Does Wireless Communication Drive the Evolution of Data Converters?
Modern wireless communication protocols are a response to the consumer demand to wirelessly convey large amounts of data from transmitters to receivers. These protocols are enabled by the significant amount of digital processing power available at both sides of the interface. However, because actual over-the-air transmission is analog by nature, advanced analog signal processing circuitry is key to achieving the ever-increasing performance demands for wireless communications systems.
Data converters are mandatory blocks within wireless communication systems and have experienced a significant evolution in recent years. The following examples illustrate the key drivers for the evolution of data converter intellectual property (IP) in handheld devices and in base stations for home/office environments.
The handheld device application case
The most common implementation of a communications system is segmented into two separate die: radio frequency (RF) and baseband (BB). These two dice are commonly integrated into the same multi-chip module package.
Communication between these two dice is achieved through an analog interface where the data converters are integrated in the BB die. In some cases, the interface between them can be digital, offering potential advantages that come from having fewer pins (especially if diversity is implemented), but there is some penalty in terms of overall power dissipation. In these cases, the data converters are integrated together with the RF die. In both cases, data converters are an integral part of the communications data path.
Mobile products are the key market driver for wireless communications. These devices are battery powered and very power conscientious. In fact, together with cost (silicon area), the most significant driver of innovation in the data conversion IP arena is power conservation, not performance or other features.
An illustration of this market trend can be found when browsing through trade publications. Interestingly, even if there is no standard defined for how a data converter should look inside a wireless communications interface, the performance specified for today’s converters has converged into a very limited number of different definitions that are common to all data converter vendors. These sets of performance definitions evolve slowly, following the pace of communications standards evolution.
So, in reality, changes in the communications systems definition do not happen very often for most systems integrators, and the cell phone illustrates this point. In fact, in the 25-plus-year timeframe in which cell phone usage has become commonplace, there have been only about four generations of communications protocols. However, every year vendors have come up with a multitude of different features and product definitions that basically changed everything except the actual definition of the communications interface. Even the advent of multi-antenna techniques to improve communication quality and reliability, as well as data rates, has not necessitated a change in the data converter definition. Instead, clever uses of the same converter have been implemented, such as taking advantage of multiplexed analog-to-digital converter (ADC) inputs or simply integrating several of them to support multiple data streams coming from a number of antennas.
In short, along with the addition of dazzling new features, the consumer has come to expect that battery life will increase with each generation of mobile products. To achieve this, the power dissipation of each component in the system has to be reduced.
The home/office base-station case
Although the evolution of the handheld device is most visible to the consumer, the home/office base-station on the fixed side of the network is rapidly evolving, as well. This is driven by the necessity to offload the network cells and provide better coverage and higher capacity in areas of concentrated load generation, such as office buildings or homes. An interesting development is the widespread adoption of femto-cells used as local wireless access hubs inside offices.
In contrast to the headset, femto-cells have to support multiple communication streams simultaneously. This naturally leads to a different definition of the system architecture. Typically, large multi-antenna arrays are used in order to increase data throughput. These systems require high-speed, high-resolution data converters to discriminate with enough precision signals on a large spectral band.
Systems-on-chip (SoCs) built for these applications need to implement multiple parallel data processing paths (one per antenna); therefore, the area of each of the components that is replicated is critical for the overall system cost. Interestingly, although these applications are not battery powered, overall power dissipation is an important design criterion for avoiding the increased packaging costs incurred when special packages are used to dissipate excessive heat generation.
Data converters for wireless communications
The field of data conversion IP has experienced dramatic advances, in several domains. In a 10-year period, a data converter providing the same level of performance has reduced its power dissipation by up to 25 times, along with a fivefold or more reduction in area. These advances are, in part, the result of using faster, deep-submicron processes and lower supply voltages. However, most of the credit should go to advances in the converter architecture. By exploring digital “performance enhancement” techniques, such as digital background calibration and other architectural tricks, it is possible to alleviate the performance limitations of each analog block within the converter while maintaining or even improving the overall performance of the converter. These techniques are pushing the limits of the high-performance data converter sampling rate to frequencies that were previously not achievable in , using CMOS processes, further enabling system architects to creatively change the way communications systems are defined.
For example, the availability of higher sampling rate converters facilitates the use of alternative RF demodulation schemes. These include using IF (intermediate frequency) demodulation rather than zero-IF demodulation, or employing architectures like the highly flexible Software Defined Radio, in which a broad section of the spectrum, including multiple communication channels, is sampled simultaneously, and channel selection and down conversion are carried out in the digital domain.
In addition to the aforementioned advantages related to reduced power dissipation, smaller area and higher sampling rates, these techniques offer the additional benefit of reducing the blocks’ sensitivity to process and temperature variations. The improved robustness achieved makes these data converter IP components more suitable for integration in complex digital SoCs, even as they migrate down to very deep-submicron process generations – beyond 45/40 nanometers (nm).
The evolution of data converters has tracked well with the new requirements set by the emerging communications standards and is enabling the development of alternative architectures. In fact, the power dissipation reduction, as well as the area improvements and higher sampling rates that are routinely announced to the market, facilitate new system architecture evolutions. In parallel, the advances in data converter technology have significantly improved their robustness, making them highly suitable for integration in complex digital SoCs, as well as on analog RF chips.
When considering the analog components of such wireless communications systems, system designers sometimes overlook data converters. However, data converters provide one of the most critical functions in a wireless communications system, and their performance defines the competitiveness of the overall solution.