Opamp based
[1] OTA based
[This work]
RB
2.RB= g1m+gRmFro ro
RF RF
A gm
1. Requires low impedance 1. Inverters can be used [3]
3. Two Regimes:
(a) Linearity limited if g1m ≪ gRmFro (b) Noise limited if g1m ≫ gRmFro output stage.
2.RB= 1+ARF 3. Linearity limited
Figure 4.2: Comparison between Opamp based TIA and OTA based TIA.
and
Rsh =
(Rs+ Rsw)1−4γ4γ for N=4, (Rs+ Rsw)1−8γ8γ for N=8.
(4.3)
The analytical expression in (4.1) is verified by the design and simulation of two N-path mixer-first receivers operating at fs = 1 GHz. Ideal switches with 5 Ω ON-resistance and a resistance RBare used in simulations. Fig. 4.1(c) shows the input resistance of 4-path and 8-path mixer-first receivers for different values of RB. From Fig. 4.1(c), the RB values should be close to 260 Ω and 370 Ω for N=4 and N=8 respectively for 50 Ω impedance matching.
4.4 Regimes of operation
From (4.1), RB or the TIA input impedance plays a significant role in input matching. In passive mixer-first receivers, a shunt-feedback TIA is typically used1. The main amplifier in a shunt-feedback TIA can be an operational amplifier or an operational transconductance amplifier (OTA). These two TIA implementations are shown in Fig. 4.2. In [59], an opamp based TIA is considered2. An OTA based TIA is more frequently found in recent implementations of highly linear mixer-first receivers [56, 58,61,62]. In this work, a single stage OTA based TIA such as the one shown in Fig. 4.2 is considered.
1Other possibility is a common-gate amplifier based TIA.
2A detailed comparison between Opamp based TIA and OTA based TIA is discussed in Appendix A
In the OTA based TIA model shown in Fig. 4.2, RF is the feedback resistance, gm represents the transconductance of the main amplifier, and ro represents the output impedance of the OTA. The input resistance of this OTA based TIA is given by
RB= 1
RF − 1
RF(ro||RF) +gm(ro||RF) RF
−1
≈ RF
gmro||RF ≈ 1
gm + RF
gmro (4.4)
The specific RB value required for input matching can be realized either by using only g1
m or by using only gRF
mro or by using both g1
m and gRF
mro. Based on this choice of gm, the receiver operates in one of the two possible regimes. If we choose ro RF so that RB ≈ g1m, then the receiver operates in the noise-limited regime. In the noise-limited regime, the receiver exhibits poor NF but a relatively good IB-IIP3 which can be tuned by adjusting the RF value. If we choose g1
m gRmFro so that RB≈ gRmFro, then the receiver operates in the linearity-limited regime. In the linearity-limited regime, the receiver exhibits poor IB-IIP3 but a relatively good NF which can be tuned by adjusting the gm value.
4.4.1 Noise-limited regime
Fig. 4.3(a) shows an equivalent LTI model of the N-path mixer-first receiver for noise analysis. The main sources of noise in the model shown in Fig. 4.3(a) are switch resistance Rsw, shunt resistance Rsh, feedback-resistance RF and noise of the OTA. Here the effect of ro is neglected as g1
m gRmFro. The noise factor of the receiver can be computed by calculating the noise contribution of each source at the output. The noise factor is given by
NF = 1 + Rsw Rs
+Rsh Rs
Ra Rsh
2
+ γRF Rs
1 G2N +γi2n,gm
RF+Rγp 1−gmRF
2 1 4KTRs
, (4.5)
where, Ra = Rs+ Rsw, Rp= Ra||Rsh, i2n,gm is the noise current at the output of the OTA, and GN is the transfer function of the receiver which is given by
GN=
1−gmRF 1 +gmγRp
Rsh Rsh+ Ra
≈
1−RRFB 1 +γRRp
B
Rsh Rsh+ Ra
as RB≈ 1 gm
. (4.6)
For amplification, we need gmRF>1. The last two terms of (4.5) represent the noise contribution
4.4 Regimes of operation
0.5 1.5 2.5 3.5 4.5 3
6 9 12 15 18 21 24
0.5 1.5 2.5 3.5 4.5 5
8 11 14 17 20
RF (kΩ) RF (kΩ)
NF(dB) NF(dB) vn,gm= 3
√nV Hz
vn,gm= 4√nVHz
vn,gm= 2√nV Hz
vn,gm= 3√nV
Hz
vn,gm= 4√nV
Hz
vn,gm= 2√nVHz
(a)
gm γ
γRF v2n,RF
i2n,gm γ
v2n,sh Rsh v2n,s
Rs v2n,sw Rsw
Rp= Rsh||Ra
ZB =γRB = gγ
m
(b) (c)
RF< 2.5 k
RF< 3.7 k
noise-limited noise-limited
regime regime
Figure 4.3: (a) Equivalent LTI model of the receiver for noise analysis in the noise-limited regime. Variation of NF with RF for different input referred noise voltage of transconductor (b) in a 4-path mixer-first receiver, and (c) in an 8-path mixer-first receiver (Solid lines indicate simulation, while dashed lines indicate (4.5)).
of the TIA to the overall receiver noise. (4.5) is verified by the design and simulation of 4-path and 8-path mixer-first receivers. Ideal switches with 5 Ω ON-resistance, and a VCCS with an appropriate input referred noise voltage v2n,gm = i2n,gmg2
m to model the OTA are used in simulations. Moreover, the OTA is assumed to have gmro = 100 3. Fig. 4.3(b) and Fig. 4.3(c) show the variation of NF with feedback resistance RF for different v2n,gm in 4-path and 8-path receivers respectively. The NF of the receiver in this noise limited regime depends on v2n,gm and RF. The gm required for input matching is≈4 mfin a 4-path receiver and is≈2.7 mfin an 8-path receiver. The noise-limited regime in the plots shown in Fig. 4.3 is separated by a dashed line where g1
m is 10 times more than gRF
mro.
3In this work, a single stage OTA is considered where gmro is a constant. A two-stage OTA restricts the receiver to the linearity-limited regime only. [56, 58, 61, 62] use single stage OTAs.
20 100 140 180 0.8
1 1.2 1.4
0.6 60
gm (mS)
NF(dB)
20 60 100 140 180 1.6
2 2.4 2.8
1.2
gm (mS)
NF(dB)
gm>37 mS
gm>25 mS
linearity-limited regime linearity-limited regime
Analytical NF Simulated NF without TIA noise
Simulated NF Analytical NF
Simulated NF without TIA noise
Simulated NF
v2n,sh
Rsh v2n,s
Rsv2n,sw Rsw
Rp= Rsh||Ra
gm
γ i2n,gm
γ
v2n,RF
roγ
RF
gmro
(a)
(b) (c)
Figure 4.4: (a) Equivalent LTI model of the receiver for noise analysis in the linearity-limited regime. Variation of NF with gm with and without TIA noise (b) in a 4-path mixer-first receiver, and (c) in a 8-path mixer-first receiver (Solid lines indicate simulation, while dashed lines indicate (4.7)).
4.4.2 Linearity limited regime
A noise equivalent LTI model of the receiver under study is shown in Fig. 4.4(a). Using Miller theorem, the feedback resistor RFcan be modeled by resistive terminations at the input and output of the OTA. Since RFro, only ro is shown at the output of the OTA in Fig. 4.4(a). The noise voltage due to RF is added to the OTA output. From Fig. 4.4(a), the noise factor can be computed as
NF = 1 +Rsw
Rs
+Rsh
Rs
Ra
Rsh
2
+γRF
Rs
1 G2L +γi2n,gmr2o
G2L
1 4KTRs
, (4.7)
where GL=
γgmroRshRF
γRFRsh+γRaRF+ gmroRshRa
≈
gmro
1 +RRa
sh +γRRa
B
as RB≈ RF
gmro
. (4.8)
4.4 Regimes of operation
2 4 6 8 10 12 14 16 18
0 4 8 12 16 20 24 28 32
0.4 0.8 1.2 1.6 2 2.4 2.8 3.2
NF(dB) Gain(dB)andIIP3(dBm) NF(dB)
Gain NF IB-IIP3
RF (kΩ)
0 3 6 9 12 15 18 21
0 4 8 12 16 20 24 28
0.4 0.8 1.2 1.6 2 2.4 2.8 3.2 RF<2.6 kΩ
Gain NF IB-IIP3
Gain(dB)andIIP3(dBm)
RF (kΩ) noise-limited
regime
(a) (b)
RF<1.9 kΩ noise-limited
regime
Figure 4.5: Variation of NF, gain and IB-IIP3 of the receiver (implemented with inverter based TIA) with feedback resistance RF in (a) a 4-path mixer-first receiver, and (b) in an 8-path mixer-first receiver.
Using RshRa from (4.3), the above expression can be further simplified to GL≈
gmro
1 +γRRa
B
. (4.9)
(4.7) is verified by the design and simulation of 4-path and 8-path mixer-first receivers. Ideal switches with 5 Ω ON-resistance, and a VCCS to model the OTA are used in the simulations. The noise current of an OTA depends on the gm value. For simulation and verification purposes, i2n,gm is assumed to be equal to 4KTγMgm, where γM ≈ 23 is the noise proportionality constant. Fig. 4.4(b) and Fig. 4.4(c) show the variation of NF with the gm value in 4-path and 8-path mixer-first receivers respectively. The RF required for input matching in a 4-path, and an 8-path mixer-first receivers respectively are 27 kΩ and 39 kΩ. The linearity-limited regime is separated in the plots of Fig. 4.4 by a dashed line where g1
m is 10 times less than gRF
mro. From Fig. 4.4(b) and Fig. 4.4(c), one can infer that the NF of the receiver in a linearity-limited regime can be brought closer to the NF of the mixer alone. The cost of reducing the NF is the increase in power consumption to increase the gm value.
By substituting γ, Rsh and RB values from (4.2) and (4.3) in (4.6) and (4.9), one can deduce that GNGL. For a given output swing, a system with lower gain will support higher input amplitude and hence will have higher IB-IIP3. Thus, the IB-IIP3 of a receiver operating in the linearity-limited regime is smaller than the IB-IIP3 of a receiver operating in the noise-limited regime.
0.8 1 1.2 1.4 1.6 1.8 2 2.2
-20 -10 0 10 20 30 40 50
1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3
-30 -20 -10 0 10 20 30 40 50
20 60 100 140 180 20 60 100 140 180
Gain(dB)andIIP3(dBm)
NF(dB) Gain(dB)andIIP3(dBm)
NF(dB)
gm(mS) gm>44 mS gm>30 mS
gm (mS)
(a) (b)
Gain NF IB-IIP3 Gain
NF IB-IIP3
linearity-limited regime linearity-limited
regime
Figure 4.6: variation of NF, gain and IB-IIP3 of receiver (implemented with inverter based TIA) with transconductancegm in (a) a 4-path mixer-first receiver, and (b) in an 8-path mixer-first receiver
20 60 100 140 180 58
62 66 70 74 78
0.4 1.2 2 2.8 3.6 4.4 5.2 6 58
62 66 70 74 78
RF (KΩ)
SFDR(dB) SFDR(dB)
gm (mS)
gm>44mS gm>30mS RF<1.9KΩ
RF<2.6KΩ
linearity-limited regime N=8
N=4
N=8 N=4
(a) (b)
noise-limited regime
Figure 4.7: (a) variation of SFDR of a 4-path and a 8-path mixer-first receiver (implemented with inverter based TIA) withRF, (b) variation of SFDR of 4-path and 8-path mixer-first receiver (implemented with inverter based TIA) withgm