RAHUL SHARMA et al.

: A COMPARATIVE STUDY OF DIFFERENT TYPES OF MIXER TOPOLOGIES

DOI: 10.21917/ijme.2016.0031

A COMPARATIVE STUDY OF DIFFERENT TYPES OF MIXER TOPOLOGIES

Rahul Sharma1, Abhay Chaturvedi2 and Manish Kumar3
Department of Electronics and Communication Engineering, Institute of Engineering and Technology, GLA University, India
E-mail: [email protected], [email protected] and [email protected]

Abstract On the basis of performance and structure, mixer can be single

In this paper, a comparative study of different types of mixer topologies balanced mixer or double balanced mixer [14]. An active mixer
is presented. Gilbert cell is widely used as core of the mixer because it provide conversion gain and requires low LO power but it has
provides high conversion gain, good port-to-port isolation and low poor linearity, whereas, passive mixer provides high linearity and
even-order distortion. It is found that the linearity of mixer is very good increased dynamic range but requires high LO drive [23]. Single
for Multi-Tanh technique by incorporating multiple differential trans- balanced mixer requires differential form of local oscillator (LO)
conductance stage but it reaches to very low conversion gain whereas,
signal and single ended form of RF signal as shown in Fig.2 [12].
use of current bleeding technique increase linearity and conversion
VDD
gain of the mixer by adding current source to increase the bias current
at the expense of power consumption. A very low value of noise figure
can be achieved with the switched biasing technique by replacing
current source with parallel connected nMOS transistors but due to use
of the transistor in place of tail current source, linearity is degraded
V_IF
and more power is consumed. Folded Cascode Technique is used to
reduce DC supply voltage by folding the LO switching stage with pMOS
transistors in switching stage but it degrades the noise figure. Bulk-
M2 M3
driven technique can be employed to lower down the power LO+
consumption by providing the switching action via the gate of LO (RF)
and amplification by the bulk of LO (RF) transistors, however it
reduces the linearity. High linearity is obtained by using CCPD (Cross LO-

coupled post distortion) technique by cancelling of third order

M1
derivatives but it decreases the conversion gain and consume more RFin
power due to increase in the number of auxiliary transistors. MGTR
enables to achieve high linearity by incorporating auxiliary transistor
but it decreases the overall conversion gain and increases noise figure
of the mixer. So it is observed that there is a trade-off among the Fig.2. Single balanced mixer [12]
performance metrics, i.e., conversion gain, noise figure linearity, and
power consumption of the mixer. Double balanced mixer operates with differential LO as well
as differential RF signal as shown in Fig.3. [20]. Single balanced
Keywords: mixer is simple in design and provide moderate gain and low
Multi-Tanh, Current Bleeding Technique, Switched Biasing, Folded noise figure but has poor port to port isolation between LO to RF
Cascode, Bulk-Driven, CCPD, MGTR and LO to IF and possesses low third order input intercept point
(IIP3). Double balanced mixer provide high port to port isolation
and has output spurious product rejection capability [7].
1. INTRODUCTION
RF mixer is an indispensable part of modern wireless
communication system. Mixer is a three port active or passive
device, designed to provide down-converted and up-converted
version of input frequency. When the desired frequency at the
output is lower than the input frequency it is called as down-
conversion and if the signal at the output is at higher frequency than
the input signal, it is known as up-conversion as shown in Fig.1.
Mixer

RF IF = RF  LO

LO
Fig.3. Schematic of conventional double balanced CMOS
Fig.1. Mixer Gilbert cell mixer [12]

182
ISSN: 2395-1680 (ONLINE) ICTACT JOURNAL ON MICROELECTRONICS, APRIL 2016, VOLUME: 02, ISSUE: 01

In mixer design, double-balanced Gilbert cell mixer is Linearity can also be measured through 1-dB compression as
preferred as a core of many mixers due to benefits, such as higher given by Eq.(3) [20]:
linearity, low even order distortion and good port to port isolation
1
[8]. The basic double-balanced CMOS Gilbert cell mixer consists A1dB  0.145 . (3)
of three stacked stages i.e., trans-conductance stage, LO switching 3
stage and output load stage. In this paper, the comparative study of different topologies of
There exists trade-offs between the performance parameters mixer such as Multi-Tanh [1], Current bleeding [8], Switched
of mixer like conversion gain, noise figure, linearity, power biasing [2], folded cascode [4], bulk driven [5], CCPD [6] and
consumption and so on as shown in Fig.4 as RF design hexagon. MGTR techniques [7] are discussed with their merits and
[23]. demerits. These techniques can be used to improve mixer
Noise Power performance parameters in terms linearity, noise figure, supply
voltage, power consumption and conversion gain.

Linearity 2. LITERATURE SURVEY

Frequency
This section summarizes the techniques that are used to
enhance the performance of Gilbert cell mixer.
Supply
Gain Barrie Gilbert proposed “Multi-Tanh” technique [1] in order
Voltage
to extend the voltage capacity of a trans-conductance (gm) of
Fig.4. Tradeoffs in RFIC design [23] mixer, by incorporating multiple differential pairs operating in
parallel as shown in Fig.6 [28],
The high linearity can be achieved at the expense of low
conversion gain and high power consumption, while high
conversion gain mixers suffer from increase in noise figure. The
highly linear mixer can increase the dynamic range of the mixer.
The linearity of mixer affects overall linearity of the system which
can be expressed by Friss equation [20]:
1 1 G GG
  1  1 2 . (1)
IIP3 IIP3,1 IIP3,2 IIP3,3
In Eq.(1), Gi, i = 1,2,3,…,n are the power gains of each
cascode stage, therefore, the designing of front-end transceivers
require a highly linear mixer. Therefore, designing highly linear
mixers has attracted a considerable amount of attention [14].
Linearity of mixer is measured by third order input intercept
point (IIIP3) as well as 1 dB compression point as shown in Fig.5
[25]. Fig.6. Schematic of Multi-Tanh technique [28]

In this technique, a constant gm along a large range of input

voltage is achieved by replacing single differential pair stage with
a multiple differential pair stages. The individual gm functions of
each differential stage is distributed along the required larger
input range, so that the resultant constant trans-conductance (gm)
can be achieved over a wider range of input voltage [25].

Fig.5. output power versus input power [25]

Third order input intercept point (IIP3) can be estimated by

Eq.(2) [20]: Fig.7. n-Multi - Tanh differential pairs [25]
4 1
AIP3  (2) Multi-Tanh technique allows the mixer to handle larger input
3 3 voltage swings, as a result, performance of mixer can be enhanced
where, 1 and 3 are the coefficients of non-linearity. in terms of wider range of constant trans-conductance with better

183
 RAHUL SHARMA et al.: A COMPARATIVE STUDY OF DIFFERENT TYPES OF MIXER TOPOLOGIES

linearity and low-distortion [1]. But power consumption is Whereas the voltage conversion gain of the basic Gilbert-cell
increased in this technique [28]. without current bleeding technique can be expressed as:
S. Douss, F. Touati and M. Loulou proposed current -bleeding 2
Av  gm RL . (5)
technique with double-Balanced mixer [8]. This technique can be 
explained more clearly using a single-balanced mixer as shown in
It is clear that conversion gain of double balanced Gilbert
Fig.8 [8], in which a bleeding current source (I_BLD) is added
mixer depends on the trans-conductance (gm) of RF stage
between the supply voltage and the source of LO switching
transistors and output load stage resistor (RL) as given by Eq.(5).
transistors, without this current source, the total bias current is IBias
Conversion gain is increased by increasing the current flow
= ID1 + ID2, but by using current bleeding technique IBias increases
through RF differential pair but it also increases the power
to IBias = ID1 + ID2 + I_BLD. Therefore, the conversion gain as well
consumption. Besides this, voltage headroom problems also
as linearity can be improved simultaneously but this increases the
arises due to higher RF current through the switching transistor.
total power dissipation of the mixer. The power dissipation of the
Current-bleeding transistors are used to minimize this problem as
mixer can be controlled by keeping the total bias current (IBias)
they provide larger current at trans-conductance stage but do not
constant. For this ID1 and ID2 can be decreased, by adding the
increase the current flowing through the switching pair transistors
current bleeding source but it requires to increase the load resistor
as well.
RL to ensure that the bias conditions of the switching pair should
not be destroyed. Improvement of conversion gain in this Eric A.M. Klumperink, Sander L.J. Gierkink, Arnoud P. Van
technique can be shown as Eq.(4) [8]. der Wel and Bram Nauta proposed a switched bias technique [2]
for reducing flicker (1/f) as well as white noise. The principle of
2
CG  RL K n I Bias (4) switched biasing technique with conventional constant bias is
 shown in Fig.10 [2].
W
where, K n  nCox .
L

Fig.10. Basic principle of switched biasing technique [2]

In switched biasing technique MOS transistor is periodically

switched between two states [2]:
1) “Active state” or “operational state”, in which transistor is
biased in strong inversion region to provide a bias current
Fig.8. Single-balanced mixer with current bleeding source [8] [2].
2) “Inactive state” or “rest state”, in which transistor is biased
in or near to accumulation region. In this region MOSFET
does not remain in fully active state. This state of
transistor is responsible of reducing flicker noise as well as
power consumption [2], [3], [10].
In double-balanced Gilbert cell mixer, tail current source of
the transistor is considered as a critical noise source. The switched
biasing technique replaces the tail current source into two half size
transistors by operating them alternatively using IF output signal
[2]. This reduces the flicker noise generated by tail current source.
The switching operation of switched bias tail current transistors
are generally asymmetric because of large difference between the
dc voltage of the output nodes and the threshold voltage of the tail
current. To overcome this problem, dc level shifter can be used,
as shown in Fig.11 [3], to provide efficient gate to source voltage
(VGS) for symmetric switching operation with a small output
swing and to make the overdrive voltage as smaller as possible,
but the overall conversion gain of the mixer is reduced [10].
Fig.9. Double-balanced mixer with current-bleeding technique [8]

184
ISSN: 2395-1680 (ONLINE) ICTACT JOURNAL ON MICROELECTRONICS, APRIL 2016, VOLUME: 02, ISSUE: 01

VDD

IF+ IF-

Gate Driven +
(RF or LO)
-
M2 M3 M4 M5

Bulk Driven +
(RF or LO) -

Vbias } Tail Current Source

(Optional)
M1

Fig.13. Schematic of Bulk driven mixer core [14]

The principle of the Bulk-driven method is that; the gate-
source voltage (VGS) of all the four transistors is set to a value
Fig.11. Schematic of switched biasing technique [3]
sufficient to form an inversion layer, and an RF input signal is
applied to the bulk terminal. Bulk-driven technique increases the
Pokuri Sravanthi and Aniruddha Chandra applied folded
threshold voltage, therefore, the condition of (VGS > Vth) is relaxed
cascode technique [4] in Gilbert mixer for reducing the power
[14]. Using Bulk driven technique number of transistors reduces,
consumption of the mixer while maintaining the values of all the
therefore, voltage supply is decreased. Transistors are biased in
parameters as of Gilbert mixer but the linearity of the mixer is
sub threshold region for decreasing current dissipation, therefore,
degraded. Basic schematic of folded cascode is shown in Fig.12
power consumption is low, and drain current is estimated by [24]:
[4].
W  V  Vth 
ID  I exp  GS . (6)
L Do  nVT 
The relationship between threshold voltage (Vth) and source-
bulk voltage (VSB) is given by Eq.(7) [24]:
Vth  Vtho    2F  VSB  2F  (7)

where, (Vtho) is the nominal zero bias threshold voltage, i.e., for
(VSB) = 0,  is the process-dependent body effect factor, and (F)
is a substrate (Bulk) Fermi potential typically in the range of 0.2-
0.4V [27, p. 258].
D. Selvathi, M. Pown, and S. Manjula presents a paper on
Fig.12. Folded cascode mixer [4] CCPD technique [6], to improve the mixer performance in terms
of linearity.
Che-Yu Wang and Jeng-Han Tsai proposed a mixer with
The expression using Taylor series expansion for the drain
reduced supply voltage and low power consumption using Bulk
current (ids) versus gate to source voltage (vgs) up to its third-order
driven technique [5]. This technique allows the RF trans-
non-linearity can be expressed as [6]:
conductance and LO switching stages of the core of the basic
Gilbert mixer to be merged to a single stage, consisting of only ids  idc  gm1vgs  gm2vgs
2
 gm3v3gs . (8)
four transistors. Schematic of this technique is shown in Fig.13
[14]. In Eq.(8), gm1, gm2, gm3 are the first order, second order and
third order of transconductance of the transistor. The coefficients
for higher order terms are typically small and can be ignored.
The main objective of using CCPD is to improve the linearity
of mixer by incorporating auxiliary transistors, operating in weak
inversion region and main transistors, operating in strong
saturation region, as shown in Fig.14 [6].

185
 RAHUL SHARMA et al.: A COMPARATIVE STUDY OF DIFFERENT TYPES OF MIXER TOPOLOGIES

Cross coupled AT Auxillary

Transistors
C1 C2

C1 M3 M4 C3

M1 M1a M2a M2
Vbias Vbias

2 54
1uF
RF+ RF-
Ibias 0C2 M1 Main M2 C4
1uF
31 Transistors
0
1kΩ
RFin

R1 R2

Fig.14. Schematic of CCPD technique trans-conductance stage Fig.16. Differential multiple gated transistor (DMGTR) [6]
[6]
The advantages of the MGTR technique is that it provides high
Bonkee Kim, Jin-Su Ko, and Kwyro Lee proposed Multiple conversion gain, moderate noise figure and higher linearity under
Gated Transistor (MGTR) [7] technique, as shown in Fig.15 [7]. lower power dissipation but the DC overdriving voltage of the
The input signal is given to the gate of the RF trans-conductance auxiliary transistors can degrades the overall linearity of the
transistors. The drains of differential main transistor are mixer. In orderI1
1A
to overcome this limitation, careful design of the
connected to the respective drains of the differential auxiliary bias voltage of the auxiliary transistors is needed so that auxiliary
transistor. The main transistors shown in Fig.16 [6] are operating transistors should operate in weak inversion region [29].
in strong inversion region and the auxiliary transistors are Comparison of the discussed proposed mixer techniques is
operating in weak inversion region. The MGTR enables the shown in Table I. Comparative results show that there is trade-off
cancellation of third order derivative due to the introduction of among different parameters of mixer ,which is in agreement with
auxiliary transistors. RF design hexagon [12], [23].

C1 M1 3. CONCLUSION
RFin
C1 M1
In this paper, various mixer topologies such as Multi-Tanh,
Vgs current bleeding, switched biasing, folded cascode, bulk-driven
RFin Vgs CCPD and MGTR techniques are discussed and compared. It can
+ be concluded that there is a trade-off among the performance
parameters of the mixer. If one of the parameters of the mixer is
C2 M2 C2 M2
RFin
improved using some technique then the other mixer parameter
may be degraded. Such as linearity can be improved by using
Multi-Tanh technique but it suffers with a very low conversion
Vgs Vgs gain. Similarly bulk driven technique is used to obtain low power
consumption but it degrades linearity. So the overall comparative
Fig.15. Single ended
8 00
MGTR [7]
45 0 0 study shows that there must be some compromise among the
R2 0 2 R4
+01
R1
36
13 0
9 0 10
values of the different performance parameters of the mixer. This
7 11 work facilitates the selection of a particular topology of the mixer
depending on the specific requirements of the mixer.

Table.1. Summary and Comparison of different types of mixer techniques

Techniques Multi-Tanh Current Bleeding Folded cascode Bulk-driven CCPD MGTR Switched biasing

Linearity High High Moderate Low High Moderate Low

Noise figure Moderate High High High Moderate Moderate Low

Conversion gain Low Moderate Moderate Moderate Moderate Low Moderate

Power consumption High High Low Low High High High

186
ISSN: 2395-1680 (ONLINE) ICTACT JOURNAL ON MICROELECTRONICS, APRIL 2016, VOLUME: 02, ISSUE: 01

REFERENCES
[1] Barrie Gilbert, “The Multi-Tanh Principle: A tutorial Proceedings of International Symposium on Integrated
Overview”, IEEE Journal of Solid-State Circuits, Vol. 33, No. Circuits, pp. 341-344, 2007.
1, pp. 2-17, 1998. [16] B. Iji Ayobami, Forest Zhu and Michael Heimlich, “A Down
[2] E.A.M. Klumperink, S.L.J. Gierkink, A.P. Van Der Wel and Converter Active Mixer, in 0.25μm CMOS Process for Ultra
B. Nauta, “Reducing MOSFET 1/f Noise and Power Wide-Band Applications”, Proceedings of International
Consumption by Switched Biasing”, IEEE Journal of Solid- Symposium on Communications and Information
State Circuits, Vol. 35, No. 7, pp. 994-1001, 2000. Technologies, pp. 28-31, 2012.
[3] Jong Ha Kim, Hee Woo An and Tae Yeoul Yun, “A Low- [17] M. Jouri, A. Golmakani, M. Yahyabadi and H. Khosrowjerdi,
Noise WLAN Mixer Using Switched Biasing Technique”, “Design and Simulation of a Down-conversion CMOS Mixer
IEEE Microwave and Wireless Components Letters, Vol. 19, for UWB Applications”, Proceedings of International
No. 10, pp. 650-652, 2009. Conference on Electrical, Electronics, Computer Telecomm.
[4] Pokuri Sravanthi and Aniruddha Chandra, “Current-Bleeding and Information Technology, pp. 937- 940, 2010.
Folded Gilbert RF Mixer Design for Wireless Applications”, [18] Mixer Evolution, The Balanced Mixer Family Circle,
Proceedings of IEEE Conference Information and Available at :
Communication Technologies, pp. 1044-1048, 2013. http://www.markimicrowave.com/assets/appnotes/balanced.
[5] Che Yu Wang and Jeng Han Tsai, “A 51 to 65GHz Low- pdf
Power Bulk-Driven Mixer Using 0.13m CMOS [19] Peter Paliwoda and Mona Hella, “An Optimized CMOS
Technology”, IEEE Microwave and Wireless Components Gilbert Mixer using Inter-Stage Inductance for Ultra
Letters, Vol. 18, No. 8, pp. 521-523, 2009. Wideband Receivers”, Proceedings of 49th International
[6] D. Selvathi, M. Pown and S. Manjula, “Design and Analysis Midwest Symposium on Circuits and Systems, Vol. 1, pp. 362-
of UWB Down-Conversion Mixer with Linearization 365, 2006.
Techniques”, WSEAS Transactions on Circuits and Systems, [20] Q. Li and J.S. Yuan, “Linearity Analysis and Design
Vol. 13, pp. 202-207, 2014. Optimisation for 0.18μm CMOS RF Mixer”, IEE
[7] Bonkee Kim, Jin Su Ko, and Kwyro Lee, “A New Proceedings - Circuits, Devices and Systems, Vol. 149, No. 2,
Linearization Technique for MOSFET RF Amplifier Using pp. 112-118, 2002.
Multiple Gated Transistors”, IEEE Microwave and Guided [21] Saeed Gholami, Hossein Shamsi, Hossein Mirzaie and
Wave Letters, Vol. 10, No. 9, pp. 371-373, 2002. Davood Mirzahosseini, “Design and Analysis of a High-
[8] S. Douss, F. Touati and M. Loulou, “Design Optimization Linear UWB CMOS Mixer”, Proceedings of 19th Iranian
Methodology of CMOS Active Mixers for Multi-Standard Conference on Electrical Engineering, pp. 1-4, 2011.
Receivers”, Proceedings of 3rd International Conference [22] Anh-Tuan Phan, et al. “A High Performance CMOS Direct
Design and Technology of Integrated Systems in Nanoscale Down Conversion Mixer for UWB System”, Proceedings of
Era, pp. 1-6, 2008. the 14th ACM Great Lakes Symposium on VLSI, pp. 377-380,
[9] Hooman Rashtian, Amir Hossein Masnadi Shirazi and 2004.
Shahriar Mirabbasi, “On the Use of Body-Biasing to Improve [23] Thomas H. Lee, “The Design of CMOS Radio-Frequency
Linearity in Gilbert-Cell CMOS Mixers”, Microelectronics Integrated Circuits”, 2nd Edition, Cambridge University
Journal, Vol. 45, No. 8, pp. 1026-1032, 2014. Press, 1998.
[10] Myoung Gyun Kim, Hee Woo An, Yun Mo Kang, Ji Young [24] E. Abiri, Mohammad Reza Salehi and Masoud Bahrebar, “A
Lee and Tae Yeoul Yun, “A Low-Voltage, Low-Power, and 2.4GHz Ultra-Low Power and High Gain Bulk Driven
Low-Noise UWB Mixer Using Bulk-Injection and Switched Mixer”, Proceedings of International Conference on Power
Biasing Techniques”, IEEE Transactions on Microwave Electronics, Systems and Applications, pp. 197-200, 2012.
Theory and Techniques, Vol. 60, No. 8, pp. 2486-2493, 2012. [25] Richard Chihsi Li, “RF Circuit Design”, 2nd Edition, John
[11] M. Tavassoli, E. Kargaran, S.I. Tous and H. Nabovati, Wiley and Sons, 2012.
“Design of 0.5V, 1.28mW CMOS UWB Mixer using the [26] Ickjin Kwon and Kwyro Lee, “An Integrated Low Power
Body Effect”, Proceedings of 55th International Midwest Highly Linear 2.4GHz CMOS Receiver Front-End Based on
Symposium on Circuits and Systems, pp. 202-205, 2012. Current Amplification and Mixing”, IEEE Microwave and
[12] B. Razavi, “RF Microelectronics”, Prentice Hall, 1998. Wireless Components Letters, Vol. 15, No. 1, pp. 36-38, 2005.
[13] K.-H. Liang, H.-Y. Chang, Y.-J. Chan, “A 0.5-7.5GHz Ultra [27] A.S. Sedra and K.C. Smith, “Microelectronic Circuits”, 5th
Low-Voltage Low-Power Mixer Using Bulk-Injection Method Edition, Oxford University Press, 2004.
by 0.18m CMOS Technology”, IEEE Microwave & Wireless [28] Hong Qi, Chen Junning, Pan Hao and Meng Jian, “A 1.2V
Components Letters, Vol. 17, No. 7, pp. 531-533, 2007. High Linearity Mixer Design”, Proceedings of 8th
[14] Ganesh Kathiresan and Chris Toumazou, “A Low Voltage International conference on Electronics Measurement and
Bulk Driven Down Conversion Mixer Core, Circuits and Instruments, pp. 274-277, 2007.
Systems”, Proceedings of the IEEE International Symposium [29] Wu Yiqiang, Wang Zhigong, Xu Jian, Wang Jian, Zhang Ouli
on Circuits and Systems, Vol. 2, pp. 598-601, 1999. and Tang Lu, “Design of a Novel Mixer with High Gain and
[15] Pan Renjing, Yeo Kiat Seng and Zheng Yuanjin, “A Low- Linearity improvement for DRM/DAB Applications”,
Voltage Low-Power High Linear and Wide-Band Mixer”, Journal of Semiconductors, Vol. 34, No. 1, pp. 015001-
015005, 2013.

187