Journal of Electrostatics 87 (2017) 26e31

Contents lists available at ScienceDirect

Journal of Electrostatics
journal homepage: www.elsevier.com/locate/elstat

Measure of the capacitance coefficients of square electrodes using

capacitance-to-digital converters
Enrico Cataldo a, Alberto Di Lieto b, Francesco Maccarrone a, *, Giampiero Paffuti b
a
University of Pisa, Physics Department, Largo Pontecorvo 7, Pisa, Italy
b
University of Pisa, Physics Department and INFN Pisa, Largo Pontecorvo 7, Pisa, Italy

a r t i c l e i n f o a b s t r a c t

Article history: We measure the coefficients of self capacitance C11 and of interelectrode capacitance C12 of a pair of
Received 31 January 2017 square parallel plates, by using commercial models of capacitance to digital converters. The measured
Accepted 7 March 2017 coefficient depends on which charge is sampled, that of the excited electrode or that of the other one.
The analysis of the data at large distances is in agreement with the asymptotic formulas for the
capacitance coefficients. The different leading behaviour as a function of the distance gives coherent
Keywords:
independent estimations of the intrinsic capacitance of a single plate, in good agreement with the
Coefficients of capacitance
calculated one.
Asymptotic formulas for capacitance
Capacitance-to-Digital-converters
© 2017 Elsevier B.V. All rights reserved.
Capacitive sensors

PACS:
41.20.Cv
84.32 Tt

1. Introduction commonly feasible. On the other hand, the design of a real sensor
often requires a complete knowledge of the absolute capacitance of
Since several years, the readout of a capacitive sensor (CS) has a set of conductors having a given geometry. Over the past 30 years,
been highly simplified through the application of Capacitance-to- a new generation of CSs and actuators has grown in the framework
Digital Converter (CDC). They directly convert an unknown capac- of microelectromechanical systems (MEMS) technology. A large
itance to a digital word. Nowadays, monolithic CDCs offer direct market of MEMS devices has been developed offering many in-
interface of analog sensors to digital systems. Currently, they are struments able to sense and regulate very small displacements and
the first choice for experimenters working with instrumentation forces through the measure and the control of a microscopic
based on CSs, both from the point of view of the cost and for the capacitor. In that kind of devices the dimensions of the micro-
limited needs of signal conditioning. Over the past few years, a electrodes are generally comparable to their separation and the
large number of papers have dealt with apparata in which the CDC field distribution around the electrodes must be considered in or-
readout of a CS is used in applications considered typical such as der to foresee the transfer function of the sensors.
with capacitive accelerometers or dynamometers [1,2], liquid level Although the analytical calculation of the capacitance of a sys-
sensors [3] or proximity detectors and also in more fundamental tem of conductors has been a challenging task since long time [6,7]
studies ranging from Earth's gravimetry [4] to the detection of tiny and exact solutions are available only for a small variety of geom-
traces of chemical substances adsorbed to one of the electrodes etries [8,9], accurate numerical techniques are largely applied for
forming the CS [5]. many typical arrangements of the electrodes [10e13].
In most of the experimental practice, an accurate accordance From the point of view of measurement technology, the changes
between the ideal model and the real device is not mandatory, in the capacitance value are detected by connecting the capacitive
since much of the interest resides into the relative changes of the sensor to an electric circuit and by applying an excitation signal, e.g
capacitance and an appropriate calibration of the sensor is a static or variable voltage, between the electrodes of the capacitor
and by registering the change of the amplitude, phase or frequency
of some electrical quantity such as current, voltage, charge at
* Corresponding author. appropriate test points of the circuit [14].
E-mail address: [email protected] (F. Maccarrone).

http://dx.doi.org/10.1016/j.elstat.2017.03.002
0304-3886/© 2017 Elsevier B.V. All rights reserved.
 E. Cataldo et al. / Journal of Electrostatics 87 (2017) 26e31 27

In this paper we present the results of a comparison of nu-

2
C11 C22
C12
merical modelisation of a capacitor with parallel plane geometry
C¼ (1)
with the capacitance measurements of a parallel pair of metal C11 þ C22 þ 2C12
plates, the basic geometry adopted in the description of a CS. We
used two models of the most popular CDC device available in the having taken into account that C12 ¼ C21 .
market, manufactured by Analog Device [15]. The aim of the study A system made of two plane conductors of the same shape and
is to clarify the principle of operation of these instruments from the area A placed parallel to each other and separated by a distance d, is
point of view of the effective physical quantities detected. Partic- the pedagogical prototype of a two conductors capacitor and also
ularly, we show that it is possible to measure separately the the simplest model of the basic capacitive sensor. The parallel plate
capacitance coefficients Cij of a couple of conductors. We checked capacitor is very interesting from a practical point of view as similar
these properties measuring the self capacitance C11 and the inter- configurations are quite common in sensor devices in a large va-
electrode capacitance C12 of two equal conducting plates as a riety of applications. For example, small displacement are detected
function of the distance. We compared the experimental data with through the variation of the capacitance of the two plates induced
the numerical results and with the asymptotic behaviour of the two by the relative movement of one plate with respect to the other
quantities at large distances [16]. fixed to a testing body, e.g. in the direction orthogonal to the planes
of the electrodes. In other sensors the change in capacitance is
driven by the variation of the pressure acting on one deformable
plate so that a capacitance measurements gives information on the
2. Materials and methods
force/pressure acting onto a body. So, the principle of operation of
those devices is very simple and if one restricts the examination to
2.1. Switched capacitor CDC
the ideal case in which the electric field is confined in the volume
between the two plates, it is also expressed in an elementary form.
Many modern CDC are based on the technology of the S
D
In fact, in these conditions, the value of the capacitance is
ADC. A particular implementation of the balancing process typical
approximated by the ratio of the area to the plate separation, ac-
of the S
D converters, named switched capacitor modulation, is
cording to the well-known formula:
the basis for using that technology for the capacitance measure-
ment [17]. A reference voltage Vref is applied periodically to one A
plate of an unknown capacitor Cx and the charge induced on the C0 ¼ ε0 (2)
d
conductor itself or on the other plate of the capacitor is summed,
using an op-amp integrator, with the charge present on a second In a real capacitor the “fringe effect” occurs due to the extension
known capacitor excited with the opposite voltage
Vref . The of the field lines outside of the volume between the plates. Its
feedback mechanism tends to balance the charge on the two ca- magnitude depends on the aspect ratio k which, for parallel square
pacitors, increasing or decreasing the charge on the sampled plate plates
pffiffiffigeometry, is just the ratio of the plate separation to the side
by reversing the sign of Vref . The ratio of charging to the total cycles L ¼ A of the square:
is used, in accordance of a built-in calibration procedure, to trans-
d
late the ADC scheme into that of a CDC. Among these instruments k¼
based on the principle of S
D conversion technique, the series of L
low-cost monolithic devices launched in 2005 by Analog Device Increasing the aspect ratio generally results in a value of the
became very popular in the last few years. In that family the capacitance larger than the ideal case and, for the parallel square
AD7475 model is designed to be used with a floating capacitor plate geometry, the correction is lower than a few per cent when k
while AD7747 modelmeasures a capacitor with a grounded leg. We is smaller than  0.01.
show that the different internal circuitry of the two devices makes Let us consider the quantity effectively measured with the
it possible to measure separately the capacitance coefficients of a Analog Device CDCs when the two electrodes of a capacitor are
capacitor made by two conductors of any shape. connected to the input channels of each CDC and a square wave
excitation signal Vref is applied to one of either conductors. In the
AD7745 model the charge induced on the other armature of the
2.2. Capacitance coefficients capacitor is compared with that of an internal reference capacitor
and the converted value is returned. In this condition, the measured
It is well known that in a two-conductors system, the charge Q1 charge is Q2 and the CDC reads the capacitance coefficient C12 .
on conductor 1 depends linearly on the potential V1 and V2 applied, Differently, using the AD7747, the excitation signal is applied to the
respectively, to the conductor 1 itself and to the conductor 2 same electrode which is connected to the comparator, while the
through: other plate is grounded, and the measure of Q1 is used to derive the
value of the capacitance coefficient C11 .
Q1 ¼ C11 V1 þ C12 V2
3. Measurements
where the coefficient Cij are called capacitance coefficients. A similar
relation holds for the charge Q2 on conductor 2: In this section we present four experiments, which clearly
highlight the operating mode of the two CDCs. The comparison
Q2 ¼ C21 V1 þ C22 V2 between measured and expected results confirms the correctness
The quantity to which we refer commonly as the capacitance C of and the accuracy of the principle of measurement described in the
a two conductors system applies when the charge of the two previous section.
conductors is opposite: Q1 ¼ Q and Q2 ¼
Q . Then, the relative
capacitance C is the ratio between the charge Q and the difference of 3.1. Coefficients of capacitance of parallel square plates
potentials V1
V2 . From the two last equations, it is easy to
calculate the relation of C with the capacitance coefficients: The parallel-plates capacitor is composed by a couple of square
28 E. Cataldo et al. / Journal of Electrostatics 87 (2017) 26e31

plates, realized in copper with a side L ¼ 25 mm and a thickness limit d/0. By symmetry, the two bodies stay equipotential. A finite
t ¼ 0.3 mm. Electrical connections are made with coaxial cables a electric field in the near limit would involve a discontinuity in the
few tens of centimetres long with the central conductor plunged charge surface densities, but the electric field between the two very
into a small connector soldered on a corner of each plate. One of the close surfaces at equal V is infinitesimal, so that both the charges
two plates is fixed vertically while the other, parallel to the first one, and the potential do not diverge at the moment of the contact
is mounted on a translational stage having both fine and coarse (d ¼ 0) at which the capacitors becomes a single equipotential
micrometer movement, spanning many centimetres with tenths of conductor. Then we have that the limit
millimetres accuracy. In the first experiment we fix the distance so
that the aspect ratio is k ¼ 0.08, and the fringe effects are not lim C11 þ C12 ¼ a
d/0
negligible.
When we use the AD7745, one plate is connected to the input is finite and the divergence of the two coefficients of capacitance
CINðþÞ linked to the internal comparator and the other to the pin are equal and opposite. Denoting with S the divergent part of C12
EXCA carrying the reference waveform, the measure returned by we have:
the device was Cm ¼ 2925 fF. If the measure is interpreted as that of
the relative capacitance given by eq. (2), neglecting the fringe effect, C12 ¼
S þ O ðdÞ C11 ¼ S þ a þ O ðdÞ ¼
C12 þ a þ O ðdÞ
the expected value should be C0 ¼ 2763fF. At first sight, the
measured value is in partial agreement with the expected one, but and the relative capacitance C of the two conductors is for this
one must take in account at least the fringe effect. The numerical special case of identical bodies:
treatment of the case of two bidimensional square plates is well-
known in literature. For example, in Ref. [10] the numerical re- C11
C12
sults are presented as a multiplying factor g greater than one which
C¼ ¼ C11 þ a ¼ C12 þ a
2
depends on the aspect ratio. For the considered value of the pa-
For very close plates [9], the diverging term overcomes a, which
rameters, g  1:25 and the corrected value of the capacitance is
can be neglected in the near limit where
C ¼ gC0 ¼ 3453 fF, in severe disagreement with the measure. A
similar discrepancy is observed also using AD7747. We put one of C ¼ C11 ¼ C12
the plate at ground and the other in the CINðþÞ. With these elec-
trical connections the measured value is Cm ¼ 3904 fF a value not This explains why the readout of the CDC for a monolithic
compatible again with the expected one. capacitor having a gap very narrow coincides with the nominal
In order to solve the problem let us refer to the discussion at the value of the relative capacitance in both mode of operation.
end of section 2. In the first experiment, i.e. using the AD7745 and
with the plates floating with respect to ground, we are measuring 3.3. Capacitance coefficients as a function of distance
the absolute value of the interelectrode capacitance coefficient C12 ,
which value is actually negative. On the other hand, with AD7747, The experiments reported in the two previous subsections
the measured coefficient is C11 . The coefficients of mutual induc- strengthen the statement that AD7745 measures the absolute value
tion are equal by definition C12 ¼ C21 and the symmetry of the of the interelectrode capacitance C12 and AD7747 the self-
plates gives C22 ¼ C11 . With these conditions the relative capaci- capacitance C11 . This is not a limitation when one of the two in-
tance C of eq. (1) simplifies to: struments is applied to a sensor where the two electrodes are very
close. In that case the two coefficients tend to the same diverging
C11
C12 value, independently from the shapes of the electrodes. On the
C¼
2 other hand, the different value returned by the CDC when the
Substituting the measured values of the two capacitance co- interelectrode distance grows must be taken into account either if
efficients and taking care to change the sign of the value of C12 one an accurate evaluation of the capacitance is needed or the sensi-
finds C ¼ 3414 fF in good agreement with the expected value. This tivity of the sensor must be maximized.
result, easily reproducible, gives the first evidence of the principle In this subsection we describe a more systematic test reporting
of operation alleged in the previous section. the measurements of the capacitance coefficients of the parallel
plates as a function of their distance. Incidentally, it is worth
noticing that we are not aware of similar experimental data in
3.2. Capacitance of a monolithic capacitor literature.
By using the same plates described above, we collect data in the
Both models of CDC can be used to measure a commercial interval of interelectrode distance between 1 mm and many cm.
capacitor such as a multilayer ceramic chip (MLCC) or a monolithic The minimum distance is limited by the full scale value of both the
disk ceramic capacitor of few pF of capacitance. The two devices converters. In the first few tens of millimetres the read value was
return the same value within the declared accuracy, in close recorded at intervals of one tenths of mm and the steps grow up to
agreement with the nominal capacitance. In fact, those capacitors one mm in the far region. Particular care is taken in removing any
are realized in a quite small packages where the electrodes are very external conductor from the measured system, consisting in the
close together. Let us consider, for the sake of simplicity, two equal CDC, the two cables and the plates. The entire system is held up by
conductors having therefore C11 ¼ C22 and by definition C21 ¼ C12 plastic holders, maximizing the distance between the apparatus
placed at a tiny distance d. In order to ascertain the behaviour of the and all the other bodies, comprising those of the experimenters.
system when d/0, let us assume that the two plates are placed at But, even if the two plates are very far from any other major source
the same potential V. As the conductors are identical one finds that of electric influence, there is an inevitable contribution from
each plate brings the same charge; connection wires and the chip mount itself.
Particularly, as the measurement of C12 consists in sampling the
Q ¼ C11 V þ C12 V
stationary charge on the plate connected to the input channel of the
Let us now detach the plates from the potential sources so that CDC caused by the charging cycles on the other plate, the influence
the charges can not change and reduce the distance toward the of the cables can be reduced by maintaining the excitation wire as
 E. Cataldo et al. / Journal of Electrostatics 87 (2017) 26e31 29

far as possible from the sampled side. Nevertheless, the two cables thickness (0.3 mm) of our plates, which is important for having
end to the very close chip pins, so a discrepancy between the accurate numerical results at very small distances. Even if a high
measured coefficient and that of the pair of isolated conductors is accuracy is time-consuming, it has been possible to make a precise
expected. Induction of the exciting wire on the sampled part of the quantitative comparison between the measured quantities and
circuit is reduced by using coaxial cables with grounded shields. those calculated with boundary element methods [19]. Details on
With that care the length of the wires has a minor influence on the our numerical algorithm will be discussed elsewhere.
readout. Fig. 1 shows a macroscopic discrepancy between measured and
In the measure of C11 the excited plate is also the sampled one calculated values: raw data tend at large distances to a value r
while the other side of the system is at ground potential. Then, a different from the calculated one. Roughly speaking, a typical
particular attention should be paid to the excitation cable: it decrease of few hundreds of fF is observed with respect to the
changes the sensed quantity, being a part of the conductor 1. The numerical calculation when the cable bringing the excitation signal
self-capacitance of an unshielded wire grows up with the length so is few tens of cm long and the value varies with the position of the
that it is possible to use only very short unipolar cables. If the plate connections. Quantitatively, the value of C11 should converge to the
1 is placed distant from the CDC a coaxial cable must be used intrinsic capacitance C1 , which for the plates considered is 1044 fF.
mandatorily together with a compensating mechanism, included in On the contrary, the asymptotic value extrapolated from the data is
the design of the AD7747 CDC. Nevertheless, we observe that if the only 720 fF. A similar offset is observed also measuring C12 although
coaxial cable is too long a systematic negative offset is introduced. of minor magnitude and with different sign. Those systematic be-
The magnitude of the shift must be taken into account if an accurate haviours are attributable to the fact that the two conductors are
measure is needed. In our experiment we estimated that the composed by the plates and the connections, mainly the wires and
discrepancy due to this effect is less than  50 fF. their attachments. In fact the magnitude of the discrepancy de-
It is well known that the two coefficients C11 and C12 , which pends mainly the length of the cables, their shielding and the
tend to equalize in the limit d/0 (see the last subsection), have a relative position of the attachments. As the scope of our actual
different asymptotic behaviour when the distance grows [18]. In experiments is to clarify the primary physical quantities measured
particular, C12 tends to vanish and C11 goes to a constant value. by the CDCs, we leave the problem of the absolute calibration of the
Physically, these limits are expected. In fact, C12 measures the capacimeter to a more refined version of the apparatus. It is worth
electrostatic induction of the conductor 1 on the conductor 2 and to note that the principal effects of these multiple causes of devi-
that mutual influence must vanish at large distances. On the other ation for each coefficient can be taken into account correcting the
hand, the coefficient C11 is the parameter describing the self- data with a uniform offset.
capacitance, i.e. the charge induced on the plate 1 when it is In Fig. 2 the same data reported in the Fig. 1 are shifted uni-
brought at a potential V1 , possibly in the presence of the second formly in the vertical direction, minimizing the square residuals
conductor. When the latter is very far C11 tends to the intrinsic between the corrected and the calculated data. For a number of
capacitance C1 of the plate 1 which is simply the ratio of the charge different geometries of the wires and of the connections the offset
Q1 to the potential V1 of the body very far from any other conductor. is of few hundreds of fF for the C11 and some tens of fF for C12. The
The two behaviours are shown in Fig. 1. right panels of the figure, showing the residual of the fitting pro-
In the absence of an analytic expressions valid in the entire cedure show a very good agreement, as in the entire range exam-
range of investigated distances, it is only possible a comparison ined the discrepancies do not exceed two per cent.
between computed and experimental data. Actually, several suit- These results show that the CDCs are suitable to be used to
able numerical algorithms apply to the calculation of the co- measure with good accuracy the coefficients of capacitance of a pair
efficients of capacitance, particularly for 2D plates. We have of plates. Of course, the same experimental procedure applies also
implemented a numerical procedure taking into account also the to more complex conductor bodies. In the hypothesis that the offset

Fig. 1. Coefficients of capacitance C11 (o) and
C12 (þ) of two square copper plates
with side L ¼ 25 mm, as a function of distance. The raw data show clearly that the two Fig. 2. Left panel: same graphs as in figure, applying to the raw data uniform cor-
coefficients converge to the same value at very small gaps and that the asymptotic rections consisting in a shift of the measured values of the capacitances. Here,
trends at large distances are compatible with a constant value for C11 and to zero for DC11 ¼ 370fF DC12 ¼
137fF. Right upper panel: Residual for C11 calculated as the
C12 . The continuous lines are the numerical calculation of the two coefficients, using a difference between the unbiased and the calculated data. Right lower panel: Residual
method of moments and taking into account the thickness of the plates. for the data of C12 .
30 E. Cataldo et al. / Journal of Electrostatics 87 (2017) 26e31

4. Discussion

Usually, the technical presentations of the operation of com-

mercial CDCs and their specification documents emphasize the
precision of the data conversion, the accuracy of the measurement
and the sampling frequency. Indeed, as these devices are largely
applied for the readout of capacitive sensors the primary goal is to
obtain a high sensitivity to the variations of the charges on the
capacitor plates resulting from a change of the relative position, the
approach of a third body, the deposition of impurities which vary
the dielectric constant of the space occupied by the electric field
around the capacitor and so on. Little attention and little infor-
mation is given on the physical quantities characterizing the system
and on their influence on the returned data. A common architecture
of the CDC produced by the primary manufacturers is based on the
measure of the charge accumulated on one of the sensor plates
during repeated cycles of charge and discharge. The measure con-
sists in the comparison of the charge to be sampled on the external
sensor with that filed on an embedded calibrated capacitor. A
Fig. 3. The data at distances d > 2L are fitted with the two models C11 ¼ a þ b=d2 and
sigma-delta ADC converts the measures in digital words. The

C12 ¼ h þ k=d of eq. (3). In the figure the curve fitted and the data, showing a very capacitive sensor is generally realized with two conductors and the
good agreement. As b≡C13 and k ¼ C12 we evaluate C1 in the two fitting procedures measure can consist in exciting and sampling the charge on the
finding, in the fit of C12 , C1 ¼ 1117fF and in that of C11 C1 ¼ 1124fF. It is noticeable same conductor, the other being connected, typically, to ground, as
that, in the same interval of distances, the two coefficients undergo a much different
in the CDC AD7747, manufactured by Analog Device. A second
change in accordance with the different power-dependence on 1=d in eq. (3).
scheme consists in the excitation of one of the two plates and in the
sampling of the other (e.g. AD7745). The two techniques read a
in the converted datum is fixed by the geometry of the connections different capacitance coefficient. Moreover, our physical descrip-
it seems possible to use the well-known and reproducible parallel tion of the principle of operation shed light on the effects of the
plate geometry to calibrate the instruments and, after that connect connections between the main conductor bodies and the IC input
it to a generic pair of conductor, obtaining with a little effort an ports and of the bodies surrounding the apparatus. With respect to
experimental accurate knowledge of the complete electrostatic the first issue, it is clear that the cable and the connection to the
properties of the system. excited plate contributes in both mode of operation at the same
level with the plate. But the magnitude of the effect is different for
the measures of C11 and of C12 . In the first case, the wire is excited
3.4. Modelling the far limit with the same waveform of the plate and it is charged at the same
time. As the intrinsic capacity of the wire contributes with hun-
When the conductors are very near or very far it is possible to dreds of fF already for length of few centimetres it is impractical to
write analytical models reproducing the asymptotic trends. At the use long unipolar cables to connect the CDC measuring C11 . As a
far limit the coefficients of capacitance of two generic conductor matter of fact, the AD7747 CDC is equipped with a second exciting
can be expressed by the first terms of a power expansion in 1=d signal which can be applied to the shield of a coaxial cable which
involving as parameters the intrinsic capacitances of the two con- internal conductor is connected to the conductor 1 when the
ductors and their polarizabilities and quadrupole moments [16]. In measure is made. An explanation of the role of this connection can
the simplified geometry of two equal parallel plates C11 ¼ C22 ¼ C1 be given on a physical basis. Let us now consider that only the
and taking the leading term in 1=d one finds: shielded cable is connected and the core and the shield are brought
! to the same potential so that the electric field between the wire and
C12 the shield is zero. This condition is realized when the inner wire has
C11 ¼ C1 1 þ
d2 no charge. This condition is perturbed when the plate is connected
and the electrical induction between the coaxial cable and the plate
is responsible for the negative offset observed. A major difficulty to
C2
C12 ¼
1 (3) keep this effect under control is that one expects that the magni-
d tude of the phenomenon could depend, even for the same cable, on
As the two asymptotic trends are expressed, for identical sides the shape of the body connected to it. The other cable is not directly
of the capacitor, in term of the same parameter C1 we fit, with the excited by the signal generated by the CDC circuit and it is main-
two simple functions of eq. (3), the raw data for distances larger tained at constant potential in both modes of operation. Moreover
than the side of the plates. The graphical result of the fit is reported it is further away from the excited one and from the first body so
in Fig. 3, where a very good agreement is shown. that the induction effects are weaker, mainly when the two prin-
It is worth to note that only the constant term in the model cipal bodies under test are far from each other and the two cable are
function is influenced by the offset. The other two parameters placed maximizing their distance for the major part of their path.
represent C13 and C12 for identical plates and wires. The value of C1 Nevertheless, an influence of the second cable on the readout of C11
determined in this way are consistent with each other and with the, and of C12 is unavoidable and could be reduced using also in the
offset-dependent, horizontal asymptote of C11 . This check confirms side 2 a coaxial cable with the shield grounded. We observe that
the procedure of data analysis and also is a clear test of the formulas when measuring C12 the inner conductor of the coaxial cable is not
in the far limit for the coefficient of capacitances which at the at the same potential of the grounded shield so that it is charged
leading level are valid for generic bodies. Moreover the evaluation with plate 2 by induction of the side 1 and an increase of the
of C1 is also independent from the absolute estimate of the offset. sampled charge is expected, as the observed positive offset signals.
 E. Cataldo et al. / Journal of Electrostatics 87 (2017) 26e31 31

5. Conclusions the apparatus and the data analysis in order to get deeper infor-
mation on less symmetric systems of two conductors.
The experiments reported in the paper demonstrate the phys-
ical mechanism underlying the capacitance-to-digital converters
operating through the sampling of the charges accumulated on one References
of a pair of conductor. We have shown that the returned data when
two parallel plate are connected to the CDC does not coincide with [1] Y. Qi, H. Wang, Y. Lin, M. Zhao, Measuring system for micro force based on
what we intend usually as capacitance C. Really, the value returned electrostatic theory, in: Electronic Measurement & Instruments, 2009.
ICEMI’09. 9th International Conference on, IEEE, 2009, pp. 524e527.
is the coefficient of self-capacitance C11 or the induction capaci- [2] R. Bekhti, V. Duchaine, P. Cardou, A novel force sensing method based on
tance C12 . We have explained also that, due to the convergence of stress imaging analysis, IEEE Sensors J. 16 (no. 7) (2016) 1926e1936.
C11 and C12 to the same value when the distance between the plates [3] O. Akhmetov, S. Pilgaev, Y.V. Fedorenko, D. Dneprovskii, A liquid microbaro-
graph, Instrum. Exp. Tech. 53 (no. 5) (2010) 750e754.
tends to vanish both the mode of operation of the CDCs return, in [4] K. Ghose, S. Dandavino, H. Meyer, B. Chamot, and J.-F. Labrecque-Piedboeuf,
this condition, the value of relative capacitance C and this is the Gravity gradient earth sensor experiment on rexus 11,” in 21st ESA Sympo-
result when a small monolithic capacitor is tested. The results show sium on European Rocket and Balloon Programmes, no. EPFL-CONF-187717,
2013.
that the instruments, basically applied for the ability to sense a [5] S. Satyanarayana, D.T. McCormick, A. Majumdar, Parylene micro membrane
change in sensor capacitance, are able to perform with a little effort capacitive sensor array for chemical and biological sensing, Sensors Actuators
accurate measure of the basic properties of an electrostatic system. B Chem. 115 (no. 1) (2006) 494e502.
[6] J.C. Maxwell, A Treatise on Electricity and Magnetism, vol. 1, Clarendon Press,
We checked with a very simple apparatus the dependence of the
1892.
coefficients of capacitance of two square plates up to a value of the [7] G. Kirchhoff, Zur theorie des condensators, Monatsber. Akad. Wiss. Berl.
distance several times the side of the plates and we made a strict (1877) 144e162.
comparison with the calculated numerical results finding a very [8] R. Love, The electrostatic field of two equal circular co-axial conducting disks,
Q. J. Mech. Appl. Math. 2 (no. 4) (1949) 428e451.
good agreement both on the global range of the recorded data, [9] G. Paffuti, E. Cataldo, A. Di Lieto, F. Maccarrone, Circular plate capacitor with
when the effect of the connections, consisting primarily in an different discs, in: Proc. R. Soc. A, vol. 472, 2016, p. 20160574.
overall offset of the measured data, are taken into account properly. [10] H. Nishiyama, M. Nakamura, Form and capacitance of parallel-plate capaci-
tors, Components, Packag. Manuf. Technol. Part A, IEEE Trans. 17 (no. 3)
This quantitative experimental determination of the fringe effects (1994) 477e484.
of a parallel plate capacitor up to a very high value of the aspect [11] M. Norgren, B. Jonsson, The capacitance of the circular parallel capacitor ob-
ratio d=L is not common in the technical literature. On the other tained by solving the love integral equation using an analytic expansion of the
kernel, Prog. Electromagn. Res. PIER 97 (2009) 357e372.
hand, it should be easy to investigate with the same apparatus [12] H. Wintle, S. Kurylowicz, Edge corrections for strip and disc capacitors, IEEE
other systems such as a couple of circular disks or two rectangular Trans. Instrum. Meas. 34 (no. 1) (1985) 41e47.
plates in the strip configuration. Moreover a very good agreement is [13] G. Paffuti, Numerical and analytical results for the two disks capacitor prob-
lem, Proc. R. Soc. A 473 (2017) 20160792.
found with the asymptotic behaviour of the coefficient of capaci- [14] L.K. Baxter, Capacitive Sensors. Design and Applications, IEEE Press, 1997.
tance. The fitting of the well known minimal model for the ex- [15] Analog devices datasheets. http://www.analog,com/zh/analog-to-digital-
pected trend give accurate value of the intrinsic capacitance of the converters/capacitance-to-digital-converters/ad7745/products/product.html.
[16] F. Maccarrone, G. Paffuti, Capacitance and potential coefficients at large dis-
plates used, and as the model resulting from eq. (3) is general the
tances, J. Electrost. 83 (2016) 22e27.
procedure should apply with minor modifications to conductors of [17] J. O'Dowd, A. Callanan, G. Banarie, E. Company-Bosch, Capacitive sensor
generic shapes and relative disposition. The fact that sub-leading interfacing using sigma-delta techniques, in: Sensors, 2005 IEEE, IEEE, 2005, p.
terms in the higher power of 1=d involve, as demonstrated in 4.
[18] L. Landau, E. Lifshitz, Electrodynamics of Continuous Media, Pergamon, 1984.
Ref. [16], other interesting parameters such as the polarizability of [19] D.K. Reitan, Accurate determination of the capacitance of rectangular parallel-
the conducting bodies makes interesting the possibility to refine plate capacitors, J. Appl. Phys. 30 (no. 2) (1959) 172e176.