IC keeps phone costs ultralow

To view an on-demand seminar on Infineon's E-Gold radio, click here.

The market for cell phones is fragmented into two main streams. One comprises countries and regions that have been using cellular handsets for years. Users here typically look for fashionable cell phones with more bells and whistles. For the entry-level, low-cost handset market alone, this sector accounts for roughly 400 million units per year at an approximate price between $40 to $80 per handset.

The other stream comprises emerging markets in which cellular technology is not as prevalent. Handset cost is the main focus in this developing wireless industry; features are secondary. Expected demand for cell phones in emerging markets, including areas like India, Africa and Latin America, is 240 million units per year at an approximate price of $30 to $40 per phone, with a move toward the $30 target. To that end, companies such as Infineon and Texas Instruments have released products that are designed to reduce the handset bill of materials by integrating the baseband and RF into a single chip.

The dual functionality can potentially reduce the three biggest challenges in mobile systems: space, power and cost. All three relate to the number of components needed to perform the necessary functions on a handset. Not only does an integrated solution reduce the number of chips, it also eliminates trace leads, allowing for considerably smaller pc boards. Since there are fewer components, they do not have to be independently powered, lowering power consumption. Finally, a shorter components list means less to purchase.

Even if the price for one component is increased slightly, if another component is eliminated altogether the OEM will come out ahead. Industry members estimate component count and board space will be reduced by as much as 50 percent, while BOM costs will decrease by 25 percent or more.

For chip companies, integration could also help lock in solutions with OEMs. Designing out an integrated solution, especially when there is only one other identified alternative solution, requires a complete redesign, as opposed to simply designing out a single device. Theoretically, one would have to introduce two devices and all of the associated traces, power lines and control circuits — and that only takes into account the baseband and RF functionality. When companies integrate further features into the solution, such as camera controllers or MIDI tones, it makes the solution even more valuable.

The Infineon E-Gold radio, which we reported on in "BOM shrinks handset costs" (www.eetimes.com, search article ID: 191901457), was the first instance of a single-chip solution we saw. We have now analyzed Texas Instruments' single-chip LoCosto solution, and we are suitably impressed with the process technology that was used, as well as with the feature advantages it offers OEMs.

Die photograph of the Texas Instruments LoCosto highly integrated, dual-processor baseband/RF phone chip shows the RF block on the right.

The LoCosto is based on TI's 90-nanometer process technology, yielding a considerably small die size — a necessity for mobile applications. The LoCosto die measures about 25 percent smaller than that of the Infineon E-Goldradio, which was designed using a 130-nm process technology. The die size savings will allow TI to produce more dice per wafer, either to sell at a lower cost or to maintain a higher profit margin.

The small process in itself is quite impressive, in light of the design constraints that the integrated functionality would impose. When circuitry is reduced to smaller processes, especially in the RF domain, crosstalk and performance glitches can result. TI has taken advantage of its DRP (digital RF processor) architecture, which was created as a general RF solution that could leverage the cost and power benefits of volume CMOS process manufacturing regardless of the air interface. As future generations of the code are revised, the architecture conforms to the necessary standards of the target application. That gives TI an advantage, as it can fine-tune the code in software to reduce the number of respins necessary.

In addition to the LoCosto solution, TI's DRP architecture has been successfully integrated into its WiLink 4.0 mobile wireless-LAN platform and three generations of Bluetooth products (the BRF6100, BRF6150, and BRF6300). The company's integrated DRP road map includes a single-chip digital-TV solution for mobile phones, GPS and other air interfaces.

Included in the RF portion of the LoCosto device are the transmitter, receiver, variable crystal oscillator and A/D converters.

When designing for the low-cost handset market, trade-offs must be made for every feature implemented. Adding features raises the system cost, so care must be taken not to price it out of the market. LoCosto integrates an ARM 7 core, for high performance with low power consumption. At first glance the choice of the ARM7 may seem odd, considering that ARM 9 cores have been used for a number of years and that ARM 11 cores have started finding their way into devices. I am sure TI recognizes the importance of ARM 9 cores for multimedia applications, but the ARM 7 core is more optimized for low-cost handsets.

Given the LoCosto solution's feature set, the ARM 7 offers a fair trade-off between cost and functionality. The core allows a system to display low-frame-rate multimedia up to video playback. While not providing sufficient multimedia capabilities to meet the more-established cell phone markets' handset expectations, the ARM 7 core is adequate for introductory systems and is widely adopted by the industry.

The ARM 7 core operates in conjunction with TI's TMS320C54x DSP core. The dual-processor platform eliminates an external processor to perform basic functionality, such as camera control and MIDI tones. TI has also incorporated enough SRAM to support most of the features required in the low-cost handset market without the need for external SRAM. While the part can connect to external SRAM, for most features it offers a large enough buffer to operate, yielding significant BOM savings and reducing the system component count even further.

Gregory Quirk (gregoryq@semiconductor.com), technology marketing manager at Semiconductor Insights Inc.(www.semiconductor.com).