Microcontrollers have become a widely accepted architecture for highly complex embedded systems on a single chip (SoC). It consists of deeply embedded heterogeneous components with poor accessibility makes their testing process a difficult task using hardware based self-test (HBST). Software-based self-test (SBST) is considered to be a promising testing technology for these systems. Almost every SoC contains at least one embedded processor, SBST utilize this processor for test pattern generation (TPG) and test response compaction (TRC) based on its instruction set, then test response will be unloaded and evaluated using external automatic test equipment (ATE). In this paper, SBST strategy disadvantages in microcontroller testing will be identified. Then, a new testing approach that combines both the HBST and the SBST, called hybrid-based self-test (HYBST) will be introduced. Based on a divide-and-conquer approach, HYBST identify microcontroller's components and their corresponding component operations. Feasibility and effectiveness of HYBST and SBST methodologies will be assessed by applying them to a Microchip ® PIC16F877A and PIC18F452 in terms of memory usage, time consumption and number of tested modules found in microcontrollers.

Testing of embedded system including microcontroller is difficult task with external Automatic Test Equipment (ATE). Therefore, empowering the microcontroller to test itself as software-based self-testing (SBST) looks the suitable solution like the microprocessor testing. Practically, the SBST is not suitable for microcontroller testing. It utilizes large space area in the program memory inside the microcontroller that has limited space area in the available memory. Also, it cannot test all microcontroller internal modules and when it test internal modules it cannot make sure that the General Purpose Input Output (GPIO) of the microcontroller work probably without using external ATE. So the Design for Testability (DFT) methodology that uses Instruction Set Architecture (ISA) of the microcontroller family to generate test subroutines and for the Test Pattern Generator (TPG) and part of the Built-In Self-Test (BIST) control unit and uses the external ATE for the other part of the BIST control unit and for the test response compaction (TRC) and evaluation. This paper introduces a hybrid testing methodology that combines both SBST and hardware-based self-test (HBST) for microcontroller testing as an efficient DFT methodology. It introduces for either in the field or as part of a production test of a microcontroller as an example of the system of chip (SoC). This DFT methodology is based on divide and conquer algorithm and requires knowledge of the ISA of the microcontroller to test not only the embedded processor found in microcontroller but also test other peripherals found in it using brute force technique. The comparison between the SBST and the presented hybrid methodology is based on memory utilization, number of clock cycles that was taken to complete each test and the number of modules that can be tested using each of them. Experimental results indicate that the presented methodology is superior in memory utilization, test time and can test all microcontroller modules for both 18F4X2 and 16F87X families.

In this paper, a new hybrid test strategy, called hybrid-based self-test (HYBST), is presented to test complex digital circuits such as microcontrollers. This test strategy integrates the signature multi-mode hardware-based self-test (SM-BST) with the software-based self-test (SBST). In this test strategy, the microcontroller is divided into a number of main modules, and then test subroutines are used to functionally test each module, based on its instruction set architecture (ISA). The ISA is used to generate test subroutines that represent test pattern generators (TPGs) and part of the test controller. The SMHBST represents the other part of the test controller and the test response compaction (TRC). The experimental results illustrate the superiority of the HYBST in the memory utilization, test application time, testing of internal modules of the microcontroller, and testing of general-purpose input-output (GPIO) pins of the microcontroller. In addition, an integrated test solution for fault diagnosis of the circuit boards including random logic integrated circuits (ICs) and microcontroller chips is presented to indicate a real practical test strategy.