EP1992090A1

EP1992090A1 - Method for trimming an electronic system

Info

Publication number: EP1992090A1
Application number: EP07786425A
Authority: EP
Inventors: Reiner Franke
Original assignee: Atmel Duisburg GmbH
Current assignee: Microchip Technology Munich GmbH
Priority date: 2006-07-31
Filing date: 2007-07-30
Publication date: 2008-11-19
Also published as: WO2008014956A1

Abstract

The invention relates to a method for trimming an electronic system (200) in which n parameters (x, y) of the system (200) can be predetermined, which correspond to an n-dimensional trimming area, with two limit values (x0, XI y0, yl) being predetermined for each parameter (x, y) at the start of the trimming process, which limit values bound a corresponding output area (B_O) in the n-dimensional trimming area, and in which the following steps are repeated until a terminate condition is reached: - evaluation (100) of a target function which quantifies the reaching of a trimming target for the limit values (x0, xl, y?, yl), which bound the output area (B_O), with the evaluation comprising the measurement and/or evlauation of at least one physical variable of the system (200) which is dependent on the respective parameter (x, y) and/or its limit value (x0, xl, y0, yl), and with corresponding target function values associated with the limit values being obtained, - definition (110) of a changed, in particular reduced, output area (B_l, B_2) for a subsequent iteration as a function of the resultant target function values. According to the invention, the target function values are allocated to two different classes, with all the target function values which correspond to one target criterion being allocated to a first class, and with all the target function values which do not correspond to the target criterion being allocated to a second class, and with the changed output areas (B_l, B_2) being defined for the next iteration as a function of those target function values which are allocated to the first class.

Description

Method for matching an electronic system

The invention relates to a method for adjusting an electronic system, in which n many parameters of the system can be specified, which correspond to an n-dimensional adjustment space, wherein at the beginning of the adjustment two limit values are given per parameter, which define a corresponding output range in the n-dimensional adjustment space. limit dimensional adjustment space, and repeat the following steps until an abort condition is met:

Evaluating an objective function quantifying the achievement of an adjustment target for the limit values limiting the adjustment space, wherein the evaluation comprises the measurement and / or evaluation of at least one physical quantity of the system dependent on the respective parameter or its limit value, and wherein corresponding target function values associated with the limit values are obtained become,

Defining an altered, in particular reduced, output range for a subsequent iteration as a function of the resulting target function values.

Alignment methods of this type are known. For example, the adjustment methods using binary search principle provide for each iteration a halving of the output range for a subsequent iteration, for example, in a one-dimensional adjustment space with a parameter, assigning an output range for the subsequent iteration to the limit whose target function value is closer to the matching target.

A disadvantage of this method is the fact that erroneous target function values, such as are typically obtained by measuring a physical quantity of the system, can make erroneous decisions with regard to the assignment of the output range for the subsequent iteration, which leads to suboptimal matching. Particularly disadvantageously, for example, during the formation of the target function values, a measurement error may occur such that the actual target function value assigned to a limit value comes less close to the matching target to be achieved than a target function value which is assigned to another limit value. This is by the conventional methods made a wrong decision regarding the further evaluation of the adjustment space or in the definition of the output range, the u. U. prevented in the subsequent iterations, a further approximation to the matching target to be achieved.

Other known adjustment methods provide for the systematic evaluation of all possible parameter values of the entire adjustment space, whereby the fault tolerance is increased in principle. Already with a discretization of two parameters with 8 bits each, however, one is here

To investigate the adjustment space containing 2 ^Λ 16 = 65536 many adjustment points. Such a

Effort can u.a. in most systems to be adjusted u.a. due to the non-disappearing

Duration of time required to acquire or evaluate the physical quantities can not be tolerated.

Accordingly, it is an object of the present invention to improve a matching method of the type mentioned in that even in the presence of measurement errors, a reliable and rapid achievement of the matching target is given.

This object is achieved in a method of the type mentioned in the present invention, that the objective function values are assigned to two different classes, all

Target function values corresponding to a target criterion are assigned to a first class, and wherein all of the objective function values that do not correspond to the target criterion are assigned to a second class, and defining the modified output range for the subsequent class

Iteration depending on those objective function values assigned to the first class.

The classification according to the invention of the target function values advantageously makes it possible to consider a plurality of target function values instead of the single supposedly best objective function value which is used in the conventional methods alone for defining the new output range and thus to avoid or make incorrect decisions, for example, caused by measurement errors with respect to the newly defined output range To reduce the impact of this. By considering a plurality of target function values contained in the first class according to the invention, a more precise definition of the output range for the subsequent iteration is possible than in the case of the conventional methods. At the same time, despite the error tolerance, the adjustment method according to the invention advantageously leads to optimum adjustment values in a minimum number of adjustment steps. In a particularly advantageous development of the method according to the invention, it is provided that a threshold value for the target function values is used as the target criterion. In this way, the target function values associated with the first class can be determined by checking whether they exceed or fall short of the predefinable threshold value. By appropriate selection of the threshold value, the fault tolerance of the balancing method according to the invention can thus be set, which can also be done dynamically, ie during a calibration process. For example, during a first number of iterations, a first threshold may be set, and for a second number of subsequent iterations, a second threshold may be set, and so forth.

If, for example, the threshold value is chosen in such a way that at least two target function values are assigned to the first class, a measurement error in one of these two target function values in the context of the method according to the invention does not have the same effect on the further comparison, as with conventional methods. In the conventional method, such a measurement error would lead to the selection or definition of a completely incorrect new output range for the subsequent iteration, while the output range in the method according to the invention is also formed as a function of the other target function value not having such a measurement error, so that at least no completely wrong output range is obtained. Overall, the fault tolerance of the method according to the invention is achieved in that a measurement error would have to be greater than the threshold value in order to make a faulty decision for the changed output range.

In a further very advantageous embodiment of the method according to the invention, it is provided that the modified output range for the subsequent iteration is chosen to be closer to each limit associated with a target function value of the first class than at a threshold corresponding to a target function value of the second Class is assigned. This advantageously ensures that, for subsequent iterations, only those regions of the adjustment space which are in the vicinity of those limit values whose corresponding target function values lie sufficiently close to the matching target to be achieved, while such regions of the adjustment space are considered less or not at all, are examined which are within the range of limit values whose target function values are not sufficiently close to the matching target to be achieved.

In a further very advantageous embodiment of the method according to the invention provided that the exit area or an altered exit area have the same length in each dimension. In this way, a systematic examination of the adjustment space can be carried out particularly efficiently, with the adjustment space being subdivided into quadratic output ranges in the case of two parameters, for example.

It is also particularly advantageously provided that the length of the changed output region preferably in each of the n many dimensions corresponds to half the length of the output region of the preceding iteration. Alternatively, other educational regulations can be used for the output area to be used in a subsequent iteration, which for example do not provide for a halving of the respective length.

In a further very advantageous embodiment of the method according to the invention, it is provided that the termination condition, after which the inventive matching method is terminated, depends on a measurement accuracy in the measurement of the physical quantity and / or the number of iterations passed through and / or the size of the current output range , Since the step of evaluating in accordance with the invention for determining the target function values comprises the measurement and / or evaluation of at least one physical variable of the system, it is particularly advantageous to discontinue the calibration method according to the invention if the determined target function values are only offset by such an amount from one another. differ from the adjustment target, which is about the order of the achievable processing accuracy in the evaluation. A conventionally known measurement accuracy can advantageously also be taken into account in order not to carry out an unnecessarily large number of adjustment steps.

The balancing method according to the invention can advantageously be used in general in any electronic system which has one or more parameters to be adjusted and, for determining the achievement of a balancing target, comprises the measurement and / or evaluation of physical variables. Particularly advantageously, the balancing method according to the invention can be used, for example, for balancing an upward mixer with respect to an undesired residual carrier, wherein two parameters of the upward mixer can be specified which influence the application of two input signals of the up-converter, preferably an in-phase component and a quadrature component, with an offset , In this case, the remaining residual carrier of the up-mixer, which is measured in a manner known per se, is accordingly used as the objective function, as a result of which the described objective function values are obtained, which indicate how well the matching target is already is reached.

The balancing method according to the invention is not limited to the balancing of a system with two parameters. It is also possible to systematically and particularly advantageously also error-tolerantly search three-dimensional or multi-dimensional calibration spaces using the calibration method according to the invention, whereby optimal calibration values are always obtained in a minimum number of iterations.

As a further solution to the object of the present invention, there is provided an electronic system configured to carry out the method of the invention. Such a system may, for example, have its own control unit for carrying out the method. The functionality according to the invention can preferably also be implemented in an already existing control unit of an existing electronic system in which, for example, the conventional balancing methods are used.

Other features, applications and advantages of the invention will become apparent from the following description of exemplary embodiments ..der invention, which are illustrated in the figures of the drawing. All described or illustrated features, alone or in any combination form the subject matter of the invention, regardless of their summary in the claims or their dependency and regardless of their formulation or representation in the description or in the drawing.

In the drawing shows:

FIG. 1a shows an output region for carrying out a first iteration of the adjustment method according to the invention in a two-dimensional adjustment space,

FIG. 1 b shows an output range for a further iteration of the calibration method according to the invention,

FIG. 1c shows an output range for a third iteration of the calibration method according to the invention;

Figure 2 is a simplified flow diagram of an embodiment of the invention Adjustment method,

3a-e further possible output ranges during the performance of a calibration method according to the invention,

Figure 4 is a simplified block diagram of an electronic system according to the invention, and

Figure 5 is a simplified block diagram of a transceiver.

FIG. 1a schematically shows a detail of a two-dimensional parameter space spanned by the parameters x, y. The parameters x, y are quantities which influence the operation of the electronic system 200 shown schematically in FIG. 4, and which are to be adjusted using the balancing method according to the invention in order to ensure optimum operation of the electronic system 200.

The electronic system 200 is presently represented by an up-converter to which two input signals not shown in Fig. 4 are applied, and which processes the input signals in a manner known per se. In this case, due to asymmetries on the output side of the up-converter 200, an undesired residual carrier is produced, which can be influenced by means of the parameters x, y. The parameters x, y in the present example correspond to quantities which are used for offset adjustment of the input signals of the up-converter 200 and are accordingly added to the input signals in order to compensate for the described asymmetries.

For a certain combination of parameters x, y, a minimum residual carrier is established, which corresponds to the matching target to be achieved. Thus, the residual carrier to be determined in a known manner, or a variable representing the residual carrier, is dependent on the parameters x, y

Target function, by means of which the attainment of the matching goal of minimal residual carrier can be quantified.

At the beginning of the adjustment method according to the invention, two limit values x.sub.θ, x.sub.l, y.sub.θ, y.sub.l are predetermined per parameter x, y, which limit a corresponding output region BO in the two-dimensional adjustment space shown by a rectangle in FIG. The subsequent adjustment determines the value combination for the parameters x, y lying in the output region B_0 at which a minimum residual carrier is obtained at the output of the up-converter 200 (FIG. 4).

Since the parameters x, y for the application of the adjustment method according to the invention are preferably in digital form with a bit width of, for example, 3 bits in each case, the result according to FIG

Ia a range of 64 possible so-called to be reached for reaching the matching target.

Matching vectors, wherein a first component of the adjustment vector is represented by the parameter x and a second component of the adjustment vector by the parameter y. In the case of a finer discretization of the calibration space to be investigated, which is finer than that described here by way of example, or in the case of a larger output range B 0, correspondingly more is required

To examine parameter values.

In a first step 100 of the inventive method, which is illustrated by the flowchart shown in Figure 2, the limit values xθ, xl, yθ, yl limiting the output range BO are first examined with regard to the achievement of the matching target. Here xθ for each of the limit values, xl, yθ, yl or (for corresponding boundary points GO (X _0, yo), Gl X _0, yθ, G2, G3 evaluated the objective function described above, wherein the objective function values corresponding to received the respective limit values be assigned.

The evaluation of the objective function is carried out by appropriately setting the parameters x, y to the respective limit values and by measuring the output signal of the up-converter 200 (FIG. 4), whereby the corresponding target function value is finally obtained. In other words, for each limit point GO, G1, G2, G3 of the initially considered output range B_0, a target function value is present in step 100 which indicates a residual carrier remaining at the respective parameter values.

After this evaluation of the objective function for all four boundary points GO, G1, G2, G3, the objective function values obtained in this case are assigned to two different classes according to the invention. For this purpose, a target criterion is specified, which is preferably a threshold value for the

Target function values is. For example, all those objective function values become the first one

Class associated with the best objective function value, i. the objective function value that the

Matching target comes closest, not further than the predetermined threshold are removed. In this way, the first class is usually assigned multiple objective function values, all in a comparable manner to an approximation to the matching target, while those target function values that deviate by more than the threshold from the best target function value are assigned to the second class.

Subsequently, in a further step 110 of the method according to the invention, a modified output range is defined for a next iteration. According to the invention, this definition is advantageously carried out as a function of those target function values which are assigned to the first class and which accordingly are closer to the matching target to be achieved than those target function values which are contained in the second class.

In the present example, the evaluation in step 100, FIG. 2, has shown that the limit points GO, G1 corresponding to the parameter or limit values xθ, yθ and xθ, y1 are contained in the first class. Accordingly, the output region B_l for a subsequent iteration of the method according to the invention, cf. Figure Ib, advantageously chosen to be closer to the boundary points GO, Gl associated with the objective function values of the first class than to those boundary points G2, G3 included in the second class.

The new parameter values limiting the changed output range B_1 or the associated limit points are chosen as shown in FIG. 1b, cf. the filled circles in the corners of the exit area B l. For a better overview, the other points of the adjustment space contained in the output area B 1 are symbolized by means of unfilled circles, while the remaining points of the output area B O considered in the context of the preceding iteration are indicated by circles having dashed borders.

The classification of the target function values according to the invention and the consideration of all target function values contained in the first class or the limit values of the parameters x, y or the limit points assigned to them make the adjustment method according to the invention error-tolerant. For example, if a wrong target function value for the boundary point GO is obtained, for example, by a measurement error in the determination of the corresponding target function values during the step 100 of the evaluation, this would be the case in a conventional adjustment method, in which only the target function value closest to the matching target for determining the position of a following Output range is used, which cause erroneous formation of the modified adjustment range B_l. In contrast, the inventive method allows the consideration of several relatively "good" parameter or limit values or limit points, which are sufficiently close to the target to be achieved with their target function values, avoids the risk of a complete misallocation of the changed output range B_l due to a measurement error Output range B_l according to the invention arranged so that it is in the vicinity of the two promising border points GO, Gl of the first class, so that even with an erroneous assignment of a boundary point of the modified output range B l is not completely wrong placed.

In the subsequent step 120 of the method according to the invention, cf. 2, a check is made as to whether an abort condition has been reached for the inventive matching method, and if this is not the case, steps 100, 110 are repeated in a next iteration. In this case, for example, as shown in FIG. 1c, a new modified output region B_2 is obtained, which is located substantially in the vicinity of the boundary point G4, because in steps 100, 110 of the subsequent iteration only the boundary point G4 or a target function value assigned to it is the required one Has reached the target criterion and belongs to the first class.

If, however, it is determined in step 120 that the abort criterion is present, the adjustment method according to the invention is aborted and it can be assumed that the parameters x, y or the corresponding target function value found thereby are sufficiently close to the matching target to be achieved. That is, in the present case as an up-converting electronic system 200, see. 4, it can then be assumed that the best possible offset matching takes place using the found parameters x, y and the undesired residual carrier is correspondingly minimized. These parameters x, y are stored in step 130 (FIG. 2) of the method according to the invention in order to be available for future use.

The abort criterion for the query 120 can be selected, for example, as a function of a measurement accuracy with which the target function values can be determined. As soon as the target function values assigned to the different limit values or limit points to be examined during the evaluation 100 only differ by amounts that are of the order of magnitude of the measurement accuracy of the residual carrier or other physical quantities, another search in the adjustment space is not expediently and the already found parameter values are stored as optimal parameter values, step 130, and for the rest

Operation of the up-converter 200 used. The absolute number of continuous iterations during the calibration process according to the invention can likewise be used to form the termination condition, so that it is ensured that the method according to the invention does not exceed a predefinable maximum number of iteration steps. Also, the remaining size or number of discrete parameter values of an output range B_l can enter into the formation of the termination criterion.

Particularly advantageously, the output range B O, B_1, B_2 has the same length in each dimension, which leads to square output ranges B0, B1, B2 in the example described above with reference to FIGS. 1a to 1c. Such a design of the output regions makes a systematic examination of the adjustment space particularly easy.

It is also very advantageous to choose the length of the modified output region B l, B_2, preferably in each of the n many dimensions, to be equal to half the length of the output region B o of the preceding iteration.

Although the above example uses the two parameters x, y and accordingly has a two-dimensional adjustment space, the inventive matching method is not limited to two-dimensional adjustment spaces. It is also conceivable to examine matching spaces with three or more dimensions or correspondingly many parameters, as well as one-dimensional adjustment spaces. In any case, the erf Klassndungsgemäße classification of the limit values or limit points associated target function values a fault-tolerant search for optimal matching parameters given, which - comparable to the principle of binary search - leads in a minimum number of iterations to optimal matching values.

FIGS. 3 a to 3 e again symbolize different cases, again based on a two-dimensional adjustment space, which may occur in the evaluation of the objective function or the classification of corresponding target function values and in the formation of modified output ranges B_1. In this case, the dashed square BO represents in each case an output region for the first iteration of the method according to the invention, and a square B 1 executed by a solid line represents the modified output region for the subsequent iteration of the calibration method according to the invention, which depends on corresponding limit values or limit points according to the invention has been determined.

In the figure 3a all four cases are shown, in which only one of the four Output range B_0 limiting limit points or the corresponding target function value of the first class has been assigned. In the present case, the relevant boundary point is always symbolized by a circle filled in black, while the further parameter values corresponding to points of the output area B_0 are symbolized by open circles. Accordingly, according to the invention, the changed output range B_l is assigned as close as possible to the respective limit point.

In Figure 3b, those cases are shown in which the classification of the objective function values has revealed that there are exactly two objective function values in the first class corresponding to adjacent limit points, i. lie on the same side of the exit area B O. Accordingly, according to the invention, the modified output regions B_1 are each arranged such that they lie closer to these limit points.

For the two cases depicted in FIG. 3c, respective boundary points or their target function values in the first class are contained within the output area B_0, from which a central arrangement of the changed output area B_l within the output area B_0 can be seen, as shown in FIG.

In the cases depicted in FIG. 3d, in each case three out of four boundary points bounding the output area B_0 or their target function values are contained in the first class, so that the changed output area B_1 is arranged in the vicinity of these three boundary points.

In the situation symbolized in FIG. 3e, all four target function values of the boundary points delimiting the output region B O are contained in the first class, and the arrangement of the modified output region B_l is accordingly centered with respect to the output region B O of the preceding iteration, cf. also Figure 3c.

The definition of the changed output range B 1 for a subsequent iteration is advantageously carried out in accordance with the above-described principle that the changed output range B_I is always preferably arranged in closer proximity to limit points whose target function values have been assigned to the first class. However, the exact selection of the limit values or limit points for the changed output range B_1 or its shape and / or size can be modified almost arbitrarily and adapted to the respective given conditions. output ranges However, B_0, B_l, B_2 of equal length with respect to each dimension and a halving of this length from iteration to iteration are preferably used, in particular due to the favorable implementation in a computing or control unit 300 (FIG. 5) carrying out the inventive matching method.

Due to the fault tolerance of the method according to the invention, with the same minimum number of iterations as in the principle of the binary search, the measurement or evaluation of the physical quantity, such as, e.g. 4 of the residual carrier of the upward mixer 200 shown in FIG. 4, with a lower accuracy, so that a corresponding expense can be reduced and the duration of the balancing process can be shortened without having to forego the optimum balancing values.

5 shows a simplified block diagram of a transmitting / receiving device for a data transmission system according to IEEE 802.16 (WiMax) with an up-converter 200, which is adjusted by the method according to the invention.

The transceiver 400 comprises a baseband unit (BB) 410, addition nodes 420, 421, an up-converter 200, an oscillator 430, a quadrature generator 431, a subtraction node 440, a power amplifier (PA) 450, an antenna 460 and a control unit (CTRL ) 300 on.

The baseband unit (BB) 410 provides a complex-valued transmission signal with an in-phase component 10 and a quadrature component QO, which is to be transmitted as far as possible without distortion. The addition nodes 420 and 421 add to the respective signal 10 and QO respectively the respective parameters x and y provided by the offset compensation control unit 300 and thus form the input signals Il and Ql of the up-converter 200.

The oscillator 430 provides a local oscillator signal from which the quadrature generator 431 derives an in-phase component (0) and a quadrature component (90) of the local oscillator signal.

The up-converter 200 mixes the input signals II, Ql with the in-phase component (0) and the quadrature component (90) of the local oscillator signal, the output signal being formed by subtracting the two resulting signals in the subtraction node 440. The transmission signal thus formed is finally amplified by the power amplifier (PA) 450 and radiated via the antenna 460.

The control unit 300 (and thus the transmitting / receiving device 400 or the electronic system) is designed to carry out the method according to the invention. For this purpose, it evaluates the output signal of the subtraction node 440, deduces therefrom parameter values x, y and supplies the input signals of the up-converter 200 with these parameters until the up-converter 200 is adjusted in the context of the above-described iterative method.

In further embodiments, a rotational extension of the input signals of the up-converter with two parameters x, y can be made.

Of course, the invention can also be advantageously used in transmitting / receiving devices which are specified according to other standards for data transmission.

Claims

claims

1. A method for matching an electronic system (200), in which n many parameters (x, y) of the system (200) can be specified which correspond to an n-dimensional adjustment space, wherein at the beginning of the adjustment per parameter (x, y ) two limit values (x θ, x l, y θ, y l) are defined, which limit a corresponding output range (B_0) in the n-dimensional adjustment space, and in which the subsequent steps are repeated until an abort condition is reached:

Evaluating (100) an objective function quantifying the achievement of a matching target for the limiting values (xθ, xl, yθ, yl) limiting the output range (BO), wherein the evaluating comprises the measurement and / or evaluation of at least one of the respective parameter (x, y ) or whose limit value (xθ, xl, yθ, yl) is the dependent physical quantity of the system (200), and where corresponding target function values associated with the limit values are obtained,

Defining (110) an altered, in particular reduced, output range (B l, B_2) for a subsequent iteration as a function of the obtained target function values,

characterized in that the objective function values are assigned to two different classes, all objective function values corresponding to a target criterion being assigned to a first class, and all objective function values not corresponding to the target criterion being assigned to a second class and defining the changed one Output range (B_l, B_2) is carried out for the subsequent iteration in dependence of those target function values that are assigned to the first class.

2. The method according to claim 1, characterized in that a threshold value for the target function values is used as the target criterion.

3. The method according to any one of the preceding claims, characterized in that the changed output range (B_l, B_2) is selected for the subsequent iteration, that it is closer to each limit associated with a target function value of the first class, as at a limit which is associated with a goal function value of the second class.

4. The method according to any one of the preceding claims, characterized in that the

Output range (B O) or a modified output range (B_l, B_2) in each dimension has the same length.

5. The method according to any one of the preceding claims, characterized in that the length of the modified output area (B l, B_2) preferably in each of the n many dimensions of half the length of the output area (B_0) corresponds to the previous iteration.

6. The method according to any one of the preceding claims, characterized in that the termination condition depends on a measurement accuracy in the measurement of the physical size and / or the number of iterations passed and / or the size of the current output range (B_0, B_l, B_2).

7. The method according to any one of the preceding claims, characterized in that the system (200) is designed as an up-mixer, and that two parameters (x, y) can be specified, wherein the parameters (x, y), the application of two input signals of the up-mixer , preferably an in-phase component and a quadrature component, with an offset.

8. The method according to any one of the preceding claims, characterized in that at least two (n = 2) parameters (x, y) can be specified, which correspond to a two-dimensional adjustment space.

An electronic system configured to carry out the method of any one of the preceding claims.

10. An electronic system according to claim 9, characterized in that a control unit (300) is provided, which is designed for carrying out the method according to one of claims 1 to 8.

11. An electronic system according to claim 10, characterized in that an up-converter (200) is provided and the control unit (300) is adapted to specify two parameters (x, y), the admission of two input signals of the up-mixer (200), preferably one In-phase component and a quadrature component, with an offset influence.