Abstract: Embodiments of a scannable flip-flop are disclosed that may reduce data hold time, which may in turn improve the performance of circuits incorporating the scannable flip-flop. The scannable flip-flop may include a slave latch and a master latch including an input multiplexer. The multiplexer may include a number of input ports, for example to receive normal operating mode data as well as scan operating mode data, and the multiplexer may be operable to controllably select one of the input ports and pass the value of the selected port to an output of the multiplexer. For example, the multiplexer may generate individual control signals for the various ports dependent upon both the clock signal and a select signal, such that each of the ports is qualified with the select signal and the clock signal before the multiplexer presents the input data of the selected port as the output of the multiplexer.

Abstract: Embodiments of a scannable flip-flop are disclosed that may reduce data hold time, which may in turn improve the performance of circuits incorporating the scannable flip-flop. The scannable flip-flop may include a slave latch and a master latch including an input multiplexer. The multiplexer may include a number of input ports, for example to receive normal operating mode data as well as scan operating mode data, and the multiplexer may be operable to controllably select one of the input ports and pass the value of the selected port to an output of the multiplexer. For example, the multiplexer may generate individual control signals for the various ports dependent upon both the clock signal and a select signal, such that each of the ports is qualified with the select signal and the clock signal before the multiplexer presents the input data of the selected port as the output of the multiplexer.

Abstract: A memory (10) such as a current sensing static random access memory (SRAM) achieves fast write recovery through bit line loads and two additional mechanisms. First, an additional load (252) on shared data lines also becomes active to speed the write recovery process. Second, multiple columns (200, 202, 204) are connected to common data lines during write recovery so that a column written to during a write cycle may be again precharged in part by charge sharing using the charge stored in other columns. These two mechanisms allow fast write recovery with minimum column pitch and avoid the problems which would be encountered if the loads were placed on the write data line.