YUNJIA TEXTILE · Material guides
Textile Sensors & E-textile Guide
Understand how fabrics and yarns sense pressure or movement, how to connect them, and what a useful test result tells you.
Start with the mechanism, follow the materials through the assembly, then choose for the task. Research examples explain design choices; they are not specifications for a commercial product.
01 · Principles
How do textiles sense?
Pressure or movement changes a textile structure. A circuit reads the resulting electrical change as a signal.
Resistive pressure sensing
Compression changes a network or its contact with electrodes.
Capacitive pressure sensing
Compression changes separation, geometry or dielectric behaviour.
Resistive stretch sensing
Extension changes paths and contacts within the structure.
One input, different ways to sense it
Pressure and stretch are inputs; resistance and capacitance are ways to read the change. Resistance describes how much a path opposes electric current. Capacitance describes how much charge separated conductors store at a given voltage. Compression can change conductive contacts or squeeze an insulating layer, called a dielectric, between electrodes. Stretch can rearrange yarns and loops.
Look for the part that changes
In a resistive textile, contacts within the fabric and at its electrodes can both contribute to the signal. In a capacitive structure, deformation changes the electrical coupling between separated electrodes. Knowing which part changes helps you decide what to keep free to move and what to hold steady.
Why similar materials can sense in different ways
A carbon-based fabric can be the pressure-sensitive path, or it can serve as an electrode. Carbon-treated nonwoven fabric provides a resistance that changes under pressure. [1] Carbonized cotton around a compressible dielectric instead forms the conductive faces of a capacitor. [6] The arrangement gives the material its role.
To understand a resistance change, consider both the conductive path and the contacts along it. Models for resistive fabric stacks include contact with the electrodes; an explanation for a knitted strain sensor focuses on loop arrangement and contact changes. These are models of the particular structures, rather than separate measurements of every contact contribution. [1][3][7]
Two mechanisms can also share one textile. Coated yarn can provide a resistance signal for stretch, while an ion-gel interface at crossing yarns provides a capacitance signal for pressure. That distinction guides the choice of readout for each channel. [5]
The sensor should change; its lead usually should not
In a resistive sensor, the sensing region needs a useful, repeatable change in resistance. The lead needs to carry that signal with as little extra change as possible. Elasticity alone does not ensure electrical stability: repeated stretching can leave a yarn with a different resistance. [8] Test the sensing region and its connections separately, then check them together in the assembled system.
This guide focuses on resistive and capacitive pressure and stretch sensing. Other mechanisms, including piezoelectric and triboelectric sensing, need their own explanations.
02 · Structure
What makes up a textile sensor?
Follow two paths: how force reaches the sensing region, and how the electrical signal reaches the readout.
Identify the roles before counting layers
- 1 + 3 · Textile electrodes
- The conductive layers on either side.
- 2 · Compressible insulating layer
- Keeps the electrodes electrically separated and deforms under pressure.
- Leads → readout
- Connect each electrode to the capacitance readout.
- Support and attachment
- Affect how force reaches the sensing region.
The arrangement follows the sensing mechanism. Two electrodes can sit on the same side of a resistive fabric, so downward pressure need not mean current travels through the thickness. [1] A continuous knit can itself sense stretch without a separate layered stack. [7]
Keep the right contacts free—or fixed
Some sensing interfaces must change contact as they deform. A permanent electrical joint usually needs stable contact. Before sewing, bonding or encapsulating, identify which behaviour each location needs.
Why attachment position changes the response
Bonding can secure an assembly while also changing the interface that senses pressure. Direct hot-melt bonding between textile electrodes and sensitive fabric severely reduced the response in a tested resistive stack. Heat damage and adhesive penetration were possible explanations; their individual effects were not isolated. [2]
Attachment away from that interface offers a different arrangement. Conductive strips can be bonded to a nonconductive backing, or the non-adhesive faces of conductive tapes can face the sensitive cloth. [4][3] These examples use different materials, so they do not identify a best adhesive. They help answer a more useful question: which part needs to stay fixed, and which contact needs to move?
In the capacitive example above, the specific choices are carbonized cotton reinforced with thermoplastic polyurethane (TPU), a porous Ecoflex dielectric and sewing to hold the three layers together. [6] The diagram simplifies dimensions, pores and fastening; its external support shows the mounting context, not an additional material layer specified by that design.
A sensing point is not yet a pressure map
An array must distinguish location as well as load. Shared electrical paths can make one contact affect another channel; a cover can spread force mechanically. Electrode spacing alone is therefore not proof of spatial resolution.
The scanning circuit and its compensation therefore belong to the sensor design. They can reduce crosstalk—one loaded point affecting another channel—as demonstrated for selected two-location loads in a fabric array. [4] A pressure map still needs checks with the simultaneous contacts expected in use.
Know what the readout includes
A two-terminal resistance reading includes the full path between its terminals. A four-terminal setup uses separate current and voltage contacts to reduce the influence of leads and contacts outside the voltage-sensing points. Interfaces inside those points remain in the measurement. [10] A voltage divider converts resistance into a voltage that also depends on the circuit, so its output cannot be compared directly with a resistance reading. [7]
03 · Materials & integration
What roles do materials and methods play?
Choose each material for its job: sensing, making contact, carrying a signal or supporting a connection.
Fabric
A sensing region, electrode or support.
Yarn & stitches
A placed sensing path, lead or contact pad.
Flexible connection
A signal route that must tolerate movement.
Read a pattern as a possible circuit
A striped or Zebra-pattern cloth needs a continuity map: which stripes conduct, which are isolated, and where do they end? Appearance alone does not establish an electrode pair or a sensing matrix. Check those relationships after cutting and attachment before choosing a layout.
For material families and manufacturing basics, see the yarn and fabric sections of the Structure Guide.
Placement can matter as much as material
Where the conductive yarn sits can change which contacts move during stretching. In a plated knit, where two yarns occupy different positions within the loops, changing that placement changed the electrical response. [7] This makes placement a design variable alongside the choice of yarn.
Fabric structure also affects how an electrode deforms. Woven and knitted electrodes produced different responses in a capacitive assembly, but their starting mass per area differed too. [6] To compare manufacturing methods fairly, account for those material and construction differences.
A sewn route needs a reliable end connection
Sewing and embroidery place conductive yarn where it is needed. It may form a sensing path, an electrode, a lead or a contact pad. The processed route, the joint and the transition into a rigid module each need attention.
Choose the construction, not just the metal name. Silver-coated polymer thread, silver-coated copper hybrid yarn and stainless-steel sewing thread are different constructions. The evidence here does not support a general silver-versus-stainless performance ranking.
A conductive route also needs a joint that stays together. Simply crossing hybrid yarns was mechanically inadequate in an embroidered connection; a larger pad and polymer fixation held multiple microwire contacts together. [10] The joining method must suit the actual thread construction, including its polymer and metal components.
Can it stretch, recover and keep carrying the signal?
Stretchability alone does not tell you whether a connection keeps a steady resistance or returns to its starting value. Repeated stretching can shift the resistance measured afterward. [8] Separate three questions when choosing an elastic connection:
- During extension: does resistance change, or does the signal briefly disappear?
- After release: do its length and electrical baseline recover?
- After repeated use: are the conductor and both terminations still reliable?
A wrapped conductor can accommodate extension by changing its geometry. [8] For a complete elastic cable, the working range also depends on channel insulation and the connections at both ends.
What movement can change along a connection
Repeated stretching can shift the baseline. A 20 cm silver-coated-PBO/rubber-core yarn with a Tactel outer wrap increased in resistance by 137.87% after 2,150 cycles from 0 to 25% strain. The absolute increase was 0.0781 Ω. [8] Keeping both values helps judge the effect on a circuit. This was a reading after cycling, not continuous monitoring during motion, so it cannot reveal brief interruptions or qualify a complete cable.
The rigid end needs its own bending test. In a comparison of woven interconnects with and without a soft transition at an encapsulated board, four copper-plated-steel specimens with the transition reached the 100,000-cycle limit under 180° bending, 1 Hz and a 100 g hanging clamp. [9] That result applies to the tested assembly and stopping criterion; sampling once per cycle could miss shorter interruptions.
A joint must retain contact as well as conduct. Ultrasonic plastic welding can soften the polyester in a silver-coated copper hybrid yarn to fix microwire contacts; the metal wires are not melted together. Broken microwires occurred among failed joints. [10] Resistance and physical damage therefore belong in the same joint evaluation.
04 · Performance & testing
How should test results be read?
A useful result tells you how much the signal changes, how consistently it returns, and under what conditions.
Hysteresis
Does the way back differ? Compare loading and unloading at the same input.
Drift
Does the output stay steady? Hold the input constant and observe the output over a stated time.
Repeatability
Does another run agree? Repeat on the same specimen and location under the same conditions.
Keep the baseline and working range
A normalized curve such as (R − R0)/R0 makes relative change easy to see, but hides the starting resistance. Keep the absolute reading and state whether the baseline was unloaded, preloaded or reset each cycle. Repeated resets can hide drift.
Sensitivity belongs to a stated output and input range. Resistance, current and capacitance slopes are different quantities. A steep response over a small pressure interval does not establish the same sensitivity across the whole range.
Read the definition behind the number
Two percentages labelled “hysteresis” may use different calculations. Response time also needs a threshold, sampling rate and filtering information. Resolution needs noise and a decision rule: the smallest load demonstrated is not automatically the smallest change the system can distinguish.
Why sensitivity and hysteresis need a definition
A large slope may cover a small interval. A resistive fabric sensor has normalized-current slopes of 0.43, 0.18 and 0.10 kPa−1 over 1–4, 4–40 and 40–150 kPa respectively. [3] Each slope belongs to its pressure interval; pairing the largest slope with the full working range would overstate sensitivity.
The same physical change can give different percentages. For an ideal resistor at constant voltage, halving resistance doubles current: −50% in resistance becomes +100% in current. This is an arithmetic example, not measured data. A divider circuit needs its own model.
| Research result | How it was calculated |
|---|---|
| Resistive fabric · about 11.7% [3] | Maximum loading–unloading gap divided by full-scale response; cycle to 150 kPa, maximum gap at 70 kPa. Supplementary Fig. S1. |
| Capacitive stack · 8.8% [6] | Difference between areas under the two curves divided by area under the loading curve; 0–100 kPa. Eq. 3 and Fig. 7. |
The first calculation measures the largest gap; the second measures separation across the curves. “Area” here means area under a graph, not contact area. Different devices, ranges and formulas—and incomplete loading-rate and dwell information—prevent these percentages from ranking the materials.
Readout settings describe different parts of a measurement. A 100 kHz AC excitation drives the capacitance measurement at a yarn interface. [5] A glove patch sampled every 50 ms and smoothed ten successive readings with a moving average. [6] Excitation frequency, sampling interval and mechanical response time answer different questions.
Match the load, area and repeat being compared
Pressure is force divided by area: P = F/A. Use the loaded area, not the whole cloth size. For a finger or curved object with an unknown contact area, report force or the task rather than inventing pressure.
Also name what repeats: another cycle at one point, another location, another specimen or another batch. They reveal different sources of variation. A repeatable trace on one sample does not establish consistency across manufactured sensors.
Which conditions make two tests comparable?
- Specimen and assembly: material, dimensions, direction, electrodes, attachment, support and independent sample count.
- Loading: contact area, preload, range, rate, dwell, recovery, cycle history and environment.
- Readout: measured quantity, contact positions, excitation, sampling, filtering, baseline and metric formula.
- Outcome: absolute and normalized readings, variation, failures and the acceptance criterion.
The loading tool sets the nominal area. A textile test used 20 × 20 mm plates and a 0.5 N preload on much larger cloth specimens. [2] Using the plate area of 400 mm², 6 N gives an average pressure of 15 kPa and the preload gives 1.25 kPa. These calculations describe nominal average pressure; local stress within the textile can differ.
Keep resistance specifications distinct: Ω describes a terminal-to-terminal path; Ω/m needs a defined conductor length and tension; Ω/□ needs an appropriate sheet geometry; Ω·m needs a meaningful bulk geometry. Do not convert between them using assumed dimensions. Leave unreported test conditions unknown.
Count the exposure and the change
A cycle count describes exposure; the change afterward tells you what that exposure did. For example, a pressure sensor lost approximately 4.3% of its response amplitude after 2,000 cycles at 30 kPa. [3] Stretching, bending and washing stress an assembly differently, so compare the load and outcome alongside the count.
A damaged conductor can regain contact without recovering its original structure. Monitoring interruptions during use and measuring resistance afterward reveal different problems. [9] For washing, keep the method, drying and conditioning with the result: electrical performance after one laundering procedure cannot set a care instruction for every assembly. [8]
05 · Design & selection
How are textile sensors designed for use?
Choose the information you need first. Then choose the structure that can produce it in the intended assembly.
Define the task before selecting a material
Detecting contact, following movement, estimating force and mapping pressure are different tasks. State the working range, spatial detail and acceptable variation before comparing materials.
Mounting becomes part of the sensor
Follow bending
A knitted sensing region can turn movement into a resistance change. Mounted on a mechanical joint, it produced a bending-cycle signal through a voltage divider. [7] Turning that signal into a joint angle on a person also requires calibration for the mounting, fit and possible slippage.
Sense a grasp
A glove-mounted capacitive patch senses loading where the finger meets an object. Cup-holding demonstrations show this use of the electrode–dielectric stack. [6] Cup mass is not a calibrated finger pressure because grip and contact area vary.
A soft surface adds its own deformation
On a curved or inflatable body, the support beneath a fabric array moves too. Moving an array from a flat calibration setup to a soft robotic link brought noisier measurements and remaining unloading error. [4] Check the surface’s own movement alongside external contact, using the intended cover and readout.
For a seat or insole, extend that evaluation to sustained loading, contact geometry and repeated positioning. Those conditions need application-specific validation beyond the bending, grasping and soft-link examples here.
When one textile needs more than one signal
Separate electrical channels can help distinguish stretch from pressure. One textile combines coated-yarn resistance for strain with an ion-gel capacitance channel for pressure. Both channels were observed under separately applied strain and pressure, and two pressure points were demonstrated. [5]
For use during simultaneous movement and contact, the next question is how independently those channels respond. Apply each input alone, then combine them under the intended environmental conditions. The separate-input results provide a starting point, not a complete check of mixed inputs.
References
These sources support the specific examples discussed above. Diagrams in this guide are original conceptual explanations; no paper figures or experimental curves are reproduced.
- Xu, H. et al. (2020). Flexible Waterproof Piezoresistive Pressure Sensors with Wide Linear Working Range Based on Conductive Fabrics. Nano-Micro Letters, 12, 159. DOI: 10.1007/s40820-020-00498-y.
- Okss, A., Valisevskis, A., Briedis, U. & Baltina, I. (2023). Functional fabrics for multilayered textile pressure sensors: comparison of structures and sensitivity characteristics. Engineering for Rural Development, 419–426. DOI: 10.22616/ERDev.2023.22.TF089.
- Zheng, X. et al. (2025). Daily carbon ink functionalized conductive fabric for simple, low-cost, and wide range flexible pressure sensor with multifunctional applications. Diamond & Related Materials, 158, 112661. DOI: 10.1016/j.diamond.2025.112661. The hysteresis comparison uses supplementary Fig. S1.
- Day, N. et al. (2018). Scalable fabric tactile sensor arrays for soft bodies. Journal of Micromechanics and Microengineering, 28, 064004. DOI: 10.1088/1361-6439/aab221.
- Keum, K. et al. (2022). Mechanically robust textile-based strain and pressure multimodal sensors using metal nanowire/polymer conducting fibers. iScience, 25, 104032. DOI: 10.1016/j.isci.2022.104032.
- Ko, Y., Vu, C. C. & Kim, J. (2021). Carbonized Cotton Fabric-Based Flexible Capacitive Pressure Sensor Using a Porous Dielectric Layer with Tilted Air Gaps. Sensors, 21, 3895. DOI: 10.3390/s21113895.
- Kim, Y.-H. et al. (2024). Assessing the Role of Yarn Placement in Plated Knit Strain Sensors: A Detailed Study of Their Electromechanical Properties and Applicability in Bending Cycle Monitoring. Sensors, 24, 1690. DOI: 10.3390/s24051690.
- Schwarz, A. et al. (2011). Electro-conductive and elastic hybrid yarns – The effects of stretching, cyclic straining and washing on their electro-conductive properties. Materials & Design, 32, 4247–4256. DOI: 10.1016/j.matdes.2011.04.021.
- Veske, P., Bossuyt, F., Thielemans, F. & Vanfleteren, J. (2023). Measuring the Flex Life of Conductive Yarns in Narrow Fabric. Micromachines, 14, 781. DOI: 10.3390/mi14040781.
- Dils, C. et al. (2022). Interconnecting embroidered hybrid conductive yarns by ultrasonic plastic welding for e-textiles. Textile Research Journal, 92(21–22), 4501–4520. DOI: 10.1177/00405175221101015. A 2023 corrigendum updates the funding statement.