1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
A Low Power Highly Linear
CMOS Current Mode
Up-Conversion Mixer
Chunhua Wang, Xiangyue Shi, Sichun Du, Qiuzhen Wan, Jingru Sun
Abstract – In order to decrease the power dissipation and supply voltage, a current mode up-conversion
mixer for RF transmitter is proposed in this paper. Different from the traditional Gilbert structure, the mixer
replaces the transconductance stage with current mirrors. This method can bring enough conversion gain
and high linearity. The way of current bleeding is also adopted here to improve the performance. The
simulation is made by Cadence Spectre in CHRT 0.18mm CMOS mixed signal/RF process. The simulation
results indicate that the mixer achieves 24.5dBm third-order input intercept point (IIP3), 5dB conversion gain
and 14.9dB SSB noise figure while its power dissipation is only 3.8mW under 1.2 V supply voltage. It is
especially suitable for low power mobile communication.
Index Terms – CMOS, up-conversion mixer, current mode, current bleeding, Volterra series
1 Introduction
With the development of the mobile communication, the
demand for low-cost, small-size and low-power equipment
increases rapidly. In order to meet the need, the CMOS single-
chip communication system is proposed, which contains both
the RFand digital modules[1]. However, this causes some new
problems such as different supply voltages of the RF and
digital circuits, thermal dissipation and intrinsic defects of the
CMOS devices. More and more researchers begin to study on
these issues.
Mixer is a very important part of the wireless communica-
tion system[2]. It performs frequency down- conversion in a
receiver and frequency up-conversion in a transmitter. The
design of mixer should make a trade-off among the linearity,
conversion gain, noise figure and power dissipation. In low
power field, the power dissipation and linearity are very
important, because they respectively decide the life of the
battery and the mixers immunity to the interference.
Usually the mixer is based on the Gilbert structure which
requests a stack of three transistors. Although the Gilbert
mixer has the advantages of good port-to-port isolation and
high conversion gain, its high supply voltage and power
dissipation prevent its application in low power field. To
improve the performance of the Gilbert mixer, several
improved structures have been proposed: a folded switching
mixer using current reuse[3], it can reduce the power
dissipation and supply voltage, but the linearity will deterio-
rate because of more transistors; a modified mixer with the
source degeneration inductors[4], this method can improve
the linearity of the mixer, but the conversion gain may
decrease as well.
Different from the voltage mode circuit, the current mode
circuit has both low impedance and low voltage at internal
nodes. So it usually has higher linearity and lower power
dissipation[5]. And now it attracts more and more attention in
low power field. A novel CMOS current mode down-
conversion mixer was proposed in 2002[6]. It has a high
linearity and low power dissipation at 900MHz work fre-
quency, but the conversion gain is less than zero. Recently a
CMOS current mode up-conversion mixer has been present-
ed[5]. It has high conversion gain and low power dissipation,
but the structure is too complex.
A low power highly linear CMOS current mode up-
conversion mixer is proposed in this paper. Both current
mirrors and the method of current bleeding are adopted here,
and the structure of the mixer is simple. The simulation is
made by Cadence Spectre in CHRT 0.18mm CMOS mixed
signal/RF process. Compared with the previous CMOS
voltage-mode mixers[7 –9] and current-mode mixers[5,6,10]
at the similar frequency, the mixer simultaneously has lower
power dissipation of 3.8mW and much higher linearity of
24.5dbm under the 1.2 V supply voltage.
2 Circuit analysis
The circuit is shown in Fig.1, which is based on Gilbert double-
balance mixer. The transistors M1, M2, M3 and M4 comprise
the input current mirrors. By this way, the mixer can directly
accept the current waveform input signal without extra V-I
conversion stage. It can also have high conversion gain with
the help of the current mirrors amplification. The inductor
L1, L2 can improve the linearity of the current mirrors. The
reason is given by Volterra-series in part B of this section. The
transistors M5, M6, M7, and M8 work as switches. Meanwhile
the current sources I1, I2 can improve the performance of the
switch transistors by adjusting the bias current. By the two
optimizing methods of adding inductors and current sources,
the total IIP3 of the mixer is as high as 24.5dbm. The LC tanks
C1 L3 C2 and L4 are helpful for increasing the conversion
gain.
2.1 Up-conversion principle
The transistors M5-M8 are driven by LO signals, and they
switch on and off by turns. If the LO signal is large enough,
transistors M5-M8 can be approximated to be ideal switches.
Fig.1. the schematic of the proposed mixer
Frequenz
64 (2010)
11– 12
1