6.776
High Speed Communication Circuits and Systems
Lecture 16
Mixers, continued
Massachusetts Institute of Technology
April 5, 2005
Copyright © 2005 by Hae-Seung Lee and Michael H.
Perrott
Image Reject Mixer, Review
RF
r(t)
y(t)
Balanced
modulator
Local Oscillator
+
-90o phase
shifter
IF Output
-90o phase
shifter
Balanced
modulator
ƒ
LPF or
BPF
z(t)
^z(t)
Rather than filtering out the image, we can cancel it out using an image
rejection mixer
-
Advantages
ƒ Allows a low IF frequency to be used without requiring a high Q filter
ƒ Very amenable to integration
Disadvantage
ƒ Level of image rejection is determined by mismatch in gain and phase different
paths
ƒ Practical architectures limited to about 40 dB image rejection
H.-S. Lee & M.H. Perrott
MIT OCW
Image Reject Mixer, Alternate Implementation
a(t)
RF in(f)
e(t)
0
f1
f
2cos(2πf2t)
2cos(2πf1t)
2sin(2πf1t)
b(t)
ƒ
ƒ
c(t)
Image Desired
Interferer channel
RF in
-f1
Lowpass
Lowpass
IF out
2sin(2πf2t)
d(t)
g(t)
Avoids 90 degree phase shift of signal
Precise 90 degree phase shift of LO outputs is much
easier by using quadrature VCO’s or frequency dividers
H.-S. Lee & M.H. Perrott
MIT OCW
Image Reject Mixer Principles – Step 1
a(t)
RF in(f)
0
f1
f
2sin(2πf1t)
Note: we are assuming RF in(f)
is purely real right now
Lowpass
1
-f1
Lowpass
0
IF out
2sin(2πf2t)
d(t)
g(t)
A(f)
1
1
f1
f
-f1
Lowpass
0
f1
f
B(f)
j
j
f1
-f1
e(t)
2cos(2πf2t)
2cos(2πf1t)
b(t)
ƒ
c(t)
Image Desired
Interferer channel
RF in
-f1
Lowpass
0
-j
H.-S. Lee & M.H. Perrott
f
-f1
0
-j
f1
f
MIT OCW
Image Reject Mixer Principles – Step 2
a(t)
RF in(f)
c(t)
e(t)
Image Desired
Interferer channel
RF in
0
-f1
Lowpass
f1
f
2cos(2πf2t)
2cos(2πf1t)
2sin(2πf1t)
Lowpass
b(t)
IF out
2sin(2πf2t)
d(t)
g(t)
C(f)
1
-f1
1
1
0
f1
f
-f1
0
f1
f
D(f)
j
j
f1
-f1
0
-j
H.-S. Lee & M.H. Perrott
f
-f1
0
-j
f1
f
MIT OCW
Image Reject Mixer Principles – Step 3
C(f)
c(t)
e(t)
1
D(f)
2cos(2πf2t)
-f2
j
0
f2
f
2sin(2πf2t)
d(t)
f
0
-j
IF out
g(t)
E(f)
2
1
-f2
1
1
0
f2
f
-f2
0
G(f)
2
f2
0
-j
f
f
f2
1
j
-f2
1
1
-f2
f
f2
-1
-1
-2
H.-S. Lee & M.H. Perrott
MIT OCW
Image Reject Mixer Principles – Step 4
E(f)
2
e(t)
1
G(f)
2
1
IF out
1
0
-f2
1
f
f2
f
-f2
g(t)
f2
-1
-1
-2
IF out(f)
Baseband
Filter
2
-f2
H.-S. Lee & M.H. Perrott
0
f
f2
MIT OCW
Image Reject Mixer Principles – Implementation Issues
a(t)
RF in(f)
c(t)
e(t)
Image Desired
Interferer channel
RF in
0
-f1
Lowpass
f1
f
2cos(2πf2t)
2cos(2πf1t)
2sin(2πf1t)
2sin(2πf2t)
Lowpass
d(t)
b(t)
IF out
g(t)
IF out(f)
(after baseband filtering)
2
0
ƒ
For all analog architecture, additional mixers introduce more
mismatch and noise (limits image rejection and noise figure)
-
ƒ
f
Can fix this problem by digitizing c(t) and d(t), and then performing
final mixing in the digital domain
Can generate accurate quadrature sine wave signals by using a
frequency divider
H.-S. Lee & M.H. Perrott
MIT OCW
What if RF in(f) is Purely Imaginary?
a(t)
Lowpass
c(t)
e(t)
RF in(f)
Image Desired
Interferer channel
j
-f1
0
f1
-j
f
RF in
2cos(2πf2t)
2cos(2πf1t)
2sin(2πf1t)
2sin(2πf2t)
Lowpass
d(t)
b(t)
IF out
g(t)
IF out(f)
(after baseband filtering)
0
f
ƒ
Both desired and image signals disappear!
ƒ
Can we modify the architecture to fix this issue?
- Architecture is sensitive to the phase of the RF input
H.-S. Lee & M.H. Perrott
MIT OCW
Modification of Mixer Architecture for Imaginary RF in(f)
a(t)
Lowpass
c(t)
e(t)
RF in(f)
Image Desired
Interferer channel
j
-f1
0
f1
-j
f
RF in
2cos(2πf1t)
2sin(2πf1t)
2sin(2πf2t)
2cos(2πf2t)
Lowpass
d(t)
b(t)
IF out
g(t)
IF out(f)
(after baseband filtering)
2
0
f
ƒ
Desired channel now appears given two changes
ƒ
Issue – architecture now zeros out desired channel when
RF in(f) is purely real
- Sine
and cosine demodulators are switched in second half of
image rejection mixer
- The two paths are now added rather than subtracted
H.-S. Lee & M.H. Perrott
MIT OCW
Overall Mixer Architecture – Use I/Q Demodulation
2sin(2πf2t)
real part of RF in(f)
1
1
a(t)
0
-f1
imag. part of RF in(f)
j
-f1
f1
f
RF in
Image Desired
Interferer channel
0
f1
IF out
I
Lowpass
f
2cos(2πf1t)
2cos(2πf2t)
2sin(2πf1t) Lowpass
2sin(2πf2t)
IF out
Q
b(t)
-j
2cos(2πf2t)
IF out(f) (I component)
(after baseband filtering)
2
2
0
ƒ
IF out(f) (Q component)
(after baseband filtering)
f
0
f
Both real and imag. parts of RF input now pass through
H.-S. Lee & M.H. Perrott
MIT OCW
Example of Double Conversion Image Reject Mixer
RF Filter
LNA
A/D
I
Receive Signal Path
Q
I
LO1
RF
12 3
LO2
IF
N
1 23
Q
N
BB
123
N
3
Figure by MIT OCW.
REF: “A 1.9-GHz Wide-Band IF Double Conversion CMOS Receiver for Cordless
Telephone Applications,”J C. Rudell, et. al. IEEE J. Solid-State Circuits, Vol. SC32, Dec. 1997 pp 2071-2088
ƒ
ƒ
ƒ
I/Q Image rejection provided by 6 mixers
IF filtering is LPF: single-chip integration is easier
LO frequency is unequal to carrier – LO leakage is not
an issue
H.-S. Lee & M.H. Perrott
MIT OCW
Mixer Single-Sideband (SSB) Noise Figure
Noise
RF in(f)
Image
band
Desired
channel
Noise
0
LO out
= 2cos(2πfot)
IF out(f)
LO out(f)
1
-fo
ƒ
0
fo
IF out
f
fo
-∆f ∆f
1
Channel
Filter
RF in
SRF
No
-fo
Image
Rejection
Filter
f
-∆f
0
Noise from Desired
and Image bands add
SRF
2No
∆f
f
Issue – broadband noise from mixer or front end filter will be
present in both image and desired bands
-
Noise from both image and desired bands will combine in desired
channel at IF output
ƒ Neither image reject filter not channel filter can remove this
The SSB noise figure computes (correctly) noise in both the desired
signal band and image band with signal only in the desired band
(SSB signal, but DSB noise)
H.-S. Lee & M.H. Perrott
MIT OCW
Mixer Double-Sideband (DSB) Noise Figure
Noise
RF in(f)
Desired
channel
Noise
0
Channel
Filter
f
fo
LO out
= 2cos(2πfot)
IF out(f)
1
ƒ
ƒ
ƒ
ƒ
0
Noise from
positive and negative
frequency bands add
2SRF
LO out(f)
-fo
IF out
RF in
SRF
No
-fo
Image
Rejection
Filter
2No
fo
f
-∆f
0
∆f
f
DSB NF assumes signal and noise in both sidebands (thus 3dB lower noise
figure) – this is misleading because there is no signal in the image band in
heterodyne receivers
For zero IF, there is no image band-DSB noise figure is appropriate
- Noise from positive and negative frequencies combine, but the signals do as well
DSB noise figure is 3 dB lower than SSB noise figure
- DSB noise figure often quoted since it sounds better
For either case, Noise Figure should be computed through simulation
H.-S. Lee & M.H. Perrott
MIT OCW
A Practical Issue – Square Wave LO Oscillator Signals
RF in
Local Oscillator
Output
= 2sgn(cos(2πfot))
LO out(t)
1
T
ƒ
IF out
W
T
t
Square waves are easier to generate than sine waves
- How do they impact the mixing operation when used as
the LO signal?
- We will briefly review Fourier transforms (series) to
understand this issue
H.-S. Lee & M.H. Perrott
MIT OCW
Two Important Transform Pairs
ƒ
Transform of a rectangle pulse in time is a sinc function
in frequency x(t)
X(f)
T
1
T
2
T
2
ƒ
t
f
1
T
Transform of an impulse train in time is an impulse
train in frequency
T
H.-S. Lee & M.H. Perrott
s(t)
S(f)
1
1
T
T
t
1
T
1
T
f
MIT OCW
Decomposition of Square Wave to Simplify Analysis
ƒ
Consider now a square wave with duty cycle W/T
y(t)
W
1
T
T
ƒ
t
Decomposition in time
x(t)
s(t)
*
W
1
t
H.-S. Lee & M.H. Perrott
1
T
T
t
MIT OCW
Associated Frequency Transforms
ƒ
Consider now a square wave with duty cycle W/T
y(t)
W
1
T
T
ƒ
t
Decomposition in frequency
W
X(f)
S(f)
1
W
H.-S. Lee & M.H. Perrott
1
T
f
1
T
1
T
f
MIT OCW
Overall Frequency Transform of a Square Wave
ƒ
Resulting transform relationship
y(t)
1
T
W
T
Y(f)
W
T
t
1
W
f
1
T
ƒ
Fundamental at frequency 1/T
ƒ
If W = T/2 (i.e., 50% duty cycle)
ƒ
If the waveform is between 1/2 and -1/2 (rather than 1 and 0)
- Higher harmonics have lower magnitude
- No even harmonics!
- No DC component (50% duty cycle)
H.-S. Lee & M.H. Perrott
MIT OCW
Analysis of Using Square-Wave for LO Signal
RF in(f)
-fo
0
LO out(f)
-3fo -2fo -fo
0
RF in
f
fo
Even order harmonic
due to not having an
exact 50% duty cycle
fo 2fo 3fo
f
DC component
of LO waveform
ƒ
IF out
Local Oscillator
Output
= 2sgn(cos(2πfot))
IF out(f)
-3fo 2fo -fo
0
f
fo 2fo 3fo
Desired
Output
Each frequency component of LO signal will now mix with
the RF input
- If RF input spectrum sufficiently narrowband with respect to f ,
ƒ
then no aliasing will occur
o
Desired output (mixed by the fundamental component) can
be extracted using a filter at the IF output
H.-S. Lee & M.H. Perrott
MIT OCW
Voltage Conversion Gain
RF in(f)
A
2
0
-fo
RF in =
Acos(2π(fo+∆f)t)
IF out
f
fo
∆f
LO ouput =
Bcos(2πfot)
LO out(f)
IF out(f)
AB
4
B
2
-fo
0
fo
f
-fo
-∆f 0 ∆f
fo
f
ƒ
ƒ
Defined as voltage ratio of desired IF value to RF input
Example: for an ideal mixer with RF input = Asin(2π(fo +
∆ f)t) and sine wave LO signal = Bcos(2πfot)
ƒ
For practical mixers, value depends on mixer topology
and LO signal (i.e., sine or square wave)
H.-S. Lee & M.H. Perrott
MIT OCW
Impact of High Voltage Conversion Gain
RF in(f)
A
2
0
-fo
RF in =
Acos(2π(fo+∆f)t)
IF out
f
fo
∆f
LO ouput =
Bcos(2πfot)
LO out(f)
IF out(f)
AB
4
B
2
-fo
ƒ
ƒ
0
fo
f
-fo
-∆f 0 ∆f
f
fo
Benefit of high voltage gain
- The noise of later stages will have less of an impact
Issues with high voltage gain
- May be accompanied by higher noise figure than could
be achieved with lower voltage gain
- May be accompanied by nonlinearities that limit
interference rejection (i.e., passive mixers can generally
be made more linear than active ones)
H.-S. Lee & M.H. Perrott
MIT OCW
Impact of Nonlinearity in Mixers
Memoryless
Nonlinearity A
Desired
RF in(w) Interferers Narrowband
RF in
Ideal
Mixer
Memoryless
Nonlinearity C
y
IF out
Signal
0
w1 w 2
W
Memoryless
Nonlinearity B
ƒ
LO signal
Ignoring dynamic effects, we can model mixer as
nonlinearities around an ideal mixer
- Nonlinearity A will have the same impact as LNA
nonlinearity (measured with IIP3)
- Nonlinearity B will change the spectrum of the LO signal
-
ƒ We already looked at an extreme case (square wave)
ƒ Changes conversion gain somewhat
Nonlinearity C will cause self mixing of IF output
H.-S. Lee & M.H. Perrott
MIT OCW
Primary Focus is Typically Nonlinearity in RF Input Path
Memoryless
Nonlinearity A
Desired
RF in(w) Interferers Narrowband
Memoryless
Nonlinearity C
Ideal
Mixer
y
RF in
z
IF out
Signal
0
w 1 w2
Y(w)
W
x
Corruption of desired signal
Memoryless
Nonlinearity B
W
w1 w2
2w1 2w2
3w1 3w2
0 w2-w1
2w1-w2 2w2-w1 w1+w2 2w1+w2 2w2+w1
LO signal
ƒ
Nonlinearity B not detrimental in most cases
ƒ
Nonlinearity C can be avoided by using a linear load (such as
a resistor)
Nonlinearity A can hamper rejection of interferers
ƒ
- LO signal often a square wave anyway
- Characterize with IIP3 as with LNA designs
- Use two-tone test to measure (similar to LNA)
H.-S. Lee & M.H. Perrott
MIT OCW
The Issue of Balance in Mixers
RF in(f)
DC component
IF out
RF in
0
-fo
f
fo
LO sig
∆f
LO sig(f)
-fo
ƒ
ƒ
ƒ
LO
IF out(f) feedthrough
DC component
0
fo
f
-fo
-∆f 0 ∆f
RF
feedthrough
f
fo
A balanced signal is defined to have a zero DC component
Mixers have two signals of concern with respect to this
issue – LO and RF signals
- Unbalanced RF input causes LO feedthrough
- Unbalanced LO signal causes RF feedthrough
Issue – transistors require a DC offset (e.g. VT) for biasing
H.-S. Lee & M.H. Perrott
MIT OCW
Achieving a Balanced LO Signal with DC Biasing
ƒ
Combine two mixer paths with LO signal 180 degrees
out of phase between the paths
LO sig
1
RF in
IF out
RF in
IF out
0
1
0
LO sig
1
LO sig
0
-1
- DC component is cancelled
H.-S. Lee & M.H. Perrott
MIT OCW
Single-Balanced Mixer
I1
VLO
VRF(t)
DC
ƒ
ƒ
ƒ
ƒ
M1
I2
M2
VLO
Io
VRF
Transconductor
Io = GmVRF
Works by converting RF input voltage to a current, then
switching current between each side of differential pair
Achieves LO balance using technique on previous slide
- Subtraction between paths is inherent to differential output
LO swing should be no larger than needed to fully turn on
and off differential pair
- Square wave is best to minimize noise from M
1
and M2
Transconductor designed for high linearity
H.-S. Lee & M.H. Perrott
MIT OCW
Transconductor Implementation 1
Io
Rs
VRF
Cbig
M1
Rbig
Vbias
ƒ
Apply RF signal to input of common source amp
- Transistor assumed to be in saturation
- Transconductance value is the same as that of the
transistor
ƒ
High Vbias places device in velocity saturation
- Allows high linearity to be achieved
H.-S. Lee & M.H. Perrott
MIT OCW
Transconductor Implementation 2
Io
Rs
Cbig
M1
Vbias
VRF
ƒ
ƒ
Ibias
Apply RF signal to a common gate amplifier
Transconductance value set by inverse of series
combination of Rs and 1/gm of transistor
- Amplifier is effectively degenerated to achieve higher
linearity
ƒ
Ibias can be set for large current density through
device to achieve higher linearity (velocity saturation)
H.-S. Lee & M.H. Perrott
MIT OCW
Transconductor Implementation 3
Io
Rs
VRF
Cbig
M1
Rbig
Ldeg
Vbias
ƒ
Add degeneration to common source amplifier
- Inductor better than resistor
-
ƒ No DC voltage drop
ƒ Increased impedance at high frequencies helps filter out
undesired high frequency components
Don’t generally resonate inductor with Cgs
ƒ Power match usually not required for IC implementation
due to proximity of LNA and mixer
H.-S. Lee & M.H. Perrott
MIT OCW
LO Feedthrough in Single-Balanced Mixers
Higher order
harmonics
Higher order
harmonics
VLO(f)-VLO(f)
I1
VLO
0
-fo
fo
f
VRF(t)
VRF(f)
-f1
Higher order
harmonics
-fo-f1
ƒ
DC
0
M2
VLO
Io
VRF
Transconductor
Io = GmVin
f
f1
Higher order
harmonics
I1(f)-I2(f)
-fo -fo+f1 0 fo-f1
M1
I2
fo
fo+f1
f
DC component of RF input causes very large LO
feedthrough
- Can be removed by filtering, but can also be removed by
achieving a zero DC value for RF input
H.-S. Lee & M.H. Perrott
MIT OCW
Double-Balanced Mixer
Higher order
harmonics
Higher order
harmonics
VLO(f)-VLO(f)
I1
VLO
0
-fo
fo
f
VRF(t)
VRF
Higher order
harmonics
-fo-f1
ƒ
ƒ
ƒ
0
f
f1
VLO
Transconductor
Io = GmVin
DC
Higher order
harmonics
I1(f)-I2(f)
-fo -fo+f1 0 fo-f1
M2
Io
VRF(f)
-f1
M1
I2
fo
fo+f1
f
DC values of LO and RF signals are zero (balanced)
LO feedthrough dramatically reduced!
But, practical transconductor needs bias current
H.-S. Lee & M.H. Perrott
MIT OCW
Achieving a Balanced RF Signal with Biasing
ƒ
Use the same trick as with LO balancing
LO sig
LO sig
1
VRF(t)
RF in
IF out
0
RF in
VRF(t)
1
DC
1
1
signal
0
IF out
0
0
DC
LO sig
LO sig
IF out
LO sig
DC component
cancels
signal component
adds
1
VRF(t)
DC
RF in
0
1
signal
0
LO sig
H.-S. Lee & M.H. Perrott
MIT OCW
Double-Balanced Mixer Implementation
I3
VLO
VRF(t)
DC
M2
Io
VRF
ƒ
M1
I4
Transconductor
Io = GmVRF
I1
VLO
VLO
VRF(t)
DC
M1
I2
VLO
M2
Io
VRF
Transconductor
Io = GmVRF
Applies technique from previous slide
- Subtraction at the output achieved by cross-coupling
the output current of each stage
H.-S. Lee & M.H. Perrott
MIT OCW
Gilbert Mixer
I1
I2
LO DC
VLO
LO DC
RF DC
RF DC
M3
M4
M5
VLO
M6
VLO
VRF(t)
M1
M2
VRF(t)
Ibias
ƒ
ƒ
ƒ
Use a differential pair to achieve the transconductor
implementation
LO signal can be sinusoidal or square wave (preferred)
This is the preferred mixer implementation for most radio
systems!
H.-S. Lee & M.H. Perrott
MIT OCW
A Highly Linear CMOS Analog Multiplier
M7
R1
+
VX
X+
M1
VO-
BIAS
M5
M3
XIN
X-
M9
GRN
M6
M2
VX
M4
+
VY
+
VO
VY
Y-
BIAS
R2
M8
YIN
Y+
VDD
M10
VDD
Figure by MIT OCW.
B. Song, "CMOS RF circuits for data communications applications," IEEE Journal of
Solid-State Circuits, vol. 21, pp. 310 - 317, April 1986.
ƒ
ƒ
Transistors are operated in triode regions
The product terms cancel out resulting in linear multiplication
H.-S. Lee & M.H. Perrott
MIT OCW
CMOS Analog Multiplier Analysis
H.-S. Lee & M.H. Perrott
MIT OCW
A Highly Linear CMOS Mixer
Cf1
VLO
M1
M2
Cb1
Cb2
VLO
M3
M4
VRF
VRF
Rf1
VIF
VIF
Rf2
Cf2
J. Crols and M. S. J. Steyaert, "A 1.5 GHz highly linear CMOS downconversion mixer," IEEE
Journal of Solid-State Circuits, vol. 30, pp. 736 - 742, July 1995
ƒ
ƒ
Transistors are alternated between the off and triode regions by the
LO signal
RF signal varies resistance of channel when in triode
Large bias required on RF inputs to achieve triode operation
High linearity achieved, but very poor noise figure
-
H.-S. Lee & M.H. Perrott
MIT OCW
Passive Mixers
RS/2
Cbig
VLO
2Ain
VRF
VIF
VLO
RS/2
ƒ
ƒ
VLO
CL
RL/2
RL/2
VIF
VLO
Cbig
We can avoid the transconductor and/or op amp
simply use switches to perform the mixing operation
- No bias current required allows low power operation to
be achieved
ƒ
Disadvantage: the RF input is low impedance
H.-S. Lee & M.H. Perrott
MIT OCW
Square-Law Mixer
LL
VRF
RL
CL
VIF
VLO
M1
Vbias
ƒ
ƒ
ƒ
Achieves mixing through nonlinearity of MOS device
- Ideally square law, which leads to a multiplication term
- Undesired components must be filtered out
Need a long channel device to get square law behavior
(no velocity saturation!)
Issue – no isolation between LO and RF ports
H.-S. Lee & M.H. Perrott
MIT OCW
Alternative Implementation of Square Law Mixer
LL
RL
CL
VIF
Cbig
M1 Cbig2
VRF
Rbig
Vbias
Ibias
VLO
ƒ
Drives LO and RF inputs on separate parts of the transistor
ƒ
Issue - poorer performance compared to multiplicationbased mixers
- Allows some isolation between LO and RF signals
- Lots of undesired spectral components
- Poorer isolation between LO and RF ports
H.-S. Lee & M.H. Perrott
MIT OCW
Flicker Noise in Gilbert-Type Mixer
I1
I2
LO DC
VLO
LO DC
RF DC
RF DC
M3
M4
M5
M6
VLO
VLO
VRF(t)
M1
M2
VRF(t)
Ibias
ƒ
ƒ
ƒ
ƒ
Let’s consider Gilbert-type mixer for direct conversion receivers
1/f noise in Gm transistors (M1 and M2): Up-converted to LO
frequency (no issue)
1/f noise in switches (M3-M6) no effect if LO signal is a square
wave
Typically, the LO output is not a square wave, and has finite slope
at the switching instant:1/f noise in M3-M6 modulates the switching
threshold of switch pairs!
H.-S. Lee & M.H. Perrott
MIT OCW
Flicker Noise Analysis
H. Darabi and J Chiu “A Noise Cancellation Technique in Active RF-CMOS Mixers,”
Digest of Technical Papers, 2005 ISSCC pp544-545.
T
Time
Differential Output
Noise
∆t
LO
RF
vn
Differential LO
Slope = S
CP
Error (noise) current at output
I
Time
2I
∆t
∆t = vn/S
vn: 1/f noise of switching devices
ƒ
Figure by MIT OCW.
1/f noise at the output is proportional to bias current I and
inversely proportional to LO slope S
H.-S. Lee & M.H. Perrott
MIT OCW
Flicker Noise Reduction in Gilbert Mixer
ƒ
Inject current at the switching moments to reduce
current through switching devices
Time
Output
Noise
Differential LO
LOn
LOP
A
LOP
ID
Voltages at nodes A & B
CP
iS1
RF
H.-S. Lee & M.H. Perrott
Time
B
ID
iS2
I
Time
Injected currents iS1,2
2(I-ID)
~0
Figure by MIT OCW.
Time
MIT OCW
Actual Implementation of Noise Cancellation
1.2V
VB
R1
MP3 ID ~
~ 2I
MP1
R2
MP2
OUTp
OUTn
M3
M4
LOP
M6
LOn
M1
RFIN
M5
I
LOP
M2
BIAS
Figure by MIT OCW.
H. Darabi and J Chiu “A Noise Cancellation Technique in Active RF-CMOS Mixers,”
Digest of Technical Papers, 2005 ISSCC pp544-545.
H.-S. Lee & M.H. Perrott
MIT OCW
Noise Cancelled Mixer Results
24
LO at 2GHz
22
W/O Injection, Measured
W/I Injection, Measured
DSB NF, dB
20
18
16
W/O Injection,
Simulated
W/I Injection,
Simulated
14
12
10
10
100
1000
IF, kHz
H.-S. Lee & M.H. Perrott
Figure by MIT OCW.
MIT OCW
Noise Cancelled Mixer Results Summary (2GHz)
Parameter
MIXER WITH
INJECTION
MIXER WITHOUT
INJECTION
Measured
Simulated
Measured
Simulated
10.5dBm
11dBm
10.5dBm
10.8dBm
11dB
11.8dB
11dB
11.6dB
Voltage Gain
0.5dB
1dB
0dB
0.5dB
Bias Current
2mA
2mA
2mA
2mA
NF at 20kHz
13.5dB
12.3dB
19.5dB
20.4dB
-1.5dBm
-0.8dBm
-1.5dBm
-0.8dBm
IIP3
White NF
1-dB Compression
Figure by MIT OCW.
H.-S. Lee & M.H. Perrott
MIT OCW