Global Positioning System Reference
In-Depth Information
ADC, then high-precision ADCs are required because precise amplitude informa-
tion is essential to these processes. Time domain transversal filter or frequency
domain FFT techniques are commonly used for frequency excision. Typically these
ADCs are 12-bit for military and 10-bit for commercial frequency excision applica-
tions. In these cases, both the quantization noise and clipping noise are negligible.
The vast majority of GPS receivers use fewer than 4 bits in the ADC because
there is little signal degradation improvement beyond 3 bits. With appropriate sam-
pling ( R S
2 B S ), the ADC degradations for wideband Gaussian noise are 1.96 dB for
1-bit, 0.55 dB for 2-bit, and 0.16 dB for 3-bit ADCs [1]. This assumes antialiasing
filtering that suppresses the RMS noise levels well below the ADC degradation levels
at and beyond B S . By contrast, with undersampling ( R S
2 B P ), the ADC degrada-
tions for wideband Gaussian noise with mismatched sample rates result in degrada-
tions of approximately 3.5 dB for 1-bit, 1.2 dB for 2-bit, and 0.6 dB for 3-bit ADCs
[2]. These results were based on a P(Y) code receiver design with passband band-
width of twice the spreading code rate ( B P
=
2 R C ) and an optimum AGC signal
amplitude into the ADC. No AGC is required for the 1-bit (limiter) case. Assuming a
typical ADC voltage reference of
=
5 VDC (a 10-V peak-to-peak ADC range), the
optimum RMS level outputs from the AGC into the ADC to achieve these minimum
ADC signal degradations for the P(Y) code receiver example are approximately
12.5V for 2-bit and 7.7V for 3-bit ADCs. A second example in [2] illustrates the
same mismatched sample rate ( R S
+
/
=
2 B P ) for a wideband C/A code receiver design
5 R C . In this case, the degradations are somewhat reduced to approxi-
mately 2.3 dB for 1-bit, 0.7 dB for 2-bit, and 0.3 dB for 3-bit ADCs. The 1-bit ADC
is clipped all of the time. The approximate RMS level outputs from the AGC into the
ADC to achieve these minimum ADC signal degradations for this C/A-code example
(still assuming a 10-V peak-to-peak ADC voltage reference) are approximately
10.1V for 2-bit and 5.6V for 3-bit ADCs. This illustrates that the optimum AGC lev-
els for the 2-bit ADC is clipped a high percentage of the time and the 3-bit ADC is
seldom clipped. This is in accordance with the reduction in the quantization noise as
the number of ADC bits is increased. Obviously, both designs suffer additional sig-
nal degradations due to aliasing noise caused by undersampling. Reference [2] also
includes 4-bit and 5-bit ADC degradations (not discussed here) that serve to illus-
trate the diminishing returns for more quantization precision in GPS applications
that do not perform digital frequency excision or frequency domain processing.
They also illustrate that the AGC level must be adapted to the antialiasing filtering,
the sample rate, and the number of bits in the ADC in order to achieve minimum sig-
nal degradation in the presence of wideband Gaussian noise.
The preferred low-precision ADC embodiment in a GPS receiver is the nonuni-
form 2-bit quantization design included in Figure 6.1. This ADC design is adapted
from [3, 4] for GPS applications because of its substantial processing gain in the
presence of CW interference plus thermal noise. 2 As a result of CW jamming, the sta-
tistics of the zero crossings of the signal are no longer determined by a combination
of thermal noise and the GPS signals buried in this random noise, but become domi-
with B P
=
2.
H. Logan Scott originally adapted Amoroso's nonuniform ADC design [3] for military GPS receivers. This
design was first used in the TI 4XOP family of military GPS receiver designs, Texas Instruments, Inc., 1985.
Search WWH ::




Custom Search