Chip-level electromagnetic interference positioning method and device based on dual-plane near-field sampling

Abstract

The invention provides a chip-level electromagnetic interference positioning method and device based on dual-plane near-field sampling, and the method comprises the steps: obtaining spatial domain radiation data from a first sampling plane and a second sampling plane, and building a wave spectrum domain conversion relation between the two planes and a target detection plane through a plane wave spectrum expansion theory. And calculating a height calibration factor and inverting target plane field distribution, and finally forming high-resolution electromagnetic imaging. According to the method, the scanning area, the scanning step length and the scanning height difference serve as core parameters, sampling conditions are accurately constrained according to the Nyquist criterion and the cut-off frequency, and spectrum leakage and aliasing errors are effectively restrained; meanwhile, no extra calibration piece is needed, the test process is remarkably simplified, and structural dependence is eliminated.

Description

Technical Field

[0001] This invention relates to the field of electromagnetic interference detection technology, and in particular to a chip-level electromagnetic interference localization method and device based on dual-plane near-field sampling. Background Technology

[0002] Electromagnetic compatibility (EMC) is a critical technical indicator in electronic system design, directly impacting equipment reliability and compliance. As modern electronic devices become increasingly integrated and miniaturized, internal electromagnetic interference (EMI) sources exhibit multi-physical coupling characteristics, posing higher demands for accurate source localization. Near-field scanning platform-based radiation source detection technology, due to its cost advantages and operational flexibility, has successfully expanded from traditional antenna testing to EMI diagnostics. This technology uses electric / magnetic field probes to perform two-dimensional planar scanning in the extremely near-field region of the device surface, theoretically reconstructing the spatial distribution characteristics of electromagnetic interference sources and thus obtaining their location information.

[0003] Modern high-density electronic devices, including integrated circuits (ICs) using flip-chip packaging technology and multilayer heterogeneous printed circuit boards (PCBs), typically embed highly non-uniform, densely packed components, exhibiting significant three-dimensional structural heterogeneity. This is typically manifested as micro- and nano-scale features such as bump arrays and multi-scale passive components. These non-planar structures create physical obstacles to the physical scanning path of near-field probes. Furthermore, when the probe is raised to a plane far from the device under test (DUT) for sampling, the electromagnetic field amplitude attenuates significantly and systematically according to the spatial attenuation characteristics of Maxwell's equations, directly leading to a decrease in the spatial resolution of near-field imaging.

[0004] In recent years, to address the problem of EMI source localization in complex scenarios such as metal-shielded cavities, the academic community has developed multimodal research methods. On the one hand, equivalent dipole models are constructed to analyze the field distribution characteristics under shielding conditions (such as inverse solutions based on the Green's function method). On the other hand, machine learning algorithms such as neural networks are combined to extract features and perform nonlinear optimization on near-field scanning data. Among these methods, PWS theory, due to its rigorous mathematical completeness, has always served as the core foundational framework. However, in practical applications, when directly using PWS theory for field transformation, the spectral leakage effect caused by the finite scanning plane size prevents the complete characterization of the spatial attenuation characteristics of higher-order modal fields, resulting in significant distortion in the calculated results. Although the compensation method of introducing probe calibration factors can improve the reconstruction accuracy, this method requires a calibration structure that is highly matched to the dielectric constant and geometric topology of the DUT. Its complex testing process and strong structural dependence significantly limit its engineering applicability. At the same time, with the continuous upgrading of the demand for high-performance electronic devices, the feature size of semiconductor devices continues to shrink, and the chip integration density has significantly increased. This trend has led to a significant increase in the transient current density of MOSFET switching, which in turn significantly enhances the electromagnetic conduction emission intensity of ICs. Currently, chip-level EMI issues have become a major factor causing circuit-level and device-level EMC failures, making chip-level EMC analysis and optimization a key research direction for researchers both domestically and internationally. In traditional IC packaging structures, electromagnetic radiation mainly originates from interconnect structures such as package leads and bonding wires; however, with the widespread application of advanced packaging technologies such as 3D integration and through-silicon vias (TSVs), the EMI characteristics of new packaging architectures are receiving widespread attention from the academic community.

[0005] Therefore, there is an urgent need for a new method to obtain accurate interference source radiation information, so as to achieve interference source localization with smaller errors. Summary of the Invention

[0006] In view of this, embodiments of the present invention provide a chip-level electromagnetic interference localization method and apparatus based on dual-plane near-field sampling, so as to eliminate or improve one or more defects existing in the prior art, and solve the problem that the existing near-field scanning method is difficult to locate chip-level electromagnetic interference sources with high precision and fast speed due to cumbersome calibration process and large truncation error.

[0007] One aspect of the present invention provides a chip-level electromagnetic interference localization method based on dual-plane near-field sampling, the method comprising the following steps:

[0008] The scanning parameters are set according to the set rules. The scanning parameters include the scanning area, the scanning step size, and the scanning height difference between the first sampling plane and the target detection plane. The scanning area is set according to the operating frequency and cutoff frequency of the device under test to determine the upper frequency limit of the sampled magnetic field spectrum. The scanning step size is limited according to the Nyquist sampling theorem and the wave vector corresponding to the cutoff frequency to prevent aliasing. The scanning height difference is set based on the height of the first sampling plane combined with the operating frequency and the cutoff frequency of the device under test.

[0009] The scanning height difference is superimposed on the height of the first sampling plane to determine a second sampling plane at a corresponding height.

[0010] Acquire first spatial domain electromagnetic radiation data collected by a near-field probe in the first sampling plane according to the scanning parameters, and second spatial domain electromagnetic radiation data collected in the second sampling plane according to the scanning parameters;

[0011] The first spatial domain electromagnetic radiation data is converted into the first spectral domain electromagnetic radiation data, and the second spatial domain electromagnetic radiation data is converted into the second spectral domain electromagnetic radiation data.

[0012] Based on the predefined conversion relationship of electromagnetic radiation data in the spectral domain between the target sampling plane, the first sampling plane and the second sampling plane according to the plane spectral expansion theory, the quotient of the electromagnetic radiation data in the second spectral domain and the electromagnetic radiation data in the first spectral domain is used as the height calibration factor. The target spectral domain electromagnetic radiation data of the target detection plane is solved by combining the height calibration factor and the electromagnetic radiation data in the first spectral domain.

[0013] The electromagnetic radiation data in the target spectral domain is transformed into electromagnetic radiation data in the target spatial domain to obtain electromagnetic imaging results, and the interference source is located by the amplitude of the electromagnetic radiation.

[0014] In some embodiments, the constraints for setting the scanning area are:

[0015]

[0016] Where A represents the scanning area, k0 represents the wave vector calculated by the device under test based on the operating frequency, and k cutoff This represents the wave vector calculated based on the cutoff frequency.

[0017] In some embodiments, the constraints for setting the scanning step size are:

[0018]

[0019] Where, k cutoffThis represents the wave vector calculated based on the cutoff frequency.

[0020] In some embodiments, the setting rule for the scanning height difference is expressed as follows:

[0021]

[0022] Where Δd represents the scanning height difference, d #2-DUT λ0 represents the height difference between the first sampling plane and the target sampling plane, λ0 is the wavelength corresponding to the operating frequency of the device under test, and k0 represents the wave vector calculated by the device under test based on the operating frequency. cutoff This represents the wave vector calculated based on the cutoff frequency.

[0023] In some embodiments, the quotient of the second spectral domain electromagnetic radiation data and the first spectral domain electromagnetic radiation data is used as the height calibration factor, calculated as follows:

[0024]

[0025] in, This represents the electromagnetic radiation data in the second spectral domain, that is, the spectral domain form of the magnetic field component at the second sampling plane; The first spectral domain electromagnetic radiation data represents the spectral domain form of the magnetic field component at the first sampling plane; z2 represents the height of the first sampling plane, z3 represents the height of the second sampling plane, and k represents the height of the second sampling plane. x Let k be the wave vector component along the x-axis. y denoted as the wave vector component along the y-axis; j represents the ordinal unit; Δd represents the scan height difference; e represents the natural base; η represents the height calibration factor;

[0026]

[0027] The formula for calculating the wave vector k is:

[0028] and

[0029] Where ω is the angular frequency, μ is the permeability of the medium, and ε is the dielectric constant of the medium; and It represents a unit vector in a rectangular coordinate system.

[0030] In some embodiments, the target spectral domain electromagnetic radiation data of the target detection plane is solved by combining the height calibration factor and the first spectral domain electromagnetic radiation data, and the calculation formula is:

[0031]

[0032] in, The z1 represents the electromagnetic radiation data in the target spectral domain, i.e., the spectral domain form of the magnetic field component at the target detection plane; the k represents the height of the target detection plane. x Let k be the wave vector component along the x-axis. y Let η be the wave vector component along the y-axis, and let η represent the height calibration factor.

[0033] In some embodiments, the method further includes: applying a Wiener filter to the electromagnetic imaging result, calculated as follows:

[0034]

[0035] Where I(m,n) represents the data before filtering, J(m,n) represents the data after filtering, noise_var represents the estimated noise variance, μ represents the mean of the data before filtering, and σ represents the standard deviation of the data before filtering.

[0036] On the other hand, the present invention also provides a chip-level electromagnetic interference positioning device, the device comprising:

[0037] A support platform is used to load the device under test; the support platform is provided with a first slide rail along the horizontal X-axis.

[0038] The first support member is slidably connected to the bearing platform via the first slide rail and is driven by the first motor; the first support beam is provided with a second slide rail along the Y-axis of the horizontal plane;

[0039] The second support member is slidably connected to the first support member via the second slide rail and is driven by the second motor; the second support member is provided with a third slide rail along the Z-axis direction;

[0040] The third support member is slidably connected to the second support member via the third slide rail and is driven by the third motor;

[0041] A near-field probe, fixed on the third support, is used to collect electromagnetic radiation data in the first spatial domain on the first sampling plane and electromagnetic radiation data in the second spatial domain on the second sampling plane according to preset scanning parameters. The scanning parameters include scanning area, scanning step size, and the scanning height difference between the first sampling plane and the target detection plane.

[0042] The processor module is connected to control the first motor driver, the second motor driver, the third motor driver, and the near-field probe. The processor module is also used to execute the chip-level electromagnetic interference localization method based on dual-plane near-field sampling as described above to obtain the interference source localization result of the device under test.

[0043] On the other hand, the present invention also provides a computer-readable storage medium having a computer program / instructions stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0044] On the other hand, the present invention also provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the above-described method.

[0045] The chip-level electromagnetic interference localization method and device based on dual-plane near-field sampling described in this invention acquires spatial domain radiation data in a first and second sampling plane, respectively. Then, using planar spectral expansion theory, it establishes the spectral domain transformation relationship between the two planes and the target detection plane, calculates the height calibration factor, and inverts the target plane field distribution, ultimately forming a high-resolution electromagnetic image. This method uses scanning area, scanning step size, and scanning height difference as core parameters, and precisely constrains sampling conditions based on the Nyquist criterion and cutoff frequency, effectively suppressing spectral leakage and aliasing errors. Simultaneously, it requires no additional calibration components, significantly simplifying the testing process and eliminating structural dependence. Compared to traditional direct transformation or calibration factor methods, this invention achieves peak position errors of less than one-tenth of a wavelength and reduces the root mean square error of amplitude by more than 50% in the localization of microscale radiation sources such as chip package pins, bonding wires, and three-dimensional through-silicon vias. It can quickly and non-contactly achieve precise millimeter-level interference source localization, meeting the electromagnetic compatibility diagnostic requirements of advanced packaged chips.

[0046] Additional advantages, objects, and features of the invention will be set forth in part in the description which follows, and will also become apparent in part to those skilled in the art upon studying the description, or may be learned by practice of the invention. The objects and other advantages of the invention can be realized and obtained by means of the structures specifically pointed out in the description and drawings.

[0047] Those skilled in the art will understand that the objectives and advantages achievable with the present invention are not limited to those specifically described above, and that the above and other objectives achievable with the present invention will become clearer from the following detailed description. Attached Figure Description

[0048] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, are not intended to limit the scope of the invention. In the drawings:

[0049] Figure 1 This is a schematic diagram of plane wave spectrum expansion based on PWS theory.

[0050] Figure 2 This is a flowchart of interference source localization technology based on dual-plane near-field sampling.

[0051] Figure 3This is a schematic flowchart of a chip-level electromagnetic interference localization method based on dual-plane near-field sampling according to an embodiment of the present invention.

[0052] Figure 4 This is a structural diagram of a printed dipole antenna according to an embodiment of the present invention.

[0053] Figure 5 (a) shows the derivation of the printed dipole antenna before filtering |H x |, Figure 5 (b) shows the derivation of the filtered printed dipole antenna |H x |

[0054] Figure 6 for Figure 4 The printed dipole antenna in the middle about |H x |and|H y |Comparison of the component values ​​at z=3mm based on formula (7), the calculated values ​​and theoretical values ​​of dual-plane near-field sampling, and the scan values ​​at z=5mm.

[0055] Figure 7 for Figure 4 The calculation values, direct derivation values, and theoretical values ​​of the magnetic field component amplitude of the printed dipole antenna in the dual-plane near-field sampling algorithm are compared, along with the RMSE calculation results. Figure 7 (a) represents the |H| at y = 43 mm. x |; Figure 7 (b) in the figure represents |H| = 41 mm. y |

[0056] Figure 8 This is a schematic diagram of the CESAME chip packaging according to an embodiment of the present invention.

[0057] Figure 9 for Figure 8 A schematic diagram of the ICEM model structure of the CESAME chip described in the article.

[0058] Figure 10 for Figure 9 The CESAME chip in the middle is about |H x |and|H y |Comparison of the component values ​​at z = 2 mm based on formula (7), the calculated values ​​and theoretical values ​​of dual-plane near-field sampling, and the scan values ​​at z = 5 mm.

[0059] Figure 11 for Figure 9 A comparison of the calculated values, directly derived values, and theoretical values ​​of the magnetic field component amplitude of the CESAME chip using the dual-plane near-field sampling algorithm, along with the RMSE calculation results, is presented. Figure 11(a) represents the |H| at y = 24 mm. x |; Figure 11 (b) in the figure represents |H| = 24 mm. y |

[0060] Figure 12 This is a schematic diagram of a 3D signal transmission model structure based on TSV in one embodiment of the present invention.

[0061] Figure 13 for Figure 12 A schematic diagram of the numbering of heavy TSVs in the 3D signal transmission model.

[0062] Figure 14 for Figure 13 3D signal transmission model about |H x |and|H y | Comparison of the component at z = 85um based on the direct derivation value of formula (7), the calculated value of dual-plane near-field sampling and the theoretical value, and the scan value at z = 90um.

[0063] Figure 15 The hybrid filter chip based on TSV process packaging described in one embodiment of the present invention relates to |H x | Comparison of the component values ​​at z = 2 mm based on formula (7), the calculated values ​​and theoretical values ​​of dual-plane near-field sampling, and the scan values ​​at z = 2.5 mm. Detailed Implementation

[0064] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the embodiments and accompanying drawings. Here, the illustrative embodiments and descriptions of this invention are used to explain the invention, but are not intended to limit the invention.

[0065] It should also be noted that, in order to avoid obscuring the invention with unnecessary details, only the structures and / or processing steps closely related to the solution according to the invention are shown in the accompanying drawings, while other details that are not closely related to the invention are omitted.

[0066] It should be emphasized that the term "including / comprises" as used herein refers to the presence of a feature, element, step, or component, but does not exclude the presence or addition of one or more other features, elements, steps, or components.

[0067] It should also be noted that, unless otherwise specified, the term "connection" in this article can refer not only to a direct connection, but also to an indirect connection involving an intermediary.

[0068] Existing technologies for near-field conversion and reconstruction of the device under test (DUT) require the acquisition of the radiation field of a calibration device for calculating the calibration factor. However, due to limitations inherent in PWS theory, high reconstruction accuracy is only achieved when the selected calibration device's structure and physical characteristics are similar to the DUT. When this algorithm is applied to heterogeneous integrated circuits (ICs) or multilayer heterogeneous printed circuit boards (PCBs) with unknown radiation modes, ensuring the structural similarity of the calibration device is difficult. Existing technologies require testing the high-plane radiation field data of both the calibration device and the DUT, and simulating the low-plane radiation data of the calibration device. These data are then used together as input for near-field-to-near-field conversion technology to finally obtain the electromagnetic radiation result of the DUT in the low-plane. The overall testing process is quite cumbersome. Furthermore, research on the localization of EMI sources within the chip-level system is lacking.

[0069] This invention aims to calculate electromagnetic imaging data of a plane lower than the DUT (Distributed Under Test) by combining near-field dual-plane sampling with PWS (Precision Waveform Sampling) unfolding theory, serving as a basis for high-precision EMI interference source localization. First, the theoretical premises of this invention are explained, including three parts: the basic theory of PWS unfolding, the dual-plane near-field sampling algorithm, and related parameter settings.

[0070] 1. Basic Theory of PWS Deployment

[0071] When the electromagnetic waves generated by the electromagnetic radiator propagate along the positive z-direction, such as Figure 1 As shown, the electromagnetic wave generated at the plane z = z0 of the radiator can be unfolded into a superposition of plane waves with different amplitudes and phases, each along a different wave vector k. This transformation can be written as the following expression:

[0072]

[0073] Where E(x,y,z0) represents the electric field vector component at the spatial point (x,y) on the plane z = z0, and H(x,y,z0) represents the magnetic field vector component at the spatial point (x,y) on the plane z = z0. Wave vector ω is the angular frequency, μ is the permeability of the medium, and ε is the permittivity of the medium. As for k and k... x k y and k z The relationship is as follows:

[0074]

[0075] in, and It represents a unit vector in a rectangular coordinate system.

[0076] γ is calculated as follows:

[0077]

[0078] The transformations in (1) and (2) above realize the conversion of the emitted electromagnetic wave from the spatial domain to the spectral domain at plane z0. E and H are the electric and magnetic field values ​​of the radiation source in the spatial domain, while E P and H P These are the electric and magnetic field values ​​in the spectral domain. Observing the formulas, it can be seen that the above transformation has the same form as the inverse Fourier transform. Given that the electric or magnetic field in the spatial domain is known, the Fourier transform can be used to solve for the field values ​​in the spectral domain:

[0079]

[0080] Based on the derivation of formulas (5) and (6) above, it can be found that the electromagnetic field values ​​of the same radiation source at different heights often have different coefficients in the spectral domain. Suppose there are two planes at different heights, denoted by z. i and z j This is represented by H, and the field values ​​on these two planes are transformed to the spectral domain. In this case, H... x Taking components as an example, the relationship between the field values ​​on two planes can be derived from their transformation equations:

[0081]

[0082] According to the derived height transformation formula (7), it is theoretically feasible to calculate the radiation data of a certain height surface using the height difference, provided that the radiation data of another height surface is known.

[0083] 2. Dual-plane near-field sampling algorithm

[0084] Since the near-field sampling plane of the DUT cannot be infinitely large in practical calculations, directly applying PWS theory to derive the radiation data of the low-plane will produce significant truncation errors. This invention proposes a derivation algorithm based on a dual-plane near-field sampling method to reduce the errors generated by PWS expansion theory during application and improve the accuracy of low-plane electromagnetic imaging data. (Refer to...) Figure 2 Suppose we need to acquire electromagnetic field data from plane #1 (height z1), which is very close to the DUT, while the near-field sampling probe can only sample from plane #2 (height z2), which is relatively far from the DUT. Now we choose plane #3, with the following height:

[0085] z3=z2+(z2-z1)=z2+Δd (8)

[0086] In the above formula, Δd is the height difference between planes #1 and #2. Electromagnetic field data of planes #2 and #3 are sampled using a near-field scanning platform. The obtained field values ​​can be converted to the spectral domain using formulas (1) and (2), and substituted into formula (7) to obtain:

[0087]

[0088] Here, η refers to the height calibration factor. The selected height difference between plane #3 and plane #2 is the same as the height difference between plane #2 and plane #1, and therefore can be calculated based on the height calibration factor η and the field value of plane #2 in the spectral domain:

[0089]

[0090] In formula (10) This is the spectral domain form of the magnetic field component at plane #1. Substituting it into equation (2) and performing an inverse Fourier transform yields the magnetic field information in the spatial domain. It must be emphasized here that the near-field conversion formula derived from the fundamental PWS theory in this invention is effective under the assumptions of a homogeneous propagation medium, no scattering effects, and negligible non-plane wave components in the radiated electromagnetic field. These conditions define the scope of this invention. For non-ideal cases, such as non-uniform propagation media or strong near-field coupling, it may be necessary to study a perturbation model incorporating a calibration factor η, thereby enabling a more comprehensive analysis of the algorithm's deviations under complex practical conditions.

[0091] Compared to near-field-to-near-field transformation methods based on calibration components, this invention exhibits a simpler testing procedure and demonstrates independence from the geometry of the EMI source. Notably, the proposed dual-plane sampling algorithm primarily focuses on predicting the near-field radiation amplitude: within the radiator-induced near-field region (d < 0.1λ), due to the evanescent wave dominance and mutual coupling effects, the electromagnetic field phase exhibits drastic spatial gradient changes. This phase complexity contributes negligibly to EMI source localization. Therefore, this invention does not explicitly process the phase information prediction and reconstruction based on the dual-plane sampling algorithm.

[0092] 3. Setting relevant parameters

[0093] 3.1 Since the sampling plane cannot be infinitely large, truncation errors will inevitably occur in the calculation of the plane wave spectrum domain. In actual testing, more attention is paid to the capture of some important wave vector components, while some high-frequency components can be appropriately ignored. Therefore, to mitigate the truncation error problem in actual testing, it is necessary to select an effective wave vector range and truncation threshold for the specific cutoff frequency of the DUT, which must satisfy the following in the spatial domain:

[0094]

[0095] in, k0 and k cutoff Both represent the magnitude of the wave vector, where the wavelength λ0 of the DUT at the known operating frequency and the wavelength λ corresponding to the selected cutoff frequency are given. cutoff In this case, k0 and k can be calculated using the formula above. cutoff .

[0096] 3.2 Scan Height Difference Settings

[0097] In the dual-plane near-field sampling algorithm, for the scan height difference Δd corresponding to the calibration factor η, since the algorithm aims to derive the radiation information of plane #1 from the near-field radiation data of plane #2, the value of Δd must obviously be less than the distance from plane #2 to the DUT. Furthermore, to avoid excessively large Δd causing ambiguity in the extrapolated data, the values ​​should be adjusted according to k0 and k... cutoff Further restrictions are imposed. The following are suggested conditions for Δd:

[0098]

[0099] Where Δd represents the scanning height difference, d #2-DUT λ0 represents the height difference between plane #2 and plane #1 where the device under test (DUT) is located, λ0 is the wavelength corresponding to the operating frequency of the DUT, and k0 represents the wave vector of the DUT calculated based on its operating frequency. cutoff This represents the wave vector calculated based on the cutoff frequency.

[0100] 3.3 Scan Step Size Setting

[0101] Since scanning sampling needs to be performed in the very near-field region of the EMI source, and the field distribution is dominated by evanescent waves, a denser sampling wavelength is required to achieve high-precision near-field derivation. Based on the Nyquist sampling theorem and k determined from the cutoff frequency... cutoff The sampling step size has the following limitations:

[0102]

[0103] Because the amplitude of evanescent waves decays rapidly, the effective high-frequency components that can actually be detected are limited, so the step size is appropriately widened based on the above formula. In the millimeter wave and lower frequency bands, the inductive near-field scanning step size is usually set to λ / 50 to λ / 100. Beyond this range, the improvement in the reconstruction of the transmission field can be ignored.

[0104] Therefore, based on the characteristics of the DUT, the settings parameters for near-field scanning, including scanning step size, sampling area, and scanning height difference, are determined, and then a near-field probe is used to sample the radiation data of the DUT. Based on the sampling results, electromagnetic data from plane #2 and plane #3 are obtained as input data for the dual-plane near-field sampling algorithm. The spatial domain electromagnetic radiation data from planes #2 and #3 are transformed into spectral domain electromagnetic radiation data. The height calibration factor is calculated using the height transformation formula (9) derived from the PWS expansion theory. Based on the height calibration factor and the spectral domain data of plane #2, the spectral domain data of plane #1 is calculated; the spectral domain electromagnetic radiation data of plane #1 is then transformed into spatial domain electromagnetic radiation data. Finally, based on the calculated electromagnetic imaging results of plane #1, the EMI source can be accurately located.

[0105] Specifically, this invention provides a chip-level electromagnetic interference localization method based on dual-plane near-field sampling, such as... Figure 3 As shown, the method includes the following steps S101 to S106:

[0106] Step S101: Set the scanning parameters according to the set rules. The scanning parameters include the scanning area, scanning step size, and the scanning height difference between the first sampling plane and the target detection plane. The scanning area is set according to the operating frequency and cutoff frequency of the device under test to determine the upper limit of the frequency of the sampled magnetic field spectrum. The scanning step size is limited according to the wave vector corresponding to the cutoff frequency based on the Nyquist sampling theorem to prevent aliasing. The scanning height difference is set based on the height of the first sampling plane combined with the operating frequency and cutoff frequency of the device under test.

[0107] Step S102: Superimpose the scanning height difference on the height of the first sampling plane to determine the second sampling plane at the corresponding height.

[0108] Step S103: Acquire the first spatial domain electromagnetic radiation data collected by the near-field probe in the first sampling plane according to the scanning parameters, and the second spatial domain electromagnetic radiation data collected in the second sampling plane according to the scanning parameters.

[0109] Step S104: Convert the electromagnetic radiation data in the first spatial domain into electromagnetic radiation data in the first spectral domain, and convert the electromagnetic radiation data in the second spatial domain into electromagnetic radiation data in the second spectral domain.

[0110] Step S105: Based on the predefined conversion relationship between the target sampling plane, the first sampling plane and the second sampling plane in the spectral domain electromagnetic radiation data according to the plane spectral expansion theory, the quotient of the second spectral domain electromagnetic radiation data and the first spectral domain electromagnetic radiation data is used as the height calibration factor. The target spectral domain electromagnetic radiation data of the target detection plane is solved by combining the height calibration factor and the first spectral domain electromagnetic radiation data.

[0111] Step S106: Transform the electromagnetic radiation data in the target spectral domain into electromagnetic radiation data in the target spatial domain to obtain electromagnetic imaging results, and locate the interference source by the amplitude of electromagnetic radiation.

[0112] In steps S101 to S106, during chip-level electromagnetic interference localization, the significant three-dimensional structural heterogeneity of integrated circuits, typically manifested as micro / nano-scale features such as bump arrays and multi-scale passive components, creates physical obstacles to the physical scanning path of the near-field probe. When the probe is raised to a plane far from the device under test (DUT) for sampling, the electromagnetic field amplitude attenuates significantly and regularly according to the spatial attenuation characteristics of Maxwell's equations, directly leading to a decrease in the spatial resolution of near-field imaging. This invention combines a dual-plane near-field sampling algorithm, performing near-field scanning on a first and second sampling plane with equal spacing above the target detection plane. Based on PWS unfolding theory, the conversion relationship between the spectral domain magnetic field data on the target detection plane, the first sampling plane, and the second sampling plane is calculated to achieve accurate detection of the target detection plane.

[0113] In step S101, the scanning parameters are set using formulas 11, 12, and 13 as described above. The constraints for setting the scanning area are as follows:

[0114]

[0115] Where A represents the scanning area, k0 represents the wave vector of the device under test calculated based on its operating frequency, and k cutoff This represents the wave vector calculated based on the cutoff frequency.

[0116] The constraints for setting the scanning step size are as follows:

[0117]

[0118] Where, k cutoff This represents the wave vector calculated based on the cutoff frequency.

[0119] In some embodiments, the rule for setting the scanning height difference is expressed as follows:

[0120]

[0121] Where Δd represents the scanning height difference, d #2-DUT The first sampling plane represents the height difference from the target sampling plane, λ0 is the wavelength corresponding to the operating frequency of the device under test, and k0 represents the wave vector calculated by the device under test based on its operating frequency. cutoff This represents the wave vector calculated based on the cutoff frequency.

[0122] In steps S102 to S105, a second sampling plane (z = z2 = z1 + Δd) is obtained by superimposing Δd along the normal direction using the first sampling plane (z = z1) as a reference. Both planes are located in the near-field region of the device under test (DUT) radiation, but are higher than the surface of the DUT to avoid physical collision between the probe and the chip under test.

[0123] Using a magnetic near-field probe (or electric field probe), the area and step size set in S101 are scanned point by point in the two planes z = z1 and z = z2 to acquire electromagnetic radiation data in the first and second spatial domains. After probe calibration and amplitude-phase consistency correction, the data are stored in a computer for further processing.

[0124] Two-dimensional Fourier transforms were performed on the two sets of spatial domain data to obtain electromagnetic radiation data in the first and second spectral domains. The transform formula follows the plane wave spectral expansion theory, expressing any field point as a plane wave superposition integral.

[0125] The quotient of the electromagnetic radiation data in the second spectral domain and the electromagnetic radiation data in the first spectral domain is used as the height calibration factor, calculated as follows:

[0126]

[0127] in, This represents the electromagnetic radiation data in the second spectral domain, i.e., the spectral domain form of the magnetic field components at the second sampling plane; This represents the electromagnetic radiation data in the first spectral domain, i.e., the spectral domain form of the magnetic field component at the first sampling plane; z2 represents the height of the first sampling plane, z3 represents the height of the second sampling plane, and k... x Let k be the wave vector component along the x-axis. y denoted as the wave vector component along the y-axis; j represents the ordinal unit; Δd represents the scan height difference; e represents the natural base; η represents the height calibration factor;

[0128]

[0129] The formula for calculating the wave vector k is:

[0130] and

[0131] Where ω is the angular frequency, μ is the permeability of the medium, and ε is the dielectric constant of the medium; and It represents a unit vector in a rectangular coordinate system.

[0132] The target electromagnetic radiation data of the target detection plane is obtained by combining the height calibration factor and the first spectral domain electromagnetic radiation data. The calculation formula is as follows:

[0133]

[0134] in, This represents the electromagnetic radiation data in the target's spectral domain, i.e., the spectral domain form of the magnetic field components at the target detection plane; z1 represents the height of the target detection plane, and k... x Let k be the wave vector component along the x-axis. y Let η be the wave vector component along the y-axis, and η represent the height calibration factor.

[0135] In step S106, the electromagnetic radiation data in the target spectral domain is inversely Fourier transformed from the frequency domain back to the spatial domain and imaged, and the electromagnetic interference source is located by the amplitude.

[0136] In some embodiments, step S106 further applies Wiener filtering to the electromagnetic imaging results, calculated as follows:

[0137]

[0138] Where I(m,n) represents the data before filtering, J(m,n) represents the data after filtering, noise_var represents the estimated noise variance, μ represents the mean of the data before filtering, and σ represents the standard deviation of the data before filtering.

[0139] On the other hand, the present invention also provides a chip-level electromagnetic interference positioning device, the device comprising: a support platform, a first support member, a second support member, a third support member, a near-field probe, and a processor module.

[0140] The support platform is used to load the device under test; the support platform is provided with a first slide rail along the X-axis of the horizontal plane.

[0141] The first support member is slidably connected to the bearing platform via the first slide rail and is driven by the first motor; the first support beam is provided with a second slide rail along the Y-axis of the horizontal plane.

[0142] The second support member is slidably connected to the first support member via the second slide rail and is driven by the second motor; the second support member is provided with a third slide rail along the Z-axis direction.

[0143] The third support member is slidably connected to the second support member via a third slide rail and is driven by a third motor.

[0144] The near-field probe is fixed on the third support and is used to collect electromagnetic radiation data in the first spatial domain on the first sampling plane and electromagnetic radiation data in the second spatial domain on the second sampling plane according to preset scanning parameters. The scanning parameters include scanning area, scanning step size and scanning height difference between the first sampling plane and the target detection plane.

[0145] The processor module connects and controls the first motor drive, the second motor drive, the third motor drive, and the near-field probe. The processor module is also used to execute the chip-level electromagnetic interference localization method based on dual-plane near-field sampling as described in steps S101 to S106 above to obtain the interference source localization result of the device under test.

[0146] On the other hand, the present invention also provides a computer-readable storage medium having a computer program / instructions stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0147] On the other hand, the present invention also provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the above-described method.

[0148] The present invention will now be described in conjunction with specific embodiments:

[0149] Example 1: Simulation Case of PCB-level Radiation Source

[0150] The verification of EMI source localization technology based on a dual-plane near-field sampling algorithm was first carried out at the PCB radiation level. To verify the effectiveness of the algorithm, a printed dipole antenna with a center operating frequency of 2.45 GHz and a balun microstrip feed structure was designed and simulated. Figure 4 As shown, this antenna employs a dipole double-arm structure formed by placing metal transmission line segments on both sides of the dielectric substrate, comprising a microstrip transmission line and a microstrip balun structure. The dielectric material is FR4, the metal transmission lines are designed as ideal electrical conductors, and the feed point uses a waveport excitation mode. Detailed dimensional parameters of this antenna are shown in Table 1.

[0151] Table 1 Structural parameters of printed dipole antenna

[0152] parameter Numerical value (mm) <![CDATA[L1]]> 60 <![CDATA[L2]]> 21 <![CDATA[W1]]> 40 <![CDATA[W2]]> 3 <![CDATA[H1]]> 1.6

[0153] Using a printed dipole antenna as the radiation source, its near-field radiation data was obtained through electromagnetic simulation. Assuming the lower surface of the printed dipole antenna is z = 0, the upper surface is located at z = 1.6 mm. To obtain electromagnetic field information of the plane closest to the dipole, a plane with a height difference Δd = 2 mm at z = 5 mm and a plane at z = 7 mm were selected, corresponding to probe sampling planes #1 and #2 in the dual-plane near-field sampling algorithm, respectively. The size of the scanning plane was set to 80 mm * 80 mm, and the step size was set to Δx = Δy = 1 mm. The amplitude and phase data of the magnetic field components Hx and Hy on the two sampling planes were simulated and exported, and input into the designed dual-plane near-field sampling algorithm to derive Hx on the plane at z = 3 mm. x and H y value.

[0154] To more accurately recover near-field emission data, filtering techniques are employed to suppress noise generated during the transformation process. Wiener filtering was chosen because, compared to other filtering techniques, it adaptively minimizes the mean square error between the noise signal and the ideal signal. Furthermore, when using plane wave spectrum theory to transform near-field radiation, additive noise may be introduced; Wiener filtering can achieve good noise suppression without requiring preset basis functions. When using Wiener filtering for image processing, the local mean and variance of each data point in its specific neighborhood are first calculated. The calculated data is then processed using the pixel value adjustment formula shown below:

[0155]

[0156] Where I(x,y) is the input data before filtering, J(x,y) is the output data after filtering, and noise_var is the estimated noise variance. When calculating the above formula, if the local variance at a certain point is close to noise_var, it indicates that the region is mainly noise, and smoothing will be performed.

[0157] Figure 5 This demonstrates the |H| of a printed dipole antenna converted using a dual-plane near-field scanning algorithm at z = 3 mm. x Data Figure 5 (a) shows the effect before filtering. Figure 5 (b) shows the effect after filtering. Based on the overall near-field data resolution and noise level, the Wiener filter was used with a local pixel window of [5,5] and noise_var of 0.48. The filtering results show that the Wiener filter ensures that values ​​in key areas with strong radiation are not processed, while effectively eliminating unwanted noise. This plays a significant role in accurately reconstructing near-field radiation results and further locating EMI sources.

[0158] Figure 6 This demonstrates the |H| of the plane with z = 3 mm derived by the algorithm. x The values ​​derived directly from formula (7), the calculated values ​​from dual-plane near-field sampling, the theoretical values ​​obtained from simulation, and the scan values ​​from the z=5mm plane are compared. The near-field emission data calculated by the dual-plane near-field scanning algorithm are processed using the Wiener filtering technique described above. The filtering results are as follows: Figure 5 As shown in Figure (b), the local field window used for filtering |Hx| is [5,5], and noise_var is 0.48; the local field window used for filtering |Hy| is [5,5], and noise_var is 0.96. It can be seen from the figure that the |Hx| obtained by scanning at the plane z = 5 mm... x|It is more blurred compared to the scan obtained at the z=3mm plane,|H y The same applies. And the |H| calculated using the dual-plane near-field scanning algorithm... x |and|H y Compared with the direct derivation using formula (7), it has a higher degree of agreement with the theoretical value, indicating that the proposed algorithm can be used to make reasonable predictions about the near-field value of the low plane.

[0159] Figure 7 Then by comparing |H at high radiation values x |and|H y The effectiveness of the algorithm was further revealed numerically. Figure 7 Part (a) uses the dual-plane near-field sampling algorithm to obtain |H| at y = 43 mm. x The curves are very close to the theoretical values, with a generally consistent trend and peaks occurring at x = 34 mm; similarly... Figure 7 (b) The calculation of |H at x = 41 mm y It also has very high accuracy, with the peak value appearing at y = 36 mm, which is close to the theoretical value. The height transformation formula (7) is directly used to derive the |H| of the lower plane from the high plane scanning data. x |and|H y |, in Figure 7 A comparison is made between the two. It can be seen that although the peak values ​​of the data obtained by directly using the formula and the peak values ​​of the theoretical values ​​appear in roughly the same positions, the numerical differences are large. Moreover, in other locations with lower radiation intensities besides the peak values, it is difficult to obtain accurate predictions by directly using the formula. This is because the application range of formula (7) is an infinitely large sampling plane, while the actual scanning plane during testing cannot achieve this condition. The root mean square error (RMSE) of |Hx| and |Hy| obtained by the dual-plane near-field scanning algorithm and directly derived by formula (7) compared with the theoretical values ​​is further calculated and shown in the figure. Figure 7 Based on the comparison results of RMSE, the algorithm proposed in this paper has a clear advantage in accurately obtaining radiation information closer to the DUT.

[0160] Example 2: Simulation Case of IC-level Radiation Source

[0161] Traditional interference source localization algorithms primarily focus on PCB-level devices in their verification and application. This embodiment will further discuss the application of the proposed algorithm in the IC radiating layer. First, a fully packaged CESAME chip model is used as the radiated interference source, and near-field simulation is performed to obtain data and verify the correctness of the dual-plane near-field scanning algorithm. The CESAME test chip contains six logic core blocks on its internal die, comprising over 700,000 transistors. It uses a 144-pin quad flat package, including one VDD pin and one VSS pin. A schematic diagram of the chip package and the relative positions of the pins are shown below. Figure 8 As shown, the leads and bonding wires of an IC package increase series inductance, which radiates and couples into external electromagnetic fields, and the intensity of this radiation increases with the length of the leads. In the emission field generated during IC operation, the radiation caused by the leads and bonding wires constitutes the majority of the emission.

[0162] To predict the noise that may be generated by this IC package model after adding excitation, we will temporarily disregard the internal design details of the IC and predict the electromagnetic interference generated by the IC by obtaining the integrated circuit emission model (ICEM) of the CESAME chip. The ICEM model of this test chip is as follows: Figure 9 As shown, the ICEM model comprises two main components: a passive decoupling network (PDN) and an internal activity (IA). The impedance network described by the PDN and the current source emitter simulating the internal IC activity, as described by the IA, allow for the modeling of the radiated conduction of the CESAME chip. The ICEM model is imported into the IC-EMC integrated circuit electromagnetic compatibility performance simulation tool, and the interface information of the CESAME chip package described by the designer's Input / Output Buffer Information Specification (IBIS) is loaded. The chip and package location information displayed by the IBIS information is shown in Table 2 below.

[0163] Table 2 Parameters of CESAME chip and its packaging structure

[0164] parameter Numerical value (mm) parameter Numerical value (mm) Chip width 2.7 Package width 20.1 chip length 2.5 Package length 20.1 Chip height 0.8 Package height 1 Starting position of chip x 31.8 Starting position of package x 23.1 Starting position of chip y 32.9 Starting position of package y 24.1 Starting position of chip z 0 Starting position of the z-package 0

[0165] By performing near-field testing on the CESAME chip package model in a simulation tool, a near-field radiation map at a set height can be obtained. In this simulation case, the scanning plane is set to 60mm*60mm, and the step size is Δx=Δy=1mm. The plane at z=5mm (Δd=2mm) and the plane at z=8mm are selected as input data for the dual-plane near-field scanning algorithm, and the radiation field map of the CESAME chip package model at the z=2mm plane is derived. Similarly, Wiener filtering technology is used to remove noise from the calculated data, and the |H| of the final processed z=2mm plane is then used. x |and|H y |Based on the direct derivation value of formula (7), the calculated value of dual-plane near-field sampling, the theoretical value, and the scan value of the z=5mm plane, the results are compared as follows: Figure 10 As shown. To clearly illustrate the electromagnetic radiation of the chip and its package, Figure 10 The coordinate range of the radiation field map is focused near the chip, with the chip outline represented by a yellow outline and the package outline by a white outline. The |H| of the z=2mm plane processed by the dual-plane near-field scanning algorithm... x |and|H y Compared to the direct derivation of the formula, which has a high degree of agreement with the theoretical value, a comparison of the pin distribution diagram shows that the strongest radiation is concentrated at the leads and bonding wires connected to the VDD and VSS pins.

[0166] The algorithmic calculated value and the theoretical value are compared at the point near the starting position of the package, at x = 24 mm on the z = 2 mm plane. Figure 11 As shown. The theoretical |H x |and|H y The location where the peak value appears is near the location of the VDD and VSS pins, i.e., the location of the EMI source. According to Figure 11 The comparison results shown are obtained using the dual-plane scanning algorithm proposed in the article, with |H... x |and|H y It can accurately reproduce the radiation distribution under real-world conditions. The peak values ​​of |Hx| calculated by the algorithm appear at y = 28 mm and 30 mm, which are completely consistent with the theoretical values; |H y The same applies. However, if the method of directly deriving using formula (7) is adopted, the obtained field value, along with the algorithm-calculated value and the theoretical value, will be... Figure 11 Comparison. Clearly, the directly derived |H x |and|H y The value differs significantly from the theoretical value, and the location of the peak also deviates from the theoretical value, which means that this method cannot accurately locate the EMI source.

[0167] Example 3: 3D TSV Transmission Model

[0168] With the rapid development of 3D integration technology, especially TSV-based chip interconnect structures, EMI diagnostic technology is urgently needed to improve designs. This invention constructs a 3D model of signal transmission in a multilayer silicon wafer stack based on TSV, used to simply simulate silicon chip interconnects. The specific model structure is as follows... Figure 12 As shown. This 3D transmission model mainly consists of two silicon substrates, metal transmission lines, and TSVs. The bottom silicon substrate has a metal ground plane on its back and 5×5 signal transmission channels on its top, which can be connected to the transmission paths of the upper silicon substrate through TSV structures. Each TSV in the top silicon substrate structure is labeled with a number, such as... Figure 13 As shown. To facilitate the viewing of the signal path and the analysis of the radiation field diagram, two transmission channels connected to the TSV structure, namely TSV(1,3) and TSV(3,2), were randomly selected and connected to the silicon substrate on top by transmission lines to form a complete signal path.

[0169] The specific parameter settings for this 3D signal transmission model are shown in Table 3. All metal parts in the model are made of copper, the dielectric material in the TSV structure is SiO2, and the substrate material is silicon. Waveport settings are added to both ends of the signal path in the simulation tool, allowing the signal to enter from port #1, pass through the upper transmission structure connected by the TSV, and finally exit from port #2. Other signal channels that do not form a transmission path are connected to the ground plane through vias. The center point of the metal ground plane is taken as the origin o of the model's coordinate system, and the height of the 3D model is 83µm. Scanning planes with Δd = 5µm and z = 90µm and z = 95µm are selected, with a scan plane size of 600×600µm and a step size of Δx = Δy = 5µm.

[0170] Table 3 Parameters of the 3D Signal Transmission Model

[0171]

[0172] Figure 14 This demonstrates the planar radiative data |H obtained using a dual-plane near-field scanning algorithm at z=85µm. x |and|H y |, and compared the theoretical value directly calculated by formula (7), the value calculated by dual-plane near-field sampling, and the data collected by scanning in the z = 90um plane. It should be noted that the data shown in the figure have all undergone noise reduction filtering: where |H x |Filtering is performed using a local field processing window of [5,5] (noise variance noise_var = 7.81e+03), |H yThe same processing window size (noise_var = 3.04e+04) was used for filtering. Analysis shows that, compared with the relatively blurry measurement data of the radiation field distribution in the z = 90µm plane, the low-plane radiation value calculated by the algorithm proposed in this embodiment provides a clearer data distribution that is closer to the actual radiation characteristics, thereby significantly improving the positioning accuracy of the EMI source.

[0173] To fully demonstrate the advantages of the dual-plane near-field sampling algorithm in EMI source localization, Table 4 details the |H| values ​​on the cross sections at y = 100 μm and y = -100 μm. y The peak position and precise data of |H. Based on the sampling information and the structural parameters of the three-dimensional signal transmission model, it can be seen that the center points of TSV(1,1) to TSV(1,5) are located in the y=100um plane, and the center points of TSV(3,1) to TSV(3,5) are located in the y=-100um plane. Based on the data in Table IV, it can be concluded that if |H is directly collected from the z=90um plane... y The data-based localization of the radiation source shows a 15µm x-axis deviation from the theoretical value at the peak position in the y=100µm section; and a 10µm x-axis deviation in the y=-100µm section, which can easily lead to insufficient positioning accuracy or even misjudgment. However, the algorithm proposed in this invention, based on high-plane data, extrapolates the z=85µm plane |H y The distribution of peak values ​​perfectly matches the key nodes TSV(1,3) and TSV(3,2) of the model signal transmission channel, and the amplitude is closer to the theoretical value. This is of great value for achieving refined EMI diagnosis.

[0174] Table 4 | H y | Comparison of peak information and TSV location information

[0175]

[0176] Example 4

[0177] To verify the practical performance of the proposed dual-plane near-field scanning algorithm, a real-world test case is presented. A test platform is provided, mainly consisting of a near-field scanner, a magnetic near-field probe, a computer, a signal analyzer, and a vector network analyzer. The Langer EMV-Technik high-frequency magnetic near-field probe, with a maximum operating frequency of 6 GHz, is used in the experiment. The measured data is corrected using the calibration factor provided by the manufacturer. The tested object is a hybrid filter chip based on TSV technology packaging, with a package size of 1.7 mm × 1.9 mm, an operating frequency of 2.9 GHz, and signal filtering capabilities.

[0178] During the measurement process, the chip mounting plane was used as the reference plane (z = 0 mm). The probe was controlled by computer scanning software to collect data from the z = 2.5 mm and z = 3 mm planes as input for the dual-plane near-field scanning algorithm, based on which the radiation field at z = 2 mm was deduced. Simultaneously, the measured radiation value at z = 2 mm was used for algorithm verification. The scanning area was set to 5 mm × 7 mm, with a step size Δx = Δy = 0.1 mm.

[0179] Figure 15 Demonstrates actual measurement of z=2mm|H x |、z=2.5mm Actual Measurement|H x | Derivation of the Dual-Plane Near-Field Scanning Algorithm | H x | and formula (7) directly calculate |H x | Comparison results (all are for the z=2mm plane). All data were processed with noise reduction filtering (using a [8,8] local field window, noise variance noise_var=5.58e-10). Figure 15 A comparison of (b) and (c) shows that, compared to the direct derivation of the formula, the algorithm proposed in this invention yields calculation results that highly match the measured data, and the radiation area perfectly matches the chip packaging location. It is worth noting that |H| is scanned in a z = 2.5 mm plane. x The value is significantly lower than the radiation intensity of the z=2mm plane, resulting in a serious deterioration in electromagnetic imaging quality and creating a substantial obstacle to the accurate positioning of areas with strong radiation.

[0180] Table 5 below lists the near-field |H values ​​of the tested chip at locations with high radiation intensity (x = 1.8 mm, x = 3.2 mm and y = 5.5 mm, y = 3.9 mm). x Peak data, these coordinate points correspond to the external pin positions of the TSV chip package. Comparative analysis shows that the near-field values ​​of the lower plane derived by the dual-plane near-field scanning algorithm have better numerical consistency with the measured values ​​of z=2mm compared to the data of the z=2.5mm plane. Particularly noteworthy is the high plane |H x The location of the maximum radiation intensity determined by the data has a positioning offset, while the H value derived by the proposed algorithm is different. x The distribution is completely consistent with the peak position of the measured data of z=2mm.

[0181] Table 5 Measured Chip | H x Comparison of peak information between calculated and scanned values ​​|

[0182]

[0183] The embodiments 1 to 4 described above can achieve precise localization of electromagnetic interference sources at the PCB and chip levels. Addressing the current issues of cumbersome testing procedures and high dependence on structurally similar calibration components, this invention proposes a near-field-near-field transformation technique based on dual-plane near-field acquisition. This technique utilizes simulated or measured dual-plane near-field scanning results of the device under test (DUT) and inputs them into an algorithm to obtain an extrapolated lower-plane radiation field value, thereby enabling more accurate identification of the interference source location and radiation information. This method achieves high-precision radiation distribution reconstruction using simpler testing procedures than traditional methods. Furthermore, experimental verification of PCB-level radiation source localization further validates the applicability of chip-level radiation sources, including pin and bond wire radiation from chip packages and TSV-packaged chips, filling the technological gap in high-precision localization of 3D-packaged chip-level radiation sources.

[0184] In summary, the chip-level electromagnetic interference localization method and device based on dual-plane near-field sampling described in this invention acquires spatial domain radiation data on the first and second sampling planes respectively. Then, using planar spectral expansion theory, it establishes the spectral domain transformation relationship between the two planes and the target detection plane, calculates the height calibration factor, and inverts the target plane field distribution, ultimately forming a high-resolution electromagnetic image. This method uses scanning area, scanning step size, and scanning height difference as core parameters, and precisely constrains sampling conditions based on the Nyquist criterion and cutoff frequency, effectively suppressing spectral leakage and aliasing errors. Simultaneously, it requires no additional calibration components, significantly simplifying the testing process and eliminating structural dependence. Compared to traditional direct transformation or calibration factor methods, this invention achieves peak position errors of less than one-tenth of a wavelength and reduces the root mean square error of amplitude by more than 50% in the localization of microscale radiation sources such as chip package pins, bonding wires, and three-dimensional through-silicon vias. It can quickly and non-contactly achieve precise millimeter-level interference source localization, meeting the electromagnetic compatibility diagnostic requirements of advanced packaged chips.

[0185] Those skilled in the art will understand that the exemplary components, systems, and methods described in conjunction with the embodiments disclosed herein can be implemented in hardware, software, or a combination of both. Whether implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this invention. When implemented in hardware, it can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this invention are programs or code segments used to perform the desired tasks. The programs or code segments can be stored in a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried in a carrier wave.

[0186] It should be clarified that the present invention is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of the present invention. In the present invention, features described and / or illustrated for one embodiment can be used in the same or similar manner in one or more other embodiments, and / or combined with or replacing features of other embodiments.

[0187] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, various modifications and variations of the embodiments of the present invention are possible. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A chip-level electromagnetic interference localization method based on dual-plane near-field sampling, characterized in that, The method includes the following steps: The scanning parameters are set according to the set rules. The scanning parameters include the scanning area, the scanning step size, and the scanning height difference between the first sampling plane and the target detection plane. The scanning area is set according to the operating frequency and cutoff frequency of the device under test to determine the upper frequency limit of the sampled magnetic field spectrum. The scanning step size is limited according to the Nyquist sampling theorem and the wave vector corresponding to the cutoff frequency to prevent aliasing. The scanning height difference is set based on the height of the first sampling plane combined with the operating frequency and the cutoff frequency of the device under test. The scanning height difference is superimposed on the height of the first sampling plane to determine a second sampling plane at a corresponding height. Acquire first spatial domain electromagnetic radiation data collected by a near-field probe in the first sampling plane according to the scanning parameters, and second spatial domain electromagnetic radiation data collected in the second sampling plane according to the scanning parameters; The first spatial domain electromagnetic radiation data is converted into the first spectral domain electromagnetic radiation data, and the second spatial domain electromagnetic radiation data is converted into the second spectral domain electromagnetic radiation data. Based on the predefined conversion relationship of electromagnetic radiation data in the spectral domain between the target sampling plane, the first sampling plane and the second sampling plane according to the plane spectral expansion theory, the quotient of the electromagnetic radiation data in the second spectral domain and the electromagnetic radiation data in the first spectral domain is used as the height calibration factor. The target spectral domain electromagnetic radiation data of the target detection plane is solved by combining the height calibration factor and the electromagnetic radiation data in the first spectral domain. The electromagnetic radiation data in the target spectral domain is transformed into electromagnetic radiation data in the target spatial domain to obtain electromagnetic imaging results, and the interference source is located by the amplitude of the electromagnetic radiation.

2. The chip-level electromagnetic interference localization method based on dual-plane near-field sampling according to claim 1, characterized in that, The constraints for setting the scanning area are as follows: Where A represents the scanning area, k0 represents the wave vector calculated by the device under test based on the operating frequency, and k cutoff This represents the wave vector calculated based on the cutoff frequency.

3. The chip-level electromagnetic interference localization method based on dual-plane near-field sampling according to claim 2, characterized in that, The constraints for setting the scanning step size are as follows: Where, k cutoff This represents the wave vector calculated based on the cutoff frequency.

4. The chip-level electromagnetic interference localization method based on dual-plane near-field sampling according to claim 3, characterized in that, The rule for setting the scanning height difference is expressed as follows: Where Δd represents the scanning height difference, d #2-DUT λ0 represents the height difference between the first sampling plane and the target sampling plane, λ0 is the wavelength corresponding to the operating frequency of the device under test, and k0 represents the wave vector calculated by the device under test based on the operating frequency. cutoff This represents the wave vector calculated based on the cutoff frequency.

5. The chip-level electromagnetic interference localization method based on dual-plane near-field sampling according to claim 1, characterized in that, The quotient of the second spectral domain electromagnetic radiation data and the first spectral domain electromagnetic radiation data is used as the height calibration factor, calculated as follows: in, This represents the electromagnetic radiation data in the second spectral domain, that is, the spectral domain form of the magnetic field component at the second sampling plane; The first spectral domain electromagnetic radiation data represents the spectral domain form of the magnetic field component at the first sampling plane; z2 represents the height of the first sampling plane, z3 represents the height of the second sampling plane, and k represents the height of the second sampling plane. x Let k be the wave vector component along the x-axis. y denoted as the wave vector component along the y-axis; j represents the ordinal unit; Δd represents the scan height difference; e represents the natural base; η represents the height calibration factor; The formula for calculating the wave vector k is: and Where ω is the angular frequency, μ is the permeability of the medium, and ε is the dielectric constant of the medium; and It represents a unit vector in a rectangular coordinate system.

6. The chip-level electromagnetic interference localization method based on dual-plane near-field sampling according to claim 5, characterized in that, The target electromagnetic radiation data of the target detection plane is calculated by combining the height calibration factor and the first spectral domain electromagnetic radiation data. The calculation formula is as follows: in, The z1 represents the electromagnetic radiation data in the target spectral domain, i.e., the spectral domain form of the magnetic field component at the target detection plane; the k represents the height of the target detection plane. x Let k be the wave vector component along the x-axis. y Let η be the wave vector component along the y-axis, and let η represent the height calibration factor.

7. The chip-level electromagnetic interference localization method based on dual-plane near-field sampling according to claim 1, characterized in that, The method further includes: applying Wiener filtering to the electromagnetic imaging results, calculated as follows: Where I(m,n) represents the data before filtering, J(m,n) represents the data after filtering, noise_var represents the estimated noise variance, μ represents the mean of the data before filtering, and σ represents the standard deviation of the data before filtering.

8. A chip-level electromagnetic interference positioning device, characterized in that, The device includes: A support platform is used to load the device under test; the support platform is provided with a first slide rail along the horizontal X-axis. The first support member is slidably connected to the bearing platform via the first slide rail and is driven by the first motor; the first support beam is provided with a second slide rail along the Y-axis of the horizontal plane; The second support member is slidably connected to the first support member via the second slide rail and is driven by the second motor; the second support member is provided with a third slide rail along the Z-axis direction; The third support member is slidably connected to the second support member via the third slide rail and is driven by the third motor; A near-field probe, fixed on the third support, is used to collect electromagnetic radiation data in the first spatial domain on the first sampling plane and electromagnetic radiation data in the second spatial domain on the second sampling plane according to preset scanning parameters. The scanning parameters include scanning area, scanning step size, and the scanning height difference between the first sampling plane and the target detection plane. The processor module is connected to control the first motor drive, the second motor drive, the third motor drive, and the near-field probe. The processor module is also used to execute the chip-level electromagnetic interference localization method based on dual-plane near-field sampling as described in any one of claims 1 to 7 to obtain the interference source localization result of the device under test.

9. A computer-readable storage medium having a computer program / instructions stored thereon, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method as described in any one of claims 1 to 7.

10. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method according to any one of claims 1 to 7.

Images