|
Ever improving non-volatile static random access memories (nvSRAMs) are here to address the shortcomings of battery-backed SRAMs. Providing a significant power management advantage, these parts integrate a high-speed SRAM cell with non-volatile elements to provide powered-off back up while eliminating batteries altogether. Here's what they can do for you.
Overview
System designers have traditionally combined static RAMs and batteries to protect data during power outages, whether planned or unexpected. Initially, the systems were discrete implementations built with low-density memory circuits and rudimentary battery technologies using discrete power sensing and switching circuits.
Good performance was a challenge. Batteries had limited capacity with questionable reliability. To meet the low power required for extended battery-backed operation, memories were designed with slow access times. Ultimately, devices improved significantly as designers came up with modules, first using dual-in-line and through-hole and, later, surface mount packages. By putting the memory and control circuits in the same package with the battery, or by making a two-package implementation where the battery snaps onto the integrated circuit, designers shifted the burden of good circuit design, manufacturing, and reliability to the component supplier.
One additional feature became almost synonymous with battery-backed SRAMs: the real-time-clock (RTC). Since many applications needed to display time and date information along with time-stamping system events, it was only natural to combine these capabilities with the battery-backed SRAM.
Even as memory densities, integrated control circuits, and battery technologies improved, however, the same fundamental deficiencies remained. Whether in modular packages or discretely assembled on the printed circuit board, battery-backed SRAM continues to suffer from low reliability, complex manufacturing, a large footprint, slow performance and, now with greater environmental awareness, difficulty in keeping with the intent of moving towards completely "green" solutions. Let's take a look at these issues.
The problems
The typical battery-backed SRAM implementation comprises four components: SRAM, voltage monitor/controller, the battery, and a battery socket, although the socket can be eliminated in modules (Fig. 1). If we simply multiply the failure rate of each component we have a reasonable first-level approximation of the system's reliability. Additionally, we should also consider the number of traces and connections required for all of the interconnects between devices. Take special note of the battery connection, too, which is frequently socketed. Indeed, the battery connection introduces the possible of additional failure mechanisms due to contact corrosion and circuit board vibration (intermittent connections).
(Click on Image to Enlarge)
Fig. 1: Traditional battery-backed SRAM
Battery life is also a big variable, dependent upon how often the system cycles, the temperature, and the type of battery used. Nominally, an SRAM operating with a standby current of 5 microamps and powered by a 165-mAh battery will last less than four years. If you put this system through temperature extremes, whether while operating or in storage, you'll see that this number may degrade dramatically.
Manufacturing with batteries has been difficult primarily because of the battery's inability to withstand the extreme temperatures used for reflow soldering, a condition which is exacerbated with the higher temperature profiles being introduced for lead-free solder. Temperature issues initially required the battery or the module to be assembled as a secondary manufacturing operation: after the printed circuit board was assembled the battery was added in a manual step. Later, modules were built with the SRAM and control ICs embedded in a surface-mount package that went through the normal reflow cycle. A separate molded package containing the battery snapped onto it as a secondary operation. The molded battery was attached with plastic clips and the electrical connections were simply pressure connections subject to the same corrosion and vibration concerns of the socketed battery.
The board space required to support battery-backed SRAM has changed with the evolution of technologies but still remains unsatisfactory. Discrete implementations have always required more area, largely because of the battery. Even with today's high-density packaging, a 44-pin TSOPII for the memory, a μDFN for the controller, and a 20-mm coin-cell require at least 555 mm2, and that doesn't account for routing and manufacturing tolerances. A popular battery-backed SRAM module today is packaged in a 27-by-27 mm BGA package, which takes 729 mm2 of board space. No battery-backed SRAM implementation today can claim the combined footprint and height of a standard single package monolithic memory.
High-speed access wasn't critical in the early life of battery-backed SRAMs, since system speeds were modest and in many cases these memories weren't used for real-time processing. Since very low-power SRAMs were required to conserve battery power, it was convenient that low power designs were also slow. Today, most battery-backed implementations are rated an access time of 100 ns, although some perform as fast as 55 ns. Again, these specifications require that the designer balance system speed requirements against data retention time, which is a function of standby current and battery capacity.
Non-green
This brings us to an issue that draws more attention than all the previous issues combined—that of being "green." This issue is highly charged, and battery-backed SRAMs can be especially sensitive because the memories are usually deeply embedded in systems and can't be easily inspected, repaired or, most importantly, properly disposed of. Within the industry there is much ado about complying with the European Union's Restriction of Hazardous Substances Directive or RoHS.
RoHS restricts the use of six substances: lead, mercury, cadmium, hexavalent chromium, polybrominated biphenyls, and polybrominated diphenyl ether. In the semiconductor industry we generally think of this as the "lead-free" initiative. It is not commonly understood, however, that batteries, except as they might use some of the restricted materials above, are exempt from the RoHS directive.
However, the Directive on Batteries and Accumulators and Waste Batteries is targeting the use of less hazardous substances in the manufacture of batteries, and improving on waste management. Other countries, including the United States with the EPA's 1996 Battery Act, are implementing similar initiatives to control the environmental impact of batteries. These initiatives will put increasing pressure on battery users and, by adding environmental-awareness pressures and costs related to monitoring and disposal, will force electronics manufacturers to look for viable alternatives.
|
