Chapter 4.

Reactive sensors
Capacitive sensors
This sensors are widely used because of their constructive simplicity, small volume and
integration, low cost and robustness, moreover they allow some non-contact measurement.
Besides these many advantages they show some major disadvantages, that are mainly related to
the large sensitivity to electrical noise.

Capacitive sensing can in principle sense any variation of in the electric displacement field
remember that the definition of capacitance relies on the following differential equation:

∇ ∙ 𝐃 = ∇ ∙ ϵ𝐄 = ρ

or in the integral form, being A a close surface, n the versor normal to the dA, and V the volume
enclosed by A:

∫ ϵ𝐄 ∙ 𝐧 dA = ∫ ρ dv
A v

∫ ϵ∇V ∙ 𝐧 dA = ∫ ρ dv = Q
A v

x1
V(x1 ) − V(x2 ) = ∫ 𝐄 ∙ 𝐝𝐥
x2

Being the capacitance C defined for a structure consisting of two conductive element separated
by a perfect insulating media as the coefficient relating the voltage difference applied to the two
conductors and the charge accumulated at the dielectric/conductor interfaces, as follows:

CΔV = Q

It is clear that C can be expressed in analytical form only for simple geometries. Otherwise it can
be only be evaluated by numerical methods or experimentally.

In this section the main applications and problems related to capacitive sensors are discussed,
with reference to the simplest possible geometries, i.e. a parallel infinite plate capacitor or a
cylindric capacitor. To understand many of the problems discussed in the following it has to be
taken into account that typical values for capacitive sensors are in the order of tenths of pF. In
fact being 0 = 8.85 pF/m, for a planar plate capacitance with a surface area of 1 cm2 and a
distance between plates of 1 mm, we get C=0.85 pF.
Displacement sensors.
The configurations below show the possible displacement sensing, the first two are based on a
moving plate, obviously in this case this plate can be replaced by a conductive target (usually
grounded) so that the actual sensor consists of a single plate (probe).

dx C1 C2

d
 d
A=wl (a) (b)
wx A=wl (c)
𝜖𝐴 𝜖𝐴 𝐶0 𝜖𝐴 𝜖𝑤(1 − 𝑥)𝑙
𝐶0 = 𝐶= = 𝐶0 = 𝐶1 = ;
𝑑 𝑑(1 − 𝑥) (1 − 𝑥) 𝑑 𝑑
𝜖𝑤(1 − 𝑥)𝑙 𝜖0 𝑤𝑙𝑥
𝐶= = 𝐶0 (1 − 𝑥) 𝐶2 = ;
𝑑 𝑑
𝐶1 𝐶2
𝐶=
𝐶1 + 𝐶2

figure 4.1

For type (a) measurement it is found that the ratio of the measurement range (dmax) with the spot
diameter (2√𝐴/𝜋) is 1:8 with an associated resolution that can be as low as 10-6 of the full
range.

Differential structure can be easily obtained as shown below:

d 𝐶0 𝐶0
𝐶1 = ; 𝐶2 =
dx
(1 − 𝑥) (1 + 𝑥)
d

A=wl

figure 4.2

Obviously these sensors allow noncontact measurements, are quite insensitive to target surface
roughness, composition or microstructure if conductive.

They are instead sensitive to all phenomena that modifies the electrical field in between the
plates, i.e. moisture (that heavily affects the dielectric constant of the dielectric gas), dust and or
deposits of the surface (if they correspond to the creation of a layer with different dielectric
constant), pressure (because the dielectric constant of gases depends on pressure up 50% changes
with 200 Bar changes).
Proximity sensors
Proximity sensors belong to a special class of displacement sensor, because they are based on
the same measurement principle as the one shown in figure 4.1. a), but they are not intended to
measure the distance, d, but only to detect the presence of a target.

They can be used to sense also non-conductive targets, in fact if the target consists of a material
with a different dielectric constant with respect to air, then its proximity perturbs the electric
field causing a variation of the sensed capacitance. Applications of proximity capacitive sensors
are touch screen matrix and anti-intrusion sensors.

figure 4.3

Pressure/acceleration sensors
A capacitive sensing can be coupled to primary sensors for pressure based on deformable elastic
membranes or thin plates to readout the average displacement due to deformation.
3
(1 − 𝜈 2 )
Deformable circular thin z 𝐷𝑧 (𝑟) = 16 (𝑅2 − 𝑟 2 )2 Δ𝑃
plate with thickness t 𝐸𝑡 3
P1 1 𝑅 (1 − 𝜈 2 ) 𝑅4 Δ𝑃
R 𝐷𝑧−𝑚𝑒𝑎𝑛 = ∫ 𝐷𝑧 (𝑟)𝑑𝑟 =
𝑅 0 10𝐸 𝑡 3
DZ ( r) 2
r 𝜋𝑅 𝐷𝑧−𝑚𝑒𝑎𝑛
d 𝐶=𝜖 ~𝐶0 (1 + )
P1 𝐷𝑧−𝑚𝑒𝑎𝑛 𝑑
𝑑(1 − )
𝑑
Rigid plate Δ𝐶 𝐷𝑧−𝑚𝑒𝑎𝑛 (1 − 𝜈 2 ) 𝑅4 Δ𝑃
= =
𝐶0 𝑑 10𝐸 𝑡 3 𝑑
figure 4.4

Capacitive pressure sensing is at the basis of capacitive microphones.

Capacitive sensing can be coupled also to other mechanical primary sensors, such as
accelerometers, based on deformable elastic structure since it senses the average displacement
due to deformation.

Figure 4.5

Level sensors
Non conductive liquids (e.g. oil)

In this case a bare rod (radius r) is immersed in the container, assuming for this latter a cylinder
shape (radius R), we have.

r 𝜋𝜖𝑙 𝑙(1 − 𝑥) 𝜋𝜖0 𝑙𝑥

R 𝐶 = 𝐶1 + 𝐶2 = 2 +2
𝑅 𝑅
ln( ) ln( )
𝑟 𝑟
lx

l

figure 4.6

For conductive liquids the probe is covered by a layer of dielectric material with thickness t and
dielectric constant d. In this case we have approximately
𝜋𝜖𝑑 𝑙(1 − 𝑥)
𝐶1 ~2 𝑟+𝑡
ln( 𝑟 )
Having neglected the capacitance of the probe in the part coupled through air to the container
walls.

The same structure can be used also for measurement of level in containers used for solid
particles or powder.

Humidity sensors

In this case a capacitor is constructed using a suitable dielectric whose dielectric constant
depends on humidity. Typically the humidity sensitive materials are porous metal oxide or
polymers. These materials, by absorbing water vapor, change their average electrical behavior,
(H2O=800). These sensors are characterized by a value of some hundreds of pF and by a linear
relationship between capacitance and Relative Humidity with a sensitivity of some pF/RH

These sensors have a typical accuracy of ±2% RH in the range 5–95% RH. Without calibration,
the accuracy is 2 to 3 times worse.

Temperature sensors
The dielectric constant depends on temperature. Some materials (ferroelectrics) have a
sensitivity large enough to be the basis for a capacitive sensor.

The relationship is described by the following equation:

𝑘
𝜖=
𝑇 − 𝑇𝑐

Where Tc is acrytical temperature.

Main problems related to capacitive sensing.

Electrical filed/charge sensors

Obviously a capacitance can sense an electrical field or a charge, actually due to the leakage
current, it can sense only a variation of the electrical field or a moving charge.

Figure 4.7 (measurement of static electrical field by means of moving shield)

Electrical noise.
The main problems related to the use of capacitive sensing is due to the effect of external electric
fields that cause charge variation on the sensing electrode or on any part of the wire or cable used
to connect the sensing plate itself to the front-end electronics, that are in all respects as a part of
the metallic plate of the capacitance. To study the problem let’s consider a grounded capacitive
sensor (Cs) driven by an excitation current Is and a worst case noise source, described by an
ideal voltage generator. With reference to the following figure, the voltage, Vin, at the input of
the front-end amplifier can be found by opening the current source as follows:

Noise source
Vd
Cds
Cdv=Cds + Cdc
Cdc cable Front – end input
cable Front – end input

sensor Ccg Ri Ci
Is Ri Ci
Vd Cs Ccg

Figure 4.8 a) b)

𝑉𝑑 𝑅𝑖 𝐶𝑑𝑣 𝑗𝜔
𝑉𝑖𝑛 = 𝑉𝑑
1 + 𝑗 𝜔𝑅𝑖𝑛 (𝐶𝑑𝑣 + 𝐶𝑠 + 𝐶𝑔𝑐 + 𝐶𝑖 )
1 𝐶𝑑𝑣
The cut off frequency being 𝑓𝑡 = 2 𝜋𝑅 (𝐶 +𝐶 +𝐶 and the attenuation given a
𝑖 𝑖 𝑠 𝑔𝑐 +𝐶𝑑𝑣 ) 𝐶𝑠 +𝐶𝑖 +𝐶𝑔𝑐
certain coupling configuration, the problem is particularly severe if the front end circuit has a
large value input resistance (impedance) or if the overall measurement system has a large value
impedance.

Note that Vd, the parasitic capacitance Cdv, and Ccg, are all variable because the environment
changes: all moving object charged by triboelectricity and the power lines are sources of
variable disturbances. For this reason it is impossible to use capacitive sensors without the
proper use of shields. Electrostatic shields are metallic structure enclosing the ‘sensor plate’
which should be closed and kept at a fixed voltage, the easiest solution are grounded shields.

Electrostatic shields
The sensor and the cable should be surrounded by a metal layer which forms a perfectly closed
surface, this will grant that the external electric fields can’t affect in any way the measured
capacitance value. Wherever the shield is interrupted a small surface of the inner measuring
surface can couple with the electrical environment, thus being subjected to its effect. The sensor
could be grounded or not, the choice sometimes is impossible there being constraints in the
application, for instance to sense the displacement of grounded objects one of the two plate is
necessarily connected to ground. Otherwise the choice has to be made in order to select the best
front end circuit. Note that all the coupling parasitic capacitance in an unshielded structure are
variable.

C1env C2env
Cs
1 2

C1g C2g

Unreferenced,
unshielded sensor

Figure 4.9 a )

In the following figure a grounded shield surrounds the sensor and the wires. This allows for
canceling (or strongly reducing) the coupling capacitances with the environment and for setting
to a constant value the capacitances towards ground . This is an almost perfect solution to the
electric noise problem, but in cases where Cs has a small value the variable sensor capacitance is
placed in a networks of parasitic capacitances with large values, representing the electrical
coupling with the shield body (ground). This is simple to be understood if looking at figure 4.8.b,
where it is clear that whatever the front-end circuit is, it senses the parallel of Cs and C1s and C1s
can be much larger (many order of magnitude) than Cs. A typical example is a non-contact
displacement sensor placed far away from the readout electronics by means of some meters of
coaxial cable. In this case the value of Cs is probably less than 1 pF, whereas the value of C1s
(100 pF/m) can reach some nF.

1 Cs
Cs
1 2 1 2
1 2
Cs C1s C1s C2s
2

Grounded single shielded Unreferenced single

sensor shielded sensor

Figure 4.9 b) c)

To solve this problem tri-axial cables can be used in order to introduce a driven shield called
‘guard’ as shown in the following figure.
 C2s -
2
2
+
1 Cshield-ground

Cs

1
Unreferenced double shielded To/from the front-end
sensor

Figure 4.10 a) b) c)

In case of an ideal buffer the capacitance C2s is kept at 0 V, hence it doesn’t affect the
measurement. In case of a real amplifier the effect is that of reducing the capacitance in parallel
with the sensor down to the value C2s/A0, being A0 the open loop gain of the amplifier.

Parasitic effects
A capacitor can’t never be described only by the capacitance value, at low frequency the
capacitor can be modeled by the parallel of a capacitance and a resistance, describing the leakage
currents in the dielectric and in the enclosure. At high frequency the electrical model becomes
more complicated

Figure 4.11

Parameters such as the loss angle tg=Pa/Pr (ratio of active power to reactive power) or its
inverse the quality factor Q = Pr/Pa can be used to describe the ideality of the capacitor.
Capacitance related to displacement sensor in which the dielectric is air are quite close to
ideality. Water vapor can cause some problems in this respect.

The displacement sensor principles were described in the previous sub-sections considering a
infinite parallel plate structure, nevertheless when realizing a capacitive probe the finite area of
the plate cause a deviation from this simple behavior that is larger if the ratio probe area/target
distance decreases. In particular the field at the border of the capacitor is distorted causing a
‘fringe effect’ that depends on the probe-target geometry and on the electrical environment.

Cf Cf
Cs
ideal

Figure 4.12

So the capacitance-distance relationship deviates from the ideal one and is affected by external
electrical fields. In order to reduce this problem a ‘triaxial probe’ can be used, i.e. the sensing
plate is surrounded by a metallic ring which is kept at the same potential as the sensing plate
itself, if the gap between the central conductor (the active one) and the guard ring is small, the
structure can be thought as a larger probe in which only the central part is used for
measurements, where the field is uniform and not sensitive to the external electric environment.

Figure 4.13
ln(R𝑐 )
With reference to the geometry and symbols defined in figure 4.13, if R g − R c >> d and if
𝜋
g<<Rc a good approximation for Cs (capacity of the probe) is:

𝑔 2
𝜋 (𝑅𝑐 + 2 )
𝐶𝑠 =
𝑑
The most accurate measurement can be obtained using triaxial probe in conjunction with driven
triaxial cables.
Displacement sensors and unreferenced targets

S
CsT
S
T
C= CsT CTg /(CsT +CTg) CsT
CTg

C1s

Figure 4.14

The displacement sensor can be used also for unreferenced targets if the surface of the target
facing ground is sufficiently large with respect to the one in front of the sensor.

Front end electronics.

Many different types of conditioning electronics are used for capacitive sensors.

We distinguish among different classes of measurement problems:

Depending on type of signal:

a) AC Signals (vibrations as for instance in capacitive microphone), in this case the

excitation can be continuous.
b) Low frequency-DC signals (static measurement of position), in this case the excitation is
an AC signal generated by an oscillator which is a part of the front-end. The AC
excitation must have a frequency well above the signal maximum frequency and it is
selected also in order to get a reasonable value of the sensor impedance. AC excitation is
usually in the range 100 kHz-500 kHz.

Depending on the measurement topology

c) Grounded sensor (e.g. displacement sensors with grounded target) this is a constraint for
the front end design.
d) unreferenced sensor.

Depending on the capacitance value

a) High impedance sensor (C in the order pF or less) when the sensor capacitance is
obtained from a sensing plate coupled to the external electrical environment
 (displacement sensors, charge sensors, electrical field sensors), then the capacitance
usually can’t assume large values
b) Medium impedance sensors, when the sensor consists of an actual capacitor (humidity,
temperature sensors or capacitive sensors in MEMS).

Typical solution for condition systems are:

Bridges (AC/DC)
Used for differential capacitive probes, this circuits provides a voltage output proportional to x,
provided it is an AC signal (for DC signals the parasitic resistances determine the actual voltage
divider). For an half bridge.

VCC

C1
Vout

C2

Figure 4.15

𝐶0
𝐶1 1 −𝑥 𝐶0 (1 + 𝑥) 1+𝑥
𝑉𝑜𝑢𝑡 = 𝑉𝐶𝐶 = 𝑉𝐶𝐶 = 𝑉𝑐𝑐 = 𝑉𝑐𝑐
𝐶2 + 𝐶1 𝐶0 𝐶0 𝐶0 (1 + 𝑥) + 𝐶0 (1 − 𝑥) 2
1−𝑥+1+𝑥
For microphones the simple following circuit is used:

VCC

C1
Vout

Figure 4.16
Charge/Current amplifiers (AM measurement chain).
Rf
Cf

Cs
- Vo Vout

Vs
Amplitude detector
Cin +

oscillator

Figure 4.17

These are low input impedance circuits, very accurate usually with a limited bandwidth (audio-
frequency). Used for instance in MEMS capacitive accelerometers. Can be used for DC signals,
Vs comes from an oscillator.

For electric field measurements or charge sensing a similar structure is used as a charge
amplifier.

Rf
Cf

Cs
- Vo
Vs

Cin +

Figure 4.18

(𝑉𝑠 − 𝑉− ) 𝑗𝜔𝐶𝑠 − 𝑗𝜔𝐶𝑖𝑛 𝑉− + (𝑉0 − 𝑉_)(1/ 𝑅𝑓 + 𝑗𝜔𝐶𝑓 ) = 0

𝑗𝜔𝐶𝑠 𝑗𝜔𝐶𝑠 𝑅
𝑉0 = 𝑉𝑠 𝐶𝑖𝑛 +𝐶𝑠 1 1 𝑉𝑠 1+𝑗 𝜔𝑅 𝑓𝐶
𝑗 𝜔 +(1+ )( +𝑗 𝜔𝐶𝑓 ) 𝑓 𝑓
𝐴0 𝐴0 𝑅𝑓

The input parasitic capacitance (particularly large if long cables are used) is reduced thanks to
the feedback circuit of a factor A0 (the open loop gain of the OP AMP). Vs is the output of an
oscillator with a fundamental frequency, fs, higher than the cut-off frequency, therefore the
output is amplitude modulated by the value Cs. Obviously if fx is the bandwidth of the signal
(variation of Cs) we must have fx<<fs.

Usually the oscillator fundamental is in the order of 200 kHz, so to ensure a measurable sensor
impedance and sufficient bandwidth.

The front end can be completed by an AM demodulator system, an amplitude detector, such as
the one shown below with a low pass bandwidth >> fx.

R1

R1
Vin - Vo
SW1
+
R1/2
vc
SW2

Figure 4.19

𝑅1 𝑅1
𝑖𝑓 𝑣𝑐 𝑖𝑠 𝐻𝑖𝑔ℎ − 𝑆𝑊1 𝐶𝑙𝑜𝑠𝑒𝑑 𝑆𝑊2 𝑜𝑝𝑒𝑛: 𝑉𝑜 = 𝑉𝑖𝑛 (1 + ) − 𝑉𝑖𝑛 ( ) = 𝑉𝑖𝑛
𝑅1 𝑅1

𝑅1
𝑖𝑓 vc 𝑖𝑠 𝑙𝑜𝑤 − 𝑆𝑊2 𝐶𝑙𝑜𝑠𝑒𝑑 𝑆𝑊1 𝑜𝑝𝑒𝑛: 𝑉𝑜 = −𝑉𝑖𝑛 ( ) = −𝑉𝑖𝑛
𝑅1

Note that the circuit in figure 4.19 is a demodulator and can recover the phase of the AM
modulating signal, tis is usually not required for capacitive sensor, and a simpler envelope
detector can be used.

In case the sensor capacitance is small and /or the cable introduce a capacitance too high a driven
shield can be used.

If a differential capacitance sensor is considered (MEM combs) then a more sensitive AM

circuit can be used exploiting two voltage generators with opposite phase.

Neglecting the effect of input capacitances:

 Rf
Cf
Cs2

Vs
- Vo

-Vs
Cin +
Cs1

Figure 4.20

𝑉𝑠 𝐶0 1 1 𝑉𝑠 𝐶0
𝑉𝑜 = 𝑉𝑠 𝑍𝑓 𝑗 𝜔𝐶𝑠1 − 𝑉𝑠 𝑍𝑓 𝑖𝜔𝐶𝑠2 ≅ ( − )≈ 2𝑥
𝐶𝑓 1 − 𝑥 1 + 𝑥 𝐶𝑓

Frequency Modulated (FM) measurement chains:

The sensor is a part of an oscillator, the variation of the capacitance determines the variation of
the resonant frequency. A FM demodulator converts the frequency variations into voltage
variations.

Figure 4.21
High frequencies (some MHz) are used for large bandwidth systems (more than 100 kHz) with a
good accuracy, usually are used for AC signals.

Low frequency (relaxation oscillators) are used for humidity sensors.

Switched capacitance.
Some ICs exploit this structure, which adapts the output impedance for easy connection, the
measurement is performed by means of a certain number of repeated cycle with a duration long
enough to grant the complete charge of the capacitances and short enough to neglect t heir
discharge. In each cycle at first the switch SW1 is closed and SW2 open (phase 1)then SW1
opens and SW2 is closed (phase 2).

SW1 SW2

VR
Cs CR

Figure 4.22

With CR>>Cs

At the end of the first phase:

𝑄𝑠 = 𝐶𝑠 𝑉𝑅

At the end of the second phase:

𝑄𝑅 (𝑛 + 1) = 𝑄𝑅 (𝑛) + 𝐶𝑠 𝑉𝑅

𝑄𝑅 (𝑛) + 𝐶𝑠 𝑉𝑅 𝐶𝑅 𝑉(𝑛) 𝐶𝑠 𝑉𝑅 𝐶𝑠 𝑉𝑅 − 𝐶𝑠 𝑉(𝑛)

𝑉(𝑛 + 1) = = + = 𝑉(𝑛) +
𝐶𝑠 + 𝐶𝑅 𝐶𝑅 + 𝐶𝑠 𝐶𝑅 + 𝐶𝑠 𝐶𝑅 + 𝐶𝑠

If V(n)<<VR

𝐶𝑠 𝑉𝑅 𝐶𝑠 𝑉𝑅
𝑉(𝑛 + 1) ≅ 𝑉(𝑛) + ≅ 𝑉(𝑛) +
𝐶𝑅 + 𝐶𝑠 𝐶𝑅

And starting with V(0)=0V, we have a linear relationship:

 𝐶𝑠
𝑉(𝑛 + 1) = (𝑛 + 1) 𝑉
𝐶𝑅 𝑅

10

6
V(n) (V)

5 Cr=500pF; Cs =1pF; VR=10V

2
LINEAR
1

0
0 500 1000 1500 2000 2500 3000 3500
n=number of cycles

Figure 4.22
Inductive and variable reluctance sensors

In this section inductive sensors or more in general sensors based on a variation of the magnetic
flux in a measuring coil are described. These devices are based on the most mature technology
for sensing devices and are still widely used in industrial applications.

The basic law exploited by all these sensors is

𝑑Φ(𝐵)
𝑣=−
𝑑𝑡
Where v is the voltage in a circuit (closed loop) caused by the variation of the flux of the
magnetic field across the area enclosed by the circuit itself.

The variation of (B) can be caused by a motion of the magnetic environment with respect to
the circuit or viceversa by a motion of the circuit. In any case motion is transduced into an
electrical quantity.

Figure 4.23

To better understand the operating principle of these sensors let’s define some fundamental
quantities related to the so called magnetic circuits. A magnetic circuit (on analogy to electric
circuits) consists of a ferromagnetic material (a core) on a region of which a coil consisting of N
turns is winded (see figure). From Maxwell equations, being:

𝜕𝐷
∇×𝐻 =𝐽+
𝜕𝑡
We consider D=0 and the integral form of the above equation
 ∮ 𝐻 ∙ 𝑑𝑙 = ∫ 𝐽 ∙ 𝑛 𝑑𝐴
Γ 𝐴

Which relates the line integral of H on a close line  with the current crossing a surface A (n is
the unit vector perpendicular to the surface) enclosed by the line. This relationship can be written
more conveniently as follows:

𝐵
∮ ∙ 𝑑𝑙 = 𝐼
Γ 𝜇

Where I is the current crossing the surface.

Now considering the circuit in the figure and a line  on the core axis, and being the cross
section area of the core much smaller than the squared length of , we can consider the field B
to be parallel to  and we can write

𝐵 𝐵
𝑙𝑐 + 2 𝑙 = 𝑁𝐼
𝜇𝑐𝑜𝑟𝑒 𝜇0 𝑔

Φ(𝐵) 𝐵Φ(𝐵)
𝑙𝑐 + 2 𝑙𝑔 = 𝑁𝐼
𝐴𝑐 𝜇𝑐𝑜𝑟𝑒 𝐴𝑔 𝜇0

𝑙𝑔 𝑙𝑐
𝑔 = ; 𝑐𝑜𝑟𝑒 =
𝐴𝑔 𝜇0 𝐴𝑐 𝜇𝑐𝑜𝑟𝑒

Φ(𝐵)(𝑐𝑜𝑟𝑒 + 2𝑔 ) = 𝑁𝐼 = 𝐹𝑚𝑚

assuming small distortion of the B field and the continuity of the perpendicular component of the
B field at the medium interface (Bn (gap)= Bn (core) and Hn (gap)=core/0 Hn (core) ) .

If the gap has a length lg much smaller than the length of , then it is possible to consider AcAg
𝑙
NI is named magnetomotive force with symbol Fmm and  = 𝐴 is called magnetic reluctance.
𝜇
In general Fmm can be taken as the equivalent of the voltage in an electric circuit, (B) can be
considered as the equivalent of the current, and finally the reluctance acts as an electric
resistance. The equation obtained for the magnetic circuit shown in the figure can be seen as the
equivalent of Ohm law, and of the Kirchoff voltage law. A law equivalent to Kirchoff current
law can be demonstrated, from ∇ ∙ 𝐵 = 0 and Gauss theorem.

Given a certain magnetomotive force the flux of B in a circuit section is proportional to the
cross section area of the core, and to the magnetic permittivity whereas it is inversely
proportional to the length (Hopkinson law). We can imaging that a coil generates a magnetic flux
only in the magnetic circuit because the core has a large magnetic permittivity value with respect
to the surrounding medium (air), such as the electrical current is confined in conductors in
electrical circuit because they have a conductivity much larger than the surrounding medium.
Nevertheless being the ratio /0 for a ferromagnetic material (<80000) smaller than the ratio
/0 (>108 and 0 is the parasitic conductance describing leakage currents in insulators) for a
conductor, the magnetic circuit described by the simple equations above is a much worse
approximation of reality with respect to an electric circuit.

Summarizing, when using the analogy between magnetic circuits and electric circuits, the limitations of
this analogy must be kept in mind. Electric and magnetic circuits are only superficially similar because of
the similarity between Hopkinson's law and Ohm's law.
Electric currents represent the flow of particles (electrons) and carry power, which is dissipated as heat in
resistances. Magnetic fields don't represent the "flow" of anything, and no power is dissipated in
reluctances.
The current in typical electric circuits is confined to the circuit, with very little "leakage". In typical
magnetic circuits not all of the magnetic field is confined to the magnetic circuit; there is significant
"leakage flux" in the space outside the magnetic cores, which must be taken into account but is difficult to
calculate.
Most importantly, magnetic circuits are nonlinear; the reluctance in a magnetic circuit is not constant, as
resistance is, but varies depending on the magnetic field. At high magnetic fluxes the ferromagnetic
materials used for the cores of magnetic circuits saturate, limiting the magnetic flux, so above this level
the reluctance increases rapidly. The reluctance also increases at low fluxes. In addition, ferromagnetic
materials suffer from hysteresis.

Moreover it is useful relating the magnetic reluctance to self-inductance. Taking into account a
coil consisting on N turns with cross section area S, length l around a core with magnetic
permittivity , we can relate the voltage drop to the current using the following equations:

𝑑Φ 𝑁 2 𝜇𝑆 𝑑𝑖 𝑑𝑖
𝑣𝑐𝑜𝑖𝑙 =𝑁 = =𝐿
𝑑𝑡 𝑙 𝑑𝑡 𝑑𝑡

𝑁2
𝐿=


So a variation of reluctance causes a variation of the inductance and of impedance.

A magnetic circuit can be exploited for the transduction of displacement/velocity into flux of B
variation, two types of devices can be distinguished: variable gap and movable core.

Variable gap devices are used for non - contact measurement of both distance (magnetic probe
and eddy current sensors) and velocity (pick up), whereas movable core are essentially contact
displacement sensors.

Variable reluctance sensors: Non-contact displacement/velocity probe

I
Sensing coil

d R

target

Figure 4.24

With reference to the figure above we have:

𝜋𝑅 2𝑅 𝑑
𝑁𝐼 = Φ(𝐵) + Φ(𝐵) + 2 Φ(𝐵)
𝜇𝑠𝑒𝑛𝑠𝑜𝑟 𝜋𝑟 2 𝜇𝑡𝑎𝑟𝑔𝑒𝑡 𝐴𝑡 𝜇0 𝜋𝑟 2

𝜋𝑅 2𝑅 𝑑
0 = + ; 𝑑 =
𝜇𝑠𝑒𝑛𝑠𝑜𝑟 𝜋𝑟 2 𝜇𝑡𝑎𝑟𝑔𝑒𝑡 𝐴𝑡 𝜇0 𝜋𝑟 2

𝑁𝐼 = 0 (1 + 2𝛼𝑑)

𝑑 1 𝜇𝑠𝑒𝑛𝑠𝑜𝑟 𝜇𝑡𝑎𝑟𝑔𝑒𝑡 𝜋𝑟 2 𝐴𝑡 𝜇𝑠𝑒𝑛𝑠𝑜𝑟

𝛼= = 
0 𝜇0 𝜋𝑟 𝜇𝑡𝑎𝑟𝑔𝑒𝑡 𝐴𝑡 𝜋𝑅 + 2𝑅𝜇𝑠𝑒𝑛𝑠𝑜𝑟 𝜋𝑟 𝑅(𝜋 + 2)𝜇0
2 2

The rightmost expression is obtained for At=r2 and target=sensor and can be used to
understand that this device operates better with ferromagnetic target.

The variation of the gap that induce a variation of the reluctance can be sensed either measuring
the inductance of the exciting coil, or by measuring the voltage across a sensing coil due to the
mutual inductance. To sense static displacement an AC excitation is provided, whereas for
moving target the flux variation can be large enough to be sensed without an AC excitation, in
this case the excitation coil is not necessary and the core can be replaced by a permanent magnet.
This is the structure of magnetic pick-up the are used to sense the speed of moving target, since
the output voltage in directly proportional to the variation of flux, hence to target speed.
 Figure 4.25

Actually if the target is not ferromagnetic but it is a conductor, the magnetic flux changes with
distance due to the presence of eddy currents. Sensors exploiting this effect have the same basic
structure of the one in figure but they are called eddy-current sensors

Eddy current sensors

In conductive target currents called eddy current are created by the variation of the flux in the
target, since the induced voltage pushes the free electrons in the conductor. These currents
create in turn an opposing magnetic field which counteracts the field being generated by the
probe coil. The interaction of the magnetic fields depends on the distance between the probe but
also on the characteristics of the target. As the distance changes, the electronics sense the change
in the field flux.

It is important to distinguish between "high-performance" eddy-current sensors and inexpensive

inductive sensors. Simple inductive sensors, such as those used in proximity switches, are
inexpensive devices that sense the ‘presence’ of a target. Proximity type sensors are
tremendously useful in automation applications and many commercially available models and
are well made for their intended applications, but they are not suited to applications needing
accuracy.

In contrast, eddy-current sensors for use in accurate displacement measurements use complex
electronic designs and exploit complex signal processing.

Being the operating principle of this device related to the presence of currents in the target, a
sufficient sensitivity is obtained only if the target itself has a minimum thickness that can be
found from the field penetration depth (“skin-depth.”, ) as a rule of thumb the minimum target
thickness is evaluated as three times this depth:
 𝜌
𝛿= √
𝜋𝑓 𝜇

Where ρ is the electrical resistivity, f the field frequency and μ is the magnetic permeability

Field density decreases exponentially with depth (1/e). At three skin-depths eddy current density
is about 5% of the surface density. Three skin-depths is the minimum required depth.

Nonferrous Materials

Minimum
Thickness
Material ρ (μcm) μr f (MHz)
(mm)

Silver 1.59 1 (1.0) 0.19

Copper 1.71 1 (1.0) 0.2

Gold 2.21 1 (1.0) 0.22

Aluminum 2.65 1 (1.0) 0.25

Zinc 5.97 1 (1.0) 0.37

304 SST 72 1.01 (1.0) 1.27

Lead 20.8 1 (1.0) 0.69

Brass 6.4 1 (1.0) 0.38

Tin 11.5 1 (1.0) 0.51

Titanium 47 1 (1.0) 1.03

 Ferrous & Magnetic Materials

Minimum
Material ρ (μcm) μr f (MHz) Thickness
(mm)

Nickel 6.4 1240 (1.0) 0.01

416 SST 57 700 (1.0) 0.04

Iron 11 100 (1.0) 0.05

Table 4.I Data from Lion Precision

The strength of the eddy current and the resulting magnetic field depend on the permeability and
resistivity of the material. These properties vary between different materials. They can also be
changed by different processing techniques such as heat treating or annealing. For example, two
otherwise identical pieces of aluminum that were processed differently may have different
magnetic properties. Between different nonmagnetic materials such as aluminum and titanium
the variance of permeability and resistivity can be small, but a high performance eddy-current
sensor calibrated for one nonmagnetic material will still produce errors when used with a
different nonmagnetic material.

The differences between nonmagnetic materials like aluminum and titanium and magnetic
materials such as iron or steel are enormous. While the relative permeability of aluminum and
titanium are approximately one, the relative permeability of iron can be as high as 10,000.
Eddy-current sensors calibrated for nonmagnetic materials are not likely to function at all when
used with magnetic materials. When using eddy-current sensors for precise measurements, it is
critical that the sensor be calibrated for the specific material used in the application.
The high permeability of magnetic materials such as iron and steel can also cause small eddy-
current sensor errors within the same piece of material. Within any imperfect material, there are
microscopic cracks and material variations. The material permeability changes slightly around
these areas. While the changes are relatively small, the extremely high permeability of magnetic
materials enables high-resolution eddy-current sensors to detect these changes. This problem is
most evident in rotating targets of magnetic materials.
 Figure 4.26 profile plot showing
actual profile in blue,
and electrical profile from
eddy-current sensor in red.

An eddy-current sensor can be mounted to measure the profile of a rotating shaft. But even if the
shaft is ideal, a high resolution eddy-current sensor will detect a repeatable pattern of changes as
the shaft rotates. These changes are a result of small variations in the material. This phenomenon
is well known and is called electrical run-out. These errors can be small, often in the micron
range. Many shaft run-out applications, especially those in hostile environments where eddy-
current sensors are the norm, are looking for much larger errors and can therefore tolerate these
errors. Other more precise applications will need to use techniques to address these errors or use
a different sensing technology such as capacitive sensors.

The electric field of a capacitive sensor uses the target as a conductive path to ground. All
conductive materials offer this equally well, so capacitive sensors measure all conductive
materials the same. Once a capacitive sensor is calibrated, it can be used with any conductive
target with no degradation in performance.

Because the electric field of a capacitive sensor does not penetrate the material, variations within
the material do not affect the measurement. Capacitive sensors do not exhibit the electrical run-
out phenomenon of eddy-current sensors and can be used with rotating targets of any conductive
material without additional error.

Eddy-current sensors should be calibrated to the same material as the target in the application
and should not be used with rotating magnetic material targets unless the electrical run-out errors
are acceptable in the application. Capacitive sensors, once calibrated, can be used with any
conductive material with no material related errors, and they work well with rotating targets.

Front –end for inductive/eddy current sensors

The method of the signal conditioning can be classified as for capacitive sensors:
using an external excitation signal (EO) and self-oscillation (SO). Principle (i) is
based on measuring the amplitude of a signal exciting the coil coming from a stable
external oscillator (the displacement of the target modulates the amplitude of the
signal). In the alternative interfacing method, the coil is a part of an oscillator due to
very large variation of impedance that can be sometimes obtained with these sensors
in case of SO the information of interest can be the frequency (FM) or the amplitude
(AM).
AM – EO front-end electronics as the bridge circuits (or auto-balanced bridge circuits)
as the ones in depicted in Figure 4. can be used for accurate measurement. A
differential impedance circuit, in which the coils of a pair of sensors form the opposite
branches of a balanced bridge circuit. This differential configuration achieves its high
resolution by eliminating any intervening summation and log amplifiers that would
normally add noise and drift to the system. Additionally, by using electrically matched
sensors on opposing legs of the same bridge, temperature effects common to the
sensors and cabling of a differential sensor pair tend to be cancelled. This is true for
the mechanical aspects of the sensor/target system also.

Figure 4.27 From Lion precision

Figure 4.27 b

The excitation frequency has to be far from the coil resonance frequency

Figure 4.28 from Kalman

An example of SOE – AM example is shown in figure 4. (from Kaman), like the inductive
bridge circuits, in the Colpitts oscillator circuit the sensor acts as the resonating coil for the
oscillator. The proximity of the targets to the sensor coil pulls the oscillator, changing its
frequency (small change not used) and amplitude (large change) of modulation. The signal is
half-wave rectified (AC to DC) and filtered to obtain an analog voltage proportional to the target
position. The analog voltage output can be varied by adjustment of the gain. The input power is
diode protected and regulated to provide a clean low noise signal. The output is short-circuit
current protected.
Non-contact displacement sensors: a comparison between capacitive and
inductive/eddy sensors

There are many factors to consider when choosing between capacitive and eddy-current
displacement sensors. Any application that involves measurement area contaminants such as
liquids or waste material requires eddy-current sensing. Capacitive sensors require a clean
environment. In contrast inductive sensors are very sensitive to temperature whereas capacitive
probe are robust to temperature changes. Fluctuations in sensor coil temperature affect the output
of inductive linear displacement sensors. This is primarily due to the affect the temperature has
on the electrical properties of the wire used to manufacture the coil.

Small targets will be more easily measured with capacitive sensors because of the comparatively
small size of the capacitive sensing field. When eddy-current sensing is required, special
calibration can be used with small targets.

For the same size capacitive or eddy-current probe, the eddy-current probe will have a larger
measurement range (ratio range/probe diameter =1:8 capacitive,=1:3 inductive), but the spot size
for a capacitive sensor has a diameter 1.3 times the one of the probe, whereas for an eddy
current probe the spot size is approximately 3 times the probe size.

Because capacitive probes interact with the surface of the target, the material thickness is not a
factor in capacitive measurements. Eddy-current sensors have minimum target thickness
requirements.

Capacitive sensors have no sensitivity to the target material provided it is conductive or surface
structure. Eddy-current sensors are sensitive to material differences and must be calibrated to the
application’s target material.

When using multiple probes, capacitive sensors must be synchronized, but can be mounted close
together without interference. Even when synchronized, eddy-current probes will interact if
mounted close together. When this is unavoidable, special calibration can be used but is only
available with digital sensors.

A capacitive probe small sensing field, which is directed only at the target, prevents it from
sensing mounting hardware or nearby objects. Eddy-current’s large, surrounding sensing field
can detect mounting hardware or other objects if they are too near the sensing area.
Two other specifications differ between the two technologies: resolution and bandwidth.
Capacitive sensors have higher resolutions than eddy-current sensors due for instance to
temperature effect making them a better choice for high resolution, precise applications.
Most capacitive and eddy-current sensors have bandwidths of 10-15kHz, but some eddy-current
sensors have bandwidths as high as 80 kHz, capacitive with FM front-end can reach bandwidth
of 200kHz.

Another difference between the technologies is cost. Generally speaking, eddy-current sensors
are lower cost.

PICK UP velocity sensors (active)

Figure 4.29

Figure 4.30

The magnetic pickup is a speed sensor which simply consists in a coil wound around a permanent
magnet, it is a self-generating sensor. The magnetic pickup produces a voltage output when any
magnetic material moves through the magnetic field at the end of the pickup. Since most engines and
turbines have flywheels or other large gears made of magnetic material (usually iron or steel), magnetic
pickups can usually be installed without adding attachments to a gear or shaft. Nonmagnetic materials,
such as aluminum, brass, and some stainless steels, will not excite the magnetic pickup. The MPU makes
use of a “stray magnetic field”. Any device which produces a dynamic discontinuity of magnetic material in
the field of the pickup will produce a variation of reluctance and a B-flux variation and hence an electrical
voltage in the coil. Although gears are the normal devices measured by an MPU, other devices such as a
vibrating surface, moving bar, crank, wheel spokes, or a steel screw head mounted on some moving
surface will work equally well if surface speed and other factors are taken into account.
The output voltage of a magnetic pickup is affected The output voltage of a magnetic pickup is affected by
three factors.
Voltage increases with increases of the surface speed of the monitored magnetic target.
Voltage decreases as the air gap between the magnetic pickup and the surface of the gear tooth is
increased.
Voltage waveform is determined by the size and shape of the gear tooth in relation to the size and
shape of the pole piece.
With any given speed and clearance conditions, a maximum power output will result when the field is
filled with a relatively infinite mass of magnetic material at one instant and a complete absence of such
material the next. For this reason the MPU is usually used in conjunction with a toothed gear that can
approximate these conditions when the cross-section of the target (tooth) is equal to or greater than that
of the pole piece, and the space between the teeth is equal to or greater than three times the diameter of
the pole piece (see Figure 1-1).

Figure 4.31 –Pickup and toothed wheel (Woodward)

LVDT (linear variable transformer) –movable core – contact

displacement sensor.

The linear variable differential transformer (LVDT) (also called just a differential transformer,
linear variable displacement transformer, or linear variable displacement transducer) is used for
measuring linear displacement (position). A counterpart to this device is used for measuring
rotary displacement and it is called a rotary variable differential transformer (RVDT).
Operation
The linear variable differential transformer has three solenoidal coils placed end-to-end around a
tube. The center coil is the primary, and the two outer coils are the top and bottom secondaries.
A cylindrical ferromagnetic core, attached to the object whose position is to be measured, slides
along the axis of the tube. An alternating current drives the primary and causes a voltage to be
induced in each secondary proportional to the length of the core linking to the secondary.[1] The
frequency is usually in the range 1 to 10 kHz.
As the core moves, the primary's linkage to the two secondary coils changes and causes the
induced voltages to change. The coils are connected so that the output voltage is the difference
(hence "differential") between the top secondary voltage and the bottom secondary voltage.
When the core is in its central position, equidistant between the two secondaries, equal voltages
are induced in the two secondary coils, but the two signals cancel, so the output voltage is
theoretically zero. In practice minor variations in the way in which the primary is coupled to
each secondary means that a small voltage is output when the core is central.
When the core is displaced toward the top, the voltage in the top secondary coil increases as the
voltage in the bottom decreases. The resulting output voltage increases from zero. This voltage is
in phase with the primary voltage. When the core moves in the other direction, the output voltage
also increases from zero, but its phase is opposite to that of the primary. The phase of the output
voltage determines the direction of the displacement (up or down) and amplitude indicates the
amount of displacement. A synchronous detector can determine a signed output voltage that
relates to the displacement.
The LVDT is designed with long slender coils to make the output voltage essentially linear over
displacement up to several inches (several hundred millimetres) long.
The LVDT can be used as an absolute position sensor. Even if the power is switched off, on
restarting it, the LVDT shows the same measurement, and no positional information is lost. Its
biggest advantages are repeatability and reproducibility once it is properly configured. Also,
apart from the uni-axial linear motion of the core, any other movements such as the rotation of
the core around the axis will not affect its measurements.
Because the sliding core does not touch the inside of the tube, it can move without friction,
making the LVDT a highly reliable device. The absence of any sliding or rotating contacts
allows the LVDT to be completely sealed against the environment.
LVDTs are commonly used for position feedback in servomechanisms, and for automated
measurement in machine tools and many other industrial and scientific applications.
 M1

I2
I1 RL
E1

M2

Figure 4.32 Front end for LVDT 4W and 5W

If R1 and L1 are the parasitic resistance and self-inductance of the primary winding whereas R2’
and L2’ are the parasitic resistance and self-inductance of the first secondary winding, R2’’ and
L2’’ are the parasitic resistance and self-inductance of the second secondary winding and R2= R2’
+ R2’+ RL and L2= L2’+ L2’’ , and finally M1 and M2 are the mutual inductances, we can write

𝐸1 = (𝑗 𝜔𝐿1 + 𝑅1 )𝐼1 + 𝑗 𝜔(𝑀2 − 𝑀1 )𝐼2

0 = (𝑗 𝜔𝐿2 + 𝑅2 )𝐼2 + 𝑗 𝜔(𝑀2 − 𝑀1 )𝐼1

𝜔(𝑀1 − 𝑀2 )
𝐼2 = 𝐸1 𝑗
(𝑗 𝜔𝐿1 + 𝑅1 )(𝑗 𝜔𝐿2 + 𝑅2 ) − 𝜔 2 (𝑀1 − 𝑀2 )
 𝜔(𝑀1 − 𝑀2 )𝑅2
𝐸𝑜𝑢𝑡 = 𝑅𝑙 𝐼2 𝑅2 𝐼2 = 𝐸1 𝑗
(𝑗 𝜔𝐿1 + 𝑅1 )(𝑗 𝜔𝐿2 + 𝑅2 )

Note that 𝑀1 = √𝐿1 𝐿′2 and 𝑀1 = √𝐿1 𝐿′′2 . It can be shown that L2 remains constant whereas the
difference of mutual inductance M1-M 2 varies with the displacement of the core, x, from the
central positions which ideally ensures a perfect equality of these parameters. With a proper
mechanical design, if the core moves in the measurement range a linear relationship exists: M1-
M 2 =kx.

𝑗𝜔𝑘𝑥𝑅2
𝐸𝑜𝑢𝑡 = 𝐸1
(𝑗 𝜔𝐿1 + 𝑅1 )(𝑗 𝜔𝐿2 + 𝑅2 )

Here x is considered constant (low frequency with respect to the excitation frequency).

Being the phase:

𝐸𝑜𝑢𝑡 𝜔𝐿2 𝑅1 + 𝜔𝐿1 𝑅2

𝜙( ) = Φ(𝑥) + 90° − atan( )
𝐸1 𝑅1 𝑅2 − 𝜔 2 𝐿1 𝐿2

2
𝐸𝑜𝑢𝑡 2
𝑅1 𝑅2
𝜙( ) = Φ(𝑥) → 𝑅1 𝑅2 − 𝜔0 𝐿1 𝐿2 = 0 → 𝜔0 = √
𝐸1 𝐿1 𝐿2

This condition sets the excitation frequency so that there is a zero phase shift from the primary
excitation voltage to the voltage output for positive displacements and -180° phase shift for
negative displacements.

𝑘𝑥
|𝐸𝑜𝑢𝑡 (𝜔0 )| | |
𝑅2

The LVDT converts the position x or linear displacement from a mechanical reference (zero, or
null position) into a proportional electrical signal containing phase (for direction) and amplitude
(for distance) information. The LVDT operation does not require an electrical contact between
the moving part (probe or core assembly) and the coil assembly, but instead relies on
electromagnetic coupling.
 Figure 4
Figure 4.34
The structure described is called four wires LVDT and it requires a carrier amplifier
(synchronous demodulator) to recover the direction of the displacement.
A variety of LVDT signal conditioning electronics is available, including chip-level and board-
level products for OEM applications as well as modules and complete laboratory instruments for
users.
A typical front-end circuit (integrated circuit AD698) is shown in the following figure, where the
blocks named A and B are synchronous demodulator channels used to detect primary and
secondary amplitude.

Figure 4.35

The output of the secondary is divided by the amplitude of the primary and multiplied by a scale
factor. This eliminates scale factor errors due to drift in the amplitude of the input or of
temperature.
To avoid the use of carrier amplifiers a 5 wires sensor can be used. As shown in the following
figures. In the right figure the IC conditioning circuit AD 598 block diagram is shown.

Figure 4.36

As AC operated LVDTs do not contain any electronics, they can be designed to operate at
cryogenic temperatures or up to 650 °C, in harsh environments, under high vibration and shock
levels. LVDTs have been widely used in applications such as power turbines, hydraulics,
automation, aircraft, satellites, nuclear reactors, and many others. These transducers have low
hysteresis and excellent repeatability.

Figure 4.36

The support electronics can also be self-contained, as in the DC-LVDT shown in Figure 4. These
easy-to-use position transducers offer practically all of the LVDT's benefits with the simplicity
of DC-in, DC-out operation. Of course, LVDTs with integral electronics may not be suitable for
some applications, or might not be packaged appropriately for some installation environments.