For the PDF version of this article, click here.

Demand for mixer linearity performance is high in modern cellular base station transmitter design. In the past, passive diode-ring type mixers have been the devices of choice to meet stringent requirements. Today, a new generation of active mixers offers significant advantages that can save cost, power and space, while solving some of the most difficult technical problems associated with high-performance transmitter design.

The primary goal for RF designers is to maximize the signal-to-noise (SNR) performance of a transmitter. This requires transmitting as much RF signal power as permitted, while pushing the noise floor as low as possible. At maximum output, the level is capped by the amount of distortion introduced by undesirable output spectral components. Thus, there is a fine balance between achieving high SNR and spurious-free dynamic range (SFDR) while attaining maximum output power.

A new generation of high-performance active mixers like LT5521 offers a combination of high linearity, low noise, easy LO drive, and little or no conversion loss, resulting in a cost-effective, high-performance solution.

Choosing the right operating point of the mixer is essential for ensuring the optimum SFDR performance, while attaining maximum output power. Referring to Figure 1, the LT5521's minimum noise figure (NF), 11.9 dB, occurs when the local oscillator (LO) power is at 0 dBm. At this LO power level, the LT5521's IIP3 (24.1 dBm) and conversion gain (-0.5 dB) are also near their respective maxima (Figure 2). Thus, all three parameters are working together to enhance the dynamic range.

With the IIP3 and noise figure known, the SFDR of the mixer circuit can be computed using the following equations:

SFDR = 2/3 (IIP3 - noise floor) (1)

where

noise floor = NF - 174 dBm + 10 Log BW (2)

For WCDMA or TD-SCDMA-type signals, the BW is 3.84 MHz. Substituting the mixer IIP3 = 24.1 dBm and NF = 11.9 dB, at a LO level of 0 dBm at 1.95 GHz, thus,

noise floor = 11.9 dB - 174 dBm + 10 Log (3.84 MHz) = -96.3 dBm

With this set of operating parameters, a circuit's spurious-free dynamic range can be

SFDR = 2/3 [24.1 dBm - (-96.3 dBm)] = 80.3 dB

With the objective of maximizing the distortion-free output signal, higher conversion gain is a real advantage (Figure 2). This makes more signal available at the output without extra gain stages. Active mixers have a conversion loss of only 0.5 dB. By comparison, passive mixers typically offer much higher loss of between 6 dB and 10 dB. Hence, with an active mixer there is a net signal improvement of 6 dB to 8 dB.

Another important advantage is its inherently low LO drive and superior LO suppression. The new generation of high linearity active mixer requires 0 dBm (or less) signal to drive its LO port. By contrast, a similarly high IIP3 passive mixer needs at least +17 dBm LO signal. A high-power +17 dBm signal on a PC board can be a strong source of undesirable radiation. At 1 GHz to2 GHz frequencies, small PC board parasitic elements can couple enough LO signal to affect other sensitive circuits in the system, needing RF shields. However, designing an effective RF shield may require several PC board spins, and could lengthen a program's development cycle.

Additionally, an active mixer's much lower LO drive simplifies the LO driver circuitry, reducing external components by eliminating one or two power amplifier stages. Thus, power consumption is much reduced, as well as cost.

Also, an active mixer is less sensitive to LO level variations. In the case of passive mixers, 2 dB to 3 dB of LO power change can significantly degrade their linearity performance. Designing a high-power LO source while holding a tight LO level tolerance is difficult, particularly in a mass production environment. Active mixers like the LT5521 can tolerate a wider operating LO power level range without significantly degrading performance.

The inherently superior port-to-port isolation of active mixers helps to greatly reduce the LO leakage to the transmitter output. Typical passive mixers offer about 30 dB isolation. But with their LO signal at +17 dBm, the output will have LO output leakage in the range of -13 dBm, an unacceptably high level that will require extensive filtering to suppress. Multiple stages of filters may be necessary to control the LO leakage, and the frequency plan of the transmitter may be affected. Depending on the frequency offset, two stages of SAW filters achieve about 40 dB of attenuation. The active mixers' advantage starts with its low LO drive power. Combined with superior LO isolation of more than 40 dB, it results in an LO leakage level of about -40 dBm to -45 dBm at the RF output. This is 30 dB lower than a passive mixer, and greatly reduces the filtering requirements that would otherwise be needed.

Having fewer filter stages not only saves cost, but also improves the signal quality. That is because sharp roll-off filters have appreciable in-band ripples. When cascaded, the ripples from each filter stage can increase in amplitude to an extent that exceeds the design limit, distorting the modulated waveform. Moreover, each filter stage contributes an appreciable insertion loss. Thus, an active mixer's relaxed output filter requirement significantly reduces the solution cost, improves signal quality and higher signal levels available at the transmitter output.

James Wong is product marketing manager for high-frequency products at Linear Technology Corp., Milpitas, Calif.

RSS    Save to Del.icio.us  Digg This