METHOD FOR MANUFACTURING SYSTEM ANALYSIS AND/OR MAINTENANCE

Abstract

A method for factory analysis and/or maintenance, preferably including receiving factory information and/or associating metrics with factory components, and optionally including acting based on defect associations and/or operating factory machines. The method is preferably associated with one or more manufacturing systems and/or elements thereof.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS

This application is a continuation of U.S. patent application Ser. No. 18/104,666, filed 1 Feb. 2023, which is a continuation-in-part of U.S. patent application Ser. No. 17/879,369, filed 2 Aug. 2022, which claims the benefit of U.S. Provisional Application Ser. No. 63/228,495, filed on 2 Aug. 2021, and of U.S. Provisional Application Ser. No. 63/306,375, filed on o3 Feb. 2022, each of which is incorporated in its entirety by this reference.

TECHNICAL FIELD

This invention relates generally to the manufacturing field, and more specifically to a new and useful method for manufacturing system analysis and/or maintenance.

BRIEF DESCRIPTION OF THE FIGURES

FIG. 1 is a schematic representation of an embodiment of the method.

FIG. 2A is a schematic representation of an embodiment of a system for performing the method.

FIG. 2B is a schematic representation of an example of a manufacturing system.

FIGS. 3A-3B are schematic representations of examples of elements of operating factory machines.

FIGS. 3C-3D are schematic representations of a first and second example, respectively, of associating defects with factory components.

FIG. 3E is a schematic representation of an example of associating metrics with factory components.

FIG. 4A is a schematic representation of an example of renumbering array position indicia.

FIG. 4B is a schematic representation of an example mapping between the different array position indicia depicted in FIG. 4A.

FIG. 5 is a schematic representation of an example of determining program links based on temporal considerations.

FIGS. 6A-6B are schematic representations of a first and second specific example, respectively, of determining program links based on product identifiers.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

The following description of the preferred embodiments of the invention is not intended to limit the invention to these preferred embodiments, but rather to enable any person skilled in the art to make and use this invention.

1. Overview.

A method 100 for factory analysis and/or maintenance preferably includes receiving factory information Silo and/or associating metrics with factory components S120, and can optionally include acting based on metric associations S130 and/or operating factory machines S105 (e.g., as shown in FIG. 1). However, the method 100 can additionally or alternatively include any other suitable elements.

The method 100 is preferably associated with (e.g., directed toward maintenance, such as repair and/or predictive maintenance, for) one or more manufacturing systems (or elements thereof), such as described below in more detail. Further, the method can optionally be performed (in part or in whole) by and/or in association with one or more computing systems, manufacturing systems, and/or any other suitable systems (e.g., as shown in FIG. 2A).

2. Technical Advantages.

Variants of the technology can confer one or more advantages over conventional technologies. In some examples, these advantages can facilitate analysis and/or maintenance of one or more manufacturing systems or elements thereof.

First, in some embodiments, the technology can enable determination and/or quantification (e.g., automated determination and/or quantification) of possible links between factory components (and/or other manufacturing elements) and defective products and/or products with one or more aspects that diverge from target properties. For example, by attributing information associated with products (e.g., blame for detected product defects, for manufacturing variance, and/or for divergence from metric targets; quantitative and/or qualitative product metrics associated with all products, defective products, and/or any other suitable subsets of products; etc.) to the factory components to which these characteristics could be attributed (e.g., factory components that could be responsible for the defects, variances, and/or divergences, such as factory components that did or may have interacted with the defective and/or high-variance regions or elements of the product; factory components that could be responsible for the metrics in question, such as factory components that did or may have interacted with the regions or elements of the product from which the metrics were determined; etc.), the factory components associated with large numbers of product defects and/or large product variances (and/or frequent divergences from desired characteristics) may automatically accumulate large shares of such blame and/or may automatically be associated with large amounts of metric variance and/or divergence from target values, which can enable and/or facilitate detection of these factory components as potential sources of manufacturing system errors.

Second, in some embodiments, the technology can enable determination (e.g., automated determination) of one or more data structures (e.g., matrices, such as stochastic matrices; graphs, such as directed and/or undirected graphs with edge weights; etc.) representative and/or indicative of links (e.g., importance metrics such as causal links, fractional blame links, etc.) between features of manufactured products (e.g., PCB assemblies) and the manufacturing elements (e.g., factory operations, programming parameters, machines and machine elements, etc.). These data structure(s) can provide one or more importance metrics associated with these links, which can enable sensitivity analyses associated with manufacturing elements represented therein. For example, this can enable analysis and/or quantification of manufacturing elements most likely to be associated with (e.g., the cause of) product defects, low production rates, manufacturing variances and/or divergences from target values, and/or other undesired manufacturing system behavior.

Third, in some embodiments, the technology can enable and/or facilitate provision of manufacturing system information (e.g., associated with defects, defect blame, defect rates, causal links, manufacturing variances, blame for variances, etc.) to one or more users. For example, the technology can include one or more user interfaces that enable exploration of the links between factory components (and/or other manufacturing elements) and products, product components, and/or manufacturing variances and/or defects thereof.

Fourth, in some embodiments, the technology can enable and/or facilitate automated determination of maintenance recommendations and/or performance of manufacturing system maintenance tasks. For example, the technology can include determining (e.g., based on the links between manufacturing elements and products, product components, and/or manufacturing variances and/or defects thereof) one or more manufacturing elements that may be problematic (e.g., factory components that may require and/or benefit from inspection and/or maintenance, programming parameters that may require and/or benefit from alteration, etc.). Based on such determinations, the technology can optionally include presenting information indicative of these determinations, acting to effect changes based on these determinations (e.g., altering programming parameters, performing inspections and/or maintenance, etc.), and/or taking any other suitable actions.

Fifth, in some embodiments, the technology can enable determination and/or quantification (e.g., automated determination and/or quantification) of defect rates and/or other production rates. For example, information reported by (and/or in relation to) one or more manufacturing systems may not include successful production information (e.g., production of defect-free products), such as production counts, rates, and/or timing information (e.g., but rather may only include information indicative of detected defects and/or the associated timing and/or rates thereof); in this example, embodiments of the technology may act to determine such information associated with successful production. Accordingly, such information may be utilized to determine defect rates and/or other production rates.

Sixth, in some embodiments, the technology can enable standardization (e.g., automated standardization) of information received from (and/or in relation to) one or more manufacturing systems. In some examples, this information may include inconsistent program names (e.g., for assembly and/or inspection programs associated with manufacturing of a particular product), product location identifiers (e.g., array position indices), and/or other information. In some such embodiments, this standardization can enable and/or facilitate further analysis and/or presentation of the associated information (and/or derivatives thereof).

However, further advantages can additionally or alternatively be conferred by the system and/or method described herein.

3. Manufacturing System.

As described above, the method can be associated with one or more manufacturing systems or elements thereof (e.g., as shown in FIG. 2B). In examples, a manufacturing system can include one or more: manufacturing lines, factories, supply chains (e.g., including multiple factories), and/or any suitable subsets thereof. In some embodiments, the method can be associated with one or more factories, manufacturing lines, and/or other manufacturing systems (and/or elements thereof) for printed circuit board (PCB) assembly (e.g., one or more surface-mount technology (SMT) PCB manufacturing lines, including a plurality of assembly and inspection machines).

In some such embodiments, the factory can include one or more: PCB assembly machines, such as pick-and-place (PNP) machines, solder printers, ovens, and/or elements thereof; PCB inspection machines, such as optical and/or X-ray inspection tools (e.g., automated optical inspection (AOI) and/or automated X-ray inspection (AXI) machines), solder paste inspection (SPI) tools, in-circuit test (ICT) tools; and/or any other suitable entities. In examples, the PCB assembly machine elements can include one or more PNP elements, such as placement heads, nozzle holders on placement heads, nozzles attached to nozzle holders, cameras (e.g., cameras associated with nozzle holders, which can be used to detect part placement on nozzle holders; separate cameras such as “flying” cameras associated with placement heads, which can be used to detect panel placement, warpage, and/or other attributes; etc.), feeders, component reels, vacuum pumps and/or feeds, and the like; solder printer elements, such as stencils, squeegees, cameras (e.g., for inspecting board position and/or registering board position with stencil position), and the like; and/or any other suitable elements.

For example, an SMT line can include one or more modules (typically, a plurality of modules). Each module can be capable of holding a plurality of feeders (with the same or different components loaded into each feeder, or into subsets thereof). Each module preferably has a placement head with a plurality of nozzle holders, each of which can hold one of a plurality of different nozzles (e.g., wherein each nozzle has a particular size and shape). In a specific example, an SMT line (e.g., with Fuji NXT modules) includes 18 modules, each with 40 feeders, 24 nozzle holders, and more than 24 nozzles (e.g., 40-100 nozzles, about 60 nozzles, etc.).

Accordingly, for any particular component defect, there may be a large number (e.g., equal to the product of the number of modules, feeders per module, nozzle holders per module, and nozzles) of separate placement paths, each of which describes a path by which a component could come from a feeder, be picked up by a nozzle held in a nozzle holder, and then placed on a board, and any of these system components could potentially be associated with causing the defect; in addition, there may also be a variety of different sensors associated with the placement of the component, and thus also potentially associated with causing the defect. In typical manufacturing systems, defect information (e.g., determined using inspection equipment, such as described below in more detail) is not indicative of which of these placement paths (or subsets thereof, such as any system components in that placement path; e.g., which nozzle, nozzle holder, and/or feeder was used) was used to place the component, nor which of the system components (or system component types, such as nozzle versus feeder) may have been responsible for causing the defect.

Further, the operation of the factory is typically controlled by a large number of programming parameters. Among these programming parameters may be parameters that are specific to particular component types, assembly tools, placement paths, and/or other aspects. Accordingly, for any particular component defect, the defect may be attributable to one or more programming parameters, rather than (or in addition to) being attributable to one or more system components.

In examples, the programming parameters can include PNP parameters, printer parameters, oven parameters, and/or any other suitable parameters. The PNP parameters can include, in some examples: expected dimensions and/or shape of component, location to pick up component from, vacuum pressure to use to hold component, air pressure used to release component onto board, height to pick up, height over board at which to drop, movement profile to use, waiting periods (e.g., wait for N milliseconds before picking up from reel, such as to compensate for slow feeder movement; wait for M milliseconds before dropping part on board, such as to halt lateral motion; etc.), force to use to push chip down into solder paste, and the like. The printer parameters can include, in some examples: force to apply to squeegee to squeeze paste through stencil, cleaning cycle interval (e.g., clean every 5 panels, every 10, etc.), motion profile and/or speed to use to lower stencil onto panel, motion profile and/or speed to use to release stencil from board after paste deposited, speed at which squeegee moves, speed at which squeegee retracts to starting point, amount of solder paste to coat onto stencil, and so on. Over parameters can include, in some examples: temperature profile of oven, speed of board through oven, dwell time at each temperature, and so on. However, the programming parameters can additionally or alternatively include any other suitable parameters.

In typical PCB factory (e.g., SMT factory) operation, product metrics can be determined (e.g., product measurements made, derived metrics determined, production defects detected) during production inspections by automated inspection machines such as automated optical inspection (AOI), automated X-ray inspection (AXI), solder paste inspection (SPI), in-circuit testing (ICT), and the like; additionally or alternatively, metrics can be determined by manual inspection processes and/or any other suitable inspection tools and/or processes. These inspection machines typically make measurements on a product (e.g., panelized array of 1 or more circuit boards, typically processed as a single unit), and can optionally use those measurements to detect production defects.

In some examples, such product measurements can include: component position (e.g., lateral displacement from a target location, such as total displacement and/or displacement along one or more reference axes, preferably in-plane axes but additionally or alternatively one or more out-of-plane axes such as a plane-normal axis) and/or orientation (e.g., angular displacement from a target orientation, such as total angular displacement and/or angular displacement about one or more reference axes, preferably a plane-normal axis but additionally or alternatively any other suitable axes); electrical metrics (e.g., resistance, capacitance, inductance, voltage, etc.) associated with one or more components, networks, circuits, traces, and/or reference points; solder paste deposit size (e.g., volume, width, length, height), position (e.g., lateral displacement from a target location, such as total displacement and/or displacement along one or more reference axes, preferably in-plane axes; distance between solder paste deposits such as smallest distance, which can be indicative of a near-bridging condition; etc.), and/or orientation (e.g., angular displacement from a target orientation, such as total angular displacement and/or angular displacement about one or more reference axes, preferably a plane-normal axis); and the like. However, the product measurements can additionally or alternatively be indicative of any other suitable metrics.

Further, in some examples, such production defects can include missing components, incorrect components (e.g., wherein a component of the wrong type is placed at a location intended for a different component), misoriented components (e.g., rotated, “tombstone”, incorrect polarity such as flipped, etc.), displaced components (e.g., laterally displaced from the appropriate location by more than a threshold amount, too close to another component or trace, etc.), electrical defects (e.g., open circuit, short circuit, resistance out of tolerance such as high or low resistance, capacitance out of tolerance such as high or low capacitance, inductance out of tolerance such as high or low inductance, analog measurement of a voltage value on a circuit node not correct, etc.), solder defect (e.g., solder bridging, solder joint not formed correctly, solder paste volume too high or too low, solder paste deposit not in the correct location, solder paste dimensions not correct, such as wherein the solder paste width, length, and/or height are not within bounds, etc.), OCR failure, and the like.

In typical factory operation, these production defects are associated with (e.g., labeled with, identified based on, etc.) a defect type and/or one or more location fields, such as the array position, reference designator, and/or pin number. For example, each defect can be labeled with a defect type, array position, reference designator, and optionally a pin number. The array position preferably specifies a circuit board (e.g., top-right, bottom-left, position #2, etc.) within the array of circuit boards. The reference designator preferably specifies the defective component (within the particular circuit board specified by the array position), such as C16, U1, R5, etc. The pin number can specify a particular component pin (of the component specified by the reference designator), such as pin 1, 2, 3, 4, etc., but can alternately be omitted in some or all cases (e.g., for defects not associated with a particular pin). The defect type preferably specifies the type of production defect that has been identified, more preferably using one of a standardized set of defect type labels (e.g., “MISSING”, “OPEN”, “TOMBSTONE”, etc.). However, the defects can additionally or alternatively be associated with any other suitable information.

In some examples, the metrics may analogously be associated with a metric type or other metric identifier (e.g., specifying the type of measurement or the specific measurement that has been made) and/or one or more location fields (e.g., such as described above regarding location fields for production defects). However, the metrics may additionally or alternatively be associated with any other suitable identifiers, and/or may be received and/or stored in any other suitable manner (e.g., wherein each specific measurement source, such as all volume measurements of a particular solder deposit location, is associated with a unique measurement code; wherein all measurements from a particular measurement source, across all products measured, are received and/or stored in association with each other; etc.).

However, the manufacturing systems can additionally or alternatively include systems for any other suitable manufacturing and/or other production tasks, and/or can include any other suitable manufacturing and/or inspection tools.

4. Method.

4.1 Receiving Factory Information.

Receiving factory information Sno preferably functions to receive information about metrics determined regarding factory products (e.g., metrics and/or defects such as described above regarding the manufacturing system), such as measurements made and/or defects detected by automated inspection machines. These metrics can include qualitative and/or quantitative metrics, such as numerical measurements, derived metrics determined based on one or more measurements, defects and/or associated characterizations (e.g., presence or absence of defects, defect types, defect severity, etc.). The factory information is preferably received by (and/or stored at) one or more computing systems and/or analysis tools, but can additionally or alternatively be received by any other suitable entities. In some examples, the received metrics are limited or substantially limited to defect information, whereas in other examples, the factory information may include any other suitable metrics (e.g., raw measurements, derived metrics, etc.).

The metric information received in Sno preferably includes, for each metric and/or defect (or a subset thereof), an identifier (e.g., type such as metric type or defect type, other metric identifier, etc.) and/or one or more location fields (e.g., such as described above). Further, the metric information can optionally include timestamps (e.g., production timestamp, inspection timestamp, defect detection timestamp, etc.) and/or other time-related information (e.g., time window associated with the metric and/or defect, such as a time window during which the product was produced and/or the defect was identified), product information (e.g., product identifier, product serial number, product type, CAD information, programming information, program name(s), etc.), and the like. However, the metric information can additionally or alternatively include any other suitable information.

Further, Silo can optionally function to receive other information about factory production (e.g., regarding passed inspections, defect-free production count, total production count regardless of defect state, etc.), about factory state (e.g., production and/or inspection tool state), about machine programming (e.g., program identifiers run on various machines in the factory, preferably received in association with timing information indicative of time periods during which the different programs were running on the machines, etc.), and/or any other suitable information. The information can be available and/or received on a per-component basis, a per-product type basis, a per-manufacturing line basis, a per-manufacturing tool basis, a time window basis, and/or any other suitable basis.

The factory information is preferably received in response (e.g., immediately or substantially immediately in response) to determination of the information, such as in real-time (or in near-real-time). In particular, defect information is preferably received in response to defect identification, more preferably received immediately (or substantially immediately) in response to defect identification. Additionally or alternatively, some or all metric information (e.g., non-defect metric information) may be received in response to metric determination (e.g., measurement at the inspection tool), such as received immediately (or substantially immediately) in response to metric determination; for example, the metric information may be provided as a data stream such as a continuous or substantially continuous data stream. However, the factory information can additionally or alternatively be received after a delay, received in batches, received periodically, received sporadically, and/or received with any other suitable timing.

The factory information is preferably received from the manufacturing system (e.g., from the inspection machines and/or assembly machines, from one or more manufacturing system controllers such as a centralized controller, etc.), but can additionally or alternatively be received from any other suitable entities.

However, S110 can additionally or alternatively include receiving any other suitable factory information in any suitable manner. Further, although referred to herein as receiving factory information, a person of skill in the art will recognize that Silo can additionally or alternatively include receiving information from and/or associated with any other suitable manufacturing system(s).

4.2 Associating Metrics with Factory Components.

Associating metrics with factory components S120 preferably functions to enable identification of factory components (e.g., factory tools, such as assembly tools, and/or elements thereof; programming parameters; etc.) to which these metrics (and/or portions thereof) may be attributable. For example, associating defects with factory components can function to enable identification of factory components that may be responsible for the defects. S120 preferably includes determining potential links S125 (e.g., links between metrics, such as defects and/or measurement values, and factory components), and can optionally include determining data mappings S123 and/or determining defect rates S127.

Although described herein as “factory components”, a person of skill in the art will recognize that S120 can additionally or alternatively include associating metrics with other manufacturing system elements (e.g., upstream factories within a supply chain and/or factory components thereof), input materials (e.g., supplies received from upstream in the supply chain, such as: raw materials, manufactured components such as SMT components, etc.), and/or any other suitable elements associated with the manufacturing process; metrics can be associated with such elements in a manner analogous to that described herein regarding the factory components.

However, S120 can additionally or alternatively include any other suitable elements performed in any suitable manner.

4.2.1 Determining Potential Links.

Determining potential links S125 preferably functions to determine potential links between metrics, such as defects and/or measurement values, and their causes (e.g., as shown in FIGS. 3C-3E). For example, S125 can function to determine links between metrics of interest (e.g., metrics exhibiting high variance and/or divergence from desired properties, defects, etc.), process characteristics associated with these metrics (e.g., associated placement paths and/or aspects thereof), and/or factory components, such as the specific machine components that were involved in operations affecting specific components (e.g., R3 on circuit 1, pin 2 of U1 on circuit 4, etc.), preferably components associated with the metrics of interest.

In a defect-specific example, S125 can function to determine (e.g., generate, preferably automatically) links between defect occurrences, defect characteristics (e.g., elements of the information received in association with the defects, the associated placement paths and/or aspects thereof, etc.), and/or factory components, such as the specific machine components that were involved in operations affecting specific components (e.g., R3 on circuit 1, pin 2 of U1 on circuit 4, etc.), preferably components associated with defects.

In some embodiments, this can include attributing blame or potential blame for defects to upstream factory components (e.g., any machine components and/or programming parameters, or a subset thereof, that affected elements associated with the defect before it was detected; any machine components and/or programming parameters that could have caused the defect; etc.). Thus, each defect can be linked not only to the associated product component (e.g., the component identified in the defect information received in S110), but also to the specific factory components (e.g., nozzles, feeders, component, programming shapes, etc.), that interact with the associated component.

For example, for each defect (e.g., each defect represented by the defect information received in S110), S125 can include noting the link between the defect (e.g., and/or the associated product component, such as U3 of circuit 2, indicated by the defect information) and the N machine components that interacted with the defect-associated product component before that defect was determined (or alternatively, noting the link between the defect and only a subset of such machine components). Noting these links can function to attribute potential blame for the defect to these machine components.

For each of the machine components, noting the link can be performed by adding the value of the link to a sum of defect links associated with that machine component. In a first example, this can include assigning blame for a defect to each of the associated machine components (e.g., adding 1 to each of the associated defect link sums), whereas, in a second example, this can include assigning fractional blame for the defect to each of the associated machine components (e.g., adding 1/N to each of the associated defect link sums, such as shown by way of example in FIG. 3C).

In some variations, in which only n of the N machine components are capable of contributing (or alternatively, are likely to contribute) to a defect of the type flagged, assigning blame for i/n of a defect to each such machine component, and no blame to the remaining (N−n) machine components. For example, an incorrect polarity defect type is not typically attributable to a nozzle or nozzle holder (e.g., as the nozzle will not mistakenly flip the product component around), and so no blame for incorrect polarity defects should be assigned to any nozzle or nozzle holder (e.g., rather, blame for such defects is potentially attributable only to programming parameters and machine components involved in product component loading), such as shown by way of example in FIG. 3D.

In some variations, blame can be assigned non-uniformly, such as assigning greater blame to machine components more likely to have caused the defect (e.g., in a manner such that the sum of the amounts added to all the defect link sums is 1); for example, a first and second machine component may each be assigned ¼ of the blame for a defect, and a third machine component may be assigned the remaining ½ of the blame. A person of skill in the art will recognize that, analogously, the values described above could be multiplied by a factor c (e.g., adding c to each associated defect link sum, adding c/N to each associated defect link sum, adding to the associated defect link sums such that the sum of the amounts added is c, etc.), where c is any suitable value greater or less than unity.

In some variations, different defects can be associated with different amounts of blame (e.g., wherein more important or critical defects can be associated with greater amounts of blame). For example, a first defect can be associated with A units of blame (e.g., 1 unit), and a second defect can be associated with B units of blame (e.g., 2 units). Accordingly, in a first specific example, in which the blame is apportioned equally between potentially-responsible factory components, the N components associated with the first defect will each be assigned A/N units of blame (e.g., 1/N units), and the M components associated with the second defect will each be assigned BIM units of blame (e.g., 2/M units). Alternatively, in a second specific example, in which each potentially-responsible factory component is assigned full blame for a defect, the N components associated with the first defect will each be assigned A units of blame (e.g., 1 unit), and the M components associated with the second defect will each be assigned B units of blame (e.g., 2 units). However, the different defects can additionally or alternatively be associated with different (and/or the same) amounts of blame in any other suitable manner, and/or the blame can additionally or alternatively be assigned to factory components in any other suitable manner (e.g., apportioning fractional blame non-uniformly between the different factory components).

Additionally or alternatively, S125 can include associating metrics such as measurement results and/or values derived therefrom (e.g., difference between a target value and a measured value) with upstream elements (e.g., factory components, configuration parameters, input materials such as SMT components, etc.), such as creating causal associations indicative of possible causal links between the upstream components and the metrics. Such associations are preferably created in a manner analogous to that described herein regarding defect blame; for example, if a particular metric can be used to detect a defect (e.g., if the metric value falls outside a threshold ‘acceptable’ range), and blame for that defect would be attributed to a first set of upstream elements, then the metric should analogously be associated with this same first set of upstream elements (e.g., with the upstream elements that had or could have had an effect on the metric in question).

For each factory product, a typical manufacturing process will include determination of a large number of different metrics (e.g., measurement of many different characteristics for each of many different components and/or aspects of the product, determination of many derived values based on these measurements, etc.). In some examples, S125 can include storing all of these metrics (or any suitable subset thereof, such as wherein measurements of only certain characteristics are stored while other measurements are not). These metrics can be stored in association with the components for which they were measured (e.g., enabling determination of a histogram, associated with the particular metric for the particular component, across all nominally-identical factory products for which the metric was determined, preferably for all such factory products manufactured in a particular time window and/or using a particular manufacturing line), the upstream elements to which those metrics may be attributable in part or whole (e.g., enabling determination of a histogram, associated with the particular metric for which the upstream element may share some responsibility), and/or any other suitable entities. In some examples, each metric value is stored in association with (e.g., labelled in a manner indicative of) each upstream element that may have a causal relationship with the metric in question (e.g., each upstream element that would be blamed, such as described above, if the metric value in question were indicative of a defect). Such associations preferably enable analysis of subpopulations of these metric values associated with individual upstream elements (e.g., as described below in more detail, such as regarding S130), but can additionally or alternatively enable any other suitable analysis.

In some examples, the metrics can be associated with (e.g., stored in association with) one or more weight factors (e.g., one weight factor associated with each upstream element associated with the metric, denoting the weight that the metric is given in determining the effect attributed to that upstream element; a single weight factor that is common to all upstream elements; etc.), which can function in a manner analogous to the fractional defect blame described herein. In one example, if a particular metric is indicative of a defect, then that defect would be attributed in fractional amounts to a set of upstream elements; in this example, the metric (e.g., whether or not it is indicative of a defect) can be associated with those upstream elements with a weight factor proportional to the fractional blame that would have been attributed (e.g., wherein the weight factor attributes the metric causation equally to all upstream elements, wherein the different upstream elements are attributed different amounts of causation, etc.).

However, the metrics can additionally or alternatively be associated with any other suitable information in any suitable manner.

In some examples, S125 can include using the metrics to generate and/or determine one or more histograms representative of the metrics. For example, each metric can be used to generate a set of histograms specific to that metric, such as a global histogram (e.g., representative of all values observed for that metric, preferably for all examples of the factory product for which the metric is determined, such as for all examples of an individual SKU), upstream element histograms (e.g., each representative of all values observed for that metric that are associated with a particular upstream element, optionally wherein the histogram is a weighted histogram representative also of the weight factors associated with the metric and the upstream element), and/or any other suitable histograms. Any or all such histograms can additionally or alternatively be restricted to one or more time windows (e.g., restricting the values used in the histogram to only the values associated with measurements made during a particular period of time). In a first example, the measured values for a metric indicative of solder paste volume at pin 2 of component U3 can be used to generate both a global histogram representative of the solder paste volume at this pin for all examples of the board design (e.g., all examples fabricated using the manufacturing system), and a set of upstream element histograms (e.g., a separate histogram for each upstream element of interest, each generated based only on values measured for solder paste deposits that could have been affected by the upstream element in question). In a second example, the measured values for a metric indicative of horizontal displacement (e.g., from a target horizontal position) of component C2 can be used to generate both a global histogram representative of the horizontal displacement of this component for all examples of the board design (e.g., all examples fabricated using the manufacturing system), and a set of upstream element histograms (e.g., a separate histogram for each feeder, nozzle holder, and nozzle, wherein each histogram is generated based only on values measured for components that could have been affected by the upstream element in question, such as only components placed by the nozzle in question).

Further, S125 can additionally or alternatively include generating and/or determining one or more histograms representative of pooling multiple metrics (e.g., across a single SKU, across multiple SKUs, etc.). For example, S125 can include determining histograms (e.g., global histograms, upstream component histograms, etc.) representative of lateral SMT component displacement (e.g., from target positions) for all SMT components (or a subset thereof, such as all ICs, all capacitors, all components of a particular size or size range, all components of a particular shape, etc.).

However, S125 can additionally or alternatively include using the metrics to generate and/or determine values of one or more statistical metrics, such as descriptive statistics (e.g., mean, variance, etc.). For example, S125 can include determining values of one or more error metrics representative of the numerical distance (e.g., norm, such as e¹ norm, e² norm, e^∞ norm, etc.) between observed values associated with a metric (or set of metrics) and a desired or target value associated with each such metric. Further, S125 can additionally or alternatively include generating any other suitable summaries and/or descriptions of the collections of metric values in any suitable manner.

Additionally or alternatively, S125 can include generating (and/or adding to) a web of links between product features (e.g., features on one or more circuit boards, such as individual SMT components or pins thereof) and the upstream factory components (e.g., factory operations, programming attributes, machines, and/or machine components, etc.). This web of links can be used to generate a matrix (e.g., stochastic matrix) describing the strength of links between the product features and the factory components. In this matrix, the values can be representative of a sensitivity analysis of which parameters and/or systems exhibit high correlation with undesired metrics (e.g., and therefore may merit maintenance and/or alteration, such as described below in more detail regarding S130). Here, undesired metrics can include metrics exhibiting high variance, metrics exhibiting divergence from desired properties, metrics indicative of defects, and the like. For example, the sensitivity analysis can indicate which parameters and/or systems are likely to be the most common causes of defects over time, and/or are likely to be the most common causes of high process variance.

In examples, S125 can include, for all machines upstream of an inspection machine (or a subset thereof), creating a list of the machine systems and parameters that were involved in placing a given reference designator on a given board at a given array position on the panel; and then (e.g., when a defect is flagged) using the associated information to distribute blame for the defect (e.g., as described above) to all of the involved systems (or to any suitable subset thereof), and/or (e.g., for each measurement) using the associated information to associate the measurement value with all of the involved systems (or to any suitable subset thereof).

However, in some embodiments, the metric information received in S110 may not be sufficient to uniquely identify the particular product component associated with the metric (and/or to determine the set of factory components to which possible causation for the metric should be attributed). For example, there may not be sufficient information to enable mapping the array position reported with the metric to a particular board (e.g., to map between the array positions reported by the inspection machine and the assembly machines); accordingly, the reference designator reported with the metric could represent any one of a number of different product components (e.g., one on each board, or a subset thereof, such as each board having such a component).

For example, array numbering may be inconsistent across different machines, such as machines from different machine vendors. In one example, all machines will agree that there are 6 circuits in an array, such as arranged in a 2×3 rectangular array. However, an inspection machine may assign array positions in a snake fashion, in ascending order from 1 at the top right to 6 at the bottom left, whereas a placement system may assign array positions in a brick laying fashion, in ascending order from 1 at the bottom left to 6 at the top right.

It may be possible to resolve this ambiguity in some cases (e.g., as described below in more detail regarding S123). However, in other cases, the ambiguity may remain (e.g., S123 is not performed, attempts to resolve the ambiguity are not made, insufficient information is available to resolve the ambiguity, etc.). Accordingly, in some examples (e.g., in which ambiguity remains), the metric can be attributed to each of these potentially-associated product components, and then, for each of these potentially-associated product components, potential causation can be attributed to factory components such as described above. In a first example, potential causation of the metric is attributed for each potentially-associated product component, whereas in a second example, potential causation is attributed fractionally, such as attributed with a 1/B weight, for B potentially-associated product components (e.g., B different boards, each including one such potentially-associated product component), can be distributed for each potentially-associated product component (e.g., such that the counts added to the corresponding bin of each associated metric histogram is reduced by a factor of 1/B, as compared with the approach described above in which there is no ambiguity regarding the identity of the associated product component).

S125 can additionally or alternatively include determining time series associated with these causal links. In a first example, a time series for a factory component can include a timestamp of linked defect detection and/or metric determination (and optionally, fractional metric association, such as fractional defect blame apportionment). In a second example, the time series for a factory component can include a series of metric value attributions, each accumulated over a short time window (equal to the temporal resolution of the time series). However, the time series can additionally or alternatively be determined in any other suitable manner.

However, S125 can additionally or alternatively include determining any other suitable potential links in any suitable manner.

4.2.2 Determining Data Mappings.

Determining data mappings S123 can function to determine mappings between elements of data received from (and/or about) the manufacturing system (e.g., as described above in more detail regarding Silo), such as mappings that can further enable and/or facilitate metric analysis (e.g., enabling and/or facilitating determination of potential links, such as described herein regarding S125, determination of defect rates, such as described herein regarding S127, and/or analysis of such information, such as described herein regarding S130, etc.). In some examples, S123 can include transforming data (e.g., based on these mappings), such as transforming the data into forms that can further enable and/or facilitate this metric analysis.

In some embodiments, S123 can include determining mappings between array position indicia (e.g., array position numbers) associated with different manufacturing system machines (e.g., different assembly and/or inspection machines), such as shown by way of example in FIGS. 4A-4B. Such mappings can be determined based on CAD data (e.g., associated with the product to be produced by the manufacturing system), based on machine-specific reference information regarding the array position numbering, and/or based on any other suitable information. In some examples, such information can be determined in an automated manner (e.g., in real time or substantially in real time), such as by requesting the information from the factory machines. Further, S123 can optionally include transforming data (e.g., received in Silo) based on these mappings, such as by standardizing all array positions in the data, such that each array position indicator (e.g., number) is consistently representative of a particular array position (e.g., the array position ‘1’ always represents the top-left board, the array position ‘2’ always represents the top-right board, the array position ‘6’ always represents the bottom-right board, etc.). In these embodiments, such mapping and/or renumbering can function to resolve ambiguities, such as those described above regarding S125, thereby facilitating more accurate determination of potential links between metrics and factory components.

Additionally or alternatively, S123 can include (e.g., for each product for which information is received, or a subset thereof) determining links between program identifiers (e.g., program names) associated with different factory machines (e.g., different assembly and/or inspection machines) corresponding to production of a product (e.g., wherein the linked program identifiers each correspond to one or more aspects of production of the same product). For example, this can include determining that a set of program identifiers (e.g., program “X12_a” associated with a first machine, program “X12” associated with a second machine, program “X12_insp1” associated with a third machine, and program “a5” associated with a fourth machine) are all associated with portions of the production of the same product. These links can facilitate further understanding of production processes, can enable determination of product counts (e.g., as described below in more detail regarding S127), and/or can be used in any other suitable manner.

In a first example, such program links can be determined based on temporal overlap (and/or proximity) of programs run on the different machines (e.g., within a single manufacturing line). For example, given an inspection program identifier and the associated time window(s) during which it was run on an inspection machine of a particular manufacturing line, the program links can be determined for the inspection program identifier based on the associated time window(s), such as shown by way of example in FIG. 5. For every other machine of the manufacturing line (or a subset thereof, such as each machine upstream of the inspection machine), the program identifier commonly running during (and/or proximal to) the associated time windows can be linked to the inspection program identifier, as it likely corresponds to the same product; if, for any machine, no single program identifier is sufficiently dominant during the associated time windows (e.g., is not running during at least a threshold fraction of the duration of the windows), a link may not be determined. Additionally or alternatively, temporal correlations between program identifiers can be determined, wherein any sets of program identifiers with a sufficiently strong (positive) correlation (or a subset of such sets) can be linked.

In a second example (which can be performed in addition to or instead of the first example), such program links can be determined based on product serial numbers (e.g., as shown in FIGS. 6A-6B). For example, given a set of one or more serial numbers (and/or other identifiers) associated with a particular product (e.g., a single board, a set of boards having identical designs, etc.), the times at which the product is present at a machine (e.g., as indicated by the serial number information) can be correlated with the program(s) run at that machine while the product is present, and the associated program identifiers can be linked to the product. This can be repeated for each machine (or a subset thereof) at which the product is known to have been present. Thus, the program identifiers linked to the product can then be linked to all other program identifiers (e.g., associated with other machines) that are also linked to the product.

However, S123 can additionally or alternatively include determining any other suitable data mappings (and/or transformations) in any suitable manner.

4.2.3 Determining Defect Rates.

Determining defect rates S127 can function to provide additional context to the defect information (e.g., received in Silo) and/or the potential defect links (e.g., potential links and/or defect link sums determined in S125). In some situations, defect rates can provide important context that is not discernable from defect counts alone. For example, in different situations, 100 defects could be associated with a high defect rate (e.g., 10%, corresponding to 100 defects out of 1000 total units produced), a low defect rate (e.g., 0.001%, corresponding to 100 defects out of 10 million total units produced), an intermediate defect rate (e.g., 0.1%, corresponding to 100 defects out of 100,000 total units produced), and/or any other suitable defect rate. Accordingly, it may be beneficial to determine the defect rates associated with various defects.

In many embodiments, the factory information received in Silo will include reports of detected defects, but will not include information indicative of successful production (e.g., counts of defect-free units produced, such as units of a particular product type or all units produced, etc.) or total production (e.g., total units produced, production “opportunities” including both units produced and defective units for which production was aborted, etc.). Accordingly, it can be difficult to determine defect rates (e.g., the ratio of the number of defects detected or defective units produced to the number of defect-free units produced, total units produced, or production opportunities) based on the received information. To overcome this difficulty, S127 can include determining one or more metrics associated with the total production (e.g., determining the number of production opportunities (“opportunity count”) for a particular product type or a particular potential defect).

For example, the opportunity count for a potential defect is determined based on the inspection program that can detect that defect. In response to receiving information indicative of a particular defect (e.g., in response to the first time that information indicative of a defect with a particular reference designator is received in association with a particular product type), the defect can be checked against and/or added to (e.g., if not already present) a defects catalog that maps that defect (e.g., and all defects of the same reference designator associated with the same product type) to the associated inspection program responsible for detecting it; in a variation, the defects catalog may further specify defects based on the defect type (e.g., wherein defects at a particular reference designator of a particular product type may be mapped to different inspection programs, depending on the defect type). This defects catalog can include multiple defect entries mapped to each inspection program (but can additionally or alternatively include inspection programs associated with only one defect entry). Thus, the opportunity count for a defect can be determined based on (e.g., is equal to) the number of times the associated inspection program is run (e.g., during a specified time window, counted only after the defect is first added to the catalog or counted to also include runs performed before the defect is added, etc.).

Additionally or alternatively, the opportunity count for a potential defect can be determined based on links determined (e.g., in S123) between different programs (e.g., assembly & inspection programs) associated with the product type for that potential defect. Thus, even if information indicative of a number of inspection program runs is not available, the opportunity count can analogously be determined based on (e.g., as equal to) the number of times that an assembly program linked to that inspection program is run. In a specific example, in which a link between assembly program “prod17” and inspection program “prod17_insp” is known (e.g., has been determined in S123), both associated with manufacturing of product type 17, the opportunity count for all potential defects in product type 17 can be determined as equal to the number of runs of assembly program “prod17”.

Additionally or alternatively, S127 can include receiving the opportunity count directly (e.g., receiving information representative of the opportunity count in Silo). However, S127 can additionally or alternatively include determining the opportunity count in any other suitable manner.

Once the opportunity count for a set of defects (e.g., all defects for a product type, all defects for a single reference designator within the product type, defects of a particular defect type at the single reference designator, and/or any other suitable set of defects) is known, S127 can include determining the associated defect rate by dividing the number of defects in the set by the opportunity count.

Additionally or alternatively, S127 can include determining defect rates associated with factory components. In some embodiments, these factory component defect rates can be determined in a manner analogous to that described above regarding S125 (e.g., regarding defect link sums and/or defect link time series). For example, rather than only determining links between defects and the upstream factory components, the method can also include determining links between each production opportunity (and/or each defect-free product, or the like) and the upstream factory components (e.g., determining an associated production opportunities sum for each upstream factory component, in a manner analogous to that described above regarding the defect link sums of S125). In other embodiments, the factory component defect rates can be determined based on estimated opportunity counts. For example, the opportunity counts can be estimated based on the production line configuration, and/or based on observed workload share of the different components (e.g., during a portion of the entire time for which opportunity count is to be determined). The estimate is preferably determined in conjunction with a known or estimated total opportunity count (e.g., for a particular product type, for a particular manufacturing line, etc.); for example, an estimated fraction of total opportunities performed by a particular factory component can be multiplied by a (known or estimated) total opportunity count to determine the estimated opportunity count for that factory component. In some examples of these embodiments, the defect rate for a factory component can be determined by dividing its defect link sum (e.g., determined in S125) by the (e.g., known or estimated) total number of production opportunities; however, the factory component defect rates can additionally or alternatively be determined in any other suitable manner. In some examples of these embodiments, the method can additionally or alternatively include acting (e.g., presenting information, performing inspections and/or maintenance, etc.) based on one or more of the factory component defect rates (e.g., in a manner analogous to actions taken based on defect link sums, such as described in more detail regarding S130), and/or performing any other suitable actions based on the factory component defect rates.

However, S127 can additionally or alternatively include determining defect rates in any other suitable manner, and/or can additionally or alternatively include any other suitable elements performed in any suitable manner.

4.3 Acting Based on Metric Associations.

The method can optionally include acting based on metric associations S130, which can function to diagnose manufacturing system faults (e.g., resulting in high defect rates, low production rates, high variance, etc.) and/or to improve, enable improvement of, and/or facilitate improvement of manufacturing system function (e.g., by decreasing defect rates, increasing production rates, decreasing variance, etc.). The actions taken in S130 are preferably informed by the metric associations determined in S120 (e.g., defect link sums, defect rates, metric histograms, etc.), and can additionally or alternatively be informed by any other suitable information determined and/or received during performance of the method. In some embodiments, S130 can include determining one or more elements of manufacturing system maintenance information (e.g., information determined based on the metric associations, such as information associated with and/or indicative of manufacturing system faults and/or possible resolutions) and acting based on the one or more elements of manufacturing system maintenance information (e.g., providing the information to users, performing inspections and/or maintenance based on the information, etc.). In examples, the manufacturing system maintenance information can include information associated with and/or indicative of: factory components that may be responsible for manufacturing system faults and/or sources of variance, maintenance actions, recommendations, and/or prioritizations, and/or any other suitable information (e.g., other information described herein, such as with respect to S130).

In some embodiments, S130 can include presenting information (e.g., derivative information, such as information determined based on the information received in S110 and/or determinations made in Silo) to one or more users. For example, S130 can include presenting potentially-problematic factory components to the users (e.g., providing information indicative of the potential problem(s) to the users).

For example, S130 can include generating and/or presenting a list of factory components that may be responsible for unacceptable (e.g., increased) defect rates, error metrics, and/or metric variance on a manufacturing line (and/or may be otherwise problematic). This list may be used in order to inform decisions about maintenance to be performed (e.g., preventative maintenance performed during a scheduled manufacturing line stop, reactive maintenance performed to eliminate root causes of high defect rates during a manufacturing line quality stop, etc.). For example, this list can include the factory components (e.g., feeders, nozzles, nozzle holders, cameras, programming parameters, etc.) with the highest overall blame for defects (e.g., highest linked defect sum and/or linked defect rate), which may be particularly useful for preventative maintenance, and/or the factory components with the highest blame (e.g., linked defect sum and/or rate) for defects of a particular type or types and/or associated with a particular component or set of components (e.g., defect type(s) and/or component(s) for which the defect rate is particularly high). Additionally or alternatively, the list can include any factory components for which the defect blame (e.g., total count, defect rate, etc.), error metrics, and/or metric variance is greater than a threshold amount (e.g., uniform threshold, threshold determined on a per-component or per-component-type basis, etc.; fixed threshold and/or threshold determined dynamically, such as based on overall factory performance, etc.). Additionally or alternatively, factory components can be selected for the list based on an automated anomaly detection system (e.g., machine learning system, such as one that accepts metric time series and/or defect rate time series for the factory components as inputs).

In a variation of this example, the list may be filtered and/or prioritized based on anticipated time, cost, and/or impacts of performing maintenance associated with the different factory components. For example, the list may be filtered and/or sorted based on anticipated return-on-investment of maintenance resources (e.g., maintenance costs, factory or factory line stoppage time, etc.). In some embodiments, S130 can include providing a suggested maintenance list for current or upcoming (e.g., planned, required or desired due to poor performance such as poor yield and/or production rate, etc.) factory or factory line stoppages. In one example, based on a planned stoppage duration (e.g., 5 minutes, 15 minutes, 2 hours, etc.), the list can be filtered to include only maintenance tasks anticipated to fit within the planned stoppage duration (e.g., wherein each task is anticipated to require less time than the planned stoppage duration; wherein the total anticipated time required for a set of tasks associated with a particular maintenance resource, such as a maintenance worker or team, is less than the planned stoppage duration; etc.) and/or sorted to prioritize the tasks with the highest anticipated value (e.g., with respect to anticipated increased product yield and/or production rate) and/or highest anticipated return-on-investment (e.g., with respect to a ratio of the anticipated value to the anticipated maintenance time).

Additionally or alternatively, S130 can include presenting information indicative of the expected effects from making a manufacturing system change (e.g., fixing or removing a problematic factory component, altering a problematic configuration value, etc.). For example, S130 can include presenting one or more metric histograms (and/or other data representations) indicative of the effect of making one or more such manufacturing system changes. In some examples, in which the effect of removing or fixing a single upstream element is to be presented, such a metric histogram could be representative of all values observed for that metric, except for those associated with the particular upstream element (or alternatively, wherein representation of all values associated with the particular upstream element is reduced by the weight factor corresponding with that upstream element); such a histogram could, in some such examples, be generated by subtracting the counts of the upstream element histogram from those of the global histogram. For example, the global variance or error of a particular metric (e.g., lateral displacement of a particular component, solder paste volume at a particular pad, etc.) or a pool of metrics (e.g., lateral displacement of all SMT components) may be within nominal operational thresholds (e.g., resulting in few or no defects), but the portion of that variance or error attributable to a particular upstream element may be abnormally high (e.g., wherein the upstream element is not yet causing a high number of defects, but may be drifting away from desired operational performance); in this example, it may be desirable to present information indicative of the upstream element's effect on this variance or error (e.g., presenting a metric histogram for that upstream element, presenting a metric histogram indicative of the removal or repair of that upstream element, etc.). Further, analogous histograms, representative of the effects of removing and/or fixing multiple upstream elements, could additionally or alternatively be determined and/or presented. However, a person of skill in the art will recognize that S130 can additionally or alternatively include presenting descriptive statistics (e.g., in a manner analogous to that described herein regarding the metric histograms) and/or presenting any other suitable information in any suitable manner.

Additionally or alternatively, S130 can include taking action in response to determining that a factory component may be problematic. For example, S130 can include alerting users, altering manufacturing line operation (e.g., to avoid or reduce use of the problematic factory component), and/or initiating maintenance (e.g., of the problematic component and/or other associated factory components). Determining that the component may be problematic can include determining that the defect blame and/or metric variance/error has exceeded a threshold amount (e.g., as described above), determining that the component's operation is anomalous (e.g., based on an automated anomaly detection system, such as the machine learning system described above).

Additionally or alternatively, S130 can include presenting the derivative information to one or more users in a manner that enables the user(s) to explore various aspects of the information, such as via a user interface (UI) that enables viewing data slices associated with various defects, product components, factory components (e.g., tools, tool elements, programming parameters, etc.), and/or any other suitable aspects.

For example, the UI can display defect heatmaps for upstream machine components (e.g., nozzles, feeders, feeder attachment locations, etc.), component types, shape codes (associated with programming parameters shared by one or more component types), and the like.

Additionally or alternatively, the UI can enable users to filter the defect data. For example, looking at defects associated with component U1 (of a particular product type), the UI can show that U1 is placed by 2 machines from 3 feeders, but that the vast majority of defects arise from only one of the machines (even though identical parts are loaded in both machines). Based on this information, the user may conclude that the problem is associated with the that one machine. Next, the user may look at additional defects associated with components of that one machine (e.g., removing the U1 filter, and filtering instead by machine component for that machine), thereby discovering a machine component associated with many defects (e.g., associated with different SMT components), which is likely the machine component at fault for most or all of these defects. Alternatively, the defects for the filtered component may not be predominantly attributed to any one machine (e.g., may be substantially uniformly dispersed between the machines). Rather, many defects may be present for the product component (e.g., SMT component, such as a particular resistor, capacitor, IC package, etc.) and/or for its shape code, across multiple machines, which may be indicative of a programming error (e.g., poorly-selected parameters for component placement).

As a person of skill in the art will recognize, some or all such functionality can additionally or alternatively be automated, wherein this analysis (or a subset thereof) may be performed without human intervention.

Further, S130 can additionally or alternatively include (e.g., in response to determining one or more factory components associated with a high defect rate and/or associated with a particular defect), the method can optionally include performing maintenance (and/or altering programming parameters) to reduce or eliminate the defect rate. In a first example, in response to a problematic defect rate, the machine components and/or programming parameters associated with the defect can be corrected (e.g., via immediate or deferred maintenance of the associated components). In a second example, the method can include determining a priority list of machine components on which predictive maintenance should be performed (e.g., wherein the highest priority component is associated with the highest defect rate, or the highest ROI such as a ratio of defect rate to typical maintenance time), and the priority list can be used to perform predictive maintenance during line downtime (e.g., shift changes, product changes, etc.).

In some examples, the manufacturing process and/or analyses thereof may be performed based on an inherent assumption that input materials (e.g., components received from upstream elements of a supply chain) of any particular type are substantially identical or the variance/error thereof is bounded by a particular specification. Alternatively, the potential effects (e.g., on metrics and/or defects) of variance and/or error in some or all such input materials may be analyzed (e.g., in a manner analogous to that described herein regarding other upstream elements). For example, based on knowledge (e.g., from supplier specifications and/or testing, from internal tests conducted on received materials, etc.) of the variance and/or error in one or more metrics associated with these input materials (e.g., variance in capacitance of an SMT capacitor), one or more analyses (e.g., ‘computed experiments’) regarding these input materials can be conducted. For example, S130 can include determining the expected effect of decreasing the tolerated or expected variance/error for one or more of these input material metrics, such as the effect of removing one or more tails of an input material metric distribution (e.g., corresponding to the effect expected from testing the input materials and rejecting those that do not meet a threshold criterion) and/or of narrowing an input material metric distribution while it maintains a substantially similar overall shape (e.g., corresponding to the effect expected from requiring the input material supplier to change the associated manufacturing process to ensure tighter tolerances).

In some such examples, the use of a large number of diverse and/or inexpensive input materials may make it impractical (e.g., due to a large cost outweighing a relatively small expected benefit) to test performance and/or tighten tolerances for all such input materials (or for a significant subset thereof); however, such computed experiments can enable identification of a small subset of input materials for which such testing and/or tolerance tightening may be worth conducting (e.g., based on the expected benefit of such actions exceeding the cost of taking these actions, wherein the expected benefit for testing this small subset may be a significant portion of the total expected benefit for testing all input materials). In one example, S130 can include: analyzing the expected effects of reducing variance associated with each of a plurality of input material metrics (e.g., electrical parameters for each SMT component, such as capacitance of each capacitor, resistance of each resistor, etc.); identifying a subset of the analyzed input material metrics as metrics of potential concern (e.g., wherein the variance and/or error observed for these metrics is likely to drive significant unwanted downstream effects, such as large variance, error, and/or defect rates for manufactured products produced using the input materials in question; and preferably acting to reduce the variance and/or error for some or all metrics of this identified subset (e.g., by implementing testing of the metric to enable selective rejection of defective input materials, by requesting tighter tolerances from the supplier, by switching to alternate input materials that offer tighter tolerances, etc.).

However, S130 can additionally or alternatively include taking any other suitable actions based on the defect associations (and/or based on any other suitable information determined and/or received in performance of the method).

4.4 Operating Factory Machines.

The method can optionally include operating factory machines S105 (e.g., as shown in FIGS. 3A-3D), which can function to manufacture and/or inspect the products (e.g., SMT boards) designated to be produced by the factory machines. S105 can include operating the assembly machines (e.g., based on the programming associated with one or more products), operating the inspection machines (e.g., based on the programming associated with the one or more products), and/or providing information (e.g., manufacturing system state, defect detection, metric values, histograms, descriptive statistics, etc.) for analysis (e.g., provided to one or more analysis tools, such as computing systems). The information is preferably received such as described above regarding Silo, but can additionally or alternatively include any other suitable information to be used in any suitable manner.

Claims

1. A method for manufacturing system maintenance, comprising:

receiving information associated with manufacturing, at a manufacturing system, of a plurality of products, the information indicative of a set of metric values determined based on products of the plurality; wherein, for each metric value of the set, the metric value is associated with a respective product component;

for each metric value of the set: based on the respective product component, automatically determining a respective set of one or more components of the manufacturing system; and for each component of the respective set, associating the metric value with the component; and

determining a set of values of a descriptive statistic, comprising: for each component of a subset of components of the manufacturing system, based on metric values associated with the component, determining a respective value of the descriptive statistic.

2. The method of claim 1, further comprising:

based on the set of values of the descriptive statistic, selecting a set of one or more potentially-problematic components from the subset of components; and

presenting information to a user, wherein the information is indicative of at least one of: a first component of the set of one or more potentially-problematic components; or the respective value of the descriptive statistic associated with the first component.

3. The method of claim 2, wherein:

the descriptive statistic is indicative of variance; and

selecting the set of one or more potentially-problematic components comprises selecting components associated with high variance.

4. The method of claim 2, further comprising:

based on the set of values of the descriptive statistic, determining a predicted result of altering the first component; and

presenting further information to the user, the further information indicative of the predicted result.

5. The method of claim 4, wherein presenting the further information comprises:

presenting an observed data histogram associated with the set of values; and

presenting a predicted histogram indicative of the predicted result.

6. The method of claim 1, further comprising, based on the set of values, controlling maintenance of the manufacturing system.

7. The method of claim 6, wherein controlling maintenance of the manufacturing system comprises:

based on the set of values, identifying a set of components of the manufacturing system that have an elevated probability of defective operation; and

in response to identifying the set of components, selecting the set of components for inspection.

8. The method of claim 6, wherein controlling maintenance of the manufacturing system comprises:

based on the set of values, identifying a component of the subset that has an elevated probability of defective operation; and

in response to identifying the component, selecting the component for repair.

9. The method of claim 1, wherein the information comprises, for each metric value of the set: a respective set of location information indicative of the respective product component associated with the metric value, wherein the metric value is associated with the respective product component based on the respective set of location information.

10. The method of claim 9, wherein the products of the plurality are printed circuit board assemblies (PCBAs), wherein the product components are surface-mount technology (SMT) components, wherein, for each metric value of the set, the respective set of location information comprises:

an array position indicative of a PCBA within the manufacturing system; and

a reference designator indicative of an SMT component of the PCBA.

11. The method of claim 1, wherein the set of metric values comprises a first subset associated with a first metric and a second subset associated with a second metric.

12. The method of claim 11, wherein the first metric is a lateral displacement metric and the second metric is a volume metric.

13. The method of claim 1, wherein, for each metric value of the set: each component of the respective set has interacted with the respective product component.

14. The method of claim 1, wherein the subset of components is a strict subset of the components of the manufacturing system.

15. The method of claim 1, wherein the manufacturing system comprises a plurality of geographically-separated factories.

16. The method of claim 1, wherein the manufacturing system consists essentially of a single factory.

17. A method for manufacturing system maintenance, comprising:

receiving information associated with manufacturing, at a manufacturing system, of a plurality of printed circuit board assemblies (PCBAs), the information indicative of a set of metric values determined based on PCBAs of the plurality; wherein, for each metric value of the set, the metric value is associated with a respective set of one or more manufacturing system components of the manufacturing system;

for each metric value of the set, associating the metric value with each manufacturing system component of the respective set; and

determining a set of values of a descriptive statistic, comprising: for each manufacturing system component of a plurality of manufacturing system components, based on metric values associated with the manufacturing system component, determining a respective value of the descriptive statistic.

18. The method of claim 17, wherein the set of metric values comprises a first subset, wherein, for each metric value of the first subset:

the metric value is associated with a respective product component; and

associating the metric value with each manufacturing system component of the respective set comprises, based on the respective product component, automatically determining the respective set of manufacturing system components such that each manufacturing system component of the respective set has interacted with the respective product component.

19. The method of claim 18, wherein the information comprises, for each metric value of the first subset: a respective set of location information indicative of the respective product component associated with the metric value, wherein the metric value is associated with the respective product component based on the respective set of location information.

20. The method of claim 19, wherein:

the product components are surface-mount technology (SMT) components; and

for each metric value of the set, the respective set of location information comprises: an array position indicative of a PCBA within the manufacturing system; and a reference designator indicative of an SMT component of the PCBA.