GC4016 Datasheet, PDF (74/83 Pages) Texas Instruments – MULTI-STANDARD QUAD DDC CHIP

GC4016 MULTI-STANDARD QUAD DDC CHIP
DATA SHEET REV 1.0
7.12.1 UMTS Filter Response
Figure 23 shows the overall response using specially tuned CFIR and PFIR filters., The PFIR filter is a specially optimized
root-raised-cosine 63 tap filter with an alpha of 0.22. The filter is designed to have less than -50dB of inter symbol interference
(ISI) noise.
2.5MHz
Figure 23. Frequency Response for the Example UMTS Application
These coefficients are available by e-mail or over the web as cfir_umts.taps and pfir_umts.taps. The CFIR_SUM for
cfir_umts.taps is 58827. The PFIR_SUM for pfir_umts.taps is 61821.
7.12.2 Oversampling Using the Resampler
This example assumes the input sample rate is equal to 4*N*B, where N is the decimation in the CIC filter (See Section 5.6)
and B is the UMTS symbol rate (3.84 MHz). The outputs are two complex samples per symbol (7.68MHz) with the resampler set
to interpolate by unity (no interpolation). The output rate can be doubled or quadrupled as desired by changing the resampler
ratio. The resampler uses the configuration res_8x64_60. (See Section 7.7). This configuration introduces 0.03dB of passband
ripple and -62dB inband noise, neither being large enough to effect UMTS demodulation.
The resampler ratio for two samples per symbol (7.68MHz) is 0x04000000 and for four samples per symbol (15.36MHz) it
is 0x02000000.
Sample rates other than 4*N*B can be accepted by adjusting the PFIR filter to match the desired RRC filter and by using the
resampler to adjust the final sample rate to be exactly 2X or 4X the symbol rate.
7.12.3 Gain
The example configuration assumes a CIC decimation of N=4, which corresponds to an ADC clock rate (CK) of 61.44 MHz.
The values of SCALE and BIG_SCALE must be chosen to satisfy: (SHIFT + SCALE + 6 × BIG_SCALE) ≤ (62 – 5log₂N) . N is 4
and SHIFT=4, so (SCALE + 6 × BIG_SCALE) ≤ 48 , which is satisfied by setting SCALE=5 and BIG_SCALE=7, which are their
maximum values. If other values of N are chosen, then SCALE and BIG_SCALE need to be modified as necessary. The overall
gain is adjusted using COARSE, FINE_GAIN and FINAL_SHIFT. The overall gain is:
GAIN
=



(
----------¹-----------)
NZEROS + 1
N⁵
2^(SHIFT
+
SCALE
+
6
×
BIG_SCALE
–
62
)
(

2^COARSE
)
(
C_---F_--I_--R_--__--S_---U_--M_---)
65536
(
P_---F_--I_-R_---__--S_--U_---M_--
65536
)
(
F_---I_-N_---E_--__--G_---A_---I_-N_--)
1024
(
3----2---7---6R---8-E---×^{-S---_-N--S--D-U----E-M--L----A----Y---}
)
(
2
FINAL_SHIFT
)
where NZEROS=0, N=4, SHIFT=4, SCALE=5, BIG_SCALE=7, COARSE=1, CFIR_SUM=58827, PFIR_SUM=61821,
© GRAYCHIP,INC.
- 69 -
August 27, 2001
This document contains information which may be changed at any time without notice