A method for phase balance control of a high-speed line layout
By adjusting the trace center distance and linewidth within the phase compensation range, the problems of phase mismatch and impedance fluctuation in non-homogeneous substrates are solved, achieving phase alignment and impedance stability for high-speed signal transmission and ensuring signal quality.
Patent Information
- Application Number
- CN202611058700.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-16
- Publication Date
- 2026-08-25
AI Technical Summary
Existing technologies struggle to achieve phase balance control under impedance continuity constraints in heterogeneous substrates, leading to phase mismatch and impedance fluctuations during signal transmission, which is particularly difficult to resolve in ultra-high-speed signal transmission.
By synchronously adjusting the trace center distance and single-ended line width within the phase compensation range, the difference in odd mode propagation speed caused by the center distance change is used to offset the accumulated phase skew, and the parasitic capacitance disturbance caused by the spacing change is offset by the reverse configuration of the line width, ensuring that the differential impedance is stable at the target nominal value.
It achieves phase-aligned energy transmission continuity in high-frequency signal transmission, avoids secondary high-frequency dispersion and mode conversion, shortens the total transmission path length and reduces the risk of common-mode radiation, and ensures the clarity of the signal eye diagram.
Smart Images

Figure CN122640930A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of hybrid substrate wiring technology, and particularly relates to a phase balance control method for high-speed line layout. Background Technology
[0002] As signal transmission rates evolve towards 112G and 224G, the phase consistency of printed circuit boards (PCBs), as the physical carriers of differential signals, has become a core indicator for ensuring eye diagram quality. The industry typically adopts the principle of geometrically equal length, using serpentine winding structures in the trace paths to compensate for physical length differences and achieve synchronous transmission of differential pair signals. However, PCB substrates are composed of periodically woven glass fiber bundles filled with epoxy resin, exhibiting significant heterogeneity in their surface structure. Existing methods, based on the ideal assumption of isotropic dielectric constant, ignore the spatial fluctuations in the ratio of glass fiber to resin within the trace coverage area. This glass fiber weaving effect causes a difference in the effective dielectric constant perceived by the two traces within the differential pair, resulting in electrical time delay hysteresis. Consequently, traces of equal physical length exhibit phase mismatch in electrical performance.
[0003] Traditional geometric compensation methods have fundamental limitations in handling such inhomogeneous skew. Simply increasing the physical length of the serpentine winding to force equal length alters the original electromagnetic edge coupling state of the differential pair, inducing impedance discontinuities and increasing common-mode noise radiation at bends. Because the compensation range setting lacks awareness of the substrate's physical pitch, the expanded electromagnetic field often cuts through the boundary region where the dielectric constant fluctuates drastically, causing signal dispersion. Relying on external overall configuration adjustments is limited by engineering boundaries. In ultra-high-speed signal interconnect systems, the underlying wiring control methods are insufficient and struggle to solve the inhomogeneous skew problem. For example, Chinese invention patent CN223142211U discloses a printed circuit board and electronic device that adjusts the circuit pattern relative to the substrate dielectric layer glass fiber bundle by 5 degrees. An 8-degree deflection creates an acute angle between the trace and the fiber bundle, allowing the trace to pass through the fiber mesh and reducing the time delay difference within the differential pair. This solution relies on idealized underlying properties, namely that the entire board routing channel has unlimited rotation space and is not affected by external rigid connection nodes. In actual high-density interconnection conditions, the chip pin arrangement direction and the physical assembly position of the high-speed connector are absolutely anchored. Forcibly implementing an overall deflection of the entire board circuit pattern leads to a decrease in panel utilization and causes physical interference of traces in local high-density lead-out areas. This static global geometric alignment mechanism has a mismatch between its core preset premise and the physical boundary conditions of the actual complex interconnection topology. It cannot dynamically adapt to random fluctuations in local intermediate gradient and cannot handle complex shaped areas with bends and turns, making it difficult for existing technologies to achieve phase self-consistency under impedance continuity constraints.
[0004] Therefore, how to couple the trace topology with the physical pitch of the substrate to achieve phase balance control under impedance continuity constraints is the technical problem to be solved by this invention. Summary of the Invention
[0005] The present invention aims to solve the problem of impedance fluctuation caused by phase skew repair of differential printed circuit lines in non-homogeneous substrates.
[0006] In this technical solution, a phase balance control method for high-speed line layout includes the following steps: Step S101: Obtain the physical parameters of the substrate of the printed circuit board and the nominal value of the target impedance. The physical parameters of the substrate include the fiber weaving pitch and the fiber cloth orientation. Step S102: Determine the initial wiring trajectory of the differential pair. Based on the relative angle between each straight line segment in the initial wiring trajectory and the fiber cloth direction, calculate the effective dielectric constant difference between the locations of the first and second traces in the differential pair, and determine the cumulative phase skew of the differential pair. Step S103: In the initial wiring trajectory, a phase compensation interval defined by the physical start coordinate and the physical end coordinate is set. The physical start coordinate and the physical end coordinate are respectively locked on the geometric center axis of the fiber bundle inside the printed circuit board or the geometric center axis of the resin enrichment area, so that the physical length of the phase compensation interval is half an integer multiple of the length of the fiber braid pitch mapped in the direction of the initial wiring trajectory. Step S104: Based on the accumulated phase skew and physical length, adjust the center distance of the differential traces within the phase compensation interval to generate a phase cancellation amount to compensate for the accumulated phase skew; simultaneously, based on the change in the value of the differential trace center distance, synchronously and in reverse adjust the single-end line width of the first and second traces within the phase compensation interval, using the change in distributed inductance and distributed capacitance generated by the line width adjustment to offset the impedance fluctuation caused by the change in the differential trace center distance, so that the differential impedance within the phase compensation interval is stabilized at the target impedance nominal value.
[0007] Preferably, step S104, which involves synchronously and inversely adjusting the single-ended linewidth, includes: establishing a correlation mapping model between the differential trace center distance, the single-ended linewidth, and the differential impedance; when the cumulative phase skew indicates that the outer trace is ahead, increasing the differential trace center distance within the phase compensation interval, and increasing the single-ended linewidth according to the correlation mapping model, using the reduction in distributed inductance caused by the increased linewidth and the increase in capacitance to ground to offset the decrease in capacitive coupling caused by the increase in differential trace center distance, thereby limiting the differential impedance deviation within the phase compensation interval to within 3% of the target impedance nominal value.
[0008] Preferably, locking the physical start coordinates and physical end coordinates in step S103 includes: obtaining the spatial distribution gradient of dielectric constant inside the printed circuit board using a preset electromagnetic scanning algorithm; identifying linear regions in the spatial distribution gradient of dielectric constant with a rate of change lower than a preset threshold, and defining the linear regions as physical safety boundaries; mapping the physical start coordinates and physical end coordinates to the physical safety boundaries to prevent the expanded edge electric field in the phase compensation interval from cutting the interface between the fiber bundle and the resin-rich area.
[0009] Preferably, after step S104, the method further includes a step of processing phase skew in the corner region: identifying the bend corner region in the initial wiring trajectory and determining the physical path difference between the inner and outer traces in the bend corner region; within the bend corner region, asymmetrically adjusting the line width of the inner and outer traces, and generating a reverse phase shift by changing the unit length distributed inductance of the inner and outer traces to counteract the time skew caused by the physical path difference.
[0010] Preferably, obtaining the fiber weaving pitch in step S101 includes: measuring the warp fiber density and weft fiber density of the sample substrate using a scanning electron microscope; calculating the orthogonal mapping component of the fiber weaving pitch in the initial wiring trajectory direction based on the warp fiber density and weft fiber density; and calibrating the physical length of the phase compensation interval based on the orthogonal mapping component in step S103.
[0011] Preferably, adjusting the center distance of the differential traces in step S104 includes: obtaining the signal wavelength corresponding to the differential signal transmission rate; setting the physical length of the phase compensation interval to 0.25 to 0.5 times the signal wavelength; and changing the center distance of the differential traces within the phase compensation interval using a gradient microstrip line structure.
[0012] Preferably, after step S104, the method further includes: monitoring the common-mode noise conversion rate of the differential pair at a preset operating frequency; when the common-mode noise conversion rate exceeds a preset threshold, obtaining the dielectric loss distribution within the phase compensation interval; and finely adjusting the ratio of the single-end linewidth to the center distance of the differential traces according to the dielectric loss distribution to balance the transverse electric field coupling strength between the two branches of the differential pair.
[0013] Preferably, obtaining the physical parameters of the substrate in step S101 includes: establishing a hybrid dielectric model based on the resin enrichment region and the fiber bundle region; determining the angle between the initial wiring trajectory and the fiber cloth direction; determining the spatial fluctuation period of the effective dielectric constant below the initial wiring trajectory based on the angle and the hybrid dielectric model, and using the spatial fluctuation period as the basis for setting the step length of the phase compensation interval.
[0014] Preferably, after step S104, the following steps are also included: Step S105, performing differential signal eye diagram quality verification for the phase compensation interval, and extracting eye height parameters and jitter parameters; Step S106, if the eye height parameter is lower than the nominal preset value or the jitter parameter is higher than the safety preset value, then return to step S103 to adjust the physical starting coordinates until the eye diagram quality verification meets the preset high-speed transmission standard.
[0015] Furthermore, compared to existing technologies, the phase balance control method for high-speed line layout of the present invention has the following advantages: 1. In the phase balance control of high-speed line layout, this method eliminates the physical conflict between phase compensation and impedance matching by synchronously adjusting the trace center distance and single-ended line width within the phase compensation range. It uses the difference in odd mode propagation speed caused by the change in center distance to offset the accumulated phase skew. At the same time, it uses the reverse configuration of line width to offset the parasitic capacitance disturbance caused by the change in spacing, so that the differential impedance is always maintained at the target nominal value. This control method changes the limitation of impedance mismatch caused by phase repair in traditional technology, and ensures the continuity of energy transmission of high-frequency signals during the phase alignment process.
[0016] 2. To avoid secondary high-frequency dispersion and mode conversion, this method forcibly anchors the physical start and end boundaries of the phase compensation interval to the geometric center axis of the substrate fiber bundle or resin-rich region. This allows the edge crosslinking electric field, which expands due to topological abrupt changes, to be completely contained within a physical space with a gentle dielectric constant. Through this physical lattice boundary alignment mechanism, the expanding electric field is prevented from cutting the boundary region where the dielectric constant fluctuates drastically. This cuts off the physical causes of high-frequency dispersion and parasitic mode generation at the surface structure level, ensuring the clarity of the signal eye diagram at speeds above 112G.
[0017] 3. Achieving in-situ repair of phase skew in corner regions: This method utilizes the asymmetrical configuration of the linewidth of the traces inside and outside the differential corner region. By changing the distributed inductance and capacitance per unit length of the trace, the actual propagation speed of electromagnetic waves in this region is adjusted. The reverse phase shift generated by this wave speed control mechanism can directly offset the physical delay caused by the difference in the geometric path of the corner. There is no need to add an extra serpentine winding structure in the straight wiring area. This not only shortens the total transmission path length, but also reduces the risk of additional common-mode radiation introduced by complex winding. Attached Figure Description
[0018] Figure 1 This is a logic flowchart of impedance and phase coupling adjustment in the phase balance control method of the present invention; Figure 2 This is a functional architecture diagram of the wiring control engine and physical scanning of non-homogeneous substrates of the present invention. Detailed Implementation
[0019] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.
[0020] It should be noted that all directional and positional terms used in this invention, such as: up, down, left, right, front, back, vertical, horizontal, inner, outer, top, bottom, transverse, longitudinal, center, etc., are only used to explain the relative positional relationship and connection between components in a specific state (as shown in the accompanying drawings). They are only for the convenience of describing this invention and do not require that this invention be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting this invention. In addition, the descriptions of "first," "second," etc., in this invention are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated.
[0021] In the description of this invention, unless otherwise explicitly specified and limited, the terms installation, connection, and linking should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; they can refer to mechanical connections; they can refer to direct connections or indirect connections through an intermediate medium; they can refer to the internal connection of two components. For those skilled in the art, the specific meaning of the above terms in this invention can be understood according to the specific circumstances.
[0022] In the description of this specification, references to the terms "an embodiment," "some embodiments," "illustrative embodiments," "examples," "specific examples," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example, and the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.
[0023] A phase balance control method for high-speed line layout includes the following steps: Step S101: Obtain the physical parameters of the substrate of the printed circuit board and the nominal value of the target impedance. The physical parameters of the substrate include the fiber weaving pitch and the fiber cloth orientation. Step S102: Determine the initial wiring trajectory of the differential pair. Based on the relative angle between each straight line segment in the initial wiring trajectory and the fiber cloth direction, calculate the effective dielectric constant difference between the locations of the first and second traces in the differential pair, and determine the cumulative phase skew of the differential pair. Step S103: In the initial wiring trajectory, a phase compensation interval defined by the physical start coordinate and the physical end coordinate is set. The physical start coordinate and the physical end coordinate are respectively locked on the geometric center axis of the fiber bundle inside the printed circuit board or the geometric center axis of the resin enrichment area, so that the physical length of the phase compensation interval is half an integer multiple of the length of the fiber braid pitch mapped in the direction of the initial wiring trajectory. Step S104: Based on the accumulated phase skew and physical length, adjust the center distance of the differential traces within the phase compensation interval to generate a phase cancellation amount to compensate for the accumulated phase skew; simultaneously, based on the change in the value of the differential trace center distance, synchronously and in reverse adjust the single-end line width of the first and second traces within the phase compensation interval, using the change in distributed inductance and distributed capacitance generated by the line width adjustment to offset the impedance fluctuation caused by the change in the differential trace center distance, so that the differential impedance within the phase compensation interval is stabilized at the target impedance nominal value.
[0024] Preferably, step S104, which involves synchronously and inversely adjusting the single-ended linewidth, includes: establishing a correlation mapping model between the differential trace center distance, the single-ended linewidth, and the differential impedance; when the cumulative phase skew indicates that the outer trace is ahead, increasing the differential trace center distance within the phase compensation interval, and increasing the single-ended linewidth according to the correlation mapping model, using the reduction in distributed inductance caused by the increased linewidth and the increase in capacitance to ground to offset the decrease in capacitive coupling caused by the increase in differential trace center distance, thereby limiting the differential impedance deviation within the phase compensation interval to within 3% of the target impedance nominal value.
[0025] Preferably, locking the physical start coordinates and physical end coordinates in step S103 includes: obtaining the spatial distribution gradient of dielectric constant inside the printed circuit board using a preset electromagnetic scanning algorithm; identifying linear regions in the spatial distribution gradient of dielectric constant with a rate of change lower than a preset threshold, and defining the linear regions as physical safety boundaries; mapping the physical start coordinates and physical end coordinates to the physical safety boundaries to prevent the expanded edge electric field in the phase compensation interval from cutting the interface between the fiber bundle and the resin-rich area.
[0026] Preferably, after step S104, the method further includes a step of processing phase skew in the corner region: identifying the bend corner region in the initial wiring trajectory and determining the physical path difference between the inner and outer traces in the bend corner region; within the bend corner region, asymmetrically adjusting the line width of the inner and outer traces, and generating a reverse phase shift by changing the unit length distributed inductance of the inner and outer traces to counteract the time skew caused by the physical path difference.
[0027] Preferably, obtaining the fiber weaving pitch in step S101 includes: measuring the warp fiber density and weft fiber density of the sample substrate using a scanning electron microscope; calculating the orthogonal mapping component of the fiber weaving pitch in the initial wiring trajectory direction based on the warp fiber density and weft fiber density; and calibrating the physical length of the phase compensation interval based on the orthogonal mapping component in step S103.
[0028] Preferably, adjusting the center distance of the differential traces in step S104 includes: obtaining the signal wavelength corresponding to the differential signal transmission rate; setting the physical length of the phase compensation interval to 0.25 to 0.5 times the signal wavelength; and changing the center distance of the differential traces within the phase compensation interval using a gradient microstrip line structure.
[0029] Preferably, after step S104, the method further includes: monitoring the common-mode noise conversion rate of the differential pair at a preset operating frequency; when the common-mode noise conversion rate exceeds a preset threshold, obtaining the dielectric loss distribution within the phase compensation interval; and finely adjusting the ratio of the single-end linewidth to the center distance of the differential traces according to the dielectric loss distribution to balance the transverse electric field coupling strength between the two branches of the differential pair.
[0030] Preferably, obtaining the physical parameters of the substrate in step S101 includes: establishing a hybrid dielectric model based on the resin enrichment region and the fiber bundle region; determining the angle between the initial wiring trajectory and the fiber cloth direction; determining the spatial fluctuation period of the effective dielectric constant below the initial wiring trajectory based on the angle and the hybrid dielectric model, and using the spatial fluctuation period as the basis for setting the step length of the phase compensation interval.
[0031] Preferably, after step S104, the following steps are also included: Step S105, performing differential signal eye diagram quality verification for the phase compensation interval, and extracting eye height parameters and jitter parameters; Step S106, if the eye height parameter is lower than the nominal preset value or the jitter parameter is higher than the safety preset value, then return to step S103 to adjust the physical starting coordinates until the eye diagram quality verification meets the preset high-speed transmission standard.
[0032] Example 1: In a high-speed backplane cabling design with a transmission rate of 224 GPA M4, the physical length of the differential link is 500 mm. The backplane uses a hybrid substrate with a periodic glass fiber braided structure, and the first and second traces of the differential pair experience a cumulative phase skew of 2.5 ps due to the glass fiber braiding effect. The system reads the physical parameters of the hybrid substrate and determines the fiber weaving pitch. The fiber diameter is 0.4 mm and the fiber direction forms a 5° angle with the initial wiring trajectory of the differential pair. The angle and fiber weaving pitch are then considered. The effective dielectric constant difference between the locations of the first and second traces was calculated. To determine the path for obtaining the dielectric characteristics of heterogeneous substrates, the system establishes a volume projection calculation model based on the Maxwell-Garnett equivalent dielectric theory. The processing unit reads the proportion of resin cross-sectional area and glass fiber bundle cross-sectional area in the vertical projection plane of the initial wiring trajectory on the hybrid substrate. Combined with the external optical scanning module to pre-obtain the intrinsic dielectric constant of pure resin and the intrinsic dielectric constant of pure glass fiber, the system substitutes them into the hybrid dielectric volume weighted equation, solves the equivalent dielectric constant values of the first and second traces in specific three-dimensional spatial coordinates, and performs differential operations to extract the final dielectric constant difference.
[0033] Regarding the cumulative phase skew of 2.5 ps Without changing the total physical length of the differential pair, a straight wiring area is selected as the phase compensation interval in the initial wiring trajectory. Electromagnetic scanning logic identifies the spatial distribution gradient of the dielectric constant within the hybrid substrate, and linear regions with dielectric constant change rates below a preset threshold are extracted as physical safety boundaries. The physical start and end coordinates of this phase compensation interval are locked on the geometric center axis of the fiber bundle or the geometric center axis of the resin-rich area within the hybrid substrate, respectively, ensuring the physical length of the phase compensation interval is... satisfy The constraint relationship, where, This represents the physical length of the phase compensation interval. It is a positive integer, in this embodiment Select 20. The physical length obtained by weaving the fiber pitch. The value is 4mm. Within the aforementioned 4mm phase compensation range, which is physically lattice locked, according to the formula... Determine the required odd-mode phase velocity adjustment amount and adjust the center distance between the first and second traces, wherein... To accumulate phase skew, The signal angular frequency, This represents the physical length of the phase compensation interval. The effective dielectric constant difference, For the speed of light in a vacuum, when the cumulative phase skew... When the outward-pointing trace leads, the center distance within the phase compensation interval is increased. Based on the correlation mapping model between center distance, single-ended linewidth, and differential impedance, the single-ended linewidths of the first and second traces are simultaneously increased. The increase in capacitance to ground generated by the increased single-ended linewidth offsets the decrease in capacitive coupling caused by the increased center distance, keeping the differential impedance deviation within the phase compensation interval within 3% of the target impedance nominal value. When constructing the correlation mapping model, this invention pre-performs multi-dimensional simulations based on a two-dimensional electromagnetic field solver for the current hybrid substrate's stack parameters, extracting the slope of the effect of center distance variation on the distributed capacitance per unit length under different dielectric thicknesses. This model uses the target impedance... The nominal value of the impedance is a constant in the constraint equation. The compensation relationship between the center distance variable and the linewidth variable is established by numerical lookup table method or polynomial fitting algorithm. When the center distance increases and causes the differential pair odd mode capacitance to decrease, the model calculates the required linewidth compensation increment according to the preset sensitivity matrix. The increase in capacitance to ground caused by the increase in linewidth compensates for the reduction in coupling capacitance caused by the center distance expansion. In the actual control process, the wiring engine calls the calculation results of the image model in real time and directly maps the change in geometric parameters to the physical size adjustment command of the microstrip line structure to ensure that the characteristic impedance per unit length of the transmission line is always locked within the process tolerance range during the phase alignment process.
[0034] By anchoring the physical start and end coordinates of the phase compensation interval to the central axis of the fiber bundle or the central axis of the resin-rich region where the dielectric gradient is stable, the edge crosslinking electric field expansion induced by the center distance adjustment is contained within a relatively homogeneous dielectric space. This avoids secondary high-frequency dispersion and parasitic mode transitions caused by the high-frequency electric field cutting the transition interface between the fiber bundle and the resin-rich region. The wiring trajectory maintains the eye diagram characteristics at the receiver at a rate of 224G, and the total insertion loss is reduced by 1.2dB compared to the geometric serpentine equal-length compensation method. This achieves a wiring topology that is consistent with non-linearity. The system employs three-dimensional physical alignment of the physical period of a homogeneous dielectric substrate to address common-mode switching impedance degradation caused by surface dielectric inhomogeneity under high-speed operating conditions. It integrates a dynamic quantization and compensation procedure for dielectric loss extraction based on time-frequency domain reflection. A vector network analyzer with a sampling bandwidth of at least 50GHz is used to apply a broadband sweep excitation signal to the differential pair input node, acquiring single-ended insertion loss parameters for the first and second traces. Based on the inherent property of microwave transmission line attenuation—that conductor loss is directly proportional to the square root of frequency while dielectric loss is linearly proportional to frequency—the processing unit... The high-frequency linear component in the insertion loss decay curve is separated and fitted, and the scalar of dielectric loss spatial distribution within the phase compensation interval is extracted. It should be noted that the above-mentioned experimental procedure based on the network analyzer is mainly applied to the process calibration stage before mass production or to the verification sample for a specific batch of substrates. The system performs time-frequency domain analysis on the dielectric loss characteristics of the verification sample to establish a loss deviation model unique to the batch of materials. In the layout design stage of the formal product, the routing engine uses this loss deviation model as a feedforward constraint and performs preventive asymmetric linewidth pre-correction before Gerber data is generated. Through this closed-loop feedback logic of experimental calibration, model feedforward and layout pre-correction, the common-mode noise risk that may occur in the later finished product is offset in advance at the design stage through the differential configuration of geometric topology. When the difference in dielectric loss between the two branches is continuously monitored and drives the common-mode noise conversion rate to exceed the preset threshold boundary, the system asymmetrically fine-tunes the single-end linewidth ratio of the two traces according to the difference in dielectric loss spatial distribution scalar value. By increasing the conductive cross-sectional area of the high-loss physical branch, the skin effect additional resistance is weakened, and the transverse electric field coupling energy dissipation bias is offset.
[0035] Example 2: In the physical verification of a high-speed link with a transmission rate of 224 GPA M4, a hybrid substrate with 6 layers and a thickness of 1.2 mm was used as the carrier. A vector network analyzer with a sampling bandwidth of 110 GHz and a real-time oscilloscope with a sampling rate of 200 GS / s were used to acquire the differential signal characteristics. Gaussian white noise with a signal-to-noise ratio of 22 dB and random power supply ripple interference with an amplitude of 5 mV were superimposed at the signal excitation source. A 500 mm long differential trace was selected as the observation object, and the fiber braiding pitch of the hybrid substrate was determined by high-resolution microsectioning. The initial cumulative phase skew of the first and second traces was measured to be 0.402 mm, with an initial wiring trajectory and fiber direction angle of 5.1°. It is 2.53ps.
[0036] To verify the phase balance control effect, a control system containing three experimental gradients was established. The sample group of this invention applied the physical alignment logic of the phase compensation interval of a straight line segment, adjusting the physical length of the phase compensation interval... The setting is 4.02mm, which meets the requirements. A physical lattice constraint of 20 was used, and the physical start and end coordinates were locked on the central axis of the fiber bundle. Within this phase compensation interval, the trace center distance was increased from 0.15 mm to 0.18 mm, and the single-end linewidth was simultaneously compensated from 0.12 mm to 0.132 mm. The first control group used a non-aligned length compensation method, setting the physical length of the phase compensation interval to 3.85 mm, which did not satisfy the half-integer multiple relationship of the fiber weaving pitch, and the physical boundary was located at the transition interface between the fiber bundle and the resin-rich region. The second control group used a partial missing scheme, adjusting only the trace center distance without adjusting the single-end linewidth. The third control group used a geometric serpentine equal-length winding method. The effective dielectric constant difference output by the sample group of this invention was... Within the phase compensation range, the differential impedance exhibits a linear distribution due to the locking of the physical boundaries. The fluctuation range of the differential impedance obtained by time-domain reflection testing within this phase compensation range is 98.7%. Up to 101.4 Between, the residual cumulative phase skew measured at the final output terminal The insertion loss was reduced to 0.16 ps, the eye opening of the receiver eye diagram reached 0.72 UI, and the total insertion loss was measured at 15.2 dB at the 56 GHz Nyquist frequency. In the first control group, the phase compensation interval boundary cuts through the substrate dielectric gradient transition region, inducing edge electric field dispersion, resulting in residual accumulated phase skew. It has a resonant loss of 0.68 ps and generates an additional 1.8 dB at 56 GHz.
[0037] Data regarding the continuity of differential impedance indicate that, in the second control group, after increasing the trace center distance, the differential impedance jumps to 114.6 within the phase compensation range due to the lack of capacitive compensation for the single-ended linewidth. This results in a reduction in reflection loss to 8.5 dB at 56 GHz, degrading link signal integrity. However, the prototype of this invention utilizes the increased capacitance to ground caused by the increased single-ended linewidth to offset the decrease in capacitive coupling caused by the increased trace center distance, thus maintaining the differential impedance deviation within the phase compensation interval at the target impedance nominal value of 100. Within 3%; by introducing a problem intensity gradient comparison, when the original cumulative phase skew amount When the phase shift increased from 2.53 ps to 5.06 ps, the sample group of this invention applied cascaded compensation over two consecutive 4.02 mm physical cycles, and the residual cumulative phase offset was... Stable within 0.22ps, with no performance degradation, physical length There is a nonlinear evolution relationship between the phase compensation effect and the phase compensation effect. When the deviation from the fiber braiding pitch exceeds 15% by half an integer multiple, the deviation rate between the phase compensation amount and the theoretical calculation value increases exponentially, indicating that the integral effect of the differential electric field on the heterogeneous medium has predictable monotonicity only within a specific physical period window. This invention anchors the physical start coordinate and physical end coordinate of the phase compensation interval on the central axis of the fiber bundle or the central axis of the resin-rich area where the medium gradient is gentle, thereby decoupling the wiring parameter adjustment from the physical structure of the substrate and achieving low time domain skew and signal pass-through at a transmission rate of 224G.
[0038] Example 3: In a hybrid substrate design scenario with 112GPAM4 high-speed interconnect links, differential traces pass through a glass fiber reinforced resin substrate with cross-braiding characteristics. Due to the heterogeneous dielectric distribution, the differential signal experiences phase skew and induces impedance discontinuities. The system initiates a physical fingerprint scanning procedure for the substrate, performing two-dimensional meshing on the dielectric space below the wiring traces. The lateral and longitudinal steps of the mesh are both equal to the fiber braiding pitch. 0.05 times; in this embodiment, The value is set to 0.4 mm, and the corresponding step value is set to 0.02 mm. The processing unit sequentially reads the intrinsic dielectric constant at each grid node. And according to the formula Calculate the spatial distribution gradient of the dielectric constant between adjacent grids, where, The gradient of the spatial distribution of the dielectric constant of the medium. The intrinsic dielectric constant is The coordinates are horizontal grid coordinates. The input data for the electromagnetic scanning algorithm, using vertical grid coordinates, comes from microwave induction measurements of the substrate surface and embedded reference layer. By extracting the phase change of the reflection coefficient of the high-frequency scanning probe under different spatial coordinates, the effective dielectric constant distribution spectrum at the corresponding grid location is obtained. The processing unit then processes the acquired discrete intrinsic dielectric constant E... nMapped to a two-dimensional physical coordinate system, the numerical rate of change between adjacent grid points is calculated using the finite difference method. The specific processing logic is as follows: spatial convolution operation is performed on the subdivided grid using the gradient operator to extract the gradient vector field reflecting the severity of the heterogeneity of the medium. This map not only contains the spatial extension trajectory of the glass fiber bundle, but also quantifies the dielectric abrupt change characteristics of the resin impregnation zone and the glass fiber interface, providing a deterministic numerical criterion for subsequent search of physical safety boundaries. Considering the difference in dimensional order between the sum of squares of the two partial derivatives and the standard deviation, the system performs a dimensional consistency conversion procedure before outputting the physical safety boundary. The processing unit performs an arithmetic square root operation on the spatial distribution gradient value of the dielectric constant of the medium output by the above formula to obtain the true amplitude of the scalar gradient, or simultaneously converts the extracted preset threshold into a square scalar dimension, establishing a closed-loop comparison benchmark with physical self-consistency, and eliminating the risk of misjudgment of the safety boundary due to dimensional mismatch.
[0039] The system determines the preset threshold using the following calibration method: a 50mm x 50mm sample area is selected in the non-wiring area of the hybrid substrate as a reference field, and the average dielectric constant of all grid nodes within this reference field is calculated. with standard deviation Set the preset threshold to the standard deviation. The threshold is set at 0.1 times the standard deviation. When the spatial distribution gradient of the dielectric constant of a medium within a continuous linear region is lower than this preset threshold, the region is determined to be a physical safety boundary. The selection of this 0.1 times standard deviation threshold is based on the quantitative assessment of the sensitivity of high-speed signals to dielectric fluctuations. Since the integral effect of the edge electric field of the 224G signal on dielectric constant fluctuations is approximately linear within a very small gradient, experimental statistics show that when the local dielectric gradient fluctuation is controlled within 10% of the overall standard deviation, the contribution of mode conversion loss induced by dielectric discontinuity to the total link loss will be reduced to below the background noise level. If the threshold is too high, the physical safety boundary cannot effectively avoid the abrupt interface between the fiber and the resin, leading to secondary jitter introduced during the phase repair process. If the threshold is too low, the locking area that meets the conditions will be too narrow, and the cabling engine will have difficulty completing physical anchoring in high-density cabling areas. In this embodiment, the fiber is extracted. A strip-shaped area with a width of 0.05mm on both sides of the central axis directly above the fiber bundle serves as the locking position for the physical start and end coordinates. To eliminate the influence of random drift of the glass fiber cloth during the pressing process on the locking accuracy, this invention establishes a dynamic compensation coordinate system by reserving L-shaped optical markers on the edge of the board. Before the actual locking action is executed, the system calls a high-resolution optical sensor to capture the true weaving phase of the substrate in the current production batch, calculates the spatial offset vector between the design coordinates and the actual fiber axis, and superimposes this offset vector onto the logic origin of the wiring engine in real time, so that the start and end coordinates of the phase compensation interval are physically aligned with the central axis of the actual fiber bundle. This in-situ locking mechanism ensures that even when the substrate undergoes nonlinear deformation, the expanded edge electric field can still be accurately contained within the preset homogeneous dielectric channel, and the physical length of the phase compensation interval is determined. Then, the system calculates the cumulative phase skew. Start-up impedance adaptive compensation logic, linewidth compensation increment According to the formula It is confirmed that, among them, This is the adjustment value for the single-ended line width. This is the adjustment value for the center distance of the trace. For the compensation operator, the compensation operator The curve is derived from the effect of center distance variation on distributed capacitance per unit length fitted in an offline simulation environment; in this embodiment... The value is 0.42. When the center distance of the traces is adjusted from 0.15mm to 0.17mm, that is... When the line width is 0.02mm, the calculated single-ended line width increases by 0.0084mm, and the differential impedance deviation within the phase compensation range is limited to 2.1% of the nominal value.
[0040] The physical start and end coordinates of the phase compensation interval are mapped to the fiber center axis with stable dielectric properties, avoiding the fiber gap region where the dielectric constant fluctuates. Combined with the calculation of the linewidth compensation increment, the impedance drop of high-frequency signals during the phase repair process is eliminated. The VSWR of the differential link at 28GHz frequency is reduced from 1.45 to 1.12, and the physical boundary of the phase compensation interval follows the physical lattice constraint. The mode conversion loss induced by dielectric discontinuity is reduced by 0.8dB, and the wiring topology is physically aligned with the physical period of the non-homogeneous substrate in three dimensions.
[0041] Example 4: In a scenario involving impedance parameter calibration of multiple batches of hybrid substrates, the system adjusts the compensation operator based on offline electromagnetic simulation and capacitance measurement data. Establish feature mapping for different medium thicknesses A test matrix containing 50 parameter combinations was established, and the nominal value of the differential impedance was set to 100 for each parameter combination. Adjust the center distance of the traces in steps. And monitor the distributed capacitance per unit length The fluctuation amount, based on The change in linewidth compensation increment is determined ,calculate Adjustment value of center distance from trace The ratio is defined as the compensation operator for the corresponding medium thickness. This allows for the construction of an offline data mapping table covering media thicknesses ranging from 0.05 mm to 0.2 mm.
[0042] When the system encounters a situation where there are fiber braiding pitch deviations in hybrid substrates from different production batches, the wiring engine initiates a pre-alignment procedure on-site before determining the phase compensation range. This uses test samples at the corners of the hybrid substrates as reference points, with a resolution of at least 0.5. Optical sensors acquire the fiber orientation angle With fiber weaving pitch The acquired physical parameters are updated to the global wiring coordinate system, so that the physical start coordinates and physical end coordinates of the wiring trajectory are synchronized with the periodic distribution characteristics of the current batch of substrates. This on-site calibration process is used to offset the periodic drift of the dielectric caused by the bonding process fluctuation, so that the differential electric field in the phase compensation interval is maintained within the physical safety boundary where the spatial distribution gradient of the dielectric constant is lower than the preset threshold.
[0043] Example 5: In a hybrid substrate deployment scenario targeting 224 GPA M4 ultra-high-speed signals, the system applies a standardized pre-calibration procedure to determine the initial mapping reference of the dielectric space below the wiring trace. An optical sensor with a resolution better than 0.5 μm is driven to perform a 3D scan of L-shaped markers at the corners of the hybrid substrate under a controlled ambient temperature of 25 ± 1 °C. The centroid coordinate offset of the markers is calculated and used as a compensation factor to correct the global coordinate system, ensuring that the lateral and longitudinal steps of the mesh are aligned with the fiber weaving pitch within the substrate. The spatial phase offset is less than 0.01mm. This initialization step eliminates scanning deviations caused by plate clamping inclination or mechanical repetitive positioning accuracy, ensuring that the physical start and end coordinates of the phase compensation interval are physically aligned with the geometric center axis of the fiber bundle, thus ensuring the accurate reading of the intrinsic dielectric constant. Corresponding to a defined physical space grid.
[0044] To address the impedance inconsistency risk caused by variations in the lamination process, the system initiates an adaptive update process for the impedance correlation mapping model. This process uses built-in time-window sliding sampling logic to capture the reflected echoes from the test traces and convert them into distributed capacitance per unit length. The processing unit calculates the ratio of the deviation between the measured capacitance data and the theoretical value in the offline lookup table, and then applies the formula... The compensation operator is modified, where... For the corrected compensation operator, The compensation operator is preset for offline use. Let be the dimensionless convergence coefficient. The measured capacitance is the rate of change relative to the theoretical value. The system limits the correction step of the compensation operator to within 0.05 and monitors the convergence status of the differential impedance deviation in real time until the differential impedance fluctuation range in the phase compensation interval stabilizes within the 3% threshold. The final determined wiring geometry parameter matrix is directly mapped to the wiring geometry engine, so that the return loss of the differential link is better than -15dB in the entire frequency band.
[0045] In high-speed differential pair routing scenarios involving multiple bends and turns, the routing engine applies a corner phase balance calibration procedure to eliminate geometric skew caused by the difference in the inner and outer trace paths. The system identifies bend and turn areas in the initial routing trajectory and calculates the geometric path difference between the first and second traces. Using the formula Determine the intrinsic skew at the corner, where, This is the intrinsic skew at the corner. For geometric path difference, For the effective dielectric constant, The speed of light in a vacuum; based on the principle of microwave transmission induced by the discontinuity of the microstrip line structure, the outer trace of the bend region accumulates additional inductive hysteresis due to geometric extension. To perform precise local phase repair, the system initiates an in-situ asymmetric linewidth adjustment procedure within the bend corner region. The operation steps are as follows: lock the linewidth parameter reference of the outer trace, synchronously reduce the physical linewidth at one end of the inner trace according to the intrinsic skew, and utilize the reduction in linewidth to induce a decrease in the capacitance to ground at the edge and an increase in the inductance per unit length, forcibly lowering the phase velocity of electromagnetic waves propagating at the inner trace. The system calculates the inverse... The cumulative phase shift and the intrinsic skew form an absolute value offset, achieving in-situ time calibration. Under the global routing constraint architecture, the system performs pre-compensation by adjusting the physical length of the phase compensation interval before and after entering the corner. This makes the artificial phase guidance generated by the phase compensation interval of the straight segment and the geometric phase lag generated by the bending corner area opposite in polarity and algebraically canceled in value. This process limits the total phase skew of the signal after passing through the corner to within 0.1ps, enabling the cabling system to maintain phase self-consistency throughout the link when dealing with complex topology paths.
[0046] The embodiments of this application have been described above with reference to the accompanying drawings. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. This application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit of this application and the scope of protection of this invention, and all of these forms are within the protection scope of this application.
Claims
1. A phase balance control method for high-speed line layout, characterized in that, Includes the following steps: Step S101: Obtain the physical parameters of the substrate of the printed circuit board and the nominal value of the target impedance. The physical parameters of the substrate include the fiber weaving pitch and the fiber cloth orientation. Step S102: Determine the initial wiring trajectory of the differential pair. Based on the relative angle between each straight line segment in the initial wiring trajectory and the fiber cloth direction, calculate the effective dielectric constant difference between the locations of the first and second traces in the differential pair, and determine the cumulative phase skew of the differential pair. Step S103: In the initial wiring trajectory, a phase compensation interval defined by the physical start coordinate and the physical end coordinate is set. The physical start coordinate and the physical end coordinate are respectively locked on the geometric center axis of the fiber bundle inside the printed circuit board or the geometric center axis of the resin enrichment area, so that the physical length of the phase compensation interval is half an integer multiple of the length of the fiber braid pitch mapped in the direction of the initial wiring trajectory. Step S104: Adjust the center distance of the differential traces within the phase compensation interval according to the accumulated phase skew and the physical length to generate a phase cancellation amount to compensate for the accumulated phase skew. Simultaneously, based on the change in the center distance of the differential traces, the single-ended line width of the first and second traces within the phase compensation range is adjusted in the opposite direction. The changes in distributed inductance and distributed capacitance generated by the line width adjustment are used to offset the impedance fluctuations caused by the change in the center distance of the differential traces, so that the differential impedance within the phase compensation range is stabilized at the target impedance nominal value.
2. The phase balance control method for high-speed line layout according to claim 1, characterized in that, The synchronous reverse adjustment of the single-ended linewidth in step S104 includes: establishing a correlation mapping model between the differential trace center distance, the single-ended linewidth, and the differential impedance; when the cumulative phase skew indicates that the outer trace is ahead, increasing the differential trace center distance within the phase compensation interval, and increasing the single-ended linewidth according to the correlation mapping model, using the reduction in distributed inductance caused by the increase in linewidth and the increase in capacitance to ground to offset the decrease in capacitive coupling caused by the increase in differential trace center distance, and limiting the differential impedance deviation within the phase compensation interval to within 3% of the target impedance nominal value.
3. The phase balance control method for high-speed line layout according to claim 1, characterized in that, Step S103, which involves locking the physical start coordinates and physical end coordinates, includes: using a preset electromagnetic scanning algorithm to obtain the spatial distribution gradient of the dielectric constant inside the printed circuit board; identifying linear regions in the spatial distribution gradient of the dielectric constant with a rate of change lower than a preset threshold, and defining the linear regions as physical safety boundaries; and mapping the physical start coordinates and physical end coordinates to the physical safety boundaries to prevent the expanded edge electric field in the phase compensation interval from cutting the interface between the fiber bundle and the resin-rich area.
4. The phase balance control method for high-speed line layout according to claim 1, characterized in that, Following step S104, the method further includes a step of handling phase skew in the corner region: identifying the bend corner region in the initial wiring trace and determining the physical path difference between the inner and outer traces in the bend corner region; within the bend corner region, asymmetrically adjusting the linewidth of the inner and outer traces, and generating a reverse phase shift by changing the unit length distributed inductance of the inner and outer traces to counteract the time skew caused by the physical path difference.
5. The phase balance control method for high-speed line layout according to claim 1, characterized in that, Step S101, obtaining the fiber weaving pitch, includes: measuring the warp and weft fiber densities of the sample substrate using a scanning electron microscope; calculating the orthogonal mapping component of the fiber weaving pitch in the initial wiring trajectory direction based on the warp and weft fiber densities; and calibrating the physical length of the phase compensation interval based on the orthogonal mapping component in step S103.
6. The phase balance control method for high-speed line layout according to claim 1, characterized in that, The adjustment of the differential trace center distance in step S104 includes: obtaining the signal wavelength corresponding to the differential signal transmission rate; setting the physical length of the phase compensation interval to 0.25 to 0.5 times the signal wavelength; and changing the differential trace center distance within the phase compensation interval using a tapered microstrip line structure.
7. The phase balance control method for high-speed line layout according to claim 1, characterized in that, After step S104, the method further includes: monitoring the common-mode noise conversion rate of the differential pair at a preset operating frequency; when the common-mode noise conversion rate exceeds a preset threshold, obtaining the dielectric loss distribution within the phase compensation interval; and finely adjusting the ratio of the single-ended linewidth to the center distance of the differential traces according to the dielectric loss distribution to balance the transverse electric field coupling strength between the two branches of the differential pair.
8. The phase balance control method for high-speed line layout according to claim 1, characterized in that, The steps in S101 for obtaining the physical parameters of the substrate include: establishing a hybrid dielectric model based on the resin enrichment region and the fiber bundle region; determining the angle between the initial wiring trajectory and the fiber cloth direction; determining the spatial fluctuation period of the effective dielectric constant below the initial wiring trajectory based on the angle and the hybrid dielectric model, and using the spatial fluctuation period as the basis for setting the step length of the phase compensation interval.
9. The phase balance control method for high-speed line layout according to claim 1, characterized in that, After step S104, the following steps are also included: Step S105, perform differential signal eye diagram quality verification for the phase compensation interval, and extract eye height parameters and jitter parameters; Step S106, if the eye height parameter is lower than the nominal preset value or the jitter parameter is higher than the safety preset value, return to step S103 to adjust the physical starting coordinates until the eye diagram quality verification meets the preset high-speed transmission standard.
Citation Information
Patent Citations
Printed circuit board and electronic device
CN223142211U