NWay Speaker Project

1
N-Way Speaker Project

• Objectives
• Test hypothesis of lower distortion by using
  multiple drivers
• Frequency-optimized drivers
• Lower IM products due to constrained input range
• Reduced Doppler distortion
• Provide a versatile test bed for evaluating
  drivers, number of crossovers and various slopes
• Motivations
• Results with steep filters shows promise
• 3 and 4-way systems with 8-pole filters (Alesis
  amps) sound very good
• Results with 3-way test bed very revealing
• Some drivers work very well over small ranges
• Additional test bed work likely to be productive

2
Initial Approach
3
Crossover Options With Initial Approach
4
Capabilities

• Selectable crossovers between 2-way and 7-way,
  with at least 19 different combinations
• See chart on previous page
• 3 different 2-way combos 4 different 3-way
  combos etc.
• Each crossover selectable between 4-pole LR and
  8-pole LR
• Electronic volume controls on each driver
• Each driver can be switched to an external driver
  for testing
• Allows testing different tweeters, midranges, etc
  with other drivers
• Some sort of user interface to make switching easy

5
Initial CircuitryFilter board
6
Some Re-thinking

• The switching (Mux blocks) will need to be done
  with a controller
• Raises several questions
• What type
• Development tools
• User interface
• This becomes a major decision point
• Need to evaluate microcontroller options and SW
  complexity
• User interface becomes a major issuePC vs
  dedicated display
• This is a lot of analog stuff
• Purchased 10 of the filter boards and used a pair
  for Saras 3-way
• Worked first time, but time-consuming to build
• Caps and other components get expensive
• Analog muxes arent that simple
• Non-linear resistance is disconcertinghigh
  distortion unless used with high impedance loads,
  which increases complexity sigh
• That TI digital audio series starts to look more
  and more attractive

7
Revised Project Flowchart (1)
Purchase MPU evaluation board
Purchase LCD display
Build LCD Interface Bd
Write code for LCD
Evaluate Controller options
Things going OK?
No
Digital/ Analog decision
Initial approach
Purchase TAS3004 eval board
Implement 8-pole LR crossover
Digital
Yes
Write a basic user interface
Analog
Design Relay Boards
Purchase analog filter parts
Add Logic for Denniss Cap tester
Design TAS3004 test boards
Build MPU boards
Build LM3886 amp boards
Design Cabinet Chassis
Path not active as of 5/23/2004
Buy more SW development tools
All tan boxes completed as of 5/23/2004
Cabinet/Chassis
Amps
Relays
Filters
CPU/Display
Completed 5/31/2004
8
Project Flowchart (2)revised per 5/2004 learning
Cabinet/Chassis
Amps/Power Supply
Relays/Driver test
Digital Filters
CPU/Display
Equipment cage approach
8-channel volume control EQ board
Evaluate TI Digital Amps
TAS3004 and PGA4311 control code
Build Relay Boards
Rotary encoder SW
Prototype power supply
Cabinet Approach
TAS3004 prototype box
Cabinet/Chassis
Amps/Power Supply
Relays/Driver test
Digital Filters
CPU/Display
Partially Completed 9/2004
Completed 7/3/2004
Switch box for Dennis Murphy
8-channel Preamp
Plate amp Prototype
Completed 1/2005 to 3/2005
Completed 9/2004
Side-tracks
9
Project Flowchart (3) Current as of 3/2005
Cabinet/Chassis
Amps/Power Supply
Relays/Driver test
Digital Filters
CPU/Display
Complete Chassis mods
Filter design/EQ for Usher 2-way
Digital amp PS TAS3103 prototype
Rev 2 CPU boards
Driver test prototype
Network the controllers for stereo
Final Amp Design (Aarons help)
TAS3103 Evaluation
Driver database (EQ/Vol)
Woodworking
Build Amps
Wire test
Completed 9/2004 to 10/2004
Completed 7/3/2004
Digital amp kit?
???
DIY Event (?)
Partially Completed 9/2004
Completed 1/2005 to 3/2005
Possible side-tracks
10
Some Learning (as of 5/2004)

• Micros have become much easier to use
• Development tools for the 68HCS08 are very nice
• Lots of flash memory, lots of RAM, lots of I/0
  and peripheralsfor 10
• LCD HCI was done well
• Table-driven design that allows rapid
  reconfiguration and export of HCI to remote
  devices (PDAs)
• The TAS3XXX series of chips is amazing
• TAS3004 isnt the most powerful in the family,
  but it can do a complete 8-pole LR crossover
  (both HPF and LPF) with just 4 of the 7 biquads
• The automatic loudspeaker equalization (ALE)
  software is nicedoes the same job as something
  like LspCAD, but is more effective because it has
  all the TAS3004 resources available
• With additional shelving filters and volume
  control, the TAS3004 will be far more effective
  than analog circuitry in a much smaller, much
  lower cost package

11
More Learning (2005)

• The TI amps allow a full digital pathno analog
  conversions
• Lots of parts and higher circuitry cost (bad)
• The digital amps require less heatsinkingthis
  may help reduce cost and make this option better
  in the long run (good)
• The TI amps require a good power supply (no PSSR)
• This means prototyping some switching supplies
• Two approaches AC to DC (purchased) or DC to DC
  (use 60Hz toroid front end to generate 45-50V
  then DC to DC convert
• DC to DC approach makes good sense for an
  on-board regulator approachjust have one large
  transformer and multiple DC to DC converters.

12
Driver Test Learning

• Attempted a driver test fixture in Oct 2004
• Active 3-way system using 8-pole crossovers at
  400Hz and 2900Hz
• Relay box allows selecting between any two of the
  midrange drivers or tweeters using an RF remote
  control

13
Driver Test Learning (2)

• TOTALLY unexpected results
• Differences didnt fall neatly into nice obvious
  bins such as clarity, smoothness, lower
  distortion, coloration, etc
• No correlation with cone materials Seas L14 and
  the Dayton RS125 didnt sound much alike driver
  closest in sound character to W14 Excel driver
  was the 8 coated paper GW-205 Goldwood woofer
  (!!).
• Tweeters were just different but no obvious
  leader
• Lessons learned
• Fixture must adjust for volume (need database of
  volume adjustment settings)
• Crossovers should be adjustable to constrain the
  testing to a desired range.
• Each driver is going to have to be measured and
  EQed (steep crossovers not sufficient). A
  database of EQ settings and computer control is
  needed
• A better mechanical mounting fixture will be
  needed to minimize the effects of driver location
• Driver testing is subjective and qualitative
  tests should be supplemented with quantitative
  distortion tests

14
Scalable Approach for Digital Amps

• Common digital input and CPU
• TAS3004 for overall EQ and budget analog I/O
• 2.5 channels for the power amp
• On-board power supply volume control
• Two power amps plus 3rd channel with analog
  output for sub

Digital Audio
TAS3103
TAS5121
TAS3004
DIR1703
I2S
I2S
SPKR
I2S
TAS5518
TAS5121
Control
Analog In
PSVC
I2C
Control
Power Supply
CPU
Low Voltage Power Supply
LCD
Connectors
Power
Modules
15
Cabinet/Equipment Chassis

• Use hexagonal speaker from Steve
• Use KLH KL-2400 receiver as the chassis
• Separate sub amp (analog)

16
Side Track 8-channel Preamp

• Uses the CPU board and TAS3004/PGA4311 board
• Added a relay input board to allow by-passing the
  PGA4311s
• Uses a THL Audio chassis
• Two high-quality rotary encoders
• Total cost is way out of controladded up to
  about 560 in total materials

17
Side-track Digital Plate Amp Prototype

• Uses TI evaluation board with mod to convert
  SPDIF stream (32-bit delay)
• Uses LT1074 switching regulator
• Part of a fixture for Dennis Murphy
• 15L cabinet
• Morel HU631 and Hiquphon OW1
• Plate amp shown at right
• Allows development of digital active crossovers
  using TI tools

18
Schematics/PWB

• Download ExpressPCB to view original files

19
CPU Board
20
PGA4311/TAS3004 Board
21
Relay Board
22
Power Amps 1
23
Power Amps 2
24
Power Supply/SPDIF for Plate Amp Prototype
25
Costs (2004)

• CPU
• TAS3004