The Digital Receiver

The digital receiver consists of a digital local oscillator, digital mixer, and a decimating
lowpass filter. Digital samples from the ADC are split into two paths
and applied to a pair of digital mixers. The mixers have digital local oscillator inputs of
a quadrature signal�"that is, sine and cosine�"to enable the sampled IF to be mixed
down to a lower frequency, usually positioned around 0 Hz (DC).

This process converts the digitized signal from a real to a complex signal, that is, a
signal represented by its I and Q phase components. Because the signal is represented
by two streams, the I and Q could be decimated by a factor of 2 at this point. However,
because the down-shifted signal is usually processed for a single channel selection at
this stage, the two decimation factors may be combined.
The LO waveform generation may be achieved by a number of different options�"for
example, by a Numerically Controlled Oscillator (NCO), also referred to as a Direct
Digital Synthesizer (DDS)�"if digital-to-analog converters are used on the I and Q outputs.
In this process, a digital phase accumulator is used to address a lookup table
(LUT), which is preprogrammed with sine/cosine samples. To maintain synchronization,
the NCO is clocked by the ADC sampling/conversion clock.
The digital samples (sine/cosine) out of the local oscillator are generated at a sampling
rate exactly equal to the ADC sample clock frequency fs. The sine frequency is
programmable from DC to fs/2 and may be 32 bits. By the use of programmable phase
advance, the resolution is usually sub-Hertz. The phase accumulator can maintain precise
phase control, allowing phase-continuous switching. The mixer consists of two
digital multipliers. Digital input samples from the ADC are mathematically multiplied
by the digital sine and cosine samples from the LO. Because the data rates from the two
mixer input sources match the ADC sampling rate (fs), the multipliers also operating at
the same rate produce multiplied output product samples at fs. The I and Q outputs of
the mixers are the frequency downshifted samples of the IF. The sample rate has not
been changed; it is still the sample rate that was used to convert the IF.
The precision available in the mixing process allows processing down to DC (0 Hz).
When the LO is tuned over its frequency range, any portion of the RF signal can be mixed
down to DC; in other words, the wideband signal spectrums can be shifted around 0 Hz,
left and right, by changing the LO frequency. The signal is now ready for filtering.
The decimating lowpass filter accepts input samples from the mixer output at the
full ADC sampling frequency, fs. It uses digital signal processing to implement a finite
impulse response (FIR) transfer function. The filter passes all signals from 0 Hz to a
programmable cutoff frequency or bandwidth and rejects all signals higher than that
cutoff frequency. The filter is a complex filter that processes both I and Q signals from
the mixer. At the output either I or Q (complex) values or real values may be selected.
An example will illustrate the processes involved in the digital receiver function (see
Figure 3.14). The bandwidths�"that is, number of channels sampled and digitized�"in
the sample may be outside a handset power budget; a practical design may convert
only two or three channels. A30 MHz (6 by 5 MHz W-CDMARF channels) bandwidth
signal has been sampled and digitized at an IF of 120 MHz.
It is required to process the channel occupying 120 to 125 MHz. When the LO is set
to 122.5 MHz, the channel of interest is shifted down to a position around 0 Hz. When
the decimating (lowpass) filter is set to cut off at 2.5 MHz, the channel of interest may
be extracted.
To set the filter bandwidth, you must set the decimation factor. The decimation factor
is a function of both the output bandwidth and output sampling rate. The decimation
factor, N, determines both the ratio between input and output sampling rates and
the ratio between input and output bandwidths.
In the example in Figure 3.15, the input had a 30 MHz bandwidth input with a
±2.5 MHz bandwidth output. The decimation factor is therefore 30 MHz/2.5 MHz�"
that is, 12.
Digital receivers are divided into two classes, narrowband and wideband, defined
by the range of decimation factors. Narrowband receivers range from 32 to 32,768 for
real outputs, wideband receivers 1 to 32. When complex output samples are selected,
the sampling rate is halved, as a pair of output samples are output with each sample
clock. The downconverted, digitized, tuned around 0 Hz, filtered channel (bandwidth
= 5 MHz) now exists as minimum sample rate I and Q bit streams. In this form it is now
ready for baseband recovery and processing. 77