Digital Signal Processing Reference
In-Depth Information
Also interesting to note is that the sidelobe levels outside the passband of the
main lobe are reduced to below
10 dB. This implies that the susceptibility
for narrowband out-of-band interference is improved by an additional factor of
10 dB, on top of the suppression ratio already offered by the duty cycle of the
receive window. In theory, dynamically controlling the shape of the receive
window offers some perspectives to position one of the nulls of the receiver
response at a frequency point with high interferer power. However, the practical
implementation of this method can be very complicated, for only a marginal
revenue and a marginal flexibility. To conclude, observe that the spectrum plot
of Figure 6.4 shows a decreasing trend line. This is due to the limited frequency
response of the lqfp-32 package that was used for the test setup.
6.3
RF input stage
The architecture of the prototype chip has already been introduced in Figure
6.1. The circuit implementation includes a two-way i/q signal path, of which
each part is constructed of a window block, a pulse-to-baseband downcon-
version mixer, a baseband variable gain amplifier and a set of analog output
drivers. The most performance-critical sections are located in the rf input of
the receiver. Figure 6.5 shows the basic structure of the downconversion mixer
used in the prototype version (the supporting biasing circuits are omitted). The
mixer is based on a double-balanced cmos Gilbert cell [Gil68] with modified
input stage. Remark that the differential input signal is injected in parallel with
the current sources of the mixer. The cascode transistors between the injection
point and the mixer switches have a twofold purpose.
Their first task is to provide a low input impedance, as seen from the entry point
of the rf antenna signal. The result is a wide input bandwidth, while most of
the incident rf current is effectively forced to flow through the switches of
the mixer. The second goal of the cascoded input stage is to isolate the mixer
and the receive window circuitry from the input of the receiver. This is to pre-
vent that lo-spurs and clock harmonics become unintentionally radiated by
the receive antenna. The differential output of the mixer is shorted by a ca-
pacitor, which is part of a low-pass filter and plays an important role in the
pulse-to-baseband conversion process (see Section 5.1). The resistive part of
this low-pass filter is formed by the real part of the mixer output impedance.
Part of this output impedance can be accounted to the pmos current sources on
the top of the circuit, while another part is caused by the effective resistance of
the four cross-coupled switches during their crossover period. 6
6 The actual value of the crossover impedance depends on the amplitude, frequency and slope of the lo-
signal applied to the mixer switches and should be determined using transient simulations.
Search WWH ::




Custom Search