The conventional approach for testing RF circuits is specification-based testing, which involves verifying sequentially all specification requirements that are promised in the data sheet. This approach is a long-time effective testapproach but nowadays suffers from significant drawbacks.First, it requires generation and capture of test signals at the DUT operating frequency. As the operational frequencies of DUT are increasing, it becomes difficult to manage signal generation and capture using ATE. As a consequence, there is a need of expensive and specialized equipment. In addition,as conventional tests target several parameters, there is a need of several data captures and multiple test configurations. As a consequence, by adding settling time between each test and test application time, the whole test time becomes very long, and the test board very complex. Another challenge regarding RF circuit testing is wafer-level testing. Indeed, the implementation of specification-based tests at wafer level is extremely difficult due to probing issues and high parasitic effects on the test interface.Moreover, multi-site testing is usually not an option due to the small count of available RF test resources, which decreases test throughput. Hence, the current practice is often to verify the device specifications only after packaging.The problem with this solution is that defective dies are identified late in the manufacturing flow, which leads to packaging loss and decreases the global yield of the process.In order to reduce production costs, there is therefore a need to develop test solutions applicable at wafer level, so that faulty circuits can be removed very early in the production flow. This is particularly important for dies designed to be integrated in Systems-In-Package (SIP).In this context, a promising solution is to develop indirect test methods. Basically, it consists in using DUT signatures to non-conventional stimuli to predict the result of conventional tests. The underlying idea is to learn during an initial phase the unknown dependency between simple measurements and conventional tests. This dependency can then be modeled through regression functions. During the testing phase, only the indirect measurements are performed and specifications are predicted using the regression model built in the learning phase.Our work has been focused on two main directions. First, we have explored the implementation of the alternate test method based on DC measurements for RF circuits and we have proposed a methodology to select the most appropriateset of DC parameters. Results from two test vehicles (a LNA using electrical simulations and a PA using real production data) indicate that the proposed methodology allows precise estimation of the DUT performances while minimizing the number of DC measurements to be carried out.Second, we have proposed a novel implementation of the alternate test strategy in order to improve confidence in alternate test predictions and to overcome the effect of limited training set sizes. The idea is to exploit model redundancy in order to identify, during the production testing phase, devices with suspect predictions; these devices are then are removed from the alternate test tierand directed to a second tier where further testing may apply. / The conventional approach for testing RF circuits is specification-based testing, which involves verifying sequentially all specification requirements that are promised in the data sheet. This approach is a long-time effective testapproach but nowadays suffers from significant drawbacks.First, it requires generation and capture of test signals at the DUT operating frequency. As the operational frequencies of DUT are increasing, it becomes difficult to manage signal generation and capture using ATE. As a consequence, there is a need of expensive and specialized equipment. In addition,as conventional tests target several parameters, there is a need of several data captures and multiple test configurations. As a consequence, by adding settling time between each test and test application time, the whole test time becomes very long, and the test board very complex.Another challenge regarding RF circuit testing is wafer-level testing. Indeed, the implementation of specification-based tests at wafer level is extremely difficult due to probing issues and high parasitic effects on the test interface.Moreover, multi-site testing is usually not an option due to the small count of available RF test resources, which decreases test throughput. Hence, the current practice is often to verify the device specifications only after packaging.The problem with this solution is that defective dies are identified late in the manufacturing flow, which leads to packaging loss and decreases the global yield of the process.In order to reduce production costs, there is therefore a need to develop test solutions applicable at wafer level, so that faulty circuits can be removed very early in the production flow. This is particularly important for dies designed to be integrated in Systems-In-Package (SIP).In this context, a promising solution is to develop indirect test methods. Basically, it consists in using DUT signatures to non-conventional stimuli to predict the result of conventional tests. The underlying idea is to learn during an initial phase the unknown dependency between simple measurements and conventional tests. This dependency can then be modeled through regression functions. During the testing phase, only the indirect measurements are performed and specifications are predicted using the regression model built in the learning phase.Our work has been focused on two main directions. First, we have explored the implementation of the alternate test method based on DC measurements for RF circuits and we have proposed a methodology to select the most appropriateset of DC parameters. Results from two test vehicles (a LNA using electrical simulations and a PA using real production data) indicate that the proposed methodology allows precise estimation of the DUT performances while minimizing the number of DC measurements to be carried out.Second, we have proposed a novel implementation of the alternate test strategy in order to improve confidence in alternate test predictions and to overcome the effect of limited training set sizes. The idea is to exploit model redundancy in order to identify, during the production testing phase, devices with suspect predictions; these devices are then are removed from the alternate test tierand directed to a second tier where further testing may apply.

Être en mesure de vérifier si un circuit intégré est fonctionnel après fabrication peut s'avérer très difficile. Dans le cas des circuits analogiques et Radio Fréquence (RF) les procédures et les équipements de test nécessaires ont un impact majeur sur le prix de revient des circuits. Une approche intéressante pour réduire l'impact du coût du test consiste à mesurer des paramètres nécessitant des ressources de test faible coût et corréler ces mesures, dites mesures indirectes, avec les spécifications à tester. On parle alors de technique de test indirect (ou test alternatif) car il n'y a pas de mesure directe des spécifications, qui nécessiterait des équipements et du temps de test importants, mais ces spécifications sont estimées à partir des mesures « faibles couts ». Même si cette approche semble attractive elle n'est viable que si nous sommes en mesure d'établir une précision suffisante de l'estimation des performances et que cette estimation reste stable et indépendante des lots de circuits à traiter. L'objectif principal de cette thèse est de mettre en œuvre une stratégie générique permettant de proposer un flot de test indirect efficace et robuste. Pour être en mesure de construire cette stratégie nous avons amenés différentes contributions. Dans un premier temps, on a développée une nouvelle métrique dans cette thèse pour évaluer la robustesse des prédictions relaissées. Dans un deuxième temps, on a défini et analysé une stratégie pour la construction d'un model optimal. Cette dernière englobe un prétraitement de données ensuite une analyse comparative entre différentes méthodes de sélections de mesures indirectes aussi l'étude d'autres paramètres tels que la taille des combinaisons de mesures indirectes ainsi que celle de la taille de set d'apprentissage. Aussi on a proposé une stratégie pour une confidente exploration d'espace de mesures indirectes afin de construire plusieurs meilleurs modèles qu'on peut se servir par la suite pour résoudre des problèmes de confiance et d'optimisation. Les études comparatives réalisées ont été effectuées sur 2 cas d'études expérimentaux et à partir de métriques classiques et de la nouvelle métrique proposée permettant ainsi d'évaluer objectivement la robustesse de chaque solution.En fin, nous avons développé une stratégie complète mettant en œuvre des techniques de redondance de modèles de corrélation qui permettent d'améliorer grandement la robustesse et l'efficacité de la prise de décision en fonction des mesures obtenues. Cette stratégie est adaptable à n'importe quel contexte en termes de compromis entre le coût du test et le niveau de confiance et de précision attendu. / Being able to check whether an IC is functional or not after the manufacturing process is very difficult. Particularly for analog and Radio Frequency (RF) circuits, test equipment and procedures required have a major impact on the circuits cost. An interesting approach to reduce the impact of the test cost is to measure parameters requiring low cost test resources and correlate these measurements, called indirect measurements, with the targeted specifications. This is known as indirect test technique because there is no direct measurement for these specifications, which requires so expensive test equipment and an important testing time, but these specifications are estimated w.r.t "low-cost measurements". While this approach seems attractive, it is only viable if we are able to establish a sufficient accuracy for the performance estimation and if this estimation remains stable and independent from the circuits sets under test.The main goal of this thesis is to implement a robust and effective indirect test strategy for a given application and to improve test decisions based on data analysis.To be able to build this strategy, we have brought various contributions. Initially, we have defined new metric developed in this thesis to assess the reliability of the estimated performances. Secondly, we have analyzed and defined a strategy for the construction of an optimal model. This latter includes a data preprocessing followed by a comparative analysis of different methods of indirect measurement selection. Then, we have proposed a strategy for a confidant exploration of the indirect measurement space in order to build several best models that can be used later to solve trust and optimization issues. Comparative studies were performed on 2 experimental data sets by using both of the conventional and the developed metrics to evaluate the robustness of each solution in an objective way.Finally, we have developed a comprehensive strategy based on an efficient implementation of the redundancy techniques w.r.t to the build models. This strategy has greatly improved the robustness and the effectiveness of the decision plan based on the obtained measurements. This strategy is adaptable to any context in terms of compromise between the test cost, the confidence level and the expected precision.

The conventional approach for testing RF circuits is specification-based testing, which involves verifying sequentially all specification requirements that are promised in the data sheet. This approach is a long-time effective testapproach but nowadays suffers from significant drawbacks.First, it requires generation and capture of test signals at the DUT operating frequency. As the operational frequencies of DUT are increasing, it becomes difficult to manage signal generation and capture using ATE. As a consequence, there is a need of expensive and specialized equipment. In addition,as conventional tests target several parameters, there is a need of several data captures and multiple test configurations. As a consequence, by adding settling time between each test and test application time, the whole test time becomes very long, and the test board very complex.Another challenge regarding RF circuit testing is wafer-level testing. Indeed, the implementation of specification-based tests at wafer level is extremely difficult due to probing issues and high parasitic effects on the test interface.Moreover, multi-site testing is usually not an option due to the small count of available RF test resources, which decreases test throughput. Hence, the current practice is often to verify the device specifications only after packaging.The problem with this solution is that defective dies are identified late in the manufacturing flow, which leads to packaging loss and decreases the global yield of the process.In order to reduce production costs, there is therefore a need to develop test solutions applicable at wafer level, so that faulty circuits can be removed very early in the production flow. This is particularly important for dies designed to be integrated in Systems-In-Package (SIP).In this context, a promising solution is to develop indirect test methods. Basically, it consists in using DUT signatures to non-conventional stimuli to predict the result of conventional tests. The underlying idea is to learn during an initial phase the unknown dependency between simple measurements and conventional tests. This dependency can then be modeled through regression functions. During the testing phase, only the indirect measurements are performed and specifications are predicted using the regression model built in the learning phase.Our work has been focused on two main directions. First, we have explored the implementation of the alternate test method based on DC measurements for RF circuits and we have proposed a methodology to select the most appropriateset of DC parameters. Results from two test vehicles (a LNA using electrical simulations and a PA using real production data) indicate that the proposed methodology allows precise estimation of the DUT performances while minimizing the number of DC measurements to be carried out.Second, we have proposed a novel implementation of the alternate test strategy in order to improve confidence in alternate test predictions and to overcome the effect of limited training set sizes. The idea is to exploit model redundancy in order to identify, during the production testing phase, devices with suspect predictions; these devices are then are removed from the alternate test tierand directed to a second tier where further testing may apply.

在臺灣的技職教育體系中，職業學校群科課程綱要是教材編撰及課程設計的參考基準；而四技二專統一入學測驗是高職學生升學的重要依據。其中，統測共同科目英文考科中的對話題，採用了間接測驗來評量學生的口語能力。本研究旨在討論統測英文科對話題與高職英文課綱之吻合程度，及統測對話題的內容效度。為了達成此研究目的，本研究分析統測英文考科對話題型以及高職英文課程綱要，同時參考與測驗口說能力之相關教學研究及論文，發展出課綱檢核表及口說能力檢核表做為研究工具。 研究結果顯示，統測對話題與高職課綱中口說能力相關指標大致符合，但主題分佈不甚平均，主要強調日常生活情境下的溝通能力。就內容效度而言，其檢驗之能力多為基礎口語技巧，並且偏重測驗考生如何傳遞訊息及維持互動，並沒有包含處理互動的技巧。根據本研究之發現，筆者針對未來研究方向及測驗實務提出了建議。 / In the vocational education system in Taiwan, the Vocational High School Curriculum Guideline (VHSCG) is the basis of teaching materials and curriculum design, and the Technological and Vocational Educational Examination (TVEE) is an important reference for students to enter college. In the English test of the TVEE, an indirect speaking test was applied in the dialogue section. This research aims to investigate: (1) the correspondence between goals of speaking ability in the VHSCG and the dialogue section of the TVEE; and (2) the construct validity of the dialogue section. To achieve the purpose, this research analyzed the current curriculum guideline and the English tests of the TVEE, and studied the previous literature about assessing speaking ability to develop two checklists as instruments to analyze the targeted test items. Results indicated that the dialogue section of the TVEE generally corresponded with the speaking ability index in the VHSCG, but the topic distribution was imbalanced. The test items primarily focused on the communicative ability in daily-life contexts and was lacked of items that examine descriptive ability. For the construct validity, the dialogue section of the TVEE mostly examined fundamental speaking skills. Informational and interactional skills were strongly emphasized, while the skills in managing interaction were completely absent. Based on the findings of this research, the researcher yielded suggestions for future research and implications for the test developers.