This work created a textile noise dosimeter suitable for military use. Overexposure to noise is known to cause permanent hearing damage, and as a result, employers are required to implement suitable health-monitoring measures when workers will be exposed to loud noises [1
]. Noise exposure can lead to a variety of disorders [4
] with tinnitus and noise-induced hearing loss (NIHL) being the most common among them.
Hearing damage is dependent on the amplitude, frequency and duration of sound exposure. Initially, NIHL results in a loss of hearing at a range of higher frequencies (3–6 kHz) [7
] and can hinder the perception of sound. NIHL is a result of either an extremely high-level sound or gradual exposure to high-levels of sound over a period of time. A secondary symptom of acoustic trauma can be tinnitus, which presents as the perception of a sound when no sound source is present. In both cases, damage can be permanent with no accepted common treatment methods. In the last 10 years there have been 1505 claims to the United Kingdom Industrial Injuries Disablement Benefit scheme relating to NIHL in the period 2007–2016 [8
], and in the United States of America more than 10 million adults under 70 (6 percent of the under 70 population) exhibit some form of hearing loss from noise exposure [9
Many military activities generate high levels of noise, which fall into two categories; impulse noise, or background noise. Impulse noise can be related to activities such as firearms use, where a single gunshot is capable of producing a sound pressure level (SPL) of 140–170 dB(A) [7
]. While the use of firearms is a major source of noise, research has also investigated sound exposure experienced by helicopter aircrews, who are exposed to continuous noise levels of up to 100 dB(A) during flying operations [10
]. The type and magnitude of noise exposure will, therefore, depend on the military role of the personnel involved. A study by Rovig et al. investigated the hearing health of a number of personnel operating on an aircraft carrier flight deck [11
]. As part of this study they showed that staff in different roles were exposed to significantly different levels of noise. It was observed that flight-deck personnel had an average exposure of 109 dB(A), compared to 92 dB(A) for engineers.
Overexposure to noise resulted in 278 personnel being discharged from the UK Armed Forces between April 2011 and March 2016 [12
]. NIHL is also a serious concern for militaries in other countries, such as the USA [13
]. There have been a number of studies into hearing damage specific to the military [14
], with research showing a far higher number of cases of hearing damage among military personnel when compared to other noise-exposed workers, such as shipbuilders [16
]. Reduced sound perception can introduce significant dangers in a military setting, where orders may not be heard correctly. As highlighted by Grantham, NIHL caused communication issues during the Battle of Fallujah [14
A noise dosimeter is the most reliable way to determine a worker’s noise exposure; however, commercially available solutions are not suitable for military use. The device must not interfere with normal military operations, which may not be true for jacket-mounted devices or helmet attachments currently available on the market. Further to this, asymmetric hearing damage is known to be more common when firearms are frequently used compared to the acoustic injuries encountered in civilian roles [17
]. Therefore, in a military setting it is desirable to monitor the noise exposure to each ear individually.
As a result, it was proposed that the dosimeter should be incorporated into a helmet cover, with sensing elements on both sides of the helmet cover. This required the creation of an electronic textile helmet cover.
Electronic textiles and smart textiles have grown in prevalence in recent years finding applications in a number of sectors [18
] including energy-storing fabrics to cool buildings [19
], energy harvesting [20
], and medical devices [21
]. Wearable electronics have also seen a recent increase in attention. A number of wearable triboelectric nanogenerators have been developed which are capable of collecting acoustic noise [22
]; while designed as generators, the same technology could be applied to allow for sensing applications [24
], such as acoustic sensing [25
Electronics can be integrated into textiles in a number of ways; by attaching them onto the surface of a fabric, integrating conductive elements into a textile, or incorporating electronics at a yarn level, where the yarns can be used subsequently to create a fabric [27
]. Integration at a yarn level hides the electronics better and can protect electronics from external forces better.
In this work, commercially available microelectromechanical system (MEMS) microphones were used as the sensing element of an electronic acoustic sensing yarn, based on Nottingham Trent University’s fibre electronic technology [28
]. The technology has previously been used to create a temperature-sensing yarn [30
]. MEMS devices were used as opposed to a more-complex triboelectric devices to simplify the manufacturing process and reduce cost.
To create the acoustic sensing yarn, microphones were first soldered onto fine multi-strand copper wire, attached to a carrier fibre to increase the tensile strength, and protected with a micro-pod crafted from an ultraviolet (UV)-curable resin to protect the microphone. The ensemble, with additional packing fibres, was then put through a small diameter circular warp knitting machine, covering it with a fibre sleeve to form the acoustic sensing yarn. The final acoustic sensing yarn was drapeable, flexible, and soft to the touch like a normal textile yarn.
While commercial sensing devices (microphones) were employed to construct the acoustic sensing yarn, the inclusion of the micro-pod and fibers may alter the sensor’s response to external stimulus. Textiles structures are commonly used in sound-absorbing applications [31
], as this type of porous structure gives sound waves many opportunities to interact with the fibers and lose energy. Both the structure of the material, and the sound-wave frequency, will affect the sound absorbance properties of a textile [32
]. It is therefore possible that the knitted textile structure of the acoustic sensing yarn would absorb sound. As it is very difficult to model a textile structure accurately, the sound absorbance of textile had to be determined and understood experimentally.
As a result, the acoustic sensing yarns embedded with microphones were carefully characterized, and design rules for the encapsulation were developed. To achieve this the yarns were tested for a range of relevant frequencies and amplitudes at each stage of the yarn production process (informed by EN 61672-1:2013 [34
], and other sources [35
Once characterised, the acoustic sensing yarns were paired with self-contained, supporting electronics to record and store the collected acoustic information (herein referred to as the hardware module). The yarns were subsequently re-tested with the hardware module. Finally, the acoustic sensing yarns, and supporting hardware, were incorporated into a knitted helmet cover created using computerised flat-bed 3D knitting technology. Two yarns were integrated into each helmet cover, one over each ear. The completed helmet cover dosimeter was tested over a range of conditions to validate the functionality of the final prototype for its intended purpose.
2. Materials and Methods
2.1. Acoustic Sensing Yarn Fabrication and Design Considerations
The acoustic sensing yarns were fabricated using a handcrafting process in three steps (as shown in Figure 1
a–c). Initially, a MEMS microphone was soldered to multi-strand copper wire using a lead-free solder paste (Nordson Solder Plus 7024454 Lead-free; Nordson Corperation, Westlake, OH, USA) and an infrared (IR) spot reflow soldering system (PDR IR-E3 Rework System; PDR Design and Manufacturing Centre, Crawley, UK). The PUI Audio VM1010 (PUI Audio, Dayton, OH, USA) was chosen as the sensing element in this work after extensive trials with other MEMS microphones for this application (not presented in this paper). Ultimately, this microphone was chosen as it was able to operate correctly without external power and because it had a degree of moisture and dust ingress protection [36
Eliminating the need for an external power source was highly desirable as it significantly simplified the construction process because only two copper interconnects, and not three, were needed. The requirement for external power could be eliminated by wiring the GA1 pin (wake-on-sound acoustic threshold adjust pin 1) to the ground on the recording device. Physically, it was believed that the energy generated by the motion of the piezoelectric element at the core of the microphone (due to incident soundwaves) was providing enough power for the device.
Once soldered, the microphone and solder joints were encapsulated within a UV curable polymer micro-pod, which protected the device from mechanical stresses and provided some chemical resilience. A synthetic yarn with a high-tensile strength (Vectran™, Kuraray America Inc., Houston, TX, USA) was included within the micro-pod to add strength to the ensemble in the direction of the copper interconnects. The micro-pod was formed by injecting resin (Multi-Cure® 9001-E-V-3.7, Dymax Corporation, Torrington, CT, USA) into a cylindrical Teflon mould and curing it using a UV light source (Blue-Wave™ 50, Dymax Corporation, Torrington, CT, USA). To allow the microphones to continue to function, a small cavity in the micro-pod was required over the microphone’s inlet, which was created using a rubber tube (o.d. ~1 mm). To avoid resin entering the cavity and impairing the microphone’s operation, resin was injected and cured in multiple stages.
The micro-pod and copper interconnects were then inserted within a circular warp-knitting machine (RIUS MC braiding machine, RIUS, Barcelona, Spain). Four polyester packing yarns were included around the micro-pod which created a final acoustic sensing yarn of a diameter of 7 mm.
2.2. Prototype Acoustic Sensing Helmet Cover
The prototype acoustic sensing helmet covers consisted of three main components: the knitted helmet cover, the acoustic sensing yarns, and the supporting hardware electronics modules. The hardware modules were created using two modified commercially available Dictaphones (Mini USB Voice Recorder, Wjiling), as this offered a low-cost and small size solution. The final hardware modules were covered in a silicon-based coating to add a degree of water-resistance to the modules and improve their mechanical resilience.
The bespoke knitted helmet cover used in this work was designed to fit on top of a standard British Army combat helmet. The cover was produced using a merino wool and Kevlar mix on a Shima Seiki computerised flat-bed knitting machine (SIR, 14 gauge, Shima Seiki, Wakayama, UK). The helmet cover had two knitted channels on either side allowing for the insertion of the acoustic sensing yarns above each ear. These yarns each attached to a hardware module, which was inserted into knitted pockets at the rear of the helmet cover (Figure 1
2.3. Testing Procedure
Acoustic sensing yarns were tested in an acoustically quiet environment achieved using a bespoke acoustic testing chamber. The testing chamber comprised of two parts: the outer chamber was an acoustically insulated rack designed for storing computer servers (Orion Mini Acoustic Rack, Orion, Leeds, UK), while the inner chamber was built from a 17 cm diameter extruded polyvinyl chloride (PVC) pipe (see Figure 2
). The PVC pipe fit over the top of a base speaker which was used to generate the test sounds (Bass Face SPL6M.2 800 W 6.5 inch Mid-Bass Car Speaker Single, Base Face, Macclesfield, UK). The PVC tube was clad in acoustic insulation foam (Adhesive PUR Foam Soundproofing Sheet, RS Components Ltd., Corby, UK) to provide sound insulation. The final inner chamber was 175 mm high. A cap for the inner chamber was built from insulation foam and reinforced with cardboard and a plastic coating. This cap was used to position the microphones under test as well as a high-accuracy calibration microphone used to determine the output sound from the speaker (Brüel & Kjær Type 4190 with a Photon + signal analyser; Brüel & Kjær, Nærum, Denmark). Sounds into the speaker were computer generated and fed through an amplifier (TeLe Hi-Fi A6 audio amplifier, Venezia, Italy).
Generated audio signals had a sinusoidal waveform at a consistent amplitude and frequency; this was confirmed using the calibration microphone. Some additional spectral features were observed near the operating limits of the system (low frequencies and very high amplitudes). Under normal operating conditions, the frequencies between 63 Hz–8000 Hz could be explored without additional (undesirable) spectral features. Additional spectral features at 31.5 Hz prevented accurate testing at that frequency.
The experimental design ensured that a slight misplacement of the lid did not have a significant effect on readings, as the input sound level was determined by the calibration microphone and not a setting on the computer. The chamber minimised the effects of external noise on the microphone pick-up during testing and provided an important layer of protection to the operator when high sound pressure level (SPL) sounds were being tested.
For yarn-level testing, the samples were attached to the top surface of the inner chamber and their copper wires were clipped to wires leading to a signal out and a ground. The signal was run through a 5 m audio extension (Maplin 3.5 mm Stereo Jack Extension Cable 5 m; Maplin Electronics, Rotherham, UK) into a sound card (Dynamode USB Sound Card; Dynamode UK Ltd., Watford, UK).
Signals were recorded and processed using a bespoke Python v2.7 (Python Software Foundation, Wilmington, DE, USA) script, which made use of the SciPy [37
], Matplotlib [38
] and PyAudio [39
] modules. A 10-second signal was recorded (based on information given in British Standard EN 61672-1:2013 [34
]), the signal was fast-Fourier transformed, and a peak picking algorithm was used to find peaks between 5 Hz and 10,000 Hz in the signal. The signal peaks were then sorted by amplitude. If the selected frequency was within ±1% of the value identified by the calibration microphone, then the sensor under test was deemed to have correctly selected the frequency.
Peak values were used throughout this work unless otherwise stated. Presented amplitude values were typically the average of five results, with error values given as the standard deviation. The amplitude of the sensors response is given in arbitrary units (arb) related to a voltage output from the microphone.
For prototype-level testing, the yarns, or the yarns within the knitted helmet cover, were attached onto the top surface of the inner acoustic testing chamber. The hardware module was used to record a series of sounds covering a range of frequencies and amplitudes. The recordings were downloaded and 10-second intervals relating to each test condition were extracted (using Audacity v2.1.2). The sound files were individually read into a bespoke Python script that analysed the data in a similar fashion to the process described for the yarn-level testing, output and amplitude value.
Data presented in this work has been prepared using either Microsoft Excel (Microsoft Corporation, Redmond, WA, USA), IGOR Pro (Version 126.96.36.199; Wavematrics, Tigard, OR, USA), or Matplotlib [38
An acoustic monitoring helmet cover has been produced. The device can measure noise exposure using small-scale MEMS microphones at the core of the yarns used to fabricate the cover. The acoustic sensing yarns have been carefully characterized and validated over a range of frequencies and sound pressure amplitudes relevant to a military setting.
This work showed that the inclusion of a small textile structure around a microphone did not have a substantial effect on the recorded signal; hence, the structure did not absorb sound over the range of conditions tested. The inclusion of an additional knitted tube (which the acoustic sensing yarn was fed into), also did not affect the microphone response.
A bespoke (handcraft) manufacturing technique has been developed that facilitates the repeatable fabrication of multiple yarns that provide consistent responses. Future work will transition the production process to an automated or semi-automated method. A helmet cover, supporting electronics, and processing software have been produced and tested with the acoustic sensing yarns to show the utility of the fully integrated acoustic monitoring helmet cover in a laboratory setting (TRL 4).
Ultimately, the use of a modified commercial hardware solution is not scalable, and, within the work reported, the solution employed introduced an additional degree of variation into the results. Future work will investigate the creation of a bespoke hardware module for the storage of data from the acoustic sensing yarns. Combined with a method to more rapidly produce acoustic sensing yarns, this will allow for the production of a larger number of helmet covers which would be necessary to conduct field trials.