Biomedical Engineering Reference
In-Depth Information
results in an appropriate frequency response may be selected. Assuming that the switching
frequency allows for a large oversampling of the desired passband,
z
can be approximated
by the continuous Laplace term (1
sT
); then
1
H
/C
S
)
1
z
1
sT
sT
(
1
C
H
Φ
in
odd;Φ
out
odd
(
z
)
→
H
(
s
)
Comparing this to the continuous frequency-domain transfer function of a simple
RC
low-
pass
fi
filter yields
1
1
)
H
(
s
)
(
s
/
ω
1
where
ω
1
1/
RC
. Then a cuto
ff
frequency of
f
3dB
1/2
π
RC
may be obtained through a
capacitance ratio of
C
C
1
3dB
T
H
S
2
π
f
The analogy to an
RC
low-pass
filter is not a mere coincidence, because the DPDT
switches together with C
S
constitute the parallel switched-capacitor resistor realization of
Figure 1.30
f
, whose value is given by
fi
R
equivalent
T
C
S
The capacitance of C
S
must be computed to comply with a desired input impedance, which
is dependent on the sampling frequency. It must be noted that because of the fact that com-
mon-mode signals are not sampled, their spectral content may well exceed the sampling
frequency without encountering aliasing.
It may be seen from the analysis presented above that after charge reorganization has
been achieved during an even-phase interval, the output voltage is held as long as the sam-
pling capacitor does not bring a new sample in contact with the hold capacitor. This prop-
erty may be used to reject stimulation artifacts (i.e., high-amplitude spikes caused by
currents caused by a pulse generator intended to cause tissue stimulation) by extending the
even-phase interval, making it slightly longer than the stimulation pulse to be rejected. This
technique e
cation and
processing circuitry, following the instrumentation stage, allowing for the immediately con-
secutive detection of biopotentials.
Although switched-capacitor instrumentation stages are not very common in patient
monitors, they are often used as the core of biopotential ampli
ff
ectively isolates the stimulation artifact from the high-gain ampli
fi
ers in implantable devices
(e.g., pacemakers). In addition, many modern analog signal processing applications rely
on switched-capacitor sampled-data processing techniques implemented through the use
of CMOS charge manipulation circuits. In these applications, CMOS application-speci
fi
c
ICs (ASICs) contain switches and capacitors that are used as an economical means of mass
producing sophisticated signal processing functions, such as ampli
fi
fi
cation, analog arith-
metic, nonlinear functions, and
filtering [Allen and Sanchez-Sinencio, 1984].
In the circuit of Figure 1.31, IC3, a monolithic charge-balanced dual switched-capaci-
tor instrumentation building block (Linear Technology's LTC1043), implements all of the
required charge manipulation functions. Within this integrated circuit, a nonoverlapping
clock controls two DPDT CMOS switch sections. If the switched-capacitor stage IC3
would be connected directly to a di
fi
erential biopotential source detected by small-area
surface electrodes, and the sampling frequency is chosen to be 100 times that of the high-
est spectral component of the signal, the optimal value of the sampling capacitor would
result in the picofarad range to present an input impedance in the gigaohm range.
ff
Search WWH ::
Custom Search