Chapter 2 A Broadband DLNA in 0.15-μm GaAs pHEMT Process for Radio
2.2 Circuit Design
2.2.5 The Proposed Broadband Active Balun
Baluns are key components in balanced transceiver blocks such as differential amplifiers, mixers, and frequency multipliers. Also, a balun may be used for differential connections close to antenna elements. There are various types of active baluns. A single device balun [32] and a common-gate common-source (CGCS) balun [33] were proposed for narrowband applications due to the limited gain in common-gate cells. A cascode-cascade configuration [34] and a modified CGCS balun [35] were proposed to solve the input matching problem of the CGCS balun. However, the output imbalances are sensitive to the transistor size variations due to process uncertainty. Active baluns using a differential pair were reported in [36] [37], but it is not suitable for high speed data applications due to the higher gain variation. Active baluns using distributed techniques demonstrate broad bandwidth with small gain variation [38] [39], but occupy large area, have relatively long group delay and the bandwidth is limited by DC-decoupling capacitors. The combining matrix baluns feature the broadband behavior of the traveling wave amplifier and the matrix amplifier. The phase rotation of the matrix balun is achieved by utilizing the fact that the phase difference is 180°
between two input ports due to common-source transistors. The matrix baluns have the potential of offering good wideband performance with small chip area [40]. For broadband radio astronomy receiver system, the matrix balun is adopted to solve the disadvantages of passive balun such as high loss, enormous imbalance and limited bandwidth. The proposed broadband matrix balun utilizes high impedance transmission line elements between the transistors to achieve high small-signal gain over decade bandwidth. The small area was achieved by using meandering of high impedance transmission lines.
VD
Vbalun
RFout
RFin2
Vctrl
RFin1
M2 M2
M1 M1
Cblock Cblock
R1
R2
R3
VD
Cblock
Cblock
Cblock
Fig. 2.27 The circuit schematic of the broadband matrix balun.
Fig. 2.28 Definition of the chain matrices used for calculating the S-parameters of the matrix balun [40].
Fig. 2.27 shows the circuit schematic of the proposed broadband matrix balun. The proposed matrix balun can be treated as 6-port with three open ports terminated by resisters R1, R2, and R3, other three ports are denoted as RFin1, RFin2 and RFout. Theoretically, the relation between the input port and output port can be described by a six-port chain matrix representation which then transformed to admittance parameters, reduced to three port and finally transformed to S-parameter matrix fully describing the matrix balun. The chain matrices T are defined as
[ ]
and multiplied to find the total 6-port chain matrix, Tb of matrix balun. The partial six-port chain matrices defined by the sub-blocks as shown in Fig. 2.28 expressed as3
1
with gm1 and gm2 being the transconductance of the individual transistors M1 and M2. The chain matrix of entire matrix balun consisting of the termination resistors, inductance and a transistor cell can be defined as
2 2 2 2
The six-port chain matrix Tb may be transformed to admittance form Yb
The three-port scattering matrix describing the matrix balun can now be derived through the transformation
(
0)(
0)
1b b b
S = U Y Z U Y Z− × + × − (2.28) where U is the identity matrix. However, it is not easy to calculate the full analytical expressions of three-port scattering matrix over frequency precisely for analysis purpose.
Therefore, to design of matrix balun, the chain matrix need to simplify first. Since target frequency of this work is low (4.6 to 8.5 GHz), the frequency is assumed to be zero (ω = 0), and it turns out that this initial assumption is suitable for this design from the final simulation results considering the frequency dependence. The S-parameters can be expressed as
1 2 1 2 3 02
where Z0 denoted as the system impedance. RFin1, RFin2 and RFout are represented with port1, port2 and port3. By dividing the (2.29) and (2.30), the gain ratio of two input ports can be expressed as
Observed from (2.31), the designed parameters gm1 and R1 are influence the imbalance of two input ports. However, it need to choosing a gm1 corresponding to zero gain and phase imbalance first due to the rare choice of device size comparing to the termination resistor R1.
As shown in the Figs. 2.29 and 2.30, by sweeping the transistor size of M1, it is observed that the gain imbalance is affected by gm1 and phase imbalance is affected by the parasitic capacitance (CGS) of M1. The transistor size of M1 are both 2-finger with total 50-µm gate width with Vd = 1 V, Vctrl = -1 V. Besides, the matrix balun can be regarded as a matrix amplifier. The input signals from the two input ports will equally change the phase and the amplitude due to two common-source transistors M2 in the matrix balun. This means that the gate and drain bias voltage (Vbalun and VD) and the size of M2 transistors can be selected more flexibility. The size and the bias condition of M2 are chosen for flat frequency response and minimum power consumption. The final size of M2 are both 2-finger with total 50-µm gate width with Vd = 1V, Vbalun= -1V. Figs. 2.31 and 2.32 show the simulated gain and phase imbalance of proposed matrix balun at different termination resistance R1. The open port termination resistors of R1 = 25, R2 = 60 Ω, and R3 = 330 Ω are adopted. As shown in the Figs 2.33 and 2.34, in the desired frequency from 4.6 to 8.5 GHz, the gain and phase imbalance are barely affected by stretching the transmission lines between transistors. The blocking capacitors are 1 pF.
Fig. 2.35 shows the simulated S-parameters of the complete matrix balun. The 3-dB bandwidth covers from 3.3 to 17.4 GHz. Higher gain or wider bandwidth may be reached by adding more stages or decreasing the transistors size to reduce the parasitic capacitances. The RFin1 to RFin2 isolation is shown in Fig. 2.36. In this design of matrix balun, there is limited isolation from RFin1 to RFin2, and it means that the two input signal will interfere with each other which in some applications can be a drawback. However, in this work, isolation is greater than 30 dB after cascading LNAs in front of the matrix balun.
2 4 6 8 10 12 14 16 18 20
Fig. 2.29 The simulated gain imbalance of proposed matrix balun at different device size of M2.
Fig. 2.30 The simulated phase imbalance of proposed matrix balun at different device size of M2.
2 4 6 8 10 12 14 16 18 20
Fig. 2.31 The simulated gain imbalance of proposed matrix balun at different termination resistors R1.
Fig. 2.32 The simulated phase imbalance of proposed matrix balun at different termination resistors R1.
2 4 6 8 10 12 14 16 18 20
Fig. 2.33 The simulated gain imbalance of proposed matrix balun at different length of transmission line between M1 transistors.
2 4 6 8 10 12 14 16 18 20
Fig. 2.34 The simulated gain imbalance of proposed matrix balun at different length of transmission line between M1 transistors.
2 4 6 8 10 12 14 16 18 20 -20
-10 0 10 20 30
S11 (Sim.) S22 (Sim.) S21 (Sim.)
S- Pa ra m et er s ( dB )
Frequency (GHz)
Fig. 2.35 The simulated S-parameters of the proposed broadband matrix balun.
2 4 6 8 10 12 14 16 18 20
-10 -5 0 5 10
Is ol ati on (d B)
Frequency (GHz)
Fig. 2.36 The simulated isolation of the proposed broadband matrix balun.
In order to demonstrate the advantage of proposed matrix balun, the conventional passive lumped LC balun is used. A higher-order lumped balun consisting of 90°
3-element T-type high-pass phase shifter and low-pass phase shifter is a good choice [41].
This structure is regarded as combinations of different types of quarter-wave transformer.
Fig. 2.37 shows the simulated gain and phase imbalance of matrix balun and passive LC balun. Within the target bandwidth from 4.6 to 8.5 GHz, passive L-C balun shows enormous phase (12° to 49°) and gain imbalance (0.1 to 2.8 dB). Thus, the proposed matrix balun exhibits the potential for application in the next-generation radio telescope of radio astronomical receiver systems.
2 4 6 8 10 12 14 16 18 20
10
Gain imblance (L-C balun)Gain imblance (matrix balun) Phase imblance (L-C balun) Phase imblance (matrix balun)
Ga in im bl anc e (dB )
Fig. 2.37 The simulated gain and phase imbalance of matrix balun and passive L-C balun.