EP1828894A2

EP1828894A2 - Method and device for predicting the lifespan of a product that comprises several components

Info

Publication number: EP1828894A2
Application number: EP05813679A
Authority: EP
Inventors: Klaus VOIGTLÄNDER; Johannes Duerr; Rolf Becker; Reinhold Muench; Ivica Durdevic; Uwe Wostradowski; Christopher Hahn; Joerg Breibach; Philippe JÄCKLE; Hendrik Ehrhardt; Thomas Rupp; Antoine Chabaud
Original assignee: Robert Bosch GmbH
Current assignee: Robert Bosch GmbH
Priority date: 2004-12-16
Filing date: 2005-11-25
Publication date: 2007-09-05
Also published as: CN101443735A; US20090119029A1; WO2006063923A2; CN101443735B; JP2008524678A; US7831396B2; DE102004060528A1; JP4629735B2

Description

Method and device for predicting a

Life expectancy of a multi-component product

The present invention relates to a method and an apparatus for predicting a life expectancy of a product comprising at least two components. The life expectancy is determined as a function of a specifiable load of the product.

In addition, the invention relates to a computer program which is executable on a computing device, in particular on a microprocessor, a data processing system.

State of the art

In various technical areas, products are used or used depending on the type of product

Load that they are exposed to during their operation, have a certain lifespan. A product comprises several components. A product can be a personal computer (PC), the components then the electrical components (power supply, motherboard, hard disk, Floppy disk drive, CD-ROM drive, DVD drive, etc.) of the PC. It would also be conceivable to consider, for example, the motherboard as the product, in which case the various electrical components, solder joints etc. on the board are the components. A product could also be any motor vehicle component, in particular a motor vehicle control device, the components then being the various electrical components (for example resistors, capacitors, coils, operational amplifiers), solder joints, conductor tracks, etc.

The life of a product is highly dependent on the exposure to the product during its life. A motor vehicle control device, for example, which is arranged in the vicinity of an internal combustion engine of a vehicle, is exposed to a significantly higher thermal load than a control device arranged in the passenger compartment of the vehicle. The thermal load can be an absolute temperature but also a temperature fluctuation. The same control unit would therefore have a shorter service life in the engine area than in the passenger compartment. The same applies to mechanical and chemical loads as well as any other form of stress.

From the prior art it is known to provide a proof of reliability for a product by being exposed to a certain predeterminable absolute or cyclic load for a certain predefinable period of time or after predeterminable intervals. In order to provide a proof of reliability, it is thus determined whether a product or the components of the product can withstand a predeterminable load for a predefinable period of time. Such proofs of reliability are provided according to manufacturer or customer-specific standards or according to legal standards. In order to provide evidence of a product, ideally the product should be exposed to the load imposed on it within the scope of its intended use or use for a required minimum service life. If the product survives the minimum service life under these loads, the proof of reliability for this product is provided. If it fails before reaching the minimum life, the

Proof can not be provided. The testing for providing the proof can be carried out on a plurality of identical products, so that the result of the proof is more representative.

However, due to time constraints, it is usually not possible to run the entire required minimum life of the product (for example, 15 to 20 years for a vehicle control unit) in real time under the imminent use or use of its loads. For this reason, it is known to increase the burdens beyond the likely impending burdens and in turn to reduce the duration of the test. In this regard, one speaks of an accelerated proof of reliability.

However, the exact relationship between increasing the load and shortening the trial period is not known and may be completely different for different loads and for different products. In addition, this procedure leads to an accumulation of errors that are a consequence of the increased loads and do not occur during normal operation or use of the product in reality. Such mistakes are also referred to as field irrelevant errors. On the other hand, due to the shortened duration of the test, certain errors that occur in reality as a result of long-term, but in terms of lower amounts, during the shortened test duration may not occur. In other words, the more the trial period is shortened, the less realistic or close to the field is the result of the proof. Thus, there is a conflict of objectives between a shortening of the test duration on the one hand and the field proximity of the result of the testing on the other hand.

The known method provides after testing a plurality of identical products, for example, statements such as: "The product achieved at absolute

Operating temperatures above 9O ⁰ C with a probability of 97% a minimum life of 5,000 operating hours. "However, the known method can not provide such statements concern, for example, the question of how long the life expectancy of the product at a given load, how long the product lives at a lower or higher load and which components have to be dimensioned so that the product reaches a higher minimum service life (which components need to be made more stable?) or less expensive (which components can be made less stable) ?).

Based on this prior art, the present invention based on the object to design a method and an apparatus of the type mentioned in such a way and further that within the shortest possible time and as close to the field as possible a particularly meaningful proof of reliability for the product provided or the lifetime of the product can be determined.

To solve this problem, the following method steps are proposed starting from the method of the type mentioned: the components of the product are subjected to different loads; the components are operated at the different loads each to their failure; the downtimes achieved are stored for the respective component as a function of the load; depending on the load-dependent

Downtime of a component is added to an end-of-life (EoL) curve of the component; an EoL curve of the product is determined so that, at the various loads, it comprises that of the EoL curves of the components which, at the respective load, each have the shortest downtime; and the expected life of the product being determined as the functional value of the EoL curve of the product as a function of the given load on the product.

Advantages of the invention

Thus, according to the invention, each component of a product is first considered individually. Based on the EoL curves recorded for the individual components, an EoL curve is determined for the entire product. This corresponds to the different loads, so to speak, always the EoL curve of the worst component, ie the component with the lowest life at a given load. This This results from the consideration that a product in its entirety fails as soon as only one of the components fails.

The consideration of the individual components has the advantage that if the structure of the product changes (replacement of one component by another, omission of a component, addition of a new component) it does not have to be determined again for the entire product, if the product of the given load for the given

Duration can withstand. It is sufficient if the EoL curve for the new or modified component is included and taken into account in the determination of the EoL curve of the entire product. When omitting a component, simply determining the EoL curve for the product simply ignores the EoL curve of the omitted component.

By recording the EoL curve for one of the components of the product, the EoL curve of the component can simply be passed through or approximated to the calculated values for the load-dependent downtime for that component. The points of the EoL curve of a component between the determined values for the load-dependent downtimes of the component can be interpolated according to any method known per se. At the edge, d. H. at the beginning and end, the EoL curve can be extrapolated.

The method according to the invention enables particularly rapid checking of the durability of a product, in particular when the EoL curves for the individual components have already been recorded in advance of the check, stored in a database and, if necessary, only retrieved. In addition, with the inventive method field irrelevant errors due to excessive burden for the proof of reliability can be prevented. The various burdens that are placed on the individual components are largely in the range of an assumed one

Field load and only slightly above it. Checking the shelf life of the product, d. H. Whether the product can withstand the predetermined load for the given period of time, thus takes place with loads that can actually occur in the field during operation of the product.

With the present invention, so to speak, the loading of the components or of the entire product with the assumed field load was separated from the determination of the lifetime of the product and the examination of the durability. While the loading of the components with the assumed field load may take place in advance of the actual determination of the life and / or reliability of the product, the actual life of the product is determined based on the previously determined EoL curves during the running time of the method according to the invention.

The consideration of the individual components also has the advantage that, based on the EoL curves for the individual components, it can be checked whether the components are suitable for withstanding the assumed field load. If a component can not withstand the field load, it will be immediately recognized by the EoL curve of the component and the component can then be swapped for a more stable one. On the other hand, it is also possible components that are clearly much more stable than actually required (because the EoL curve of these components significantly above assumed field load) to replace with less stable less expensive components, without running the risk that the product of the assumed field load can no longer withstand.

Finally, with the present invention, the assumed EoL curve of the product can be used to determine an assumed field load (by magnitude of load and duration of load) which, from the point of view of minimizing the test duration and field irrelevant errors, allows optimal verification of product reliability. For this purpose, the assumed field load is chosen so that the duration of the stress has a representative value (this corresponds to approximately 50 to 3,000 cycles). The value for the field load is then selected as the load at which the selected duration of the load is just below the EoL curve of the product. Thus, the duration and the stress are set for optimal test conditions. For safety reasons, it is not necessary to carry out the test at a higher load or to carry it out for longer than the determined duration.

The method according to the invention is not only suitable for determining the life of the product (how long does the product withstand a given load?) But can also be used to provide proof of the reliability of the product (keeps the product at a predetermined load for a given period of time was standing ?) . Therefore, it is proposed according to an advantageous embodiment of the invention that the method is used to provide a proof of reliability for the product and it is checked whether the EoL curve of the product is above a predetermined load for a predetermined period of time. The proof of reliability for the Product is considered to be provided if the EoL curve of the product is above the load in the Statement of Reliability. In this case, at the given load, the failure of the component therefore occurs at a later point in time than the component has to withstand according to the specifications of the proof under this load.

According to a preferred embodiment of the invention, it is proposed that the EoL curves of the components be extrapolated towards smaller loads. Under certain circumstances, a very long service life of the component up to its failure can occur, in particular for small loads, ie the test duration can be very long. With the training, the test duration can be reduced and thus the detection accelerated by the failure time of the component is taken at higher loads and is extrapolated to smaller loads and longer test periods out. For extrapolation, there are a number of known methods. It is conceivable, for example, that the recorded downtimes at different loads put an approximation function, which is continued beyond the recorded downtimes to smaller loads and longer test periods. As an approximation function, for example, an arbitrary exponential function (e ^A (A + Bx); A (le ^A (-Bx)) + C; Ae ^A (B / x)) or an equalizing polynomial of the xth order may be considered.

According to another preferred embodiment of the invention, it is proposed that the EoL curves of the components be interpolated between the discrete values for the downtime at certain loads at which the EoL curves were taken. With the embodiment, the number of component tests to be performed (operation of the component with a certain load to failure of the component) by taking the component downtime only at a few discrete loads and interpolating the EoL curve therebetween. For interpolation, there are a plurality of methods known per se. It is conceivable, for example, that one places an approximation function through the recorded failure durations at different loads, which is, for example, an arbitrary exponential function, a spline function or an offset polynomial of the xth order.

The load of the component can be any type of load, in particular a mechanical, thermal, chemical, electrical, magnetic or electromagnetic load. Each kind of load has a life-changing, usually life-shortening, effect on the considered component. Advantageously, however, the load is a certain absolute operating temperature of the component and / or a temperature fluctuation of a certain size within a certain period of time.

In addition to mechanical loads, these are the most common and greatest burdens for products used in the automotive sector.

The downtime of the components of the product is absorbed at various discrete loads. Between the recorded downtimes, the EoL curve can be interpolated and extrapolated beyond the recorded downtime. One speaks also of different load classes, with which the downtime of the

Components are detected. The manufacturer of a product gets concrete specifications from his customer in which load class the product has to withstand for how long. For example, if the load is the operating temperature of the components, the downtime of the Component in different discrete temperature classes A first temperature class covers for example the temperature range from 100 ⁰ C to 12O ⁰ C, a second temperature class the range from 12O ⁰ C to 14O ⁰ C. According to specification of the customer the product must be 120 hours and in the first temperature class endure in the second temperature class for 20 hours. For example, if the tested product could run 150 hours to failure in the first temperature class and only 10 hours to failure in the second class, the product or component of the product would not meet the customer's specifications. At least the under-dimensioned component needs to be replaced by a more stable one. Even if the product in the second temperature class can be operated for up to 25 hours before failure, ie longer than specified by the customer, the total allowable load of the product will be exceeded, so that it will reach the customer's required level minimum life of for example 180 hours fails.

To determine this, it is proposed according to a preferred embodiment of the invention that for each load class, the distance of the duration of a predetermined load (for example, the assumed field load) of a component to the corresponding value on the EoL curve of the component that is the corresponding downtime the component is determined and added to Palmgren-Miner. According to the so-called Palmgren-Miner damage accumulation hypothesis, the quotients from the value of the duration of a given load on the component and the corresponding downtime are summed up for all load classes. The component only meets the requirements if the sum is less than 1. If the sum is greater than or equal to 1 must be with the premature failure of the component can be expected.

If the customer of a manufacturer of a product as a proof of reliability is given a certain load, which must withstand the product for a predetermined period of time (so-called burden in the proof of reliability) can be easily demonstrated by means of the method according to the invention that the product the proof of reliability can provide. First, the EoL curve of the product is determined. Then a damage accumulation according to Palmgren-Miner is carried out. The EoL curve of the product can even be used to determine a different (usually higher) load than the customer's intended load for a shorter period of time to still meet the required proof of reliability. In this way, the required time for providing the reliability proof can be reduced without disadvantages of the validity of the proof.

The finished product is exposed over its entire lifetime seen a certain load profile, that is, the product is loaded with different load classes each for a certain duration. In order to provide a proof of reliability, the actual load profile can not be traversed for reasons of time. For this reason, accelerated tests are used in which the load is increased and the duration of the test is reduced accordingly. In order to achieve as realistic a simulation as possible of a thermal stress of the product over its life cycle even with accelerated tests, it is proposed that at a certain temperature, which is above a predeterminable field load of the product, for a a certain period of time, which is below the duration of the field load, an accelerated proof of reliability is carried out, wherein the determined temperature and the determined time period are coordinated with one another such that a mechanical test

Load of the product and a thermomechanical load of the product can be accelerated by about the same factor.

Advantageously, an EoL curve of a component includes at least two points resulting from the downtime of the component at different loads. If, for example, the load is the operating temperature of the component, the downtime of the component is recorded at at least two different discrete temperatures or in at least two different temperature classes, for example at 100 ^° C. and 175 ^° C.

As a further solution of the object of the present invention, it is proposed starting from the device of the type mentioned at the outset that the device comprises:

- means for loading an end-of-life (EoL) curve for at least one of the components, the EoL curve having been previously recorded in dependence on determined load-dependent downtimes of the component;

Means for determining an EoL curve of the product so that, at different loads, it

Includes curves of the components which each have the shortest downtime for the corresponding load; and

- means for determining the expected life of the product as the functional value of the EoL curve of the product as a function of the specified load on the product. The EoL curves for the components of the product are included in advance of determining the life of the product. To record the EoL curves of the components, each component is subjected to different loads individually and each operated until their failure. It can also be said that the components are operated in different load classes until failure. This results in load-dependent downtimes for each component, which are the interpolation points of the EoL curve

Form component. These interpolation points are stored. To actually determine the lifetime of the product, the stored interpolation points of the EoL curves are simply called up for the components of the product, the EoL curves are determined by interpolation or extrapolation and used to determine the life expectancy of the product.

However, it is also conceivable that the interpolation points are interpolated or extrapolated in advance and the complete EoL curves for the components are stored. For the actual determination of the lifetime of the product, the complete EoL curves for the components of the product are then immediately available, without first having to wait between runtimes during runtime

Interpolated points or extrapolated beyond the bases.

According to an advantageous development of the present invention, it is proposed that the at least one EoL curve for a component has been recorded in advance as a function of determined load-dependent downtimes of the component by a manufacturer of the component. The EoL curve of a component can then be a manufacturer of the entire product, for example a datasheet or online. The manufacturer of the product can then use the EoL curves of the various components that he may receive from different manufacturers to determine the EoL curve of the entire product and verify that the overall product meets the requirements set by his customers.

It is further proposed that the device according to the invention serve to provide a proof of reliability for the product and means for checking whether the EoL curve of the product is above a predeterminable load for a predefinable period of time, and, if so, the proof of reliability for the product is considered to be supplied.

The method according to the invention can be realized in the form of a computer program that can run on a computing device, for example on a microprocessor or a microcontroller. The computer program runs on the computing device and performs the inventive method fully automatically. In this case, therefore, the invention is realized by the computer program, so that this computer program in the same way represents the invention as the method to whose execution the program is programmed. Strictly speaking, the method according to the invention is divided into two parts. A first part is responsible for recording the EoL curves for each component of the product. This can be done for example by the manufacturers of the components. A second part of the method according to the invention is responsible for determining the EoL curve for the entire product and for determining the life expectancy of the product. This can be done, for example, by the manufacturer of the product, following his calculations based on component manufacturers' EoL curves.

Assuming that, in the future, EoL curves are taken up by default for each manufactured component and made available to customers, the second part of the method is particularly important. Thus, in the future, it will be possible to provide reliability certificates for any products in a fast, simple and reliable way

Estimate life expectancy. Changes or adjustments to the product can be quickly and easily incorporated into the reliability and life expectancy calculations.

drawings

The invention will be explained in more detail with reference to the figures. Show it:

Figure 1 is a method known from the prior art for providing a proof of reliability for a product at a given assumed field load;

FIG. 2 shows a first embodiment of a method known from the prior art for providing a reliability check in the case of an assumed field load increased compared with the embodiment of FIG. 1;

3 shows a second embodiment of a method known from the prior art for providing a reliability check in a compared to the embodiment of Figure 1 increased assumed field load;

Figure 4 possible communication paths between suppliers and customers made possible by the present

Invention;

Figure 5 is an end-of-life curve for a component of a

product;

Figure 6 end-of-life curves for four components and for an unreliable product;

Figure 7 End-of-life curve for four components and for a reliable product;

FIG. 8 shows the end-of-life curve for a component of a product from FIG. 5 with a distance plotted in a certain stress class between a predetermined duration of the field load and the corresponding downtime of the component for producing a lifetime prediction of the Palmgren-Miner component;

Figure 9 ways to define failure criteria and to design components of a product;

FIG. 10 Realistic acceleration of the testing of a

Products with equal acceleration factors for a thermal and a thermomechanical

Burden; and

11 shows a device according to the invention for carrying out the method according to the invention. Description of the embodiments

The testing of products or certificates of reliability for products is carried out according to the state of the art using standards and standardized procedures proposed by customers or standards bodies. These standards describe accelerated tests that must be performed on a product for a specific service life t or a certain number of cycles N in order to demonstrate the reliability of the product. A test is said to be accelerated if it is performed under higher loads than the loads that occur in the field and only for a lesser time t or number of cycles N. The evaluation criterion for passing the reliability test is then usually proof of the functioning of the product after completion of the test. In this way, a yes / no statement can be made as to whether the product withstands the required load for the required period of time or not. However, a statement about the life of the product under any load can not be made in this way.

The tests are accelerated by the fact that the load acting on the products to be tested is selected to be higher, sometimes even significantly higher than an assumed field load for the product. However, this leads to more and more field irrelevant errors occurring as the load increases in the proof of reliability. This means that errors occur that do not occur during normal operation of the product in practice. In other words, as the burden of proof of reliability increases beyond the assumed field load, the validity of the test decreases the practice. The result is that the products or the components of the products are made much more stable and durable, in order to be able to provide the proof of reliability in any case even at the increased load, ie also the occurrence of field irrelevant defects of the product during the proof of reliability avoid. However, this is associated with additional weight, additional space and above all additional costs for the components.

On the other hand, a reduction of the burden in the proof of reliability, so that less field irrelevant errors occur during the test, would lead to a significant extension of the required test duration. In practice, therefore, attempts have always been made to find a suitable compromise between the shortest possible test duration, on the one hand, and the lowest possible number of field-irrelevant errors, on the other hand.

More recently, customers have been demanding ever more stringent reliability requirements due to the higher quality requirements, longer product lifetimes required, and higher loads imposed on the products during the practice. This leads to increased test costs and test times, as well as failures during the testing tests due to field irrelevant errors. This often means that products have to be repaired shortly before the series introduction, because the required proof of reliability can not be provided and the customer refuses to release the product. Due to the improvement conditional recursions (redesign of the entire product) take place, whereby the quality of the product is often inferior due to the then prevailing time pressure. In Figure 1, the current state of the art of a reliability is described by way of example. The customer specifies the reliability test or the conditions (load and duration of the load) under which the test is to be carried out. In FIG. 1, the assumed field load (AFB) is designated by the reference numeral 1 and the load specified in the reliability report (BZN) by the customer is designated by the reference numeral 2. The load is plotted on the x-axis in the form of an operating temperature T or an operating temperature deviation ΔT and on the y-axis the duration of the load as time t or number of cycles N.

The assumed field load 1 corresponds to an estimated or empirically determined load of the product during normal operation. The load 2 in the reliability check is an estimated load which is above the assumed field load 1. The load 2 in

Proof of Reliability is chosen so that if the product is still functional after test operation with Load 2 in the Reliability Statement for the indicated duration, it can be assumed that the product would survive the practical use with the assumed field load 1 for the desired period of time.

In the illustrated embodiment, the load is formed as an absolute operating temperature T or as a temperature deviation .DELTA.T. Of course, any other types of loads are conceivable. For the operating temperature T, the duration of the load corresponds to a time t. For the temperature deviation ΔT, the duration of the load corresponds to the Number of cycles N, which indicates how often a certain temperature deviation ΔT is passed through.

If, for example, the assumed field load 1 increases due to increased quality requirements, problems arise for the method illustrated in FIG. FIG. 2 shows the case in which the assumed field load 1 has been extended by an additional period of time 3 at different load values. In order to take account of this increased assumed field load 1, 3, the burden 2 in the reliability check is increased by an additional time period 4, i. The product must be operated for a longer period of time with the load 2 in the proof of reliability. However, this leads to an inappropriate prolongation of the test period, which is not practical in practice.

In the second exemplary embodiment shown in FIG. 3, the additional increased field load 1, 3 is taken into account by not extending the time duration of the reliability check, but rather by increasing the load T, ΔT. This is shown in Figure 3 on the basis of the new load 5 in the proof of reliability. However, it is disadvantageous that due to the load 5 in the reliability report far above the field load 1, 3 during the trial phase of the product increased field irrelevant errors can occur. The field irrelevant failure mechanisms can prevent the release of the product, although they do not occur in practical operation. By eliminating the field irrelevant errors, which usually means a revision or redesign of the product, there are increased product costs. While in the known method described in Figure 2, the test duration is extended (theoretically up to one year), are in the 3 generates a large number of field irrelevant errors due to extremely high loads,

In addition, the simple increase of the test times or the load does not lead to the desired correlation with the field loads 1, 3, since these can not be analyzed from the initial state. Thus, in the method known from the prior art, no life expectancy is possible. Loads can not be differentiated.

Therefore, a method for determining the life of a product is proposed according to the invention, the method steps are explained in more detail below with reference to FIG 5. An important aspect of the invention is the fact that several components of a product, preferably all components, are initially considered individually. The components of the product are subjected to a predeterminable load. It is not necessary to apply all the components of the product, but it is advisable to include at least those components of the product in the process according to the invention and consequently to apply a predeterminable load which has an effect on the life of the product. The components are operated at different loads (so-called load classes) until their failure. This results in load-dependent downtime for the various components.

The results of this measurement are shown in Figure 5 for a particular component. Those blocks (so-called load classes) of field loading 1 in which the component was operated until its failure and where a measurement of downtime was made, are shown hatched in Figure 5 and provided with the reference numeral 6. These are the load values HO ⁰ C, 12O ⁰ C, 14O ⁰ C, 16O ⁰ C and 18O ⁰ C, or 100 ° C, 125 ⁰ C, 15O ⁰ C and 175 ⁰ C or any other equidistant or non-equidistant Division in the interest

Temperature range (eg up to 175 ⁰ C, 200 ⁰ C or 233 ⁰ C). The measurement of the downtimes is carried out in each load class preferably at the corresponding highest load, these are the points 19 in Figure 5. Of course, the measurement could also be at an average load within the respective load class or at a low load within the load class. The measured values for the downtime of the component are shown as points 7. All measured and recorded in this way downtime 7 are on a so-called end-of-life (EoL) curve 8 of the component.

The values of the EoL curve 8 between the measured downtimes 7 are determined by means of an interpolation. Some of the interpolated values are designated by reference numeral 9 in FIG. The lower the loads, the longer the component would actually have to be operated until its failure. To avoid too long test times, the downtimes are only limited to a minimum load

Embodiment to 100 ⁰ K or 100 ⁰ C) detected. In order to obtain the values of the EoL curve 8 also below this load, the EoL curve 8 is extrapolated over the measured downtimes 7 in the direction of lower loads. The extrapolated values of the EoL curve 8 are designated by the reference numeral 10 in FIG.

The component is accelerated from near field to strong at different loads, ie at higher loads, which shortens the test duration, up to operated their failure. In this process, defect images are analyzed and combined so that a defect-specific EoL curve 8 is created for the component, which makes possible correlations to the field, ie for practical use of the component. Examples of defects are cracking and creeping in solders, diffusion in boundary layers (eg Kirkendal in bonding wires), delamination of a so-called MoId compound, bond fatigue, increasing a transient thermal resistance Z_th, leakage in electrolytic capacitors, etc.

The various fault patterns are expressed in different EoL curves 8 of the components, as shown in FIG. There, the various EoL curves are exemplified for four components Kl, K2, K3 and K4. Of course, the inventive method can be used for less than four components or for any number of components Kl, K2, K3 to Kn. It can be clearly seen in FIG. 6 that the EoL curve 8 for the fourth component K4 can not withstand a load which lies approximately in the middle load range over the required period of time. The component K4 is thus unsuitable for the planned use in the field. Before using the product in the field, the component K4 must be replaced by a more stable or durable one. In Figure 7, the end-of-life curves 8 are exemplified for four components Kl, K2, ... K4, wherein the component K4 in the course shown in Figure 7 over the course of Figure 6 by a more stable or durable Component K4 has been replaced. In this way, the reliable product of FIG. 6 becomes a reliable product (see FIG. 7) without much effort. An in-depth error analysis and a complete redesign of the product is - different as in the prior art - not required in the present invention.

At a glance, it can be seen in FIG. 6 that the component K4 does not fulfill the requirements placed on it, since the EoL curve 8 of the component K4 intersects the assumed field load 1 or the additional field load 3. Those components of the product which do not meet the requirements, for example the component K4 in Figure 6, are simply replaced by more stable and / or durable ones (see Figure 7), so that all EoL curves 8 for all components Kl to K4 (resp Kl to Kn) are above the assumed field load 1 plus the additional field load 3.

Based on the EoL curves 8 for the various components Kl to K4 (or Kl to Kn) an EoL curve 11 of the product is determined. The EoL curve 11 of the product always comprises the EoL curves 8 of those components K1, K2,... K4, which have the shortest downtimes in the various load classes. In the exemplary embodiment illustrated in FIG. 6, the EoL curve 8 for the fourth component K4 has the shortest downtimes in all load classes. For this reason, the EoL curve 11 of the product includes only the EoL curve 8 of the fourth

Component K4. Thus, the EoL curve 11 of the product also cuts the assumed field load 1 or the additional field load 3 and thus does not meet the reliability requirements placed on it.

In the embodiment shown in FIG. 7, the EoL curve 11 of the product at low load classes comprises the EoL curve 8 of the fourth component K4, for medium load classes the EoL curve 8 of the first component K1, and finally for high Load classes of the EoL curve 8 of the second component K2. The EoL curves 8 of all components Kl, K2,... K4 and the EoL curve 11 of the product all run at a distance from the assumed field load 1 or the additional field load 3, so that a first approximation can be assumed in that the components Kl, K2, ... K4 and the product meet the reliability requirements. The proof of reliability would therefore be provided for all components Kl, K2, ... K4 (or Kl, K2, ... Kn) and thus also for the entire product.

However, in order to have the final certainty, the components whose EoL curves 8 are above the assumed field load 1 or the additional field load 3 have actually met the requirements, damage accumulation according to Palmgren-Miner can be carried out. This will be explained in more detail below with reference to FIG. For each load class (eg 60-70 ⁰ C, 70-80 ⁰ C, ... 180-190 ° C), the distance 13 of the duration (time t or number of cycles N) N _{1 of} the assumed field load of a component to the corresponding value (the downtime) N _{1 determined} on the EoL curve 8 of the component and summed up according to Palmgren-Miner. The corresponding values on the EoL curve 8 of the component were either measured

(Recording downtime) (points 7), interpolated between measured values (points 9) or extrapolated beyond measured values (points 10). If the sum S ₁ of

Quotient is less than 1 (S ¹ is a reliable component. If the sum S _1, however

Danger, that the component Kl, K2, ... Kn fails prematurely before reaching its predetermined minimum service life. Therefore, the component must be replaced by a more stable or durable.

One advantage of the method according to the invention is that the failure criteria can be set product-specifically and adapted to the application. The failure criteria need not be known a priori, but may even be determined after the end of the determination of the downtime and the EoL curves 8, 11 for the components and the product.

For contact resistances R ₁ in the field of connection technology, this is shown by way of example in FIG. The assumed field load is again denoted by the reference numeral 2. For different sized contact resistances R ₁ different EoL curves 8.1 to 8.3 are shown. The various EoL curves 8.1 to 8.3 are located at a different safety distance from the assumed field load 1. Thus, if, for example, thought of, a control device (product), in which the contact resistance R ₁ (component), for in Figure 9 the different EoL curves 8.1 to 8.3 are shown to use in a warmer environment than originally planned, for example. closer to the internal combustion engine, the contact resistance R _{1 should} preferably be designed according to the criterion Krit3 with R ₃ = 20mΩ, so that the safety distance of EoL -Curve 8.3 to the assumed field load 1 is sufficiently large.

However, if it can be ensured that the component is used exclusively in the intended environment, so that the actual field load the assumed field load 1 is not significantly exceeded, the contact resistance R _{1 can} also be easily interpreted according to the criterion Kritl with Ri = 5mΩ. Thus, with the present invention, depending on the intended use or depending on changing conditions of use, in each case the optimal components are selected and the product is assembled according to the.

With the present invention, it is possible to convert customer-specific load profiles into a corresponding service life of the product, preventive life statements are possible.

With the aid of the present invention, not only can a product be tested for reliability, but also the optimal load 12 for the proof of reliability can be determined. What constitutes a major problem in the prior art, namely the determination of the load 12 in the reliability report (see Figure 2: relatively low load 2, 4, therefore too time-consuming; see Figure 3: relatively high load 5, therefore issue field irrelevant error) , According to the invention can be carried out without problems. The load 12 in the proof of reliability must be chosen so that it lies below the EoL curve 11 of the product. In addition, care should be taken that the test duration is chosen to be long enough (for example, between 50 and 3,000 hours or cycles N) that the test result is representative. With the burden 12 on the proof of reliability determined in this way, these products could be checked in the future, whereby it would be ensured that the check is as short as possible and the occurrence of field-irrelevant errors is prevented as far as possible. In FIG. 10, a specific load profile to which a finished product is exposed over its entire service life is designated by reference numeral 14. The stress profile 14 is either estimated or simulated or recorded under realistic conditions. The load profile 14 comprises in the simplified embodiment shown here relatively short periods of time .DELTA.ti, during which a relatively low load Ti is applied. In addition, the load profile 14 includes longer periods of time Δt2, during which a higher load T ₂ is applied. The load section 14 comprises a so with a certain frequency (1 / (Δti + At ₂₎₎ recurring load change ΔTi. The stress on the product with the load profile 14 leads to both a thermal load and to a thermomechanical load of the product, with thermal and thermomechanical load in a certain relationship to one another. The thermal stress occurs, for example, in the form of recrystallization or diffusion on the product. The thermomechanical load is mainly due to different coefficients of thermal expansion of the substances and components of the product (so-called TCE mismatch).

The reference numeral 15 is an accelerated

Designated loading profile, which includes in the simplified embodiment shown here, the relatively short periods of time .DELTA.ti, during which the relatively low load T ₁ is applied. In addition, the load profile 15 includes the longer periods of time Δt ₂ , during which a higher load T ₃ is applied, which is greater than the load T _{2 of} the first load profile 14. In this way, while the frequency of the load fluctuations is equal to the first load profile 14, but it results in a larger load stroke .DELTA.T ₂ than in the first Loading profile 14 (ΔT ₂ > ΔTi). The exposure profile 15 thus comprises a at a specific frequency (1 / (Δti + .DELTA.t ₂₎₎ recurring larger load change .DELTA.T. ₂ The accelerated load profile 15 means that the thermal load is over-emphasized compared to the thermomechanical load. The resulting acceleration factor of the thermal load or the degree of increase of the probability of failure due to the thermal load of the load profile 15 can be calculated according to the Arrhenius rule.

Reference number 16 in FIG. 10 designates a further accelerated load profile which, in the simplified exemplary embodiment illustrated here, comprises relatively short durations Δt ₃ during which the relatively low load Ti is present. The time duration Δt ₃ is shorter than the time duration Δti of the load profiles 14 and 15. In addition, the load profile 16 includes longer periods of time Δt ₄ during which the higher load T ₂ is applied. The time duration Δt ₄ is shorter than the time duration Δt ₂ of the load profiles 14 and 15. In this way, while the load fluctuations .DELTA.Ti with respect to the load profile 14 remain unchanged, but the test is accelerated in time, so that the load profile 16 one with a higher frequency (1 / (At ₃ + At ₄ )) includes periodically recurring loading stroke ΔTi. The accelerated load profile 16 results in that the thermomechanical load is over-emphasized compared to the thermal load. The resulting

Acceleration factor of the thermomechanical load or the degree of increase in the probability of failure due to the thermo-mechanical loading of the load profile 16 can be calculated according to the Coffin-Manson rule. In order to provide a proof of reliability, the actual load profile 14 can not be completely traversed for reasons of time. For this reason, accelerated tests (see load profiles 15 and 16) are used in which the load is increased and the duration of the test is correspondingly reduced. In order to achieve as realistic a simulation as possible of the actual stress of the product over its life cycle even in accelerated tests, it is proposed according to the invention that at a specific temperature T ₄ , which lies above a predeterminable field load T _{2 of} the product, for a certain period of time Δt ₆ , which is below the duration .DELTA.t _{2 of} the field load T ₂ , an accelerated proof of reliability is performed. In this case, the elevated temperature T ₄ and the shortened duration Δt _{6 are} coordinated with one another in such a way that a mechanical loading of the product and a thermomechanical loading of the product are accelerated by approximately the same factor. Of course, this method can not only be used for absolute loads (eg temperature T ₁ ) but also for load cycles or load strokes (eg temperature strokes AT ₁ ). In this case, at a certain temperature lift AT ₃ , which is greater than a predeterminable field load ATi of the product, with a specific frequency 1 / (At ₅ + At ₆ ) which is greater than the frequency 1 / (Δti + At ₂ ) the field load ATi is performed, an accelerated proof of reliability. The increased temperature lift AT ₃ and the greater frequency 1 / (At ₅ + At ₆ ) are coordinated in such a way that a mechanical load and a thermomechanical load of the product are accelerated by approximately the same factor. For an elevated temperature T ₄ or an increased temperature swing AT ₃ , the resulting acceleration factor for the thermal load is first determined according to Arrhenius. According to Coffin-Manson, the resulting reduced duration At ₅ , At ₆ or increased frequency 1 / (At ₅ + At ₆ ) for the test is then determined on the basis of the determined acceleration factor for the thermal load such that the acceleration factor for the thermomechanical load is approximately is equal to the acceleration factor for the thermal load.

With the method according to the invention, the test times can be shortened and the product costs can be reduced. In addition, the reliability can be increased. In addition, accurate predictions can be made about the life expectancy of a component or the entire product. The test conditions are chosen as close to the field as possible in order to be able to represent the correlation to the product in real terms. The generation of defect images that can not occur in the field is prevented as much as possible.

Carrying out near-field tests by operating components or components to failure at various temperatures T and temperature strokes AT, analyzing the failure mechanisms and correlating them with predetermined waveforms and functions is a significant advantage of the invention. The test criteria are chosen close to the field to ensure the correlation to the actual load in the field (in real use).

The termination criteria can be dynamically adapted to the respective site after the end of the test and need not be known a priori. This is a cost-effective, product-specific design of the products possible, at the same time increased reliability with sufficient life expectancy. So-called over- or under-engineering can be prevented. By avoiding recursions, the delivery quality of the products can be improved at the same time.

With increased field loading, an insufficient design of the products can be reliably avoided with the method according to the invention. The evidence of reliability may be due to scalability (previously performed measurements and

Setting up the test conditions) can be performed within a reasonably short time without having to carry out tests on the new product itself. The described, simplified tests on prototypes are sufficient.

Based on the present invention, sound communication between suppliers and customers can be performed. The communication sequences are shown by way of example in FIG. During the normal operation of a product, it is exposed to an actual field load (TFB), which is an objectively ascertainable quantity and is therefore the same from the perspective of the supplier and from the point of view of the customer. However, the actual field load TFB is not known before using a product, ie at the time of product development. Based on the expected actual field load TFB, the supplier determines an assumed field load AFBi, which may differ significantly from an assumed field load 1 determined by the customer. This is particularly due to the fact that suppliers and customers usually pursue different objectives. While the customer (for example, an automobile manufacturer) desires the highest possible quality and therefore of a relatively high assumed field strain AFB ₂ proceeds, at the supplier are in addition to quality requirements and low cost in the foreground, so that the supplier is usually of a lower assumed field loading AFBI will end. The reason for the assumed field loads AFB from the supplier and customer perspective is in particular the fact that during product development the actual field load TFB for the product in the field is still unknown.

On the basis of or taking into account the assumed field load AFBi of the supplier, the load 12 is determined in the proof of reliability BZN. With the aid of the method according to the invention, the determination of this load 12 can be optimized in the manner shown in FIG. 6 and described above, so that the optimum load 12 is obtained in the reliability statement BZN.

The customer determines from the assumed field load AFB ₂ a test load EB demanded by him, which the product must withstand from the perspective of the customer to secure a given quality standard. As a rule, the test load EB required by the customer is also greater than the burden on the proof of reliability BZN determined by the supplier. Previously, the supplier had no arguments as to why the load 12 determined by the supplier was completely sufficient for the BZN certificate of reliability so that the product could meet the requirements demanded by the customer

Quality requirements met; In addition, if the product complied with the test load given by the customer, the product would under certain circumstances be oversized. With the aid of the present invention, the supplier is now provided with a means by which he can present to the customer comprehensible and reconstructable arguments as to why the load 12 determined by the method of FIG. 6 is completely sufficient for providing the reliability certificate BZN ,

The determination of the end of life of components or components is done by measuring and observing observables (observable quantities). For example, these can be Delamination MoId Compound or Bonder Fatigue for an active device. Further examples are summarized in the following list. Of course, further observables are possible.

In the following some error pictures with their observables (observable quantities) are described. These are, for example, in active components (for example, an integrated circuit IC, an operational amplifier OP, a

Microprocessor, a microcontroller, a MOSFET, etc) the delamination of a so-called MoId compounds, which can be detected by means of ultrasound. In addition, it could be a bonder fatigue or an increase in a thermal resistance Z _th , an increase in the current I _SU bthreshoid or an increase in the internal resistance R _0n .

In the case of passive components (eg, an ohmic resistance, a coil or a capacitor, etc.), for example, the capacitance, the loss factor tan α or an electrical series resistance ESR is an observable variable; as well as the insulation resistance R _1SO at -40, +25, +125 and + 150 ° Celsius or other freely selectable temperatures. Also faulty images could be detected by optical control or by the drift evaluation of characteristics. In printed circuit boards, a resistance measurement, eg. The insulation resistance R ₁₈₀ ^ be performed. Likewise, as observable quantities, delamination, cracks in the insulating lacquer or cracks in the glass fiber structure can be recognized optically.

In the setup connection technique AVT (for example, in soldering), a resistance measurement, a shear force measurement or an X-ray inspection can be carried out to determine typical defect images. In addition, it is possible to abrade the connection point and to detect faulty images in this way.

The reliability tests must be geared to the types of loads occurring in the field where passive heating, active heating and impulse loads can occur. In the following, these types of load will be explained and an example will be given with which test procedures these loads can be reproduced.

Passive heating:

This is, for example, caused by temperature changes during driving in a motor vehicle. In a single chamber temperature change system (e.g., +5 K / min), these loads can be replicated.

Active warming:

This is caused by an average power loss, eg of a semiconductor, during the driving operation. The experimental simulation takes place by active heating of the component at 0.1 to 0.001 Hz (period 10 to 1,000 seconds) and 5 to 10 W power dissipation instead. Active heating is superimposed on passive heating.

Pulse loads:

These cause temperature changes with frequencies between 1 to 50 Hz (period of 20ms to 1 second) as a result of switching operations with short-term power losses up to several 100 W. The experimental simulation is done by a pulse operation of

Components. The pulsed operation can be superimposed on the passive and active heating.

To simplify and improve the reproducibility of the proof of reliability, the operation of the components during the detection is limited to periodic processes and fixed corner temperatures. Of course, also varying temperatures and temperature strokes can be interleaved. For this purpose, two near-field corner temperatures T _bottom and T _are selected _above . At the lower corner temperature _T.sub.n , the mean minimum temperature in winter is averaged over populous regions with high motor vehicle content, resulting in -15.degree. As upper corner temperatures T _above for the various EoL experiments, for example, 100, 125, 150, 175 and 200 ° Celsius are selected.

In contrast to many standard test methods, these experiments are performed with slow temperature changes, as they are mainly caused by passive heating. The holding times at the lower corner temperature T _nte n is limited to a minimum, provided there are no defects at low temperature. At the above corner temperatures T _above , the already known failure mechanisms (eg creep of solders, diffuse Mass transfer, etc.) in the design of the holding times are always considered. This can lead to different holding times being necessary for different error mechanisms. It is conceivable, for example, that the upper corner temperatures T abut the _top for 15 or even 60 minutes.

The following results can be obtained from the EoL curves 8 for the components:

only field-relevant mechanisms are recorded, a failure distribution, and the failure time (N 63%).

The evaluation of the EoL curves 8 of the components gives a:

Determination of activation energies, - determination of Coffin-Manson coefficients,

Determination of the correlation between near field

Testing and standard testing, and product-specific interpretations by variable

Failure criteria.

The present invention has the following advantages:

- shorter development time for the products, avoidance of recursions during the development of products, no over-engineering: the most favorable components, technologies and components can be selected,

a particular product can also be offered for higher technical usage profiles if its Components are sufficiently stable and durable, and it will be detected only field-relevant failure mechanisms.

While providing evidence of reliability for a given prior art product may take up to a year, with the present invention it is possible to make the same reliable and representative statements after just a few hours. The invention also makes it possible to build better customer relationships, as key issues and issues related to the life and design of components and products can be more quickly addressed. In addition, end-of-life testing can explicitly demonstrate the technology limitations of products and components. An acceleration of the testing is possible, whereby the acceleration factors (of field and laboratory) can be determined precisely. A lifetime estimate may also be made, with the life of the end-of-life curve 8 components directly given, and the lifetime of the entire product determined from the EoL curve 11 for the product.

The results obtained can be used in the development departments of the supplier or customer for:

- better utilization and assessment of the reliability of components and products;

- application-specific design of components and products; and

- optimized characterization of future developments. In addition, well-founded risk assessments in case of crisis and special approvals for specific products and components are possible. Certain lifetime models of the products can be built so that improved and more reliable lifetime models are available for simulation.

The present invention has been further described in the context of a high temperature application. But it is also applicable to all other evidence of reliability at "normal" temperatures, mechanical stress, exposure to moisture and / or chemicals.

In FIG. 11, a device according to the invention is designated by the reference numeral 20. In the device 20 EoL curves 8 are available for many different components from a database 21. The database 21 can be connected to the device 20 directly or via a client-server network, for example the Internet. The device 20 can access desired EoL curves 8 in the database 21 and load the desired EoL curves 8 in whole or in part.

At an input 22 of the device 20 is a predetermined load collective or load profile for a particular component or a specific product. The applied load profile is, for example, the assumed field load AFB 1 and optionally an additional load 3 in the sense of FIGS. 2 and 3. An output 23 of the device 20 is connected to an output signal that provides information about the reliability and / or service life of a particular load Component or the entire product. While the evidence of reliability provides only "yes" or "no" statements, ie the component or the product may or may not provide evidence, the lifespan will provide an accurate statement of the durability of the component or product. For example, the life provides a value in hours, or a cycle number, after which the component or product is likely to fail when subjected to a given load.

It is conceivable that the EoL curves 8 for the various components of the manufacturers and suppliers of

Components are added and delivered to the customer together with the product, for example as additional information on a data sheet or as retrievable via the Internet information. The customer can then use the device 20 according to the invention to assemble any products from the available components and to determine a corresponding EoL curve 11 for the components from these components from the EoL curves 8 of the components. With the help of the provided or self-determined assumed field load 1 (AFB) and the additional field load 3, by means of a predetermined load 2 in the proof of reliability or a previously determined optimal load 12 in the proof of reliability can be checked whether the product meets the requirements.

In the device 20, a storage element 24 is provided, which is designed for example as a flash memory. On the memory element 24, a computer program 25 is stored on a

Computing device 26 of the device 20 is executable. The computing device 26 is designed, for example, as a microcontroller or as a microprocessor. For executing the computer program 25 on the computing device 26, it is commanded over a data connection 27 or as a whole to the Calculator 26 transmitted. In the context of processing the computer program 25 determined data can be transmitted in the opposite direction via the data link 27 from the computing device 26 to the memory element 24 and stored there.

When the computer program 25 is executed on the computing device 26, it carries out the method according to the invention. In the illustrated embodiment, the computer program executes in particular the following method steps:

- Based on the EoL curves 8 from the database 21, an EoL curve 11 is determined for a particular product, so that the EoL curve 11 at the various loads T, .DELTA.T includes that of the loaded into the device EoL curves 8 of the components which has the shortest downtime t, N at the respective load T, ΔT, respectively; and the expected life of the product is determined as the functional value of the EoL curve 11 of the product as a function of the load 1, 3 of the product given at the inlet 22.

In addition, a proof of reliability for the product can be provided by checking that the EoL curve 11 of the product is above a load 2 set at input 22; 12 for a predetermined period of time t; N is, and, if so, the

Proof of reliability for the product is considered provided.

In advance of the execution of the computer program 25 on the computing device 26, the following method steps can be carried out for the determination of the EoL curves 8 for the components K1 through Kn: the components K1 to Kn of the product are subjected to a predeterminable load T, ΔT; the components Kl to Kn are operated at different loads T, ΔT, respectively, until their failure; - The achieved downtime t, N are stored for each component Kl to Kn as a function of the load T, .DELTA.T;

- Depending on the load-dependent downtime t, N of a component Kl to Kn an associated end-of-life (EoL) curve 8 of the component Kl to Kn is added and stored in the database 21.

The above-described interpolations between the interpolation points of the EoL curves 8 and extrapolations beyond the interpolation points are also possible.

Claims

claims

A method for predicting a life expectancy of a product comprising at least two components (Kl, K2, K3, ... Kn), the life expectancy being determined as a function of a predeterminable load (1, 3) of the product, characterized by following process steps: the components (Kl, K2, K3, ... Kn) of the product are subjected to different loads (T, ΔT); - The components (Kl, K2, K3, ... Kn) are operated at the various loads (T, ΔT) each to their failure;

- The achieved downtime (t, N) are stored for each component (Kl, K2, K3, ... Kn) as a function of the load (T, .DELTA.T); Depending on the load-dependent downtimes (t, N) of a component (Kl, K2, K3, ... Kn), an associated end-of-life (EoL) curve (8) of the component (K1, K2, K3, ... Kn) recorded; an EoL curve (11) of the product is determined, such that it comprises, at the various loads (T, ΔT), that of the EoL curves (8) of the components (Kl, K2, K3,... Kn), which has the shortest downtime (t, N) at the respective load (T, ΔT); and - it is the expected life of the product determined as the function value of the EoL curve (11) of the product as a function of the predetermined load (1, 3) of the product.

2. The method according to claim 1, characterized in that the method for providing a

Proof of compliance is provided for the product and a check is made to see if the EoL curve (11) of the product is above a predeterminable load (2; 5; 12) for a predefined period of time (t; N) and, if so, the proof of reliability is considered to be provided for the product.

3. The method according to claim 1 or 2, characterized in that the EoL curves (8) of the components (Kl, K2, K3, ... Kn) are extrapolated to smaller loads (T, .DELTA.T) out (10).

4. The method according to any one of claims 1 to 3, characterized in that the EoL curves (8) of the components (Kl, K2, K3, ... Kn) between the discrete values for the downtime (t, N) at certain Loads (T, ΔT) at which the EoL curves (8) were recorded (7) are interpolated (9).

5. The method according to any one of claims 1 to 4, characterized in that the load is a certain absolute operating temperature (T) of the component (Kl, K2, K3, ... Kn) and / or a temperature fluctuation (.DELTA.T) of a certain amplitude within a certain period of time.

6. The method according to claim 5, characterized in that for each load class, the distance of the duration of a predetermined load of a component to the corresponding value on the EoL curve of the component is determined and summed up to Palmgren-Miner.

7. The method according to claim 5 or 6, characterized in that within a thermal stress of the product at a certain temperature, which is above a predetermined field load of the product, for a certain period of time, which is below the duration of the field load, an accelerated proof of reliability is carried out, wherein the specific temperature and the predetermined time period are coordinated so that a mechanical load of the product and a thermomechanical load of the product are accelerated by about the same factor.

8. The method according to any one of the preceding claims, characterized in that an EoL curve of a component comprises at least two points resulting from the downtime of the component at different loads.

9. A device (20) for predicting a life expectancy of a product comprising at least two components (Kl, K2, K3, ... Kn), the device (20) the

Life expectancy depending on a predefinable load (1, 3) of the product determined, characterized by

Means for loading an end-of-life (EoL) curve (8) from at least two of the components (Kl, K2, K3, ... Kn), the EoL curves (8) being pre-determined as a function of determined load-dependent Downtime (t, N) of the components (Kl, K2, K3, ... Kn) have been recorded; - means for determining an EoL curve (11) of the product such that at different loads (T, ΔT) it comprises that of the EoL curves (8) of the components (Kl, K2, K3, ... Kn) which at the respective load (T, ΔT) has the shortest downtime (t, N) respectively; and

Means for determining the expected lifetime of the product as the functional value of the EoL curve (11) of the product as a function of the specified load (1, 3) of the product.

10. Device (20) according to claim 9, characterized in that the at least two EoL curves (8) for the components (Kl, K2, K3, ... Kn) in advance depending on determined load-dependent downtime (t, N ) of the components (Kl, K2, K3, ... Kn) have been recorded.

11. Device (20) according to claim 10, characterized in that the at least two EoL curves (8) for the components (Kl, K2, K3, ... Kn) by a manufacturer of the component (Kl, K2, K3, ... Kn) have been recorded.

12. Device (20) according to any one of claims 9 to 11, characterized in that the device (20) for

Providing a proof of reliability for the product and means for checking whether the EoL curve (11) of the product is above a predefinable load (2; 5; 12) for a predeterminable period of time (t; N), and if so proof of reliability is provided for the product.

13. Computer program (25), which is executable on a computing device (26), in particular on a microprocessor, a data processing system (20), characterized in that the computer program (25) for

Execution of a method according to one of claims 1 to 8 programmed when it runs on the computing device (26).