Digital Signal Processing Reference
In-Depth Information
RF section
IF section
baseband section
ceramic filter IF amp. 2
nd
IF
LNA
SAW filter
mixer
low-pass
A
D
I
0
A
agc control
band select
90
D
Q
channel select
PLL
loading factor
XT
local oscillator
fixed oscillator
frequency
control
Figure 1.7.
Architecture of a narrowband superheterodyne receiver.
followers of purely software defined radio. However, the previous considera-
tions made clear why skipping those frequency conversion and filtering steps
would impose serious demands on the linearity and dynamic range of the front-
end in a purely software-based implementation.
Some receiver implementations have managed to get rid of the external com-
ponents, though. In [Muh05] for example, the external saw filter located at the
if frequency has been replaced by a direct-sampling mixer. The discrete-time
continuous-amplitude output samples are subsequently processed by three on-
chip iir filters which are implemented using on-chip capacitor banks. A good
linearity is still achieved since the iir filters are located in front of the variable
gain amplifier and the ad converter. The dynamic range of the iir filter itself is
much less of a problem since it largely depends on some passive components
such as capacitors and cmos switches. Doing the same in software, after the
vga and the ad converter, would place you one step ahead of your competitors
concerning the reduction of power efficiency.
At the same time, the question is raised what to do with the ever-increasing
bandwidth of telecommunication devices. In previous sections, it was found
that for higher spatial efficiency, a small cell size should be combined with a
large spectral bandwidth. From this point of view, the Shannon theorem indeed
predicts that it is more advantageous to spread the total transmission power
over a wide frequency band rather than to compress it in a small one. When
bringing interferers into the game, this choice becomes much less evident: the
large spectral footprint of the transmitted signal makes it a much more chal-
lenging task to remove an accidental in-band interferer. The poor selectivity
inherent to wideband receivers, combined with a lower power spectral den-
sity (psd) of the signal-of-interest, make that classic narrowband transceiver
architectures are not the ideal candidate for porting to a wideband system.
