Upload
marius-ferdy
View
212
Download
0
Embed Size (px)
Citation preview
7/29/2019 PCIRF_5_MIXER_4
1/35
EECS 142
Lecture 20: BJT/FET Mixers/Mixer Noise
Prof. Ali M. Niknejad
University of California, Berkeley
Copyright c 2008 by Ali M. Niknejad
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 1/35
7/29/2019 PCIRF_5_MIXER_4
2/35
A BJT Mixer
C
L1
C1
L2
C2
L3
C
C3
LO
RF
IF
The transformer is used to sum the LO and RF signals at the
input. The winding inductance is used to form resonant tanks atthe LO and RF frequencies.
The output tank is tuned to the IF frequency.
Large capacitors are used to form AC grounds.
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 2/35
7/29/2019 PCIRF_5_MIXER_4
3/35
AC Eq. Circuit
L1
C1
L2
C2
L3C3LO
RF
IF
The AC equivalent circuit is shown above.
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 3/35
7/29/2019 PCIRF_5_MIXER_4
4/35
BJT Mixer Analysis
When we apply the LO alone, the collector current of the mixeris given by
IC = IQ
1 + 2I1(b)
I0(b)cos t + 2I
2(b)I0(b)
cos2t +
We can therefore define a time-varying gm(t) by
gm(t) =IC(t)
Vt=
qIC(t)
kT
The output current when the RF is also applied is thereforegiven by iC(t) = gm(t)vs
iC =qIQ
kT 1 +2I1(b)
I0(b)
cos t +2I2(b)
I0(b)
cos2t + Vs cos stA. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 4/35
7/29/2019 PCIRF_5_MIXER_4
5/35
BJT Mixer Analysis (cont)
The output at the IF is therefore given by
iC|IF = VsqIQkTgmQ
I1(b)
I0(b)cos(0 s
IF)t
The conversion gain is given by
gconv = gmQI1(b)
I0(b)
2 4 6 8 10
0.2
0.4
0.6
0.8
1
gconv
gmQ
b =Vi
V
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 5/35
7/29/2019 PCIRF_5_MIXER_4
6/35
LO Signal Drive
For now, lets ignore the small-signal input and determine theimpedance seen by the LO drive. If we examine the collectorcurrent
IC = IQ
1 +
2I1(b)
I0(b)cos t +
2I2(b)
I0(b)cos2t +
The base current is simply IC/, and so the input impedanceseen by the LO is given by
Zi|0
=Vo
iB,0=
Vo
iC,0=
Vo
IQ 2I1(b)I0(b)=
bVt
IQ 2I1(b)I0(b)
=b
2
gmQ
I0(b)
I1(b)=
Gm
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 6/35
7/29/2019 PCIRF_5_MIXER_4
7/35
RF Signal Drive
The impedance seen by the RF singal source is also the basecurrent at the s components. Typically, we have a high-Qcircuit at the input that resonates at RF.
iB(t) =iC(t)
= 1
qIQkT
Vs cos st + 2I
1(b)I0(b)
cos(0 s)t +
The input impedance is thus the same as an amplifier
Rin =Vs
|component in iB at s|=
kT
qIQ=
gmQ
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 7/35
7/29/2019 PCIRF_5_MIXER_4
8/35
Mixer Analysis: General Approach
If we go back to our original equations, our major assumptionwas that the mixer is a linear time-varying function relative tothe RF input. Lets see how that comes about
IC = ISevBE/Vt
where
vBE = vin + vo + VA
or
IC = ISeVA/Vt eb cos0t e
VsVt
cosst
If we assume that the RF signal is weak, then we canapproximate ex 1 + x
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 8/35
7/29/2019 PCIRF_5_MIXER_4
9/35
General Approach (cont)
Now the output current can be expanded into
IC = IQ1 + 2I1(b)I0(b) cos t + 2I2(b)I0(b) cos2t + 1 +
VsVt
cos stIn other words, the output can be written as
= BIAS + LO + Conversion Products
In general we would filter the output of the mixer and so the LOterms can be minimized. Likewise, the RF terms are undesiredand filtered from the output.
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 9/35
7/29/2019 PCIRF_5_MIXER_4
10/35
Distortion in Mixers
Using the same formulation, we can now insert a signal with twotones
vin = Vs1 cos s1t + Vs2 cos s2
IC = ISeVA/Vt eb cos0t e
Vs1Vt
coss1t+Vs2Vt
coss2t
The final term can be expanded into a Taylor series
IC = ISeVA/Vt eb cos0t
1 + Vs1 cos s1t + Vs2 cos s2t + ( )
2 + ( )3 +
The square and cubic terms produce IM products as before,but now these products are frequency translated to the IFfrequency
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 10/35
7/29/2019 PCIRF_5_MIXER_4
11/35
Another BJT Mixer
L3
C
C3LO
RF
IF
C0
CsLs
+
v0x
The signal from the LO driver is capacitively coupled to the BJTmixer
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 11/35
7/29/2019 PCIRF_5_MIXER_4
12/35
LO Capacitive Divider
CsLs
+
v0x
C0
+
v0x
C0
C
s
Assume that LO > RF, or a high side injectionNote beyond resonance, the input impedance of the tankappears capacitive. Thus Cs is the effective capacitance of thetank. The equivalent circuit for the LO drive is therefore a
capacitive divider
vo =Co
Co + Csvox
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 12/35
7/29/2019 PCIRF_5_MIXER_4
13/35
Harmonic Mixer
Second
Harmonic
Mixer
LO
RF I F
We can use a harmonic of the LO to build a mixer.
Example, let LO = 500MHz, RF = 900MHz, and IF = 100MHz.
Note that IF = 2LO RF = 1000 900 = 100
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 13/35
7/29/2019 PCIRF_5_MIXER_4
14/35
Harmonic Mixer Analysis
The nth harmonic conversion tranconductance is given by
gconv,n
=|IF current out|
|input signal voltage|=
gn
2
For a BJT, we have
gconv,n = gmQ In(b)I0(b)
The advantage of a harmonic mixer is the use of a lower
frequency LO and the separation between LO and RF.The disadvantage is the lower conversion gain and higher noise.
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 14/35
7/29/2019 PCIRF_5_MIXER_4
15/35
FET Large Signal Drive
vo
VQ
ID
Consider the output current of a FET driven by a large LO signal
ID =Cox
2
W
L(VGS VT)
2(1 + VDS)
where VGS = VA + vLO = VA + Vo cos 0t. Here we implicitlyassume that Vo is small enough such that it does not take thedevice into cutoff.
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 15/35
7/29/2019 PCIRF_5_MIXER_4
16/35
FET Large Signal Drive (cont)
0.5 1 1.5 2 2.5 3
That means that VA + V0 cos 0t > VT, or VA V0 > VT, orequivalently V0 < VA VT. Under such a case we expand thecurrent
ID
(VA VT)
2 + V20 cos2 0t + 2(VA VT)V0 cos 0t
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 16/35
7/29/2019 PCIRF_5_MIXER_4
17/35
FET Current Components
The cos2 term can be further expanded into a DC and secondharmonic term.
Identifying the quiescent operating point
IQ =Cox
2
W
L(VA VT)
2(1 + VDS)
ID = IDQ + CoxW
L
1
4V20
bias point shift+ (VA VT)Vo cos 0t
LO modulation+
V204
cos(20t)
LO 2nd harmonic
(1 + VDS)
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 17/35
7/29/2019 PCIRF_5_MIXER_4
18/35
FET Time-Varying Transconductance
The transconductance of a FET is given by (assuming stronginversion operation)
g(t) = IDVGS
= Cox WL
(VGS VT)(1 + VDS)
VGS(t) = VA + V0 cos 0t
g(t) = CoxW
L(VA VT + V0 cos 0t)(1 + VDS)
g(t) = gmQ1 + V0VA VT cos 0t (1 + VDS)
This is an almost ideal mixer in that there is no harmoniccomponents in the transconductance.
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 18/35
7/29/2019 PCIRF_5_MIXER_4
19/35
MOS Mixer
vs
VQ
vo
IF FilterWe see that we can build a mixer
by simply injecting an LO + RFsignal at the gate of the FET (ig-nore output resistance)
i0 = g(t)vs = gmQ
1 +
V0VA VT
cos 0t
Vs cos st
i0|IF =gmQ
2
V0VA VT
cos(0 s)tVs
gc =
i0|IF
Vs =
gmQ
2
V0
VA VT
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 19/35
7/29/2019 PCIRF_5_MIXER_4
20/35
MOS Mixer Summary
But gmQ = CoxWL (VA VT)
gc =
Cox
2
W
L V0
which means that gc is independent of bias VA. The gain iscontrolled by the LO amplitude V0 and by the device aspect
ratio.Keep in mind, though, that the transistor must remain in forwardactive region in the entire cycle for the above assumptions tohold.
In practice, a real FET is not square law and the above analysisshould be verified with extensive simulation. Sub-thresholdconduction and output conductance complicate the picture.
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 20/35
7/29/2019 PCIRF_5_MIXER_4
21/35
Dual Gate Mixer
+
vs
M1
M2
VA
+
vo
VA
io
G1
G2D2
S1
shared junction
(no contact)
The dual gate mixer, or more commonly a cascode amplifier,can be turned into a mixer by applying the LO at the gate of M2and the RF signal at the gate of M1. Using two transistors inplace of one transistor results in area savings since the signalsdo not need to be combined with a transformer or capacitively .
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 21/35
D l G Mi O i
7/29/2019 PCIRF_5_MIXER_4
22/35
Dual Gate Mixer Operation
0.5 1 1.5 2 2.5 3
trio
de
saturation
LO
Without the LO signal, this is simply a cascode amplifier. Butthe LO signal is large enough to push M1 into triode during partof the operating cycle.
The transconductance of M1 is therefore modulated periodically
gm|sat = CoxW
L(VGS VT)
gm|triode = CoxWL
VDS
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 22/35
7/29/2019 PCIRF_5_MIXER_4
23/35
Dual Gate Waveforms
Active Region
Triode Region
gm,max
gm(t)
vLO,1
vLO,2
VGS2 is roughly constant since M1 acts like a current source.
VD1 = vLO VGS2 = VA2 + V0 cos 0t VGS2
g(t) =
Cox
WL (VGS1 VT) VD1 > VGS VT
Cox(VA2 VGS2 |V0 cos 0t|) VD1 < VGS VT
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 23/35
7/29/2019 PCIRF_5_MIXER_4
24/35
Realistic Waveforms
A more sophisticated analysis would take sub-thresholdoperation into account and the resulting g(t) curve would besmoother. A Fourier decomposition of the waveform would yield
the conversion gain coefficient as the first harmonic amplitude.
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 24/35
Mi A l i Ti D i
7/29/2019 PCIRF_5_MIXER_4
25/35
Mixer Analysis: Time Domain
A generic mixer operates with a periodic transfer functionh(t + T) = h(t), where T = 1/0, or T is the LO period. We canthus expand h(t) into a Fourier series
y(t) = h(t)x(t) =
cnej0ntx(t)
For a sinusoidal input, x(t) = A(t)cos 1t, we have
y(t) =
cn2
A(t)ej(1+0n)t + ej(1+0n)t
Since h(t) is a real function, the coefficients ck = ck are even.That means that we can pair positive and negative frequency
components.
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 25/35
Ti D i A l i ( t)
7/29/2019 PCIRF_5_MIXER_4
26/35
Time Domain Analysis (cont)
Take c1 and c1 as an example
= c1
ej(1+0)t + ej(1+0)t
2A(t)+c
1
ej(10)t + ej(10)t
2A(t)+
= c1A(t) cos(1 + 0)t + c1A(t)cos(1 0)t +
Summing together all the components, we have
y(t) =
cn cos(1 + n0)t
Unlike a perfect multiplier, we get an infinite number offrequency translations up and down by harmonics of 0.
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 26/35
Frequency Domain Analysis
7/29/2019 PCIRF_5_MIXER_4
27/35
Frequency Domain Analysis
Since multiplication in time, y(t) = h(t)x(t), is convolution in thefrequency domain, we have
Y(f) = H(f) X(f)
The transfer function H(f) =
cn(f nf0) has a discrete
spectrum. So the output is given by
Y(f) =
cn( nf0)X(f )d
=
cn
( nf0)X(f )d
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 27/35
Frequency Domain (cont)
7/29/2019 PCIRF_5_MIXER_4
28/35
Frequency Domain (cont)
noi
sefoldingfrom3LO
noisefoldingfrom
2LO
noise
fold
ingf
rom-2LO
noise
foldin
gfrom-3LO
fo fsfiffofs fif 2fo 3fo2fo3fo
By the frequency sifting property of the (f ) function, wehave
Y(f) =
cnX(f nf0)Thus, the input spectrum is shifted by all harmonics of the LOup and down in frequency.
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 28/35
Noise/Image Problem
7/29/2019 PCIRF_5_MIXER_4
29/35
Noise/Image Problem
Previously we examined the image problem. Any signalenergy a distance of IF from the LO gets downconverted in aperfect multiplier. But now we see that for a general mixer, any
signal energy with an IF of any harmonic of the LO will bedownconverted !
These other images are easy to reject because they are distantfrom the desired signal and a image reject filter will be able to
attenuate them significantly.
The noise power, though, in all image bands will fold onto the IFfrequency. Note that the noise is generated by the mixer sourceresistance itself and has a white spectrum. Even though thenoise of the antenna is filtered, new noise is generated by thefilter itself!
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 29/35
Current Commutating Mixers
7/29/2019 PCIRF_5_MIXER_4
30/35
Current Commutating Mixers
+LO LO
VCC
Q2 Q3
Q1+RF
IF
RIF
+LO LO
+RF
IF
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 30/35
Current Commutating Mixer Model
7/29/2019 PCIRF_5_MIXER_4
31/35
Current Commutating Mixer Model
+LO LO
VCC
Q1+RF
IF
RIF
If we model the circuit with ideal elements, we see that thecurrent IC1 is either switched to the output or to supply at the
rate of the LO signal.When the LO signal is positive, we have a cascode dumping itscurrent into the supply. When the LO signal is negative, though,we have a cascode amplifier driving the output.
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 31/35
Conversion Gain
7/29/2019 PCIRF_5_MIXER_4
32/35
Conversion Gain
t
TLO
+1
0
s(t)
We can now see that the output current is given by a periodictime varying transconductance
io = gm(t)vs = gmQs(t)vs
where s(t) is a square pulse waveform (ideally) switching
between 1 and 0 at the rate of the LO signal. A Fourierdecomposition yields
io = gmQvs0.5 +2
cos 0t
2
1
3cos30t +
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 32/35
Conversion Gain (cont)
7/29/2019 PCIRF_5_MIXER_4
33/35
Conversion Gain (cont)
So the RF signal vs is amplified (feed-thru) by the DC term andmixed by all the harmonics
ioVs
= gmQ2
12
cos st + 2
cos(0 s)t 23
cos(30 s)t +
The primary conversion gain is gc =1 gmQ.
Since the role of Q1 (or M1) is to simply create an RF current, itcan be degenerated to improve the linearity of the mixer.Inductance degeneration can be employed to also achieve animpedance match.
MOS version acts in a similar way but the conversion gain islower (lower gm) and it requires a larger LO drive.
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 33/35
Differential Output
7/29/2019 PCIRF_5_MIXER_4
34/35
Differential Output
+LO LO
+RF
IF
This block is commonly known as the Gilbert Cell
If we take the output signal differentially, then the output currentis given by
io = gm(t)vs = gmQs2(t)vs
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 34/35
Differential Output Gain
7/29/2019 PCIRF_5_MIXER_4
35/35
Differential Output Gain
t
TLO
+1
1
s2(t)
The pulse waveform s2(t) now switches between 1, and thushas a zero DC value
s2(t) =4
cos 0t
1
3
4
cos30t +
The lack of the DC term means that there is ideally no RFfeedthrough to the IF port. The conversion gain is doubled since
we take a differential output gc/gmQ =2
A. M. Nikne ad Universit of California Berkele EECS 142 Lecture 18 . 35/35