Cirrus Logic AN168 - Manual (Page 14)

AN168
14 AN168REV2
typically employ the sidetone only on the analog
interface, leaving the digital interface
sidetone-free.
If it is not possible to disable the sidetone or work
around it by using a different interface, then the
presence of the sidetone must be considered in the
design of the system. To minimize the loop gain as
well as to prevent the local mic signal from being
echoed out the speaker, it is necessary to enable the
Network Echo Canceler and to allocate Taps to it.
These are the two parameter changes necessary
from the recommendations given above.
Specifically, NECD should be set to '0', and Taps
should be set to either 47.5 ms/16 ms or
39.5 ms/24 ms. Either setting will work;
47.5 ms/16 ms is slightly more desirable since the
network path delay inside the phone is easily
handled by 16 ms of taps, allowing more taps to be
used for modeling the acoustic path.
1.4.4 Loop Gain
The presence of a network sidetone places
constraints on the total loop gain allowed in the
system. The network interface gains are
determined by matching the full scale level of the
phone to the full-scale level of the CS6422. The
speaker driver, acoustic coupling, and mic preamp
gains are determined as described earlier in this
note. The remaining gains are TVol and RVol
inside the CS6422.
Note: TGain is treated as part of the mic preamp
gain, and RGain is treated as part of the network
receive gain.
When the echo cancelers are enabled, they provide
additional loss in the loop that is equal to the ERLE
(Echo Return Loss Enhancement) of each echo
canceler. Because the network path is strong and
stable, the ERLE of the network echo canceler can
be relied upon to reduce the system loop gain.
Because the acoustic path is highly variable, the
ERLE of the acoustic echo canceler should NOT be
relied on to reduce the system loop gain.
The network echo canceler can provide, worst case,
15 dB of additional loss in the loop, provided that
the network path is linear and lossy, that is, the
signal level at NO is reflected back at NI at a lower
amplitude than it originated.
Here's an illustrative loop gain example. Assuming
that the worst-case acoustic coupling is -9 dB and
the network sidetone amplitude, here defined as the
gain between the NO and NI pins on the CS6422, is
-6 dB, and the network ERLE is 15 dB, then
maximum. This +30 dB number can be distributed
arbitrarily between RVol and TVol. The placement
of the gain does not affect the stability of the
system. Usually, the gain is distributed to provide a
balance between the transmit path volume and the
receive path volume.
1.5 Tests
In this section, we present some tests which are
useful to identify and solve system-level problems.
These tests should be performed in the listed order.
The first set of tests, Acoustic Coupling, Acoustic
Distortion, and Acoustic ERLE, should be
performed in a lab and in a car. These tests verify
the electrical and the mechanical design of the
system.
The last set of tests, which are actual call test
scenarios, are used to fine-tune the CS6422 register
settings.
NOTE: These tests should be performed in a car.
The latter tests can be performed in a lab, however,
it is likely that the set of optimum coefficients
derived in lab testing will not be optimal for car
use.
RVol TVol + 9 6 15 + + 30dB = =