
AD8551/AD8552/AD8554
Rev. C | Page 15 of 24
+
A
B
B
B
C
M2
V
IN+
V
NB
C
M1
V
OA
–B
A
V
NA
Ф
B
Ф
A
A
A
V
OSA
Ф
B
Ф
A
V
OUT
V
IN–
0
Figure 50. Auto-Zero Phase of the AD855x
Amplification Phase
When the φB switches close and the φA switches open for the
amplification phase, this offset voltage remains on C
M1
and,
essentially, corrects any error from the nulling amplifier. The
voltage across C
M1
is designated as V
NA
. Furthermore, V
IN
is
designated as the potential difference between the two inputs to
the primary amplifier, or V
IN
= (V
IN+
V
IN
). Thus, the nulling
amplifier can be expressed as
[ ]
[ ]
(
t
V
V
A
t
V
OSA
IN
A
OA
]
[
)
[ ]
t
V
B
NA
A
=
(3)
+
A
B
B
B
C
M2
V
IN+
V
NB
C
M1
V
OA
–B
A
V
NA
Ф
B
Ф
A
A
A
V
OSA
Ф
B
Ф
A
V
OUT
V
IN–
0
Figure 51. Output Phase of the Amplifier
Because φA is now open and there is no place for C
M1
to
discharge, the voltage (V
NA
), at the present time (t), is equal to
the voltage at the output of the nulling amp (V
OA
) at the time
when φA was closed. If the period of the autocorrection switching
frequency is labeled t
S
, then the amplifier switches between
phases every 0.5 × t
S
. Therefore, in the amplification phase
[ ]
t
=
S
NA
NA
t
V
V
2
1
(4)
Substituting Equation 4 and Equation 2 into Equation 3 yields
[ ]
[ ]
[ ]
t
A
S
OSA
A
A
OSA
A
IN
A
OA
B
t
V
B
A
V
A
V
A
V
+
+
=
1
2
1
(5)
For the sake of simplification, assume that the autocorrection
frequency is much faster than any potential change in V
OSA
or
V
OSB
. This is a valid assumption because changes in offset
voltage are a function of temperature variation or long-term
wear time, both of which are much slower than the auto-zero
clock frequency of the AD855x. This effectively renders V
time invariant; therefore, Equation 5 can be rearranged and
rewritten as
[ ]
t
[ ]
t
(
)
A
OSA
A
A
OSA
+
A
A
IN
A
OA
B
V
B
A
V
1
B
A
V
A
V
+
+
=
1
(6)
or
[ ]
t
[ ]
t
+
+
=
A
OSA
B
IN
A
OA
V
1
V
A
V
(7)
From these equations, the auto-zeroing action becomes evident.
Note the V
OS
term is reduced by a 1 + B
A
factor. This shows how
the nulling amplifier has greatly reduced its own offset voltage
error even before correcting the primary amplifier. This results
in the primary amplifier output voltage becoming the voltage at
the output of the AD855x amplifier. It is equal to
[ ]
[ ]
(
)
OSB
IN
B
OUT
V
V
A
V
NB
B
V
B
+
+
=
(8)
In the amplification phase, V
OA
= V
NB
, so this can be rewritten as
[ ]
t
[ ]
t
[ ]
t
+
+
+
+
=
A
OSB
B
IN
A
B
OSB
B
IN
B
OUT
V
V
A
B
V
A
V
A
V
1
(9)
Combining terms,
[ ]
t
[ ]
(
)
OSA
B
A
OSA
A
A
1
B
B
B
IN
OUT
V
V
A
B
V
B
+
A
B
A
A
V
+
+
+
=
(10)
The AD855x architecture is optimized in such a way that
A
A
=
A
B
and
B
A
=
B
OS
B
and
B
A
>> 1
Also, the gain product of A
A
B
B
B
is much greater than A
B
B
. These
allow Equation 10 to be simplified to
[ ]
[ ]
A
A
A
IN
OUT
A
B
A
t
V
t
V
(
)
OSB
OSA
V
V
+
+
≈
(11)
Most obvious is the gain product of both the primary and
nulling amplifiers. This A
A
B
B
A
term is what gives the AD855x its
extremely high open-loop gain. To understand how V
OSA
and
V
OSB
B
relate to the overall effective input offset voltage of the
complete amplifier, establish the generic amplifier equation of
(
EFF
OS
IN
OUT
V
V
k
V
,
)
+
×
=
(12)
where
k
is the open-loop gain of an amplifier and
V
OS, EFF
is its
effective offset voltage.
Putting Equation 12 into the form of Equation 11 gives
[ ]
[ ]
A
A
IN
OUT
B
A
V
V
A
A
EFF
OS
B
A
V
,
+
≈
(13)
Thus, it is evident that
A
OSB
OSA
EFF
OS
B
V
V
V
+
≈
,
(14)
The offset voltages of both the primary and nulling amplifiers
are reduced by the Gain Factor B
A
. This takes a typical input
offset voltage from several millivolts down to an effective input
offset voltage of submicrovolts. This autocorrection scheme is
the outstanding feature of the AD855x series that continues to