3-20
Additionally, the Programmable FIR filter provides for
decimation factors, R, from 1 to 16. The processing rate of
the Filter Compute Engine is PROCCLK. As a result, the
frequency of PROCCLK must exceed a minimum value to
ensure that a filter calculation is complete before the result is
required for output. In configurations which do not use
decimation, one input sample period is available for filter
calculation before an output is required. For configurations
which employ decimation, up to 16 input sample periods
may be available for filter calculation.
For real filter configurations, use Equation 11 to calculate the
number of taps available at a given input filter sample rate.
for real filters, and
for complex filters, where floor is defined as the integer
portion of a number; PROCCLK is the compute clock; f
SAMP
= the FIR input sample rate; R = Decimation Factor; SYM =
1 for symmetrical filter, 0 for asymmetrical filter; ODD# = 1
for an odd number of filter taps, 0 = an even number of taps.
Use Equation 12 to calculate the maximum input rate.
for real filters, and
for complex filters, where floor[x], PROCCLK, f
SAMP
, R =
Decimation Factor, SYM, and ODD# are defined as in
Equation 11.
Use Equation 13 to calculate the maximum output sample
rate for both real and complex filters.
The coefficients are 22 bits and are loaded using writes to
Control Words 128 through 255 (see Microprocessor Write
Section). For real filters, the same coefficients are used by I
and Q paths. If the filter is configured as a symmetric filter
using Control Word 17, Bit 9, then coefficients are loaded
starting with the center coefficient in Control Word 128 and
proceeding to last coefficient in Control Word 128+n. The
filter symmetry type can be set to even or odd symmetric,
and the number of filter coefficients can be even or odd, as
illustrated in Figure 20. Note that complex filters can also be
realized but are only allowed to be asymmetric. Only the
coefficients that are used need to be loaded.
Automatic Gain Control (AGC)
The AGC Section provides gain to small signals, after the
large signals and out-of-band noise have been filtered out, to
ensure that small signals have sufficient bit resolution in the
Resampling/Interpolating Halfband filters and the Output
Formatter. The AGC can also be used to manually set the
gain. The AGC optimizes the bit resolution for a variety of
input amplitude signal levels. The AGC loop automatically
adds gain to bring small signals from the lower bits of the 26-
bit programmable FIR filter output into the 16-bit range of the
TAPS
floor PROCCLK
SYM)
(
[
–
F
SAMP
R
)
]
(
)
R
]
–
(
)
1
+
(
=
(EQ. 11A)
SYM
)
ODD#
(
TAPS
floor (PROCCLK
F
SAMP
R
(
)
R
)
2]
–
=
(EQ. 11B)
(EQ. 12A)
F
SAMP
PROCCLK
(
)
R
( )
R
floor Taps
)
[
+
[
+
=
SYM
(
)
ODD#
(
)]
1
SYM
+
(
)
]]
(EQ. 12B)
F
SAMP
PROCCLK
(
)
R
( )
[
]
R
floor Taps
)
2
( )]
+
[
]
=
(EQ. 13)
F
FIROUT
F
SAMP
(
)
R
=
EVEN SYMMETRIC
EVEN TAP FILTER
ODD SYMMETRIC
EVEN TAP FILTER
EVEN SYMMETRIC
ODD TAP FILTER
ODD SYMMETRIC
ODD TAP FILTER
C0
CN-1
C
C
C0
CN-1
C0
CN-1
COEFFICIENT
NUMBER
COEFFICIENT
NUMBER
COEFFICIENT
NUMBER
COEFFICIENT
NUMBER
C
C
ASYMMETRIC
ODD TAP FILTER
C0
CN-1
COEFFICIENT
NUMBER
C
ASYMMETRIC
EVEN TAP FILTER
C0
CN-1
COEFFICIENT
NUMBER
C
COMPLEX FILTERS
C
Q(0)
C
I(N-1)
COEFFICIENT
NUMBER
I
REAL COEFFICIENT VALUE
C
C
Q(N-1)
REAL FILTERS
Definitions:
Even Symmetric: h(n) = h(N-n-1) for n = 0 to N-1
Odd Symmetric: h(n) = -h(N-n-1) for n = 0 to N-1
Asymmetric:
A filter with no coefficient symmetry.
Even Tap filter:
A filter where N is an even number.
Odd Tap filter:
A filter where N is an odd number.
Real Filter:
A filter implemented with real coefficients.
Complex Filters: A filter with quadrature coefficients.
FIGURE 20. DEMONSTRATION OF DIFFERENT TYPES OF
DIGITAL FIR FILTERS CONFIGURED IN THE
PROGRAMMABLE DOWNCONVERTER
C
V
C
Q
C
I
C0
CN
HSP50214B