Global Positioning System Reference
In-Depth Information
degradation in the resulting processing is less than 2 dB; see Bastide et al. (2003).
Further, if conservative 2- or more bit sampling is utilized with proper quanti-
zation, the degradation is less than 1 dB. The minimum number of bits on most
commercial ADCs is 8 as is the case for the ADS830. Thus, in designing a GNSS
front end, it is most convenient to either utilize a hard limiter to obtain a single
bit or use a commercial ADC taking all or just a subset of the resulting bits of
each sample. It is also important to recognize that if multibit sampling is em-
ployed, then some form of gain control must be implemented to provide proper
quantization.
One might ask if the penalty for using single bit sampling is less than 2 dB, why
any front end would utilize multibit sampling and then incorporate the overhead
associated with automatic gain control? The key point to remember is that the
less-than 2 dB penalty is for the ideal case. If, for example, there exists narrow-
band interference within the GNSS L1, then single bit sampling will be captured
by the interference source and prevent GNSS processing. Thus, although the theo-
retical penalty for single-bit sampling is less than 2 dB, the nature of the operating
environment may dictate the need for multibit sampling.
The maximum sampling frequency is an interesting parameter. This frequency
needs to accommodate the bandwidth of the desired signal. Continuing to use the
ADS830 part as an example, the maximum sampling frequency is 60 MHz and
thus can provide a resulting sampling bandwidth of 30 MHz, more than sufficient
for the narrowband L1 navigation signals. However, recognize that the IF in Fig-
ure 4.2 is at 47.74 MHz, which is greater than the resulting [0-30] MHz sampled
information bandwidth. In this case, the sampling process acts as a second fre-
quency translation stage.
Although the ADC has a sampling frequency that provides an upper limit of
30 MHz on the resulting sampling bandwidth, then analog input bandwidth of this
ADC really determines what signals will be captured. For the ADS830, this value
is an impressive 300 MHz. What this means is that any frequency component
input to the ADC up to 300 MHz will be aliased according the sampling theorem.
Should the analog input bandwidth have been as high as 1.6 GHz, which is not im-
possible (see pdfserv.maxim-ic.com/en/ds/MAX104.pdf ), then it would be pos-
sible to directly sample and alias the original RF signal. Such an implementation
has been demonstrated [Akos (1997) and Akos et al. (1999)] yet there remain
many technical hurdles to overcome with such an approach. The approach out-
lined does provide the means to compute an appropriate sampling frequency and
the resulting sampling IF.
Based on the preceding discussion, the role of the final filter in the RF chain be-
comes clear. It must be a bandpass filter and limit the band to only those frequen-
cies to be preserved through the sampling process. Recognize that the aliasing
does not only occur for the desired IF, but all frequencies within the analog in-
put bandwidth of the ADC. Thus, it is critical for minimal noise that the last filter
prior to the ADC allows only those frequencies of interest and attenuates all others
within the analog input bandwidth.
Search WWH ::




Custom Search