Manufacturing of soft capacitive strain sensor based on dielectric elastomer material for an elastic element of a jaw coupling

In this work, we present a procedure for manufacturing a soft capacitive strain sensor in the form of a multi-layer electrical capacitor for further integration into the elastic gear rim of a jaw coupling. The dielectric elastomer sensor (DES) is based on an elastomeric substrate with alternating layers of conductive carbon black based ink as electrode layers and elastomeric film as a dielectric and electrical insulation. A variety of thin multi-layer sensors were produced to evaluate the manufacturing process. Additionally, using an LCR meter and a tensile test machine, the equivalent electrical capacitance (C) at the two sensor contacts and the applied force are measured, respectively. It is shown that C varies depending on the amount of sample strain caused by the applied force. By testing two versions of DES, a maximum change in capacitance of ΔC = 1.55 pF was achieved. The obtained characteristics show that the presented manufacturing process for the DES can be used as a soft strain sensor to measure the strain caused by the force applied to the elastic element between the jaw couplings.


Introduction
Machines can generally work under human control or independently.They consist of many different elements, some of which work perfectly for years, while others wear out much more quickly.They all need a regular technical inspection, for which both destructive [1] and non-destructive methods [2,3] are used.However, if the machine contained elements with integrated sensors, this would allow the operator to get all the necessary information about the operation of a separate section of the machine.Thus, it is possible to monitor the operating parameters and the condition of those parts with limited service life.
In recent years, attention has shifted from ordinary machine elements to sensor-integrated machine elements [4,5].Examples of the first approaches of sensor-integrated machine element are a sensor-integrated gear, published by Bonaiti et al [6] and prototypes of smart couplings proposed by Schork et al [7].Kirchner et al presented an extensive overview of various machine elements with integrated sensors from bolts, couplings and bearings to gears and gear shafts [8].In mechanical engineering, couplings are essential connecting elements, as they are one of the central elements used in most transmissions [9].Their primary function is to transfer power between two shafts.Jaw couplings are rubber-elastic couplings of medium elasticity, which compensate for shaft misalignment and absorb starting shocks [9,10].To implement a sensor into the coupling, common sensors are not suitable since they were initially designed for external placement and as they are comparable in size to an elastic element or larger.A promising option is to use dielectric elastomer sensors (DESs) [11].Since they are soft and deformable, they can also be adapted to the required tasks.Dielectric elastomers (DE) are multi-functional, elastic and highly deformable electromechanical structures commonly used as actuators [12][13][14], sensors [15][16][17], for power generation [18,19] or for signal processing [20].There exist various concepts for the use of dielectric elastomers as capacitive sensors [21][22][23] or even as self-sensing actuators [24].If a compliant elastomer structure is subjected to a mechanical load, it will deform significantly.When a dielectric elastomer sensor is built into the structure, it will also deform in the same order of magnitude.This change in the physical dimensions of the DES leads to a corresponding change in the electrical parameters, such as capacitance or resistance.capacitance is preferred to resistance to measure the deformation since it is reported to be more reliable, see Xu et al [25].For this reason, many authors employ dielectric elastomer sensors as capacitive sensors, cf [15,21,22] On the one hand, the elastic coupling is an ideal interface for torque measurement with soft strain sensors due to their high deformations under load.Furthermore, on the other hand, the elastomeric element offers the possibility to use mechano-sensitive sensors to record the process-relevant data 'in situ' with a high degree of reliability.The sensor-integrated coupling could, therefore, also be used temporarily for the commissioning and tuning of drive systems.In the present research, we introduce the concept and design of a compliant sensor for integrated measurement, which minimizes the required equipment.
DESs can be manufactured using a wide variety of materials and methods.For the large deformation requirements of the DES, many classes of dielectric materials have been investigated [11,26].A material such as ceramics, which is used in multi-layer solid capacitors [27], is not suitable in this case because it breaks at large strains.The electrodes of the DES should have a good stretching ability and flexibility to endure large deformations (compress and stretch).Due to the high electrical resistance of the sensor elements (for example, conductive ink electrodes with a diameter of d = 3 mm in the unloaded state have a resistance range of R electrodes = 10 ...50 kΩ), the electrodes are energy efficient.The research aims to find the optimal process to manufacture and characterize a soft strain sensor.The novelty of our approach lies in designing the combined structure of a multi-layer capacitor and a dielectric elastomer technology to create a soft capacitive strain sensor suitable for integration into the elastic gear rim of a jaw coupling.
The task of integrating the sensor into the jaw coupling involves many steps.In order to reduce the complexity, we present a substitute system which focuses on the sensor part of the sensor-integrated machine element [5].In order to achieve this, the complex geometry of a tooth of the gear rim, made of thermoplastic polyurethane (TPU), is simplified as a cylindrical specimen.The specimen is made of the same material to resemble the expected properties.The dielectric elastomer sensor will be embedded in a borehole in the TPU cylinder completing the sensor system.Two strain amplifiers will also be embedded to hold the DES inside the TPU.This structure serves as a test bench for the rapid development of the sensor system, where different setups and manufacturing methods can be quickly tested.
The present paper is structured as follows.First, in section 2 we will present the experimental setup as well as the design and manufacturing of the sensors.Then, we will present the obtained results for a series of loadingunloading tests in section 3 and discuss them in section 3.4.Finally, we will give a short conclusion and an outlook in section 4.

Materials and methods
The present section describes the progress in finding the optimal dielectric elastomer sensor design using two different versions of a dielectric elastomer sensor, (i) the DES 1 and (ii) DES 2.

Thermoplastic polyurethane samples manufacturing
The TPU specimen is punched out of an elastic gear rim with a Shore-A hardness of 92.The punched TPU specimen has the shape of a cylinder with a diameter of d TPU specimen = 16 mm and a height of h TPU specimen = 9.3 mm.The punch jig is displayed in figure 1 (A, C).One of the teeth of the gear rim after extraction is displayed in figure 1 (B).The TPU specimens are extracted from a gear rim with an outer diameter of d 120 mm gear rim = . The hole for the sensor implementation is drilled with a spiral drill using a drill size of d drill = 5.95 mm.
An extracted TPU specimen is displayed in figure 2 (A).The strain amplifiers were made of PMMA and are displayed in figure 2 (B).PMMA was chosen because of its higher stiffness than the surrounding TPU, its good processability and its isolation properties.The dimensions of the amplifiers are given in table 1. Due to the soft nature of TPU, the drilled hole is undersized.This hole is used for the integration of the amplifiers, which are pressed into the borehole.Both the amplifiers and the TPU specimen have a small groove in which enameled copper wires with a diameter of d 0.3 mm wire = are embedded.To ensure contact between the DES and the amplifier, a copper pad with a height of h copper pad = 1 μm is attached to one of the surfaces of the amplifier.The wires are soldered to a copper pad with a height of h copper pad = 1 μm and placed in the grooves.The strain amplifiers are used for multiple purposes: (i) they amplify the strain on the DES.Based on the higher stiffness of the PMMA, compared to the surrounding material, the compliant DES and TPU deforms more.This leads to an increase of strain on the DES, while the absolute deformation of the system remains unchanged.(ii) a place for the location of the contacts on the surface of the strain amplifiers, which is directed towards the sensor and (iii) to provide space for the copper wire, to which the LCR meter is subsequently connected through the groove on the side of the amplifier.Figure 2 depicts TPU specimen, amplifiers, DES and their position.Amplifier 2 plays the role of support for the DES, in turn, amplifier 1 receives and transmits the load from the tensile test machine to the DES, cf figure 2 (D).

Experimental setup
Load tests are conducted on a ZwickRoell Z005 tensile test machine with a 5 kN force cell.The tensile test machine has a measurement frequency of f = 10 Hz.The TPU specimen is placed between metal parallel plates  and subjected to a compressive force, cf figure 3.In order to reduce friction, a Polytetrafluoroethylene (PTFE) foil with a thickness of h PTFE = 50 μm separates the TPU specimen from the tool, cf [28].Experiments are conducted in a displacement-controlled manner.The traversing speed is set to v = 0.1 mm s −1 .Various maximal deformations are chosen, as explained in section 3.During the experiments, it is found that the preload speed should be equal to the traversing speed.Thus, the start of the force sensor data recording coincides with the moment of the force application.Otherwise, a hysteresis cut off at the beginning would be obtained.The preload force is set to F preload = 0.1 N. Force and displacement of the TPU specimen are recorded in the test software TestExpert from ZwickRoell and exported in an Microsoft Excel file for further processing.
Capacitance of the DES is measured using an IM3523 LCR meter from HIOKI.The DES is clamped between amplifier 1 and amplifier 2 and connected to the amplifier's copper pad via a physical contact, cf figure 3. Copper wires are soldered to the amplifier's copper pads and to a common board, and the LCR meter is connected to the soldered contacts using double-ended tester cables with crocodile clips.The board is held by clips on a stand.The data is recorded in an Excel file using the application software "LCR meter Sample Application" by HIOKI.The recorded data is processed in MATLAB and Python for further presentation.An "Interval Measurement" mode in the application is used for every measurement.This mode is used to make measurements at a userspecified time interval.The LCR meter is also set in R p − C p mode, i.e. the capacitance is measured in parallel.The excitation frequency of the LCR meter is set to f = 1 kHz and the operating voltage is set to U = 1 V for all measurements according to [29].The experimental setup is depicted in figure 3.

Dielectric elastomer sensor structure
The typical structure of the DES is formed by sandwiching an elastomer material between two compliant conductive electrodes, which can produce a large strain response and a high electromechanical efficiency from an electric field.The deformation of the dielectric elastomer produces a change in capacitance of the sensor [11].
In our case, we use the deformation of the sensor caused by the tensile test machine to obtain a corresponding change in capacitance of the sensor.The DES will have to respond to actual changes in the operation of the machine.Therefore, an LCR meter will be used first to monitor the change in capacitance of the sensor during  the experiments and the tensile test machine will record the strain and the applied stretch.The smaller the measured quantity, the larger the relative quantified error [30].Consequently, by increasing the measured value, the relative error decreases, e.g. a capacitance range of ΔC = 50 pF can be measured with higher accuracy than a smaller change, e.g. of ΔC = 0.1 pF.That is why, in order to improve the accuracy of capacitance change measurement and the sensitivity of the DES, it is necessary to increase the total capacitance of the sensor.There are several ways to increase the capacitance of DESs.Based on the definition of capacitance for a circular parallel plate capacitor the capacitance C increases with (i) increasing relative permittivity ε r , (ii) increasing diameter of the plates d, (iii) decreasing distance between the plates h or with (iv) the connection of multiple capacitors.However, the maximal diameter is often fixed through limitations of the allocated space.The relative permittivity ε r of a material can change though different stimuli such as prestretch or temperature [31,32].However, this is not feasible for our case since the permittivity decreases through a prestretch, and the change in temperature during the use of the jaw coupling can not be controlled.Therefore, a structure of circular parallel plate capacitor consisting of 2 and 3 plates was chosen as the main concept of the DES.The DES has a cylindrical shape, and deformation occurs along the vertical axis of the cylinder.To determine the optimal sensor design and manufacturing methodology, two approaches were used: (i) connecting single fabricated two-electrode structures and (ii) initially fabricating a three-electrode structure in a single manufacturing process.
To embed the sensor into the TPU specimen, a cylindrical area with a diameter of d space = 6 mm and a maximum height of h space = 9 mm, equal to the thickness of one TPU tooth, is provided.Due to the fact that the DES will deform during use, some free space is needed.Therefore, the diameter of the sensor d DES = 5 mm has to be smaller than the diameter of the open space.Due to the available space, a cylindrical form of the DES was chosen as the main concept.

Dielectric elastomer sensor 1
The first version of the DES (DES 1) is a sensor consisting of three single two-electrode structures placed on top of each other.Each single structure is a soft circular capacitor made of an elastomeric material.Such a single structure consists of the following layers, starting from bottom to top: a silicone substrate, first electrode layer, first dielectric layer made of two-component silicone and carbon black powder, second electrode layer and second dielectric layer consists of silicone, cf figure 4. As mentioned above, each electrode is circular in shape, resulting in a circular parallel plate capacitor.Electrode layers are made of conductive ink (CI).The first dielectric layer serves as a dielectric in a regular capacitor.The second dielectric layer serves as (i) a protective layer of the second electrode and as (ii) an adhesive layer for bonding single structure to each other.According to the DES 1 design, three single structures were connected with CI to increase the overall capacitance of the sensor.The CI and dielectric layer recipe is described in section 2.4.1.

Dielectric elastomer sensor 2
The second version of the DES (DES 2) has a three-electrode structure.Due to the structure, there are two soft circular capacitors with one mutual plate, so these capacitors are connected in parallel.The second electrode layer serves as the mutual plate.To avoid manufacturing errors when applying the dielectric layer, instead of liquid two-component silicone, we used an adhesive silicone layer and an elastomeric film.
Such a three-electrode structure consists of a silicone substrate, on which layers are applied in the following order, starting from bottom to top: CI-dielectric layer-CI -dielectric layer-CI -dielectric layer, where each dielectric layer consists of silicone adhesive and elastomeric film, cf figure 4. The height of the DES depends on the number of layers and is limited by the dimensions of the tooth.

Dielectric elastomer sensor manufacturing
The following section will describe the production process and the layers for the DES 1 and DES 2.

Dielectric elastomer sensor 1
The materials used for the manufacturing of the single two-electrode structure are: (i) Ecoflex 00-30, (ii) CB KETJEN BLACK EC-600 JD, (iii) an Elastosil film from Wacker with a thickness of h base = 200 μm and (iv) conductive ink (CI).CI is used as the electrode layer for the sensor.

Creating the dielectric layer
According to Lacasse et al [33], the best sensitivity for silicone/carbon black composite is achieved when a weight fraction of carbon black (CB) is in the range of 0 wt% to 1.5 wt%.It was confirmed by Mahmoudinezhad et al [29] that the best sensitivity of the sensor based on Ecoflex/CB is achieved at 0.4 wt% CB.Ecoflex was mixed in the ratio between part A and part B 1:1 by weight.Depending on the mass of Ecoflex, 0.4 wt% CB was weighed, and then pre-ground with the addition of metal balls with a diameter of d metal ball = 8 mm in a planetary mixer Thinky ARE-250 for 3 minutes at a speed of n = 2000 rpm.This is necessary in order to grind large carbon agglomerates to obtain a fine CB powder and then a homogeneous mixture.The components are mixed in the following sequence: parts A and B of Ecoflex are alternately added to the crushed CB with metal balls.It should be noted that it is necessary to add silicone to CB.In this case, all the fine particles of the powder will remain in the container and mix with the silicone.Otherwise, pouring carbon black into silicone, part of the powder will remain on the walls of the container.Further, all components are mixed in a planetary mixer for 3 minutes at the same parameters.The Ecoflex/CB mixture is then stored in the refrigerator.

Creating the electrode layer
There are different ways to create a conductive ink to form a carbon-based electrode layer [23,34], especially a flexible and stretchable electrode [35].CI for the electrodes in the form of a polymer matrix with dispersed conductive particles was made of (i) Ecoflex 00-20, (ii) carbon black (CB) KETJEN BLACK EC-600 JD as conductive particles and (iii) an ozone-safe volatile methylsiloxane (VMS) fluid Dowsil OS-20 as a solvent [36].Due to the lower viscosity, Ecoflex 00-20 was chosen as the polymer matrix.Ecoflex was first weighed in a ratio of part A and part B 1:1 by weight.According to the weight of Ecoflex, 5.88 wt% of CB was weighed and preground, as described above.The ratio of OS-20 to the amount of Ecoflex is 2:1.The components are mixed in the following sequence: Ecoflex parts A and B are alternately added to the pre-ground CB, and the solution is mixed once with the same settings for the planetary mixer, as described above.Then, the solvent is added to the mixture, and everything is mixed once again with the same parameters.The metal balls can be removed from the container using tweezers.The conductive ink is then stored in the refrigerator.

Creating the sensor: layering process
The layering of the single two-electrode structure is depicted in figure 5 and conducted in the following order.First, a base layer of Elastosil film with a thickness of h base = 200 μm was chosen.After that, a PET mask with a thickness of h mask = 50 μm was positioned on the substrate.The mask was cut on a Trotec Speedy 100 CO 2 Laser Cutter.The cutouts have a diameter of d electrode cutouts = 3 mm and a connection-pin for the electrode with a length of l pin = 1 mm and a width of w pin = 0.5 mm, cf figure 6.
The pins perform several functions: (i) they serve as a connection point for the LCR meter and (ii) enable the connection of the two-electrode structures to each other.Then the electrode layer was applied to the base layer using doctor blade technology with a blade speed of v = 4 mm s −1 leaving no space between the blade and the mask.Subsequently, the mask was removed, and the component was placed on an aluminium board for transportation to the oven and cured at ϑ = 100 °C for 15 minutes.After curing, it was cooled at room temperature.Since there is 65 % of a solvent in the ink, the calculated thickness of the electrode layer after the curing is h electrodes = 0.35 • h mask = 17.3 μm.The next step is the application of the first dielectric layer using Ecoflex 00-30 with 0.4 wt% CB.To apply the dielectric layer on the electrode layer, there is no need to use an additional adhesive layer since the two-component silicone itself has sufficient adhesiveness.Again a mask with a thickness of h mask = 50 μm is positioned.The mask cutouts for the dielectric layer have a diameter of d dielectric layer−cutouts = 4 mm.The area above the pin was not covered with a dielectric layer in order to maintain access to the bottom plate of the capacitor.The dielectric layer is applied similarly to the electrode layer using doctor blade technology.Again the mask was removed, and the component was cured using the same parameters.To calculate the height of the dielectric layer above the electrode, it is necessary to subtract the height of the electrode layer from the height of the mask used: h dielectric layer = h mask − h electrodes = 32.7 μm, cf table 1.
Further, the procedure for applying a layer of electrodes was repeated once again using a PET mask.The mask was rotated 180 degrees to have pins on opposite sides of the capacitor.Thus, as a result, a soft circular capacitor is obtained, consisting of silicone substrate, two circular electrodes and one dielectric layer between these electrodes.

Dielectric elastomer sensor 2
The materials used for the manufacturing of the DES 2 with a three-electrode structure are: (i) Elastosil film with a thickness of h base = 200 μm and (ii) h Elastosil = 50 μm, (iii) SilGel 612 from Wacker and (iv) CI.

Creating the sensor: layering process
The layering of the DES 2 is displayed in figure 7 and conducted in the following order.The first step of applying the layer of electrodes is identical to manufacturing the DES 1.The electrode layer was applied to the surface of an Elastosil film with a thickness of h base = 200 μm using doctor blade technology and then cured as indicated in section 2.4.1.The further procedure is distinguished by the next stage of electrode lamination using a SilGel adhesive layer.The SilGel was mixed in a ratio between part A and part B of 1:1 by weight.The components were mixed using a planetary mixer for 3 minutes at a speed of n = 2000 rpm.After that, a PET mask with a thickness of h mask = 50 μm was positioned on the edges of the substrate.Then the adhesive layer was applied to the open surface of the Elastosil base using doctor blade technology with a blade speed of v = 10 mm s −1 leaving no space between blade and mask.Thus, the entire surface of Elastosil with applied electrodes was covered with a SilGel adhesive layer.Subsequently, the mask was removed, and the Elastosil dielectric layer with a thickness of h Elastosil = 50 μm was applied to the Elastosil base using a roll laminator.The laminated structure was then placed on an aluminium board for transportation to the oven and cured for 15 minutes at ϑ = 100 °C.Commercial Elastosil is sold in a roll and is applied on a PET film for ease of use.After curing, the laminated structure was cooled at room temperature, and the PET film was separated from the Elastosil dielectric layer.This way, the first layer of electrodes is completely isolated from the next layer of electrodes.To calculate the height of the dielectric layer between the electrodes for DES 2, it is necessary to subtract the height of the electrode layer from the height of the mask used and add the height of the Elastosil film: Further, the procedure for applying a layer of electrodes and their lamination was repeated twice.Thus, as a result, a DES 2 structure of two circular capacitors with one mutual plate is obtained, consisting of silicone substrate, three circular electrodes separated by two dielectric layers and a dielectric protective layer at the top.It should be noted that the same PET mask was used for the first and third layers of the electrode.Thus, the pins of the first and third electrode layers are directed to the left.To apply the second layer of electrodes, an electrode mask was used, rotated 180 degrees, so the pin of the second layer of electrode was directed to the right, cf figure 4.
The gluing of each layer was chosen as the best way to connect the layers.SilGel adhesive layer was used to ensure good adhesion of the Elastosil base layer with the electrodes and the subsequent dielectric layer.

Assembly of the substitue-system
To prepare the DES 1 for testing, three single two-electrode structures were placed on top of each other and glued with a small amount of Ecoflex 00-30 for better fixation.Then, these three single structures were connected in parallel using the CI.In order to fix the contacts, they were covered with Ecoflex and cured.Then the resulting sensor with a height of h DES 1 = 1.3 mm was placed in the TPU specimen with a height of h TPU specimen = 9.5 mm and a diameter of d TPU specimen = 16.8 mm and connected to the strain amplifiers at the top and bottom.Two amplifiers with heights of h amplifier1 = 2.19 mm and h amplifier2 = 6.025 mm and copper contacts on them were used.
Regarding the DES 2 version, after the production phase of the three-electrode structure was completed, the individual sensors were cut using a laser cutter.Furthermore, for each individual sensor, the protruding side contacts were cut off with a sharp scalpel, dipped in CI, and brought to the upper and lower surfaces of the sensor for connection to amplifiers, as it is depicted in figure 4.
In addition, by connecting the DES to one of the amplifiers, another capacitor is formed between the upper electrode layer and the amplifier contact, cf figure 4. When connecting the DES to the opposite amplifier, no capacitor is formed between the bottom plate of the sensor and the top surface of the amplifier.This is due to the fact that the amplifier's top surface and the bottom plate are electrically connected in series and act as the bottom contact of the entire sensor.The dimensions of all components and layers of the DES 1 and DES 2 are given in table 1.

Results
The two versions, DES 1 and DES 2, were tested using a tensile machine to apply the load, record the applied displacement and a resulting force, and using an LCR meter to record capacitance changes.

Determination of the relative permittivity
For pure Ecoflex 00-30 and Ecoflex 00-30 with a mass concentration of carbon black of 0.4 wt%, the dielectric constant value of ε r = 2.85 and ε r = 2.95, respectively, was experimentally determined.For this, a series of flat capacitors with a thin layer of dielectric with a thickness of h = 100 μm and h = 50 μm, respectively, were manufactured.In comparison, silicone rubber has a dielectric constant of ε r = 2.9... 4 [37], as shown in table 2. However, at the same time, this silicone sealant has a higher hardness, cf [38], which may negatively affect the sensor's sensitivity.

Evaluation of the initial capacitance
In order to evaluate the manufacturing process of the DES, the capacitance of the single two-electrode structures DES 1 and three-electrode structures DES 2 were measured.For DES 1, N DES 1 = 15 samples and for DES 2, N DES 2 = 16 samples were tested.The results are shown in figure 8.For these measurements, the sensors only comprise the dielectrics and electrodes, without the strain amplifiers, shown in figure 3.
The median capacitance of the single two-electrode structures DES 1 is C = 3.84 pF, cf figure 8.In comparison, using equation (1) for a circular parallel plate capacitor gives a capacitance of C = 5.64 pF, showing a difference between calculations and experiments.This difference could be explained by manufacturing errors caused (i) during the application of a layer of Ecoflex with carbon black and (ii) some inaccuracy in the position of the circular electrodes relative to each other so that the area of the plates may be less than expected.For example, with an increase in the height of the dielectric layer by 20 %, i.e. by Δh dielectric layer = 6.5 μm and a decrease in the diameter of the electrodes by 10 %, i.e. by Δd electrodes = 0.3 mm then such a capacitor will have a capacitance of C = 3.81 pF.
For the three-electrode structures DES 2, the median capacitance was measured as C = 6.21 pF, cf figure 8.The analytically calculated capacitance yields C = 4.22 pF using the desired dimensions for the DES 2 from table 1.The higher measured capacitance of the DES 2 is probably due to the laminating process, when a liquid adhesive layer of SilGel becomes thinner after two layers of Elastosil are pressed together (see section 2.4.2).

Evaluation of the capacitance due to mechanical loading
As already mentioned, the aim of the sensor is to measure a strain, or a torque, respectively, when it is included in a coupling via a change in the capacitance.Testing DES 1, where three single two-electrode structures were connected in parallel, cf figure 4, a maximum capacitance of C = 6.58 pF was obtained.This capacitance value corresponds to a deformation of the TPU specimen of Δu = 1.2 mm.Considering that the initial capacitance is C 0 = 5.03 pF, a capacitance change of ΔC = 1.55 pF was obtained.According to this, the relative capacitance change is C 30.8 % D = .The relationship between the applied stretch λ and the measured capacitance C for DES 1 is shown in figure 9. To clarify the noisiness of the results, it is necessary to pay attention to the frequency of the capacitance measurements.For testing both versions of the DES, the measurement interval in 'LCR meter Sample Application' was set to the lowest value t = 0 s, that means fast data recording, according to the instructions for the device [42].This measurement interval corresponds to approximately 43...45 measurements per second, i.e. the measurement frequency is f ≈ 44 Hz.That is why there are more measured points and, therefore, more noise, as can be found in figures 9 and 10.The data was smoothed by employing a moving average.Due to the high noise of the capacitance measurement, a kernel size of 100 was used.It can be seen from figure 9 that different deformations Δu = {0.9,1.0, 1.1, 1.2} mm correspond to different capacitance maxima C = {6.28,6.36, 6.53, 6.58} pF.This means that DES 1 has a different capacitance at different stretches, which characterizes the strain sensor.Due to the fact that the frequency of the stretch measurement is lower than that of the capacitance (see section 2.2) the stretch was interpolated for the times of the capacitance measurement.
However, we could demonstrate that the presented principle is working.The hysteresis, which can be seen in figures 9 and 10 is a result of the viscoelastic behavior of the TPU and the dielectric used in the DES.This is a drawback as the information about the current capacitance value is not enough to accurately determine the respective stretch since the relationship between stretch and capacitance is ambiguous.Calibration can help to determine the stretch corresponding to a specific capacitance value.
It should be noted that in DES 2, the maximum capacitance is close to the one from the DES 1.However, the capacitance is achieved by a lower DES height, cf table 1.For one three-electrode structure of DES 2, the height is  Comparing the versions of DES 1 and DES 2 in terms of their manufacturing process, DES 2 has one complete manufacturing process, while DES 1 has an additional stage of stacking three single structures.This additional step in the process of creating DES 1 increases the probability of production errors, since this step is carried out manually.Considering the capacitance of the sensors at a certain maximum deformation, it can be seen that the difference in capacitance between adjacent points of maximum deformation for DES 2 is greater than for DES 1, that means the higher sensitivity of the sensor, cf figure 11 (a).As can be seen from figures 9 and 10, the capacitance values at the beginning and end of the compression test are different; this is due to the viscoelastic properties of the dielectric elastomer.This difference is depicted in the figure 11 (b), where for DES 1 the difference between the capacitance at the beginning and end of the test increases with increasing stretch, demonstrating some instability in the behavior of the sensor.In turn, DES 2 has a greater instability at the maximal deformations of Δu = 0.8 mm and 0.9 mm, although with increasing stretch the sensor becomes more stable.
In order to evaluate the two versions of the DES, we compare a capacitive Gauge Factor (GF) of each sensor at the different stretch value, cf figure 11 (c).The GF is determined as GF = (ΔC/C 0 )/(1 − λ).Considering the general stretching area for two sensors, it can be seen that at first the GF is at the same level: at the maximal deformation of Δu = 0.9 mm the sensors have gauge factor of GF DES1 ≈ GF DES2 ≈ 9.09, for the deformation of Δu = 1.0 mm the gauge factor is GF DES1 = 9.08 and GF DES2 = 8.94.With increasing stretching, the GF of the DES 2 becomes much larger than with the same stretch values for DES 1: at the maximal deformation of Δu = 1.1 mm the gauge factor is GF DES1 = 9.50 and GF DES2 = 11.61.
The DES 2 structure shows a higher GF at higher stretch values, more stable results and higher sensitivity compared to the DES 1 structure.

Discussions
This study presented two variants of the soft strain sensor fabrication process.Both variations have a design acceptable for the integration in an elastic gear rim of a jaw coupling.To test the DES, a minimal model of the gear rim in the form of extracted cylinders was used.Two versions of the DES based on the technology of dielectric elastomers and conductive carbon black based ink were successfully fabricated and tested.Due to the low stiffness of the elastomer material, the DES were easily deformable.Two versions of the sensor were compared with each other, and the progress in the manufacturing process was shown when moving from the two-electrode structure (DES 1) to the three-electrode structure (DES 2).In addition, the strain sensor exhibits good linearity.This can be seen from the graphs of capacitance versus applied stretch, cf figures 9, 10 and 11 (a).The presence of hysteresis is due to the viscoelasticity of both the TPU material and the dielectric elastomer.Most importantly, the strain sensor has been successfully placed into the TPU specimen of the gear rim, and measurements have been conducted for this configuration.

Conclusion and outlook
In this work, we described two manufacturing processes for a capacitive strain dielectric elastomer sensor (DES), the DES 1 and the DES 2. Both variants were successfully manufactured and then tested, using an LCR meter and a tensile test machine.It was determined that the DES 2 structure is more stable and promising for further development than the DES 1 structure.Considering the practical application, the strain sensor still needs further improvement.Both the electrical stability of the strain sensor and the repeatability of the measurement results have to be further improved to meet the needs of machine elements used in industry.The improvement of electrical stability can be achieved by developing a layering process, conducting sensor lifetime studies, and testing reproducibility after many cycles.In turn, the repeatability of the measurements can be achieved through the further elaboration of the experimental setup and synchronization of both the LCR meter and the ZwickRoell machine times.Increasing one of the main parameters of the strain sensor-sensitivity, and hence the capacitance of the sensor can be achieved by developing a multi-layer design of DES.Furthermore, in order to realize the practical application of the strain sensor, the signal acquisition, processing and transmission must be highly integrated.In addition to improving the above parameters, the next steps will be the creation of a numerical model for predicting both the mechanical and electrical behavior of the DES under the load.

Figure 1 .
Figure 1.Punch jig for extraction of TPU specimen, (A) punch jig, (B) gear rim tooth with extracted specimen, (C) drawing of the punch tool (the dimensions are given in mm.).

Figure 2 .
Figure 2. Manufactured parts of the soft strain sensor, (A) extracted from an elastic gear rim TPU specimen, (B) DES 1 on amplifier with cooper contact pad, (C) DES 1 with amplifier inside TPU specimen, (D) CAD model of the DES 1 and (E) DES 2 with strain amplifiers implemented into TPU specimen.

Figure 3 .
Figure 3. Experimental setup for integrated measuring system with substitute system.

Figure 4 .
Figure 4. Conceptual design and CAD model of the DES layering setup.Single two-electrode structure: one soft capacitor C 1 on an Elastosil base; DES 1: three single two-electrode structures stacked on top of each other to form three soft capacitors C 1 , C 2 , C 3 connected in parallel; DES 2: three-electrode structure consisting of two soft capacitors C 1 , C 2 connected in parallel.Please note that the dimensions are given in μm and the amplifiers are only connected to the outermost layer.

Figure 6 .
Figure 6.Three single two-electrode structures of the DES 1.

Figure 7 .
Figure 7. Manufacturing process of DES 2. The layering and lamination processes are repeated three times to create a three-electrode structure.CI-conductive ink.

Figure 8 .
Figure 8. Box plot comparison for initial capacitance measurements of DES 1 single structures and DES 2. N DES 1 = 15 specimens for the DES 1 and N DES 2 = 16 for the DES 2 were tested.

Figure 9 .
Figure 9. Measured capacitance C over the stretch λ of the TPU specimen for the DES 1 for various maximum deformation values.The data were smoothed via a rolling mean with a kernel size of 100.

Figure 10 .
Figure 10.Measured capacitance C over the stretch λ of the TPU specimen for the DES 2 for various maximum deformation values.The data were smoothed via a rolling mean with a kernel size of 40.

Figure 11 .
Figure 11.Comparison of DES 1 (blue) and DES 2 (red) design: (a) Capacitance maxima over the stretch of the TPU specimen; (b) difference between the initial and final capacitance over the stretch of the TPU specimen; (c) GF over the stretch of the TPU specimen.

Table 1 .
Main parameters of DES 1 and DES 2.

Table 2 .
Values of the relative dielectric permittivity ε r for different types of silicone