Feasibility analysis of wireless power delivery to implanted sensors of XLIF patients

Representing an unoptimized solution, the current hardware employed which features successive approximation ADC is the EFM8BB52 microcontroller developed by Silicon Labs [94]. When operating at an ambient temperature of 25°C with a 1.8 V supply voltage (V_VDD), the Table 1 represents typical supply current consumption. The decision to utilize an external voltage reference for ADC was done so as to using V_VDD as the reference voltage (V_VDDREF) requires a voltage divider that consumed 16 μA in comparison to the 5 μA of I_EXTVREF. Additionally, there exists an integrated temperature sensor with a supply current consumption of 50 μA (I_TSENSE). However, in an effort to save power, this hardware element can be disabled.

The piezoresistive sensors used for pressure sensing are the P122 High Silicon Pressure Sensor Die developed by Amphenol Sensors [95] as shown in Figure 3. When excited with a 1 mA input, the sensor will produce an output that is proportional to the pressure exerted upon the die. For full scale output, a minimum voltage supply of 320 mV is required. Although there exist seven P122 sensors on the piezoresistive sensor array, it is not necessary to power all components simultaneously.

Figure 3:

Piezoresistive sensor array.

While not currently realized in this project, data must be able to be transmitted wirelessly back to the user for real-time analysis of the fusion process. LSK is a means of backscatter that allows data to be transmitted back to the primary coil by shorting the secondary capacitive tank. Analysis of this technique allows for data transmission without the use of extra components which would require powering. If LSK is determined to be infeasible for this project, active components will be necessary for data transmission. UWB is the most promising technology for this, and it has been shown that data has been transferred using millimeter-size components across centimeter distances. Two examples in literature have transferred data across 10 cm, requiring a 1.2 V power supply and 1.675 mA [83] or 1.833 mA [82]. As such, the summation of all considered elements is 1.867 mA if LSK can be utilized, or 3.7 mA if active data transmission hardware is necessary [110, 111].

With the module developed by the same company, the high power transmission circuit also makes use of two ICs created by Taidecent—the XKT-801 and XKT-1511, as shown in Figure 4.

Figure 4:

High power transmitting coil and circuitry.

Arising from complexities in the geometry of the screwed interbody cage design as shown in Figure 5, it was not possible to wrap coils around the horizontal beams of the implant. Rather, coils could only be wrapped around one external vertical beam and internal vertical beam. Following the formula for self-inductance Eq. (1), a reduction in turns results in drastically lower inductance. Ultimately, affecting the mutual inductance between the transmitter and receiver Eq. (2). Hence, this project focuses on the feasibility analysis for the standalone implant architecture before exploring coil architectures for the screwed variant.

Self-inductance: (1) L=N2μrμ0AI {\rm{L}} = {{{N^2}{\mu _r}{\mu _0}A} \over I}

Mutual inductance: (2) M=kL1L2 M = k\sqrt {{L_1}{L_2}}

Initial receiver coil designs were developed around a prototype created that mirrored the architecture of the commercially available PEEK interbody cage Coroent XL developed by NuVasive [96]. As explained earlier, for the purpose of feasibility analysis, preliminary receiver coil architectures were created around the standalone design seen in Figure 5.

Figure 5:

Implant standalone version (left) and screwed version (right) [96].

Using the dimensions of the interbody implant, a CAD model was developed in Creo Parametric [97] to assist with the hand-winding of coils. The design featured extrusions which guided the positioning of the wire. Only minor changes were made in an attempt to mirror the original implant dimensions as closely as possible [112].

To create the coils, the coil winding helper was fastened in place with a clamp and manually wound using 0.4 mm copper enamel-coated wire. Wire of this diameter was chosen due to three reasons:

Commercial availability.

The three coil architectures can be seen in Figure 6. The added complexity of V2 and V3 significantly impacted the manufacturing time with the V3 coil requiring 60 min of winding time compared to 15 min for V1, and 45 min for V2. Unfortunately for the experimental results, only the V3 coil resulted in reasonable levels of induced power. As such results from the other two coils will not be mentioned.

Figure 6:

Coil V1, V2, and V3 designs using standalone implant geometry.

While not ideal, dependent on the results conducted in the vertical misalignment experiments, copper enamel-coated wire with a diameter of 0.25 mm also represents a suitable choice for coil material. A wire of this dimension would allow for an almost four-fold increase in coil turns if a multilayer design was employed, resulting in a greater self-inductance despite the increased coil length Eq. (1).

Consulting the American Wire Gage Conductor Size table for copper, it can be observed that 22 gage wire (consisting of a 0.254 mm diameter), has the maximum current carrying capabilities of 0.92 A. Following the current requirements outlined in success measure 5, this proves to be a suitable wire choice if necessary.

Vertical misalignment refers to the vertical spacing between the primary and secondary coils. When the two coils are perfectly aligned with no misalignment, the coupling factor and mutual inductance have the highest efficiency [52]. From the point of origin, magnetic waves undergo attenuation when propagating through any medium, with the magnetic permeability of the medium influencing this factor [48, 51]. As such, the effects of vertical misalignment must be explored to understand the capabilities of the transmitting coil.

Practical analysis for the vertical misalignment experiments was conducted using the setup seen in Figure 7. The transmitting coil was affixed to the elevated wooden beam using a fishing line, and the coupling distance between the receiver and the transmitter was varied with a scissor jack. Due to the metallic composition of the scissor jack, a wooden spacer was placed between the receiver coil and the scissor jack plate to ensure no interference with the magnetic field. Vertical adjustments were made by turning the knob attached to the scissor jack.

Figure 7:

Vertical misalignment testing setup.

As demonstrated in literature [52], inductive coupling WPT is most efficient when resonant coupling is employed. When the secondary coil loosely resonates at the frequency of the wavelength, the coupling factor increases in strength. This technique requires integrating the optimal receiver LC circuit as explored below.

Using the Rigol DS1054Z Oscilloscope [98], it was seen that the working frequency of the high-power transmitting coil was 141 kHz with an average time period of 7.09 μs. It is unknown how the working frequency is dictated, and a discussion with the IC manufacturer, Taidacent, yielded no results. The inductance of the coils was measured using a Keysight U1731C handheld LCR meter [99]. The following formula can be used to determine the necessary capacitor value for the LC circuit. (3) ω0=2πf0=1LC;f0=1T0 ∴C=T024π2L=14π2f02L \matrix{{{\omega _0} = 2\pi {f_0} = {1 \over {\sqrt {LC} }};{f_0} = {1 \over {{T_0}}}} \hfill \cr {\;\;\;\;\;\;\therefore C = {{T_0^2} \over {4{\pi ^2}L}} = {1 \over {4{\pi ^2}f_0^2L}}} \hfill \cr }

In an attempt to offset the instantaneous forward voltage drop introduced with rectifying diodes, testing was conducted using Schottky IN5822 diodes from Onsemi [100]. For instantaneous forward current values of 50 mA and below at 25°C, each Schottky diode exhibits only a 0.22 V forward voltage drop. For minimum forward voltage drop, a half wave rectifier circuit was implemented using the IN5822 and a large smoothing capacitor. The circuit can be seen in Figure 8.

Figure 8:

Half-bridge receiver perfectly tuned for 0.4 mm standalone v3 coil.

At varying heights, by varying loads, the voltage and current levels were tested with a half-wave Schottky diode bridge rectifier. It was determined that to meet success measure 4 and 5, a 463.8 Ω was ideal, as demonstrated in Figure 9. At 100 mm, 1.828 V and 3.941 mA were achieved. While the minimum voltage was achieved at later distances using a smaller DC load, the induced current was not sufficient for powering of an active antenna.

Figure 9:

Receiver half-bridge distant dependent CC results through air.

Inductive links have the tightest coupling factor when the primary and secondary coils are parallel to one another, with no angular, planar, or horizontal misalignment present. Due to the vertebral alignment of the spine, natural angle is introduced and results in angular misalignment if the transmitting coil is held parallel to the back region. Figure 10 demonstrates the natural curve of the spine, along with spinal deformities which XLIF aims to provide relief from [1].

Figure 10:

X-ray of a natural spine [119], kyphosis [118], and lordosis [117].

Utilizing the same testing setup outlined before, the effects of angular misalignment were explored across the 40–120 mm range at 20 mm intervals as shown in Figure 11. The primary coil was attached to a dowel constructed from wood and was rotated clockwise along specified angles the range of 0–90°, with 90° being perpendicular to the receiving coil. The transmitting circuit uses the high-power transmitting coil provided with 32 V, and the receiving circuit makes use of a Schottky diode half wave rectifier. The DC load connected to the filtered rectified output was 463.8 Ω.

Figure 11:

Receiver CC results at different distances and angles.

Table 2 demonstrates the losses in PTE and power delivery load (PDL) at varying misalignments and distances in reference to the baseline 0° values, whereby the two coils are perfectly aligned in a parallel manner.

Angular misalignment results in greater WPT disruption when comparing CC values. Thus, it would be preferable for the patient, or doctor, to rotate the transmitting coil to be parallel with the interbody cage. In doing so, it is important to note that reducing angular misalignment results in a greater level of vertical misalignment [114,115,116, 120].

While there have been inductive coupling receiver architectures designed which limit the effects of angular misalignment [54, 55], these systems are not suitable for the project as the designs necessitate coils traveling along all three-dimensional planes. As necessitated by success measure 1, increasing the implant size with further coils is not a feasible solution. A reduction in wire size would facilitate a 3D receiver coil architecture but would require timely manufacturing methodology. Rather, it is suggested that a more optimized transmitter coil and transmitter circuit design is employed.

Considering the fact that the current unoptimized equipment meets the minimum threshold defined by feasibility success measure 4 and 5, angular misalignment is not an element that can be present in the system. As suggested, when designing the optimized primary coil, it is essential that an ‘overhead’ is implemented, whereby the transmitting coil is able to be rotated by a reasonable degree while still providing the necessary power to the IMD.

As the scope of the project revolves around the creation of an in vivo biomedical implant, it is essential for testing to be conducted through the medium of tissue substitutes to determine whether the biomechanical properties of the body affect the propagation of magnetic waves.

To simulate adipose and muscular tissue, chicken breast was used. In addition, there were two synthetic bone mediums developed by Sawbones [102] which were created to mimic the mechanical properties of human bone. The synthetic vertebrae consisted of a 0.3 mm layer of cortical bone and a 5 mm layer of cancellous bone. Cortical bone is the hard compact type of bone found on the outer layer of long bones, with a typical 0.3 mm shell seen on vertebrae in the lumbar spine [103]. The width of cancellous bone was determined based on the average dimension (29.7 mm) of the lumbar intervertebral disk [104].

Although substitutes for tendon were not integrated into the phantom tissue design, the three mediums chosen represented a tissue design from which assumptions could be formed. Figure 12 demonstrates the tissue substitutes, forming a 120 mm vertical misalignment between the two coils.

Figure 12:

Vertical misalignment testing setup using tissue substitutes.

Closed circuit tests were conducted three times and averaged at distances of 40 mm, 80 mm, and 120 mm with varying loads as outlined in the closed circuit results through air. A slight variance was observed in the closed circuit voltage measurements but can be attributed to a small shift in the transmission position when contact was made between the coil and the silicone rubber. Overall, negligible differences were observed when comparing Figure 13 with Figure 11.

Figure 13:

Receiver CC results at different distances through tissue.

It is seen that reducing the frequency at which the secondary coil resonates will provide PDL gains; however, at the expense of PTE [52]. As the perfectly tuned circuit is only capable of providing 1.8 V at 100 mm, it is not possible to reduce PTE. It is crucial to note that increasing PTE should remain a major focus on any further project developments as a lack of efficiency indicates heat dissipation in coils, tissue exposure to AC magnetic field, and interference with nearby electronics [50, 51, 47]. Furthermore, it is not recommended to reduce PTE in favor of increasing PDL as the current capabilities at 100 mm are already sufficient for the project following success measures 4 and 5.

For both the preliminary and secondary receiver circuits, a 1000 μF capacitor was chosen for the smoothing filter because a capacitor of this value highly minimizes DC voltage ripple. It is to be noted that the form factor of this capacitor is quite large, and in the context of an IMD where miniaturization is required, does not represent the optimal solution. This project limitation must be explored to determine the maximum allowable level of voltage ripple while still meeting the measures of success.

Regarding AC to DC voltage rectification, we must highlight the importance of developing a circuit that features minimal voltage drop. When observing the results from the experiments, using the IN5822 Schottky diode which featured a 0.22 V forward voltage drop, it is observed that the maximum powering distance is 100 mm when considering the same factors for success. Further optimized methods of rectification should be explored to optimize the PTE and PDL across the DC load on the secondary coil.

Temperature measurements were taken using a commercially available, low cost, high accuracy TMP117 sensor developed by Texas Instruments [105]. Presenting itself as the ideal solution for this analysis, the TMP117 provides 16-bit resolution with an accuracy of ±0.1°C across the range of −20°C to 50°C with no calibration. Electronics company Core Electronics [106] integrated the sensor in a package that allows for simple connectivity between the TMP117 sensor and a Raspberry Pi using the I2C data transfer protocol programmed in Python version 3+ [107].

Direct contact was made between the TMP117 and the receiving 0.4 mm standalone V3 coil under the following test conditions:

High power transmitter operating at 32 V and 0.07 A

In these conditions, the receiver coil outputted 12.26 V and 26.43 mA resulting in a power output 32.41 mW. While it is unexpected that power will be transmitted to the receiver for long periods of time, the aforementioned testing conditions were maintained for a 15 min time period to demonstrate the suitability and power carrying capabilities of the 0.4 mm solid copper wire.

Software application PuTTY [108] was used in tandem with a Raspberry Pi Zero W [109] to allow transmission of the I2C data back to the author’s computer over Secure Shell Protocol. The equipment setup is demonstrated in Figure 14.

Figure 14:

Temperature sensor setup.

The metallic elements in the TMP117 module as well as the wiring did not create interference in the magnetic field generated by the high-power transmitter. This was observed through the Rigol DS1054Z Oscilloscope [98], with no change in the CC voltage regardless of the module’s presence.

As observed in Figure 15, a strong positive correlation was seen which related wire temperature to time and power. Temperature variance (ΔT) was measured to be 0.242°C, starting at 21.367°C (T_min) and ending at 21.609°C (T_max). (6) ΔT=Tmax−Tmin \Delta T = {T_{max }} - {T_{min }}

Figure 15:

Receiver coil temperature as a function of time.

The results observed from the experiment highlight the power transmission capacity of the 0.4 mm solid copper wire, with a ΔT of 0.242°C. This temperature fluctuation is deemed negligible, as it remains well under the 2°C threshold at which protein denaturation becomes a risk in the human body [37, 85].

It is to be noted that there are limitations involved with this preliminary testing which do not account for the coil starting temperature when encased within the human body. It remains unexplored whether the relationship between temperature, power, and time behaves differently if the receiver coil’s starting temperature is elevated. Despite this, test conditions involving the delivery of excessive power over an extended time period affirm the attainment of success.