Third-generation (3G) wireless communications systems are designed to improve upon the performance and services available from their first- and second-generation (1G and 2G) predecessors. The goal of 3G systems is true seamless global mobile communications sharing full compatibility with selected access technologies such as wireless local loop (WLL), cellular, cordless, and satellite-communications

systems. One technical challenge to the advent of seamless global-terminal mobility is the difficulty in achieving a common global-frequency plan. In every world region, at least part of the necessary spectrum is already allocated for other radio services.

Following a spectrum allocation around 2 GHz (roughly 1880 to 2200 MHz, depending upon geographic region) by the World Administrative Radio Conference (WARC) in 1992, the International Telecommunication Union—Radio-communication sector (ITU-R) began to define a wish list for 3G-system requirements. A range of technologies were proposed to meet these requirements, including orthogonal frequency-division multiplex (OFDM), opportunity-driven multiple access (ODMA), time-division synchronous code-division multiple access (TDSCDMA), and wideband CDMA (WCDMA). A technical body called the Third-Generation Partnership Project (3GPP) was organized to analyze the proposed technologies, with WCDMA selected as the preferred technology for 3G systems. The 3GPP standards organization has since written a technical-requirements specification in which chapter 25.101 includes the key performance requirements for the RF hardware portion of a WCDMA mobile terminal.

The 3GPP also defined two choices of operation for a WCDMA terminal: a frequency-division-duplex (FDD) mode and a time-division-duplex (TDD) mode. In the former, physical channels are defined by the RF channel number and the channelization code. The FDD mode is suitable for fast-moving mobile use. The uplink and downlink functions are separated in the frequency domain, and the approach offers greater downlink capacity then uplink capacity. The FDD approach employs a 100-percent duty cycle on both the uplink and downlink functions. In the TDD mode, physical channels are defined by the RF channel number, the channelization code, and the time slot. This approach is suitable for indoor or slow-moving mobile use. The uplink and downlink functions have similar capacities and occupy the same channel, with discontinuous transmission (DTx) on both the uplink and downlink.

DTx is a method for optimizing the efficiency of wireless voice-communications systems by momentarily powering-down or muting a mobile or portable telephone in the absence of voice input. In a typical two-way conversation, each party speaks slightly less than one-half the time, so if a transmitter (Tx) is on during voice input only, the telephone's duty cycle can be cut to less than 50 percent. This conserves battery power, eases the workload of Tx components, and frees time for the channel—allowing the system to take advantage of available bandwidth by sharing the channel with other signals. DTX circuits operate with voice-activity detection (VAD), which in wireless Txs is sometimes called voice-operated transmission (VOX). The 3GPP specifications also include FDD terminals for 60-MHz chunks only, with 190-MHz duplex spacing: 2110 to 2170 MHz for mobile receive and 1920 to 1980 MHz for mobile transmit.

Chapter 25.101 of the 3GPP specification covers the receive and transmit electrical requirements for FDD 3G mobile terminals. Before exploring WCDMA Tx requirements, it may help to review some of the key Tx parameters and their importance in Tx design. Adjacent-channel power ratio (ACPR), for example, is a measure of the amount of interference or power in an adjacent-frequency channel. Usually defined as the ratio of the average power in the adjacent frequency channel (or offset) to the average power in the transmitted-frequency channel, ACPR describes the amount of distortion due to nonlinearities in the Tx hardware.

ACPR is critical for WCDMA Txs, because CDMA modulation produces closely spaced spectral components in a modulated carrier. Intermodulation of those components causes a spectral regrowth of "shoulders" around the center-carrier frequency. Nonlinearities in the Tx can disperse those spectral-regrowth components into adjacent channels.

Error vector magnitude (EVM) is the vector (magnitude and phase) difference at a given instant between an ideal error-free reference and the actual transmitted signal. Because it changes continuously during every symbol transition, EVM is defined as the root-mean-square (RMS) value of the error vector over time. EVM is critical for WCDMA Tx performance because it indicates modulation quality in the transmitted signal. A large value of EVM results in degraded transceiver performance by causing poor detection accuracy.

Frequency error is the difference between specified and actual carrier frequencies. A large frequency error degrades transceiver performance by causing adjacent-channel interference and poor detection accuracy. Spurious and harmonic signals are tones produced by different signal combinations in the Tx, and harmonics are distortion products produced by nonlinear behavior in the Tx. Harmonics occur at integer multiples of the transmitted signal.

Page Title

Table 1 shows key requirements for specifying and designing 3G WCDMA Tx terminals. These requirements are met by a variety of ICs from Maxim Integrated Products (Sunnyvale, CA), including the MAX236X superheterodyne IC, using a typical Tx IF of 380 MHz. The company's MAX2383 upconverter driver has also been developed for superheterodyne architectures; it handles transmit IFs to 570 MHz.

As part of a complete WCDMA transceiver, a Tx reference design was developed based on commercial hardware (Fig. 1). The reference design includes the MAX2388 receive front end, the MAX2309 intermediate-frequency (IF) quadrature demodulator, the MAX2363 quadrature modulator/upconverter Tx, and the MAX2291 RF power amplifier (PA). The Tx assumes an IF of 380 MHz and a transmit frequency of 1920 to 1980 MHz. A duplexer filter allows full-duplex operation by connecting the transmit path (and receive path) to an antenna.

At the back-end of the Tx, the MAX2363 accepts baseband-transmit I and Q differential-input signals as inputs, and performs quadrature modulation, IF and RF LO synthesis, and RF upconversion. The IF LO is synthesized by an internal VCO and PLL running at 760 MHz. An external RF VCO module allows high-side injection of −7 dBm into the MAX2363 upconverter IC. On-chip RF drivers allow the chip to drive an external PA directly.

At the front-end of the Tx, a chip-scale-packaged linear PA (model MAX2291) provides 28 dB gain and as much as +28 dBm output power at 1-dB compression. With a post-PA insertion loss of approximately 4 dB, the system achieves maximum antenna output of +24 dBm. Since mature WCDMA systems are expected to operate typically at mid-power rather than full-power levels, the MAX2291 addresses this requirement via two (low- and high-power) optimized output-power modes. Measurements in the high-power mode, with Vcc at +3.5 VDC, show output power of +28 dBm at 1.95 GHz, ACPI of −39 dBc (measured at a 5-MHz offset within a 3.84-MHz channel), power-added efficiency of 37 percent, and idle Icc of 97 mA. Measurements in the low-power mode, with Vcc at +3.5 VDC, reveal output power of +16 dBm at a frequency of 1.95 GHz, ACPI of −38 dBc (measured at a 5-MHz offset in a 3.84-MHz bandwidth), power-added efficiency of 14 percent, and idle current, Icc, of 30 mA.

The 3GPP specification calls for WCDMA Tx power between −50 and +24 dBm, a 74-dB dynamic range. Allowing for some margin, version 1 of the WCDMA Tx reference design achieves better than 80-dB dynamic range. The dynamic range of a Tx chip is limited, usually by ACPR at the high-power end and by the noise floor at the low-power end. To obtain more than 15 dB carrier-to-noise (C/N) ratio at the low-power end, an additional 20 dB of variable attenuation (introduced by a gain-control attenuator for the PA) was designed into version 1 of the reference design. Key performance parameters extracted from extensive test results (Table 2) verify the suitability of Maxim's Tx reference design.

An EVM of approximately 5.7 percent was measured at +24 dBm Tx output power from the reference design (attributing 3.5 percent to the MAX2291 PA and 4.6 percent to the MAX2363 Tx chip). The overall EVM value is well within the 3GPP requirement (less than 17.5 percent). Measurements on the EVM and ACP for the transmit strip were as follows: transmit strip EVM at −20 dBm (Fig. 2), transmit strip EVM at +24 dBm (Fig. 3), transmit strip ACP at +24 dBm (Fig. 4), and transmit strip ACP at −20 dBm (Fig. 5).

Based on a suburban voice-output power-distribution function (a statistical figure of merit that describes power variation according to urban versus rural, data versus voice, etc.), the transmit strip current measures 550 mA at maximum output power, and 365 mA at +22 dBm. The Tx noise in the receive band measures −137.0 dBm/Hz at maximum Tx power. If the isolation between the Tx and receiver (Rx) is 50 dB, the calculated Tx noise in the receive path is −187.0 dBm/Hz, which is much lower than the thermal noise. That is, the Tx contributes almost nothing to the total Rx noise. (This calculation has been verified by supporting measurements at maximum and reduced power.)

Typical frequency-domain spectral masks (not shown) reveal signal levels expected at the reference design's antenna port. For +24 dBm antenna output power, the conditions are Icc of 490 mA for transmit only and 535 mA for transmit and receive, a MAX2363 IF DAC setting of 110, and VGC of 2.4 V. For −53 dBm antenna output power (i.e., low Tx output power for which VGC = 1.35 V and the absolute output power is −38 dBm) electrical conditions include Icc of 166 mA (for Tx only), VGC of 1.35 V, IF DAC at 000, PA bias setting of 1, and output-power attenuation set at maximum.