Method for producing at least one object in layers, with step-by-step updating of the coordinate transformation of scanners

Abstract

The invention relates to a method for producing at least one object (2) on a building platform (6) in layers by locally solidifying pulverulent material (3) in a layer (7), wherein: at least in a plurality of the layers, N high-energy beams (8a, 8b) are at least temporarily used simultaneously with N scanners, where N ≥ 2; each scanner is assigned a scanner coordinate system; a control device (10) for an exposure of a layer for each scanner - provides (102) exposure data of a machining pattern in a reference coordinate system, - converts (103) the exposure data in the reference coordinate system into exposure data in the scanner coordinate system by means of a programmed coordinate transformation (PKT) and - directs (104) the obtained exposure data in the scanner coordinate system to the associated scanner so that the scanner exposes (105) the machining pattern on the building platform in the layer; while producing the at least one object in layers, measurements are repeatedly taken, by means of each of which the current actual coordinate transformations (MTKT) of at least N-1 scanners are determined; between two successive determinations of the current actual coordinate transformations, M layers are produced, where M ≥ 2; and, while producing the at least one object, the programmed coordinate transformations for the at least N-1 scanners are updated taking into account the current actual coordinate transformations (205.1, 205.j, 205.A; 305.1, 305.j, 305.A). The invention is characterised in that a plurality of updates of the programmed coordinate transformations of the at least N-1 scanners are performed between two successive determinations (201, 206; 301, 306) of the current actual coordinate transformations. The invention provides a method which makes it possible to produce the at least one object on the building platform in layers so that the object is high-quality while minimising non-productive time.

Description

[0001] Method for layer-by-layer production of at least one object, with step-by-step updating of the coordinate transformation of scanners

[0002] The invention relates to a method for the layer-by-layer production of at least one object on a construction platform by local solidification of powdered material in a respective layer, wherein at least in a plurality of the layers with N scanners, at least temporarily, N high-energy beams are used simultaneously, with N>2, wherein a respective scanner is assigned a respective scanner coordinate system, wherein a control device for an exposure of a respective layer for each scanner

[0003] - provides exposure data of a processing pattern in a reference coordinate system,

[0004] - converts the exposure data in the reference coordinate system into exposure data in the scanner coordinate system using a programmed coordinate transformation, and

[0005] - instructs the obtained exposure data in the scanner coordinate system to the associated scanner so that the scanner exposes the processing pattern on the build platform in the layer, wherein during the layer-by-layer production of the at least one object, repeated measurements are taken with which the current actual coordinate transformations are determined in each case by at least Nl scanners, wherein between two successive determinations of the current actual coordinate transformations M layers are produced, with M > 2, and wherein during the production of the at least one object the programmed coordinate transformations for the at least Nl scanners are updated taking into account the current actual coordinate transformations.

[0006] Such a process is known from EP 3 907 021 A1.

[0007] By manufacturing objects layer by layer through the local solidification of powdered material using high-energy beams (usually laser beams or electron beams), three-dimensional objects can be produced relatively easily and quickly. Geometric limitations of conventional manufacturing processes such as milling or injection molding can be overcome. Layer-by-layer manufacturing is often used for prototypes or for objects that are only produced in small quantities or even only once (such as dental crowns).

[0008] For particularly fast production, multiple high-energy beams can be used simultaneously on a single build platform. Each high-energy beam is assigned a scanner (also called a scanner system), which, for example, in the case of laser beams, can include a mirror adjustable by piezo actuators.

[0009] To produce a large object on the build platform, multiple high-energy beams are often used simultaneously on the same object. Within each layer, the resulting surface of the layer to be processed (solidified) is divided into sub-areas, each of which is processed with one of the high-energy beams. The sub-areas to be processed by the different high-energy beams touch at boundary lines.

[0010] The various high-energy beam scanners are controlled by an electronic control unit. Depending on the desired object to be manufactured, exposure patterns are generated for each scanner from its 3D data for each layer to be produced. These exposure patterns are initially specified in a reference coordinate system. Each scanner is assigned a scanner coordinate system, and each scanner receives from the control unit the coordinates (exposure data) to be approached with the high-energy beam with reference to its scanner coordinate system. For this purpose, the exposure data for this scanner is converted from the reference coordinate system to the respective scanner coordinate system, using a coordinate transformation programmed for the respective scanner.The reference coordinate system is usually linked to the processing machine (machine coordinate system), but the scanner coordinate system of a selected scanner ("guide scanner") can also be selected as the reference coordinate system.

[0011] The coordinate transformation of a scanner, i.e. the relationship between the reference coordinate system and its scanner coordinate system, can be measured experimentally, as stated, for example, in EP 3 907 021 A1 or in WO 2018 / 086996 A1 or in WO 2019 / 173000 A1 or in DE 10 2018 205 403 A1. In general, the scanner is used to generate one or more measuring points (e.g. reflections or melted test patterns) with its high-energy beam on the build platform or a calibration object, the positions of which are determined relative to the reference coordinate system.

[0012] However, during the production of a three-dimensional object on the build platform, the relative orientation of the scanners to the build platform can change over time. This can be caused, for example, by changes in temperature, gas pressure, and humidity. Different scanners are generally affected differently by changes in orientation. In other words, the current, actual coordinate transformations can differ from the programmed coordinate transformations for the scanners (with the exception of the guide scanner, if one is installed). As a result, the surfaces on the build platform processed by different scanners or their high-energy beams can be misaligned with each other.This can result in defects in the structure of the finished object, such as pore seams, local density fluctuations, or even geometric errors, at the boundary lines between the individual layers. This impairs the quality of the finished object.

[0013] From EP 3 907 021 A1, it has become known in a variant that, during the production of an object on a construction platform, after every twelve produced layers, the current, actual coordinate transformations of the other scanners are re-determined relative to a guide scanner, and in an update, the programmed coordinate transformations of the other scanners are set to the current, actual coordinate transformations of the current determination and are used for the production of the next twelve layers.

[0014] This approach allows the programmed coordinate transformations to be tracked and adapted to changes in the scanner orientation. However, it should be noted that each determination of the current, actual coordinate transformations inherently requires time that is then not available for the production of the object itself (so-called non-productive time). Furthermore, there is a risk that between two consecutive determinations of the coordinate transformations, such significant changes in the current, actual coordinate transformation may occur that updating the programmed coordinate transformation itself leads to an error in the structure of the object to be manufactured (for example, creating a visible offset in the object, also known as a "step").

[0015] The object of the invention is to present a method for the layer-by-layer production of at least one object on a construction platform, which enables a high quality of the object with low idle time.

[0016] Description of the invention This object is achieved according to the invention by a method of the type mentioned at the outset, which is characterized in that between two successive determinations of the current, actual coordinate transformations, several updates of the programmed coordinate transformations of the at least Nl scanners are carried out.

[0017] The present invention proposes that between two successive determinations of the coordinate transformations of the at least Nl scanners for a respective scanner, the programmed coordinate transformation is not updated in a single update, but in several updates.

[0018] This makes it possible to divide an accumulated or expected relative adjustment of a particular scanner due to disruptive influences such as temperature fluctuations (e.g. due to solar radiation), pressure fluctuations or humidity fluctuations (e.g. due to weather influences), which manifest themselves in particular in a deviation between the last programmed coordinate transformation and the current, actual coordinate transformation, into smaller individual corrections and to carry out several updates with these. The programmed coordinate transformations of the scanners can thus be adjusted smoothly. The invention makes it possible to minimize or completely avoid errors in the structure of the at least one object to be manufactured on the build platform at the boundary lines between two partial surfaces in a layer that are manufactured with different scanners.

[0019] In this way, structural errors can be minimized or completely avoided by adjusting the scanners relative to each other and / or to the build platform itself, as well as by tracking the programmed coordinate transformations of at least 11 scanners during the processing of a load on the build platform. In particular, pore seams or density fluctuations, or general geometric errors and offsets (which would be visible on surfaces) are avoided in the areas of the boundary lines between the layers or in areas of corresponding cross-layer interfaces.

[0020] Determinations of the current, actual coordinate transformation can be performed comparatively rarely, since even larger accumulated relative adjustments can be tracked and compensated for without introducing noticeable errors into the structure of the object being manufactured. This can save non-productive time and thus increase productivity.

[0021] The determination of the current, actual coordinate transformations of the scanners (or scanner systems) is typically carried out on the basis of a measurement between the production of two layers; however, it is also possible for the determination to be carried out on the basis of several partial measurements that take place over the production of several layers. Measurements with which the current, actual coordinate transformation of a scanner can be determined are known as such in the prior art and can be used within the scope of the present invention. In particular, the procedure within the scope of the invention can be as described in EP 3 907 021 A1 or WO 2018 / 086996 A1; the content of these documents is hereby incorporated by reference into the disclosure of the present invention. For example, one or more test builds can be constructed at defined locations on the build platform, and measurements are carried out in the powder using all of the high-energy beams or lasers involved.Scanners are used to introduce defined structures into the powder or already welded areas by re-melting them; these structures can then be captured with a camera (the structures then represent detectable measurement points). Alternatively, such structures can also be applied to separate areas (with their own build cylinders) outside the actual build area / build platform. It is also possible to scan fixed markers or known bodies with laser beams using all scanners and to determine their current position, for example by detecting the laser light backscattered, reflected back, or generally re-remitted by them and linking it to the scanner's position stamps. In particular, a sapphire sphere can be attached to the coater (which applies the powder to the build platform) as a quasi-retroreflector and can be occasionally introduced into the build space by the coater and scanned.Depending on the number of independent measuring points (per scanner), the relative position (with one or more measuring points) or the orientation / rotational position (with two or more measuring points) of the scanners relative to one another or relative to the machine coordinate system can be determined. When determining the adjustment of the N scanners relative to one another, i.e. for N-1 scanners relative to a guide scanner (whose scanner coordinate system is then permanently defined as a reference coordinate system or with a fixed relationship to the reference coordinate system), updates of the programmed coordinate transformations are usually carried out for N1 scanners, and when determining the adjustment of the N scanners to a machine coordinate system, updates of the coordinate transformations are usually carried out for all N scanners.

[0022] The tracking of the programmed coordinate transformations in the multiple updates between two determinations of the current, actual coordinate transformations of at least one N1 scanner takes into account at least the measured current, actual coordinate transformations of the current determination and can also take into account the measured results of previous determinations. In particular, algorithms from conventional controllers can be used for tracking, especially to limit the manipulated variable (change in the programmed coordinate transformation) in a respective control step (update).

[0023] Within the scope of the invention, it can be provided to limit the tracking of the programmed coordinate transformation per update to a maximum value (with regard to offset and / or rotation), in particular to a specific maximum offset, for example, to 5 pm, 10 pm, or 15 pm per update for a spot diameter of approximately 80-100 pm; for larger spot diameters, a larger maximum offset can be selected, and vice versa. The tracked maximum offset per update is usually in a range of 5%-15% of the spot diameter of the associated high-energy beam. The spot diameter can be determined, for example, according to the 86% criterion, so that 86% of the beam power lies within a circle with the spot diameter.

[0024] Between two determinations of the current, actual coordinate transformations, A updates take place, with A>2, preferably A>5, and particularly preferably A>10. Between two determinations of the current, actual coordinate transformations, a total of M layers are manufactured, with M>2, preferably M>5, particularly preferably M>10, and most preferably M>20. Note that M>A. Each update (or a newly stored set of programmed coordinate transformations) then applies to the subsequent production of one or more layers.

[0025] Preferred variants of the invention

[0026] Particularly preferred is a variant of the method according to the invention which provides that after a respective current determination of the current, actual coordinate transformations for the at least Nl scanners for a respective one of the at least Nl scanners

[0027] - a target coordinate transformation is determined taking into account the current, actual coordinate transformation of the current determination,

[0028] - and with the multiple updates of the programmed coordinate transformation between the current determination and the next determination, the programmed coordinate transformation is gradually transferred to the target coordinate transformation. By gradually transferring the programmed coordinate transformation to the target coordinate transformation, accumulated or expected adjustments of the scanners are smoothly compensated, and errors in the structure of the at least one object to be manufactured on the build platform are minimized. The coordinate transformation reprogrammed with each update (a respective tracking step) then applies to the production of one or more layers until the next update (the next tracking step).Between the current determination and the next determination, the programmed coordinate transformation is generally not changed again after the target coordinate transformation has been reached until the next determination of the current, actual coordinate transformation. If desired, a maximum change (offset and / or rotation) of the programmed coordinate transformation permitted in a single update can be specified for a particular scanner, typically with a maximum value between 2.5 and 25 pm for an offset and a maximum value between 0.05° and 1.0° for a rotation (rotation) in the plane of the build platform.

[0029] A preferred further development of this variant provides that after a respective current determination of the current, actual coordinate transformations for the at least Nl scanners for a respective one of the at least Nl scanners

[0030] - an initial deviation between the target coordinate transformation and the programmed coordinate transformation on which the current determination was based is determined,

[0031] - the determined initial deviation is divided into several deviation components,

[0032] - and for the layers produced after the current determination until the next determination of the current, actual coordinate transformation, the programmed coordinate transformation is changed step by step in the multiple updates, with each update adding a further deviation component to the last programmed coordinate transformation applied by the control device. This procedure is simple and has proven itself in practice.

[0033] A sub-variant of this further development is advantageous, in which the determined initial deviation is divided into equal deviation components. This allows for a particularly smooth adjustment of the programmed coordinate transformation of a particular scanner to the corresponding target coordinate transformation. In this sub-variant, the multiple updates between two consecutive determinations typically occur after the production of an equal number of slices.

[0034] Alternatively, it is also possible to select unequal deviation shares. In particular, earlier updates within a determination interval can have larger deviation shares than later updates. It is also possible to select the first deviation shares in a determination interval according to a maximum permissible deviation share (with respect to offset and / or rotation), and to compensate for any remaining initial deviation in a final update.

[0035] In a further sub-variant, a respective deviation component is limited by a maximum offset and / or a maximum rotation. This allows for highly reliable prevention of defects in the finished object (such as pore seams). A typical maximum offset per update is usually between 2.5 pm and 25 pm, and a maximum rotation per update is usually between 0.05° and 1.0° in the plane of the build platform.

[0036] A sub-variant of the above development is also preferred, which provides that a number of A updates are carried out between two determinations, where A=M is ​​selected, and that the output deviation is divided into M equal deviation components. In other words, an update of the programmed coordinate transformation is carried out for each layer, and the output deviation is distributed evenly among the updates. This also makes it possible to achieve particularly smooth tracking of the programmed coordinate transformation of a respective scanner to the associated target coordinate transformation. A development in which the target coordinate transformation corresponds to the current, actual coordinate transformation of the current determination is particularly preferred.The programmed coordinate transformation of a particular scanner is adjusted to the most recent, available experimental value of the current, actual coordinate transformation until the next determination of the current, actual coordinate transformation. In most applications, this already results in good quality for at least one manufactured object and is also easy to implement. No forecast calculations or forecast models are required, and results from previous determinations of the current, actual coordinate transformations do not need to be consulted or saved. This approach is particularly useful for changes in the orientation of the scanner relative to the build platform that are unpredictable or can only be predicted with great uncertainty.If the current actual coordinate transformation cannot be achieved as a target transformation within a planned determination interval, for example because the initial deviation is too large for the planned number of updates and a maximum adjustment of the programmed coordinate transformation per update, an alternative is to select a coordinate transformation as the target transformation that lies between the current actual coordinate transformation and the coordinate transformation on which the current determination was based.

[0037] Also preferred is a further development in which the target coordinate transformation is determined as a predicted coordinate transformation, taking into account the current, actual coordinate transformations of the current determination and at least B, with B>2, previously determined, current, actual coordinate transformations. This allows the quality of the manufactured, at least one, object to be further increased in many applications. For a respective layer, the programmed coordinate transformation of a respective scanner can be kept closer to the current, actual coordinate transformation; in most cases, the programmed coordinate transformation is then particularly close to the current, actual coordinate transformation at the time of a respective determination.By observing the current, actual coordinate transformation over a large number of (known) points in time in the recent past, an expected coordinate transformation at a specific point in the future ("predicted coordinate transformation") can in many cases be determined with good accuracy through model calculations. The future development of a scanner's adjustment can be anticipated by converting the programmed coordinate transformation in the period up to the said specific point in time to this predicted coordinate transformation. Models such as those known from so-called observers in control engineering can be used to determine the predicted coordinate transformation. Usually, B is chosen with B>3 or B>5 or B>10.If the predicted coordinate transformation cannot be achieved as a target transformation within a planned determination interval, for example because the initial deviation is too large for the planned number of updates and a maximum adjustment of the programmed coordinate transformation per update, an alternative is to select a coordinate transformation as the target transformation that lies between the predicted coordinate transformation and the coordinate transformation on which the current determination was based.

[0038] A sub-variant of this further development involves determining the predicted coordinate transformation using a trend analysis. This allows for slow but consistent changes in scanner alignment, such as those that typically occur due to heating processes during the processing of a load on the build platform, to be compensated quite accurately.

[0039] Preferably, the trend analysis comprises a regression of the current, actual coordinate transformation of the current determination and the at least B, with B>2, previously determined, current actual coordinate transformations. Regression is particularly easy to perform. The regression can, in particular, be a linear regression, polynomial regression, or exponential regression. In the case of a polynomial regression, four or fewer orders (per coordinate direction) are usually sufficient, and often three or fewer orders, and in some cases even two orders.

[0040] In another sub-variant, the target coordinate transformation is determined as an average of the current, actual coordinate transformation of the current determination and the at least B, with B>2, previously determined, current actual coordinate transformations. In other words, it is assumed that the actual coordinate transformation fluctuates around the mean and, as a forecast, will approach the mean again in the future. This procedure is comparatively simple. It leads to good results if the current, actual coordinate transformations used fluctuate primarily statistically, for example, due to unsystematically and rapidly changing external conditions, or due to inherent statistical measurement errors. Note that a separate mean is always determined for each of the at least Nl scanners.

[0041] A particularly preferred variant is one in which the multiple updates between two consecutive determinations are performed after the production of an equal number of slices. This contributes to the particularly smooth tracking of the programmed coordinate transformations and maintains a high quality of the manufactured object.

[0042] A preferred variant is one in which the multiple updates between two consecutive determinations are performed after the production of exactly one layer. In other words, an update only applies to a single layer to be produced. This allows for particularly precise tracking.

[0043] A variant is also preferred in which the multiple updates between two consecutive determinations are distributed evenly across the slices produced between two consecutive determinations. In other words, the entire period between two consecutive determinations is used to track the programmed coordinate transformations, and the updates occur at (at least approximately) equal intervals. This can also contribute to setting up the tracking particularly smoothly. Preferably, a number of A updates are carried out between two determinations, with A>2, with an update being carried out after each production of M / A slices, with M / A: an integer. Typically, equal (additional) deviation components are also provided for each update to correct the programmed coordinate transformation.

[0044] Particularly preferred is a variant in which the number M of manufactured layers is variable between two consecutive determinations of the current, actual coordinate transformations during the production of the at least one object. This allows the frequency of determinations of the current, actual coordinate transformations to be adapted and optimized to the specific needs of the specific application, particularly with regard to a (specified or best possible) manufacturing accuracy and a (as low as possible or specified) proportion of non-productive time for measurements.If desired, M can be selected depending on temporal changes of measured values ​​(e.g., pressure or temperature) within and / or in the environment of the 3D printing system; for example, if a strong change in a measured value (typically associated with a strong adjustment of the scanners) is observed, M should be selected smaller than for a small change.

[0045] A further preferred development of this variant provides that the number M of manufactured layers or a moving average Mgld of the number M of manufactured layers between two consecutive determinations is selected to be lower at the beginning of the layer-by-layer production of the at least one object than during the further course of the layer-by-layer production of the at least one object. Experience has shown that at the beginning of production (i.e., after the fresh build platform has been provided in the coating chamber), transient processes initially occur that suggest greater changes in the alignment of the scanners, so that more frequent monitoring of the coordinate transformations is recommended for precise tracking.Later in the production process, typically when a predetermined proportion of layers (selected, for example, between 10% and 35% of the layers) or a predetermined minimum number of layers (selected, for example, to be 100 or more, preferably 200 or more) or layers have been produced on the build platform during a predetermined minimum time since the start of production (selected, for example, between 1 h and 3 h), only small changes occur, which can be easily tracked with less frequent checks of the coordinate transformation. The moving average Mgld can, for example, be formed over 4 or more, preferably 8 or more, and further, for example, over 24 or fewer, and preferably 12 or fewer, determination intervals (i.e., values ​​of M, typically the most recent values ​​of M).

[0046] A further development is also preferred in which the number M of layers to be produced between a current determination and a next determination of the current, actual coordinate transformation is selected depending on the size of the neighbor deviation between the current, actual coordinate transformation of the current determination and the current, actual coordinate transformation of the determination that preceded the current determination for the at least Nl scanners. This allows for flexible response to disturbances that occur during production.

[0047] In a preferred sub-variant of this further development, the number M of layers to be produced is chosen to be smaller, the greater the neighbor deviation is for the at least Nl scanners. The proposed procedure allows the changes in the programmed coordinate transformations per determination interval to be kept small without making the determination interval unnecessarily short. Typically, an average is calculated for the absolute neighbor deviations of the coordinate transformations of the at least Nl scanners, or the largest deviation occurring in each of the Nl scanners is determined and used for the size estimate.A preferred variant provides that one of the N scanners is selected as the guide scanner, that the scanner coordinate system of the guide scanner or another coordinate system having a fixed relationship to this scanner coordinate system is selected as the reference coordinate system, and that between two successive determinations of the current, actual coordinate transformations, several updates of the programmed coordinate transformations of only the N1 remaining scanners are carried out. For the guide scanner, in the first alternative, the exposure data in the scanner coordinate system simultaneously correspond to the exposure data in the reference coordinate system, so that by selecting the guide scanner, the associated programmed coordinate transformation of the guide scanner is simultaneously set to an identical mapping.In a second alternative, the scanner coordinate system of the lead scanner has a fixed relationship (for loading the build platform) with the additional coordinate system (reference coordinate system), so that the exposure data from the additional coordinate system is converted into the scanner coordinate system using a fixed programmed coordinate transformation. For example, the additional coordinate system can include the machine coordinate system or be identical to it. The programmed coordinate transformation of the lead scanner (e.g., the identical image) is not updated between two determinations. The coordinate transformations of the other NL scanners are defined relative to the lead scanner or relative to its scanner coordinate system, and the coordinate transformations of the other NL scanners are updated multiple times between two determinations. This procedure is particularly simple and reliable.Disturbances in the measuring system generally affect measuring points of all scanner systems (lead scanner and other scanners) equally and therefore have no effect on the (relative) tracking of the N1 other scanners. However, it should be noted that with this procedure, any changes in the relative orientation of the lead scanner to the build platform during the production of at least one object remain undetected and are not tracked. It should also be noted that, within the scope of the invention, a 3D printing system can also be set up with several sets of N scanners (with N>2, where N can be different for different sets), with each set comprising its own lead scanner, and the coordinate transformations of the other scanners of the respective set are tracked to the associated lead scanner or with respect to its reference coordinate system.

[0048] In an alternative variant, the reference coordinate system is a machine coordinate system of a processing machine that includes the N scanners and the build platform. Between two consecutive determinations of the current, actual coordinate transformation, several updates of the programmed coordinate transformations of the N scanners are performed. With this approach, changes in the orientation of all scanners can be detected and compensated for by updating the programmed coordinate transformations of the scanners. However, the measuring system on the processing machine should be particularly stable and not experience any noticeable disturbances that could distort the determination of measurement points in the machine coordinate system.The machine coordinate system is typically defined directly or indirectly by the position of a machine component, whereby the machine component is stationary on the machine or is at least moved to a defined position (typically always the same within the scope of delivery accuracy) for a respective determination, for example, with the feeder. In particular, the machine coordinate system can be defined by a structure scanned by the scanners during a measurement, in particular by a retroreflector, such as a reflector sphere, or by a measurement structure generated by a machine component, such as a light cross of an alignment laser.

[0049] A preferred variant is one in which the programmed coordinate transformations and the current, actual coordinate transformations only comprise displacement information in two orthogonal directions. The two orthogonal directions (x, y) generally lie in the plane of the build platform. Determining the displacement information is comparatively easy and is usually sufficient in practice, since scanner rotations usually do not occur at all or only to a small extent. The displacement is also referred to as offset. In an alternative variant, the programmed coordinate transformations and the current, actual coordinate transformations comprise displacement information in two orthogonal directions as well as rotation information in a plane spanned by the two orthogonal directions.The two orthogonal directions (x, y) are usually in the plane of the build platform, and the rotation occurs around the orthogonal direction (z). By taking scanner rotation into account, the programmed coordinate transformation can be tracked with particular precision.

[0050] The present invention also includes a system for the layer-by-layer production of at least one object on a build platform by locally solidifying powdered material in a respective layer, comprising a build platform, N scanners, where N>2, and a control device configured to carry out a method according to the invention described above. With the system according to the invention, the at least one object can be manufactured with high quality and with low downtimes in a respective loading of a build platform using multiple scanners or associated high-energy beams.

[0051] Also within the scope of the present invention is a computer program product which, when used on a system for the layer-by-layer production of at least one object on a construction platform, carries out a method according to the invention as described above by locally solidifying powdered material in a respective layer.

[0052] Further advantages of the invention will become apparent from the description and the drawings. Likewise, the above-mentioned and further-described features can be used individually or in combinations. The embodiments shown and described are not to be understood as an exhaustive list, but rather as examples for describing the invention. Detailed description of the invention and drawings

[0053] The invention is illustrated in the drawing and is explained in more detail using exemplary embodiments.

[0054] Fig. 1 shows a schematic cross-section of an exemplary system on which the method according to the invention can be carried out and with which an object is manufactured on a construction platform;

[0055] Fig. 2 explains, using a flow chart, the production of a layer of the object to be manufactured for the invention;

[0056] Fig. 3 shows a flowchart of a first variant of the method according to the invention, in which the target coordinate transformation corresponds to the current, actual coordinate transformation of the current determination;

[0057] Fig. 4 shows a flowchart of a second variant of the method according to the invention, in which the target coordinate transformation corresponds to the predicted coordinate transformation;

[0058] Fig. 5 explains four further variants of the method according to the invention using four layer diagrams, with variations after how many manufactured layers a determination is carried out and with variations in the number and distribution of updates between two determinations;

[0059] Fig. 6 shows an example of the implementation of a seventh variant of the method according to the invention, with a layer diagram, example calculations and a selection table;

[0060] Fig. 7a shows, using a diagram for an eighth variant of the method according to the invention, the determination of the predicted coordinate transformation using a polynomial regression; Fig. 7b shows, using a diagram for a ninth variant of the method according to the invention, the determination of the predicted coordinate transformation using a linear regression;

[0061] Fig. 8 shows, using a diagram for a tenth variant of the method according to the invention, the determination of the predicted coordinate transformation by means of the formation of an average value.

[0062] Fig. 1 schematically shows a system 1, comprising a processing machine 1a and a control device 10, for the invention. The system 1 is used to manufacture a three-dimensional object 2 from a powdered material 3. A computer program product that executes the method according to the invention can be used on the system 1.

[0063] The processing machine 1a comprises a processing chamber 4 located within a housing. Shown here schematically is a cover 5a with additional components (see below) and a floor 5b, which define the processing chamber 4. The processing chamber 4 is further defined by side walls and a rear wall (not shown in detail). The processing chamber 4 can be accessed via an access door (not shown in detail).

[0064] The manufacturing process is carried out on a build platform 6. The build platform 6 can be lowered along an axis in a direction Z relative to the floor 5b in order to gradually arrange new layers of powdered material 3 for the layer-by-layer production of the object 2 on the build platform 6; the powdered material 3 is applied using a feeder (not shown in detail). In Fig. 1, the build platform 6 is already lowered relative to the floor 5b.

[0065] To produce the object 2, here a prototype 2a, a top layer 7 of the powdered material 3 is irradiated with two high-energy beams 8a, 8b, here laser beams, which are directed onto the top layer 7 by two scanners 9a, 9b. The scanners 9a, 9b comprise tiltable mirrors (not shown in detail). The high-energy beams 8a, 8b are directed to predetermined positions in the powdered material 3. The high-energy beams 8a, 8b are absorbed by the powdered material 3. The powdered material 3 melts and solidifies again when it is no longer irradiated. The solidified material then forms another part of the object 2 to be manufactured.

[0066] The two scanners 9a, 9b are connected to the control device 10. For exposure of a respective layer 7, the control device 10 provides exposure data of a processing pattern in a reference coordinate system for each scanner 9a, 9b. The control device 10 can determine this exposure data from CAD data of the object 2 to be manufactured. A scanner coordinate system is each assigned to the scanners 9a, 9b. In Fig. 1, the reference coordinate system corresponds to the scanner coordinate system of scanner 9a. Scanner 9a is then referred to as the guide scanner 11. Alternatively, a machine coordinate system of the processing machine 1a can also be selected as the reference coordinate system.

[0067] In the control device 10, the exposure data in the reference coordinate system for each scanner is converted into exposure data in the scanner coordinate system using a programmed coordinate transformation. For the scanner coordinate system of the guide scanner 11, the exposure data in the reference coordinate system and in the scanner coordinate system are identical. For the scanner coordinate system of the scanner 9b, the exposure data in the reference coordinate system differ from those in the scanner coordinate system. The exposure data are forwarded in the respective scanner coordinate system to the associated scanner 9a, 9b. The scanners 9a, 9b then expose the processing pattern on the build platform 6 in layer 7.

[0068] In Fig. 1, the processing machine 1a further comprises a monitoring device 12, here a camera 12a, with which production is monitored. The monitoring device 12 is connected to the control device 10. During the production of the object 2, measurements are repeatedly carried out using the camera 12a. These repeated measurements result in repeated determinations of the current actual coordinate transformation of scanner 9b. The current actual coordinate transformation of scanner 9b can deviate from the programmed coordinate transformation on which the associated determination is based, since the ambient conditions and influences in the processing chamber 4 can change during production (for example, the processing chamber 4 can heat up during processing or there can be a change in the differential pressure or the air humidity). As a result, scanner 9b orshift or rotate its scanner coordinate system relative to the guide scanner 11. To correct these shifts or rotations, the programmed coordinate transformation of scanner 9b is updated, taking into account the current actual coordinate transformation of scanner 9b.

[0069] Note that in case the reference coordinate system is a machine coordinate system, the instantaneous actual coordinate transformations for all scanners (9a, 9b) are repeatedly determined and updated by repeated measurements.

[0070] Fig. 2 shows a flowchart explaining how the production of a single layer of an object to be manufactured takes place within the scope of the invention. The production process is controlled via the control device.

[0071] Initially, in step 100, the planning for exposing a layer of the powdered material takes place to produce a layer of the object. For this purpose, it is determined which surface areas of the entire layer of powdered material must be exposed to the scanner's high-energy beams in order to produce the corresponding layer of the object (layer processing area). The layer processing area is essentially determined from the object's CAD data and the position of the current layer in the object.

[0072] In a next step 101, the layer processing area is distributed among the N scanners involved in the production of the object. Accordingly, N processing patterns are determined for the N scanners. Typically, the distribution among the N processing patterns is carried out in such a way that the processing time of each scanner is as equal as possible. The following steps 102, 103, 104 are carried out separately for each of the N scanners. In step 102, the exposure data (target position of the beam spot of the high-energy beam for a plurality of times) of the respective processing pattern of the respective scanner are first made available in the reference coordinate system.

[0073] In the subsequent step 103, the exposure data in the reference coordinate system are converted into exposure data in the scanner coordinate system of the respective scanner. The conversion is performed using the programmed coordinate transformation for this scanner.

[0074] In step 104, the exposure data in the scanner coordinate system is instructed to the respective scanner (passed on for execution).

[0075] Then, in step 105, the layer on the build platform is exposed using the N scanners and the corresponding layer of the object is manufactured.

[0076] After the exposure is completed, production of the next layer continues (the process starts again at step 100).

[0077] Within the scope of the invention, the programmed coordinate transformations are updated in a specific manner, as described below.

[0078] Fig. 3 shows a flowchart of a first variant of the method according to the invention. If the reference coordinate system is defined as the machine coordinate system, this process is applied to all N scanners. If the reference coordinate system is defined as the scanner coordinate system of a lead scanner, this process is applied to N1 scanners (i.e., all scanners except the lead scanner).

[0079] In step 200, in the variant shown, layers have already been produced on the build platform, and the object to be manufactured has already been partially manufactured. This can be referred to as previous production.

[0080] In a next step 201, the current determination of a current, actual coordinate transformation MTKTAB of the scanner takes place. The scanner is used in a measurement, whereby one or more measurement points were generated with the high-energy beam. Known coordinates were specified in the reference coordinate system, which were converted into coordinates of the scanner coordinate system using a programmed coordinate transformation PKT AB underlying the current determination, and then instructed to the scanner. The actual location of the measurement point(s) was then measured in the reference coordinate system ("actual coordinates"). The current, actual coordinate transformation MTKTAB can then be determined from the programmed coordinate transformation PKTAB of the current determination plus any deviation, if any, of the specified known coordinates from the measured, actual coordinates.Typically, during production, due to changes in the ambient conditions in the processing chamber and the resulting adjustments (shifts and possibly also rotations) of the scanners relative to each other or of the scanner relative to the processing machine, the current, actual coordinate transformation MTKTAB will differ from the previously programmed coordinate transformation PKTAB.

[0081] In order to correct these adjustments, a target coordinate transformation ZKT is determined in step 202. In the variant shown here, the target coordinate transformation ZKT is set equal to the current actual coordinate transformation MTKTAB of the current determination. This means that

[0082] ZKT = MTKTAB. (Eq. 1)

[0083] Note that the coordinate transformation contains values ​​for each coordinate direction (x, y) and possibly also a rotation ( <p) umfassen, die einzeln betrachtet werden können, hier aber zur Vereinfachung in der Darstellung zu einem Formelzeichen zusammengefasst sind.

[0084] In step 203, an output deviation OFF is then determined. The output deviation OFF is given by

[0085] OFF = ZKT - PKTAB (Eq. 2) and indicates the deviation of the target coordinate transformation ZKT from the programmed coordinate transformation PKTAB. In the following step 204, the determined output deviation OFF is divided into deviation components ABWj for A updates, with i = l,...,A, where i = update index. The following then applies:

[0086] OFF = ^ =1 ABW ir(Eq. 3) ie the sum of all deviation components ABWi results in the output deviation AUS. Furthermore, the M layers to be manufactured until the next determination of the current, actual coordinate transformation are divided into A production blocks, each with nrii layers for the A updates. The following then applies:

[0087] M = Zf =1 m . (Eq. 4)

[0088] In step 205.1, a new programmed coordinate transformation PKTi is determined after the first update j = l (with j = l,...,A, and j: index of the current update). The programmed coordinate transformation PKTi is calculated as follows:

[0089] PKTi = PKTAB + DEV1. (Eq. 5)

[0090] The new programmed coordinate transformation PKTi replaces the last programmed coordinate transformation (here PKTAB) and is stored in the control unit. The updated programmed coordinate transformation PKTi is then used to produce the mi layers of the production block belonging to the first update.

[0091] The same procedure is followed for subsequent updates. Step 2O5.j shows the general case of determining a programmed coordinate transformation PKTj and manufacturing the production block of mj layers after update j. The programmed coordinate transformation PKTj is calculated as follows:

[0092] PKTj = PKTj-i + ABWj (Eq. 6) which is equivalent to

[0093] PKTj = PKTi + { =1 ABWi. (Eq. 7)

[0094] The programmed coordinate transformation PKTj is stored and then applied to the production of the rnij layers of the production block belonging to the update j.

[0095] In step 205. A, a programmed coordinate transformation PKTA is determined after the last update j=A. The programmed coordinate transformation PKTA is calculated as follows:

[0096] PKTA = PKTA-I + ABWA (Eq. 8) which is equivalent to

[0097] PKTA = PKTi + f =1 ABWi. (Eq. 9)

[0098] The programmed coordinate transformation PKTA is saved and applied during the production of the mA layers of the production block belonging to the last update j=A.

[0099] This is followed by step 206, in which the next determination of the current, actual coordinate transformation is performed. Subsequently, additional layers can be manufactured and the programmed coordinate transformation updated as described in steps 202-205.A, and further determinations of the current, actual coordinate transformation are made, and so on. This is summarized in step 207. Manufacturing continues, i.e., steps 201 to 205.A are repeated until the object is finished. In step 208, processing of the object to be manufactured on the build platform is terminated.

[0100] During the execution of the first variant of the method according to the invention, several updates (number A) of a programmed coordinate transformation PKT take place between two consecutive determinations. This allows accumulated adjustments of a scanner to be divided into small individual corrections, and the programmed coordinate transformation PKT of the scanner can be smoothly adjusted. This can improve the quality of the object to be manufactured. Equating the current, actual coordinate transformation MTKTAB of the current determination with the target coordinate transformation ZKT is easy to implement and results in good quality of the object to be manufactured.

[0101] Fig. 4 shows a flowchart of a second variant of the method according to the invention. If the reference coordinate system is defined as the machine coordinate system, this process is applied to all N scanners. If the reference coordinate system is defined as the scanner coordinate system of a lead scanner, this process is applied to N1 scanners (i.e., all scanners except the lead scanner). The process of the second variant largely corresponds to the process of the first variant (see Fig. 3), so only the essential differences are explained here.

[0102] In a step 300, layers have already been created on the build platform, and the object to be manufactured has already been partially manufactured. This can be referred to as previous production. During previous production, several current, actual coordinate transformations of the scanner have already been determined, see step 300. a. The B (with B>2) most recently determined, current actual coordinate transformations MTKTZB with k=1,...,B and k: counting index, are stored in the control device, along with the time of their determination.

[0103] In a next step 301, the current, actual coordinate transformation MTKTAB of the scanner is determined.

[0104] In step 302, the target coordinate transformation ZKT is determined. In the variant shown here, a predicted coordinate transformation PNSKT is first determined for this purpose. The predicted coordinate transformation is a function of the currently determined, instantaneous actual coordinate transformation MTKTAB and the other previously determined, instantaneous actual coordinate transformations MTKTZB , with k=l,...,B (previous step 300.a). The predicted coordinate transformation PNSKT is thus determined via

[0105] PNSKT = f (MTKTAB, MTKTZB,I,...,MTKTZB,B). (GL. 10)

[0106] Here, f indicates a forecast function. The forecast coordinate transformation PNSKT can be determined, for example, via a trend analysis (e.g., a regression, see Fig. 7a, 7b) or by determining a mean value (see Fig. 8). The target coordinate transformation ZKT is then equated with the forecast coordinate transformation PNSKT. Therefore,

[0107] ZKT = PNSKT. (Eq. 11)

[0108] With the target coordinate transformation ZKT determined in this way, production blocks of mj layers each can be manufactured, with j = l,...,A, as described in steps 303-305. A, until the next determination of the current, actual coordinate transformation 306 by making A updates of the programmed coordinate transformation to values ​​PKTj A. Steps 303-308 correspond to steps 203-208 (see above).

[0109] During the execution of the second variant of the method according to the invention, several updates of the programmed coordinate transformation PKT take place between two consecutive determinations. This allows accumulated and expected adjustments of a scanner to be divided into small individual corrections, and the programmed coordinate transformation PKT of the scanner can be smoothly adjusted. This can improve the quality of the object to be manufactured. Equating the predicted coordinate transformation PNSKT with the target coordinate transformation ZKT enables effective countermeasures against future developments in scanner adjustment. A good quality of the object to be manufactured can be achieved.

[0110] Fig. 5 shows further variants of the method according to the invention using various slice diagrams for a single scanner. The slice diagrams are plotted against time t and show the index numbers of the respective slices (Arabic numerals) produced between two determinations (determinations are shown as dashed lines) of the current actual coordinate transformation, as well as the number and distribution of updates (updates are shown as dotted lines) of the programmed coordinate transformation performed between two determinations.

[0111] It is understood that only characteristic determination intervals are shown, and not the entire production in one loading of the build platform.

[0112] Partial image a) shows a third variant of the method according to the invention. The current actual coordinate transformation is determined at times t1a, t2a, and t3a. Between times t1a and t2a and between times t2a and t3a, five layers (M=5) of the object to be manufactured are produced. An update is performed after each layer is produced; thus, five updates are performed between times t1a and t2a and between times t2a and t3a.

[0113] After determining the current, actual coordinate transformation at time t1a, the initial deviation is determined and divided into five equal deviation components, which are then distributed among the five updates until the next determination at time t2a. The first update of the programmed coordinate transformation is then performed, and layer 1 is manufactured. Between layer 1 and layer 2, the programmed coordinate transformation is updated again, and layer 2 is manufactured. Layers 3 to 5 are manufactured in the same way. After layer 5 has been manufactured, the current, actual coordinate transformation is determined at time t2a.

[0114] The production of layers No. 1 to No. 5 between times t2a and t3a is carried out analogously to the production of layers No. 1 to No. 5 between times t1a and t2a.

[0115] In this variant, the programmed coordinate transformation is updated at equal intervals in particularly small steps. Partial figure bj shows a fourth variant of the method according to the invention. The current actual coordinate transformation is determined at times tlb, t2b, and t3b. Between times tlb and t2b and between times t2b and t3b, eight shifts (M=8) are produced. The shifts are divided into production blocks of two shifts per production block. After each production block has been produced, an update is performed; thus, four updates are performed between times t1a and t2a and between times t2a and t3a.

[0116] After determining the current, actual coordinate transformation at time t1b, the initial deviation is determined and divided into four equal deviation components, which are then distributed among the four updates until the next determination at time t2b. The first update of the programmed coordinate transformation is then performed, and layers 1 and 2 are manufactured. Between layer 2 and layer 3, the programmed coordinate transformation is updated again, and layers 3 and 4 are manufactured. Layers 5 to 8 are manufactured in the same way. After layer 8 has been manufactured, the current, actual coordinate transformation is determined at time t2b.

[0117] The production of layers No. 1 to No. 8 between times t2b and t3b is carried out analogously to the production of layers No. 1 to No. 8 between times tlb and t2b.

[0118] In this variant, the updates also occur at equal intervals, but less frequently than in the variant of sub-figure a).

[0119] Partial image cj shows a fifth variant of the method according to the invention. In the variant shown here, the production of the object has only just begun. The current actual coordinate transformation is determined at times t1c, t2c, t3c, and t4c. Five layers are produced between times t1c and t2c and between times t2c and t3c. Eight layers are produced between times t3c and t4c. After each produced layer, an update is performed; five updates are performed between times t1c and t2c and between times t2c and t3c, and eight updates are performed between times t3c and t4c.

[0120] After determining the current, actual coordinate transformation at time t1c, the initial deviation is determined and divided into five equal deviation components, which are then distributed among the five updates until the next determination at time t2c (determination interval B11). The first update of the programmed coordinate transformation is then performed, and layer 1 is manufactured. Between layer 1 and layer 2, the programmed coordinate transformation is updated again, and layer 2 is manufactured. Layers 3 to 5 are manufactured in the same way. After layer 5 has been manufactured, the current, actual coordinate transformation is determined at time t2c.

[0121] The production of layers No. 1 to No. 5 between the times t2c and t3c (determination interval BI2) is carried out analogously to the production of layers No. 1 to No. 5 between the times tlc and t2c.

[0122] As shift-by-shift production progresses, the number of manufactured layers is increased to eight between times t3c and t4c (determination interval BI3). After determining the current, actual coordinate transformation at time t3c, the initial deviation is determined and divided into eight equal deviation components, which are then distributed among the eight updates until the next determination at time t4c. The first update of the programmed coordinate transformation is then carried out, and layer 1 is manufactured. Between shift 1 and shift 2, the programmed coordinate transformation is updated again, and layer 2 is manufactured. Layers 3 to 8 are manufactured in the same way. After layer 8 has been manufactured, the current, actual coordinate transformation is determined at time t4c.

[0123] The determination intervals B11 and BI2 represent early (first processed), shorter determination intervals, and the determination interval BI3 represents late (later processed), longer determination intervals. Processing the entire loading height of a build platform typically requires at least 20 determination intervals, and often at least 40 determination intervals.

[0124] This variant allows for more precise and faster tracking of the programmed coordinate transformation at the beginning of the object's production. Typically, more significant changes in the scanner alignment are to be expected at the beginning of production, which is why determinations are performed more frequently at this stage to monitor these changes and counteract them as quickly and effectively as possible. The later determination intervals are usually at least 1.5 times longer than the earlier determination intervals, and often at least twice or even three times longer than the earlier processing intervals.

[0125] Part d) shows a sixth variant of the method according to the invention. The determination of the current actual coordinate transformation occurs at times t1d and t2d. Between times t1d and t2d, 16 layers are produced. Updates are only performed at the beginning, namely before the production of layers 1 to 5; thus, a total of five updates are performed.

[0126] After determining the current, actual coordinate transformation at time tld, the initial deviation is determined and divided into five equal deviation components, which are then distributed among the five updates. The first update of the programmed coordinate transformation is then performed, and layer 1 is manufactured. Between layer 1 and layer 2, the programmed coordinate transformation is updated again, and layer 2 is manufactured. Layers 3 to 5 are manufactured in the same way. After layer 5 has been manufactured, layers 6 to 16 are manufactured without further updates, before the next determination of the current, actual coordinate transformation takes place at time t2d.

[0127] In all the variants shown above, the initial deviation between two determinations was always divided into equal deviation components. Alternatively, it is also possible to select different deviation components between two determinations (not shown in detail). It is important to note that the deviation components must not be chosen too large to avoid the formation of steps or offsets during production of the object to be manufactured.

[0128] Fig. 6 shows an example of the implementation of a seventh variant of the method according to the invention. In this variant, the currently determined actual coordinate transformation is equated with the target coordinate transformation.

[0129] Part a) shows a layer diagram for a single scanner; the representation is analogous to Fig. 5 (see above).

[0130] The current actual coordinate transformation is determined at times t1, t2, t3, and t4. Ten layers are produced between times t1 and t2 and between times t3 and t4, and five layers are produced between times t2 and t3. An update is performed after each layer is produced; thus, ten updates are performed between times t1 and t2 and between times t3 and t4, and five updates are performed between times t2 and t3.

[0131] In the variant shown here, the number of layers produced between determinations is determined based on the neighboring deviations of the current, actual coordinate transformations between adjacent times t1, t2, t3, t4. The neighboring deviations are determined based on the offset values ​​of the coordinate transformations. The exact procedure is explained in more detail in sub-figures b) and c).

[0132] Part b) lists the determined current actual coordinate transformation and the programmed coordinate transformation underlying the determination, each at times t1 to t4 for the selected example. Also listed are the respective neighboring deviation and the amount of the offset between times t1 and t2, t2 and t3, and t3 and t4. In the variant shown here, the respective neighboring deviations NAB - Wtg / h are determined as follows: NABWtg / h = MTKTth - MTKTtg (Eq. 12) with tg and th: adjacent times, and with MTKT tg and MTKTth: Determination of the instantaneous, actual coordinate transformation at times tg and th; h and g are the time indices, with h: 2,...,4 and g: 1,...,3.

[0133] In the variant shown here, the coordinate transformations or the associated displacement information and the neighboring deviations are considered in one direction x and one direction y. The x-direction and the y-direction are spanned orthogonally to each other and lie in the plane of the build platform. The determination of the offset BV tg / h of neighboring deviations between the neighboring times tg and th is carried out as follows (via the Pythagorean theorem):

[0134] BVtg / h = x(NABW tg / fl ) 2 + y(NABW tg / fl ) 2 (Eq. 13)

[0135] Figure c) shows a table with selection criteria for the selected example. The determination interval, i.e., the interval of shifts between determinations, is adjusted according to the amount of offset. Here, the selection criteria are chosen so that for BV<5pm, 15 shifts are produced between two determinations, and for 5pm <BV<15pm zwischen zwei Bestimmungen zehn Schichten gefertigt werden und dass bei BV> 15pm between two determinations five shifts are manufactured.

[0136] This variant of the method according to the invention will be further explained using three examples.

[0137] Example 1 :

[0138] Between times t1 and t2, ten shifts were produced and ten updates of the programmed coordinate transformation were made, as shown in sub-figure a); this was determined without applying the table in sub-figure c). A comparison of the instantaneous, actual coordinate transformation MTKTti at time t1 and the instantaneous, actual coordinate transformation MTKT t 2 at time t2 results in a neighbor deviation NABWti / 2 for the x-coordinate of x=12|jm (since 17|jm-5|jm = 12|jm) and for the y-coordinate of y=18|jm (since 25|jm-7|jm = 18|jm). The offset BVti / 2 between times t1 and t2 is then BV t i / 2=21.6pm (since A / (12|in) 2 + (18|in) 2=21.6 pm). This offset is quite large. According to the selection criteria from the table in Figure c), this means that the interval of completed layers between times t2 and t3 is selected to be only five layers. This allows for an earlier determination. The determination interval of M = 5 layers is relatively small, so presumably only a small offset will accumulate until the next determination.

[0139] Example 2:

[0140] Between times t2 and t3, five shifts were produced as previously defined, and five updates of the programmed coordinate transformation were made, as shown in sub-figure a). A comparison of the current, actual coordinate transformation MTKT t 2 at time t2 and the current, actual coordinate transformation MTKTts ​​at time t3 results in a neighbor deviation NABW t2 / 3 for the x-coordinate of x=-8pm (since 9pm-17pm=-8pm) and for the y-coordinate of y=-7pm (since 18pm-25pm=-7pm). For the offset amount BV t 2 / 3 between times t2 and t3 then follows BVt2 / 3=10.6pm (since y'f -8 ! 1111 ) 2 + (-7|in) 2 = 10.6 pm). According to the selection criteria from the table in sub-figure c), this means that the interval of completed shifts between times t3 and t4 is chosen to be ten shifts. Thus, more shifts are completed before a further determination is made than in the first example.

[0141] Example 3:

[0142] Between times t3 and t4, ten shifts were produced as previously defined, and ten updates of the programmed coordinate transformation were made, as shown in sub-figure a). A comparison of the current, actual coordinate transformation MTKTts ​​at time t3 and the current, actual coordinate transformation MTKT t 4 at time t4 results in a neighbor deviation NABW t 3 / 4 for the x-coordinate of x=2pm (since ll|jm-9|jm=2|jm) and for the y-coordinate of y=-3|jm (since 15|jm-18|jm=-3|jm). For the offset BV t 3 / 4 between times t3 and t4 then follows BVt3 / 4=3.2pm (since (2|im) 2 + (-3|in) 2 = 3.2 pm). According to the selection criteria from the table in sub-figure c), this means that the interval of completed shifts between time t4 and the following time (not shown in detail) is selected as 15 shifts.

[0143] This variant of the method according to the invention allows for flexible adaptation to different situations during the production of the object to be manufactured. For large neighboring deviations, the measurement intervals are shortened, enabling more precise monitoring of production. For small neighboring deviations, the measurement intervals are increased, allowing less time for measurements and faster completion of production.

[0144] Fig. 7a shows the determination of the predicted coordinate transformation PNSKT by means of a polynomial regression for an eighth variant of the method according to the invention.

[0145] The diagram plots the x-coordinate of the current, actual coordinate transformation MTKT against the number of completed shifts (the number of completed shifts corresponds to one time coordinate). Three determinations or measurement points (filled circles) are shown in the diagram: two previously determined, current, actual coordinate transformations MTKTZBI and MTKTZB2, and the currently determined, current, actual coordinate transformation MTKTAB. The same number of completed shifts lie between all measurement points. A polynomial regression is performed, and a second-degree polynomial is fitted to the measurement points as a regression curve. The regression curve can be used to determine the predicted coordinate transformation PNSKT at a future point in time or after a future completed shift SZ (see open circle).

[0146] Note that the determination of the predicted coordinate transformation PNSKT is not only applied to the x-component, but also to the y-component and possibly a rotational component of the current, actual coordinate transformation MTKT (not shown in detail).

[0147] Fig. 7b shows the determination of the predicted coordinate transformation PNSKT using linear regression for a ninth variant of the method according to the invention. Only the essential differences from the variant in Fig. 7a are explained.

[0148] A linear regression is performed, and a regression curve in the form of a regression line is fitted to the measurement points. This also allows the predicted coordinate transformation PNSKT to be determined at a selected future point in time, or after the future production of a selected shift SZ.

[0149] Fig. 8 shows the determination of the predicted coordinate transformation PNSKT by calculating an average for a tenth variant of the method according to the invention. Only the essential differences from the variant shown in Fig. 7a are explained.

[0150] A mean value is determined from the three measurement points: The mean value is marked by the horizontal straight line. It is assumed that the current, actual coordinate transformation will settle around this mean value or return to this mean value. This allows the predicted coordinate transformation PNSKT to be determined at the selected future point in time or after the future production of a selected shift SZ (see empty circle).

[0151] List of reference symbols

[0152] 1 system la processing machine

[0153] 2 objects

[0154] 2a Prototyp

[0155] 3 powdered material

[0156] 4 processing chamber

[0157] 5a Cover

[0158] 5b Floor

[0159] 6 Construction platform

[0160] 7th layer (top layer here)

[0161] 8a, 8b High-energy beam

[0162] 9a, 9b Scanner

[0163] 10 Control device

[0164] 11 guide scanners

[0165] 12 Monitoring device

[0166] 12a Camera

[0167] A Number of updates between two provisions

[0168] OFF Output deviation

[0169] ABWi i-th deviation share

[0170] B Number of previously determined, current actual coordinate transformations used for the forecast

[0171] BI1-BI3 determination interval

[0172] B tg / h Amount of the offset of the current, actual coordinate transformation between the neighboring times tg, th g Index h Index i Index j Index k Index

[0173] M Number of layers produced between two determinations mi Number of layers in the production block after update i

[0174] MTKT current, actual coordinate transformation MTKTAB current actual coordinate transformation of the current determination

[0175] MTKTtg instantaneous actual coordinate transformation at time tg

[0176] MTKTth instantaneous actual coordinate transformation at time th

[0177] MTKTZB previously determined, current actual coordinate transformation

[0178] N Number of scanners and number of high-energy beams

[0179] NABWtg / h Neighbor deviation of the current actual coordinate transformation between the times t gand th

[0180] PKT programmed coordinate transformation

[0181] PKTi programmed coordinate transformation after update 1

[0182] PKTA programmed coordinate transformation after update A

[0183] PKTAB programmed coordinate transformation underlying the current determination

[0184] PKTj programmed coordinate transformation after update j

[0185] PNSKT predicted coordinate transformation

[0186] SZ selected shift that will be manufactured in the future t time tl-t4 time tla-t3a time tlb-t3b time tlc-t4c time tld, t2d time tg time th time x direction y direction

[0187] Z direction

[0188] ZKT target coordinate transformation

Claims

Claims: A method for the layer-by-layer fabrication of at least one object (2) on a build platform (6) by local solidification of powdered material (3) in a respective layer (7), wherein at least in a plurality of the layers (7) N high-energy beams (8a, 8b) are used simultaneously at least temporarily with N scanners (9a, 9b), with N>2, wherein a scanner coordinate system is assigned to each scanner (9a, 9b), wherein a control device (10) for exposure of a respective layer (7) for each scanner (9a, 9b) - provides exposure data of a processing pattern in a reference coordinate system (102), - converts the exposure data in the reference coordinate system into exposure data in the scanner coordinate system using a programmed coordinate transformation (PKT) (103), and - instructs the obtained exposure data in the scanner coordinate system to the associated scanner (9a, 9b) (104), so that the scanner (9a, 9b) exposes the processing pattern on the build platform (6) in layer (7) (105), wherein measurements are repeatedly taken during the layer-by-layer production of the at least one object (2), with which the instantaneous actual coordinate transformations (MTKT) of at least Nl scanners (9a, 9b) are determined, wherein between two successive determinations of the instantaneous actual coordinate transformations (MTKT) M layers (7) are produced, with M > 2, and wherein during the production of the at least one object (2) the programmed coordinate transformations (PKT) for the at least N- 1 scanners (9a, 9b) are determined taking into account the instantaneous actual Coordinate transformations (MTKT) are updated (205.1, 205.j, 205.A; 305.1, 305.j, 305.A), characterized in that between two successive determinations (201, 206; 301, 306) of the instantaneous, actual coordinate transformations (MTKT), several updates of the programmed coordinate transformations (PKT) of the at least 1 scanners (9a, 9b) are performed. Method according to claim 1, characterized in that after each current determination (201; 301) of the instantaneous, actual coordinate transformations (MTKTAB) for the at least 1 scanners (9a, 9b), for each of the at least 1 scanners (9a, 9b) - taking into account the current, actual coordinate transformation (MTKTAB) of the current determination (201; 301), a target coordinate transformation (ZKT) is determined (202; 302), - and with the multiple updates of the programmed coordinate transformation (PKT) between the current determination (201; 301) and the next determination (206; 306), the programmed coordinate transformation (PKT) is gradually converted to the target coordinate transformation (ZKT) (205.1, 205.j, 205. A; 305.1, 305.j, 305. A). Method according to claim 2, characterized in that, after each current determination (201; 301) of the instantaneous, actual coordinate transformations (MTKTAB) for the at least Nl scanners (9a, 9b) for each of the at least Nl scanners (9a, 9b) - an initial deviation (AUS) between the target coordinate transformation (ZKT) and the programmed coordinate transformation (PKTAB) on which the current determination was based is determined (203; 303), - the determined initial deviation (AUS) is divided into several deviation components (ABWi) (204; 304), - and for the layers (7) which according to the current determination (201; 301) The programmed coordinate transformation (PKT) is modified stepwise in several updates (205.1, 205.j, 205.A; 305.1, 305.j, 305.A) until the next determination (206; 306) of the current, actual coordinate transformation (205.1, 205.j, 205.A; 305.1, 305.j, 305.A), with each update adding a further deviation component (ABWQ) to a programmed coordinate transformation (PKT) last applied by the control unit (10). A method according to claim 3, characterized in that the determined initial deviation (AUS) is divided into equal deviation components (ABWQ). A method according to claim 3 or 4, characterized in that each deviation component (ABWi) is limited by a maximum offset and / or a maximum rotation.A method according to any one of claims 3 to 5, characterized in that a number of A updates are performed between two determinations (201, 206; 301, 306), where A=M is ​​selected, and that the initial deviation (AUS) is divided into M equal deviation components (ABWi). A method according to any one of claims 2 to 6, characterized in that the target coordinate transformation (ZKT) corresponds to the instantaneous, actual coordinate transformation (MTKTAB) of the current determination (201). A method according to any one of claims 2 to 6, characterized in that the target coordinate transformation (ZKT) is determined taking into account the instantaneous, actual coordinate transformations (MTKTAB) of the current determination (301) and at least B, with B>2, previously determined (300. a), instantaneous actual coordinate transformations. (MTKTZB ) is determined as a predicted coordinate transformation (PNSKT).

9. Method according to claim 8, characterized in that the predicted coordinate transformation (PNSKT) is determined by means of a trend analysis.

10. Method according to claim 9, characterized in that the trend analysis comprises a regression of the instantaneous actual coordinate transformation (MTKTAB) of the current determination (301) and the at least B, with B>2, previously determined (300. a), instantaneous actual coordinate transformations (MTKTZB ).

11. Method according to claim 8, characterized in that the target coordinate transformation (ZKT) is determined as an average of the instantaneous actual coordinate transformation (MTKTAB) of the current determination (301) and the at least B, with B>2, previously determined (300. a), instantaneous actual coordinate transformations (MTKTZB ).

12. Method according to one of the preceding claims, characterized in that the multiple updates between two successive determinations (201, 206; 301, 306) each take place after the production of an equal number (mi) of layers (7).

13. Method according to one of the preceding claims, characterized in that the multiple updates between two successive determinations (201, 206; 301, 306) take place after production of exactly one layer (7).

14. Method according to one of the preceding claims, characterized in that the multiple updates between two successive determinations (201, 206; 301, 306) are distributed evenly across the inter- The method according to one of the preceding claims is characterized in that the number M of manufactured layers (7) is distributed between two successive determinations (201, 206; 301, 306) of the instantaneous, actual coordinate transformations (MTKT) during the manufacturing of the at least one object (2). The method according to claim 15 is characterized in that the number M of manufactured layers (7) or a moving average Mgld of the number M of manufactured layers (7) between two successive determinations (201, 206; 301, 306) is selected to be lower at the beginning of the layer-by-layer manufacturing of the at least one object (2) than in the further course of the layer-by-layer manufacturing of the at least one object (2).Method according to claim 15 or 16, characterized in that the number M of layers (7) to be produced between a current determination (201; 301) and a next determination (206; 306) of the instantaneous, actual coordinate transformation (MTKT) is selected depending on how large a neighbor deviation (NABW) is for each of the at least Nl scanners (9a, 9b). tg / h) between the instantaneous, actual coordinate transformation (MTKTAB) of the current determination (201; 301) and the instantaneous, actual coordinate transformation of the determination that preceded the current determination. Method according to claim 17, characterized in that the number M of layers (7) to be produced is chosen to be smaller the greater the neighbor deviation (NABW) is for the at least Nl scanners. tg / h). Method according to one of claims 1 to 18, characterized in that one of the N scanners (9a, 9b) is selected as the guide scanner (11), that the scanner coordinate system of the guide scanner (11) or a coordinate system fixedly related to this scanner coordinate system is selected as the reference coordinate system, and that between two successive determinations (201, 206; 301, 306) of the instantaneous, actual coordinate transformations (MTKT), several updates of the programmed coordinate transformations of only the N1 remaining scanners (9a, 9b) are performed. Method according to one of claims 1 to 18, characterized in that the reference coordinate system is a machine coordinate system of a processing machine (1a) comprising the N scanners (9a, 9b) and the build platform (6), and that between two successive determinations (201, 206; 301, 306) of the instantaneous, actual coordinate transformation (MTKT), several updates of the programmed coordinate transformations of the N scanners (9a, 9b) are performed.A method according to any one of claims 1 to 20, characterized in that the programmed coordinate transformations (PKT) and the instantaneous, actual coordinate transformations (MTKT) comprise only translation information in two orthogonal directions (x, y). A method according to any one of claims 1 to 20, characterized in that the programmed coordinate transformations (PKT) and the instantaneous, actual coordinate transformations (MTKT) comprise translation information in two orthogonal directions as well as rotation information in a plane spanned by the two orthogonal directions (x, y). A system (1) for the layer-by-layer production of at least one object (2) on a build platform (6) by local solidification of powdered material (3) in a respective layer (7), comprising a build platform (6), N scanners (9a, 9b), with N>2, and a control device (10), configured to carry out a method according to one of the claims 1 to 22. Computer program product which, when applied on a system (1) for the layer-by-layer production of at least one object (2) on a build platform (6), performs a method according to one of claims 1 to 22 by local solidification of powdered material (3) in a respective layer (7).

Images