Chip Implementation and Measurement Results

This LNA is designed with 1.0V supply voltage and fabricated in TSMC 0.13μm RF CMOS process. Figure 4-17 shows the micrograph of the LNA under test. The core circuit without pads occupies 0.49 mm² chip area. RF signals are measured with on-wafer probing while all DC voltage is provided via bonding wires. In the test cases, small signal S-parameters are obtained with Agilent 8364B network analyzer, and noise figure from Agilent 8974A noise figure analyzer with cable loss compensation. Two tone tests are performed to test the linearity

Figure 4-17. Micrograph of the fabricated reconfigurable LNA under test.

performances of IIP2 and IIP3. The accurate power level at LNA input is calibrated by use of a power meter.

Performances of this LNA are measured in three configured performance modes in five frequency bands, carrying out 15 test cases. The five frequency bands include the designed upper and lower tuning limits and three MB-UWB mode-1 bands, listed as 2.4-, 3.43-, 3.96-, 4.49-, and 5.4-GHz. The performance modes include one 20-dB high-gain mode (HG) and two 10-dB low-gain modes (LG1, and LG2), representing the three operation modes shown in Figure 4-2, in which the LG1 mode is for better linearity whereas the LG2 mode toward the lowest power consumption.

The S-parameter test results are shown in Figure 4-18 and 4-19. The broadband input/output reflection performances for the 15 configurations, the 5 frequency bands tuning with 3 gain modes switching, are successfully kept consistent. The power gains at the HG mode are in the range from 22- to 25-dB whereas at the LG1 and the LG2 modes from 10- to 13-dB.

Gains at different frequency bands for each gain configuration are within 3-dB variation and approximately consistent. Noise figures are lower than 3.1-dB for HG mode and lower than

2 3 4 5 6

1 7

-20 -15 -10 -5

-25 0

Figure 4-18. Measured input and output reflection ratio (S11 & S22) of the LNA in all listed configurations under test.

5-dB for LG1 and LG2 modes. The tested curves for the HG mode are shown in Figure 4-20 as representative. The capability of continuous frequency tuning is verified as shown in Figure 4-21, in which L1 to L4 represent the different switched inductance values.

The linearity of this LNA is characterized by the two-tone test for in-band IIP3 and out-of-band IIP2 and IIP3. For the IIP2 test two CW interferer tones are input at the frequencies of 2.599- and 2.601-GHz, while the LNA is configured to operate at 5.2 GHz where the IM2

lands. The output IM2 is measured and referred to the input by dividing the measured power

2 3 4 5 6

Figure 4-19. Measured power gain of LNA under typical bias condition.

1 2 3 4 5 6 7

Noise Figure @ HG mode (dB)

Frequency (GHz)

Figure 4-20. Measured noise figure for the five frequency configurations in HG-mode. Most of them are lower than 3-dB.

gain at 5.2 GHz to calculate the corresponding IIP2. Similar procedure applies to characterize the out-of-band IIP3 with the frequency set of {f0, f1, f2}={2.4, 3.8, 5.2} and {5.2, 3.8, 2.4}, in which the f0 is the configured operation frequency of the LNA. The IIP2 measurement results are shown in Figure 4-22, in which the gain desensitization at 5.2 GHz due to the two 2.6 GHz interferers is also included. The measured IIP2 and in-band/out-of-band IIP3 all range from -21 to -5 dBm, depending on which gain mode is chosen. The IIP2 control for the three gain modes is consistent with the design in Figure 4-7.

2 3 4 5 6

Figure 4-21. Continuous frequency tuning with coarse inductor switching and fine varactor tuning.

-35 Input Referred IM2 Distortion @ 5.2GHz (dBm)

Input Power Level of 2.6GHz Tones (dBm)

Gain Desensitization @ 5.2GHz (dB)

Figure 4-22. Measurement result of the 2nd order inter-modulation distortion and the gain desensitization at 5.2GHz because of the 2.6GHz interferers.

correlated to the band-selective character of the LNA. This is because the linearity bottleneck of this LNA is in stage before the LC band-selective filter. However the band-selective filtering of LNA should be still favorable to alleviate the stringent linearity requirement of the succeeding stages.

The measured performance is summarized in Table 4-3. The performance corner matrices are shown in Figure 4-23 and Figure 4-24, using IIP2 and in-band IIP3 as the linearity indicator respectively. Performances at the five representative frequencies with the same performance

6 5 4 3 2 1

Figure 4-23. Performance corner matrix with IIP2 as the linearity indicator.

6 5 4 3 2 1

Figure 4-24. Performance corner matrix with IIP3 as the linearity indicator.

configuration are grouped. With the linearity control on IIP2 the Figure 4-23 shows a nearly consistent result as illustrated in Figure 4-1. Therefore the performance reconfiguration conducted by switching transistor with bias control is verified. Finally, Table 4-4 compares works of multistandard LNAs published to date.

4.6 Summary

A 2.4- to 5.4-GHz wide-tuning-range performance- reconfigurable LNA is demonstrated.

The broadband input stage is verified to be adequate in providing steady input matching and noise performance. The performance reconfiguration on gain, linearity, and power consumption is achieved. By use of the proposed inductance switching configuration, the multi-tapped switching inductor can be well designed to provide wide tuning range with good performance consistency. The proposed switching configuration is advantageous with the quality factor boosting but suffers from limited frequency tuning ratio. This limitation has been identified and the design reference is given in (4-21). As a result, this prototype has proven the high

TABLE4-3

PERFORMANCE SUMMARY OF RECONFIGURABLE LNA

Frequency (GHz) 2.40 3.43 3.96 4.49 5.40

Performance Mode HG LG1 LG2 HG LG1 LG2 HG LG1 LG2 HG LG1 LG2 HG LG1 LG2

Technology TSMC 0.13μm RF CMOS process with UTM

DC Power of 2nd

Stage HG: 2 mW; LG1: 1 mW; LG2: 0.5 mW

Total DC Power

(mW) HG: 4.6 mW; LG1: 3.6 mW; LG2: 3.1 mW (1.0V supply voltage, buffer stage excluded)

performance flexibility and wide-range tuning of LNA within the below-average low power

in which two poles and one zero can be found on the positive ω-axis. The two poles are located at frequencies as solved in (4-20). We can determine which pole frequency is our case by testing their continuity to ωmin as LX,MIN is approaching LX,MAX, by setting L1 → LX,MAX and L2 → 0. In this test the higher pole frequency can be found approaching infinity whereas the lower pole meets the ωmin. Therefore the minus sign is applied in (4-20).

There are two singular cases not applicable to the above discussion, k = 0 and k = 1. In the

TABLE4-4

COMPARISON OF LNAS FOR MULTISTANDARD APPLICATIONS Ref.–Year Technology Frequency

a In frequency response, BB: broadband, MB: concurrent multiband, TSB: tunable/switchable single band.

b Performance of entire receiver front-end including LNA and quadrature mixer.

c Performance in high gain mode as representation. ^d Continuous frequency tuning.