Experimental Results

Before measuring the receiver, the loss of passive components should be measured in the first place. The loss of external input matching network, output transformers and DC blocking capacitors are measured by network analyzer. The loss of the input matching network is about 1.7 dB in the desired band. The loss of the output transformer and DC blocking capacitors of each channel is about 0.2 dB in the desired band. The cable loss is about 2.7 dB. Besides, the parasitic resistances of metal lines on the bonding board are so little that the loss of them is neglected in measurement.

Including the input matching network, composed of external capacitors and microstrip transmission lines, the receiver performs an S11 of lower than -10 dB in a frequency range of 5.07 GHz to 5.3 GHz, exhibited in Fig. 53. It is found by several tested chips that the optimum input matching can be achieved by moving the capacitor to a certain location on the microstrip transmission line. The equivalent inductance of bond-wire is estimated with an approximate value of 2 nH.

Fig. 53 S11 of the receiver.

The spectrum of the QVCO (quadrature VCO) can be analyzed by the QVCO buffer. The measured spectrum is presented in Fig. 54, where the power is only -48.67 dBm under a supply voltage of 0.7 V because the output terminals of the QVCO buffer are not impedance matching. The oscillation frequency can be tuned from 3.617 GHz to 3.797 GHz under a tuning voltage of 0 V to 1 V, and the KVCO is 180 MHz/V. The oscillation frequency shifts downward by 1.5 GHz compared with post-simulation results due to parasitic inductances of the interconnections connected to the inductors and the core of the QVCO. The parasitic inductances are not extracted by Calibre, a verification tool used to extract parasitics. In order to measure the receiver in the desired band, laser-cut technique is adopted to cut off the connection between the buffer and the core of the QVCO and a part of varactors. Thus, the capacitances at each output terminals of the QVCO are reduced so that the oscillation frequency of the QVCO shifts upward. After coping with several tested chips with this procedure, a resultant oscillation frequency tuned from 4.564 GHz to 4.684 GHz under a tuning voltage of 0 V to 1 V is obtained by measuring the LO leakage at the IF output, and the KVCO is reduced to 120 MHz/V. Fig. 55 shows the microphotograph of the QVCO after laser-cut, and Fig. 56 plots the resultant tuning range compared to the original measurement.

In conclusion, the resultant frequency range is applied to measure the performance of the tested receiver.

Ref Lvl

Center 5 GHz Span 10 GHz

-110

Fig. 55 Microphotograph of the QVCO after laser-cut.

Fig. 56 Measured tuning range of the QVCO.

Two-port network of S-parameter analysis cannot be applied for gain estimation because the frequencies of input and output terminals are different. Spectrum observation is a substitutive way. Fig. 57 presents an IF output spectrum of the receiver while a RF signal with a frequency of 4.65 GHz and a power of -65 dBm is introduced and the oscillation frequency of the QVCO is 4.65 GHz. Compensating back with the loss of cable and external components, the receiver performs a conversion gain of 12.6 dB under a supply voltage of 0.65 V. The measured conversion gain is too low because the frequency of RF signal is not located at the desired band of the LNA, which is a gain stage in chief. To measure the conversion gain at the lower frequency, a RF signal with larger power is introduced to the receiver. Fig. 58 displays the measured conversion gain of the receiver by sweeping the frequency of RF signal with an LO frequency of 4.65 GHz. The corner frequencies at 150 kHz and 10 MHz are also observed.

Ref Lvl

Center 10 MHz 2 MHz/ Span 20 MHz

RBW 200 kHz

Fig. 58 Measured conversion gain of the receiver.

The measurement setup for noise figure is shown in Fig. 59. The signal analyzer with product number FSIQ26 is made by Rhode & Schwarz. The noise source with product

number 346B, having a SNR of 15.2 dB, is made by HP. Software for noise measurement named FS-K3, having a start frequency of 100 kHz, is utilized to calculate the noise figure.

Fig. 60 presents the noise performance of the tested receiver. Selecting the noise bandwidth from 150 kHz to 10 MHz, the SSB noise figure is 24 dB after calculation. Similarity, the measured noise performance is terrible since the frequency of RF signal is not located at the desired band of the LNA, which dominates total noise performance. Fig. 61 plots the IF output power related to RF input power, and the 1-dB compression point is obtained with a value of -24 dBm. The measured result is better than post-simulation due to the measured low conversion gain of the receiver. Also, DC offset is measured, and the results are presented in Fig. 62. After measuring several tested chips, the DC offset voltage at the differential outputs is less than 3 mV and 25 mV with an injected input power of -50 dBm and -30 dBm, respectively.

The values of fine-tuned supply voltages, gate biases and bias resistors are listed and compared with post-simulation in Table XI. The power consumption of the tested receiver is 8.14 mW due to a raise of the supply voltages of each sub-circuit. Table XII lists a summary of the tested receiver, including a comparison between post-simulation and measurement. The measured performance differs from the post-simulation and thus it is discussed in detail in the next section.

Fig. 59 Measurement setup for noise figure.

Fig. 60 Measured noise figure of the receiver.

Fig. 61 Measured 1-dB compression point of the receiver.

Fig. 62 Measured DC offset of the receiver.

Table XI Comparison on Bias Status between Post-Simulation and Measurement Supply voltage Post-simulation Measurement

LNA and downconverter 0.6 V 0.65 V

Quadrature VCO 0.6 V 0.7 V

Gate bias Post-simulation Measurement

Mn1, Mn2 0.6 V 0.63 V

Mai1-Mai4, Maq1-Maq4 0.6 V 0.65 V Mbi1-Mbi4, Mbq1-Mbq4 0.6 V 0.65 V

Bulk bias Post-simulation Measurement Mbi1-Mbi4, Mbq1-Mbq4 0.35 V 0.38 V

Mi5-Mi7, Mq5-Mq7 0.3 V 0.33 V Bias resistor Post-simulation Measurement

Rci, Rcq 2 kΩ 1.8 kΩ

Rbi1, Rbi2 582 Ω

Rbq1, Rbq2 600 Ω

554 Ω

Table XII Summary of the Tested Receiver

Design target Post-simulation Measurement

Technology TSMC 0.18-μm CMOS 1P6M

Supply voltage 0.6 V 0.6 V 0.65 V/0.7 V Power consumption < 10 mW 4.4 mW 8.14 mW

QVCO tuning range 5.15-5.35 GHz 5.14-5.36 GHz 4.56-4.68 GHz (after laser-cut)