Information Technology Reference
In-Depth Information
2
Approximated Exponential Functions
To achieve a wide dynamic range, it is necessary that the gain variation be an
exponential function of the control voltage. As MOSFET characteristics are not
exponential, different pseudo-exponential functions have been proposed [2,11,12] that
enable implementation in MOSFET circuits. The following are three such functions:
()
(
)
a
+
ax
1
2
ax
2
f
x
=
e
(1)
f
(
x
)
=
[
+
(
+
2
ax
)
]
(2)
(
)
a
ax
2
2
[
k
+
(
+
ax
)
]
()
f
x
=
(3)
2
[
k
+
(
ax
)
]
Of these, the first two functions provide a linear range of about 15dB; the third,
obtains a range of more than 90 dB, for k = 0.12 [2]. We use the third function.
3
VGA Circuit Design
3.1
Control Circuit Design
Figure 1 shows a control circuit [2] which generates (3). The numerator and
denominator of (3) are proportional to the currents I C2 and I C1 in Fig. 1, respectively.
To minimize channel length modulation, the lengths of M1 and M2 must be kept
large. The current ratio (I C2 / I C1 ) as a function of the control voltage V C is given in (4)
2
I
V
o
c
+
1
+
(
)
(
)
2
V
V
K
V
V
I
dd
TH
dd
TH
c
2
=
(4)
2
I
c
I
V
o
c
+
1
(
)
(
)
2
V
V
K
V
V
dd
TH
dd
TH
which is equivalent to (3) with
I
1
o
k
=
and
a
=
(
)
(
)
2
V
V
K
V
V
dd
TH
dd
TH
Equation (4) is derived by assuming V DD = V SS , with V C in the range (V SS + V thn ) to
(V DD -│V thp │). If V DD and V SS are V supply and 0 V, respectively, then V C stays in the
range V thn to (V DD -│V thp │). Circuit simulation results of the circuit of Fig. 1 are
shown in Fig. 2, which shows that a 90dB range of √(I C2 /I C1 ) is achieved.
3.2
Proposed VGA Design
Figure 3 shows the circuit of the proposed variable gain amplifier (VGA). The circuit
includes a common-mode feedback (CMFB) circuit [2]. In the figure, the core
amplifier consists of the differential pair (M 9 and M 12 ) and diode connected loads
(M 10 and M \11 ). The sizes of M 9 and M 12 are equal, and the sizes of M 10 and M 11 are
equal. The differential gain of this amplifier cell is (g m-M9 /g m-M10 ), with the assumption
that the R out of the current sources is high. Note that this configuration is different
Search WWH ::




Custom Search