Design and fabrication of Temperature Sensing Fabric

Abstract

A temperature sensing fabric is described, along with the manufacturing techniques required to produce the fabric on a computerised flat-bed knitting machine. Knitted sensing fabrics with copper, nickel and tungsten wire elements have been produced with resistances ranging from 3 to 130 Ω. The most successful samples have been created using textile-wrapped, enamelled wire and not only the textile character of the sensing element was enhanced, but also its tensile strength. A mathematical relationship has been derived between the temperature and resistance of the knitted sensors and this can be used to optimise its dimensions to achieve a targeted reference resistance. The temperature-resistance curves demonstrate a linear trend with a coefficient of determination in the range of 0.99–0.999 and can be integrated into garments to monitor skin temperatures.

Keywords

Temperature sensor resistance temperature detector textile sensors wire inlaying into knitted structure electronic flat-bed knitting machine

Introduction

In recent years, intensive research has been carried out in the area of electronic textiles, particularly in respect of the incorporation of sensing and actuating functions into textile products. Being close to the body, textiles can offer a comfortable, flexible platform for the embedding of sensing functions. The direct consequence of applying this technology is the development of wearable monitoring systems intended for continuous usage in monitoring the human body, for vital signs over extended periods of time. One of the key components of this system is the sensor which gathers data from the wearer and passes it for further processing. This information may be used to help creating a general health picture for various medical purposes [1]. Human body temperature is one of the four vital signs used in conjunction with heart rate, blood pressure and respiratory rate for medical assessment of the state of health. It is an important indicator of the physical condition of the human body and relates to comfort, performance, and heat or cold stresses [2].

Within the domain of wearable monitoring systems, most of the research has been focused in the development of textile-based electro-cardiography (ECG) and respiration sensors. To measure human body temperature, the majority of researchers have relied on commercially available temperature sensors (mainly thermistor and temperature measuring minute integrated circuits), which are usually attached externally to the garment [3 –6]. Previous research also demonstrates that the fabrication of sensing fabric for temperature measurement has been little explored in a few individual studies [7 –11].

De Rossi et al. [9] developed sensing fabric by coating a polypyrrole on a Lycra fabric; Locher et al. [10,11] developed a woven sensing fabric by incorporating insulated copper wires into a woven fabric and devising a routing technique; Ziegler and Frydrysiak [7] developed various types of textile thermocouple by utilising conductive fabric, non-conductive fabric and conductive yarn/wire and Sibinski et al. [12] developed a ‘thermistor yarn’ by depositing a thermo-resistive paste, on a polyvinylidene fluoride yarn. All of them are preliminary studies; their products were manufactured by cumbersome manual processes and are limited in their depth of characterisation and performance of the sensor in either the laboratory or in the real-life environment.

This study presents an improved method of embedding temperature sensing functionality into a textile substrate by using industrial scale textile process, and preliminary characterisation of its laboratory performance. The developed sensing fabric can be deployed for continuous measurement of the human body temperature in non-clinical settings, e.g. sports, military, general healthcare, firefighting situations and studies related to biorhythms and assessment of thermal strain in extreme environments.

Conceptualisation of temperature sensing fabric

In order to imbue a textile substrate with temperature sensing functionality, a detailed review has been conducted to identify what are currently the most commonly used temperature sensing technologies, i.e. the resistance temperature detector (RTD), thermistor and thermocouple.

Thermistors are fragile semiconductor devices, usually encapsulated in glass, whose resistance varies as a function of temperature [13]. Thermocouples are made of two dissimilar conducting wires which have been welded together and require relatively complex ‘conditioning’ electronics. Neither thermistors nor thermocouples are as accurate as the RTD [14]. The RTD measures temperature by correlating the resistance of a fine coiled wire (wrapped around an insulator) which changes with the temperature. The wrapped wire is further protected from the environment by placing it inside a sheathed probe [15].

The review has demonstrated that the RTD is probably the most suitable type of temperature sensor, whose design can be integrated into a textile substrate and could ultimately be fabricated on an industrial scale by a fabric forming machine with minimal manual input. The associated electronic circuit (to measure resistance) is also relatively simple. Thermistors and thermocouples, in contrast, have to be sewn into the fabric and do not become part of the cloth.

Technically, textile-based RTD sensors can be developed by integrating a piece of metal wire (as a sensing element) into a textile fabric. However, various factors have to be taken into account. For example, an appropriate alloy for the sensing element and the most appropriate material for the base fabric have to be selected. The most suitable textile structure for embedding the sensing element must be found. It is also necessary to identify which textile process is most appropriate for undertaking the manufacture.

Embedding of metallic wires into textile structures for sensing purposes has been documented by several researchers. Various fabric forming techniques such as knitting, weaving and embroidery have been exploited for these purposes. Various forms of stainless steel wire have also been used to develop textile-based ECG electrodes [16 –18] and pressure sensors [19]. An elastic belt with embedded copper wire has been developed for inductance plethysmography-based respiration sensing [17]. A number of studies have been conducted in respect of the development and characterisation of routing methods for power and signal transmission in woven electronic textiles, by introducing networks of metallic wires at specified distances along the warp and weft [19 –23].

Both knitted and woven fabrics have been used for the development of sensors and as sensing platforms for wearable health monitoring systems [24, 25]. However, in those sensor applications where close contact to the body is required, e.g. respiration or ECG sensors, the knitted structures are usually preferred over woven structures due to their ability to conform to body shape. Moreover, the breathability of knitted structures makes them comfortable to wear. The knitting technology offers significant advantages over other fabric forming systems in respect of the development of textile sensors [26], e.g. simple preparatory processing, the processability of a wide selection of conductive yarns, one-step garment manufacturing and exact positioning of sensing patches in the garment. Considering the above-mentioned benefits and the required application (human body skin temperature measurement), the knitting technology has been found to be an obvious choice to produce a temperature sensing fabric (TSF) by embedding a sensing wire into a knitted structure.

TSF has therefore been developed by embedding fine metallic wire into the structure of textile material on a computerised industrial flat-bed knitting machine. The operational principle of the TSF is based on the inherent propensity of metal wire to respond to changes in temperature with corresponding variation in its electrical resistance. A modified relationship between the resistance of the TSF and temperature has also been established. Various types of the metal wire identified as potential sensing elements for the TSF were compared in conjunction with the selection criteria. The TSF samples were tested under laboratory conditions within the temperature range of 20–50℃, to observe the relationship between the temperature and the resistance of the sensing fabric. The resistance–temperature (R–T) relationship demonstrates a linear trend with a coefficient of determination (r²-value) of over 99.9%.

Materials and methods

Figure 1. presents the concept of the TSF showing a metal wire inlaid in the middle of a rib knitted structure. By supplying a constant current to the inlaid wire and measuring the resulting voltage drop across it, its resistance can be calculated, and the temperature can be determined.

Figure 1.

TSF showing inlaid sensing wire in a rib knitted structure.

The temperature–resistance relationship

The standard equation showing the relationship between the resistance of metal wire and its length, diameter and resistivity is stated in equation (1) [14]:

R_{ref} = \frac{4 l ρ_{ref}}{π d^{2}}

(1)

where

R_{ref}

and

ρ_{ref}

are the resistance (wire function) and resistivity (metal function), respectively, at a reference temperature.

R_{ref}

is also known as the nominal resistance. While, the length (l) and diameter (d) are the dimensions of the wire. A mathematical relationship between the length of the metal wire (sensing element) and the dimensions of the sensing fabric has been derived as:

l = (W_{s} L_{s} D_{i}) + L_{s}

(2)

where L_s and W_s are the length and width of the TSF and D_i represents the inlay density of the sensing wire. From equation (2), it can be seen that, a 4 × 4 cm² patch of the TSF with an inlay density of 6 cm⁻¹ might easily contain a 1 m length of sensing wire. Combining equations (1) and (2) gives a modified relationship for the TSF resistance in terms of its dimensional parameters:

R_{ref} = \frac{4 ρ_{ref} [(L_{s} W_{s} D_{c}) + L_{s}]}{π d^{2}}

(3)

Equation (3) could be used to optimise the dimensions of a TSF patch for a required reference resistance, once the resistivity and diameter of the wire are known. The higher the resistivity and the finer the sensing wire, the smaller may be the sensing area within which a target reference resistance may be achieved. It is thus important to lay-in the sensing element as densely as possible to achieve the maximum ratio of reference resistance to sensor area. High reference resistance is desirable in order to achieve high sensitivity while a small sensor area is desirable for a fast response time. The effect of temperature (T) on the resistance of a metal (R_T) conductor can be expressed as [27]:

R_{T} = R_{ref} (1 + α_{ref} (T - T_{ref}))

(4)

where

α_{ref}

is the temperature coefficient of resistivity for a given temperature range starting from the reference temperature (

T_{ref})

. After combining equations (3) and (4), the temperature-resistance relationship of a TSF can be described in the form:

R_{T} = \frac{4 ρ [(L_{s} W_{s} D_{c}) + L_{s}]}{π d^{2} (1 + α (T - T_{ref}))}

(5)

The sensing element

In RTD sensors, a very fine and relatively short sensing element (platinum wire of less than 25 µm in diameter) is employed, which changes its resistance upon change in temperature. Although platinum is the most commonly used sensing element in RTDs, copper- and nickel-based RTDs are also available in the market. Various metal wires could be used as the sensing elements of TSFs in the form of bare wire, enamelled wire or textile wrapped (braided) wire. The desirable properties for the sensing element of a TSF include the following items:

Nominal resistance: Nominal resistance is one of the most important baseline specifications of the RTD. The RTDs are available from 10 Ω to 10,000 Ω of nominal resistance but 100 Ω is the most common figure. The nominal resistance depend upon the resistivity of the sensing element and the dimensions of the TSF as stated in equation (3). High nominal resistance is desirable and could be achieved by choosing a long, fine wire of high resistivity.

Sensitivity: Sensitivity is directly related to the temperature coefficient of resistivity and nominal resistance. High sensitivity of the sensor is desirable for creating a detector with high accuracy, resolution and minimal electronic circuitry.

Response time: Temperature sensors with fast response times are desirable especially during the measurement of the temperature in a dynamic environment. The response time of a sensor may be minimised by keeping the thermal mass of the sensing element as low as possible. The thermal mass is calculated as the product of the mass of the sensing element and the specific heat capacity of its material, and is typically measured in units of J/℃.

Self-heating: Sensing elements may be prone to a self-heating (Joule heating) error if excitation currents of 10 mA or above are used. High thermal conductivity of the sensing element is desirable to minimise this error by dissipating the heat rapidly.

Repeatability: Continuous and smooth dependence of resistance on temperature without any hysteresis is important for high precision, accuracy and interchangeability.

Mechanical properties: Sensing wire should have sufficient ductility for smooth bending along the edges and high tensile strength to meet the tension requirements of the inlaying process performed during machine knitting, without any breakages.

Availability and reproducibility: Sensing wire should be low in price and readily available in fine diameters, so the sensors are readily and consistently reproducible.

Table 1 presents the important properties of sensing elements i.e. platinum, nickel, tungsten and copper. For ease of understanding, these properties are also compared relatively in Figure 2. Platinum exhibits the highest resistivity of the range of metals. Therefore, a sensing element made of platinum would require a reduced length in order to achieve any desired nominal resistance; but copper would require a much longer length to achieve the same nominal resistance. High resistivity, along with a high-temperature coefficient of resistivity, also contributes to high sensitivity, as exhibited by both the nickel and platinum elements. Because of their low thermal mass, the response time of tungsten and platinum would be faster than nickel and copper. The tungsten is the strongest sensing element, in terms of its tensile properties. This would be beneficial for processing on a knitting machine. The platinum is regarded as the best sensing element for RTDs due to its high sensitivity, low thermal mass, its availability in its purest form and its stability over a wide range of temperatures; however, it is relatively expensive.

Figure 2.

Comparative chart of important relative properties of TSF sensing elements.

Table 1.

Important properties of TSF sensing elements [28 –31].

Property	Unit	Copper	Tungsten	Nickel	Platinum
Temperature resistance coefficient	(1/ ℃)	0.0038	0.0044	0.0069	0.00392
Electrical resistivity	Ωm	1.67e−08	5.6e−08	7.5e−08	10.6e−08
Thermal mass^a	J/K	0.39	0.13	0.54	0.13
Thermal conductivity	(W/mK)	386	180	86	72
Sensitivity^b	(mΩ/ ℃)	9	32	60	53
Tensile strength	MPa	260	2500	500	130
Price	$/oz	0.26	1.6	0.64	1835

TSF: temperature sensing fabric.

g of sensing element.

1 m long sensing element with diameter of 100 µm.

In terms of price, copper is the least expensive of the viable sensing elements and is readily available in range of diameters in bare and insulated form. The tungsten and nickel elements may not be readily available in their purest form which may result in large variations in the values of resistivity and the temperature coefficient of resistivity when purchased from different sources. In comparison to copper, it is relatively difficult to find tungsten and nickel sensing elements of the required purity and diameter so they might have to be manufactured by order.

In respect of the properties presented in Figure 2, the sensing elements i.e. copper, nickel and tungsten possess some advantages over each other. It was thus decided to make use of all these metals as sensing elements for the TSF. Platinum was completely ruled out for purchase due to its high cost. These sensing elements, in the form of bare and insulated wire, were purchased from different sources in various diameters, as shown in Table 2. The temperature-resistance curves of all the selected metals are fairly linear over the required temperature range i.e. 20–50℃ and can be represented by equation (4).

Table 2.

Sensing elements of temperature sensing fabric.

Wire Tag	Form	Metal	Diameter (µm)	Source
N100	Bare	Nickel	100	Scientific Wire
N90	Bare	Nickel 270^a	90	Alloy Wire
W80	Bare	Tungsten	80	Good Fellow
W50	Bare	Tungsten	50	Good Fellow
NC125	Bare	27% Nickel coated copper	125	Temco
NC127	Bare	4% Nickel coated copper	127	Temco
C150	Bare	Copper	150	Scientific Wire
EC150	Enamelled^b	Copper	150 (170 including enamel)	Scientific Wire
BEC150	Braided^c and Enamelled^b	Copper	150 (220 including enamel and braid)	Scientific Wire
BEN61	Braided^d and Enamelled	Nickel	61 (120 including enamel and braid)	Scientific Wire

A high purity grade of nickel.

Solderable polyurethane 156℃ enamel.

Double artificial silk covered.

Double nylon covered.

TSF structure and manufacturing

The main requirement of the TSF structure includes the smooth embedding of wire in the middle of the textile layers so that it neither contacts itself, nor protrudes from the main body or the edges of the fabric. It was also desirable to pack a high density of wire in order to achieve high nominal resistance in a small-sized TSF. The most suitable way for smooth embedding of wire into knitted structure is by the inlaying technique which involves trapping the inlay between the needle and the sinker loops on a double needle-bed machine. This technique offers the possibility of introducing those yarns which are difficult to process in the knitting machine due to their physical properties.

Various types of knitted structure can be used as a platform for laying-in metal wire. Double layer knitted structures are more suitable than single layer structures due to their better cover for the metal wire from wear and tear. Secondly, the metal wire would not be visible and does not affect the aesthetic properties of the fabric. Moreover, the wire will assume a straight configuration which avoids any distortion or flow towards an area of the fabric under tension.

In order to meet the requirement, a double layer structure of the TSF was devised using Shima Seiki Knit design software as shown in Figure 3; it comprises knit courses, spacer courses and metal inlays. The samples were fabricated on a 10 gauge Shima Seiki flat-bed knitting machine (as shown in Figure 4) by employing two feeders. First feeder was responsible for knitting spacer and knit courses, by taking five ends of highly textured 167 decitex polyester yarn, as drawn in blue lines. The second feeder was used for inlaying metal wire along with three ends of the same polyester, as shown by the magenta lines. It was difficult to pack the sensing element in every course of fabric without it shorting to itself. Therefore, a few courses of spacer yarn were introduced to keep the sensing element separated in between courses. After much experimental manipulation of the spacer, knit and inlay densities, the highest element concentration that was successfully achieved was of 46 wire inlays in an 8 × 8 cm² knitted TSF structure (≈3.8 m length of element) as shown in Figure 5. The structure had to be constantly varied to achieve such a high density whilst avoiding the shorting of adjacent wires and minimising the protrusion of wire along the edges. For ease of understanding, the concept of the TSF structure may be represented in a simplified form by reducing the number of knitting and spacer courses as shown in Figure 6.

Figure 3.

Needle notation of temperature sensing fabric designed in Shima Seiki Knit Paint.

Figure 4.

Shima Seiki computerised flat-bed knitting machine.

Figure 5.

TSF sample of 8 × 8 cm² sensing area along with a magnified view of the knit structure.

Figure 6.

A schematic representation of temperature sensing fabric course notation.

The TSF sample comprised 39 wales; 33 for the inlay area and six for the edges. Since it is a double layer structure, 39 needles on the front needle bed and 39 needles on the lower needle bed were utilised. The non-sensing part of the TSF comprised full cardigan stitch (top and bottom) to minimise relaxation of the fabric. Samples developed with various types and thicknesses of metal wire (as mentioned in Table 2) provided a useful range of nominal resistances from 3 to 130 Ω.

The wires purchased from the various manufacturers were on small spools due to the shortness of the length. It was not prudent to use these small spools directly on the knitting machine because of two problems. Firstly, excessive wire breakage occurred due to interaction with the flanges of the spool while withdrawing the wire strand. Secondly, it was difficult to maintain constant tension in the wire. In order to address these issues, the metal wire was wound manually on a hank winder from small spools as shown in Figure 7.

Figure 7.

Wire transfer from small spool to hank winder.

Care was exercised while winding on the hank winder in order to maintain constant tension in the wire. The hank winder was fastened to a stand and placed at the top of the machine.

Smooth withdrawal and constant tension in the feed material is a prime requirement for inlaying. In the initial phase of manufacturing, frequent breakages of the wire were experienced. Occasionally, extensive tension in the wire and interference between the needles and the slack wire caused issues. Reduced tension in the wire was also responsible for excessive bending at the fabric edges which resulted in wire protruding from the edges and sometimes from the main body of the knitted sample.

In order to rectify this issue a number of remedial actions were implemented:

The wire feed was moved to the side from the top of the machine; this reduced the angular path, resulting in more uniform tension and smoother feeding.

A negative let off mechanism was developed at the hank winder, comprising tension plates and a spring system, to minimise discontinuities in the feed and to ensure even tension in the wire before, after and during inlaying.

Extra polyester yarn was used as an inlay along with the metal wire to increase the strength of the inlay.

A special feeder (see Figure 8) was used to inlay the metal wire, as the ordinary feeder did not meet the fundamental requirements of straight laying-in and smooth bending of wire at the edges of the TSF. This special feeder was tubular and was set at a lower height than the machine bed so that during the knitting cycle, the needles go over the wire without causing breakages. This inlay feeder must not be used for knitting as it would crash into the needles because of its lower adjusted height. The stitch presser was not used due to the risk of it interfering with the inlay wire.

The special feeder responsible for metal inlaying was programmed to run at an extremely slow speed of 0.05 ms¹.

Figure 8.

Special feeder for wire inlay.

Results and discussion

Processability of metallic wire during TSF manufacture

Table 3 presents an overview of the processability of metallic wire during the TSF manufacture, in terms of wire breakage, manual handling and the quality of inlaying. ‘Wire breakages means the average number (or percentage) of machine stoppages because of wire breakage during knitting. Manual handling of the wire includes all manual processes prior to and during manufacture and preparation of samples for testing. The quality of inlaying was considered good if the metal wire was smoothly embedded exactly in the middle of the double layer structure without any projection from the surface or at the edges.

Table 3.

Metallic wire processability during temperature sensing fabric manufacture.

Metallic Wire	Wire breakage	Quality of inlaying	Manual handling	Process-ability (overall)
BEC150	0%	Good	Easy	Good
EC150	0%	Good	Easy	Good
C150	0%	Good	Easy	Good
NC127	0%	Good	Easy	Good
NC125	0%	Good	Easy	Good
BEN61	10%	Good	Easy	Good
W80	10%	Poor	Medium	Poor
N100	20%	Medium	Medium	Moderate
N90	20%	Medium	Medium	Moderate
W50	50%	Poor	Difficult	Poor

It was observed that the frequent wire breakages were related to excessive tension whilst poor quality of inlaying was related to slackness in the wire. For smooth bending of the wire at the edges, and a straight configuration of inlay, relatively high tensions were found to be preferable; considering that the wire was able to sustain them. That is the reason that coarser wires (over 100 µm) such as BEC150, EC150, C150, NC127 and NC125 showed better overall processability than their finer counterparts.

Among the medium diameter wires, i.e. N100, N90 and W80, nickel wires demonstrated smooth bending at the edges (because of its high ductility) but slightly more breakages (because of its low tensile strength) in comparison with tungsten wire. The tungsten wire is inherently strong and brittle in nature; therefore, inlaying of W80 suffered fewer breakages but displayed the poor behaviour in respect of bending at the edges.

The fine diameter wires such as BEN61 and W50 showed contrasting behaviour. Although the metallic core diameter of the BEN61 wire is 61 µm, due to its increased tensile strength resulting from the addition of the enamel and the braiding layers, the wire showed good processability during manufacture. The 50 µm tungsten wire suffered the highest number of breakages and was found to be relatively difficult to handle and to set into the feeder, as it gathered whenever any breakage occurred. In some of the samples, wire protruded from the surface of the edges, as shown in Figure 9.

Figure 9.

A W50 temperature sensing fabric sample showing Tungsten wire protruding near the edges.

RTDs are commonly specified with their nominal resistance which is usually 100 Ω at 0℃. In order to achieve 100 Ω a very fine and relatively short sensing element (<25 µm in diameter) is employed in the RTDs. Similarly, it was desirable to produce TSF samples with high nominal resistance, which can be achieved by employing very fine metallic wires. However, the results of the TSF manufacturing demonstrated that it would be difficult to process the finer wires in the knitting machine. This problem was solved by replacing a bare sensing element with a braided–enamelled sensing element. The textile wrapping around the wire improved not only the textile character of the sensing element but also its tensile strength. Insulation would help in protecting the sensing element from external influences such as moisture and corrosion.

TSF structure

All TSF samples were made of polyester as a base material with 46 inlays of sensing element. The TSF structure was finalised following numerous development cycles, in order to meet the fundamental requirement of smooth embedding of sensing wire in the middle of fabric, without electrical short circuits. The developed TSF structures may not be the fully optimised one but the structures do meet the basic requirements of the research.

Highly textured polyester yarn was employed as the base material of the TSF. Polyester was preferred over cotton because of its extremely low moisture regain and high tensile strength. The low absorbency and good wicking properties of polyester fabrics assist the transfer of moisture from the skin to the outer surface of a garment where it evaporates. Moreover, the textured characteristics of the yarn increased the fabric bulkiness and provided better cover of the sensing element.

Modified temperature–resistance relationship

A modified temperature–resistance relationship for the TSF in terms of its dimensional parameters could be used to optimise the dimensions of a TSF patch for a required reference resistance. That will facilitate the specification and visualisation of the sensor characteristics. For example, the nominal resistance of an 8 × 8 cm² TSF could be calculated. Similarly, the area of TSF could be predicted that would be required in order to make a 100 Ω sensor. The derived relationship in equation (2) between the length of sensing element and the dimensions of the TSF is only valid for rectangular sensing areas.

Manufacturing variation

The variations among the TSF samples may (manufacturing uncertainty) be represented in terms of their nominal resistance, however, for ease of understanding and to simplify comparison, the nominal resistance was converted into a calculated length of sensing element. The length of the sensing element $(l_{cal})$ in each sample was calculated using equation (1):

l_{cal} = \frac{π d^{2} R_{20}}{4 ρ_{20}}

(6)

where R₂₀_. stands for the experimental reference resistance of TSF at 20℃.

The calculated length of sensing element was also compared with the target length $(l_{tar})$ of sensing element as defined in equation (2):

l_{tar} = (W_{s} L_{s} D_{c}) + L_{s}

(7)

After inserting the 8 × 8 cm² dimensions and 46 inlays into equation (7), the target length of the TSF samples was calculated to be 3.8 m.

Figure 10 presents the uncertainty amongst various sample types in terms of the calculated length of inlaid sensing element. Variations within the sample type are expressed by error bars of standard uncertainty. It is evident from Figure 10 that length variations exist not only amongst the various sample types but also within examples of the same sample type. It can, nonetheless, be seen that the mean of the calculated length is found to be relatively close to the target length. In all the TSF samples, the 95% confidence deviation of the calculated length from its mean was found to be less than ±4 cm; which is not substantial considering the tolerance of the manufacturing process of the TSF samples. However, ±4 cm variations in the length of the sensing element will shift the reference resistance of a particular TSF sample. This means the TSF samples may not be used as interchangeable sensors; or, the calibration equation for one sample may not be applied to another sample. In other words, each individual sample should be calibrated before using it in an application environment.

Figure 10.

TSF manufacturing variations in terms of length of sensing element.

R–T testing

In order to calibrate the TSF samples, a laboratory oven with temperature homogeneity of ±1℃ was employed to create a steady thermal environment at the temperature points of 30℃, 40℃, 50℃ and 60℃. An Agilent multimeter with a four wire resistance measurement setup was used to measure resistance at steady temperature points. The reference resistance (R₂₀) of the samples was measured at a room temperature of 20℃ using the same Agilent multimeter.

Experimental R–T curves of all the TSF samples show a linear relationship. One such relationship belonging to sample N100 is shown in Figure 11. The R–T data from all the samples have been curve-fitted with straight line equations. Variance between the data points and the fitted line has been calculated with a coefficient of determination (r²-value). The r²-values of all the experimental repeats of TSF samples were found to be in the range from 0.99 to 0.999.

Figure 11.

Temperature–resistance relationship of a TSF sample (N100).

Figure 12 presents the resistance-ratio curves of all the samples. The resistance ratio was calculated by dividing the fitted resistance $(R_{T}$ ) at temperature (T) by its reference resistance (R₂₀) at a base temperature of 20℃. Defining the TSF temperature in terms of resistance ratio instead of resistance has several advantages. TSF samples comprising the same kind of sensing element but with different dimensions and inlay densities could be more easily compared in this way by normalising their resistance values with respect to their reference resistance [28].

Figure 12.

Resistance ratio curves of TSF samples in the temperature range of 20–50℃.

All the samples with the same kind of sensing element wire followed the same trend with insignificant sample-to-sample variations as shown in Figure 12. As expected, nickel-based TSF samples exhibited higher values of resistance ratio in comparison with the copper- and tungsten-based TSF samples.

Due to the higher nickel sensitivity, within the category of copper-based TSF samples, the nickel-coated copper-based TSF samples showed marginally higher resistance ratios in comparison with the pure copper-based TSF samples.

Conclusions

This paper presents the concept of TSF (a textile-based RTD) and the manufacturing techniques required to produce it on a standard industrial computerised flat-bed knitting machine. The TSF samples were produced with nominal resistances ranging from 3 to 130 Ω by using copper, nickel and tungsten wires of different diameter. Coarser sensing wires showed better overall processability in comparison to their finer counterparts. However, the coarser wires significantly reduced the overall sensitivity of the TSF. This problem was solved by replacing a bare wire with a braided, enamelled wire. The textile wrapping around the wire improved not only the textile character of the sensing element but also its tensile strength. A modified mathematical relationship between the temperature and resistance of a TSF was derived. This relationship can be used, to design sensing fabric with optimised dimensions, in order to achieve a targeted reference resistance. Manufacturing uncertainty (sample-to-sample variation) was found to be ±4 cm in terms of the length of the sensing element. The R–T curves showed a linear trend with a coefficient of determination in the range of 0.99–0.999. The resistance ratio curves described the insignificant sample-to-sample variations. The TSF can be integrated into garments for continuous measurement of human body temperature in non-clinical settings, e.g. sports, military, general healthcare, firefighting situations and studies related to biorhythms and assessment of thermal strain in extreme environments.

Footnotes

Funding

The authors would like to acknowledge the funding provided by the NED University of Engineering & Technology, Pakistan through the Higher Education Commission of Pakistan, to carry out this study.

References

Funnell R, Koutoukidis G and Lawrence K. Chapter 21 – Vital signs. In: Tabbner's nursing care: theory and practice. Livingstone, Australia: Churchill, 2008, pp.251–274.

Ring

EFJ

. Progress in the measurement of human body temperature. IEEE Eng Med Biol Mag 1998; 17: 19–24.

Curone D, Dudnik G, Loriga G, et al. Smart garments for emergency operators: results of laboratory and field tests. In: 30th Annual international IEEE engineering in medicine and biology society (EMBS) conference. Vancouver, BC, Canada, 2008.

Derchak PA, Ostertag KL and Coyle MA. LifeShirt® system as a monitor of heat stress and dehydration. Ventura, CA: VivoMetrics, Inc., 2004.

Noury N, Dittmar A, Corroy C, et al. VTAMN – a smart clothe for ambulatory remote monitoring of physiological parameters and activity. In: 26th Annual international IEEE engineering in medicine and biology society (EMBS) conference, San Francisco, CA, 2004.

Pandian

Mohanavelu

Safeer

. Smart vest: Wearable multi-parameter remote physiological monitoring system. Med Eng Phys 2008; 30: 466–477.

Ziegler

Frydrysiak

. Initial research into the structure and working conditions of textile thermocouples. Fibres Text Eastern Eur 2010; 17: 84–88.

Bielska

Sibinski

Lukasik

. Polymer temperature sensor for textronic applications. Mater Sci Eng B 2009; 165: 50–162.

De Rossi

Della Santa

Mazzoldi

. Dressware: Wearable hardware. Mater Sci Eng C 1999; 7: 31–35.

10.

Locher I, Kirstein T and Troester G. Routing methods adapted to e-textiles. In: 37th International symposium on microelectronics (IMAPS 2004), Long Beach, CA, 2004.

11.

Locher I, Kirstein T and Troester G. Temperature profile estimation with smart textiles. In: International conference on intelligent textiles, smart clothing, well-being, and design, Tampere, Finland, 2005.

12.

Sibinski

Jakubowska

Sloma

. Flexible Temperature sensors on fibers. Sensors 2010; 10: 7934–7946.

13.

Sapoff M. 32.3 Thermistor thermometers. In: Webster JG (ed.) The Measurement, Instrumentation, and Sensors: Handbook. Boca Raton, FL, USA: CRC Press LLC, 1999, pp.3225–3241.

14.

Burns J. 32.2 Resistive thermometers. In: Webster JG (ed.) The Measurement, Instrumentation, and Sensors: Handbook. Boca Raton, FL, USA: CRC Press LLC, 1999, pp.3213–3224.

15.

John Fontes. Chapter 20 – Temperature sensors. In: Wilson JS (ed.) Sensor technology handbook. Burlington: Newnes, 2005, pp.531–561.

16.

Catrysse

Puers

Hertleer

. Towards the integration of textile sensors in a wireless monitoring suit. Sens Actuators A: Physical 2004; 114: 302–311.

17.

Pirotte F and Klefstad-Sillonville FP. MERMOTH: MEdical Remote MOnitoring of cloTHes. In: Proceedings of pHealth, LUZERNE, 31 January 2006.

18.

Caldani L, Pacelli M, Gamboni PG, et al. Real-time monitoring through wearable systems. In: International workshop on wearable micro and nanosystems for personalised health, Valencia/Spain, 2008.

19.

Martin T, Jones M, Chong J, et al. Design and implementation of an electronic textile jumpsuit. In: International symposium on wearable computers (ISWC 2009), Linz, Austria, 4–5 September 2009, pp.157–158.

20.

Zysset C, Cherenack K, Kinkeldei T, et al. Weaving integrated circuits into textiles. In: International symposium on wearable computers (ISWC 2010), Seoul, South Korea, 10–13 October 2010, pp.1–8.

21.

Cherenack

Zysset

Kinkeldei

. Woven electronic fibers with sensing and display functions for smart textiles. Adv Materi 2010; 22: 5178–5182.

22.

Locher

Troester

. Fundamental building blocks for circuits on textiles. IEEE Trans Adv Packaging 2007; 30: 541–550.

23.

Cottet

Grzyb

Kirstein

. Electrical characterization of textile transmission lines. EEE Trans Adv Packaging 2003; 26: 182–190.

24.

Lam Po Tang

. Recent developments in flexible wearable electronics for monitoring applications. Trans Inst Meas Control 2007; 29: 283–300.

25.

Pantelopoulos

Bourbakis

. A survey on wearable sensor-based systems for health monitoring and prognosis. IEEE Trans Syst Man Cybern, Part C: Appl Rev 2010; 40: 1–12.

26.

Wijesiriwardana R, Dias T and Mukhopadhyay S. Resistive fibre-meshed transducers. In: Proceedings of seventh IEEE international symposium on wearable computers, White Plains, NY, 21–23 October 2013, pp.200–209.

27.

Dogan I. Chapter 4 – RTD temperature sensors. In: Microcontroller based temperature monitoring and control. Oxford: Newnes, 2002, pp.87–106.

28.

McGee TD. Electrical resistance temperature measurement using metallic sensors. In: Principle and method of temperature measurement. Wiley, 1988, pp.141–168.

29.

Davis JR and Committee AIH. Metals handbook. ASM International, 1998.

30.

http://www.matbase.com (accessed 20 October 2012).

31.

http://www.metalprices.com (accessed on 20 October 2012).