Download this article in .PDF format
This file type includes high resolution graphics and schematics when applicable.

The prototype was evaluated with a +4.8-VDC supply voltage and 2.5-GHz nominal test frequency and found to meet the target sub-1-dB NF at the design frequency. The experimental NF was found to be 0.95 dB at midband, with negligible variability in NF (<0.05 dB) for five samples over a 1-GHz span. The predicted NF follows the same trend as the measured NF, but with value that is lower by about 0.3 dB. This discrepancy is probably due to a modelling error, since the device’s NFmin is close to the measurement limit of the noise characterization equipment.

Compact LNA Drives 2.5-GHz Base Stations, Fig. 6

The design capably meets the TMA’s gain and input match requirements. The experimental gain is 18 dB at midband (Fig. 6). The predicted gain is in good agreement with the measured over a 1-GHz span. The experimental input and output return losses are 19 and 17 dB, respectively. In theory, optimum matching should occur at the center frequency of the couplers. The experimental input match for the couplers is optimum at 2.6 GHz, which coincides with the input coupler’s center frequency. The best output match is at 2.3 GHz, however, which does not coincide with prediction (probably due to coupler tolerance). The return-loss amplitude is primarily a function of the coupler’s isolation, although how identical the amplifiers are—as well as discontinuities in the microstrip transmission lines—can contribute to variations in return-loss amplitude.

Compact LNA Drives 2.5-GHz Base Stations, Fig. 7

Because the amplifier behaves like a nonreflective attenuator during shutdown, an LNA bypass switch can be potentially eliminated. When the MMIC shutdown is activated, the circuit exhibits 26-dB attenuation, 26-dB input return loss, and 10.5-dB output return loss at 2.5 GHz (Fig. 7). Matching remains good in shutdown mode because reflections from the unpowered amplifiers are self-cancelled in the couplers. The 2.5-GHz output match is worse than expected because the minimum output return loss shifts to 2.15 GHz, possibly because of coupler tolerance. Because of the good match during shutdown, an LNA bypass switch is not required to prevent aerial or filter detuning.

Compact LNA Drives 2.5-GHz Base Stations, Fig. 8

The fabricated balanced LNA is unconditionally stable. Both modelled and measured stability factors, μ, exceed unity from 50 MHz to 20 GHz (Fig. 8). The accuracy of the simulated μ is poor above 3 GHz since the simple equivalent circuits used to model the amplifier’s passive components do not account for the higher resonances above about 3 GHz. What is remarkable is the potential instability of the constituent amplifiers, as indicated by less than unity μ from 7 to 18 GHz, although the balanced topology’s self-stabilizing promise is validated in this design. It is also remarkable that the frequency range where stability is improved, from 7 to 18 GHz, is well above the couplers’ passbands.

Compact LNA Drives 2.5-GHz Base Stations, Fig. 9

This amplifier design has sufficient linearity to operate reliably in a cellular tower’s noisy RF environment. The 2.5-GHz experimental OIP3 is +38 dBm, about 1-dB lower than the 2.0-GHz peak (Fig. 9). The OIP3 peaks away from 2.5 GHz because the amplifiers’ output networks are tuned for maximum gain. At the expense of reduced gain, it should be possible to improve the 2.5-GHz OIP3 to about +39 dBm by matching for linearity. The OIP3 simulated with load-pull data agrees well with the experimental results, with less than 0.2 dB error at 2.6 GHz. The linearity figure of merit calculated from the ratio of OIP3 to DC power is about 12.4. The midband gain compression point, P1dB, is +21.1 dBm. A high P1dB implies immunity to strong blockers (adjacent channel interference).

Among GaAs MMICs intended for balanced cellular LNA applications, this design has one of the smallest footprints, at 16 mm2. With three functions on the device, the area per function is only 5.3 mm2 compared to previous work7-10 with footprints ranging from 16 to 419 mm2, and even a MMIC footprint of only 16 mm2 at Avago for a single-function device.11 The current design is also the only MMIC cellular LNA with an integral shutdown function.

This design has demonstrated the best performance-to-size ratio among TMA-capable balanced LNAs in the 2-to-3-GHz range. Comparison of different designs can be facilitated by a figure of merit (FOM), such as LNAFOM = OIP3 9mW)/PDC (mW) × NF (dB).17 The current design has an LNAFOM of 13.2. In comparison, designs with large couplers8,9,12 have the highest LNAFOMs but also occupy large PCB areas. The current design combines high performance and compactness—a high value of LNAFOM per unit area.

The proposed balanced LNA design successfully marries high performance and small size, usually conflicting TMA requirements. It achieves substantial reduction in component count by integrating dual amplifiers, bias regulators, and shutdown circuits in a MMIC. When the highly integrated MMIC is coupled with miniature hybrid couplers, the balanced LNA’s PCB size can be significantly reduced. This MMIC’s low-noise performance allows input coupler loss to be traded off for size reduction. A bonus feature of this balanced LNA is the good impedance match during shutdown, which may eliminate the need for switches to bypass the LNA during shutdown. This new design may overcome traditional barriers to the use of balanced amplifier topologies in TMAs.


The author wishes to thank H.A. Zulfa, W.H. Chiok, and R. Chuah for the MMIC design and characterization, M.D. Suhaiza and S. Punithevati for assembling the prototypes, and S.A. Asrul and the management of Avago Technologies for approving the publication of this work. Anaren Communications (Suzhou) provided the couplers at no charge.

Chin-Leong Lim, Engineer

Avago Technologies, 11900 Bayan Lepas, Penang, Malaysia; (604) 610-2525.

Download this article in .PDF format
This file type includes high resolution graphics and schematics when applicable.