
www.national.com
14
C
The 2’s complement format represents full-scale negative as
10000000 and full-scale positive as 01111111 for an 8-bit
example.
The 16 bit phase offset is set by loading the PHASE register
according to the formula PHASE = 2
16
P/2
π
, where P is the
desired phase in radians ranging between 0 and 2
π
. PHASE
is an unsigned 16-bit number. P ranges from 0 to 2
π
(1-2
-16
).
Phase dithering can be enabled to reduce the spurious sig-
nals created by the NCO due to phase truncation. This trun-
cation is unavoidable since the frequency resolution is much
finer than the phase resolution. With dither enabled, spurs
due to phase truncation are below -100 dBc for all frequen-
cies and phase offsets. Each NCO has its own dither source
and the initial state of one is maximally offset with respect to
the other so that they are effectively uncorrelated. The phase
dither sources are on by default. They are independently
controlled by the DITH_A and DITH_B bits. The amplitude
resolution of the ROM creates a worst-case spur amplitude
of -101dBc rendering amplitude dither unnecessary.
The spectrum plots in Figure 17 show the effectiveness of
phase dither in reducing NCO spurs due to phase truncation
for a worst-case example (just below F
S
/8). With dither off,
the spur is at -86.4dBFS. With dither on, the spur is below
-125dBFS, disappearing into the noise floor. This spur is
spread into the noise floor which results in an SNR of
-83.6dBFS. The channel filter’s processing gain will further
improve the SNR.
Figure 18 shows the spur levels as the tuning frequency is
scanned over a narrow portion of the frequency range. The
spurs are again a result of phase quantization but their loca-
tions move about as the frequency scan progresses. As
before, the peak spur level drops when dithering is enabled.
When dither is enabled and the fundamental frequency is
exactly at F
S
/8, the worst-case spur due to amplitude quanti-
zation can be observed at -101dBc in Figure 19.
Four Stage CIC Filter
The mixer outputs are decimated by a factor of N in a four
stage CIC filter. N is programmable to any integer between 8
and 2048. Decimation is programmed in the DEC register
where DEC = N - 1. The programmable decimation allows
the chip’s usable output bandwidth to range from about
2.6kHz to 1.3MHz when the input data rate (which is equal to
the chip’s clock rate, F
CK
) is 52 MHz. For the maximum sam-
ple rate of 78MHz, the CLC5903’s output bandwidth will
range from about 4.76kHz to 1.95MHz. A block diagram of
the CIC filter is shown in Figure 20.
The CIC filter is primarily used to decimate the high-rate
incoming data while providing a rough lowpass characteris-
tic. The lowpass filter will have a sin(x)/x response (similar to
the AGC’s CIC shown in Figure 36 on page 24) where the
first null is at F
S
/N.
The CIC filter has a gain equal to N
4
(filter decimation^4)
which must be compensated for in the “SHIFT UP” circuit
shown in Figure 20 as well as Figure 16. This circuit has a
gain equal to 2
(SCALE-44)
, where SCALE ranges from 0 to 40.
This circuit divides the input signal by 2
44
providing maxi-
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
-160
-140
-120
-100
-80
-60
-40
-20
0
Frequency Normalized to F
S
M
Figure 18. NCO Spurs due to Phase Quantization
Complex NCO Output
Phase Dither Disabled
NCO frequency swept
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
-160
-140
-120
-100
-80
-60
-40
-20
0
Frequency Normalized to F
S
M
Figure 19. Worst Case Amplitude Spur, NCO at F
S
/8
Complex NCO Output
Phase Dither Enabled
DIN
OUT
D
B
O
SCALE
S
22
Figure 20. Four-stage decimate by N CIC filter
66
Data @ F
CK
= F
S
(F
SAMPLE
)
Data @ F
CK
/N
N = DEC + 1
22-bit input to SHIFT_UP is aligned
at the bottom of the 66-bit path when SCALE=0.
Detailed Description
(Continued)