TWI227329B - Test language conversion method - Google Patents

Abstract

A method of converting test vectors in an original cycle based test language into a target cycle based test language. The method includes the steps of forming a set of templates depicting waveforms defined in the target test language; decomposing a waveform in the original test language into a set of constituent events where each event includes data showing at least a starting value and a number of subsequent edges of the waveform; comparing the template and the set of events; storing the waveform data in the target test language when a match is detected and retrieving corresponding parameters of the waveform in the original test language; and repeating the above steps for all of the test vectors in the original test language, thereby forming a test vector file of the target test language.

Description

1227329 A7

The Intellectual Property Bureau of the Ministry of Economic Affairs' employee consumer cooperative prints and prints the hairpin in the domain J. The present invention relates to a method for converting test data for testing semiconductor devices by using automatic test equipment, and more specifically, it is related to a method of writing as The STIL (Standard Test Interface Language) digital test vector is converted into a digital test vector that is unique to the test language of the automatic test equipment. [Background of the Invention] When using automatic test equipment (ATE) or a 1C tester to test semiconductor devices such as ICs and LSIs, a semiconductor IC device to be tested is attached to its appropriate pin and provided by the IC tester to Test signals or test patterns at predetermined test timings. The IC tester receives an outgoing signal from the IC device under test in response to the test signal. The output signal is selected or sampled by the strobe signal generated by the I c tester at a predetermined timing, and compared to the expected data to determine whether the 1C device is correct; 1¾ does not. The meaning of the test signal and the strobe signal is typically defined relative to the start timing of each test cycle of the 1C tester. As noted above, a 1C tester generates test patterns and strobes, that is, test vectors specific to the tester based on digital test vectors described in a language (format). The language used for these automated test equipment varies from manufacturer to manufacturer. Recently, the IEEE has proposed a test language STIL (Standard Test Interface Language K) as a standard test interface language (IEEE Standard 1450-1999). STIL provides an interface between, for example, a computer-aided engineering (CAE) for a logic test simulator and automated test equipment. In a CAE environment or EDA (Electronic Design Automation) environment, a semiconductor device is designed with the help of a computer system (please read the note on the back? Write this page first)

This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) -4- 1227329

V. Description of the invention (2), and this design is tested by a logic test simulator or test bench. Compared with (h), it uses / j < ί C test ρ type to measure the type of semiconductors Digital test vectors for logical simulation of the device. STIL is designed to complete the transfer of a large number of digital test vectors from the CAE environment to the automatic test environment (ATE) environment. The STIL test language has recently become the standard. However, most ΛTE systems are now STIL is not used as an inherent language. The idiom “g” provided by the test equipment manufacturer is not compatible with each other. Therefore, it is necessary to effectively convert STIL to an ATE native language. (Please first Read the note on the back 1 (I install --- this page) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs [Inventory Summary] Therefore, the purpose of the present invention is to provide a digital test vector with a loop as the main test language. It is efficient and highly accurate to convert to another loop-based test vector. Another object of the present invention is to provide a test language conversion method for efficient and accurate conversion to STIL (standard (Test interface language) format test vector becomes a target. In the present invention, a method for converting a test vector in a test language-based original loop into a target loop-based test language includes the following steps: reading the definition in the target Test the available waveforms in the language and form a set of templates describing the waveforms, each of which is equivalent to the waveform of the target test language, and contains data that shows the beginning of a waveform and some subsequent ones in the waveform Edges; read the test vectors of the original test language and decompose the waveforms in the test vectors in the original test language into a set of constituent events, where each event contains data showing at least one beginning and subsequent edges of the waveform; compare The paper size of the target test paper is in accordance with the Chinese National Standard (CNS) A4 specification (210 X 297 mm). · 1227329 Printed by the Intellectual Property Bureau of the Ministry of Economic Affairs and Industrial Cooperative Cooperative A7 Β7 V. Description of the invention (3) Test ^ ri ' The waveform-derived template and the event coin 11 derived from the original test language were detected in the fei shi step-one At the time, the waveform data stored in the target measurement δ formula is retrieved and the relevant parameters of the waveform in the test vector in the original test language are retrieved, and the parameters are stored with the matched waveform data; and for all the test vectors in the original test language Repeat the above steps to form a test vector file of the target test language. Typically, the original test language is a STIL (Standard Test Interface Language) specified by IEEE Std 1 450 1 999. Preferably, compare the template with the event group The steps include the steps of changing different levels of the test vector to become a signal level, a wave level, where the signal is structured by most wave types, and a word level, where the wave type is structured by multiple digital elements . The event groups derived from the original test language are stored in a tabular format with rows that are designated to display the number of subsequent edges and information about the beginning. The table of stored event groups is simplified by reading the beginning of a specific event based on the end state generated by the previous event and optimizing it, thereby simplifying the data in the table. [: Brief description of the diagram] · Figure 1 is a chart showing an example of the STIL format, which describes the signal or signal group in the digital test vector, and the test vectors constitute the desired test vector. Figure 2 is -·-The graph shows an example of the STIL format, which describes the edge timing in each of the signals that make up the waveforms. Figure 3 is a chart showing an example of the STIL format, which is described in each paper size as applicable to the Chinese National Standard (CNS) A4 (210 X 297 mm) -6-n- n ϋ n 1% 1§ ni ·· _-1 1 nnnnn emm§ 一 ″ 口, 1_1 > 1 · «n _1 ϋ 1 1 (Please read the precautions on the back to write this page)

1227329

V. Description of the invention (4) Vector patterns in the signals constituting the waveform. Fig. 4 is a diagram showing an example of the STIL format, which describes the flow of the pattern in the test vector constituting the desired waveform. FIG. 5 is a chart showing an example of a TDL (Test Target Language) format, which is a schematic diagram of an inherent test phrase table for the ATE system developed by the assignee of the case, showing a language according to the present invention. The basic principles of conversion, and FIG. 6B is a schematic diagram showing an example of the functional architecture in the conversion of the test language of the present invention. Figure 7 is a diagram showing an example STIL architecture and an equivalent TDL representative, and a waveform template based on the TDL representative. Fig. 8 is a diagram showing other examples of the STIL architecture, corresponding to the waveform of the STIL structure, and the TDL representation. Fig. 9 is a-·-diagram showing another example of the STIL architecture, a waveform corresponding to the STIL structure, and a TDL representation to explain the best waveform. Figures 10A and 10B are tables showing examples of arrays, which describe the number of edges and the beginning of the decomposition events that are represented in the STIL test vector for pattern matching in the present invention. FIG. 11 is a flowchart showing a procedure in which the wave types of the present invention are matched with different bite test vectors. Fig. 12 is a flowchart showing the procedure from the previous cycle to the next cycle in the wave type matching of the present invention. Figure 13 is a chart showing the basic idea of converting the STIL test vector to the TDL multi-clock test. When the pin type of the component under test is applicable to this paper, the Chinese National Standard (CNS) A4 specification (210 X 297 mm) -------------- Installation (please read the notes on the back to write this page) ---- Order --------- β > Ministry of Economy Wisdom Printed by the Consumer Cooperative of the Property Bureau 1227329 A7 Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economy --- R7______ V. Description of the invention (5) When there are multiple clock signals. Figure 14 is a ▲ waveform chart showing the basic idea of converting STIL test vectors to Ί D L Z pin multiplex test vectors. [Detailed description of the preferred embodiment] The test language of S Π L (standard test interface language) uses a loop as the main representative to define the pattern and timing information for device testing. The format used does not represent the language of any ATE system, but has been constructed to provide maximum flexibility in designing test programs. STIL has only recently been adopted as a standard, but it is not widely used. Several STIL-based applications have been proposed in recent years, including general test procedures, scanning test methods, and ATPG (Automatic Test Pattern Generator). Currently, most ATE systems do not use STIL as an inherent language. Therefore, the memory needs to be converted from STIL into the native language of the target ATE system. The term ri system is generally cycle-based, so the conversion system _ one cycle is the main representative to the other cycle is the main representative. In this specification, the inventor reveals some basic steps involved in the method of converting from one cycle to the other format. The inventor also presents some methods to allow STIL conversion in an effective manner. The analysis of Mu Liming 屮 / B conversion system is based on the analysis of S TIL data. As a sample 13 markup language, the test description language (TDL) developed by the Japanese assignee Advantest in Tokyo is used for illustration purposes, but the principles described here can be equivalently applied to other test languages. . Brothers 1 to 4 do not need to describe examples of the stil format of the digital test vector. Figure 1 shows the format of signals and signal groups. Figure 2 shows the Chinese National Standard (CNS) A4 specification (210 X 297 mm) applicable to each paper size -------- (Please read the precautions on the back AH! Write this page) 丨 丨Fill ml *

1_§ n aw am WB I I I

1227329 A7 B7 V. Description of the invention (6) The timing format of the edges in the signal. Figure 3 shows an example of the s TIL format, which describes the pattern of the test vector, and Figure 4 shows an example of the STIL format, which describes the flow of the pattern. Figure 5 is a diagram showing an example of the TDL format, which is the target test language. [Boilerboard matching] The easiest way to switch from-. Loop to main representative 値 to another representative is to use event-based intermediate format. Here, the event is any change, such as the edge in the test vector or no change from the timing definition. These two cycles are mainly used to illustrate that the general system is not very different. Regarding the conversion of one cycle to the main description and another cycle to the main description, an insertion event is not used to decompose the data into basic building blocks. That is, the input loop main format (STIL) system is decomposed into constituent events, and these events are reconstructed from the target description format (TDL). The basic processing system is illustrated in FIG. 6A, which performs vector main to vector main conversion processing. In the present invention, the test vector in the STIL format is decomposed into each event, and these are compared with a template generated based on a waveform defined in a TDL, which is an example of a target test language. This idea is shown by the dotted arrow in Figure 6A. When a matching template is found, the matched template is listed in the file, and the parameters in the relevant STIL test vectors are transferred to complete the waveform. In this way, the TDL test vector is created by the dashed arrows in Figure 6A. FIG. 6B is a functional representative diagram of the test vector conversion of the present invention, which is converted from a STIL test language to a TDL test language. A STIL vector file 2 1 This paper size is applicable to China National Standard (CNS) A4 (210 X 297 mm) -9---------------- (Please read the back (Note UI written on this page)

ϋ I 1 I 1: mouth, * n n n n H ϋ ϋ I

Printed by the Intellectual Property Bureau of the Ministry of Economic Affairs of the Consumer Consumption Cooperative 1227329 A7 B7 Printed by the Intellectual Property Bureau of the Ministry of Economic Affairs of the Consumer Cooperative of Consumers . Typically, the STIL test vector in the STIL file 21 is derived from the design stage of the semiconductor device, that is, the CAE (Computer Aided Design Environment) or EDA (Electronic Design Automation) environment, and becomes the result of logic simulation. The STIL vector is decomposed into constituent events and stored in a decomposition event file 24. As noted above, TDL is a test language developed by Advantest, the assignee of this case, to create a Logical Test Pattern (LPAT) file. The TDL format is stored in a TDL wave file 22. Each wave germline defined in the target test language TDL is converted into a related template with a set of components. These wave seed templates are stored in the template file 25. A pattern is executed by a comparator 26 to compare the event from the event file 24 with the template from the template file 25. When matching, the waveforms related to TDL are listed in file 28, which stores the template matching data representative 値. Furthermore, the details of the parameters used to match the template vector are transferred from the STIL vector to the T D L vector. The details of these vectors include timing, pattern characters, and edge types. Therefore, in the TDL and LPAT file 29, it is equivalent to the STIL test direction: the TDL test vector of Europe is created through the conversion processing described above. For example, referring to Figures 7 and 8, considering the following STIL structure, an equivalent TDL represents chirp and related waveforms 01 {'400ns'D / U;} 2 >NRZ; Tl 400ns; T2 = 400ns, where the STIL structure is 0 1 {40 () ns' D / U;} on the left indicates that the character '0' indicates the falling edge (D) at 400ns and the character '丨' indicates the rising edge (U) at 400ns. This STIL represents the waveform equivalent to the ..t part shown in Figure 8 and the non-return-to-zero (NRZ) wave defined in TDL ------------- tr! Read the notes on the back 1¾ Write this page) «— — — — — — I.

This paper size applies to China National Standard (CNS) A4 (210 X 297 mm)-10- Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 1227329 A7 B7 V. Description of the invention (8). In Fig. 8, the waveform of the character '0' shows a shaded portion from the start edge of the test cycle to the falling edge T1 (from the start edge of 400 ns). The shaded area indicates that the logical state is undefined in this area. Therefore, character 0 defines a falling edge T 1 of 400 ns from the start of the test cycle, regardless of the current logic state. Similarly, the waveform of character 1 is defined by a falling edge T 1 of 400 n s at the beginning of the test cycle, regardless of the current logical state. It should be noted that this STIL construction example is equivalent to the NRZ waveform of TDL. Therefore, in the conversion method of the present invention, the structure of the TDL waveform from the decomposed STIL waveform description is performed using a template matching method. The waveforms that can be used for a given tester are read in execution time and stored in a way that makes template matching easy to perform. Therefore, in the first step, the template column is established in each waveform for the NRZ defined, for example, in TDL. The template system is characterized by the {pattern characters, starting 値, subsequent margins} pairs describing the waveform. The representative of the NRZ waveform in the example described in j is shown in the upper part of Figure 7. For example, in the template in Figure 7, the first element in each row is a pattern character, and the next two elements represent the beginning of the subsequent edge 値 number} pairs, and the other inputs represent the names for specific edges. Using the above example, we can use the NRZ waveform with the following 加以 to represent: 01 = > {0,0,0}, {0,1,1 400ns D}, U, 〇, l 400ns U}, {1,1, 0} where 丨 pattern characters are displayed, starting 値, and the number of subsequent edges 丨. In this example TDL, all the waveforms defined in the target test language are stored in this way, allowing the capacity of the STIL waveform to be simply compared with a given waveform. Therefore, the template library is prepared in the template file 25 in Figure 6B. In comparison, the STIL waveform and TDL template system are compared with each other. This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) -11-H ϋ ϋ n I ϋ n ϋ ϋ n * n I · ϋ n · n · ϋ n ϋ n ϋ ϋ · ϋ ϋ nn (Please read the precautions on the back of Jmu page first)

Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 1227329 A7 B7 V. Description of the Invention (9) The S TIL waveform is decomposed into each event, which is indicated by the combination of the beginning and the number of edges. The data describing the onset of each event and the number of edges in the STIL waveform are stored in the event file 24 in Figure 6B. Therefore, the comparison between the decomposition event and the template is completed by comparing the start and the number of edges to complete. This is done as a query on the waveform. Basically, you can start to support these many edges? It should be noted that the STIL waveform map to template needs to match all the resulting triplets, which contain the waveform. In the above example, the character '01' from STIL results in four triples shown in the template of Fig. 7. All four must be mapped so that these characters are fully represented. For example, the NRZ system can support all the required triplets. For another example of basic template matching, refer to Figure 8 for the STIL characters below. TDL stands for chirp and waveform. 〇 {} 1 {^ OOns1 U; ^ OOns1 D;} two> RZ; T 1 = 2 0 〇ns; 12 two 4 0 0 ns; This is equivalent to the return-to-zero (RZ) waveform in TDL, and its waveform is shown At the center of Figure 8. Therefore, the template of the RZ waveform of TDL is set in the template file 25 in the figure. The template matching example shown in the previous paragraph is simple. We can directly match the standard waveforms N R Z and R Z ◦ Suitable waveforms can be used on the test equipment, which is executed in TDL without penalty on resources. However, Yu Zhen; the actual implementation of i: | i is used for measurement, and more complicated waveforms are used. For example: 01 丨 ‘1 00ns’ D; ‘200ns’ D / U 400ns D; 丨 has the waveform shown in the T portion of FIG. 8 and the upper half of the waveform in FIG. 9. This example indicates 々yuan, 〇 'means that this paper size is applicable to the Chinese National Standard (CNS) A4 specification (210 X 297 mm) -12- II — — — — — — · 1111111 ^ · 111111. (Please read the (Notes? Matters on this page) • Line 1227329 Α7 Β7 V. Description of the invention (10) The falling edge at 100ns, the falling edge at 200ns, and the falling edge at 40ns. Substantially included. The character '1' indicates a falling edge at 100ns, a rising edge at 200ns, and a falling edge at 400ns. (Please read the note on the back first to write this page) The pattern character '0' can be matched by an NRZ or RZ waveform. The pattern character T will require more complex waveforms, such as X〇R (mutual exclusion or gate) with T 3 = ί 0 0 n s, T 1 = 2 0 011 s, T 2 = 4 0 0 n s. There are several reasons why this is an unwanted situation. First of all, if two characters are actually used for the pattern block (the definition is similar), it will cause the switching of the wave type in progress. In many ATTE systems, this creates a limit on the resources available. At the same time, in some systems, the use of X0R waveforms under certain conditions may result in a reduction in tester resources. Obviously, this is not a desired answer to the waveform matching problem. Therefore, in the following, some techniques will be detailed to simplify complex matching cases. Unused 値 Printed in STIL by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs. All information about the waveforms that can be used appears in the waveform table structure. This information can be analyzed to determine the characteristics of the available waveforms. This information department is used to complete the optimization of being informed. In this paragraph, the thoughts of "Unused Qiqi" are tested to help optimize. First, it should be noted that a set of pattern characters presenting a given signal defines a continuous waveform, regardless of the use of discrete characters. As a result, the start state experienced by a given character is based on the end state generated by the previous character. Therefore, the end of the group is included in the waveform table information. The exception to this description is the signal start, which can be arbitrarily set by the user. In order to optimize the model matching deduction method, it is important that the user adopts the Zhongguanjia Standard (CNS) Ai specification (210x 297 mm) ~ \ TZ-1227329 A7 for making smart choices on paper sizes. B7 V. Description of Invention (11). Referring back to the above example: 01 noons 'D;' 200n s 'Dchuan;' 400ns, D; 丨 and related waveforms are shown in Figures 8 and 9. It has been shown that the complete processing of this set of characters causes unwanted results. When analyzing this group of characters, it is known that none of them end with a "U" (rising) 値. As a result, by starting with a "0" (falling) signal, these are available characters, and the characters never begin any cycle with a "U" state. The waveform shown in the lower left of FIG. 9 is reduced to that shown in the lower right of FIG. 9. Therefore, a simple RZ waveform (Tl = 200ns; T2 = 400ns) can be used to match this set of characters. Table-based analysis The analysis of waveform tables based on STIL characters without using the initiation concept is explained with reference to Figures 10A and 10B and Figures 11 and 12. This details the classification requirements for all pictograph characters and forms the composition representative of these materials at several levels (S 1 1 in Figure 11). Note that this information does not have to be compiled into all pins or pin groups, but only in one signal. For each signal, the waveform requirements data is compiled in three layers (S12 in Figure 11). 1. The entire system 2. Waveform tables, and 3. Individual pattern characters Ideally, you want to use a single wave type to characterize the behavior of the entire signal, so the data is compiled in this layer. If this is not done, the next logical layer is the waveform table. Assume that the waveform table can be switched from one pattern square to another pattern (please read the precautions on the back-installation --- this page). The paper printed by the employee consumer cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs applies the Chinese national standard (CNS) A4 specification (210 X 297 mm) _ n A7 1227329 " ----- _B7 _ V. Description of the invention (12) When the pattern is changed, the waveform table will not switch frequently. Furthermore, if the switching between graphs and blocks occurs in the PatternExec block layer, this will be transferred to different tests in the TDL code, and the differences in the wave types will not cause the ongoing Of switching. Finally, if the behavior of pattern characters in the waveform table cannot be represented by a single wave type, matching will be performed on individual character layers. If this is not the case, it is indicated that some of the graphic characters in the STIL wall case contain requirements that will not be satisfied by the target ATE system test language. In some cases, the more advanced features of the target ATE system can be used to eliminate these problems. In other examples, this is simply M reported as a fatal error in the conversion process. The data to be stored in each layer is the same: the beginning and the number of edges required by the STIL pattern. This is stored in the array 歹 ij, whose size is based on the characteristics of the target ATE system. Capacity can be reduced using the "Unused Startup" technology described above. The number of driver edges supported by each time group is read in execution time, which is used for one dimension of the array. The other dimensions are two, which may start from the binary logic system, 1, 1, or, 〇 ,, 'mi to support the start, and each combination of the number of edges is set to true (T) ° Other It is false (F). Therefore, for the above example: 〇 {}, 1 {'200, ns U ;, 400ns' 1) "= > ruler 〇; ding 1 = 2 0〇1 ^; ding 2 = 40〇1 ^, then the array It will be as shown in Fig. 108 八 7K, where it is assumed that at most four driving edges can be supported per time group. Another ~~ · Example 23 {'100ns, D;' 200ns, D / U; '400ns, D;} The table in Figure 10B contains other inputs 値. This table describes all relevant paper standards applicable to the Zhongguanjia Standard (CNS) A4 specification (210 X 297 public love) —. — — — — — — 1111111 * — — — — — — — — (Please read the back first (Notes on this page) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs -16- 1227329 Α7 Β7 V. Description of the Invention (13) Information on the four waves of Yuyuan Techniques that do not use the starter can be applied to this table, because '' Γ is the starter (except when set at the beginning). As a result, the entire line labeled "1" can be set to false, which causes a situation where the phase θ is in the left and right lower waveforms of the aforementioned reference to FIG. 9. This mechanism can be applied to the tables created in any layer, that is, the aforementioned signals, waveform tables, and pattern characters. A more detailed description of the template matching procedure is as follows. As mentioned above, the array (matching matrix) of each decomposition event that forms the beginning and the number of edges is created in the event file in FIG. 6B and compared with the relative waveform data in the template file 25. Once the above "matching array" is created, then, the analysis of the "Unused Start" is reduced by 1: 丨 1, and the matching system for the available waveforms is completed (: S 1 3 in Figure Η). As discussed above, the waveforms (read in execution time) that can be used by the target ATE system are stored based on the number of edges and edges that they can support in the template file 25. The structure in the template is similar to that described above The matching array, except for the input in each box in the tables in Figs. 10A and 10B, is a set of wave target indicators, which can satisfy the representation combination of the starting frame and the number of edges. It should be noted that A given wave type will appear in several columns of these columns, because they can support various {starting 値, number of edges} pairs. For example, the RZ0 waveform will be ((), 0M 0, 2M 1, 0 } And u, 1 丨 are found in the array element, because it can support all these combinations. Note that it is not suitable for the R Z 0 waveform in TDL, and there is no activity for pattern character 0. Therefore, There are {1,0} combinations. The industry standard RZ waveform will return To zero, it is impossible to combine the pattern characters '0' and '1', and {1,0 丨. L_____ 一 ~ The national standard (CNS) A4 specification of this paper (21〇χ 297 mm) n ϋ nnn If i · — IB — & ϋ I n ϋ 1-(Please read the precautions on the back ymic page first)

Order --------- line J 1227329

V. Description of the invention (14) Wave-type object Π "to be asked about all the combinations it supports. For example, in order to fetch RZO objects via the {(), ()} k, the matching process can simply ask the Rz〇 object about other combinations "Can you support 丨 〇, 2}? This will return to true," You can support the board (1, 2}, / ki. Μ to false. Therefore, the matching process will find the object through the lowest number (W ίπ other edge M 丨 d), and then ask the object what other wants _. If any of these failures occur, this wave type will not enter the workpiece and matching processing into the full original column and try the processing again. Once successful matching is completed for a given wave type, then The parameter used for the parameter must be set (¾ S11 in Figure 1). The decomposed STIL character contains the time shirt used for conversion, and the wave type object contains the name stream of its related resources. For the waveform, (0 , 2 丨 Λ B matches 'Yuancheng' with related stream "T1", time 値 200ns and 400ns with fi-rate stream "T2". These pairs are stored in STIL equivalent. In the table, it is stored in the waveform table. Waveform main analysis method When the source format is cyclic, the use is economical. Related w] 用Information in the entire processed format can be used in the first half, and can be processed, analyzed, and optimized before the pattern conversion begins. Once completed. The vector processing becomes very simple. STIL pattern characters are accessed ( In S2 in Figure 12), the previous signal (stored at the end of the previous cycle) is not recalled (S2 2 in Figure 12), and these data are used to access the stored information Give {1 Do not start, STIL · pattern character) pair, that is, the wave type and parameter information determined by the matching process (S23 and S24 in Figure 12). Then, the end system is stored for the next The beginning of a cycle (S25 in Figure 12). This is quite different from the situation where the source format is event-based, such as VerlGg's VCD (値 Change Dump) data. In this example, the wave format information ( Please read the notes on the back first

(This page) · The paper size printed by the Consumer Cooperative of the Bureau of Intellectual Property of the Ministry of Economy applies the Chinese National Standard (CNS) A4 specification (210 X 297 public love) -17- A7 B7 1227329 5. The description of the invention (15) is not the first half Field available. The matching of the source data to the target waveform must be completed, when it is buried. This may cause poor selectivity, and avoid using appropriate tables (please read the notes on the back first to write this page) analysis techniques. Assume two cycles in the-waveform. If the template matching is done on a cycle-by-cycle basis (as in the example without table-based analysis), the first cycle will appear to end with a map to NRZ; this is the simplest waveform that can satisfy this And so on. The second cycle is the RZO of Chu Chu. If you know the second cycle when dealing with the first cycle, you should note that the first cycle can be mapped to RZO, and it will prevent a wave switching in progress. A table provides this capability for the primary analysis method. gLidl signal conversion Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs The foregoing description is for inputting the STIL character map to the TDL equivalent. “Enter” here means the test pattern applied to (driving) the pins of the device under test (DUT). As mentioned earlier, the test vector includes the test pattern (input) and the gate (output or comparison). A gate It is a timing pulse with edges (no pulse width) or window (predetermined pulse width) to output the signal by the sampling device. Here, the output mapping strategy is described as which one has a strobe signal in the STIL format (comparison) ) Become a TDL format conversion. Behaviors specified in STIL files can be directly converted into pattern characters. For example: LHZX {'〇ns'Z;' 〇ns' X; '260s' L / H / T / X; } Can be mapped directly to the edge gates for the appropriate frame, if all are available. In other words, the template matching is not necessary in this process. | Sample: LHZX {'〇ns' Z; 'o〇ns' X;' 260ns, 1 / h / t / x; JOOns, X} can be mapped to the appropriate window gating degree Applicable to China National Standard (CNS) A4 (210 X 297 mm)- 18-1227329

V. Description of the invention (16) ___ "At present, there are a few points to be noticed in the output conversion. First, some b allow on-going gating switching. Therefore, edge and window gating may not be '- ---------- Installation (please read the precautions on the back and write this page). The solution in this example is to make all strobes into window strobes. Allow it to be as narrow as possible. Second point, stomach k-STIL pattern characters that require multi-point gating in the loop may or may not be inconsistent ', the target ATE system. The capacity of the ATE system family is established to have resource ^ conversion tool of 1, the file system can be read at execution time, to indicate that ^ ^ ^ of which one set of lines actually appear on a given target ate system. this will, for example, the behavior of conversion and dual strobe modes. Bi-directional signal conversion W, the bit-to-bidirectional signal conversion provides the most complexity in the entire process. In 1 ^ >

The Ministry of Economic Affairs ’wisdom, wealth, bureau, Jr, a consumer masterpiece, and printed seals” means the conversion from the STIL format to the native language format used for test vectors. The test vectors are assigned to the input and output device pins. All features required for stomach input and output matching appear, and other features for bidirectional signals also appear. In addition, bidirectional signals often have other restrictions on test information placed by simple input or output signals. For example, available edges The quantity can be reduced due to the need of the driver control signal to determine the direction of the two-way signal. The consideration of driver control is based on the characteristics of STIL pattern characters and the capacity of the target ATE system. A standard model provides two driver enable modes,- One mimics the behavior of NRZ, one acts as RZ. In the former, the loop becomes "driven" (input) at some point and remains unchanged at the end of the loop. For RZ, the loop is "driven" and then in the loop " Compare, (output). Note that the paper size of the "Comparison" template applies the Chinese National Standard (CNS) A4 (210 X 297 mm) -19- 1227329 A7 _ B7 V. Description of the invention (17) The cycle of 忒 屮 does not indicate that a For comparison, only the pin is treated as an out. This distinction is important in terms of the ability to drive and compare in a cycle, the target A T E capability. The driver enable mode is determined by the STIL pattern feature [I to determine. It is better to note that the NR Z driver enable mode is better because it requires less tester resources. This mode is selected unless the RZ mode is specifically required. This can only happen when a "drive" area is surrounded by a "comparison" area in the same cycle. Again, a substantial comparison may not occur, but the device pins are used as outputs. The time used for the driver's control edge is determined by the switching time for the signal direction. The inclusion of driver type information makes the aforementioned "matching array" quite complicated. In the entire character of the · loop, the driving part of the signal is matched to the available waveform, and the capabilities of the target A T E system are compared with the 'driving' and 'comparing' parts. STIL contains the thought of a "previously driven" event. This indicates the most recently used driver included in the system. For input signals, the previous drive is always the last state of the signal, so this need not be taken into account. The output signal has no drive state, so this is irrelevant. Therefore, the previous drive is only related to the bidirectional signal, which represents the last drive state obtained by the signal, regardless of any inserted gating activity. In some ATE systems, the presence of strobe characters affects the state of the drives in the system. For example, if the last "drive" state is "D", and then followed by the strobe "H", the drive can be set to the "u" state in the next drive cycle. "Previous drive" is what you want to handle This state is related to the conversion process. This previous state must be maintained as the actual current state of the drive. The result is that the above-mentioned "matching array" contains four rows instead of two, and applies to all private paper sizes in China Standard (CNS) A4 specification (210 X 297 mm) _ 201-(Please read the precautions on the back Η This page) Printed by the Intellectual Property Bureau Employee Consumer Cooperative of the Ministry of Economic Affairs 1227329 Α7

V. Description of the invention (18) Taiwan dagger \ lk ί 'Dagger seal 丨 previous one, previous drive}, that is, the head of the line (pictures 丨 〇Α and 1 OB) is {(K〇K {OJ}, {!, 0 }, {1,1} 〇 The data used in these inputs must be properly used so that the idea of “lju run first” is properly actuated. The thought given by “previously driven” 値The initial state must be the same as the actual state of the driver, that is, the previous one. Use 1 ... Escape Example +, the driver is in the "U" state, and the STIL pattern character is based on the description of the "previously driven" or "P" ”Event, and the device is required to act as a 'D ,,' state. The waveform that matches this pattern character must be able to meet these / r jli " ~ within the required time, these are caused by the" u "state Drive this pin to the "D" state. This can become another edge, which is not obvious for the description of STIL pattern characters. (Please read the note on the back first 'mmm ·-ϋ JU page staff of Intellectual Property Bureau of the Ministry of Economy Printed by Consumer Cooperatives ^^ In the following, test language conversion is discussed further. Treatment of special characteristics in the system. These characteristics represent those found on the ATE system of the assignee AdVailtest model T66 00 1C test series, but they are quite common and may be found in various test systems. Therefore, they are suitable for use. A brief explanation is given by the deduction method of reflecting _ STIL information to these characteristics. The signal multi-clock (MCLK) signal type is often used to provide more pulses per cycle than the ATE system H | i 38, one nH " Many. This is done for repetitive repetition, by breaking the tester cycle (internally) into 1 cycle, and providing one of most copies of the basic waveform to h-1 for each cycle. The number of edges that may appear more than the fixed-rate test cycle Chinese National Standard (CNS) A4 specification (210 X 297 mm) 21

------ Order --------- Line I 1227329 A7 B7 V. Description of the invention (19) The main point of using the MCLK paradigm is that the waveform must contain a basic repeatable unit in a cycle. In the template matching deduction method, a STIL pattern character contains too many edges and the MCLK format is used to match the character. For this reason, the number of edges must be even (MCLK format is pulse-based). It must meet the limitations of the single pulse repeating waveform. It is related to the force D of the waveform on Figure 13 to derive. In order to have a repeatable unit The width of all pulses needs to be the same. Furthermore, the interval between pulses must be equal to the sum of the interval before the first pulse (A) and the interval between the ends of the last pulse (B). The meter tracking does not have to have a basic repeating unit, which looks like RZ, with T1 = A, T2 = A + PW, and the length of the secondary cycle is a + pw + b. If these conditions are not met, a single pulse repeating unit cannot be provided. It is possible to create a double-pulse basic repeatable unit. Requirements are derived based on the waveforms in Figure 13 below. In this example, each repeatable block has two pulses, and it requires that the relevant pulses have the same width, and β 卩 is labeled with pulses of PW1 and PW2. The sum of the start and end intervals must match the interval between repeating units, as described above (repeatable unit interval = A + Β). At the same time, the interval (C) between pulses must be the same for all repeatable units. Therefore, at this time, the length of one cycle is A + PW1 + C + PW2 + B. Pin Multiplexing In the PMUX method, two test channels are combined to drive a single pin. This allows the resources used for the two tester channels to be used for the same signal, to complete the drive (input) and comparison in the same cycle on the ATE system (the output is based on the Chinese National Standard (CNS) A4 specification (210 X 297) (Li) ------------- !! (Please read the note on the back page first)

Order ------------ Line I Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs-22- 1227329 A7 B7 Printed by the Employee Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 5. Description of the Invention (2〇)) , The system may not have this feature. Although using PMUX may make ATE planning easier, it adds complexity to the conversion process by providing additional degrees of freedom. Although this seems to be desirable in that it provides flexibility, it makes the search of the wave interval difficult, because the requirements are not detailed in detail. The relationship between PMUX on various pin types will be discussed below. ⑴Input PMUX: The input pin map is executed by searching the wave type interval, in order to match the ability of the wave type and the STIL pattern characters. The details of this method are discussed above. When an input signal is used in PMUX, the responsibility of matching the STIL pattern characters is to share the channel between the two testers. The matters represented by the booster α elasticity are matters that determine the method of "shared requirements". One of the methods adopted is the simplest sharing deduction method is not sharing. If possible, one pin does all the work and the other pin does not. In this regard, it is intended to map the edge designated by the STIL pattern character as the first tester channel. The matching that occurs here is the same as described for the non-PMUX input signals described above. If the matching attempt fails (that is, no wave pattern satisfies a subgroup of STIL pattern element requirements), a difference occurs. Previously, this situation also caused situations where one wanted to use MCK or report an error. At P M U X, one edge is shifted from the first channel to the second channel, and the matching system wants to be performed separately on the first channel and the second channel. The shift of the edges is continued until one match is possible for all two channels of all the element groups of the STIL pattern character. Increasing the complexity of this deductive method needs to be dimensioned between the behavior of the two constituent channels (please read the precautions on the back first to write this page).

This paper size is in accordance with China National Standard (CNS) A4 (210 X 297 mm) -23- 1227329 A7 B7 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs V. Description of Invention (21) Sustainability. "A given channel" remembers the final state of its last cycle. However, the state of the drive has been set by a signal from another channel. For example, if channel 1 drives the pin to the "H" state, it will cause "H" as its previous state. Now, suppose channel 2 drives the pin to "L" state. When channel 1 is driven again, it will "remember" the "H" state, but the driver will actually be "L" The state is due to the effect of channel 2. The coordination between the two channels containing the signal needs to prevent this state. For example, suppose the following signal in the uppermost waveform of Figure 14 where T. represents a cycle boundary. This signal The system is simple and has an NRZ character for the first signal. In fact, the second signal obviously does not need to do anything except to keep it low. Assume that the first pass solution to this problem is shown in Figure 14. The mentioned signal for the channel (pin 1) is provided on the rising edge of the drinking Ta (the second waveform above the figure 14). Because the second part of the signal has no edge, Pin 2 does not require edges (From the third waveform above the top of Figure 14) ° In the second cycle, pin 1 remains high because there is no edge in it. In the second part of the cycle, there is a downward edge. Because pin 2 It is already in the "L" state, so no edges will be generated. It is clear from this example that the simple method of shai fails in this example. This signal will have no edges as shown by the waveforms on pin 1 and pin 2. The reason for this is that 丨 big 为 is because the frontal chase / connectivity has not been maintained. Pin 2 needs to know the pins! The result is that the subsequent edges on the system will be properly processed. In this example, the transition from T a to “L” in the first cycle is shown. This can be shown at the bottom of Figure 15 The waveform of pin 2 replaces the waveform of pin 2. Here, the waveform of pin 2 is brought to "H," state at the time of separation. This will apply the Chinese national standard (CNS / A4 specifications (210 X ----- and '-y Δ.- ------- ------- installed ------) for this paper size ------ --Order --------- M (Please read the note on the back page first)

1227329 A7 B7 status. However, it does. 5. The description of the invention (22) constitutes a signal; it has no effect because the driver is already in the state of "H 'adjustment pin 2 and pin 1 due to the effect of pin 1. When pin 2 is the first When a downward edge occurs during a bad cycle, it will cause a directional conversion like pin 2 of Tb. In order to show the channel consistency problem, this must be handled by the PMU) (signal. (2) Output PMUX during general pattern matching The output signal conversion at the time of PMUX is better than $ 1 input conversion. The loop is divided into two parts, and the gate edge is assigned to two channels (pins) based on the position relative to the separation time. In simple terms, the separation time is selected as the middle of the cycle. The gated edges that occur before this time are assigned to the first channel, and the latter are assigned to the second channel. Note that one of this rules The exception is window gating. This may be separated on the boundary between channels. Therefore, the separation time is selected so that the entire window gating falls within one of the channels. (Please read the precautions on the back before this page ) Install ---- Order --------- Line Economic Affairs Intellectual Property Office employees consumer cooperatives printed

(3) Bidirectional PMUX The use of PMUX with bidirectional signals allows unobtained characteristics. This is the main purpose of the P M U X structure, not for pure input or output signals. Possibly, the most important use of the PMXX signal is to allow driving and comparison within a cycle, if the target ATE system does not allow it. For STIL pattern characters of bidirectional signals indicating pure driving or comparison behavior, the processing is substantially the same as the input and output pin multiplexing. When behaviors are mixed, the separation time used to drive the cycle between channels is based on the time of the direction switch. This results in a two-pass paper ruler of 7 A. Applicable to China National Standard (CNS) A4 specifications (210 X 297 mm) -25- A7 1227329 _B7_ V. Description of the invention (23) The responsibility is naturally divided. The idea we have discussed for the continuity between the channels of the input signal is also applicable here, but the effect of insert gating must be considered. Because the above has been explained in accordance with the present invention, the test vectors in the original pulp test language are converted into the target test language with high efficiency and high accuracy. Although only a preferred embodiment is described here, it can be understood that modifications and changes to the present invention may be completed within the scope of the teaching and patent application of the above technology without departing from the scope and spirit of the present invention. • — — — — — — — — VI — —VI I I I (Please read the note on the back page first)

Employees 'Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs of the People's Republic of China.' R This paper size applies to China National Standard (CNS) A4 (210 X 297 mm)-26-

Claims (1)

12 A month_correction / twisting / sample charging A8 B8 C8 D8 Annoying reading committee, express Qiu Yue Japanese cuisine ii 洚 t No change in substance? Is amendment allowed? Printed by the Employees' Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs. Appendix 2A: Patent Application No. 89 1 1 9684. Chinese Patent Application Replacement. Replaced on September 16, 1993. 1 The method of converting the test vector of the main test language into the target loop. The method of converting the test vector of the main test language includes the steps of: reading the available waveforms defined in the target test language and forming a set of templates describing the waveforms. The board is equivalent to one waveform of the target test language and contains data, which at least shows the beginning of the waveform part and the number of subsequent edges in the waveform; reads the test vector of the original test language and stores it in the original test language. The waveform in the test vector is decomposed into a set of constituent events, where each constituent event contains data, which at least shows the number of the beginning edge and subsequent edges of the waveform; compares the template derived from the waveform in the target test language with the The group of events derived from the original test language; storing waveform data in the target test language, When a match is detected in the comparison step, and the equivalent parameters of the waveform in the test vector of the original test language are retrieved, the matched waveform data is combined to store the parameters; repeat for all the test waveforms in the original test language The above steps are used to represent the waveforms in the target test language. 2. The method of converting test vectors as described in item 1 of the scope of patent application, wherein the steps of comparing the template and the set of events are included in different abstraction layers (please read the precautions on the back before filling this page) This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) 8 8 8 8 ABCD 1227329 7. The application of the f 3 patent scope applies a comparative deduction method, which uses sequence as a signal layer; a wave A seed layer in which the signal system is structured by most wave types; and a character layer in which the wave system is structured by multiple digital elements. 3. The method for converting a test vector as described in item 1 of the scope of the patent application, wherein the set of constituent events is stored in a table format, which has rows to specify to display subsequent edge numbers and start data. 4. The method for converting a test vector as described in item 3 of the scope of the patent application, wherein the table storing the set of constituent events is based on the end state generated by the previous constituent event and investigates the start of a particular constituent event値 Optimize to simplify the information in the table. 5. The method of converting a test vector as described in item 1 of the scope of patent application, wherein the test vector includes a driving signal, which is applied to the device under test (DUT) as an input and a strobe signal to sample the output of the DUT As an evaluation, the strobe signal in the original test language is directly converted to the target test. When the language is tested, the driving signal in the original test language is detected by comparing the template and the set of component events to detect the match and converted. Test the language for the target. 6 · The method of converting test vectors as described in item 1 of the scope of the patent application, wherein the waveform in the original test language is a plurality of cycles assigned to the target test language for resource limitation, and most of the cycles are performed by multiple The test cycle clock in the industrial test system is created, and the test system can be operated by the target test language. 7 · The method of converting test vectors as described in item 1 of the scope of patent application, wherein the waveform in the original test language is assigned to most test channels of the target test language, and most of the test channels are based on the target test language. Paper size applies to China National Standard (CNS) A4 specification (210X297 mm) (Please read the precautions on the back before filling this page), 1T Printed by the Consumers' Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs 1227329 A8 B8 C8 D8 ~, the patented test system is constructed in a way that is multiplexed to a single pin of the DUT. 0 ·· A kind of standard test interface language (STIL) The method of converting a test vector into a target loop as the main test language, including the steps of: reading the available waveforms defined in the target test language and forming a set of templates describing the waveforms, where each board is equivalent to one A waveform of the target test language and contains data, which at least shows the beginning of the waveform part and the number of subsequent edges in the waveform; reads the test waveform in STIL format and decomposes the waveform in the test vector in STIL format Is a set of component events, where each component event contains data, which at least shows the number of the beginning edge and subsequent edges of the waveform; compares the template derived from the waveform in the target test language with the template derived from the waveform in STIL This group constitutes an event; when a match is detected in the comparison step, the waveform data is stored in the target test language, and Retrieve the equivalent parameters of the waveform in the STIL test vector and combine the matched waveform data to store the parameters. Repeat the above steps for all the test vectors in STIL to form a test vector file for the target test language. (Please read the notes on the back before filling this page), tT Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs This paper is sized to the Chinese National Standard (CNS) A4 (210X297 mm)