A High-Speed, Low-Power, High-Reliability and Fully Single Event Double Node Upset Tolerant Design for Magnetic Random Access Memory

Jiaxin Yang,Ziyang Guo,Shixuan Wang,Jilong Liu,Yue Zhang,Weisheng Zhao,Lang Zeng,Deming Zhang
DOI: https://doi.org/10.1109/tcsi.2024.3438282
2024-01-01
Abstract:Magnetic Random Access Memory (MRAM) has enormous application potential in the aerospace field due to its nonvolatile, high speed, low power, and inherent radiation resistance characteristics. Due to its high sensing reliability, pre-charge differential sense amplifier (PCDSA) has been proposed and widely used in MRAM products. However, such PCDSA is based on traditional CMOS technology, and as the size of CMOS technology continues to shrink, its sensing result is easily affected by single event upset (SEU) or even the single event double node upset (SEDU). Recently, a TSC-PCDSA has been proposed to fully tolerate SEDU. However, it still suffers from slow speed, high power consumption and low reliability during normal sense operation. To address these issues, this paper proposes a novel PCDSA circuit that uses 6 three-input approximate C-elements (TACs) and 2 three-input standard C-elements (TSCs) to provide SEDU-tolerance. By reducing the number of transistors on the discharge path and increasing the difference in discharge current, the proposed PCDSA can achieve high speed, low power and high reliability. By using a physics-based STT-MTJ compact model and a commercial CMOS 40 nm design kit, hybrid simulations have been performed to demonstrate its functionality and evaluate its performance. Simulation results show that when the TMR is 150%, the width of N1-N12 is 480 nm and the \(\text{V}_\text{DD}\) is 1.1 V, the proposed PCDSA sensing error rate (SER) is close to 0% during normal sense operation, achieving a high sense speed of 123.6 ps and a low sense energy of 1.6533 fJ. Compared with the previously proposed TSC-PCDSA, the sense reliability is greatly improved, and the sense time and sense energy are reduced by 1.84 times and 1.27 times, respectively. Moreover, the proposed PCDSA can fully tolerate SEDU by optimizing the layout design. In the worst case where deposited charge \(Q_\text{inj}\) is 2 pC, it can achieve a shorter recover time of 1.28244 ns and a lower recover energy dissipation of 2.1604 pJ than the previously proposed TSC-PCDSA.
What problem does this paper attempt to address?