Wireless Power and Data Transfer
for
Sonar Array Applications

Final Report for the Period Jan.-Dec. 2003

Table of Contents

TopicPage Number

Executive Summary 3

Introduction4

Waveguide Layout 6

Waveguide Terminations 11

Power Accounting 22

Future research 35

Conclusion 35

References 36

Acknowledgements37

Executive Summary

This project will identify an efficient method to power multiple hydrophones wirelessly. In this research,the inside of an aluminum waveguide is energized with continuous wave, CW (unmodulated), radio frequency (RF) signals. Power will be extracted from the waveguide with stub monopole antennas located along the length of the waveguide. The CW RF will be converted to direct current voltage (VDC) via a rectifying antenna (rectenna) which is composed of an impedance matching circuit, a shunt diode, and a low pass filter. The VDC will be used to power the hydrophone and data telemetry circuitry. The sound acquired from each hydrophone will be processed and sent wirelessly to another location in the waveguide for further processing.

It is shown that the corporate/parallel layout is a low-cost method to distribute power to various staves in the array. Also, several schemes which will increase the number of elements on each stave have been discussed.

To prevent standing waves, it is necessary to terminate a waveguide with a multilayered/tapered material. A sample of material from ARC-Tech has been able to lower the SWR to about 1.17. It is a wedge fabricated from ML-79 material.

By using high-speed diodes it is possible to increase the efficiency of a rectenna. These diodes should have a low junction capacitance,Cjo, and a low series resistance, Rs. To handle large power fluctuations, a diode with a high reverse breakdown voltage, Vbr, should be used. The Skyworks SMS3924 diode with a Vbr of 70 Volts, Rs of 11 ohms, and Cjo of 1.5pF has been chosen as a good candidate for this project.

Introduction

This annual report will update the research on wireless power and data transfer, being conducted at the University of Connecticut. The report will provide possible schemes for extending the current research into a large scale design with the possibility of hundreds of hydrophoneelements.

The specific issues addressed in this report are:

1)Waveguide Layout. This portion of the report will give possible solutions on how to distribute power to multiple elements.

2)Waveguide Terminations. These may be necessary in order to prevent standing waves in the waveguide

3)Power Accounting. This is necessary in order to properly and safely power each hydrophone-element.

As a quick refresher, this project will identify an efficient way to power multiple hydrophone elements wirelessly. The proposed research consists of illuminating the inside of an aluminum waveguide with continuous wave, CW (unmodulated), radio frequency (RF) signals. The TE10 mode will be excited in the waveguide. This mode was selected due toits simpler power density distribution along the x-axis of the waveguide (Fig. 1). Power from the waveguide will be extracted by stub monopole antennas located in the center of the waveguide. The CW RF will be converted to direct current voltage (VDC) via a rectifying antenna (rectenna) to power the audio pre-amplifier and the data telemetry circuitry. The acoustical data acquired from each hydrophone will be digitized and it will be sent wirelessly to a receive antenna via the waveguide.

A simple graphical representation of the setup is shown below:

Fig. 1 Simplified Stave Layout

1) Waveguide Layout

The current design consists of a small waveguide with 4 elements, where each element is composed of a rectenna (power supply), hydrophone, audio pre-amplifier, and wireless telemetry circuitry. With this setup it is easy to assign a different RF frequency to each element in order to have multiple channels operating simultaneously. The current prototype uses the TI TRF6901 demo board which has 16 channels preconfigured to be tuned from 902-917 MHz, with a 1MHz channel separation. The TRF 6901is capable of transmitting at a maximum baud rate of 96 kbpsand its RF spectrum is displayed below:

Fig. 2 TRF6901 RF Spectrum (Ref. 13)

The total bandwidth of the channel, BWC, required to prevent cross-modulation is approximately 200 kHz. To increase the signal to noise ratio, the bandwidth has been increased to 1 MHz. The increase to 1 MHz is necessary to avoid interference from spurious frequencies which occur at the center frequency plus 350 kHz and at the center frequency minus 350 kHz.

Given the channel separation of 1 MHz for the TRF6901 and a need for 400 elements, a bandwidth of 400 MHz will be required to accommodate the data telemetry. One may argue that there are more “economical” modulation schemes which may provide more channels, such as Single-Side-Band (SSB), but they also require more elaborate circuitry and circuitry demands more power consumption. Therefore, frequency management becomes important as the number of channels increases.

Also, the CW power density inside the waveguide decreases along the z-axis as each element extracts a small amount of power from the waveguide. This implies that the elements near the CW transmitting antenna in the waveguide, z=0, will have to cope with a very large electric field while elements near the end of the waveguide, z=L, will see a much smaller electric field.

These two problems, frequency and power management, have led to the explorations of different methods that reuse RF channels and allow the electric field to be much smaller at z=0 . The proposed concept is similar to what the cellular telephone companies use by dividing a very big array into smaller cell arrays. By having a smaller number of elements per cell, say 50, it becomes easier to manage both the power and the frequency assignments. Such an implementation is envisioned below:

Fig. 3 Hull penetrator and cell arrays

Each cell array would then be made with one of the following architectures:

A)Serpentine/Series Feed (Ref. 2, page 478)

Fig. 4 Serpentine (series) architecture

With the serpentine configuration, a very long waveguide is folded at various lengths. The CW power is injected at one end of the waveguide (fig.4). It then travels down the first stave where it smoothly couples to the beginning of the next stave as long as the angle of the bend is not too steep. If the angle of thebend is too steep, it will cause some of the incoming CW RF to reflect back to the source antenna. The radius (r) of the bend must be large, approximately greater than 1.5 * wavelength in W/G(Ref. 3 page 39).

Fig. 5 Serpentine bend radius

B)Manifold/Space Feed (Ref. 4, page 166)

Fig. 6 Manifold /Space Feed

This design consists of a small piece of waveguide which is then coupled to multiple n waveguides by allowing the CW RF to split between n waveguides. The transition between the first short waveguide and the manifold should be gradual to prevent the first waveguide from behaving like an open ended waveguide which in turn would lead to high standing waves in the first waveguide.

C)Corporate/Parallel Feed (Ref. 4, page 166)

Fig. 7 Corporate/Parallel Feed

With this configuration, the power is split in a simple power divider with the output of the power divider being fed to each stave. The power divider will provide impedance matching between the 1 to n waveguides and it maybe built with microstrip technology or regular off-the-shelf coaxial power dividers.

This design is the easiest to build and it is also one of the easiest to analyze, and unlike the serpentine design, this design allow the power density at z=0 to be substantially lower than with z=0 for the serpentine designwith the same number of elements. Additionally, due to the small size of the power distribution device, the staves can be accommodated side by side without any decrease in performance.

Overall, this design seems to be the most efficient and the most practical of the three. As a result, this design looks like a very feasible solution for future multi-stave arrays.

2)Waveguide Terminations

When an electromagnetic wave travels in a waveguide, it has ane-jβz behavior with β being the phase constant and z beingdistance along the axis of propagation. Therefore, the power density along the z axis remains constant. If the end of the waveguide is shorted (no termination) the traveling wave will reflect off the end and it will create standing waves with a sin(βz) property, leading to localized voltage maxima and voltage minima.

Fig. 8 Waveguide with standing waves

Ideally, the power density would decrease as a function of z because each element would extract a small amount of power from the waveguide and eventually the last element would extract the remaining power in the waveguide. This would also imply that the elements working at the very end of the waveguide would have to be designed for much smaller power levels in order to operate properly.

To simplify this project and to maintain symmetry amongst the rectennas, it is important to minimize/cancel any standing waves in the waveguide. Again, the lack of standing waves in the waveguide will represent a constant power density along the z-axis and it will facilitate the positioning of the rectenna devices.

By inserting a termination at the end of each waveguide, reflected waves will be greatly attenuated with a reduction of standing waves. Terminations are either lossy or resonant. In lossy type terminations, energy is given off in the form of heat. When properly matched (impedance) the reflected waves from the lossy material will be greatly attenuated. With resonant type terminations, the thickness of the material is made to be a suitable fraction of the incoming wavelength. A portion of the signal is reflected at the face of the material and another portion is transmitted into the material. At the right thickness, the resonant material will transmit a wave which is 180 degrees out of phase with the face reflected wave. The two waves cancel each other and it appears as if the material prevents EM waves from reflecting off the end of the waveguide.

A)Conical Rod

Fig. 9, Conical Rod (Ref. 5 page 316)

  • The conical rod is made of a lossy material.
  • If l1 is greater than a couple of wavelengths, the SWR could be less than 1.04
  • The length l2 is adjusted to provide a total absorption greater than 20 dB
  • Low power applications
  • Broadband, bandwidth of about 40%

B)High Power Load

Fig. 10, High Power Load (Ref. 5 page 316)

  • The termination is made of a lossy material.
  • l1 is greater than a couple of wavelengths
  • The length l2 is adjusted to provide a total absorption greater than 20 dB
  • High power applications
  • Broadband, bandwidth of about 40%

C)Water Load

Fig. 11, Water Load (Ref. 5 page 316)

  • The termination is made of a glass rod with circulating water.
  • l1 is greater than a couple of wavelengths
  • By measuring the water temperature rise, power in the waveguide can be calculated.
  • Incident power is dissipated by the circulating water

D)Step Load

Fig. 12, Step Load (Ref. 5 page 316)

  • The termination is made of a lossy material.
  • l1 acts as a ¼ wave transformer
  • l1 is determined experimentally
  • The length l2 is adjusted to provide a total absorption greater than 20 dB
  • Bandwidth of about 10% with a SWR < 1.1.
  • Compact

E)Resonant Tile

Fig. 13, Resonant Tile (Ref. 6)

  • The ¼ wave layer produces an emergent wave (R2) which cancels out theface reflected wave (R1).
  • Depending on the dielectric constant of the ¼ wave layer, d may be very thin.
  • Reflections are 20 dB lower than normal incident waves
  • Reflections increase as the angle from normal incidence increases.
  • Very compact

F)Multilayer

Fig. 14, Multilayer (Ref. 7)

  • The termination is made of a lossy material with a very low concentration at the wave/material interface and a gradual increase in concentration towards the back of the material.
  • l1 is a function of how much absorption is required
  • l1 is determined experimentally/look up tables
  • Compact

G)Physical Tapering

Fig. 15, Physical Tappering (Ref. 7)

  • The termination is made of an evenly lossy material.
  • l1 should be greater than the wavelength to provide a smooth impedance change for the incoming wave.
  • l1 may be increased to reduce the SWR

H)¼ Wave Antenna w/Load

Fig. 16, ¼ Wave Antenna w/Load

  • Energy is coupled out of the W/G via the ¼ monopole and it is dissipated in the resistor
  • Inexpensive
  • Simple to build

Experimental data has been obtained for different waveguide terminations. Localized power measurements were obtained byusing a waveguide with a perforated top. The SWR was then calculated with the following test setup:

Fig. 17 Test Setup for measuring forward power (Pf) and reflected power (Pr).

Fig. 18 Test setup for measuring power along the W/G z-axis.

The data collected, suggested that standing waves were present inside of the waveguide. A quick check was made to determine if the power fluctuations along the W/G z-axis separation occurred at a distance predicted by theory. The following was performed:

TE10

m=1

n=0

ω= 2πf = 2π1GHz

μ=4*π*10-7 H/m (air)

ε=8.854*10-12 F/m (air)

a= 8” = 20.32 cm

b= 4”= 10.16 cm

So,

= 0.444m

Therefore, there should be a repetition of maxima and minima every ½ λg = 0.222m. By analyzing the Excel data, one can see that the maximum power values appear every 22 cm to every 23cm.

Since the experimental data coincides with theoretical data, it is now safe to say that there is a standing wave present in the waveguide. Please see chart below:

Fig. 19, Power Fluctuations in a Waveguide

The SWR based on the z-axis power measurements was determined from Fig. 19. One of the power measurements for the shorted waveguide sits at the noise floor (lowest possible reading by the power meter), Pmin =0. This means that there is no measurable power at this location and if we compute the SWR as a ratio of Emax/Emin = (sqrt(Pmax)*k) / (sqrt(Pmin)*k), where k is a constant (let k=1), then we can see that the SWR is equal to infinity for when the waveguide is shorted.

The following SWR values were then computed:

Material SWR

Short31622(no termination)

ARC Wedge1.93(resonant type material in a triangular shape similar to fig. 14)

Dense Urethane2.75(lossy material in a triangular shape similar to fig. 14)

ARC Perpendicular16788(setup similar to fig. 13)

Based on the experimental data obtained, the ARC wedge has provided decent results so far. ARC technologies (Ref. 7) has since recommended a combination of Fig. 14 and Fig. 15 (Multilayer and Physical tapering) to further reduce the SWR to values close to 1.1-1.3. A vertical-wedge shaped termination was fabricated fromARC ML-79 multi-layer foam and the following results were obtained:

Fig. 20, Power Fluctuations with ARC ML-79 RAM

The corresponding SWR was then computed to a very good value of 1.17. Based on these results, future terminations will be manufactured with this material.

3)Power Accounting.

Power accounting is necessary in order to properly and safely power each hydrophone-element. It should be done in an efficient manner to maximize the number of sensors per available power.

A)Sensor power requirements

The sensor will be composed of a hydrophone which produces an audio output in the order of a couple of millivolts. This output will be amplified by an OP-AMP, LMV751, which consumes 20 mW. The output of the LMV751 will be fed to the TI-Demo Board. From the Texas Instruments documentation, (Ref. 1), the board should consume approximately 79 mW when transmitting at a 0 dBm level. The TI board consists of an RF module and an analog to digital converter (ADC). The total power required will be approx. 100 mW, and to allow a margin of safety, assume that the total power required is 150mW. This power will be supplied through a switching regulator, which based on past work, will operate at 50% efficiency while regulating a voltage. Higher efficiencies maybe achieved if there is a small difference between the input and the output voltages. Based on these values, the required input power to the voltage regulator will have to be around 300 mW. With proper selection, the rectenna is expected to have an RF to DC conversion efficiency greater than 30%. Again, to be safe, assume that the rectenna has an efficiency of 30%. Based on these assumptions, the input power to the sensor will have to be in the order of 1 Watt. Please seeFig.21 below:

Fig. 21, Power Requirements for Sensor

B)Rectenna efficiency

One of the key parts of this project is the development of an efficient rectenna. The rectenna receives RF energy (antenna) and converts it to DC via a shunt diode. The diode configuration is similar to a diode clamper circuit, where it takes the input waveform and it shifts it down to a level where the diode just barely turns on. If the rectenna didn’t have a load,the input waveform would shift down to a point where the diode would just barely turn on (Fig.22-A). But with the introduction of a load, the input waveform turns the diode on for longer periods of time. Please see the pictures below:

Fig. 22, A) No Load B)Rectenna voltage and input wave w/load (Ref. 8)

V = input wave

Vf= diode turn-on voltage

Vd= voltage across the diode

= Phase when diode is turned on, function of load value

In Fig. 22-B, an input waveform is represented by a full sinusoid with a DC reference shifted down to a level represented by a dashed line. This signal would be present only if the diode didn’t exist in the circuit. A second waveform is then superimposed to represent the diode switching action in the circuit and it is represented by the sinusoid with a chopped upper half. The missing upper half, θ, is due to the amount of charge being discharged by the load and with an increase inthe load there will be an increase in θ. Ideally, with no load, the input waveform would be completely shifted to the dashed line and θ would be very small. As can be seen from Fig. 22-B, the diode almost sees a peak-peak input waveform while it is turned off (reverse-bias). This leads to one dilemma; it isn’t possible to simply raise the input voltage in order to increase the output power. There is a point when the input voltage will exceed the reverse breakdown voltage (Vbr) and the diode will begin to conduct while reverse biased. If the input voltage continues to increase, the diode will eventually suffer a catastrophic breakdown in its junction layer. As a general rule of thumb developed by McSpadden (Ref. 9), the output DC voltage should be less than Vbr / (2.2) to avoid operating the diode at its breakdown points. McSpadden has also developed a formula which takes into account the max. output power of a rectenna as a function of Vbr and as a function of the load resistance (RL) , , from the fact that, P = V2 / R, and that the output DC voltage should be ½ Vbr. Without a load,the diode will see approx. 2*Vpeak which should be less than the Vbr since without a load, the charge held in the capacitor will not discharge rapidly and the “Phase-on” (Fig. 22-B) will be very small, shifting the Vd signal on Fig. 22-B further down.