Abstract:
The invention concerns a circuit protected against transitory perturbations, comprising a combinatorial logic circuit (10) having at least an output (A); a circuit (20) generating an error control code for said output, and a storage element (24) provided at said output, controlled by the circuit generating a control code to be transparent when the control code is correct, and to maintain its status when the control is incorrect.
Abstract:
The invention concerns a digital circuit architecture comprising combinational circuits (10, 12), short-term memory circuits (11) not capable of storing data for more than k operating cycles, long-term memory circuits (13) capable of storing data for more than k operating cycles of the circuit. Systems for protection against different perturbations are used for the different types of circuits and based on the functionality of said circuits.
Abstract:
One embodiment provides a runtime programmable system which comprises methods and apparatuses for testing a multi-port memory device to detect a multi-port memory fault, in addition to typical single-port memory faults that can be activated when accessing a single port of a memory device. More specifically, the system comprises a number of mechanisms which can be configured to activate and detect any realistic fault which affects the memory device when two simultaneous memory access operations are performed. During operation, the system can receive an instruction sequence, which implements a new test procedure for testing the memory device, while the memory device is being tested. Furthermore, the system can implement a built-in self-test (BIST) solution for testing any multi-port memory device, and can generate tests targeted to a specific memory design based in part on information from the instruction sequence.
Abstract:
One embodiment provides a runtime programmable system which comprises methods and apparatuses for testing a multi-port memory device to detect a multi-port memory fault, in addition to typical single-port memory faults that can be activated when accessing a single port of a memory device. More specifically, the system comprises a number of mechanisms which can be configured to activate and detect any realistic fault which affects the memory device when two simultaneous memory access operations are performed. During operation, the system can receive an instruction sequence, which implements a new test procedure for testing the memory device, while the memory device is being tested. Furthermore, the system can implement a built-in self-test (BIST) solution for testing any multi-port memory device, and can generate tests targeted to a specific memory design based in part on information from the instruction sequence.
Abstract:
The invention concerns an integrated circuit chip comprising a large number of semiconductor components having parasite components through which a short circuit between the supply voltage and the circuit ground is likely to be triggered, said semiconductor components being distributed into elementary blocks, each elementary block being independently connected to supply or ground lines of the main supply network of the integrated circuit by a current limiting device capable of checking a short circuit triggered in the block concerned, and each block being dimensioned such that software errors likely to occur in said block are capable of being corrected by means of error correction.
Abstract:
A technique for transparent testing of integrated circuits has been described. It allows to test the integrated circuits without losing the applications' execution context. The technique considers two kinds of blocks, combinational and sequential blocks on the one hand and RAMs and register files on the other hand. For the first kind of blocks we propose a save, test and restore technique. It takes advantage from a circular scan path implementation and consists on shifting (via the scan path) the contents of various registers until the data inputs of a RAM, it saves the registers' contents in the RAM, performs the test session, loads back the saved data to the scan path and shifts them until the corresponding registers. For the second kind of blocks we propose a transparent test technique allowing to test the RAMs without destroying their initial contents. To do that we propose a technique allowing to derive RAM transparent test processes. These thest processes can be implemented by means of BIST techniques, or by means of scan path techniques. The transparent test processes are composed by a signature prediction or a signature initialisation process and a basic transparent test process. The signature prediction test process is obtained from the basic transparent test process by removing all the write operations. The read operations such that the read data are not injected to the output compaction scheme can also be removed. This technique avoids fault masking due to errors produced during both the signature prediction test process and the basic transparent test process. The basic transparent test process can be obtained from any standard (i.e. non transparent) test process by using several transformation steps. These transformation steps are such that for all the RAM fault models verifying some symmetric property, the basic transparent test process offers the same fault coverage as the standard test process. This property is verified by nearly all the RAM fault models. The signature initialisation process is derived from the signature prediction process by reversing its ordering. When this process is used it is coupled with some new data compactors named Up/Down data compactors. These compactors allow for the whole process (i.e. signature initialisation process plus basic transparent test process) to give a predicted signature.
Abstract:
L'invention concerne une architecture de circuit numérique comprenant des circuits combinatoires (10, 12), des circuits de mémoire de courte durée (11) non suseptibles de stocker des données pendant plus de k cycles de fonctionnement, des circuits de mémoire de longue durée (13) susceptibles de stocker des données pendant plus de k cycles de fonctionnement du circuit. Des systèmes de protection contre les perturbations distincts sont utilisées pour les différents types de circuits et selon la fonctionnalité de ces circuits.
Abstract:
Static memory comprising memory cells (MC) arranged in rows and columns, each row of cells constituting of at least one word and a corresponding error control code. Each cell column comprises at least one specific power line (Vcc', GND') connected to a general power line (Vcc', GND') of the memory through a current detector (10, 12). Each detector activates an error signal if the associated specific supply line current exceeds a predetermined threshhold.
Abstract:
The present invention relates to a fail-safe control interface including branches for providing signals having a safe state or a non-safe state. Each branch comprises inputs for receiving at least two binary control signals (Si, Si*); a source of a non-safe state (Fe) connectable through a basic chain of elements (14, 15) to an output (Oi) when the control signals realize a predetermined combination; a concurrent checker (17) providing an error detection signal (g1, g2) if the inputs of a pair of its inputs are at predetermined states; and means (14*) for providing a first input of said pair of inputs with a signal corresponding to the state of said output and the second input of said pair of inputs with a signal corresponding to the output of a duplicate chain of the basic chain, this duplicate chain reacting like the basic chain in response to the control signals.
Abstract:
Method of self-checking arithmetic units and data paths using the double rail code for the arithmetic operators and a parity code for the other blocks. The method uses a single block for the generation of both codes, thus avoiding the need of code translators. The ripple-carry adders, ALU's, multiply and divide arrays are implemented with DCVS or static differential gates, to avoid overhead (Output Checking/Parity Generation Scheme). When the adders or ALU's are of a certain length, the schemes used are the Carry and Output Checking/Parity Generation and Carry Checking/Parity Prediction.