Hi,
This is my first attempt at a web site.  So things,  
will no doubt be, dynamic for a while. Change is
the only constant.
admin@testvector.com
Xit: Exit
Sub '--
Xit: Exit
Sub '--
Bass Redirection & Management Test Automation.
In my words.  Bass Management is a process where by the low frequency content of the
decoded outputs  of a multichannel audio signal processors is sent to the system speakers
that can best make use of those signals.  

Typical  consumer stereo systems can produce from 2 to 8 output streams and reproduce them using
several types of speakers.  
This can be as simple as "large, (L)", "small, (S)" or "off, (0)".

The intent is that regardless of the number of speakers used,  that the tonal balance of the
reproduced audio is as close as possible to the way it was recorded.

Testing Bass Management systems is as easy as measuring speaker system crossover network
characteristics.  Send a signal in and see where it comes out.
Bass Management
At : Wikipedia
http://en.wikipedia.org/wiki/Bass_management
Part of looking for proper output signals and locations is also to make sure they do not appear where
unexpected.

To be done properly, for each different input signal condition all available output channels should be
monitored.

There are about 32 generic speaker combinations that can be used with 8 speakers.  Using 6 source
channels with 8 speakers gives 48 source to output signal path combinations. So measuring Bass
Management in a system that implements all 32 speaker combinations gives about 48*32 = 1,536 potential  
data points for a complete evaluation.

The output streams from typical consumer audio equipment is in groups of 2, optical or electrical channels or
pairs.  Most digital audio test equipment is native one or two channel instruments.

This usually necessitates some sort of signal multiplexing.  For testing by hand a simple 4-way switch for
electrical signals,  about the only way to switch optical signals was to do it by hand using an optical cable.
B
D
A
I designed and the company manufactured about 50, Dolby ET-142 digital
BitStream switch for engineering test purposes.  

A programmable 1 of 4 input BitStream selection switch :
Each of the 4 inputs are both optical and transformer coupled.
The output consists of 2 TORX-173 optical and 2 Pulse Engineering  output
transformers.
Control from keypad or RS-232.
Supply, 9vdc @ 448mA.

The
ET-142 allowed me to make FFTs on 8 digital channels in about 47
seconds.  Using this switch and the software shown on this page,  I could
make the necessary 1,200 to 1,500 measurements in 2 to 3 hours,  not weeks.

After creating the ET-142,  I programmed my Audio Precision 2322,  to collect
raw data used to calculate the characteristics of the Bass Management
crossover networks.
I was told that it took the first person in the Products Test Group to
complete the BM testing,  (that when AC-3 was 5 inputs to six
speakers or 30*32 = 960 measurements),
two weeks !

There were no real switch options available for test purposes.

That was the case when I started working with the newly formed  
digital signal processor test group.  

This was  after personally being the analog test group for about 4
years.  I had taken the original manual 2 week Pro Logic IC
evaluation process  down to about 4 hours by use of test
automation..

I was not interested in spending days just switching cable for one
test,  especially when the number of available channels was
increasing almost daily.  Well almost :-)
C
Stereo
At : Wikipedia
http://en.wikipedia.org/wiki/Stereo
Stereophonic sound, commonly called
stereo, is the reproduction of sound,
using
two or more independent audio
channels, through a symmetrical
configuration of loudspeakers, in such a
way as to create a pleasant and natural
impression of sound heard from various
directions, as in natural hearing.
This GUI shows the  results of one
Bass Management test iteration.

The test signal is Center to Left Front,
monitoring Front Left & Right, test
iteration 41

It is indicating that the filter is :
Low Pass.
127.9 Hz corner frequency.
-9.89 dB/Octive slope.
This GUI shows the  results of one
Bass Management test iteration.

The test signal is Center to Center,
monitoring Center and Low Frequency
Effects Channel, test iteration 43

It is indicating that the filter is :
High Pass.
130.99 Hz corner frequency.
9.39 dB/Octive slope.
This GUI shows the  results of one
Bass Management test iteration.

The test signal is Left Front to Left
Front, monitoring Front Left & Right,
test iteration 01

It is indicating that the filter is :
Band Pass.
No corner frequency.
No slope.
Above is the graphical results of a single Bass Management calculation.  The
large cursors on the plot show data points used in the calculation.
This is a zoom view of the top graph.  The large cyan and green index
crosses show actual measured data points.   The red index represents a
virtual filter -3dB point and is a linear interpretation between the closest
measured data points.

Note, the large graphic indexes are not part of the Audio Precision software
package.  The indexes are formed by adding additional virtual test data to the
measured data that when plotted as audio data assumes the shape of the
index crosses.
Above is a GUI showing the results of 30 test iterations used to evaluate one speaker combination configuration.

This display is of filter type and pass band level.

All data is GREEN so it a GO !  Red areas should never see a signal under this configuration.
This GUI shows the  results of one
Bass Management test iteration.

The test signal is Center to Center,
monitoring Center and Low Frequency
Effects Channel, test iteration 43

It is indicating that the filter is :
High Pass.
130.99 Hz corner frequency.
9.39 dB/Octive slope.
Above is a GUI showing the results of 30 test iterations used to evaluate one speaker combination configuration.

This display is of filter type and slope.

All data is GREEN so it a GO !  Red areas should never see a signal under this configuration.
Above is a GUI showing the results of 30 test iterations used to evaluate one speaker combination configuration.

This display shows the raw measured data.

All data is GREEN so it a GO !  Red areas should never see a signal under this configuration.
Above is a GUI showing the results of 30 test iterations used to evaluate one speaker combination configuration.

This display is of test iteration conditions.

All data is GREEN so it a GO !  Red areas should never see a signal under this configuration.
Above is the file listing GUI for a utility I wrote to allow rapid viewing of large volumes of AP test data.  The listings
are of Bass Management tests, (.At2) and test iteration data files, (VuDat_##).
Active Matrix
Decoders
Test Automation
Matrix Channel Separation.
Matrix Channel Separation.
In Stereo Mode :
Viewing L and R main outputs.
360' constant power pan.
The signal generators are driven off the sine and cosine of the steering vector.
CS Movie Mode :
Viewing : L & R mains, surrounds and back channel outputs.
360' constant power pan.
PL Mode :
Viewing : L & R Mains, Center & Surround
PL-II Mode :
Viewing Lm & Rm, Ls & Rs, Cm
CS Movie Mode :
Viewing Lm & Rm
CS Movie Mose :
Viewing Ls & Rs
CS Movie Mode :
Viewing Center & Center Surround
Slow / Fast Steering Threshold
The above text is the Comments Text of the AT2 test file.
Park Level : Low Level Steering Threshold
Steering Control Path Band Width
An active matrix decoder, showing Steering action.
Lt/Rt level difference = 10.0 dB.
Lt is ??.? kHz at 0 dBr. 20 Hz to 20 kHz 1000 point sweep.
Rt is 2.5 kHz at 0 dBr. Tone burst mode low level = -10 dBr
The Rt generator is run in burst mode, 400 ms ON 400 ms OFF.
The low level burst amplitude is held at a consistent 10 dB.
When the Rt level drops, L becomes dominant and is steered away from Center.
The level of the generator driven tracking bandpass filter output is plotted verses frequency.  

The depth of the notch indicates the magnitude of the steering action at that frequency.  
The top of the notch represents the un-steered level of the ??.? kHz tone as marked by the
overlaid measurement done with the burst mode OFF.  Notice how this decoder actually adds
the ??.? signal into the mix below about 500 Hz.  Notice that the depth of the notch is not
seriously affected by the combined Lt/Rt signal power. This processor parks at about -90 dBr.
An active matrix decoder, showing Steering action.
Lt/Rt level difference = 10.0 dB.
Lt is ??.? kHz at 0 dBr. 20 Hz to 20 kHz 1000 point sweep.
Rt is 2.5 kHz at 0 dBr. Tone burst mode low level = -10 dBr
The Rt generator is run in burst mode, 400 ms ON 400 ms OFF.
The low level burst amplitude is held at a consistent 10 dB.
When the Rt level drops, L becomes dominant and is steered away from Center.
The level of the generator driven tracking bandpass filter output is plotted verses frequency.  
The depth of the notch indicates the magnitude of the steering action at that frequency.  
The top of the notch represents the un-steered level of the ??.? kHz tone as marked by the
overlaid measurement done with the burst mode OFF.  Notice how the measured level does
not exceed the un-steered level. Notice that the depth of the notch is not seriously affected by
the combined Lt/Rt signal power but is influenced by the frequency of the tone being steered
not the 2.5 kHz tone that is changing in level to cause the steering action.  This processor
parks at about -63 dBr.
An active matrix decoder, showing Steering action.
Lt/Rt level difference being incremented by -1.0 dB.
Lt is 2.5 kHz at 0 dBr.
Rt is 1.0 kHz at 0 dBr.
The Lt generator is run in burst mode, 500 ms ON 500 ms OFF.  The low level burst amplitude is incremented from 0 dBr to -10 dBr in -1.0 dB steps.
When Lt = Rt the condition is un-steered so both 2.5 kHz and 1.0 kHz are present in the C and Cs outputs.
The original Lt and Rt signals are still present at the L Main and R Main outputs because there is no common in or out of phase signals to steer Lt or Rt to C or Cs.

As the low level burst amplitude is reduced there is an increasing Lt/Rt imbalance.  This will produce steering action across the L/R axis.  When the steering action takes
place the tone being steered is the one with the higher level.  Because the 2.5 kHz generator level is being reduced by the burst mode the Rt 1.0 kHz tone becomes
dominant.  Thus the 1.0 kHz tone will be removed from the C and Cs outputs as it is steered away.  Because the C and Cs
outputs contain a mixed tone it is necessary to observe the 1.0 kHz tone through a narrow band pass filter.

The 1.0 kHz bandpass amplitude is plotted verses time.  Notice the shape of the top 5 traces representing 0 to -4 dB of Lt/Rt difference.  This is slow steering.  
Traces 6 to 10 represent -5 to -10 dB of Lt/Rt difference.  This is fast steering.  The transition point is very sharp at -4.7 dB.  Why is all this important ?  Because -6 dB is
considered "Off Center Stage".  Typical dialog in a movie is mono and thus would steer to C.  It is desirable to have slow steering action during
dialog to prevent it from becoming "to busy", but when effects take place surrounding you fast steering action is more impressive. :-)