Digital Signal Processing Reference
In-Depth Information
Error vector magnitude
determines the noise
figure of the receiver.
Return-to-zero signal is
caused by offset comp-
ensation in the VGA.
DC-offset
DC-offset caused by
window circuit, located
after the downconversion
mixer. Result: in-band
spurs and clipping.
Figure 6.9.
qpsk constellation result captured from a live measurement of the
pulse-based prototype receiver. Remark the dc voltage offset of the con-
stellation center. For weak signals and high vga gains, this causes clip-
ping further on in the receive chain which in turn reduces the sensitivity
of the receiver.
reference constellation point is defined as the error vector. The error vector
magnitude (evm) is obtained by scaling the rms power of the error vector to
the rms power of the received qpsk signal. In the absence of channel noise,
the evm can be used to measure the performance of the receiver front-end.
This topic is discussed in more detail by [Has97].
During operation over a live wireless link, there is a direct relationship be-
tween thermal channel noise, the evm and the raw ber performance of the
receiver. The situation becomes a little bit more complicated for narrowband
interference: if the baseband qpsk symbol stream is first patched up in an issr
decoder before the demapping stage, the ber will depend on two other factors:
the desensitization of the receiver as a result of the back-off from the ideal op-
erating point and the equivalent amount of noise power added to the signal
due to the removal of the affected frequency bands. The final figure here de-
pends on a lot of factors (such as the gain setting of the baseband section, the
bit depth and loading factor of the adc), which makes that numerical simula-
tions are more appropriate than complex theoretical calculations. The author
decided that such high-level system simulations are out of the scope of this
work, though.
Also noteworthy to mention is the return-to-zero phase between two consecu-
tive samples of the receiver (Figure 6.9). This behaviour is caused by the reg-
ular interventions of the differential offset compensation circuit in the signal
