This application note describes a design procedure
coupled with some testing procedures to enable a
system designer to implement a low cost
full-duplex cellular hands-free system for cars
using the CS6422 Enhanced Echo Cancelling IC.
This application note focuses on the design of the
acoustic path, that is, the path between the acoustic
output (AO) and the acoustic input (APO) of the
CS6422. The acoustic path contains the speaker
driver, the speaker, the air path between the
speaker and the microphone, the microphone, and
the microphone preamp.
Additionally, a suggested set of CS6422
configuration parameters is presented as well as
some system-level tests that can be used to
optimize the parameters for a particular
environment.
1. DESIGN PROCESS AND
CONSIDERATIONS
There are four parts to the hands-free design
process: mechanical design, electrical design, echo
canceler coefficient optimization, and testing. This
note will investigate all four.
1.1Design Flow
The design flow for full-duplex systems is as
follows:
1) Design the mechanical and electrical systems
for low distortion, specifically less than
2% THD across frequency.
2) Install the equipment in the target test system,
usually a car.
3) Tweak the mic preamp gain to achieve -9 dB
acoustic coupling.
4) Load the starting point example CS6422 register configuration.
5) Perform parameter optimization tweaking
tests.
6) Test under actual driving conditions. If necessary, modify speaker/mic placement and test
again.
For all product questions and inquiries contact a Cirrus Logic Sales Representative.
To find the one nearest to you go to www.cirrus.com
IMPORTANT NOTICE
Cirrus Logic, Inc. and its subsidiaries ("Cirrus") believe that the information contained in this document is accurate and reliabl e. However, the informatio
is subject to change without notice and is provided "AS IS" without warranty of any kind (express or implied). Customers are advised to obtain the lates
version of relevant information to verify, before placing orders, that information being relied on is current and complete. All products are sold subject t
the terms and conditions of sale supplied at the time of order acknowledgment, including those pertaining to warranty, indemnification, and limitation o
liability. No responsibility is assumed by Cirrus for the use of this information, including use of this information as the basis for manufacture or sale of an
items, or for infringement of patents or other rights of third parties. This document is the property of Cirrus and by furnishing this information, Cirru
grants no license, express or implied under any patents, mask work rights, copyrights, trademarks, trade secrets or other intellectual property rights
Cirrus owns the copyrights associated with the information contained herein and gives consent for copies to be made of the information onl y for us
within your organization with respect to Cirrus integrated circuits or other products of Cirrus. This consent does not extend to other copying such a
copying for general distribution, advertising or promotional purposes, or for creating any work for resale.
CERTAIN APPLICATIONS USING SEMICONDUCTOR PRODUCTS MAY INVOLVE POTENTIAL RISKS OF DEATH, PERSONAL INJURY, OR
SEVERE PROPERTY OR ENVIRONMENTAL DAMAGE (“CRITICAL APPLICATIONS”). CIRRUS PRODUCTS ARE NOT DESIGNED, AUTHORIZED
OR WARRANTED FOR USE IN AIRCRAFT SYSTEMS, MILITARY APPLICATIONS, PRODUCTS SURGICALLY IMPLANTED INTO THE BODY,
UTOMOTIVE SAFETY OR SECURITY DEVICES, LIFE SUPPORT PRODUCTS OR OTHER CRITICAL APPLICATIONS. INCLUSION OF CIRRUS
PRODUCTS IN SUCH APPLICATIONS IS UN DE RS TOO D TO BE FULLY AT THE CUS TOMER ’ S RISK AND CIRRU S DI S CLA IMS AND MA KES NO
WARRANTY, EXPRESS, STATUTORY OR IMPLIED, INCLUDING THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
PARTICULAR PURPOSE, WITH REGA RD TO A NY CI R RUS PRODUCT TH A T IS US E D IN SUCH A MAN NER. IF THE CUSTOM ER OR
CUSTOMER’S CUSTOMER USES OR PERMITS THE USE OF CIRRUS P R O DUCTS IN CRITICAL APPLICATIONS, CUSTOMER AGREES, BY
SUCH USE, TO FULLY INDEMNIFY CIRRUS, ITS OFFICERS, DIRECTORS, EMPLOYEE S, DI S TRIB UTORS AND OTHER AGENTS FROM ANY
ND ALL LIABILITY, INCLUDI NG ATTORNEY S ’ FEES AND COS TS, THAT MA Y RESULT FROM OR ARISE IN CONNECTION WITH THESE USES.
Cirrus Logic, Cirrus, and the Cirrus Logic logo designs are trademarks of Cirrus Logic, Inc. All other brand and product names in this document may b
trademarks or service marks of their respective owners.
AN168REV23
AN168
1.2Mechanical Design
The performance of full-duplex hands-free designs
is strongly influenced by the mechanical hardware,
far more so than comparable half-duplex systems.
Upgrading a half-duplex design by adding a
full-duplex echo controller without changing the
half-duplex mechanical hardware typically results
in a system whose performance is unacceptable.
This section describes the critical parameters of the
mechanical design that ensure quality full-duplex
operation.
The mechanical design consists of speaker and
microphone component selection, speaker housing
and mounting, and speaker and microphone
placement in the car..
1.2.1Selecting the Acoustic Components
1.2.1.1Speaker Requirements
The quality of the speaker in a full-duplex system
is critical to system performance because echo
cancelers are sensitive to signal distortion. Because
the echo canceler uses a linear filter to model the
acoustic path, the acoustic path to be modeled must
be linear in order for the echo canceler to work
well. The total worst-case distortion in the acoustic
path, which includes the speaker driver, the
speaker, the microphone, and the microphone
preamp, should be less than 2% THD across
frequency.
The speakers in automotive hands-free systems are
typically driven with a maximum RMS power
between 0.5 and 5 Watts. In order to maintain
2% THD or less, it is necessary to install a speaker
whose RATED power is at least twice as large as
the maximum power to be driven. For example, if
we wish to drive 2 Watts into the speaker, then the
speaker's RATED power should be 4 Watts or
greater. The RATED power of a speaker is the
power at which the distortion performance is
specified. The typical distortion specification for a
speaker operating at its RATED power is either 5%
or 10% THD, depending on how the manufacturer
specifies distortion.
Speakers are also specified with a MAX power
rating. The MAX power is the power level above
which the speaker can be damaged. The RATED
power, if it is given, is typically about half as large
as the MAX power. Thus if the RATED power is
not given, a good rule is to assume that the RATED
power is about half of the MAX power.
NOTE: The above RATED power/MAX power
generalization does not hold for new generation
ultra-thin Mylar speakers. The poor distortion
performance of these thin speakers makes them
unsuitable for full-duplex car designs. Thick
speakers exhibit a more linear behavior than thin
speakers of equal diameter and are preferred in
full-duplex designs.
1.2.1.2Microphone Requirements
Less care is needed in microphone selection than in
Speaker
Figure 1. Speaker Distortion
4AN168REV2
speaker selection. Almost any standard
inexpensive electret microphone will work because
microphones are inherently fairly linear devices.
Microphones that cancel background noise due to
their mechanical construction are preferred over
those that do not. Microphones that are
omnidirectional are preferred over those that are
directional.
AN168
1.2.1.3Speaker Housing Requirements
The quality of the speaker housing affects the
performance of the system because the speaker can
induce vibrations in its housing if it is not properly
mounted. These vibrations tend to create “buzzing”
artifacts which are not linear and result in poor
echo canceler performance.
Speakers that are supplied after-market in a
housing and speakers that are part of the car's radio
system generally do not present problems. It is the
speakers that are glued or otherwise rigidly affixed
to their plastics that create nonlinear buzzing
artifacts.
Speakers should be soft-mounted to their housings
by using soft pliable acoustic foam. Care should be
taken to minimize any physical means by which the
speaker can induce vibrations in the plastics.
The Test section contains a test procedure for
testing and eliminating buzzing artifacts.
1.2.2Placing the Speaker and Microphone
The placement of the speaker and the microphone
affects the gain selection portion of the electrical
design of the system which will be covered shortly.
The microphone should be placed as close as
feasible to the talker's mouth. This maximizes the
signal-to-noise ratio (SNR) of the talker's speech.
In a car, the optimal place for the microphone is
near the rear view mirror, usually attached to the
driver’s visor.
There are two considerations for the speaker
placement. The more important of the two is that
the speaker be placed such that there is a minimum
of movement in the air space between the speaker
and the microphone. This will minimize the
number of updates and corrections that the adaptive
filter makes during the call, resulting in the
transmission of very little residual echo to the
far-end listener. The second consideration is that
the speaker should be placed as far from the
microphone as possible. This minimizes the
acoustic coupling between the speaker and mic and
allows the mic preamp gain and speaker driver gain
to be maximized.
The optimum placement for the speaker in a car is
the top of the dashboard. Whereas this does not
minimize the distance between the speaker and the
mic, it does limit the changes in the acoustic path,
allowing the adaptive filter to update less often,
resulting in less residual echo transmission. Other
placement options, below the dash, driver's side
door, and passenger's side door, favorably decrease
the acoustic coupling, but result in the driver or the
passengers being positioned directly in the path
between the speaker and the microphone.
1.3Electrical Design
The electrical design process consists of the
component selection of the speaker driver, the gain
selections of the speaker driver and the mic
preamp, and the implementation of an acoustic
sidetone. The primary design consideration of the
electrical design process is to limit the distortion in
the acoustic path to less than 2% THD.
1.3.1Selecting the Speaker Driver
Many system designers overestimate the quality of
their speaker drivers. For example, a speaker driver
that claims to be “5 Watts” on the cover of its data
sheet is not suitable to drive 5 Watts of power into
the speaker of a full-duplex echo cancelling
system. The reasons are two-fold:
1) The “5 Watts” number is usually a Typical
number, not a Maximum or a Minimum specification
2) “5 Watts” is specified with a THD of 10%, not
the 2% number that we are designing to.
AN168REV25
AN168
Speaker Driver
Vout
Rx Out
VCC
GND
Figure 2. Speaker Driver Distortion
The Appendix lists five example speaker driver
circuits that are suitable for full-duplex hands-free
systems.
1.3.2Setting the Speaker Driver Gain
V
Gain
-------- -
=
Vin
where Vin = full-scale voltage at the AO pin of the
CS6422, which is 1 Vrms, or 0 dBV.
Additionally, the gain can be expressed in dB using
the following relationship:
Gain dB() 20Gain()log×=
The following example shows how to derive the
gain required to drive 2 Watts of RMS power into
a 4 Ω speaker. Keep in mind that the RATED
power for this speaker should be 4 Watts or greater,
and the MAX power should be 8 Watts or greater.
2
V
PIV×
==
----- -
R
The speaker’s RATED power and the power driven
into the speaker are RMS powers. The RMS power
is given by the product of the RMS current and the
RMS voltage, or the square of the RMS voltage
over the speaker resistance.
The maximum speaker driver gain is determined by
the square root of the product of the RMS power
and the speaker driver resistance:
2
V
PVI×
==
----- -
R
VPR×=
where P = RMS power delivered to speaker, V =
RMS voltage across speaker terminals, I = RMS
current through speaker, and R = resistance of
speaker in Ohms.
VPR×24×82.828====
Gain
V
-------- -
Vin
2.828
------------1
2.828== =
Gain dB() 202.828()log×9dB==
Many speaker drivers suitable for hands-free
full-duplex design have fixed gains of 20 dB or
more, or are not stable for gains less than 20 dB.
Adding 20 dB of gain to the full-scale output of the
CS6422 (=1 Vrms, =0 dBV, =2.8 Vpp) results in a
huge signal at the speaker terminals (=10 Vrms,
=20dBV, =28Vpp). Because of this, the speaker
driver gain is implemented in two stages, an
attenuator stage followed by a gain stage. The
attenuator can be implemented using a simple
resistor voltage divider network as shown in
Figure 3.
6AN168REV2
AN168
CS6422
A0
R
+
1
R
2
AttenuationGain
+20 dB
+
Figure 3. Generic Speaker Driver Configuration
CS6422
RVol
TVol
Acoustic
Echo
Canceler
-
Σ
+
D/A
A/D
Speaker
Driver
Mic
Preamp
3
Microcontroller
Figure 4. Using RVol to Implement Volume Control
1.3.3Volume Control
In most half-duplex systems, volume control is
implemented by changing the gain of the speaker
driver. In a full-duplex system, this is undesirable
because gain changes in the acoustic path require
the echo canceler to readapt, resulting in elevated
levels of residual echo during the training process
or a temporary drop to half-duplex operation.
In CS6422 systems, it is best to implement volume
changes by using the RVol control. The RVol
control provides up to +30 dB of AGC'ed gain to
the receive path, and because the output of the
RVol control is fed both to the echo canceler and to
the DAC driving the speaker, changes in RVol do
not cause changes in the acoustic path, which keeps
the echo canceler from having to readapt. This
portion of the signal flow diagram is shown in
Figure 4.
In general, the RVol control should be set to a value
between +6 dB and +30 dB. In systems which have
a network sidetone (a coupling path between NO
and NI supplied by the phone), the maximum RVol
value may need to be limited due to loop gain
concerns. See the sections entitled NetworkSidetone and Loop Gain for more information.
AN168REV27
AN168
1.3.4Acoustic Coupling
Figure 5 shows the three most common places for
distortion to be introduced into the acoustic path.
These are the speaker driver, the speaker, and
clipping at the A/D converter after the mic preamp.
With careful choice of the speaker and speaker
driver gain, we can eliminate the first two by using
the techniques previously discussed. The third
distortion source, clipping at the A/D converter, is
controlled by limiting the amount of acoustic
coupling.
The acoustic coupling is defined as the gain (or
loss) between the AO pin and the APO pin on the
CS6422, with TGain set to 0 dB. If TGain is set to
a non-zero value, then the TGain value is added to
the AO/APO gain number to compute the amount
of acoustic coupling.
The acoustic coupling is determined by 5 factors:
the speaker driver gain, the speaker efficiency, the
air coupling between the speaker and the
microphone, the microphone sensitivity, and the
mic preamp gain. Assuming the speaker and mic
have been chosen, the remaining design variables
are the speaker driver gain, the mic preamp gain,
and the speaker and mic position.
Usually, the speaker driver gain is chosen based on
the linearity requirements previously described.
The speaker and mic placement are determined by
ergonomic factors and the desired acoustic path
stability described above. The remaining variable
is the mic preamp gain, which is typically set such
that the worst-case acoustic coupling is between
-9 dB and -15 dB, the first number being the
preferred design target, as shown in Figure 6.
The acoustic path response is highly frequency
dependent. The contributions of the speaker driver
and the mic preamp to the frequency response are
essentially negligible since both of these amplifiers
typically have a stable and well-behaved frequency
response. The dominant factors in the frequency
response of the acoustic path are the speaker's
inherent frequency response, the microphone's
inherent frequency response, and the frequency
response of the path between the speaker and the
mic which is strongly affected by the speaker's
housing. The flatter the frequency response, the
better the echo cancellation.
Figure 7 shows an example acoustic path frequency
response for a speaker and microphone separated
by approximately one meter.
The signal at APO will visibly clip for signals
greater than +5 dBV (5 Vpp). Keep in mind that the
acoustic A/D converter clips at 0 dBV (2.8 Vpp)
when TGain is set to 0 dB.
1.3.5Setting the Mic Preamp Gain
As stated above, the design goal is to have the
worst-case value for the acoustic coupling, the
highest value across the frequency band of interest,
less than or equal to -9 dB. Strictly speaking, it
need only be less than 0 dB to avoid clipping at the
acoustic A/D converter. The additional 9 dB
provides margin for component tolerance variation
(dominated by speaker variation), component
installation (dominated by speaker/mic placement),
and acoustic path variation (dominated by the
position of the driver, passengers, and objects in the
car). The mic preamp gain is adjusted to achieve
the desired level of acoustic coupling.
There are two methods that can be used to set the
acoustic coupling: the frequency response method
and the loop gain method. The frequency response
method is good because it provides frequency
response information that can be used to increase
the quality of the system (flat frequency response is
desired). The loop gain method is quick, easy, and
requires no additional test hardware beyond the
ability to configure the CS6422's registers.
In the frequency response method, the acoustic
path frequency response, the gain between the AO
and the APO pins on the CS6422, is measured by
automated test equipment and plotted. The
8AN168REV2
AN168
Speaker Driver
Speaker
AO
DAC
1
2
Air
CS6422
Coupling
3
APO
ADC
Mic Preamp
Microphone
Figure 5. Three Common Sources of Acoustic Path Distortion
Speaker Driver
AO
DAC
Speaker
CS6422
ADC
Acoustic Path = -9dB
APO
Mic Preamp
Microphone
Figure 6. Acoustic Coupling Design Target
Acoustic Coupling (dB)
05001000150020002500300035004000
0
-10
-20
-30
-40
-50
-60
-70
-80
Frequency (Hz)
Air
Coupling
Figure 7. Example Acoustic Coupling Frequency Response
AN168REV29
AN168
maximum value of this curve is then noted, and
gain is added or subtracted at the mic preamp in
order to set this maximum value to -9 dBV (with
TGain set to 0 dB). This procedure is further
described in the Tests section of this note.
The loop gain method uses howling to determine
the optimum mic preamp gain. In short, the phone
network is disconnected from the CS6422, and
TVol, RVol and NSdt are used to create a +9 dB
path between APO and AO inside the CS6422. The
system will howl, go into regenerative feedback, at
the point that the total loop gain reaches a factor of
'1', or 0 dB. This happens whenever the gain
between AO and APO outside the CS6422 reaches
-9 dB. The frequency of the howl is the frequency
of the maximum loop gain, which is dominated by
the speaker, microphone, and the air path between
the two. Figure 8 illustrates.
The loop gain procedure is as follows:
1) Configure the CS6422 with its default configuration, with the exception of the following:
a) Mic = ‘1’ or ‘0’, depending on whether the
internal mic preamp is used or not
b) TSD = RSD = HDD = ‘1’, transmit and re-
ceive suppressors and half-duplex mode are
disabled
c) ACC = NCC = 'cleared', echo cancelers are
forced to a cleared state to prevent updates
d) AECD = NECD = ‘1’, echo cancelers are
disabled
e) TVol = +12 dB
f)NSdt = -12 dB
g) RVol = +9 dB
2) Adjust the mic preamp gain (or the speaker
driver gain) until the system is just on the verge
of howling. At this point the gain between AO
and APO will be the desired -9 dB.
Disconnect
Network
CS6422
-12dB
NSdt - Network
Sidetone
+9dB
RVol - Receive
Volume
TVol - Transmit
Volume
12dB
Figure 8. Setting the Acoustic Coupling
Goal is -
9dB
A
O
APO
0dB
Speaker
Driver
Adjust
?dB
Mic
Preamp
Air Coupling
-?dB
10AN168REV2
AN168
The register settings to accomplish the above are as
follows:
reg 0: 47a0 (or c7a0 if internal mic preamp is
used)
Note: If the mic preamp gain is not easily
adjustable in the test circuit, coarse amounts of gain
can be added by using the TGain control, which can
be set to 0 dB, +6 dB, +9.5 dB, or +12 dB.
1.3.6Acoustic Sidetone
When the coupling path between the speaker and
the microphone is relatively consistent, linear, and
has a high signal-to-noise ratio (SNR), the CS6422
provides good echo cancellation and makes good
training decisions. In the car environment, the SNR
of the acoustic path can be degraded significantly
by road and engine noise and the separation
between the speaker and the mic. In these systems,
it is often useful to introduce a strong, linear,
predictable coupling path electrically by using an
acoustic sidetone.
The acoustic sidetone provides 3 main benefits:
1) The presence of a strong path decreases convergence time, meaning it decreases the time the
CS6422 spends in half-duplex.
2) The linear path enhances stability in systems in
which the strongest real (air) path is distorted.
Note that even though the echo canceller will
not cancel the nonlinear elements of the acoustic echo, it will make better decisions regarding
when to engage the supplementary suppression
algorithms to mask such echo. This results in
improved performance during far-end single-talk.
3) The consistent path provides an echo path that
is independent of the acoustic environment,
making the system less sensitive to path changes and noise. This enhances full-duplex performance by reducing the tendency of the CS6422
to drop to half-duplex when the driver moves.
The amount of sidetone required depends on
several factors. Typically, a good number is
between -24 dB and -12 dB. To be useful, the
electrical coupling should be about as strong as the
strongest typical air coupling, but not much
stronger. A good starting point for systems whose
peak acoustic coupling is -9 dB is -18 dB of
acoustic sidetone. The acoustic sidetone can be
implemented in CS6422 systems by using the ASdt
control, which is configurable to none, -24 dB,
-18 dB, or -12 dB.
1.4Echo Canceler Parameter
Optimization
One of the benefits of the CS6422 is its high degree
of configurability. Whereas the number of
parameters may seem daunting at first, there are
only a few that need to be tweaked to optimize
performance. The rest can be set once and left
alone.
1.4.1Starting Example
The following is an example register configuration
that is useful as a starting point for cellular car
hands-free systems.
Note: Actual performance testing should be
performed in a car, not a lab. This is because the car
and the lab present different acoustic environments
to the echo canceler, and the goal is to optimize the
parameters for the target environment, which
requires testing in that target environment.
The following parameter set assumes that there is
no coupling on the network interface to the phone.
If there is a network coupling path, see the NetworkSidetone and Loop Gain sections below.
AN168REV211
Loading...
+ 25 hidden pages
You need points to download manuals.
1 point = 1 manual.
You can buy points or you can get point for every manual you upload.