult to ensure both performances
simultaneously: the filters with excellent amplitude
characteristic exhibit poor phase linearity and vice
versa.
FIGURE 3.146
Typical filters and their frequency characteristics: LP – low-pass filter,
BP – band-pass filter, HP – high-pass filter.
5
The ideal filter should pass the signals in assumed Exist also filters for example Cauer filter with ripples in the
frequency bandwidth and stop the signals of other stopband.
6
frequencies. The real analogue filter can exhibit ripples Octave it is the ratio of frequency equal to 1:2, while the decade
corresponds with the ratio of frequency 1:10.
 Handbook of Electrical Measurements 91

most commonly used filters: Butterworth, Chebyshev,

0,1 1,0 10 /o Bessel and Cauer.
3dB
Uout/Uin
20dB/dec
-10 1.0

-20 0.1
Cauer
log(Uout/Uin) [dB] (elliptic)
Bessel
0,1 1,0 /o 0.01 Chebyshev

Butterworth f/fo

-30o
-66o/dec 0.1 1.0 10

-60o FIGURE 3.150

The comparison of the frequency amplitude characteristics of the main
filters.
(Uout/Uin)
The Butterworth filter is designed to obtain the
optimal parameters – almost flat characteristic in the
FIGURE 3.148
An example of the amplitude and phase frequency characteristics of
passband and modest slope in the transition band.
the lowpass filter. Therefore it is most commonly used. The Chebyshev
filter has much better steepness of the transfer
In the time domain usually is analyzed answer of the characteristic but at the cost of ripples in the pass band.
filter to stepwise or pulse input signal. The examples of It is possible to design various values of the ripples but
such characteristics are presented in Figure 3.149. as smallest ripples as weaken slope at the transition
band (in the Figure 3.150 is presented the
characteristics for 0.5dB ripples). Exists also inverse
Chebyshev filter with flat passband and ripples in
stopband. The best steepness exhibits the elliptic Cauer
response

step input change

filter but this filter has ripples in both passband and
stopband. The Bessel filter exhibits poor amplitude
characteristic but the best phase characteristic what
time results in the small signal distortion (Figure 3.151).
response

time
response

Bessel
pulse input change Butterworth
0.5 dB Chebyshev
time
FIGURE 3.149
The examples of the time domain characteristics of typical filter.

It has been designed several types of the filters FIGURE 3.151

optimized by taking into account requirements. Figure The time response of the main filters after stepwise change of the
3.150 presents typical amplitude characteristics of the input signal (after Zumbahlen 2008).
92 Basic Electrical Measurements

cascade (more than three) because each of the filters

a) b) significantly attenuates also the useful signal.
C Moreover, the cascade type filter is not a simple sum of
R single filters because each following filters loads the
previous one, thus influencing the frequency
Uout Uin R characteristic (the dependence (3.163) was derived
Uin C Uout
under assumption that the source resistance is
negligibly small and the load resistance is large).
It is possible to obtain better performances by using
LC filters instead of RC one because simple LC filter is
FIGURE 3.152 the second order filter with slope of 40 dB/dec.
The passive RC filters: a) low-pass filter, b) high-pass filter.

Figure 3.152 presents the examples of realization of C

simple passive RC filters. The low-pass RC filter
R2
presented in Figure 3.152a is described by the equation
R1 R
- -
G j  
1
(3.162) C
1  jRC + +

The cut-off frequency of such filter is o = 1/RC.

Because it is the first order filter the attenuation is only
20 dB/dec. We can improve performances of the filter FIGURE 3.154
by connecting several filters in the cascade (Figure The active first order lowpass RC filters.
3.153).
The effect of attenuation of the signal by the passive
RC filters can be decreased by supporting such filters
R1 R2 with the amplifier circuit – as presented in Figure
3.154. But such filters do not ensure sufficient
steepness of the frequency characteristic in the
Uin C1 C2 Uout transition state. It is possible to obtain improvement of
the analogue filter performances by application of the
special active filters presented in Figure 3.155.
L1 L2
a) C1
R1 R2
Uin C1 C2 Uout
+

C2 -
FIGURE 3.153
The passive RC filter of second order and LC filter of fourth order.

For cascade connected RC filters (Figure 3.153) the

b)
frequency characteristic is described in the form R2 C2
R1
G j  
1 -
1  j R1C1  R2C2  R1C2    2 R1C1R2C2 R3
+
(3.163)
C1
By connecting several filters in series we obtain an
increase of the order of the filter and the same the
steepness of the characteristic in the transition band. FIGURE 3.155
But it is not reasonable to connect many filters in the The active second order lowpass RC filters: a) Sallen-Key, b)
multifeedback MFB.
 Handbook of Electrical Measurements 93

For the special case when R1 = R2 = R the

a) R2 capacitances can be calculated from the following
conditions (Jamal 2003)
R1 C1 C
2
1
+
C ; C1  2QC ; C2  C / 2Q (3.165)
2 Rf o
R3 -
R4
R

R5 R
LP
R1 R2 R3
b) R
- - - BP
C1 R3
R1 + + +
HP
R
-
C2 +
C
R4
R2
-
+

FIGURE 3.156
The active second order bandpass RC filters: a) Sallen-Key, b) FIGURE 3.157
multifeedback MFB. The active second order State variable RC filters.

Figure 3.157 presents other commonly used active

The active filters use the operational amplifiers and
RC filter – state variable filter. This filter enables to
the RC elements in the feedback circuit. Figure 3.155a
obtain simultaneous three types of filters: low-pass LP,
presents one of the most commonly used filter – Sallen-
high-pass HP and band-pass filters BP. The
Key filter (called also filters with voltage controlled
modification of this filter known as biquadratic filter is
source - VCS). Figure 3.,.155b present other active
used for design of Cauer and inverse Chebyshev filter.
filter - MFB filters (multi- feedback filters). Note that
The design of the analogue filters requires quite
the low-pass filters and high-pass filters are obtained
complicated calculations (Huelsman 1993, Thede 2004,
simply by swapping around the R and C elements. The
Schaumann 2001, Winder 2002, Volkenburg 1995).
band-pass filters are obtained by the combination of
Fortunately, there are available various design
low-pass and high-pass filters – Figure 3.156.
procedures enabling fast design of the filter with
For the Sallen-Key lowpass filter the transfer
desired performances. For example, the Matlab
characteristic can be described by the following s-
software offers various tools for the design of the filters
operator function:
(Lutovac 2000). Also, in the LabVIEW platform there
1 are ready to use design procedures of the filters.
Usually, the filter is described using s-operators. A
R1 R2 C1C2 o2
H( s )   simple RC filter (equation 3.162) can be described in
1 1  the form:
s2  s s 2  o  o2
R1C2 R1 R2 C1C2 Q
1
(3.164) H s  (3.165)
1  sRC
where
For design of filters commonly it is used normalized
1 R1C2 form with assumption that R = 1 , C = 1 F, o = 1
o  ; Q rad/s and the dependence (3.165) can be rewritten as:
R1 R2 C1C2 R2C1
1
H( s )  (3.166)
1 s
94 Basic Electrical Measurements

Similarly the fourth order filter is described by the Table 3.11 presents the Butterworth denominator
equation: polynomials.
TABLE 3.11
1
H s  (3.167) Butterworth filter denominator polynomials.
a  bs  cs 2  ds 3  s 4 Order Polynomials
1 s 1
The general transfer equation of the filter has the
form: 2 s 2  2s  1
3 s 3  2s 2  2s  1
a s m  a s m 1  ...  a1 s  ao
H  s   m n m 1 n 1 (3.168) 4 s 4  2.61s 3  3.414s 2  2.61s  1
bn s  bn 1 s  ...b1 s  bo
The active filters presented in Figure 3.156
In analysis of the properties of the filter usually are represent the second order filters. The simplest way to
determined roots of the denominator (poles) and obtain higher order of the filter is to connect the filters
numerator (zeros). In the s-plane (Figure 3.158) all in a cascade form – for example, to obtain the fourth
poles should be in the left side to guarantee the order filter usually there are connected two filters of the
stability. second order, and to obtain the fifth order filter
additional first order filter is connected. Table 3.12
b presents the denominator polynomials derived under
the assumption that we dispose to connect the first and
Butterworth second order filters.
1 dB Chebyshev TABLE 3.12
a Butterworth filter denominator polynomials for the filter composed
from the first and second order components.
-1 1
Order Polynomials
1 s 1
2 1  1.414s  s 2
3 1  s  1  s  s 2 
FIGURE 3.158
The poles of the 4-th order Butterworth and 1 dB Chebyshev filters.
4
1  1.848s  s 1  0.765s  s 
2 2

All most popular filters are described as the

5
1  s  1  1.618s  s 1  0.618s  s 
2 2

polynomials representing the denominator in equation

(3.168). Usually the poles of all filters are completed as Let us design the low-pass Butterworth filter of the
tables in various publications. Knowing poles it is fourth order with the cut-off frequency 1000 Hz
possible to calculate coefficients of polynomial. For (Maxim 2003). The characteristic of the Sallen-Key
example the poles of Butterworth filter of 4th order are filter can be described using the following equation:
(Figure 3.158):
G1G2
0.9239  j0.3827 G( s)  (3.171)
a  jb   (3.169) G1G2  G1  G2 C2 s  C1C2 s 2
0.3827  j0.9239
where resistances R are presented as conductances G
The polynomial of denominator can be calculated (for simplifying further calculations). With normalized
as [Maxim 2003] resistors 1 the equation (3.171) can be rewritten in a
form:
s 2  2as  a2  b2 (3.170)
1
Combining equation (3.169) and (3.170) we obtain G( s)  (3.172)
the polynomial: 1  2C2 s  C1C2 s 2

s 2
 1.848s  1 1  0.765s  1 According to the polynomial (s2+1.8478s+1)
2
(s +0.7654s+1) and the equation (3.172) in the first
 Handbook of Electrical Measurements 95

filter C1C2 = 1 and 2C2 = 1.8478 thus the capacitances Q C u1  u 2 

in the first filter are C2 = 0.9239 F and C1 = 1.08 F. I  (3.174)
T T
Similarly in the second filter C1C2 = 1 and 2C2 =
0.7654 thus the capacitances in the first filter are C2 = The circuit with a switched capacitor is equivalent to
0.3827 F and C1 = 2.61 F. the circuit with the resistor R
If we assume that the resistors are 1 k (multiplied
by factor 1000) the capacitances should be divided by u u T 1
1000. To obtain the cut-off frequency as 1kHz (instead R 1 2   (3.175)
I C fC
of 1 rad/s) we should divide the capacitances by
21000. As result we obtain the circuit of the filter Thus it is possible to design the filters with the R
presented in Figure 3.159. element substituted by the switched capacitor. The
important advantage of this solution is the possibility of
171 nF 415 nF the tuning of filter by the change of the frequency of
switching. Figure 3.161 presents the design of the
1 k 1 k 1 k 1 k
+ +
switched capacitor filter.

Uin - - Uout R C C
147 nF 60.9 nF C
- -
+ +
Uin Uin
Uout Uout
FIGURE 3.159
The 4th-order Butterworth low-pass filter with the cut-off frequency
1 kHz - after (Maxim 2003).

In the integrated circuit technology it is rather FIGURE 3.161

difficult to obtain precise values of the RC elements. The RC filter and equivalent filter with switched capacitor.
Therefore, instead of resistors there are developed filter
circuits with switched capacitors because in the Figure 3.162 presents the state variable universal
integrated circuit technology it is relatively easy to filter circuit designed for the programmable active
design the switch element. filters with the switched capacitors. This filter can
operate as the low-pass filter LP, high-pass filter HP,
band-pass filter BP and additionally band-reject filter
a C a I R
BR. The cut-off frequency can be changed of the
switching frequency of capacitor C1 simulating resistor:

u1 u2 u1 u2 1 1
b b o   (3.176)
RF C fC1C

FIGURE 3.160
The switched capacitor as the simulator of the resistance.
- - -
Consider the circuit presented in Figure 3.160. If -
+ +
+
the capacitor is switched with the period T (first time to +
the input u1 the second time to the output u2) the charge
transfer is

Q  Cu1  u2  (3.173) HP f BP f LP

and the equivalent current is

FIGURE 3.162
The state variable filter with the possibility of application of the
frequency tuning.
96 Basic Electrical Measurements

- application of the phase sensitive detectors – in lock-

R2 in amplifiers to rejection of the noises;
R4
- elimination of the temperature zero drift by
R1 application of the auto-zero function;
- correct grounding and shielding of various parts of the
-
+   circuit, for example application of the Wagner earth in
the AC bridge circuits.
Modern signal recovering techniques enable us to
eliminate the noises and interferences significantly
larger than the measured signal. But we can look at the
R3
problem of noises and interferences from other point of
HP BP clock LP view. By using incorrect connections of the signal or
by applying the incorrect grounding we can deteriorate
the quality of the measuring signal.
FIGURE 3.163
The functional circuit of the high performance dual switched capacitor One of the most difficult to elimination interference
filter model LMF100 of National Semiconductor. is the signal deteriorated by the zero drifts. The zero
drift can be caused by several sources: the connections
Figure 3.163 presents the integrated filter circuit of two metals (thermoelectric voltage), instable
developed by the National Semiconductor - model contacts, vibrations of various parts of the circuit, bad
LMF100. This circuit utilizes the operation principle of quality insulation, piezoelectric effects, electrochemical
the state variable filter presented in Figure 3.162. The effects, etc. The basis of the zero drift is very often
integrated circuit of the filter with switched capacitors technological (quality of contacts, uniformity of
enable to design the second order filter (high-pass, low- materials) and it can be amplified by the temperature
pass or band-pass). The external resistors can be used differences of various parts of the circuit. For that
to the setting of the band-pass gain or low-pass gain. reason it is important to exclude the formation of the
The center frequency is tuned by the external clock temperature differences (for example non-uniform
frequency. heating of various parts). Sometimes, conversion of the
DC into the AC signals can be helpful (carrier
3.8 The noises and interferences amplifiers, chopper amplifiers, etc.).
The measured signals are usually accompanied by One of the basic sources of the noises is the resistor
some noises and interferences, sometimes of the level R. The thermal Johnson noise UnT is generated due to
comparable to the level of the measured signal. As the the chaotic thermal movement of the charges. This
noises and interferences we assume all signals other noise can be described by the Nyquist relation
than the measured signal – the noises are the stochastic
signals with indefinite frequency and magnitude (the U nT  4kTRf (3.177)
white noise is the signal with theoretically all frequency
components). The interferences are the signals coming where k is the Boltzman constant (k = 1.38 10-23 Ws/K),
from various external sources; very often these signals T is the temperature, and f is the frequency
are of the main frequency 50 Hz (or 60 Hz) and bandwidth.
harmonics of this frequency. The thermal noises can be reduced by the decrease of
The typical interference signals are generated by the the resistances, by the limitation of the bandwidth and
electric power lines, electrical machines, lighting of course by the control of the temperature.
equipment, commutating devices, radio communication Another important source of noises is the
transmitters, atmospheric discharges or cosmic noises. semiconductor junction. When the current is on the
The sources of noises are usually internal – all resistors potential barrier a number of charges q are randomly
and semiconductors intrinsically generate noises. crossing this barrier and these random current
In the previous chapters various methods of rejection fluctuations are the source of the shot noises Iss
of noises or interferences have been described: described by the Schottky relation
- application of the differential input of amplifiers
(common mode signals rejection) or application of
I ss  2qIf (3.178)
differential sensors (for example to eliminate the
external temperature influence);
- application of the galvanic separations of the circuits where q is the electric charge (q = 1.6 10-19 C).
(isolation amplifiers);
 Handbook of Electrical Measurements 97

In the low frequency range there are 1/f type noises Figure 3.164 presents the example 0f the noise
(sometimes called low frequency noise, flicker noise or characteristics of Hall sensor – spectral density and low
excess noise). There are many sources of these noises, frequency signal. When there are many sources of
most of them unknown. Such noises are inversely noises necessary is to perform noise budget calculation,
proportional to the frequency, and the power spectral as it is presented for amplifier in Section 3.6.4 (Eq.
density S(f) of this noise is: 3.128). As noises are characterized by V / Hz finally
it is necessary to calculate Unrms as:
E 2f
S( f )  (3.179)
U nrms  U n   f
f (3.182)

where Ef is the rms voltage of noises and  coefficient The low frequency 1/f noises are usually
characterized by Unrms determined for certain frequency
is  = 0.8 – 1.3.
bandwidth, for example 0.1 – 10 Hz.
Because the noise level depends on the frequency
these noises are not the white noises (noises containing
every frequency of signals) and are called pink noise – 1 1 C12
2
random noise having the same amount of energy in C12 C2 U2
each octave.
Due to the random character of the noises usually R2
R2 U1
they are described not by the voltage level Un but by C1 C2
C1
the spectral density S(f) (see Figure 2.16):
2
2
U n2  U n 
S( f )   (3.180) U1
f  f 

Rs C12
The U n / f value is called the spectral density
of noises and it means the rms value of the voltage
signal in relation to the square root from the frequency Rw
U1 Us
bandwidth. Often the noises are characterized by the
SNR factor – signal to noise ratio:

Un
SNR  20 log (3.181)
U FIGURE 3.165
The interferences connected to the amplifier by the capacitive
coupling: the access to the adjacent wire and access to the amplifier
10-14 2
noise [V Hz] input.

The external interferences can penetrate the

measuring circuit by the capacitive, inductive or
10-16 conductive coupling. Figure 3.165 presents two
examples of the interferences connected to the
f [Hz]
1
measuring circuit by the capacitive coupling. The wire
10 100
2 infiltrates the signal from the wire 1 by the inter-wire
Vout [mV] capacitance C12. Or the amplifier connected to the
30 measured signal Us is additionally connected to the
signal U1 by the coupling capacitance C12.
For the large frequency is created the capacitance
0 voltage divider and the interference signal in the wire 2
is:
6 time [hours]
C12
FIGURE 3.164 U2  U1 (3.183)
An example of the noises in Hall sensor (From Schott et al. 1997). C2  C12
98 Basic Electrical Measurements

For lower frequencies the interference signal depends Figure 3.168 presents the inductive coupling when
on the frequency f and the resistance R2 (Pallas-Areny the current in adjacent wires can induce an additional
1999) interference voltage:

U 2  jR2C12U1 (3.184) U L  MI1 (3.185)

A typical example of capacitance coupling is the A typical example of inductive coupling is the
penetration of the interferences by the inter-turns penetration of the measuring circuit by the voltages
capacitance of the separation transformer. The induced by the external sources of electromagnetic
researcher performing experiments would expect that fields, from radio transmitters or mobile telephony
the connection of the galvanic separation (by use of the communication. Therefore, such interferences are often
separation transformer) makes the circuit free from the called as RFI – radio frequency interferences or
50 Hz interferences. To effectively obtain such electromagnetic interferences IMI.
rejection of interferences it is necessary to introduce Problem of generation of interferences by
between turns the copper grounded electrostatic shield electromagnetic fields increases due to increase of
(this way we eliminate the capacitive coupling of electromagnetic pollution. It is main subject of interest
signals). electromagnetic compatibility EMC. Generally EMC
looks at the interferences from both point of view:
- how to design the electrical devices to reduction the
emitted electromagnetic field,
C = 5 pF - how to design the electric devices to obtain
immunity to EMI.
Fortunately recently are in power many international
Rw standards strictly prohibited emission of introduced to
the market electric devices.
U1 = 220V
Rs R1

FIGURE 3.166 Rw
The interferences connected to the amplifier by the capacitive Us
coupling and human body (after van Putten 1996). R2
Other example is presented in Figure 3.166 (taken
from the book of van Putten – van Putten, 1996). It
illustrates what happens when we touch the input of
Rg Ug
oscilloscope with typical impute impedance > 1 M.
Assuming impedance of the human body as about 1
M and the coupling typical capacity as 5 pF we
FIGURE 3.168
obtain that in the input of oscilloscope appears voltage The interferences connected to the amplifier by the conductive
signal as high as about 4 V. coupling.

I1 A typical example of the conductive coupling of

interferences is voltage Ug (Figure 3.168) appeared in
the case of connection of two various grounding points.
M The inter-ground difference of potentials can be
introduced as the voltage drop Ui on the resistor R2:
Rs Rw
U1 R2
Us Ui  U g (3.186)
R2  Rg

There are several universal methods and tools of

rejection or limitation of the interference. The
FIGURE 3.167
The interferences connected to the amplifier by an inductive coupling.
capacitive coupling can be reduced by the application
of the electrostatic shield. A conducting plate or foil
 Handbook of Electrical Measurements 99

grounded at one point can be used as the electrostatic

shield (Figure 3.169). After introduction of this shield Shielding
most of the currents coming from the interference effect
source U1 are shorted to the ground by the shield and
aluminum
do not penetrate the measuring circuit (Figure 3.169b). 100

10
a) b) NiFe
f [Hz]
1
0.1 1 10 100

U1 U2 U1 U2 FIGURE 3.171
Comparison of effectiveness of shielding of ferromagnetic and non-
magnetic materials versus frequency (Yamazaki 2006).

The effectiveness of shielding is described by the

shielding factor S as the relationship between external
He and internal Hi magnetic field

FIGURE 3.169 He
The reduction of the capacitive coupling by the electrostatic shield. S (3.187)
Hi
Figure 3.170 presents the method of reduction of
magnetic coupling by the application of a magnetic For simple shapes of the shield (sphere, cylinder or
shield. The magnetic shield is prepared from high- cube) there were calculated approximate formulas
permeability magnetic material. The lines of magnetic enabling determination of the shielding factor for DC
field are closed in the shield and do not penetrate the magnetic field. For a long cylinder with diameter D the
area of the measuring circuit. shielding factor is:

t
S  1 (3.188)
D
The most important factor is the relative permeability
µ of the material. That is why for shields are usually
used materials of high permeability as NiFe (mumetal)
device
with relative permeability of about 500 000 or
amorphous materials reaching permeability as high as
800 000. The second factor influencing the shielding
effect is the wall thickness. However it was proved that
instead of increasing of the wall thickness more
effective is to use multiple shells. For example a
cylinder consisting of two layers with diameter D1 and
FIGURE 3.170 D2 shielding factors S1 and S2 the resultant shielding
The reduction of the magnetic coupling by the magnetic shield. factor is
To obtain effective shielding the material of the   D 2 
shield should be properly chosen – for low magnetic S  1  S1  S2  S1 S2 1   2   (3.189)
fields it is necessary to use different material than for   D1  
high magnetic field since the magnetic permeability
strongly depends on the level of magnetic field. Also A mumetal alloy commonly used as the shield
for DC magnetic field different magnetic materials material is very sensitive to stress and mechanical
should be used than for the AC magnetic fields (Figure shocks, so it should be carefully annealed. Both
3.171). Therefore, sometimes the magnetic shield is mumetal and amorphous materials are rather expensive
composed of several shields prepared from various and in many cases the shield made from much cheaper
materials. SiFe electrical steel can be an alternative [Okazaki et al
100 Basic Electrical Measurements

2009]. At higher frequencies it can also be effective to Many of the interferences penetrate the measuring
use a shield made from aluminum in which eddy- circuits through the connecting cables. Generally, the
currents aid the shielding. Also for high frequencies measuring signals should be connected using shielded
(for example in microwave range) the Faraday cage (a wires. As it is demonstrated in Figure 3.172 the simple
mesh prepared from conducting material) is commonly twisted wire in certain cases can be more effective than
used as a shield. Ideally the shield should be designed coaxial cable. In the twisted pair the interferences are
in such a way as to make possible its periodic reduced because the voltages e’ and e” induced in
demagnetization, because magnetized shield exhibit adjacent wires compensate each other and potential
lower performance. remaining of induced voltages e exhibits opposite
We can destroy the signal by not correct connection direction in the neighbouring loops of the twisted
to the measuring instrument. Figure 3.172 presents the wires.
results of experimental investigation of the various For transmission of the digital data more and more
connecting cables presented in well-known book of Ott important are fiber-optic cables. The optical system is
(Ott 1988). The connecting cables were influenced by immune to the electromagnetic interferences. The
artificially generated external 50 kHz magnetic field. transmission of the data is extremely fast, theoretically
At the end of the cable the attenuation of parasitic 50 the speed of frequency THz is possible and in practice
kHz signal was measured. the transmission of hundreds MHz is achieved.

0,3 A attenuation fiber optic wire

50 kHz of
interference LED PD
OC OC
100 100M 0 dB
cladding
13 dB

55 dB
FIGURE 3.174
The fiber-optic cable as the communication wire for transmission of
70 dB the measuring data.

The principle of data transmission using the fiber-

optic connection is presented in Figure 3.174. The
FIGURE 3.172
electrical signal is converted to the light by means of
The attenuation of the magnetic interference signal depending on the
method of the connection of the cables (Ott 1976). the LED or laser diode, then the optical system OC –
optical coupler is formed. Typical length of the light
The simple single wire in the coaxial shield wave is about 1300 nm. The optical signal should be
grounded at the end practically did not attenuate the converted again to the electrical form by means of a
interference. Similarly the twisted wire grounded at photo-detector device PD, before it is fed to the
both ends exhibited poor attenuation of interferences. receiver. The effect of internal reflection of the light
The same twisted pair cable but grounded at the (the multimode transmission) is utilized for
receiver end attenuated the interferences much better. transmission of the signal through the fiber-optic wire.
Satisfying attenuation of interferences exhibited double The core of the fiber-optic cable is surrounded by a
wire cable with a coaxial shield grounded at the special glass material called cladding. This part of
receiver end. fiber-optic cable exhibits the reflection coefficient
ensuring that the light does not leave the wire.
In Internet communication commonly are used
optical cables known as 100BASE-FX with speed up to
e' 100 Mb/s and distance to 2 km, and 1000BASE-SX
U1 e with speed up to 1 Gb/s and distance 200 – 500 m.
e"
There is no one simple receipt for the grounding,
shielding and cabling. Practically, it is always
recommended to perform a series of experiments to
FIGURE 3.173 find the best solution – sometimes a simple and cheap
The twisted pair of wires and the principle of reduction of
interferences in such connection.
twisted pair of wires can exhibit better performances
than expensive shielded wire.
Handbook of Electrical Measurements 101

Generally we should avoid the connection of the ) impedance of leakage Zsg. In the input of amplifier
signal source and the receiver in such a way that the the interference signal is:
shield of the cable is a current-carrying part of the
circuit (Figure 3.175) (Do not allow shield current to R2 R
Uz Ug Ug 2 (3.190)
exist and do not allow the shield to be at a voltage with Rg  R2  Rsg Rsg
respect to the reference potential) (Rich 2005).
The interference signal is significantly attenuated
because the impedance Zsg is large. If it is not possible
to separate the signal source from the ground we can
use the amplifier with floating point. Modern digital
Rs Is Rw voltmeters generally have the floating ground terminal
– practically the ground terminal is not introduced at
the front panel of the instrument.
Us

Gs Gr shield
guard
Rs R1
HI

Rs Is Rw
Us Z1 Rw Uout
R2
LO
Us

Gs Gr Is Z2
G

FIGURE 3.175 Ug Z3
Not recommendable connection of the signal source and the receiver. Rg

Similarly, it is not recommended to connect the FIGURE 3.177

signal source to the amplifier of different grounding The connection of the grounded signal source to the amplifier with
potentials (Figure 3.175). In such case this potential additional shield “guard”.
difference Ug implies that in the R2 resistance exists the
equalizing current Is. This current generates the In modern amplifiers or voltmeters a special kind of
conductive interference as the voltage Uz at the shielding is used – a double shield. The first shield
terminals of the amplifier. “ground shield” is connected to the ground, but the
second shield “guard shield” is with floating ground –
Rs R1 it is connected to the ground and to the “minus”
+
terminal by large leakage impedances Z2 and Z3 (Figure
3.177). The equalizing current Is is now in a closed
Us Rw
R2 - loop outside the input signal connection.
Usually in the measuring instruments the grounds of
the digital part, supply voltage and analogue part are
Zsg Is separated. The first one (digital part and supply
voltage) is called power GND (power ground) while
Rg
the second is called LL GND (low level ground). The
Ug
input terminals are usually with floating ground and are
called HI (high) and LO (low). In high accuracy
FIGURE 3.176 instruments the terminal GUARD is additionally
The connection of the signal to the amplifier with floating ground of introduced.
the source.
When small resistance Rx is connected to the
Figure 3.176 presents the connection of the signal measuring device as the rule the four-wire connection
source with isolated ground (floating ground). The is used – with separate current and voltage terminals
source is connected to the ground by large (108 – 1010 (Figure 3.178). The current terminals are usually larger
and more massive and the connections wires Rp2 are
outside the measuring circuit. Because the voltage drop
102 Basic Electrical Measurements

is practically measured without current (large input U out R2 1

resistance of the amplifier) the resistances of  (3.191)
Ug R2  Rg L2
connection Rp1 can be neglected. In digital multimeters 1
the voltage terminals are indicated as “Sense HI” and Rg  R2
“Sense LO” while the current terminals are indicated as
“Input HI” and “Input LO”. The interference is attenuated and additionally a low-
pass filter L2/(Rg+R2) is formed. For higher frequencies
Rp2 (above 1 MHz) instead of toroid yoke sometimes the
A ferrite ring (ferrite bead) thread on the cable is used.
INPUT HI
More difficult to attenuate are the low frequency
SENSE HI interferences (industrial frequency interferences).
Rp1 Usually, various kinds of filters are used in order to
Rx V suppress interferences. Practically all devices of power
Rp1 electronics are equipped with special filters to fulfill
SENSE LO requirements of EMI standards – known as EMI filters.
An example of EMI filter is presented in Figure 3.180.
INPUT LO
Rp2
LINE
FIGURE 3.178
The connection of the measured low resistance to the four-terminal
multimeter. N
GND
Figure 3.179 presents the application of the toroid
choke for attenuation of the interferences. This method
is used in computer connections and it is realized in FIGURE 3.180
such a way, that the signal cable is wound on the An example of EMI filter for rejection of high frequency
interferences.
toroidal ferrite yoke.
It can be important to match the noise to the
amplifier. The noises of the amplifier are represented
by the voltage en and current in noises. The noise factor
Rs Rw F is the ratio of the amplifier noises to the thermal
noises
Us
e2  i 2 R 2 e2  i 2 R 2
F n n s  n n s (3.192)
eT2 4kTRs
L1 R1
The optimal value of Rs which makes F minimal can
M Rw Uout be calculated as
Rs
F
Us L2 R2 0 (3.193)
Rs

Ug Rg From (3.192) and (3.193) the optimal source

resistance is
FIGURE 3.179
en
A toroidal longitudinal choke used for attenuation of the interferences
(Northrtop 1997).
Rsopt  (3.194)
in
The interferences are represented by the source Ug.
The ratio of the signals to the interference is Usually the Rsopt is relatively large, even up to
several M, while the typical voltage sources exhibit
rather small resistance. Therefore, matching the
Handbook of Electrical Measurements 103

resistance by using a transformer can result in Hagel R., Zakrzewski J., 1984, Dynamic measurements (in
reduction of the noises. Polish), WNT
Hamilton C.A., Burroughs J., Benz S.P., 1997, Josephson
voltage standard – a review, IEEE Trans. Appl.
Rs 1:n Superconductivity, 7, 3756-3761
+ Huelsman L.P. 1993 Active and Passive Filter Design,
McGraw-Hill
n2Rs Jamal R., Steer R. 2003 Filters, Chapter 22 in Electrical
Us Ri,Re Measurements, Signal Processing and Displays, CRC
Press
Jeckelmann B., Jeanneret B., 2001, The quantum Hall effect
- as an electrical resistance standard, Rep. Prog. Phys., 64,
1603-1655
Jung W.G. 2004 Op Amp Application handbook, Newnes
FIGURE 3.181
Kester W., 2005, The data conversion handbook, Newnes
The noise matching by transformer coupling. King G., Fukushima T., 2004, RTD interfacing and
linearization using an ADuC8xx microconverter, Analog
If the ratio of turns of the transformer is n the Devices AN-709
primary resistance Rs is reflected at the secondary Kitchin C., Counts L., 1986, RMS to DC conversion
winding as: application guide, Analog Devices
Kitchin C., Counts J., 2006, A designers guide to
instrumentatiuon amplifiers, Analog Devices
Rs  n 2 Rs (3.195) Kohlmann J., Behr R., Funck T., 2003, Josephson voltage
standards, Meas. Sc. Technol., 14, 1216-1228
Thus the optimal transformer turns ratio is: Levinsen M.T., Chiao R.Y., Feldman M.J., Tucker B.A.,
1977, An inverse AC Josephson effect voltage standard,
Rsopt Appl. Phys., Lett., 31, 776
1 en
nopt   (3.196) Lutovac M.D., Tosic D.V., Evan B.L. 2000 Filter Design for
Rs Rs in Signal Processing using MATLAB and Mathematica,
Prentice Hall
The transformer inserted between the source and Malik R., 2010, Thermocouple linearization when using the
AD8497, AN-1087 Application Note of Analog Devices
the amplifier matches the resistances, but also plays the
Maloberti F., 2007, Data converters, Springer
role of isolation device (and sometimes as the Manabendra Bhuyan, 2011, Intelligent instrumentation, CTC
symmetrization device – converting the single-ended Press
source to the differential input). Mancini R., 2002, OpAmps for Everyne, Texas Iinst. Appl.
Note
Maxim 2003, Analog Filter Design Demystified, Maxim
References (Dallas Semiconductors) Technical Note No. 1795
Moghimi R., 2011, To chop or auto-zero: that is the question,
AD8551 – The data sheet of the AN 8551 amplifier of Analog
Analog Devices Technical Note, MS-2062
Devices
MT-066, 2009, In-Amp bridge circuit error budget analysis,
Bauch A., 2003, Caesium atomic clocks: functions,
Analog Devices Tutorial
performances and applications, Meas. Sc. Technol., 14,
Nash E., Greichen J., 1999, Revolutionary RF IC performs
1159-1173
rms to DC conversion, Microwaves & RF, 38, 140-149
Bez S.P., Hamilton C.A., 2004, Application of the Josephson
NIST 1458 2003 NIST Measurement Service for DC Standard
effect to voltage metrology, Proc IEEE, 92, 1617-1629
Resistors, NIST Technical Note 1458
Chevtchenko O.A., et al, 2005, Realization of a quantum
Northrop R.B. 1997 Introduction to Instrumentation and
standard for AC voltage: overview of a European
Measurements, CRC Press
Research Project, IEEE Trans. Instr. Meas., 54, No.2,
Novoselov K.S. et al, 2007, Room temperature quantum Hall
628-631
effect in graphene, Science, 315, No. 5817, 1739
Coughlin R.F., Driscoll F.F. 2000 Operational Amplifiers and
Nuce D.S., 2004, Linear position sensors, Wiley Interscience
Linear Integrated Circuits, Prentice Hall
Okazakio Y., Yanase S., Nakamura Y., Maeno R., 2009,
Duff M., Towey J., 2010, Two ways to measure temperature
Magnetic shielding by grain-oriented electrical steel sheet
using thermocouples feature simplicity, accuracy and
under alternating fields up to 100 kHz, Przegl. Elektr., 85,
flexibility, Analog Dialogue, 44-10, 1-6
No.1, 55-59
Gilber B., 1975, Translinear circuits: a proposed
Ott H.W. 1988 Noise Reduction in Electronic System, John
classification, Electron. Lett., 11, 14-16
Wiley & Sons
Fraden J., 2003, Handbook of modern sensors, Springer
Padmanabhan T.R., 2000, Industrial instrumentation, Springer
Franco S. 2001 Design with Operational Amplifiers and
Analog Integrated Circuits, McGraw-Hill
104 Basic Electrical Measurements

Pallas-Areny, Webster, 1999, Analog signals processing, John

Wiley & Sons
Pallas Areny R., Webster J.G, 2001, Sensors and signal
conditioning, John Wiley & Sons
Park P.G., Kim Y.G., Shifrin V.Y., 2005, Maintenance of
magnetic flux density standards on the basis of proton
gyromagnetic ratio at KRISS, IEEE Trans. Instr. Meas.,
54, No.2, 734-737
Rich A. 2005 Shielding and Guarding, Analog Devices
Application Note AN-347
Schaumann R. 2001 Design of Analog Filters, Oxford
University Press
Scholes R., 1970, Application of operational amplifiers to
magnetic measurements, IEEE Trans. Magn., 6, 289-291
Schott C., Blanchard H., Popovic R., Racz R., Hrejsa J., 1997,
High accuracy analog Hall probe, IEEE Trans. Instr.
Meas., 46, 613-616
Scofield J.H. 1994 A frequency domain description of a lock-
in amplifier, American Journal of Physics, 62, pp. 122-
133
Seebeck T., 1822, Magnetische polarization der Metalle und
Erze durch Temperature Differenz, Abh. Der Preusischen
Academic der Wissenschaft
Stanley W.D. 2001 Operational Amplifiers and Linear
Integrated Circuits, Prentice Hall
Stata R., 1967, Operational integrators, Analog Dialogue, 1,
11-16
Thede L. 2004 Practical Analog and Digital Filter Design,
Artech House
Van Putten A.F.P., 1996, Electronic measurement systems,
IOP Publishing
Volkenburg M.E. 1995 Analog Filter Design, Oxford
University Press
Wassenar R., Seevinck E., van Leeuwen M.G., Speelman
C.J., Holle E., 1988, New technologies for high frequency
rms to DC conversion based on a multifunctional V to I
converter, IEEE J. Solid State Circuits, 23, No.3, 802-814
Weyand K., 2001, Magnetic field standards – trace to the
maintained units, Int. J. Apl. Electromagnetism and
Mechanics, 13, 195-202
Williams J., 1988, Thermocouple measurement, AN28
Application Note of Linear Technology
Winder S. 2002 Analog and Digital Filter Design, Newnes
Witt T.J., 1998, Electrical resistance standards and the
quantum Hall effect, Rev. Sc. Instr., 69, No.8, 2823-2843
Wong V., 2011, Lowest noise zero drift amplifier has 5.6
nV/rtHz voltage noise density, Analog Device
Application Note AN-1114
Yamazaki K., Muramatsu K., Hirayama M., Haga A., Torita
F., 2006, Optimal structure of magnetic and conductive
layers of a magnetically shielded room, IEEE Trans.
Magn., 42, 3524-3526
Zhang Y., Tan Y.W., Stormer H.L., 2005, Experimental
observation of the quantum Hall effect and Berry’s phase
in grapheme, Nature, 438, 201-204
Zumbahlen H. (Ed), 2008, Linear circuit design handbook,
Elsevier