GridProbe: Passive Gating & Inversion Probing

• GridProbe is a methodology that uses structured grid perturbations to reveal otherwise inaccessible transport, field, or topological parameters.
• It enhances ion back-flow suppression and electron transmission in TPCs by employing a passive Bi-Polar Grid that avoids dead time and gain fluctuations.
• Its applications extend to proton deflectometry and inverter probing for power grids, offering precise spatial resolution and topology recovery through differential readouts.

In the supplied arXiv literature, GridProbe denotes, or is used as a closely related label for, several probe-mediated diagnostic strategies in which a grid, mesh, or controlled perturbation is used to recover otherwise inaccessible transport, field, or topology information. Its most explicit detector interpretation is a passive Bi-Polar Grid (BPG) used as a gating grid for a TPC: a wire grid biased as $(V_g \pm \Delta g)$ that is always opaque to ions but, in the presence of an external magnetic field, can remain transparent enough to electrons, thereby targeting ion back-flow (IBF) suppression without the dead time of active gating (Zakharov et al., 2023). In the same supplied corpus, analogous probe-centric constructions appear in proton deflectometry with an in situ X-ray fiducial image of the same mesh, in inverter probing for electric distribution topology inference, and in several related platforms for spectroscopy, mesoscopic transport, pixel-TPC readout, and large-area electrode quality assurance (Malko et al., 2021, Cavraro et al., 2018).

1. Passive Bi-Polar Grid as a TPC gating concept

In the TPC context, GridProbe is the use of a passive Bi-Polar Grid as a gating element that is permanently present, rather than dynamically opened and closed (Zakharov et al., 2023). Neighboring wires are held at alternating potentials, written as $(V_g \pm \Delta g)$. The purpose is to obtain strong $\vec{E}$-field ratios that block ions while avoiding the dead-time penalty of active gating. The supplied description contrasts four approaches. GEMs and Micromegas reduce IBF through geometry and electric-field ratios, but IBF suppression depends on field settings and geometry and can introduce gain fluctuations; even then, IBF can remain a major source of space charge. An active BPG can block all IBF ions, but its slow drift speed creates too much dead time. The passive BPG is intended to combine zero ion transparency, high electron transparency in a magnetic field, no dead time, and no gain fluctuations from gating action.

The electron and ion transport is formulated through the Langevin equation,

$$m\frac{d\vec{v}}{dt} = q\vec{E} +q(\vec{v} \times \vec{B}) -\kappa\vec{v},$$

together with the charge-velocity relation in combined $\vec{E}$ and $\vec{B}$ fields,

$$\vec{v} = \frac{\mu}{1+\omega^2\tau^2} \left(\vec{E} +\frac{\vec{E} \times \vec{B}}{|\vec{B}|}\omega \tau +\dfrac{(\vec{E} \cdot \vec{B})\vec{B}}{B^2}\omega^2\tau^2 \right).$$

For $\vec{E}=(E\cos{\alpha},0,E\sin{\alpha})$ with $\vec{B}=(0,0,B)$, the supplied form is

$$v_x = c(E \cos{\alpha}) \qquad v_y = c(E \omega \tau \cos{\alpha}) \qquad v_z = \mu E \sin{\alpha},$$

whereas imperfect wire alignment introduces an $(V_g \pm \Delta g)$0 component and yields

$(V_g \pm \Delta g)$1

with

$(V_g \pm \Delta g)$2

A central point is that the grid does not simply transmit electrons. Because electrons are light and lie in the regime $(V_g \pm \Delta g)$3, their motion is dominated by the magnetic field rather than by the local electric field of the BPG. Electrons coming from above a wire receive a first Lorentz-force push along the wire direction, then a second weaker push that helps them move past the wire structure. Their paths are therefore shifted sideways, pinched toward the wire centers, and distorted out of plane. The distortion is symmetric along the wires, making the alignment of the wires relative to the readout pads a specific and previously unstudied resolution question.

2. IBF figure of merit, alignment sensitivity, and experimental studies

The passive-BPG analysis is organized around a figure of merit that balances ion suppression against electron transmission rather than treating either in isolation (Zakharov et al., 2023):

$(V_g \pm \Delta g)$4

Here $(V_g \pm \Delta g)$5 and $(V_g \pm \Delta g)$6 are ion transparencies through the mesh and grid, $(V_g \pm \Delta g)$7 is the electron transparency through the grid, and $(V_g \pm \Delta g)$8 is the field ratio. The supplied discussion emphasizes the trade-off: increasing field ratio can help manipulate ion and electron transport, but if $(V_g \pm \Delta g)$9, additional gain is needed to preserve signal, and that additional gain can itself increase IBF.

The experimental IBF program at the Weizmann Institute of Science (WIS) used a detector with MWPC gain wires, a strong $\vec{E}$0Fe X-ray source, picoammeters to detect small current changes, and a mesh providing a uniform drift field and a controllable region for field-ratio studies. Ion and electron transparencies were measured as functions of magnetic field and voltage offset $\vec{E}$1 on the linear BPG. The supplied summary notes two coupled effects: higher grid ratios decrease mesh ratios and can pull more ions through the mesh; and if the grid lowers electron transparency, compensating gain increases can create more IBF. The completed WIS analysis included ion transparency, electron transparency, and the FoM behavior for different field settings, and the stated conclusion is qualitative but explicit: the passive BPG demonstrates ion blocking with high electron transparency, and the completed data support the effectiveness of the passive gating concept.

The same paper begins a separate spatial-resolution study. A BPG with 2 linear and 2 radial wire configurations was placed above a pad plane in a prototype TPC. The test used zig-zag pads to improve charge sharing and spatial precision, cosmic rays to measure track position resolution, and a prototype TPC at Argonne National Lab (ANL) inside a 3 T magnetic field. The motivation was not only to quantify a global resolution loss, but to determine whether the periodic distortion imposed by the wires could be understood and perhaps reduced through wire-pad geometry.

The alignment issue is structurally important because the BPG-induced distortion is periodic and wire-locked. In a misaligned case, if the grid position varies relative to the pads, the distortions may average out over the track. In an aligned case, the distortion pattern and pad-response nonuniformities may interact in either direction: the interaction could improve the effective correction or worsen the resolution if the effects reinforce each other. The supplied description identifies the resulting error as a form of Differential Non-Linearity (DNL) that repeats with the wire periodicity. The authors suggest it may be correctable, but they also state the main physics tension directly: stronger IBF suppression requires a grid that is more opaque to ions and suitably biased, yet that same grid inevitably distorts electron trajectories and can harm spatial resolution.

3. Proton-deflectometry GridProbe with an in situ X-ray fiducial

A second explicit use of GridProbe in the supplied corpus is a proton-deflectometry platform that records an in situ X-ray image of the same mesh/grid used to form proton beamlets (Malko et al., 2021). The experimental aim is to measure non-uniform magnetic fields in expanding plasmas. Instead of inferring undeflected beamlet positions from a separate null shot or from apparently unperturbed regions, the diagnostic records a simultaneous X-ray reference of the mesh through the same geometry. That X-ray image provides the undeformed beamlet centers required for magnetic-field inference.

The underlying deflection relation is

$\vec{E}$2

with $\vec{E}$3. For the experiments described, electric fields are neglected, so the deflection is essentially magnetic. Using a point-like proton source and source-to-object and object-to-detector distances $\vec{E}$4 and $\vec{E}$5, the deflection angle is related to the difference between the deflected proton position $\vec{E}$6 and the undeflected reference position $\vec{E}$7:

$\vec{E}$8

which gives the path-integrated magnetic field as

$\vec{E}$9

The measurement problem is therefore dominated by the localization of $$m\frac{d\vec{v}}{dt} = q\vec{E} +q(\vec{v} \times \vec{B}) -\kappa\vec{v},$$0 and $$m\frac{d\vec{v}}{dt} = q\vec{E} +q(\vec{v} \times \vec{B}) -\kappa\vec{v},$$1, making image contrast a first-order concern.

At OMEGA, the platform used D$$m\frac{d\vec{v}}{dt} = q\vec{E} +q(\vec{v} \times \vec{B}) -\kappa\vec{v},$$2He exploding pusher capsule implosions as a source of 3 MeV protons, 14.7 MeV protons, and continuum X-rays. The capsule was about 10 mm from target chamber center, the detector stack about 154 mm on the opposite side, and a Ni mesh was placed 4 mm from the backlighter source. The mesh had 125 µm pitch, 90 µm opening, and 35 µm bar thickness. The detector stack included CR-39 for 3 MeV protons, CR-39 for 14.7 MeV protons, and an image plate (IP) for X-rays, with 15 µm Ta and 200 µm Al filters positioned to place proton energies at appropriate Bragg peaks. For a vacuum magnetic-field measurement, the new CR-39/IP method achieved an average error of about 1.2 T·mm, compared with 5.7 T·mm for a conventional two-shot CR-39/CR-39 method, corresponding to about a factor of 4 improvement in accuracy.

At OMEGA EP, the same concept was implemented with TNSA protons. A 1 ps, 50 J laser pulse was focused onto a 20 µm Au foil, producing broadband protons up to about 25–30 MeV and bremsstrahlung X-rays from hot electrons. The mesh was changed to an Au mesh with 340 µm pitch, 285 µm opening, and 55 µm bar thickness. The detector stack used 12 layers of HD-V2 RCF, 12 Al filters, and an MS-type image plate at the back, designed to detect protons from about 5.5 to 32.5 MeV. The paper reports that the IP image clearly showed the mesh structure, demonstrating that the X-ray reference can still be recorded even in the TNSA configuration.

Configuration	Source and mesh	Reported result
OMEGA	D$$m\frac{d\vec{v}}{dt} = q\vec{E} +q(\vec{v} \times \vec{B}) -\kappa\vec{v},$$3He capsule; Ni mesh, 125 µm pitch	Average error about 1.2 T·mm versus 5.7 T·mm for CR-39/CR-39
OMEGA EP	TNSA protons from 20 µm Au foil; Au mesh, 340 µm pitch	IP image clearly shows the mesh structure

The paper places particular weight on contrast and blurring. At OMEGA, raw contrast was modest: about 7 for the 3 MeV proton image, about 8 for the 15 MeV proton image, and about 2 for the X-ray image; after post-processing these became about 42, 60, and 19, respectively. FLUKA studies compared Ni and Au meshes at large, medium, and small scales. For the large mesh, proton contrast improved strongly and X-ray contrast improved especially for high-$$m\frac{d\vec{v}}{dt} = q\vec{E} +q(\vec{v} \times \vec{B}) -\kappa\vec{v},$$4 material: the supplied summary states a 4–6× enhancement in contrast using a high-Z mesh with larger grid spacing, which would lead to a further factor of two improvement in magnetic-field accuracy. The analytic X-ray model,

$$m\frac{d\vec{v}}{dt} = q\vec{E} +q(\vec{v} \times \vec{B}) -\kappa\vec{v},$$5

with contrast ratio

$$m\frac{d\vec{v}}{dt} = q\vec{E} +q(\vec{v} \times \vec{B}) -\kappa\vec{v},$$6

reaches $$m\frac{d\vec{v}}{dt} = q\vec{E} +q(\vec{v} \times \vec{B}) -\kappa\vec{v},$$7 for Ni and $$m\frac{d\vec{v}}{dt} = q\vec{E} +q(\vec{v} \times \vec{B}) -\kappa\vec{v},$$8–$$m\frac{d\vec{v}}{dt} = q\vec{E} +q(\vec{v} \times \vec{B}) -\kappa\vec{v},$$9 for Au, depending on IP type.

4. Inverter probing for electric distribution topology processing

A different but formally related use of GridProbe in the supplied material is inverter probing for distribution network topology inference (Cavraro et al., 2018). The central idea is to use smart inverters as controllable actuators. Instead of relying on passive voltage and load variability, the operator intentionally perturbs inverter injections and observes the resulting instantaneous voltage deviations. Under an approximate radial-grid model, these perturbation-response pairs encode the feeder structure through a matrix $\vec{E}$0, whose inverse is a weighted tree Laplacian.

The network is modeled as a rooted tree with non-substation voltage vector $\vec{E}$1, active and reactive injections $\vec{E}$2, and the Linearized Distribution Flow approximation

$\vec{E}$3

together with

$\vec{E}$4

The entries of $\vec{E}$5 have the tree interpretation

$\vec{E}$6

so each entry equals the total resistance common to the paths from the root to buses $\vec{E}$7 and $\vec{E}$8.

If $\vec{E}$9 denotes the buses with controllable inverters and $\vec{B}$0 the perturbation vector at time $\vec{B}$1, the probing model is

$\vec{B}$2

where $\vec{B}$3. Stacking $\vec{B}$4 probing actions gives

$\vec{B}$5

The paper recommends an asynchronous “up/down” pattern,

$\vec{B}$6

so each inverter first changes and then reverts its injection.

The main structural identity is

$\vec{B}$7

which makes $\vec{B}$8 a reduced weighted Laplacian of the radial graph. Recovering $\vec{B}$9 therefore recovers both the support of the tree and the line resistances. The paper studies two tasks. The topology recovery problem estimates $$\vec{v} = \frac{\mu}{1+\omega^2\tau^2} \left(\vec{E} +\frac{\vec{E} \times \vec{B}}{|\vec{B}|}\omega \tau +\dfrac{(\vec{E} \cdot \vec{B})\vec{B}}{B^2}\omega^2\tau^2 \right).$$0 from

$$\vec{v} = \frac{\mu}{1+\omega^2\tau^2} \left(\vec{E} +\frac{\vec{E} \times \vec{B}}{|\vec{B}|}\omega \tau +\dfrac{(\vec{E} \cdot \vec{B})\vec{B}}{B^2}\omega^2\tau^2 \right).$$1

while line status verification introduces a binary vector $$\vec{v} = \frac{\mu}{1+\omega^2\tau^2} \left(\vec{E} +\frac{\vec{E} \times \vec{B}}{|\vec{B}|}\omega \tau +\dfrac{(\vec{E} \cdot \vec{B})\vec{B}}{B^2}\omega^2\tau^2 \right).$$2 with

$$\vec{v} = \frac{\mu}{1+\omega^2\tau^2} \left(\vec{E} +\frac{\vec{E} \times \vec{B}}{|\vec{B}|}\omega \tau +\dfrac{(\vec{E} \cdot \vec{B})\vec{B}}{B^2}\omega^2\tau^2 \right).$$3

and solves a binary detection problem over energized versus open lines.

A key theorem in the paper states that, with noiseless probing data and full-rank $$\vec{v} = \frac{\mu}{1+\omega^2\tau^2} \left(\vec{E} +\frac{\vec{E} \times \vec{B}}{|\vec{B}|}\omega \tau +\dfrac{(\vec{E} \cdot \vec{B})\vec{B}}{B^2}\omega^2\tau^2 \right).$$4, the topology is identifiable if the grid is probed at all leaf nodes and voltage data are collected at all buses. The proof uses the level sets of leaf nodes and shows that the combination of the leaf set and all leaf level sets uniquely determines a tree. This is an exact identifiability statement under strong observability assumptions; the same paper also makes clear that practical recovery in noise uses convex surrogates rather than a direct solution of the original non-convex problems.

The non-convex programs are relaxed using an $$\vec{v} = \frac{\mu}{1+\omega^2\tau^2} \left(\vec{E} +\frac{\vec{E} \times \vec{B}}{|\vec{B}|}\omega \tau +\dfrac{(\vec{E} \cdot \vec{B})\vec{B}}{B^2}\omega^2\tau^2 \right).$$5-type surrogate and a $$\vec{v} = \frac{\mu}{1+\omega^2\tau^2} \left(\vec{E} +\frac{\vec{E} \times \vec{B}}{|\vec{B}|}\omega \tau +\dfrac{(\vec{E} \cdot \vec{B})\vec{B}}{B^2}\omega^2\tau^2 \right).$$6 barrier. For topology recovery, the convex surrogate is

$$\vec{v} = \frac{\mu}{1+\omega^2\tau^2} \left(\vec{E} +\frac{\vec{E} \times \vec{B}}{|\vec{B}|}\omega \tau +\dfrac{(\vec{E} \cdot \vec{B})\vec{B}}{B^2}\omega^2\tau^2 \right).$$7

with

$$\vec{v} = \frac{\mu}{1+\omega^2\tau^2} \left(\vec{E} +\frac{\vec{E} \times \vec{B}}{|\vec{B}|}\omega \tau +\dfrac{(\vec{E} \cdot \vec{B})\vec{B}}{B^2}\omega^2\tau^2 \right).$$8

This problem is solved by ADMM with closed-form updates, including a proximal log-det eigendecomposition step. The relaxed verification problem is solved by projected gradient descent over a convex hull of binary line-status variables.

The numerical results use the IEEE 13-bus and IEEE 37-bus feeders. On the IEEE 37-bus system, 200 Monte Carlo trials with relative measurement noise 0.01% produced average line-status errors reported as 5.07, 3.92, 3.73, and 2.69 for the identification task at $$\vec{v} = \frac{\mu}{1+\omega^2\tau^2} \left(\vec{E} +\frac{\vec{E} \times \vec{B}}{|\vec{B}|}\omega \tau +\dfrac{(\vec{E} \cdot \vec{B})\vec{B}}{B^2}\omega^2\tau^2 \right).$$9, and 0.32, 0.21, 0.08, and 0.01 for verification. The supplied discussion emphasizes that verification is much easier than full identification and that, with probing at about 40% of the nodes, line-status error probabilities can be driven to around $\vec{E}=(E\cos{\alpha},0,E\sin{\alpha})$0 to $\vec{E}=(E\cos{\alpha},0,E\sin{\alpha})$1. The same description states that topology recovery typically needs about 16 to 160 one-second probing actions, depending on the case.

5. Related probe-centric platforms and adjacent usages

Several additional papers in the supplied corpus are not titled GridProbe, but are explicitly framed there as GridProbe-like or GridProbe-style systems. They extend the same operational pattern—localized actuation or sensing, differential readout, and inference of hidden structure—across cryogenic spectroscopy, mesoscopic transport, gaseous pixel TPCs, and electrode metrology (Das et al., 2019, Jura et al., 2010, Ligtenberg et al., 2020, Deisting et al., 14 Nov 2025).

Platform	Core instrumentation	Salient reported capability
Modular PCS probe	JANIS He$\vec{E}=(E\cos{\alpha},0,E\sin{\alpha})$2 cryostat, 7 T magnet, attocube ANPx101 / ANPz101	Sample space about 347 mK; more than 48 hours at lowest temperature
SGM of a 2DEG	QPC injector and movable biased tip	Non-equilibrium region within $\vec{E}=(E\cos{\alpha},0,E\sin{\alpha})$3 of injection point
GridPix detector quad	Four Timepix3-based GridPix chips	Setup resolution 41 $\vec{E}=(E\cos{\alpha},0,E\sin{\alpha})$4; total systematic error 24 $\vec{E}=(E\cos{\alpha},0,E\sin{\alpha})$5
GRANITE	Gantry robot on $\vec{E}=(E\cos{\alpha},0,E\sin{\alpha})$6 granite table	$\vec{E}=(E\cos{\alpha},0,E\sin{\alpha})$7 relative electrostatic precision; $\vec{E}=(E\cos{\alpha},0,E\sin{\alpha})$8 corrected absolute precision

The modular point-contact spectroscopy probe is a detachable, plug-n-play system for sub-Kelvin operation in a JANIS He$\vec{E}=(E\cos{\alpha},0,E\sin{\alpha})$9 cryostat with a 7 T superconducting magnet (Das et al., 2019). Its architecture comprises a load-lock chamber, a vertical manipulator of non-magnetic SS-316, a probe head, and a sample space can thermally anchored to the He$\vec{B}=(0,0,B)$0 pot. Exchange of the sample/tip rig is performed without breaking the inner vacuum insulation, using differential pumping and gas trapping between two Viton O-rings. The cooling stages reach a He$\vec{B}=(0,0,B)$1 pot base temperature of about 292 mK and a sample space temperature of about 347 mK, maintained for more than 48 hours if the 1 K pot is kept below 2 K. The probe uses a piezo-driven three-stage coarse positioner for x-y-z motion, employs lock-in modulation for PCS, includes a 50 $\vec{B}=(0,0,B)$2 sample-stage heater and a RuO$\vec{B}=(0,0,B)$3 sensor, and was validated on PdTe$\vec{B}=(0,0,B)$4/Ag, Zr/PtIr, and Sn/Ag, including a BTK fit giving $\vec{B}=(0,0,B)$5 for Zr/PtIr.

The scanning-gate microscopy study of electron-electron scattering in a 2DEG uses a split-gate QPC to inject a narrow beam into a GaAs/AlGaAs 2DEG 100 nm below the surface, while a negatively biased metallic tip held about 30 nm above the surface acts as a movable local scatterer (Jura et al., 2010). The measured quantity is $\vec{B}=(0,0,B)$6. At zero dc bias, $\vec{B}=(0,0,B)$7 images current flow. At finite bias, low injection energies yield a fading negative signal consistent with predominantly small-angle scattering, with angular scale

$\vec{B}=(0,0,B)$8

At larger $\vec{B}=(0,0,B)$9, the surprising regime $$v_x = c(E \cos{\alpha}) \qquad v_y = c(E \omega \tau \cos{\alpha}) \qquad v_z = \mu E \sin{\alpha},$$0 appears. The interpretation is a highly non-equilibrium region near the QPC that modifies subsequent scattering. The spatial extent inferred from the data is within about $$v_x = c(E \cos{\alpha}) \qquad v_y = c(E \omega \tau \cos{\alpha}) \qquad v_z = \mu E \sin{\alpha},$$1 of the injection point, with a more quantitative bound $$v_x = c(E \cos{\alpha}) \qquad v_y = c(E \omega \tau \cos{\alpha}) \qquad v_z = \mu E \sin{\alpha},$$2. The model approximates this region by an effective temperature $$v_x = c(E \cos{\alpha}) \qquad v_y = c(E \omega \tau \cos{\alpha}) \qquad v_z = \mu E \sin{\alpha},$$3, with fitted slopes $$v_x = c(E \cos{\alpha}) \qquad v_y = c(E \omega \tau \cos{\alpha}) \qquad v_z = \mu E \sin{\alpha},$$4 and $$v_x = c(E \cos{\alpha}) \qquad v_y = c(E \omega \tau \cos{\alpha}) \qquad v_z = \mu E \sin{\alpha},$$5 for $$v_x = c(E \cos{\alpha}) \qquad v_y = c(E \omega \tau \cos{\alpha}) \qquad v_z = \mu E \sin{\alpha},$$6, depending on the scattering model.

The GridPix detector quad is a modular gaseous pixel readout unit intended as a building block for a large TPC readout plane (Ligtenberg et al., 2020). A single GridPix combines a Timepix3 CMOS pixel chip with an integrated amplification grid supported by 50 $$v_x = c(E \cos{\alpha}) \qquad v_y = c(E \omega \tau \cos{\alpha}) \qquad v_z = \mu E \sin{\alpha},$$7 SU8 pillars; the grid is 1 $$v_x = c(E \cos{\alpha}) \qquad v_y = c(E \omega \tau \cos{\alpha}) \qquad v_z = \mu E \sin{\alpha},$$8 thick aluminum with 35 $$v_x = c(E \cos{\alpha}) \qquad v_y = c(E \omega \tau \cos{\alpha}) \qquad v_z = \mu E \sin{\alpha},$$9 holes aligned to the pixel input pads, and the chip is protected by a 4 $(V_g \pm \Delta g)$00 silicon-rich silicon nitride layer. The quad combines four such chips on a common cooled base plate, has external dimensions $(V_g \pm \Delta g)$01, and an active surface coverage of 68.9%. Tested in a small TPC at ELSA in Bonn with 2.5 GeV electrons, it operated in the diffusion-dominated regime,

$(V_g \pm \Delta g)$02

with time-walk correction

$(V_g \pm \Delta g)$03

After distortion correction, the systematics in the pixel plane were better than 13 $(V_g \pm \Delta g)$04 over the full plane and 9 $(V_g \pm \Delta g)$05 in the central fiducial region. The setup resolution was 41 $(V_g \pm \Delta g)$06, and the total systematic error of the quad detector was 24 $(V_g \pm \Delta g)$07.

The GRANITE platform, short for Granular Robotic Assay for Novel Integrated TPC Electrodes, is a fully automated metrology and inspection system for large-area dual-phase xenon TPC electrodes (Deisting et al., 14 Nov 2025). It is built around a gantry robot over a $(V_g \pm \Delta g)$08 granite table and carries a confocal microscope, a high-resolution industrial camera with a telecentric lens, a laser distance sensor, and a profile laser scanner. Wire tension is derived from the fundamental resonance frequency,

$(V_g \pm \Delta g)$09

while sagging is modeled by a parabolic approximation with gravitational and electrostatic components,

$(V_g \pm \Delta g)$10

The platform achieves $(V_g \pm \Delta g)$11 precision for relative electrostatic displacement, $(V_g \pm \Delta g)$12 capability for absolute sag measurement, and $(V_g \pm \Delta g)$13 precision after model-based correction. It also scanned the XENON1T cathode—about 124 parallel wires, 7.5 mm pitch, 95 cm diameter, 216 $(V_g \pm \Delta g)$14 wire diameter—and used an undercomplete convolutional autoencoder with 6299 trainable parameters and MSE-based thresholds $(V_g \pm \Delta g)$15, $(V_g \pm \Delta g)$16, and $(V_g \pm \Delta g)$17 to classify anomalous wire images.

This collection suggests that, within the supplied corpus, “GridProbe” functions less as a single device class than as an organizing label for experimentally controlled probing schemes. In each case, the instrumentation is designed so that a localized perturbation or structured reference converts an inverse problem into a more constrained estimation problem.

6. Common design logic, limits, and recurrent misconceptions

Across the supplied literature, the common logic is a known perturbation or fiducial plus differential readout plus model-based inversion. In the passive-BPG TPC case, the controlled structure is the alternating-potential wire grid; in proton deflectometry, it is the mesh plus simultaneous X-ray reference; in inverter probing, it is the injected inverter perturbation; in PCS, it is the mechanically controlled tip-sample contact; in SGM, it is the movable tip-induced backscatterer; in GridPix, it is the integrated amplification grid and per-pixel timing; in GRANITE, it is a calibrated non-contact scan under gravity and high voltage. This suggests a family resemblance grounded in experimentally imposed structure rather than in a single hardware lineage.

The dominant limitations are also structurally similar, although they differ in mechanism. The passive BPG confronts a direct IBF suppression versus spatial-resolution trade-off: electron transparency can be high, but trajectory distortion, alignment sensitivity, and DNL remain intrinsic issues (Zakharov et al., 2023). Proton-deflectometry GridProbe is limited not by the inversion formula itself, but by the precision with which beamlet centers are found, so contrast and blurring dominate the achievable field accuracy (Malko et al., 2021). Inverter probing offers exact recovery only under specific conditions—all leaf nodes probed and voltage data collected at all buses—and practical recovery depends on the accuracy of the linearized model and on convex relaxation quality (Cavraro et al., 2018). The modular PCS probe depends on careful thermal engineering, vacuum handling, manipulator alignment, and mitigation of radiative heat leak (Das et al., 2019). GRANITE attains its best sag metrology through relative measurements taken close in time, because gantry-induced offsets largely cancel there, whereas absolute measurements require model-based corrections (Deisting et al., 14 Nov 2025).

Several misconceptions are explicitly contradicted in the supplied descriptions. The passive BPG does not merely “let electrons through”; it changes electron trajectories in a wire-periodic way (Zakharov et al., 2023). In SGM, a tip that backscatters electrons does not always reduce conductance; at high injection energy it can produce $(V_g \pm \Delta g)$18 because the tip perturbs a localized non-equilibrium electron distribution near the injector (Jura et al., 2010). In the XENON1T optical-inspection study, bright or anomalous wire features are not established to be the cause of field emission, and optical inspection alone cannot determine which anomalies are harmful (Deisting et al., 14 Nov 2025). In the power-grid case, good verification performance does not imply that full topology-and-parameter identification is equally easy; the reported Monte Carlo data show verification is substantially easier than full identification (Cavraro et al., 2018).

Taken together, the supplied literature presents GridProbe as a technically heterogeneous but methodologically coherent research theme. Its unifying principle is the deliberate insertion of a grid, fiducial, or active probe into the measurement chain so that hidden transport, field, or topology variables become inferable through structured response data.