3.0 Applications Information
(Continued)
For most applications, the outputs are AC coupled and DC
offsets are not bothersome unless large signals are applied
to the filter input. However, larger offset voltages will cause
clipping to occur at lower AC signal levels, and clipping at
any of the outputs will cause gain nonlinearities and will
change f
O
and Q. When operating in Mode 3, offsets can
become excessively large if R2 and R4 are used to make
f
CLK
/f
O
significantly higher than the nominal value, espe-
cially if Q is also high. An extreme example is a bandpass
filter having unity gain, a Q of 20, and f
CLK
/f
O
e
250 with
pin 12 tied to ground (100:1 nominal). R4/R2 will therefore
be equal to 6.25 and the offset voltage at the lowpass out-
put will be about
a
1V. Where necessary , the offset voltage
can be adjusted by using the circuit ofFigure 20. This allows
adjustment of V
OS1
, which will have varying effects on the
different outputs as described in the above equations. Some
outputs cannot be adjusted this way in some modes, how-
ever (V
OS(BP)
in modes 1a and 3, for example).
3.5 SAMPLED DATA SYSTEM CONSIDERATIONS
The MF10 is a sampled data filter, and as such, differs in
many ways from conventional continuous-time filters. An im-
portant characteristic of sampled-data systems is their ef-
fect on signals at frequencies greater than one-half the
sampling frequency. (The MF10’s sampling frequency is the
same as its clock frequency.) If a signal with a frequency
greater than one-half the sampling frequency is applied to
the input of a sampled data system, it will be ‘‘reflected’’ to
a frequency less than one-half the sampling frequency.
Thus, an input signal whose frequency is f
s
/2
a
100 Hz will
cause the system to respond as though the input frequency
was f
s
/2
b
100 Hz. This phenomenon is known as ‘‘a(chǎn)lias-
ing’’, and can be reduced or eliminated by limiting the input
signal spectrum to less than f
s
/2. This may in some cases
require the use of a bandwidth-limiting filter ahead of the
MF10 to limit the input spectrum. However, since the clock
frequency is much higher than the center frequency, this will
often not be necessary.
Another characteristic of sampled-data circuits is that the
output signal changes amplitude once every sampling peri-
od, resulting in ‘‘steps’’ in the output voltage which occur at
the clock rate (Figure 21). If necessary, these can be
‘‘smoothed’’ with a simple R–C low-pass filter at the MF10
output.
The ratio of f
CLK
to f
C
(normally either 50:1 or 100:1) will
also affect performance. A ratio of 100:1 will reduce any
aliasing problems and is usually recommended for wide-
band input signals. In noise sensitive applications, however,
a ratio of 50:1 may be better as it will result in 3 dB lower
output noise. The 50:1 ratio also results in lower DC offset
voltages, as discussed in Section 3.4.
The accuracy of the f
CLK
/f
O
ratio is dependent on the value
of Q. This is illustrated in the curves under the heading
‘‘Typical Performance Characteristics’’. As Q is changed,
the true value of the ratio changes as well. Unless the Q is
low, the error in f
CLK
/f
O
will be small. If the error is too large
for a specific application, use a mode that allows adjustment
of the ratio with external resistors.
It should also be noted that the product of Q and f
O
should
be limited to 300 kHz when f
O
k
5 kHz, and to 200 kHz for
f
O
l
5 kHz.
TL/H/10399–32
FIGURE 21. The Sampled-Data Output Waveform
17