betaFeedback

WorksPhysics and Astronomy

Rotating Capacitive System for Measurement of a Third Order Gravity Gradient Based on a Single-Axis Mechanical Oscillator. Part I, Theory

Abstract

This paper is a feasibility study on a single-axis mechanical oscillator, mounted on a miniature circular rotating platform, that is proposed to be used for measuring a quantity proportional to the second vertical derivative of gravitational acceleration projected onto the oscillator's sensitivity axis. The latter is aligned with the axis of rotation. Part I establishes theoretical feasibility only. Design constraints, quantitative sensitivity, error budgets, and mobile platform performance will be addressed in Part II.

Introduction

Gravity gradients serve as highly valuable sources of information across a range of strategic applications. These fields of application include oil and gas exploration, mineral exploration, the detection of subsurface voids, tunnels, and similar underground structures, heavy mass detection, passive GPS-denied navigation, and various defence-related operations [2]. Gravity gradients represent the first spatial derivatives of the gravitational acceleration vector g⃗\vec{g}

∂∂xgx=Gxx,∂∂ygy=Gyy,∂∂zgz=Gzz∂∂ygx=∂∂xgy=Gyx=Gxy,∂∂zgx=∂∂xgz=Gzx=Gxz∂∂zgy=∂∂ygz=Gzy=Gyz(1)\begin{aligned} \frac{\partial}{\partial x}g_x &= G_{xx}, \qquad\frac{\partial}{\partial y}g_y = G_{yy}, \qquad\frac{\partial}{\partial z}g_z = G_{zz} \\ \frac{\partial}{\partial y}g_x &= \frac{\partial}{\partial x}g_y = G_{yx} = G_{xy}, \\ \frac{\partial}{\partial z}g_x &= \frac{\partial}{\partial x}g_z = G_{zx} = G_{xz} \\ \frac{\partial}{\partial z}g_y &= \frac{\partial}{\partial y}g_z = G_{zy} = G_{yz} \tag*{(1)} \end{aligned}

Gravity gradient components obey the following Laplacian identity

Gxx+Gyy+Gzz=Ω2=Ωx2+Ωy2+Ωz2(2)G_{xx}+G_{yy}+G_{zz}=\Omega^2=\Omega_x^2+\Omega_y^2+\Omega_z^2 \tag*{(2)}

where Ωx,Ωy,Ωz\Omega_x,\Omega_y,\Omega_z the cartesian components of the angular rotation vector in the rotating coordinate frame.

The airborne deployable Falcon generation of rotating gravity gradiometers, developed by Lockheed Martin Corporation (LMC), is able to simultaneously measure the GxyG_{xy} component and the combination GyyG_{yy}-GxxG_{xx} which is called the differential gravity field curvature [3]. LMC’s Full Tensor Gradiometer (FTG) rotating type systems are able to measure all gravity gradient components but their stable operational time is much slower than the later Falcon’s versions one [4]. Classic and Quantum free fall gravity gradiometers can measure the GzzG_{zz} component under stationary conditions [2, 5], and there are proposals for measuring different components as well [6]. The distributed type of gravity gradiometers developed by Gravitec Instruments (UK, NZ, AU) group of companies in 2000-2009 [8], and later, in 2016-2023, by the University of Western Australia, can measure GzxG_{zx} and GzyG_{zy} components or their combinations [9]. A full classification of all gravity gradiometer systems and their operational principles can be found in [2].

Currently, the Falcon and FTG generations of gravity gradiometers are the only mobile types of precision gradiometric instruments capable of being used on airborne surveying platforms. However, they are high-cost, bulky, cumbersome, heavy, and have high power requirements. These types of gravity gradiometers are used on medium-sized aircrafts, helicopters, and strategic submarines but cannot be installed on drones, AUVs, UAVs, and other unmanned platforms, including submersibles. Miniaturized gravity gradiometers suitable for such carriers are seen as a strategic new direction in mobile gravity gradiometry that can revolutionize the technology and its applications [10, 11]. All gravity gradiometers intended for use on mobile platforms suffer from the same problem: what is measured is not pure gravity gradients, but rather quantities containing non-inertial terms arising from the platform's angular motion. In order to cancel out such contamination, specifically designed mounting frames are used, which provide active compensation for any unwanted degree of freedom of the mobile platforms. These frames are called stable tables, and they may cost much more than the gradiometers themselves.

The system under consideration in this study aims to be compatible with deployment on manned and unmanned mobile platforms and to directly measure the signal proportional to the following quantity

Gzzz=∂∂zGzz(3)G_{zzz}=\frac{\partial}{\partial z}G_{zz} \tag*{(3)}

where the Z-axis is assumed to be aligned with the plumb line and the platform’s axis of rotation. This quantity is called the vertical spatial derivative of the GzzG_{zz} gravity gradient component which is almost impossible to measure directly among the other components shown in Eq. (1). If the vertical gravity gradient GzzG_{zz} is measured in a coordinate frame experiencing rotations with respect to stationary frame of reference, then the so-called dynamic gravity gradient is one which is measured, containing the squared angular rotation frequencies as shown below

G~zz=Gzz+Ωx2+Ωy2(4)\widetilde{G}_{zz}=G_{zz}+\Omega_x^2+\Omega_y^2 \tag*{(4)}

However, the angular rotation terms do not depend on spatial coordinates and this makes the quantity in Eq. (3) independent of the rotational contamination which otherwise, in practice, needs to be mitigated by using complex and expensive countermeasures. The authors of this study are not aware of any proposals for measuring the third order vertical gravity gradient except the one discussed in [12, 13].

A single-axis mechanical oscillator in a rotating coordinate frame

A single-axis mechanical oscillator is mounted in a radial position r0r_0 on a disk rotating at angular rate Ω\Omega, with its sensitivity axis parallel to the vertical axis ZZ, as depicted in Figure 1. For the oscillator test mass displacement ΔZ\Delta Z, velocity Vz=dΔZ/dtV_z=d\Delta Z/dt, mass mm, resonant angular frequency ω0=2πf0\omega_0=2\pi f_0 and relaxation time τ\tau, one has

Figure 1. A single-axis mechanical oscillator is positioned along a radial direction of the rotating disk that coincides with the X-axis of the coordinate system XYZ rotating with the disk. The coordinate system (X0,Y0,Z0)(X_0,Y_0,Z_0) is a non-rotating local coordinate frame. The axes Z and Z0_0 are both aligned perpendicular to the disk’s plane. Here the standard right-handed Cartesian coordinate system is used where the rotation angle is positive anticlockwise.

d2dt2ΔZ+2τddtΔZ+ω02ΔZ=az(t)−az(t)+1mNL(t)(5)\frac{d^2}{dt^2}\Delta Z+\frac{2}{\tau}\frac{d}{dt}\Delta Z+\omega_0^2\Delta Z=a_z(t)-a_z(t)+\frac{1}{m}N_L(t) \tag*{(5)}

where az(t)a_z(t) is the vertical linear acceleration of the rotating platform, NL(t)N_L(t) is the random Langevin noise term and gz(t)g_z(t) is the projection of gravitational acceleration vector on the oscillator’s sensitivity axis ZZ. Furthermore, we consider an example of such a mechanical oscillator as a cantilever flexor fixed at one end and having the test mass at the other, as shown in Figure 3. However, different types of single-axis mechanical oscillators can be used in practical implementation of the system under consideration. If the oscillator’s sensitivity axis is exactly aligned with the axis of rotation, then Eq. (5) does not contain any non-inertial terms originating from Coriolis, Centrifugal and Euler forces which are present in any rotating coordinate frame [14]. The case when the latter does not hold is considered in the error budget analysis section in Part II of this study.

Figure 3. Three-dimensional representation of the rotating capacitive gradiometer. A single-axis oscillator is mounted at radial offset r0r_0 on a platform rotating at angular velocity Ω\Omega. Its floating test-mass electrode undergoes axial displacement ΔZ(t)\Delta Z(t) while sequentially traversing four stationary electrode pairs. The overlap-dependent coupling and displacement-dependent gap jointly modulate the measured capacitance.

For the sake of simplicity, the vertical linear acceleration can be considered as the strongest spectral harmonic at an angular frequency ω\omega

az(t)=Fa(ω)az(ω)Sin⁡(ωt)(6)a_z(t)=F_a(\omega)a_z(\omega)\operatorname{Sin}(\omega t) \tag*{(6)}

where Fa(ω)F_a(\omega) is a spectral attenuation factor in case the rotating platform is mechanically stabilized against possible linear disturbances.

By differentiating (5) one has

d2dt2Vz+2τddtVz+ω02Vz=ddtgz(t)−ddtaz(t)+1mddtNL(t)(7)\frac{d^2}{dt^2}V_z+\frac{2}{\tau}\frac{d}{dt}V_z+\omega_0^2V_z=\frac{d}{dt}g_z(t)-\frac{d}{dt}a_z(t)+\frac{1}{m}\frac{d}{dt}N_L(t) \tag*{(7)}

where VzV_z is the oscillator test mass velocity and ddtgz(t)\frac{d}{dt}g_z(t) is the total time derivative of the gravity acceleration component. In the case depicted in Figure 1, the latter can be expressed as [2]

ddtgz(t)=∂∂tgz(t)−r0ΩGzxSin⁡(Ωt+ϕ0)+r0ΩGzyCos⁡(Ωt+ϕ0)(8)\frac{d}{dt}g_z(t)=\frac{\partial}{\partial t}g_z(t)-r_0\Omega G_{zx}\operatorname{Sin}(\Omega t+\phi_0)+r_0\Omega G_{zy}\operatorname{Cos}(\Omega t+\phi_0) \tag*{(8)}

where ϕ0\phi_0, GzxG_{zx}, GzyG_{zy} are consequently the initial oscillator’s centre of mass angular position and the local values of the gravity gradient components with respect to the non-rotating reference frame (X0,Y0,Z0)(X_0,Y_0,Z_0).

Then (7) can be written as follows

d2dt2Vz+2τddtVz+ω02Vz=−r0ΩGzxSin⁡(Ωt+ϕ0)+r0ΩGzyCos⁡(Ωt+ϕ0)−ddtaz(t)+1mddtNL(t)(9)\frac{d^2}{dt^2}V_z+\frac{2}{\tau}\frac{d}{dt}V_z+\omega_0^2V_z=-r_0\Omega G_{zx}\operatorname{Sin}(\Omega t+\phi_0)+r_0\Omega G_{zy}\operatorname{Cos}(\Omega t+\phi_0)-\frac{d}{dt}a_z(t)+\frac{1}{m}\frac{d}{dt}N_L(t) \tag*{(9)}

The ∂gz(t)∂t\frac{\partial g_z(t)}{\partial t} term is excluded in (9) from further consideration, since the temporal variations of gravitational acceleration, like tidal ones, does not significantly affect the measurement process. However, this term can be a calibration signal generated at the same rotational frequency Ω\Omega by a variable gravity field source.

Let kΩ=Ω/ω0k_\Omega=\Omega/\omega_0, kω=ω/ω0k_\omega=\omega/\omega_0 and Q=ω0τ/2Q=\omega_0\tau/2, then the general solution of (9) is as follows

Vz(t)=r0Ωω021(1−kΩ2)2+(kΩQ)2××{Gzx[kΩQCos⁡(Ωt)−(1−kΩ2)Sin⁡(Ωt)]+Gzy[kΩQSin⁡(Ωt)+(1−kΩ2)Cos⁡(Ωt)]}−1ω02ωFa(ω)az(ω)(1−kΩ2)2+(kΩQ)2××[kωQSin⁡(ωt)+(1−kω2)Cos⁡(ωt)]+vz(t)(10)\begin{aligned} V_z(t)={}&\frac{r_0\Omega}{\omega_0^2}\frac{1}{(1-k_\Omega^2)^2+\left(\frac{k_\Omega}{Q}\right)^2}\times\\ &\times\left\{G_{zx}\left[\frac{k_\Omega}{Q}\operatorname{Cos}(\Omega t)-(1-k_\Omega^2)\operatorname{Sin}(\Omega t)\right]\right.\\ &\left.+G_{zy}\left[\frac{k_\Omega}{Q}\operatorname{Sin}(\Omega t)+(1-k_\Omega^2)\operatorname{Cos}(\Omega t)\right]\right\}\\ &-\frac{1}{\omega_0^2}\frac{\omega F_a(\omega)a_z(\omega)}{(1-k_\Omega^2)^2+\left(\frac{k_\Omega}{Q}\right)^2}\times\\ &\times\left[\frac{k_\omega}{Q}\operatorname{Sin}(\omega t)+(1-k_\omega^2)\operatorname{Cos}(\omega t)\right]+v_z(t) \tag*{(10)} \end{aligned}

where vz(t)v_z(t) is the velocity noise term [14]

vz(t)=1mω∫−∞tdξddξNL(ξ)Sin⁡[ωˉ(t−ξ)]e−t−ξτ=1mω∫−∞tdξddξNL(ξ)Sin⁡[ωˉ(t−ξ)]e−t−ξτ(11)\begin{aligned} v_z(t)&=\frac{1}{m\omega}\int_{-\infty}^{t}d\xi\frac{d}{d\xi}N_L(\xi)\operatorname{Sin}[\bar{\omega}(t-\xi)]e^{-\frac{t-\xi}{\tau}}\\ &=\frac{1}{m\omega}\int_{-\infty}^{t}d\xi\frac{d}{d\xi}N_L(\xi)\operatorname{Sin}[\bar{\omega}(t-\xi)]e^{-\frac{t-\xi}{\tau}} \tag*{(11)} \end{aligned}

and

ωˉ=ω02−1τ2.\bar{\omega}=\sqrt{\omega_0^2-\frac{1}{\tau^2}}.

By taking the integral in (11) by parts, one has

vz(t)=ddt{1mωˉ∫−∞tdξ NL(ξ)Sin⁡[ωˉ(t−ξ)]e−t−ξτ}=ddtδZ(t)(12)\begin{aligned} v_z(t)&=\frac{d}{dt}\left\{\frac{1}{m\bar{\omega}}\int_{-\infty}^{t}d\xi\,N_L(\xi)\operatorname{Sin}[\bar{\omega}(t-\xi)]e^{-\frac{t-\xi}{\tau}}\right\}\\ &=\frac{d}{dt}\delta Z(t) \tag*{(12)} \end{aligned}

By using (12) and (10), one can find the general solution of (5) for the oscillator mechanical displacement ΔZ(t)\Delta Z(t) including the Langevin noise term

ΔZ(t)=r0ω021(1−kΩ2)2+(kΩQ)2×{GZx[kΩQsin⁡(Ωt)+(1−kΩ2)cos⁡(Ωt)]+GZy[−kΩQcos⁡(Ωt)+(1−kΩ2)sin⁡(Ωt)]}+1ω02ωFa(ω)az(ω)(1−kΩ2)2+(kΩQ)2×[−kωωcos⁡(ωt)+(1−kω2)sin⁡(ωt)]+δZ(t)(13)\Delta Z(t)=\frac{r_0}{\omega_0^2}\frac{1}{(1-k_\Omega^2)^2+\left(\frac{k_\Omega}{Q}\right)^2}\times \left\{G_{Zx}\left[\frac{k_\Omega}{Q}\sin(\Omega t)+(1-k_\Omega^2)\cos(\Omega t)\right] +G_{Zy}\left[-\frac{k_\Omega}{Q}\cos(\Omega t)+(1-k_\Omega^2)\sin(\Omega t)\right]\right\} +\frac{1}{\omega_0^2}\frac{\omega F_a(\omega)a_z(\omega)}{(1-k_\Omega^2)^2+\left(\frac{k_\Omega}{Q}\right)^2}\times \left[-\frac{k_\omega}{\omega}\cos(\omega t)+(1-k_\omega^2)\sin(\omega t)\right] +\delta Z(t) \tag*{(13)}

where

δZ(t)=1mωˉ∫−∞tdξ NL(ξ)sin⁡[ωˉ(t−ξ)]e−t−ξτ(14)\delta Z(t)=\frac{1}{m\bar{\omega}}\int_{-\infty}^{t}d\xi\,N_L(\xi)\sin[\bar{\omega}(t-\xi)]e^{-\frac{t-\xi}{\tau}} \tag*{(14)}

is the corresponding oscillator test mass response to the Langevin noise source. The latter obeys the standard auto-correlation rule

⟨NL(t)NL(t′)⟩=4mkBTτδ(t−t′)(15)\langle N_L(t)N_L(t')\rangle=\frac{4mk_{\mathrm{B}}T}{\tau}\delta(t-t') \tag*{(15)}

where kBk_{\mathrm{B}} is the Boltzmann constant and T is the ambient temperature.

Capacitive modulation geometrical model

Consider four stationary capacitors C₀ formed by pairs of adjacent flat metal strips, which are connected in parallel to one sensing rail and the companion rail is grounded as shown in Figure 2. The variable gating capacitors C(t) are formed by an effective overlap area Aeff(t)A_{\mathrm{eff}}(t) between the oscillator test mass and the stationary strips separated by gap d0+ΔZ(t)d_0+\Delta Z(t) while the former is crossing the latter along a circle, as depicted in Figure 3. The gap depends on the oscillator’s test mass mechanical displacement ΔZ(t)\Delta Z(t).

Figure 2. Equivalent circuit of the capacitive sensing system. Four fixed capacitors C₀ are connected in parallel between the common sensing rail and ground. The electrically floating rotating test-mass electrode forms two variable coupling capacitors C(t) in series, contributing C(t)/2 to the total capacitance when their values are treated equal. The case when this does not hold will be considered in Part II.

The equivalent total capacitance is as follows

Ctotal(t)=4C0+12C(t)=4C0+12ε0εAeff(t)d0+ΔZ(t)(16)C_{\mathrm{total}}(t)=4C_0+\frac{1}{2}C(t)=4C_0+\frac{1}{2}\frac{\varepsilon_0\varepsilon A_{\mathrm{eff}}(t)}{d_0+\Delta Z(t)} \tag*{(16)}

where ε0\varepsilon_0 is the free air dielectric constant and ε\varepsilon is the relative dielectric constant which is greater than one if the capacitive plates are coated by thin layers of advanced dielectric materials as is widely used in MEMS capacitive sensors [16].

Without loss of generality, let the rotating electrode occupy the radial interval r1≤r≤r2r_1\leq r\leq r_2 with angular width β\beta, and let each stationary capacitive electrode have an angular width α\alpha (see Figure 4). Both widths are expressed in radians. Figure 3 above shows a particular case when α=β\alpha=\beta. The rotor angle is

Plan view of the electrode geometry

Figure 4. Plan view of the electrode geometry. The rotating electrode occupies r1≤r≤r2r_1 \leq r \leq r_2 and angular width β\beta, while each of the four stationary electrode pairs has angular width α\alpha and is centred at an azimuth separated by π/2\pi/2 from its neighbours. The nominal oscillator radius is r0r_0. The diagram also indicates electrode clearances and the gap geometry. Fourfold symmetry suppresses all geometrical overlap harmonics except integer multiples of 4Ω4\Omega in the ideal device.

ϕ(t)=Ωt+ϕ0\phi(t)=\Omega t+\phi_0

Let the four stationary electrode pairs be centred at

θn=nπ2,n=0,1,2,3…(17)\theta_n=\frac{n\pi}{2},\qquad n=0,1,2,3\ldots \tag*{(17)}

Further, define the radial area coefficient

KA=r22−r122(18)K_A=\frac{r_2^2-r_1^2}{2} \tag*{(18)}

For an angular overlap of γ\gamma radians, the geometrical overlap area between the rotating and stationary electrodes is exactly

Aeff(t)=KAγ(t)(19)A_{\mathrm{eff}}(t)=K_A\gamma(t) \tag*{(19)}

The overlap function γ(t)\gamma(t) is the circular cross-correlation of two angular windows. For the electrodes’ widths satisfying α+β≤2π\alpha+\beta\leq2\pi, one can define the shortest signed angular separation

δn(t)=wrap⁡(−π,π)(Ωt+ϕ0−θn)\delta_n(t)=\operatorname{wrap}_{(-\pi,\pi)}(\Omega t+\phi_0-\theta_n)

Then the angular overlap is then as follows

γn(t)={min⁡(α,β),∣δn∣≤∣α−β∣2α−β2−∣δn∣,∣α−β∣2<∣δn∣<α+β20,∣δn∣≥α+β2\gamma_n(t) = \begin{cases} \min(\alpha,\beta), & |\delta_n| \leq\frac{|\alpha-\beta|}{2} \\ \frac{\alpha-\beta}{2} - |\delta_n|, & \frac{|\alpha-\beta|}{2} < |\delta_n| < \frac{\alpha+\beta}{2} \\ 0, & |\delta_n| \geq\frac{\alpha+\beta}{2} \end{cases}

and

An(t)=KAγn(t)(20)A_n(t) = K_A\gamma_n(t) \tag*{(20)}

The overlap rises linearly as the rotating electrode enters a stationary sector, remains constant if α≫β\alpha\gg\beta, and decreases linearly as the rotating electrode leaves it.

The effective geometrical overlap is as follows

Aeff(t)=∑n=03An(Ωt+ϕ0−nπ2)A_{\mathit{eff}}(t) = \sum_{n=0}^{3} A_n\left(\Omega t+\phi_0-\frac{n\pi}{2}\right)

where An(t)A_n(t) denotes the single-pair overlap function defined in Eq. (20).

For exactly identical pairs one has

Aeff(t+π2Ω)=Aeff(t)A_{\mathit{eff}}\left(t+\frac{\pi}{2\Omega}\right)=A_{\mathit{eff}}(t)

Consequently, although the rotor completes one revolution in 2π/Ω2\pi/\Omega time interval, the effective overlap repeats for times per revolution. So, its fundamental modulation frequency is 4Ω4\Omega.

The Fourier series coefficients can be derived analytically without approximating the general overlap trapezoidal waveform. Let A0(θ)A_0(\theta) be the overlap area of one stationary electrode pair centred at zero angle. Its Fourier expansion is

A0(θ)=c0+2∑k=1∞ckcos⁡(kθ)A_0(\theta)=c_0+2\sum_{k=1}^{\infty}c_k\cos(k\theta)

where coefficients are

cm=12π∫−ππA0(θ)e−jkθ dθc_m=\frac{1}{2\pi}\int_{-\pi}^{\pi}A_0(\theta)e^{-jk\theta}\,d\theta

Because the overlap function is the circular convolution of the rotating and stationary angular windows, its Fourier coefficients are products of their respective window coefficients

ck=2KAπk2sin⁡(kα2)sin⁡(kβ2),k≠0c_k=\frac{2K_A}{\pi k^2}\sin\left(\frac{k\alpha}{2}\right)\sin\left(\frac{k\beta}{2}\right),\quad k\ne0

The dc coefficient is as follows

c0=KAαβ2πc_0=\frac{K_A\alpha\beta}{2\pi}

The complete Fourier series of one electrode pair is

Aeff(θ)=KAαβ2π+4KAπ∑k=1∞sin⁡(kα2)sin⁡(kβ2)k2cos⁡(kθ)A_{\mathit{eff}}(\theta)=\frac{K_A\alpha\beta}{2\pi}+\frac{4K_A}{\pi}\sum_{k=1}^{\infty}\frac{\sin\left(\frac{k\alpha}{2}\right)\sin\left(\frac{k\beta}{2}\right)}{k^2}\cos(k\theta)

This result applies to the periodic angular-overlap model, including the wraparound at the end of each revolution.

The four shifted by π/2\pi/2 angle electrode pairs, connected in parallel, produce the harmonic sum

∑n=03e−jknπ/2\sum_{n=0}^{3}e^{-jkn\pi/2}

This sum is zero unless kk is an integer multiple of four. Replacing k=4nk=4n (n=1,2,3…∞n=1,2,3\ldots\infty), one finds the complete analytical Fourier expansion for the assumed ideal geometry

Aeff(t)=2KAαβπ+KAπ∑n=1∞sin⁡(2nα)sin⁡(2nβ)n2×cos⁡(4nΩt+4nϕ0)(21)A_{\mathrm{eff}}(t)=\frac{2K_A\alpha\beta}{\pi}+\frac{K_A}{\pi}\sum_{n=1}^{\infty}\frac{\sin(2n\alpha)\sin(2n\beta)}{n^2}\times\cos(4n\Omega t+4n\phi_0) \tag*{(21)}

By substituting Eq. (21) into Eq. (16) and assuming that ΔZ(t)≪d0\Delta Z(t)\ll d_0, one has

Ctotal(t)=4C0+ε0εKAαβd0π+12ε0εKAd0π××∑n=1∞sin⁡(2nα)sin⁡(2nβ)n2cos⁡(4nΩt+4nϕ0)−ε0εKAαβd02πΔZ(t)−12ε0εKAd0πΔZ××∑n=1∞sin⁡(2nα)sin⁡(2nβ)n2cos⁡(4nΩt+4nϕ0)(22)\begin{aligned} C_{\mathrm{total}}(t)={}&4C_0+\frac{\varepsilon_0\varepsilon K_A\alpha\beta}{d_0\pi}+\frac{1}{2}\frac{\varepsilon_0\varepsilon K_A}{d_0\pi}\times\\ &\times\sum_{n=1}^{\infty}\frac{\sin(2n\alpha)\sin(2n\beta)}{n^2}\cos(4n\Omega t+4n\phi_0)-\\ &\frac{\varepsilon_0\varepsilon K_A\alpha\beta}{d_0^2\pi}\Delta Z(t)-\frac{1}{2}\frac{\varepsilon_0\varepsilon K_A}{d_0\pi}\Delta Z\times\\ &\times\sum_{n=1}^{\infty}\frac{\sin(2n\alpha)\sin(2n\beta)}{n^2}\cos(4n\Omega t+4n\phi_0) \tag*{(22)} \end{aligned}

where ΔZ(t)\Delta Z(t) is presented in Eq. (13). One can see that the leading capacitance modulation harmonics in Eq. (22) are ω\omega, Ω\Omega, 4Ω4\Omega, and the corresponding intermodulation sidebands are 3Ω3\Omega, 4Ω±ω4\Omega\pm\omega, 5Ω5\Omega. The lowest intermodulation harmonic 3Ω3\Omega provides the basis for the synchronous demodulation and further signal processing as will be shown in Part II of this study. It is worth noting that the 3Ω3\Omega component is a readout frequency conversion product and not, by itself, is associated with a mechanical force.

Parametric modulation of the oscillator response

In the practical measurement system, which will be presented in Part II of this study, the total sensing capacitor is an integral part of the capacitive-to-phase converter [1] pumped with a RF carrier. The latter produces a mean-square voltage ⟨V2⟩\langle V^2\rangle across the capacitor’s electrodes and this results in the variable electrostatic force between them. From Eq. (16) one has for ΔZ(t)≪d0\Delta Z(t)\ll d_0 [17]

Fze(t)=12⟨V2⟩∂Ctotal∂ΔZ=−14⟨V2⟩ε0εAeff(t)[d0+ΔZ(t)]2≅Fz,0e(t)+ke(t)ΔZ(t)(23)F_z^e(t)=\frac{1}{2}\langle V^2\rangle\frac{\partial C_{\mathrm{total}}}{\partial\Delta Z}=-\frac{1}{4}\langle V^2\rangle\frac{\varepsilon_0\varepsilon A_{\mathrm{eff}}(t)}{[d_0+\Delta Z(t)]^2}\cong F_{z,0}^e(t)+k_e(t)\Delta Z(t) \tag*{(23)}

where Aeff(t)A_{\mathrm{eff}}(t) is presented in Eq. (21), and

Fz,0e(t)=−14⟨V2⟩ε0εAeff(t)d02(24)F_{z,0}^e(t)=-\frac{1}{4}\langle V^2\rangle\frac{\varepsilon_0\varepsilon A_{\mathrm{eff}}(t)}{d_0^2} \tag*{(24)}

The time-dependent electrostatic stiffness is

ke(t)=12⟨V2⟩ε0εAeff(t)d03(25)k_e(t)=\frac{1}{2}\langle V^2\rangle\frac{\varepsilon_0\varepsilon A_{\mathrm{eff}}(t)}{d_0^3} \tag*{(25)}

Eq.5 is now replaced by the following

d2dt2ΔZ+2τddtΔZ+ωeff2(1−h(t))ΔZ=gz(t)−az(t)+1mNL(t)−14⟨V2⟩ε0εAeff(t)md02(26)\frac{d^2}{dt^2}\Delta Z+\frac{2}{\tau}\frac{d}{dt}\Delta Z+\omega_{\mathrm{eff}}^2(1-h(t))\Delta Z=g_z(t)-a_z(t)+\frac{1}{m}N_L(t)-\frac{1}{4}\langle V^2\rangle\frac{\varepsilon_0\varepsilon A_{\mathrm{eff}}(t)}{md_0^2} \tag*{(26)}

where the dimensionless parametric modulation factor h(t)h(t) is as follows

h(t)=12⟨V2⟩ε0εKAπmωeff2d03×∑n=1∞sin⁡(2nα)sin⁡(2nβ)n2cos⁡(4nΩt+4nϕ0)(27)h(t)=\frac{1}{2}\langle V^2\rangle\frac{\varepsilon_0\varepsilon K_A}{\pi m\omega_{\mathrm{eff}}^2d_0^3}\times\sum_{n=1}^{\infty}\frac{\sin(2n\alpha)\sin(2n\beta)}{n^2}\cos(4n\Omega t+4n\phi_0) \tag*{(27)}

and

ωeff2=ω02−12⟨V2⟩ε0εKAαβπmd03(28)\omega_{\mathrm{eff}}^2=\omega_0^2-\frac{1}{2}\langle V^2\rangle\frac{\varepsilon_0\varepsilon K_A\alpha\beta}{\pi md_0^3} \tag*{(28)}

The effective mechanical resonant frequency results from the electrostatic spring softening effect as the effective overlap area function Aeff(t)A_{\mathrm{eff}}(t) in Eq. (25) contains a constant term.

Eq. (26) is a non-homogeneous Hill-type equation similar to that of discussed in [18]. A theoretical conceptual design for a spaceborne radially vibrating, rotating gravitational gradient sensor was introduced by Chobotov in 1968. It was based on a single mechanical oscillator positioned on a rotating disk with its sensitivity axis aligned with radial direction, i.e. aligned with the surface of the disk under rotation. This proposal has not resulted in any further developments but it is an interesting example of diverging from the traditional gravity gradiometry, as introduced here as well.

If the parametric modulation factor satisfies the condition h(t)≪1h(t)\ll1, which will be provided in the practical case, then the solution of Eq. (26) can be represented as

ΔZnp(t)+ΔZp(t),ΔZp(t)≪ΔZnp(t)(29)\Delta Z_{\mathrm{np}}(t)+\Delta Z_p(t),\quad\Delta Z_p(t)\ll\Delta Z_{\mathrm{np}}(t) \tag*{(29)}

where ΔZnp(t)\Delta Z_{\mathrm{np}}(t) is the non-parametric oscillator response represented in Eq. (13) where ω0\omega_0 is replaced by ωeff\omega_{\mathrm{eff}}, and ΔZp(t)\Delta Z_p(t) is the oscillator response to the parametric electrostatic modulation

d2dt2ΔZp+2τddtΔZp+ωeff2ΔZp=−14⟨V2⟩ε0εAeff(t)md02+ωeff2h(t)ΔZnp(t)(30)\frac{d^2}{dt^2}\Delta Z_p+\frac{2}{\tau}\frac{d}{dt}\Delta Z_p+\omega_{\mathrm{eff}}^2\Delta Z_p=-\frac{1}{4}\langle V^2\rangle\frac{\varepsilon_0\varepsilon A_{\mathrm{eff}}(t)}{md_0^2}+\omega_{\mathrm{eff}}^2h(t)\Delta Z_{\mathrm{np}}(t) \tag*{(30)}

In order to further simplify the treatment of the mechanical displacement of linearly and parametrically driven oscillator’s test mass, let the angular size of the stationary capacitive electrodes α\alpha be equal exactly to π/4\pi/4, or 45 degree. Then one has

Sin⁡(2nα)=1+(−1)n+12\operatorname{Sin}(2n\alpha) = \frac{1 + (-1)^{n+1}}{2}

which is equal to that for n=1,3,5,…n = 1, 3, 5,\ldots, and equal to zero otherwise. This means that in Eq. (21) and Eq. (27), the effective overlap area modulation frequencies are 4Ω4\Omega, 8Ω8\Omega, 12Ω,…12\Omega,\ldots On the other hand the leading intermodulation sidebands in Eq. (22) and Eq. (30) are 3Ω3\Omega, 4Ω±ω4\Omega\pm\omega, 5Ω5\Omega, 7Ω,…7\Omega,\ldots In order to Eq. (29) hold true, none of the above modulation frequencies should coincide with the liner and parametric resonant conditions, which means

ωeff3<Ω<ωeff(31)\frac{\omega_{eff}}{3} < \Omega< \omega_{eff} \tag*{(31)}

and Ω≅3ωeff4\Omega\cong\frac{3\omega_{eff}}{4} is an obvious choice.

One can show that the lowest driving frequency 3Ω3\Omega in Eq. (30) under the above conditions is more than twice that of the resonant frequency ωeff\omega_{eff} so all the higher harmonics can be ignored due to the high roll-off rate ∼12\sim12 dB per octave. One can prove that a low enough RMS voltage across the APF sensing capacitor, the ΔZp(t)\Delta Z_p(t) correction in Eq. (29) can be reduced to a second order one, and that a trade-off can be established between the required sensitivity and the sensing capacitor design features. As was pointed out in § 2 above, the 3Ω3\Omega harmonic in Ctotal(t)C_{\mathrm{total}}(t) forms the basis for the IF demodulation of the terms proportional to GZXG_{ZX} and GZYG_{ZY} gravity gradient components. It is also worth noting that the 4Ω±ω4\Omega\pm\omega harmonic content in Ctotal(t)C_{\mathrm{total}}(t) can be used to demodulate the term proportional to the parasitic linear acceleration az(t)a_z(t) in Eq. (6) and use this channel to actively remove its potential leakage into the desired gravity gradient signal.

Analogue Front-End Interferometric Readout and Capacitance-to-Phase Conversion

The proposed instrument employs a coherent radio-frequency (RF) Mach–Zehnder-type interferometric readout. An oscillator disciplined by a precision oven-controlled crystal oscillator (OCXO) provides the reference for a direct digital synthesiser (DDS). A power splitter generates two mutually coherent signals. The sensing arm contains a modified resonant all-pass filter (APF), here designated KAPFK_{\mathrm{APF}}, whose phase response is controlled by the total sensing capacitance Ctotal(t)C_{\mathrm{total}}(t) in Eq. (22). The other arm serves as the RF reference. The arms are recombined so that their nominal carrier components cancel, leaving a residual signal sensitive to changes in the sensing-arm complex transfer function. The sensing capacitance is the single electrical quantity presented by the four stationary electrode positions connected to a common sensing rail; the companion rail is grounded. A rotating, electrically floating gating electrode modifies the equivalent capacitance. Thus the RF front end reads one time-dependent capacitance, Ctotal(t)C_{\mathrm{total}}(t), rather than four independent channels as depicted in Figure 5 below.

Simplified block diagram of the analogue front-end interferometric readout

Figure 5. A simplified block diagram of the analogue front-end interferometric readout showing the APF capacitive-to-phase interface included into a measurement arm of the Mach–Zehnder RF interferometer. The output of the latter goes straight into FPGA-based IF demodulation stage for further signal processing.

Let ωc\omega_c denote the RF carrier angular frequency. Let us write the KAPFK_{\mathrm{APF}} transfer function as a complex quantity H(ω,C)H(\omega,C) where a(ω,C)a(\omega,C) is its voltage-amplitude transmission and φ\varphi is its phase

H(ω,C)=a(ω,C)exp⁡[iφ(ω,C)](32)H(\omega,C) = a(\omega,C)\exp[i\varphi(\omega,C)] \tag*{(32)}

For a capacitance perturbation δC(t)\delta C(t) about the nominal value C0C_0, at fixed carrier frequency ωc\omega_c, the leading phase response is

δφ(t)≃KAPF δC(t),KAPF≡(∂φ∂C)(ωc,C0)(33)\delta\varphi(t) \simeq K_{\mathrm{APF}}\,\delta C(t), \qquad K_{\mathrm{APF}} \equiv\left(\frac{\partial\varphi}{\partial C}\right)_{(\omega_c,C_0)} \tag*{(33)}

Likewise, the amplitude perturbation is δa(t)≃Ka δC(t)\delta a(t) \simeq K_a\,\delta C(t), where Ka=(∂a/∂C)K_a = \left(\partial a/\partial C\right) evaluated at the same operating point. The full differential response is therefore

δHH0≃[Kaa0+iKAPF]δC(t)(34)\frac{\delta H}{H_0} \simeq\left[\frac{K_a}{a_0} + iK_{\mathrm{APF}}\right]\delta C(t) \tag*{(34)}

The phase coefficient KAPFK_{\mathrm{APF}} has units rad/pF and its sign depends on the adopted phase convention. It must be distinguished from the sensitivity of the complete interferometer, which also depends on carrier amplitude, arm balance, combiner response and post-combiner gain. Using complex RF phasors, let the sensing arm signal at the combiner be Es=es0exp⁡(iφs)E_s = e_{s0}\exp(i\varphi_s) and the reference-arm signal be Er=er0exp⁡(iφr)E_r = e_{r0}\exp(i\varphi_r). The combiner polarity is absorbed into the definition of the reference phasor.

The residual carrier is

EINT=Es+Er(35)E_{\mathrm{INT}} = E_s + E_r \tag*{(35)}

At the intended cancellation point, Es0≈Er0E_{\mathrm{s0}} \approx E_{\mathrm{r0}} and θ=ϕs−ϕr≈π\theta= \phi_{\mathrm{s}}-\phi_{\mathrm{r}} \approx\pi (mod 2π2\pi). Near this operating point, let us introduce an amplitude mismatch δa/a0\delta a/a_0 and a differential phase error δθ\delta\theta, with Es=E0(1+δa)exp⁡[i(π+θ)]E_{\mathrm{s}} = E_0(1+\delta a)\exp[i(\pi+\theta)] and Er=E0E_{\mathrm{r}} = E_0.

To first order,

EINT≃−E0(δa/a0+iδθ)E_{\mathrm{INT}} \simeq-E_0(\delta a/a_0+i\delta\theta)
∣EINT∣≃E0(δa/a0)2+δθ2(36)|E_{\mathrm{INT}}| \simeq E_0\sqrt{(\delta a/a_0)^2+\delta\theta^2} \tag*{(36)}
Image of the latest version of the experimentally implemented capacitance-to-phase interface along with a RF Mach-Zehnder architecture

Figure 6. An image of the latest version of experimentally implemented capacitance-to-phase interface along with a RF Mach-Zehnder architecture.

The result illustrates an important qualification: the magnitude of the nulled RF signal is not a globally signed linear phase detector. At perfect balance it varies as ∣θ∣|\theta| and a signed, linear response requires a defined phase-sensitive detection quadrature, an intentional calibrated offset, or another suitable coherent demodulation method. The RF residual is amplified before ADC acquisition, while coherent phase-sensitive processing can be implemented digitally.

For a selected, calibrated phase quadrature, let QϕQ_{\phi} represent the net conversion from differential phase (radians) to the demodulated voltage, including the RF chain and digital scaling. The small-signal voltage response is

δVQ(t)≃Qϕ δϕ(t)=KAPFQϕ δC(t)(37)\delta V_{\mathrm{Q}}(t) \simeq Q_{\phi}\,\delta\phi(t)=K_{\mathrm{APF}}Q_{\phi}\,\delta C(t) \tag*{(37)}

Eq.(38) defines the effective capacitance-to-voltage conversion coefficient KCV=KAPFQϕK_{\mathrm{CV}}=K_{\mathrm{APF}}Q_{\phi} of the complete measurement channel. It should be established experimentally, rather than inferred from KAPFK_{\mathrm{APF}} alone. An amplitude-sensitive quadrature may be used simultaneously to monitor imbalance and diagnose parasitic amplitude modulation. The measured KAPFK_{\mathrm{APF}} phase response, as depicted in Figure 7 below, exhibits a steep transition in the vicinity of its resonant operating region. The carrier can be placed at a point where the sensing-arm phase is nominally opposite to that of the reference arm, avoiding a separate conventional RF phase shifter when the remaining fixed path phase is compatible with cancellation. Fine

Experimental phase-versus-frequency response of the KAPF under two sensing-capacitance conditions

Figure 7. Experimental phase-versus-frequency response of the KAPFK_{\mathrm{APF}} under two sensing-capacitance conditions. The steep transition illustrates capacitance-to-phase conversion and the possibility of selecting a carrier frequency near the phase-opposition condition.

DDS frequency control provides a practical phase-balancing degree of freedom. The carrier frequency should not be chosen solely from the largest slope ∣∂ϕ/∂C∣|\partial\phi/\partial C|. The operating point must also provide adequate arm-amplitude matching, stable residual carrier, acceptable response linearity over the expected capacitance excursion, and tolerable conversion of carrier-frequency fluctuations into differential phase noise. Frequency tuning and amplitude balancing are therefore coupled calibration tasks.

A perturbation in the synthesiser frequency produces differential phase noise whenever the sensing and reference arms have unequal phase dispersion. For small fluctuations δω(t)\delta\omega(t), the phase error contains the term

δθ(t)≃KAPF δC(t)+Dω δω(t)(38)\delta\theta(t) \simeq K_{\mathrm{APF}}\,\delta C(t)+D_{\omega}\,\delta\omega(t) \tag*{(38)}
Dω≡∂θ∂ω∣ωc(39)D_{\omega}\equiv\left.\frac{\partial\theta}{\partial\omega}\right|_{\omega_{\mathrm{c}}} \tag*{(39)}

A common OCXO/DDS source suppresses phase noise that is strictly common to both paths, but it does not automatically cancel differential phase noise caused by unequal group delay, phase dispersion, or environmental changes. The relevant source-noise spectrum must be evaluated at offsets corresponding to the recovered modulation frequencies and with the actual interferometer transfer function. An ideal all-pass network has constant magnitude, but a practical resonant KAPF includes loss, finite component quality factors and parasitic coupling. Despite the very low experimental ripple level, shown in Figure 8, amplitude fluctuations can leak into the selected phase quadrature through imperfect I/Q orthogonality, imbalance or amplifier nonlinearity. These effects will be discussed in the quantitative noise and error budget analysis in Part II of this study.

Figure 8. Experimental amplitude-versus-frequency response under the corresponding capacitance conditions.

A practical calibration sequence can use a controlled capacitance perturbation, or a traceable equivalent RF test stimulus, to determine KCV at the chosen operating point. A coherent digital receiver can estimate the complex residual phasor, separate its in-phase and quadrature components, and adjust the DDS carrier frequency to minimise the phase mismatch. Amplitude balance may be achieved through a stable electronically controlled gain or attenuation element if required. Alternatively, its measured residual can be monitored and corrected within a validated linear range. A periodic, rather than continuously intrusive, calibration is preferable if long-term stability permits. The low-noise RF amplifier drives the ADC. Secondary demodulation and subsequent numerical differentiation are performed in the FPGA/digital domain, not by the analogue KAPF itself. The known deterministic overlap modulation is removed or rejected by synchronous processing before interpreting the remaining signal as gravity gradients related test-mass motion.

The estimator of the vertical spatial derivative of the GZZ gravity gradient component

As will be disclosed in the Part II of this study, the proposed GZZZ estimator is characterised by a triple modulation-demodulation frequency scheme because the RF carrier phase is modulated at both the rotation frequency and its higher-order harmonics (primarily the third and fourth harmonics) as the rotating gating electrode successively traverses the four stationary electrodes. By sampling the capacitive variation proportional signal at four distinct, 90-degree spaced angular positions, as depicted in Figure 4, the estimator is physically configured to measure the local gravity gradient components GZX and GZY at four orthogonal angular coordinate points during a single revolution. This spatial sampling is equivalent to establishing a local measurement time grid synchronised with the rotation frequency Ω\Omega

tn,k=2πk+πn2Ω(40)t_{n,k} = \frac{2\pi k + \frac{\pi n}{2}}{\Omega} \tag*{(40)}

where k=0,1,2,…k=0,1,2,\ldots is the integer revolution number and n=1,2,3,…n=1,2,3,\ldots represents the specific electrode index corresponding to the locations of stationary electrodes. This orthogonal time sampling allows the estimator to calculate a linear combination of the gravity gradient components GZX and GZY proportional to the vertical derivative of the vertical gravity gradient component GZZ according to (2). One can show, this vertical derivative can be mathematically approximated by a finite difference equation [12]

GZX(tn,k)−GZX(tn+2,k)+GZY(tn+1,k)−GZY(tn+3,k)≃−2r0(GZXX+GZYY)=2r0GZZZ(41)G_{ZX}(t_{n,k}) - G_{ZX}(t_{n+2,k}) + G_{ZY}(t_{n+1,k}) - G_{ZY}(t_{n+3,k}) \simeq-2r_0(G_{ZXX}+G_{ZYY}) = 2r_0G_{ZZZ} \tag*{(41)}

To perform this calculation in real-time, the measurement system can use a precision optical encoder providing a high-resolution timing trigger, allowing the downstream processing electronics to sample the measurement signal precisely at the four sequential alignment instants tn,kt_{n,k}, tn+1,kt_{n+1,k}, tn+2,kt_{n+2,k} and tn+3,kt_{n+3,k}. Also, it is assumed that the time interval Δtk=2π/Ω\Delta t_k = 2\pi/\Omega is much smaller compared to potentially present time variations of the rotation angular frequency Ω\Omega, which can be treated as a constant quantity during a single revolution. This assumption, of course, depends on choosing the optimum mechanical oscillator design.

Discussions

The proposed rotating capacitive architecture provides a theoretical route to the third-order gravity gradient GZZZ using one mechanically sensitive axis, four stationary capacitive electrode pairs and a rotating floating gate. Its defining feature is that information about the spatial derivatives of GZX and GZY is obtained from measurements at orthogonal locations on a circular path. These derivatives are related to GZZZ by the differentiated Laplace identity. The method therefore differs conceptually from instruments that directly compare vertically separated accelerometers to determine GZZ. The fourfold symmetry of the electrode configuration provides a useful frequency-domain structure. The leading geometrical overlap modulation occurs at 4Ω4\Omega, and its product with the oscillator response at Ω\Omega generates a 3Ω3\Omega electrical component. For high enough Ω\Omega this permits synchronous signal extraction away from the lowest-frequency readout band where most of linear and angular disturbances may occur. A principal potential benefit of the proposed rotating capacitive system is suppression of certain inertial terms. For a rigid rotor and ideal axial alignment, the dominant horizontal centrifugal and Coriolis terms have no direct axial projection, and spatial differentiation eliminates position-independent angular-velocity-squared contributions.

The interferometer architecture combines high-slope resonant phase transduction with coherent carrier suppression. The key physical advantage is that a small capacitance perturbation can be expressed as a measurable differential RF phase variation, and ultimately as a calibrated voltage signal suitable for digital processing. The existing hardware and historical phase/amplitude measurements support the feasibility of the conversion principle, while not by themselves establishing the ultimate gravity-gradient resolution. The cancellation point, frequency-to-phase coupling, residual amplitude response, coherent detection quadrature and stability of the signal measurement and processing chain are essential system-level considerations. The practically measured values, together with amplifier and ADC noise, calibration uncertainty and practical operating parameters, are reserved for Part II. This separation keeps the present chapter focused on the readout principle and its connection to the instrument’s theoretical measurement chain.

The present analysis is intentionally restricted to the measurement principle. It does not establish a numerical GZZZG_{ZZZ} sensitivity or demonstrate operation on a drone, AUV or other moving platform. Part II will address the detailed demodulation and reconstruction chain, oscillator Brownian noise, capacitive-to-phase and voltage noise, ADC and digital-processing contributions, finite measurement-time variance, rotation-frequency optimization and systematic motion-induced error budget. Experimental validation will ultimately be required to determine whether the theoretical benefits translate into a practical instrument. In summary, the rotating capacitive system offers a compact and physically distinct approach to higher-order gravity gradiometry. The central theoretical results are the exact fourfold overlap modulation, its electrical sideband structure, and the spatial finite-difference relation to GZZZG_{ZZZ}. Their practical realization depends on controlled electromechanical coupling, a calibrated reconstruction of GZXG_{ZX} and GZYG_{ZY}, and rigorous treatment of sequential sampling and platform dynamics.

The measurement of the third-order vertical gravity-gradient component GZZZG_{ZZZ} may be particularly valuable in applications requiring discrimination of localized subsurface density contrasts. Potential targets include shallow tunnels, cavities, buried infrastructure, and other compact structures whose gravity fields exhibit appreciable spatial curvature. Because higher-order derivatives emphasize shorter spatial scales, the observable may help distinguish relatively localized anomalies from slowly varying regional gravity backgrounds. This potential advantage is accompanied by more rapid attenuation of the signal with increasing source distance; consequently, shallow targets and close-range surveys are likely to offer the most favorable conditions. Further plausible applications include near-surface geological and engineering investigations, mineral exploration, and selected geothermal surveys where faults, contacts, or other density boundaries produce sufficiently strong and spatially localized signatures. In these settings, GZZZG_{ZZZ} would most naturally complement conventional gravity and second-order gravity-gradient observations rather than replace them. The practical value of the additional spatial information will depend on target geometry and depth, survey altitude, geological background variability, instrument sensitivity, and the ability to control or correct platform-motion disturbances.

Acknowledgements

One of the authors (Veryaskin) would like to thank Dan DiFrancesco of Niagara Gravity Consulting LLC for his valuable comments and a reference provided.

References

  1. [1]Frank J. van Kann and Alexey V. Veryaskin, A novel capacitive sensor interface based on a simple capacitance-to-phase converter, 2024 Meas. Sci. Technol. 35 115101DOI
  2. [2]Alexey V. Veryaskin, Gravity, Magnetic and Electromagnetic Gradiometry: Strategic Technologies in the 21st Century. Institute of Physics Publishing (IOP, UK) Second Edition, 2021, 190p.
  3. [3]Dransfield M. and Christensen A. N., Performance of airborne gravity gradiometers. The Leading Edge 32(8), 908-922, 2013DOI
  4. [4]Murphy C.A., The Air-FTG airborne gravity gradiometer system. Airborne Gravity 2004, Abstracts from the ASRG-PESA Airborne Gravity Workshop, p.7, 2004
  5. [5]Snadden M.J. et al, Measurement of the Earth's Gravity Gradient with an Atom Interferometer-Based Gradiometer. Physical Review Letters, 81(5), 971, 1998
  6. [6]Thomas R.J., Legge S. and Close J.D., Tensor gravity gradiometry with a single-axis atom gradiometer. Physical Review A, 112(5), 1-9, 2025DOI
  7. [8]Veryaskin A.V., A novel combined gravity and magnetic gradiometer system for mobile applications. SEG 2000 Expanded Abstracts, Calgary, Canada, 2000, 4p.DOI
  8. [9]Veryaskin, A.V., Bourhill J., Ivanov E.N. and Tobar M.E. Intrinsic gravity gradiometer and gravity gradiometry, US Patent 11,002,878, 2021 B2, May 11, 2021 (granted)
  9. [10]Bjørnar Dale, David Becker, Tim Enßlberger Jensen, and René Forsberg, DroneSOM: Towards a Drone-Based Gravity Observation System. International Symposium on Gravity, Geoid and Height Systems 2024, International Association of Geodesy Symposia, vol 158. Springer, 2025DOI
  10. [11]C. Li, Z. Liu, H. Zhou et al., Feasibility study of gravity gradient measurement based on multi-rotor UAV platform. Geodesy and Geodynamics, https://doi.org/10.1016/j.geog.2026.01.007
  11. [12]Alexey Veryaskin and Wayne McRae, On Combined Gravity Gradient Components Modelling for Applied Geophysics. Journal of Geophysics and Engineering, Vol.5, 2008, pp. 348-356DOI
  12. [13]Alexey V. Veryaskin , Khyl J. McMahon and Howard C. Golden, Apparatus and Method for Measuring Gravity Gradients. Australian Patent Application 2026907747, MBIB, 11 September 2026
  13. [14]DiFrancesco D., et al., Gravity gradiometer systems – advances and challenges. Geophysical Prospecting, 57(4), 615–623, 2009
  14. [16]D’Alessandro A., Scudero S. and Vitale G., A Review of the Capacitive MEMS for Seismology. Sensors, 19, 3093, 2019DOI
  15. [17]Nastro A., Electrostatic-Capacitive MEMS Force Sensors: A State-of-the-Art Review. Journal of Microelectromechanical Systems, 35(1),196-205, 2025DOI
  16. [18]Chobotov V. A., Radially vibrating, rotating gravitational gradient sensor. Journal of Spacecraft and Rockets, Vol.5(4), 434-441, 1968DOI

Paper details

vlnc:W_vhsernn8
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents