IEEE Solid-States Circuits Magazine - Winter 2021 - 39

in-band insertion loss, which eats into
the RFIC noise budget.
This architecture relaxes the RFIC
linearity requirement at the expense
of tighter noise requirements. To improve the NFRx, the mixer is preceded
by one stage of voltage amplification;
the gain of the LNA suppresses the
mixer NF (NFMX):

NFRx = NFLNA + 6NFMX - 1@
.(2)
G LNA

In [10], the first stage is a commongate LNA with transconductance
boosting to improve the noise figure.
The node n1 in Figure 9 is a high-impedance node and, therefore, is sensitive to the blocker signal. The SAW
filter rejects the OBB at 20 MHz offset
by ~20-dB, attenuating it to the level
of the 3-MHz IBB.
In the SAW-less [11] approach, described in Figure 10, the circuits in
the Rx signal path must handle the
0-dBm blocker. Since the supply voltage constrains the maximum voltage
swing, the trick lies in processing
the desired and blocker signal in the
" current domain " until the LPF filters
the blocker. Both [10] and [11] employ
a current-driven passive mixer, followed by a transimpedance amplifier (TIA). However, in [11], the mixer
presents a small input impedance on
node N1. The mixer input impedance
is dominated by the virtual ground
node of the succeeding TIA; therefore, for all frequency range where
the loop gain of the TIA operational
amplifier is high, the input impedance of the TIA (ZIN, TIA) is low. The
blocker does not compress the first
stage, now called a low-noise transconductance amplifier (LNTA), and is
filtered by the TIA before it reaches
the ADC.
The absence of a voltage amplification before the mixer makes this circuit more sensitive to the mixer noise
profile. The noise margin afforded by
avoiding the in-band insertion loss of
the SAW filter can be assigned to mixer noise. Effectively, in the SAW-less
approach, the RFIC linear requirement is relaxed at the expense of
lower NF. Another subtle point in this

circuit is that ZIN, TIA has a high-pass
frequency response. For power-sensitive mobile applications where the
unity-gain bandwidth of the TIA is
determined by the signal bandwidth
(200 kHz in the case of the 2G), ZIN, TIA
could begin to rise at a 20-MHz offset,
resulting in linearity degradation. To
circumvent this, a current mode filter, CLPF, RLPF, is placed between the
signal and blocker frequency. For the
blocker signal, the CLPF presents a low
impedance, and the signal circulates
in the capacitor.
In commercially available smartphones, both SAW-based and SAWless approaches have been used. The
SAW-based designs are prevalent in

high-end smartphones, in which the
additional cost of the SAW filter can
be absorbed with the bill of material.

Tx VCO Phase Noise
A commonly encountered coexistence
problem, also called the near-far problem, is described in Figure 11. Users
UEA and UEB operate simultaneously
at different frequencies (FC1 and FC2)
and communicate with different BSs
(BSA and BSB, respectively). Both users are " far " away from their BSs but
" near " one another. The Rx of UEB is
operating at sensitivity, and the Tx of
UEA is transmitting the maximum permissible power level. In this scenario,
the UEB Rx could be saturated by the

MX
INP

LNA

Q
Q

n1
To
TIA

VB2

VB4

VB1
M1

VB3

INP

-1

FIGURE 9: The SAW-based 2G Rx [10]. TIA: transimpedance amplifier; INP: input.

MX
INP

TIA

n1
VB2
INP

BBI
BBI-

Q I

C
RLPF
+
+

M2
CLPF

M1
Low
ZIN

RLPF

R
-

BBI

-
BBI

+
+
R
C

FIGURE 10: The SAW-less 2G Rx [11].

IEEE SOLID-STATE CIRCUITS MAGAZINE

W I N T E R 2 0 2 1

IEEE Solid-States Circuits Magazine - Winter 2021

Table of Contents for the Digital Edition of IEEE Solid-States Circuits Magazine - Winter 2021