A zero-IF approach enables direct conversion of analog RF signals to a digital baseband format. This dramatically reduces component count and thus handset footprint and cost. Reducing the number of parts also simplifies the supply chain and manufacturing and improves yield.

For example, a traditional quad-mode heterodyne architecture for a CDMA phone uses an IF CDMA SAW filter, an IF FM surface-acoustic-wave (SAW) filter and an IF Global Positioning Satellite (GPS) SAW filter. A quad-mode zero-IF architecture would replace those components. It also would eliminate the need for an IF-to-baseband conversion circuit, an external IF switch, a receive IF voltage-controlled oscillator (VCO), a dual phase-locked-loop synthesizer, a transmit IF VCO, a transmit IF filter, electromagnetic interference shields and a multitude of resistors, capacitors and inductors as well as the board space required for all those components.

Zero IF is not a new concept: chip designers have managed to develop the first zero-IF chip sets for GSM phones. But until recently, technical barriers, particularly in receiver design, have prohibited its practical implementation for high-performance multimode, multiband CDMA phones. A more costly alternative, the heterodyne design, has until now remained the de facto radio topology.

Historically, many of the benefits offered by a zero-IF architecture have been offset by a series of problems related to input IP2, instantaneous dynamic range, signal dynamic range, dc offset and LO leakage. Second order intercept point (IP2) is a key specification for down-conversion mixers and reflects a system's susceptibility to second-order distortion. The higher the value of input IP2, the higher a system's immunity to second-order interference for the targeted baseband signal.

The spec is a key indicator of the receiver's behavior in the presence of a very strong AM jammer relative to the receive signal. When IP2 is too low, an off-channel jammer will interfere with the receiver's operation because second order distortion will lead to an unwanted baseband signal, which will smear the wanted baseband information.

The energy of this signal is proportional to the second-order coefficient of the front-end gain and inversely proportional to input IP2. Accordingly, one of the more challenging tasks for cell phone designers is to ensure that their system delivers a high input IP2 to protect it against strong off-channel AM jammers.

The IF filter handles this task in the traditional heterodyne receiver used in most CDMA cellphone designs. One of the problems a designer of a zero-IF chip set faces is that, since that filter is eliminated, any strong off-channel jammer is fed through to the baseband and eventually saturates the receiver.

Dc offset has also presented a serious design challenge to zero IF. Ideally, only undistorted information appears at output when the RF signal is down-converted to baseband. But the circuit mismatch inherent in both RF and baseband analog circuits typically introduces a dc error, which is then added to the baseband signal. This offset error can be affected by both temperature and time.

The removal of the IF section also poses a problem in gain control. In a CDMA system each mobile receiver must be able to adjust its gain by up to 90 dB via the Automatic Gain Control (AGC) loop. Typically, that control loop is split between the front end and IF sections of the design; in a zero-IF design that function must be implemented elsewhere. The logical choice is to replace it entirely in the front end of the design, but that option is not desirable because it complicates the design on the LNA.

Unlike heterodyne designs, zero IF also places severe restrictions on local oscillator (LO) leakage and reradiation. Since both the LO and RF receive channels operate on the same frequency, any LO reverse leakage from the mixer will travel backwards to the antenna from where it is radiated into the RF passband, causing potential interference to other spectrum users.

Using a silicon germanium (SiGe) BiCMOS process, engineers at Qualcomm developed a zero-IF chip set that avoids the signal degradation by building a baseband processor with high off-channel rejection and a lower noise floor capable of tolerating higher signal levels. Together with LNA gain control, Qualcomm's chip set offers the advantages of a zero-IF architecture while ensuring a high instantaneous and signal dynamic range capability for the receiver.

At the same time, IC designers are now taking advantage of process advances to reduce dc offsets with advanced signal processing. The high-input IP2 and low LO leakage exhibited by the latest zero-IF architectures help minimize the dc offset due to time-varying components of jammers and externally reflected LO. At the same time IC designers have solved the gain control problem in the latest zero-IF architectures by performing a larger portion of the task digitally in the baseband processor.

Complementing improvements in digital IC processing technology, cellphone IC designers are also implementing subtle RF design improvements at both the board and IC levels to minimize LO leakage and reradiation.