MCPA in 3G Apps

Multicarrier power amplifiers (MCPA) will allow service providers to drive 3G performance.

Not long ago, some of us shared a naive vision that the 3G/IMT-2000/UMTS standardization activities would lead to widely accepted requirements and specifications, resulting in equipment commonality with multiple sources and lower costs. Although value is trending up, the 3G revolution is more notable as a period of system innovation and variety, rather than hardware standardization. Designing power amplifiers to meet 3G requirements is technically challenging, and issues in power-amplifier selection and planning have become more complex. Here are the 3G implications and trade-offs for power-amplifier hardware.

Why Use MCPA?
The two most common approaches to integrating power amplifiers in base-station transmitters are to use parallel signal paths including several power amplifiers (individual single-channel power amplifiers in an output multiplexer) or to multiplex signals at a low power level and amplify all signals in a single higher-power MPCA.

The multicarrier approach reduces the number of power amplifiers, but places a larger burden on the amplifier in terms of peak power capability and linearity. This results in a higher cost for the amplifier and a possible single point of system failure. However, service providers and OEMs have recognized the significant advantages of MCPA in terms of greater network flexibility and lower cost of network ownership. As a result, end users have caused the industry to bring performance-driven MCPA to market that are reliable and cost-effective. Today, MCPA are the accepted technology for wireless networks.

The single-channel, power-amplifier approach often requires less investment in initial deployments, but the multicarrier approach ultimately supports higher capacity and significantly greater flexibility. Because many 3G applications must achieve higher base-station capacity in limited space, the multicarrier approach is expected to be the 3G configuration of choice. With this approach, the service provider can deploy a network that meets the initial capacity demands and have the flexibility to increase the network capacity as demand increases. This has the advantage of making network upgrades less complicated, but more importantly, it extends the life of the installed network.

What's Different About 3G Amplifiers?
The innovative proposals for 3G base-station configurations are the source of considerable development effort by power-amplifier manufacturers, as their IR&D programs adapt to what sometimes appear to be moving targets. Amplifier designers must translate these new system requirements to the physical parameters (that is, frequency, power, linearity and efficiency) that drive the amplifier design. It's apparent from even a cursory assessment of the wireless-business environment that major commitments have been made to cdma2000 (1X and 3X), GPRS, EDGE and UMTS. The CDMA evolution path involves progression from cdmaOne (IS-95B) to cdma2000-1X to cdma2000-3X. The GSM evolution path is from GSM to GPRS to EDGE and then to UMTS. The TDMA path also includes EDGE and UMTS. A common thread of all 3G-evolution plans is the end goal of increased data rates and higher capacity. Specific provider requirements may involve different spectrum assignments as well as a range of output power. The range of output power may vary by 10:1, depending on the application, from lower-power microcells and repeaters to high-power macrocells in urban environments. Third generation generally requires wider bandwidth and higher linearity of base-station equipment, often at higher power levels. This constitutes a new generation of engineering challenges for designers of power amplifiers for wireless base stations.

Protocol Impact
Major differences in protocols and spurious specifications drive new amplifier linearity and peak power requirements. The protocol-dependent signal statistics of most interest to the amplifier designer is the complimentary cumulative distribution function (CCDF), which gives the probability a given power level will be exceeded for an associated peak-to-average value. The CCDF allows estimation of the maximum (peak) instantaneous power required of the amplifier. This is critical because all 3G protocols have stringent linearity demands, and because an amplifier linearity degrades abruptly when its peak power capability is exceeded. A general design guideline is that the amplifier peak power must equal the specified average power times the peak-to-average at which the CCDF is 1 X 10^-2%(0.0001). A 1 X 10^-2% CCDF means that one has a 0.9999 probability that the power is below the level implied by that peak-to-average ratio.

Three example CCDFs for different protocols are shown for single-channel IS-95 standard CDMA, for multicarrier EDGE (8 channels in the example) and for 2-channel UMTS (wideband CDMA).

These CCDFs are chosen as representative examples — the precise CCDF of each protocol may vary according to a number of additional parameters not included in this discussion. The peak-to-average indicated at the 1 X 10^-2% level ranges from 9dB to 12dB. That is to say, the peak transistor power (for the same average power) will vary by approximately 2:1 for these different applications. Besides affecting peak power requirements, the standards associated with the particular protocol also specify the spurious and linearity requirements for base-station amplifiers. These two parameters, peak power and linearity, are priority drivers for amplifier designers.

Amplifier-Design Differentiators
Specification of a power amplifier starts with determination of the signal conditions presented to and expected of the amplifier. Frequency band of operation affects power-amplifier design and cost, primarily due to escalating power-transistor cost with frequency. Other than transistor cost, the novelty of higher-frequency operation (for example, PCS or IMT-2000 vs. cellular) is diminishing, and the circuit-design and manufacturing challenges associated with a jump in frequency have greatly diminished. At all frequencies, differentiation between base-station amplifiers is strongly dependent on signal complexity and output linearity requirements.

Design Issues
A base-station designer is concerned about capacity and coverage. Therefore, the critical requirements a power-amplifier designer must determine are:

• Frequency band of operation
• Peak output power
• Linearity and spurious requirements
• Power consumption (efficiency)
• System-interface requirements.

More than 50% of the amplifier cost will be due to the cost of the power transistors. The determining factor in the selection of transistors is the peak power and the linearity expected of the uncorrected devices. Once a class of transistor has been chosen (Si LDMOS vs. Si Bipolar or GaAs FET, for example), the issue affecting output transistor lineup becomes one of peak power required. A starting estimate is provided by the following equation:

Transistor total peak power = (average output power)*(signal peak-to-average at 10^-2% CCDF probability)/(output circuit losses)

Examples of calculation of relative transistor costs estimations are given in Table 1. These include several specific cases: single-channel PCS, multicarrier cellular, moderate-power multicarrier 2G CDMA, IMT-2000 multicarrier (2.11GHz to 2.17GHz), and multicarrier EDGE at PCS frequencies. Because power-transistor cost is the largest contributor to overall cost, there clearly will be a major difference in P/A costs for these different approaches.

These calculations indicate that the power-transistor cost for an IMT-2000 60W multicarrier is 3.2 times that of a single-channel 30W PCS amplifier, on a per-watt basis. This difference is contributed to by the large peak-to-average of the signal (protocol dependent), the sophistication of correction circuitry required (and therefore of associated output losses) and the frequency of operation. Fortunately, the capacity gain with the 3G MCPA is far greater than the ratio of transistor costs, and value is added with the more complex amplifier.

Total transistor power and cost are one indication of amplifier complexity. The second major contributor to amplifier complexity is the linearity required above and beyond that achievable with an uncorrected transistor circuit. The final amplifier linearity is a result of the basic power-module (uncorrected transistor circuit) performance and the "correction" offered by some schemes for linearization. The linearization architecture can take many forms. Predistortion and envelope-feedback schemes are examples of approaches with modest correction circuitry and, most significantly, where the associated insertion loss occurs ahead of the costly power stages. These pre-power stage circuits tend to be cost-effective, but they achieve limited correction — typically 5dB to 20dB. Many high-power (>40W) multicarrier applications with high peak-to-average signal environments require a feed-forward architecture. Feed-forward offers superior correction (³30dB typically) but at the expense of added circuit-insertion loss after the output power stages (dissipating more power and using more RF power devices) and of an additional error amplifier. Thus feed-forward amplifiers are more complex and more costly than amplifiers with predistortion alone. Many 3G systems involve multicarrier operation with high peak-to-average signals, a combination that requires a feed-forward approach. Some dynamic range requirements are sufficiently severe (multicarrier GSM and EDGE, for example) that feed-forward alone in insufficient, and combinations of predistortion and feed-forward will be necessary. Requirements for 3G that involve the highest power levels and greatest linearity are sufficiently challenging that they are currently the subject of intense IR&D efforts by amplifier suppliers.

Many of the 3G specifications for adjacent-channel power and spurious measurements are at a preliminary release level, and may be evolving. Spurious measurements involve precise definition of integration bandwidths and offsets. There is limited consensus on measurement methodology, but engineers and technicians often simplify requirements to "delta-marker" spurious rejection levels. Delta-marker refers to the difference in power spectral density measured using a spectrum analyzer, at band center compared to a stated offset. Typical uncorrected power stages would produce "spectral regrowth" of from -25dBc to -35dBc for multicarrier CDMA outputs, whereas the specifications might require -50dBc to -60dBc. The difference between the final amplifier system performance and the output stage performance is the required correction (i.e. 25dB correction to bring -30dBc uncorrected to -55dBc corrected).

By way of example, if you have uncorrected power-stage linearity of -30dBc, and the HPA system must achieve -55dBc, then a single-loop, feed-forward system should be adequate. If the requirement were for -65dBc, then some combination of predistortion and feed-forward probably is required. If, as in the case of eight simultaneous GSM or EDGE channels, you need -80dBc correction (the amount varies according to geographical area), then you must use either sophisticated predistortion plus feed-forward, or possibly dual-loop feed-forward (an inefficient and costly approach). It appears that some 3G approaches will demand a combination of high peak-to-average signals and extremely low spurious measurements (i.e. -80dBc). These higher levels of performance will require not-yet-available test equipment for verification, as well as significant advances in amplifier design and manufacturing technology. Whereas most 3G requirements are achievable with predictable refinement of today's technology, some approaches will require extension of the state of the art.

Good designs demand excess performance margin to assure high manufacturing yields and robust service experience. Besides the techniques discussed above (feed-forward and predistortion), amplifier architectures also are distinguished by the degree of adaptive capability. These products must maintain high performance over many years. Many 3G-correction systems will be expected to monitor their own effectiveness and adapt to changes due to device aging and environmental changes. The current generation of multicarrier products has sophisticated DSP at the heart of its correction, and this allows for more sophisticated correction, adaptive control and software-programmability as 3G product advances are announced.

Equipment for 3G must enable higher capacity, higher data rates and more efficient spectrum usage. This will result in protocols that involve wider bandwidths, higher peak-to-average signals, often higher peak power levels and lower spurious emissions. This means that transmitter power amplifiers will be required to produce higher peak-power levels simultaneously with lower spurious products. Amplifier designers will achieve these demands through usage of improved power transistors and application of more sophisticated correction technology. These products also will be more adaptive and more programmable. All of these requirements constitute exciting challenges for the amplifier-design community. Most importantly, this is an opportunity for amplifier suppliers to introduce 3G products with increased performance and value as their contribution to the wireless infrastructure community. The explosion in user demand for higher data rates and more pervasive access to wireless services is creating welcome challenges and opportunities.

Spear (james_spear@spectrian.com) is Spectrian product line manager for MCPA, and Crescenzi (james_crescenzi@spectrian.com) is Spectrian principal scientist.

---

Key Factors Affecting Power-Amplifier Design

Operating Frequency — 3G-base-station protocols commonly involve frequencies in the cellular, PCS and IMT-2000 band (2.11GHz to 2.17GHz).

Output Power — Typically 20W to 100W. Varies greatly from microcell to macrocell applications.

Signal Statistics (peak-to-average and distribution) — Protocol-dependent. Typically range from 7dB to 12dB for multicarrier 3G protocols.

Peak Output Power — Determined by average output power and signal statistics. Power-transistor costs are driven by the peak power required.

Multicarrier Operation — Number of operating channels. Linearization circuit complexity increases with the number of channels.

Linearity Requirement — Adjacent Channel Power & Spurious Requirements — Often expressed as a ratio of transmitter channel power to the undesirable power in adjacent channels. The simplest measurement form is “delta-marker” using a spectrum analyzer, in dBc. Requirements can vary from 40dBc to 80dBc, depending on application.

Signal Bandwidth — Multicarrier performance over wide operating bandwidths is the most challenging.

DC/RF Efficiency — High efficiency is a challenge, particularly for MCPA. “Good” efficiency may be as low as 7% for high-peak power amplifiers processing complex signals. Base-station power-amplifier efficiency often is sacrificed to achieve difficult linearity and spurious requirements.

Physical Size & Environment — Mechanical design is dominated by thermal considerations and the operating environment. High power in fanless outdoor enclosures is most challenging.