LF2242
DEVICES INCORPORATED
DEVICES INCORPORATED
12/16-bit Half-Band Interpolating/
Decimating Digital Filter
DESCRIPTION
The
LF2242
is a linear-phase, half-
band (low pass) interpolating/
decimating digital filter that, unlike
intricate analog filters, requires no
tuning. The LF2242 can also signifi-
cantly reduce the complexity of
traditional analog anti-aliasing pre-
filters without compromising the
signal bandwidth or attenuation. This
can be achieved by using the LF2242
as a decimating post-filter with an
A/D converter and by sampling the
signal at twice the rate needed.
Likewise, by using the LF2242 as an
interpolating pre-filter with a D/A
converter, the corresponding analog
reconstruction post-filter circuitry can
be simplified.
The coefficients of the LF2242 are
fixed, and the only user programming
required is the selection of the mode
(interpolate, decimate, or pass-
through) and rounding. The asyn-
chronous three-state output enable
control simplifies interfacing to a bus.
Data can be input into the LF2242 at a
rate of up to 40 million samples per
second. Within the 40 MHz I/O limit,
the output sample rate can be one-
half, equal to, or two times the input
12/16-bit Half-Band Interpolating/
Decimating Digital Filter
LF2242
FEATURES
u
40 MHz Clock Rate
u
Passband (0 to 0.22f
S
)
Ripple:
±0.02
dB
u
Stopband (0.28f
S
to 0.5f
S
)
Rejection: 59.4 dB
u
User-Selectable 2:1 Decimation or
1:2 Interpolation
u
12-bit Two’s Complement Input
and 16-bit Output with
User-Selectable Rounding, 8- to
16-Bits
u
User-Selectable Two’s Complement
or Inverted Offset Binary Output
Formats
u
Three-State Outputs
u
Replaces TRW/ Raytheon/
Fairchild TMC2242
u
Package Styles Available:
• 44-pin PLCC, J-Lead
• 44-pin PQFP
sample rate. Once data is clocked in,
the 55-value output response begins
after 7 clock cycles and ends after 61
clock cycles. The pipeline latency
from the input of an impulse response
to its corresponding output peak is 34
clock cycles.
The output data may be in either
two’s complement format or inverted
offset binary format. To avoid
truncation errors, the output data is
always internally rounded before it is
latched into the output register.
Rounding is user-selectable, and the
output data can be rounded from 16
bit values down to 8 bit values.
DC gain of the LF2242 is 1.0015
(0.0126 dB) in pass-through and
decimate modes and 0.5007 (–3.004
dB) in interpolate mode. Passband
ripple does not exceed
±0.02
dB from
0 to 0.22f
S
with stopband attenuation
greater than 59.4 dB from 0.28f
S
to
0.5f
S
(Nyquist frequency). The
response of the filter is –6 dB at 0.25f
S
.
Full compliance with CCIR Recom-
mendation 601 (–12 dB at 0.25f
S
) can
be achieved by cascading two devices
serially.
LF2242 B
LOCK
D
IAGRAM
TCO
RND
2–0
3
3
55-TAP
FIR
FILTER
ROUND
AND LIMIT
CIRCUIT
SI
11–0
12
INTERPOLATION 12
CIRCUIT
16
DECIMATION
CIRCUIT
16
SO
15–0
3
3
CLK
TO ALL REGISTERS
OE
INT DEC SYNC
Video Imaging Products
1
08/16/2000–LDS.2242-K
LOGIC [ LOGIC DEVICES INCORPORATED ]
