草榴社区 Cloud
Cloud native EDA tools & pre-optimized hardware platforms
Request a Free Trial →
Introduction
The world is more connected than ever; it is not limited to human-to-human connection using mobile-phone or social media, but also millions of connected devices collecting and transmitting data. Today we live in a smart everything world where we may instruct our dishwasher to run at a specific time or our refrigerator to monitor temperature based on the food we store. Our car can optimize the car seat based on the driver. Microcontrollers (MCU) are a complete system containing a processing unit and in-built storage; that power this day-to-day automation. MCUs are the most cost-effective solution designed for programmability, making them ideal for IoT and Automotive applications.
Embedded Flash (eFlash) memory is a commonly used non-volatile memory (NVM) for in-built storage. Due to an increase in on-chip memory capacity and eFlash reaching its limit to scale below 28nm, this memory space is rapidly changing. Even though there are multiple contenders in-line for eFlash position: spin-transfer torque MRAM (STT-MRAM), phase-change RAM (PCRAM), resistive RAM (ReRAM), and Ferroelectric RAM (FeRAM), MRAM and ReRAM are ahead on the market adoption. These inflection points in the technology shift are exciting; many variables contribute to the success: market demand, manufacturability, reliability, and pros and cons of new technology. Embedded MRAM (eMRAM) will replace eFlash with its unique advantages.
Memory Hierarchy in MCUs
In a typical MCU system-on-chip (SoC), there are four types of memories (Figure 1):
- Cache memories: Small capacity memories designed using flips-flops or SRAMs. They require high-speed read and write operations and very-high endurance.
- Data memories: High-capacity memories that feed data to the processing units, designed to improve overall system efficiency. These are high-density memories, typically SRAMs.
- Key-store memories: Small capacity NVM used to store chip IDs, security codes, SRAM repair signatures, or trimming information for analog circuitry. They do not require high endurance; hence ROM, anti-fuse-based one-time programmable (OTP) memory, or eFuses are commonly used for these memories.
- Code and data-store memories: High-capacity NVM stores bootup code, firmware, and data. Memory capacity is increasing with AI/ML, IoT, and automation advancements, and power is becoming critical. Historically, eFlash is the choice for this purpose.
Figure 1: Block diagram of an MCU design
eFlash is Not Enough for Future MCUs
eFlash is a NVM solution integrated within an SoC. Thus, it is faster than having NVM outside the system. Based on the IP provider eFlash is developed using different flash technology like the floating gate, SONOS, or SG-MONOS. It requires additional masks (5-13); high voltage for program/erase (>5V); supports only page/block level erase; can’t do byte-write, making it an expensive high-power solution. These limitations were acceptable to the industry until 28nm when eFlash stopped scaling.
With increasing computational requirements, designers want to use smaller process geometries that provide higher performance with a lower area and power footprint. eFlash scaling limitation brings the designer two choices (Figure 2):
1. Keep NVM outside the compute SoC
- Pros: no need to scale memories with the compute die
- Cons: variable cost, higher latency, higher system power, and security
2. Keep NVM inside the compute SoC
- Look for alternative advanced NVM solutions, that support secure, high-performance, and low-power designs.
Figure 2: Advanced NVM solution is a better choice for future MCUs
eMRAM Leading the Emerging NVM Market
Unlike conventional embedded memories like SRAM and Flash, which store information via an electric charge, eMRAM stores data via its spin. The spintronic nature of eMRAM is comprised of ferromagnetic as well as non-magnetic materials which form a magnetic tunnel junction (MTJ). The MTJ maintains its polarization even when the power supply is removed, retaining stored data. eMRAM offers a smaller area, lower leakage, higher capacity, and better radiation immunity.
Compared to PCRAM and ReRAM, eMRAM has lower temperature sensitivity, provides better production-level yields, and offers longer endurance (retaining data over multiple read/write cycles over many years). It allows word-level erase and program operation, making it a power-efficient NVM solution. Even though eMRAM’s manufacturing cost is higher than ReRAM, its higher reliability and lower variability lead to area-efficient and robust design, which offsets the higher wafer cost. A single die can boast more memory with eMRAM, or a design utilizing eMRAM can be smaller and power-efficient with the same amount of memory. eMRAM is already in production with leading foundries at 22nm and now moving to FinFET nodes.
Figure 3: Unified eMRAM solution is the solution for advance MCUs
草榴社区 is Ahead on eMRAM
草榴社区’ eMRAM offering is a complete NVM memory IP solution that helps designers achieve higher memory capacity with a smaller area footprint and power. Our product is a configurable, fully integrated memory IP allowing designers to choose the most optimal instance per their needs (Figure 4). It includes a memory compiler, error correction code (ECC), and memory BIST, accelerating the design cycle and allowing faster time to market (TTM).
Figure 4: Configurable eMRAM IP provides flexibility for multiple word widths, aspect ratios, and floorplan optimizations
While eMRAMs present attractive advantages, designers should use a reliable silicon-proven solution with seamlessly integrated BIST and ECC support. To enhance the reliability and endurance of the solution, 草榴社区 eMRAM provides differentiated features like read-modify-write, write-verify-write, and programmable reference voltage generation. Our eMRAM design supports 150?C junction temperature making it an ideal choice for automotive applications.
MCU designers must consider eMRAM magnetic immunity while integrating the solution into the SoC. This involves testing the MRAM for its sensitivity level, reported in Gauss or Oersted, and informing their customers of this spec. Any elements near the chip that can become magnetic - such as inductor coils - can impact eMRAM performance, so system designers must protect against those elements by keeping them at a sufficient distance from the eMRAM. At 草榴社区, we test for MRAM magnetic immunity at static and dynamic conditions along with foundry-provided guidelines. Chip packaging with a magnetic shield can also protect the eMRAM from devices with a large magnetic field.
草榴社区 eMRAM solution uses low-risk design techniques to avoid read-disturb penalties. 草榴社区 SMS BIST and STAR ECC solutions provide a broad range of features to optimize manufacturing yield and reliability. STAR ECC supports configurable ECC implementation that allows flexibility and reliability in the design. Features like DECTED improves failure rate detection and corrections, which is highly important for Automotive applications. Time to market is another essential consideration. 草榴社区’ solution eases the IP integration process and provides a silicon validation report and silicon bring-up guidelines for faster turnaround time (TAT).
草榴社区 Memory Solution Meets Advanced MCUs Unique Needs
As a trusted IP partner with experience in volatile and non-volatile memory design, 草榴社区 offers a low-risk solution to help accelerate the development of high-quality MCUs:
- 草榴社区 eMRAM Compiler IP provides a configurable memory IP solution with options to optimize instance size, power-gating and redundancy features, and ECC schemes. The just-in-time compilation of eMRAM hard macros provides designer floor-planning flexibility, reducing design turnaround time which accelerates time to market.
- 草榴社区 SRAM & ROM Compiler IP provides designers with an extensive offering of high-speed, high-density, ultra high-density memory compilers to explore tradeoffs between power, performance, and area (PPA), leading to optimal memory configurations for the design.
- 草榴社区 NVM OTP Solution based on anti-fuse technology and enhanced with security features, delivers attack resistance and is the preferred choice of high-reliability applications.
- 草榴社区 Self-Test and Repair (STAR) Memory System? which provides a full suite of test, repair, and diagnostic capabilities for SRAM and eMRAM, optimizing test time without sacrificing test coverage. Configurable memory BIST and repair algorithms mitigate SRAM and MRAM defects.
- 草榴社区 STAR ECC Compiler IP improves in-field reliability by enabling multi-bit detection and correction, for example, DECTED. The IP maximizes manufacturing yield by infield error correction, which is essential due to the stochastic nature of eMRAM technology.
- 草榴社区 Silicon Lifecycle Management Family provides insight into the silicon to allow tweaks to performance levels or margins for better operation.
Memory will remain integral to every electronic device or system we use. eMRAM is poised to take on even more significant roles in next-generation, high-performing embedded applications by delivering high capacity, low power, and process technology scaling.
Contact your 草榴社区 sales representative to learn more about how 草榴社区’ eMRAM solution can improve your next product while reducing design risk.
Additional Resources
Blog:
Web page: 草榴社区 eMRAM Compiler IP
草榴社区 IP Technical Bulletin
In-depth technical articles, white papers, videos, webinars, product announcements and more.
Continue Reading
Article
Article
Newsletter
草榴社区 IP Technical Bulletin
In-depth technical articles, white papers, videos, webinars, product announcements and more.
Explore all articles →