It’s Time To Fix The NVM Shortfalls In Mobile Devices
The world of mobile devices owes much of its phenomenal progress to semiconductor advances. From the bulky Motorola “brick” in 1980 to today’s palm-size handset, integration, power, and cost have seen similar advances to those of computers. Today’s smart phones will continue to pack more functionality at the same sales price, while feature phones will continue to break low-water marks on bill-of-materials (BOM) cost.
In a teardown of a typical smart phone, we can see that it is already a very integrated system. There are very few DRAM and NAND flash components and fewer digital logic chips, which are easiest to integrate because of available EDA tools and methodologies. Also, the analog/mixed-signal chips are becoming multi-functional. But there are 10 to 20 nonvolatile memory (NVM) chips, such as EEPROMs and serial flash chips, spread around the printed-circuit board, each next to a digital or mixed-signal chip.
After each power reset, the EEPROMs provide software programs, user IDs, configuration settings, and so on—a small amount of code, but critical to phone initialization. At approximately $0.20 per chip, total EEPROM cost adds up to a non-trivial $2 to $4 per system. What’s more, these NVMs take up space, draw power, and likely contribute to the industry’s field returns of greater than 1%.
The Role Of NVM
One solution to this problem is for all components to share one larger NVM, such as NOR flash. But because each digital and analog/mixed-signal chip that needs NVM comes from a different manufacturer, the complexity of developing software and validating protocols creates a significant testing and integration challenge—the biggest time-to-market barrier to a carrier’s deployment of a new phone.
Another solution is integrating the NVM functionality into each digital and analog/mixed-signal chip. This seems obvious, because the NVM size of less than 1 million bits or transistors is small compared to that of digital chips, frequently greater than 100 million transistors. However, the memory and logic chips often use different manufacturing process technologies and have different design considerations, resulting in higher but limited integration, as well as higher cost. The time has come to create an embedded NVM technology that conforms to the basic CMOS semiconductor process technology and still delivers the density and capacity to satisfy the integration imperative.
In the basic feature phone, the cost of this NVM as a percentage of total BOM is probably similar to that of current architectures, and in this market every penny counts. But the integration imperative transcends cost. For example, the target market for feature phones in emerging economies requires this integration to deliver a more secure mobile device for financial transactions.
In much of Africa, in rural India, and in other developing countries, the mobile phone may be the only infrastructure available for such transactions, with no bank branches nearby, no Internet connection, and no storefronts in sight. Innovations in Africa have enabled people to make transactions using phone credits as currency. As a result, the mobile phone has become the backbone of financial transactions, acting as an ATM card, bank depository, and wire-transfer station all in one device.
Continue to next page.
Want to use this article? Click here for options!
© 2012 Penton Media Inc.
Acceptable Use Policy blog comments powered by Disqus
advertisement
Latest Issue
Features:- Android Opens Up The Operating System For Innovation
- The Future Of Apps Lies In The Enterprise And On TV
- Engineering The Differentiation Into Smart Phones
Most Popular Stories
advertisement
advertisement
