This page last updated December 13, 2002
This page is best viewed with Internet Explorer 6.0 or later at
1024x768 or higher resolution.
This site contains
information on my senior design project at the University of Kentucky
Fall 2002
Active, Stereo Audio Crossover Network
Who am I? My name is Gus
Dattilo.
I am an Electrical Engineering senior at the University of Kentucky.
I will be working on this project alone.
I can be reached at amdatt0@engr.uky.edu
if any one is interested in contacting me about this project.
My Postal Address is: Gus
Dattilo
209
Simpson Ave #302
Lexington
KY, 40504
This is the second draft of this page.
The first draft was compiled on September 22, 2002
THE IDEA:
The basic idea of this project is to be able to split an audio signal
into various components, by frequency.
Have you ever been driving in your car and notice how those small
speakers seem to rattle and the music gets distorted when there is a lot of
bass in the music. Now if there weren’t
a lot of bass in the signal those small speakers could be pretty loud, but that
wouldn’t sound very good. The common
solution to this is to split the signal up into various components. We could send the left side high frequency
signals to the small left side speaker and we could do the same with the right
side. We could send the bass
frequencies, from both the left and right channels to a single large speaker. This way we could still preserve the stereo
effect but send all of our bass to a single “subwoofer.” Now we can have music that is both loud and
sounds good and we only need one big speaker that we could hide in the trunk or
under a seat. This applies not only to
car audio but also to home audio. At home
we could use two small speakers for the high frequencies on the left and right
channels but then hide one large low frequency speaker behind the sofa, or
where ever.
Abstract:
Most home stereo speakers incorporate some sort of crossover network to
send high frequencies to the tweeters and the low frequencies to the
woofers. The vast majority of these
crossovers are what is called passive networks. This means that these crossovers, or filters as it would be
better to call them, are located after the power stage of the audio
system. This would mean that the audio
signal leaves the amplifier and enters the filter network where the audio is
split between the tweeters and woofers.
My scheme will require more amplifiers than this passive scheme. My scheme will locate the filter network
before the amplifiers. The audio signal
will enter the filter network as a left and right signal but leave as a
left-hi, right-hi and a summed-low. The
high frequency signals can go to small amplifiers and drive the high frequency
speakers. The summed low frequency
signal can go to one large amplifier to drive the subwoofer.
Even this scheme is not anything new.
This has been done for years in many high-end audio applications and in
recent years we have begun to see this in home theater application. Many new home theater systems featuring what
is called Dolby Digital 5.1 use a scheme like this. The five in 5.1 alludes to that fact that the system uses five
high frequency speakers. The one
alludes to the single low frequency speaker, which plays the low frequency
sounds from all 5 channels. Most active
crossovers use the typical filters that most electrical engineers are familiars
with such as the Bessel or Butterworth implementations. These filters are most often constructed
from resistors and capacitors, with the exception of Dolby Digital, which uses
a DSP. I intend to build my filter
network using a different type of filter called the “switched capacitor”
filter. To explain it simply this is a
digital filter network but the input signal never actually goes through an A-D
conversion and the output never goes through a D-A conversion. I intend to provide more information on this
topic as I learn about it myself.
EMBODIMENT:
Here is a simple block diagram to show how my proposed system will
work.
To be more technical, lets take a look inside the Active Crossover
Network.
The filter sections are where the real work on the project will occur.
The switched capacitor filter action will be achieved using MAX264 chip.
I may have to implement some parts of some filters using OP Amp
circuits.
Time will tell.
The buffers will be simple OP Amp circuits.
UPDATES:
October 13, 2002
It does not appear that I will be using any filter sections other than
those in the MAX264 chip. After doing a
great deal of research on this device I have found that it is almost perfectly
suited to the task at hand. It provides
me with a pair of second order high and low pass filters. I had initially intended to build this
network with 3rd order filters and I may do this depending on time,
but for now I am going to limit my filter sections to 2nd order.
I have not yet had the opportunity to test my filter sections with the
MAX264. Due to this I am way behind
schedule. I am still having trouble
obtaining the 40KHz crystal that I need.
It has been ordered from Digikey, but it has not yet arrived. To make matters worse I am beginning to
believe that I will need 50KHz crystal instead in order to have the sampling
rate that I need for good sound quality.
Studying the MAX264 data sheets I have learned that the clock frequency
is divided in half before it drives the sampling stages. This means that with a 40KHz crystal I would
only have 20KHz-sampling rate. That is
not going to be enough. I will probably
encounter some aliasing. Never the less
I can test the circuit with 40KHz crystal.
It will not be much of a problem to switch only the crystal at a later
time.
On the brighter side I have taken the time to pick my various
parameters to use with MAX264 chip. My
frequency divide ratio will be 109.96 which when used with a 40KHz clock will
result in a crossover frequency at about 350Hz. I have also selected my Q factor to be 1.12. I will be operating the MAX264 in mode 3
since this is the only mode which provides a high pass output.
I have also completed a prototype power supply for the system. I discovered some very small switch mode
power supply built be TI, they were also free which made them look even nicer. I am using two supply one +5V the other
–5V. The negative regulator has the
nice ability to take a positive supply as an input and output a negative
voltage. This is wonderful for my
MAX264 and my LF347 OP amps which all need dual supplies. These converters are also very compact and
with all the need support circuitry are only taking up about 2 square inches of
board space. There are 3 electrolytic
Caps a dummy load Resistor and a clamping Diode also in the circuit. The clamping diode was my idea. I learned the hard way that the negative
regulator does not like to had a reverse voltage across its input, there was a
quite a bit of smoke. Another nice
feature of this circuit is its wide input range. The converters will both work with an input voltage between 7 and
26V. Since they are switching type,
they don’t even need heat sinks. Here
are some pictures of the power supply prototype. They aren’t great but they get the point across.
Front Side
Back Side
Schematic
From the back side you can see the dummy load Resistor (RL) and the
clamping Diode (D Clamp)
I have created a preliminary draft of the board that I will want to
create for the final design. This board
design will most likely change, as I am able to test the circuit. It is still a good building block. If you would like to see the board layout
here is the link.
www.engr.uky.edu/~amdatt0/first.brd
I have been unable to convert the board image to something that I can
display on this web page. To look at
this image you will need Eagle Layout Software. The evaluation version can be downloaded free of charge at
November 1, 2002
Last Friday, a week ago today, I built and tested my filter sections
using the MAX_264 chip. I had some
problems and successes.
Problems:
·
Internal oscillator
would not function. According to the
data sheets I should have been able to connect a crystal directly to the
MAX_264 to achieve my clock signal. I
could not get this to work. Perhaps it
is a problem with my crystal.
·
I saw some aliasing at
higher frequencies. After some close
inspection of the data sheets I determined that the MAX_264 divides the clock signal
by two before the signal enters the sampling circuits. With a 50kHz crystal this meant that I was
only sampling at 25kHz and because of the Nyquist theorem my maximum input
frequency would be 12.5kHz, not exactly as high as I would like.
Successes:
·
Using a function
generator as the clock signal I was able to drive the MAX_264 and see that
filter sections work quite well.
·
I discovered that the
MAX_263 has a much higher clock signal divide ratio so I will be able to use a
70kHz clock and still be able to crossover at 350Hz. I have the MAX_263 ordered; the pin out is the same as the
MAX_264 so only some minor circuit changes were required.
·
I have determined that
I will use a 1MHz TTL oscillator as a clock signal to the MAX_263. I will use a 4-stage divider to reduce this
frequency to roughly 70kHz. I performed
some experiments using a 555 timer as my clock generator, but its output was
too unstable. I would like to have the
stability of a crystal in the project.
I have a new board layout complete.
This one supports the changes needed for the MAX_263. It also has support for off board level
controls. This makes use of the 3
OP-Amps that were unused in the previous version. I have also added surface mount decoupling caps to the bottom of
the board at each chip. There is one
point where this may be a problem, under the TTL oscillator where the cap has
to cross the –5V rail. I am going to
talk to someone about this today.
Hopefully I will have a board ready to be populated by the middle of
next week.
Most fun of all, I got my scanner working and now you can see pictures
of my layout and component placement.
This is not actually my most recent version of the board, but it has
all of the features that I intend to incorporate.
If you would like to see the actual board file itself here it is.
www.engr.uky.edu/~amdatt0/499_ver21.brd
It is in the eagle layout format.
All of my images should now be JPEGs so every one should be able to
view them.
November 17, 2002
Today I made quite a bit of progress.
I had indeed tested the filter sections some weeks ago with an
oscilloscope, but today was the first time I had actually listened to filtered
audio to determine if the sound quality was good. It was actually very good.
I also updated my prototype power supply to include the clock
generator. The clock generator is kind
of cobbled together since the only TTL oscillator I could put my hands on was
8MHz. That was too high of a frequency
to run on the breadboard. Putting it on
the prototype board with the power supply helped to make operation more
stable. I am using a CD4060B binary
ripple counter to divide the 8MHz down to 62.5KHz that I need to drive the
Maximum chip. This whole circuit, TTL
oscillator and CD4060B, is located on the power supply board at this time. This counter replaces the 74HC93. I needed more divide stages to use the 8MHz
oscillator than the 74HC93 provided.
Please note that the parts list has changed dramatically.
I also fabricated some “probes” today.
I cutting up some RCA and eighth inch cables and soldering short
breadboard jumpers to them. This will
allow me to do more tests with the audio signals and use my ear as the test
equipment. After all, in this project
the ear will be the final judge.
To listen to the signals I went banging around in my closet and found
an old receiver, I was able to use this as an amplifier. I snagged the center channel speaker off of
my home theater system so I could listen.
It worked pretty well since it can reproduce both high and low
frequencies pretty well.
I took some pictures of the whole setup and also a close up picture of
the modified power board.
In this picture of the modified power supply board you can see the TTL
oscillator (the silver can) and the counter chip next to it. The clock signal travels to the breadboard
via the back and white wire pair. Sorry
this picture is so blurry.
The whole rig is being powered by 12V output of a computer power
supply.
The walk man is being used as the audio source.
Some things to note are that I have not yet built the OP Amp buffers
and I am still using a MAX_264 chip. I
requested samples of a MAX_263 weeks ago, but I have not yet received it. I fear I may have to use the 264 in my final
design. The only draw back to this is I
will not be able to get the crossover frequency that I want. I do intend to socket the MAX_264 so if the
MAX_263 does arrive I will be able to switch it out right up to the very
end. The chips have the same pin
out. I am not too concerned about the
buffer circuits. When I prepare my
etched circuit board I will test the buffers.
The broken link to my latest board layout is also fixed!
Here is a hand drawn schematic of the whole circuit, minus the power
supply. It may be it hard to see.
December 10, 2002
IT LIVES!
Well I believe I am almost finished.
Last week I completed the circuit boards. The entire circuit worked on the first try. I guess all of that prototyping paid
off. I have not updated this web page
in some time, but I have been quite busy, and a great deal of my design has
changed. Due to the large amounts of
noise I was hearing in the prototypes I decided to build the power supply and
clock generator circuits on a separate board.
Now only the filter sections and their associated amplifiers are located
on main board. I also added a large amount
ground to the both of the boards. This
paid off with substantially reduced noise when compared to the breadboard
prototypes. Here are my new layouts,
note the large amounts of copper in the ground plane, especially on the power
supply / clock generator board.
Main Board
Power Supply
Note that the above schematic is still valid. The only change is that the clock generator circuit has been
moved to the power supply board. There
are also a large number of surface mount components in this design. As I learned more about the PCB fabrication
facility here at the University of Kentucky surface mount components became
very attractive. The biggest reason for
this would be that the surface mount components require no holes be
drilled. This is very nice since our
facility requires that all holes be drilled by hand. The surface mount components help to make the whole package
smaller.
There are some electrical changes that should be noted. I added a low pass filter to the high
frequency output section. This
eliminated most of the high frequency noise that I was hearing, the improvement
was dramatic. The cutoff frequency of
this filter, actually there are two (one for each stereo channel), is around
9kHz. This makes it nearly transparent
when listening to music. Components in
this filter include R1, R2, C3, C4 and part of IC3.
I also added another low pass filter at the subwoofer output. This brings the order of the low pass filter
up to 3rd. The results of
this are quite pleasing to the ear when listening to music. This low pass filter includes components
R12, C12 and part of IC2.
Note the complex circuit in the upper right corner of the main
board. This was intended to be a 3rd
board, which would be used for volume controls. In the interest of simplicity I did not populate this board. I decided to use an old piece of stereo
equipment for level controls.
I decided to add one more bell, and a whistle, to this project to make
demonstration easier. I constructed one
more PCB, I was on a roll, that has several relays I can use to switch the
outputs off and on. To control this relay
board I have it connected to the LPT port of my laptop and wrote a small
interface in C++ for control. I am not
into graphical programming so the interface is all text based. It works quite well, and since I will be
using my laptop as an audio source for demonstration it doesn’t add very much
bulk to the project. This source code
will be located in the .zip file I will create of all the files used in this
project. I don’t even have a schematic
for this relay board, but here is the layout. The circuit is very simple; it is
actually the same circuit four times over.
Relay Board
Here are some pictures of my setup as of today. Tomorrow I will be mounting all of the
various boards onto a Plexiglas fixture, which will make the package more
portable and rugged.
Note the five Caps soldered to the back of the RCA board. These are DC blocking capacitors to protect
the amplifiers and audio source from any DC that may come from this
project. They are here because I forgot
to make provisions for them on any of the circuit boards.
I went and stayed at my parents’ house over Thanksgiving weekend and
while I was there I built a test fixture so I could demonstrate my system
easily. The fixture houses three
speakers, one 8” subwoofer and two 4” full range drivers. There are two amplifiers attached to the
fixture along with a 12-volt power supply.
The whole unit is about 12” x 18” x 10”. I do not have a picture of this unit at this time.
Here is the .zip file with all of the various files I have used in the
construction of this project. There are
both the Eagle .brd files and .ps files for all three layouts. Note that neither this .html file nor any
.jpg files are included in this .zip.
www.engr.uky.edu/~amdatt0/allfiles.zip
I plan to update at least one more time and will hopefully be able to
include more pictures of the completed project.
December 11, 2002
It is finished!
Tonight I mounted all of the various boards on their fixtures and I
will do no more. Here is a picture of
the completed unit.
I decided to leave the relay board sitting off to the side since it is
more for demonstration than functionality.
It is also heavy enough that it will stay in place on its own.
AND, I say it works. Do you
believe me? Well here is the
proof. Using a two channel function
generator I pumped a 100Hz sine wave into channel A and a 1kHz sine wave it to
channel B, at the same time. The
following image is what the output of the two high frequency channels and the
subwoofer channel looked like. Nice
scope, isn’t it?
This image is a bitmap. Some
browsers might not be able to display it properly. I tired to convert it to a .jpg but my image processing software
keeps giving an out of memory error. I
haven’t the heart to fight with it.
December 13, 2002
Here is one final bit of information for those of you interested in a
more technical description of this entire system, I warn you it is heavy
reading.
www.engr.uky.edu/~amdatt0/Xfer.doc
December 17, 2002
This will be my final update.
Most of the errors in this page have been corrected. There were also some errors in the transfer
function documents. Those have also
been corrected. The following are the
schematics for the main board and power supply board. They are correct, matching the final layouts. These images will be hard to see if your
resolution is set to less than 1024x768.
Main Board
Power Supply
TIME LINE: (Organized
by the Monday of each week, this is a work in progress)
Sept. 30: Publish preliminary
web page.
Build and test low
pass filter with MAX264 chip.
Build and test hi pass
filter with MAX274 chip.
Receive LM347 quad
OP Amp chips. (on order from National
Semiconductors)
Oct. 7: Work up schematics for
all filter sections based on results with MAXIM prototype circuits.
Work on board
layouts.
Update web page.
Oct. 14: Finalize board layouts.
Verify complete
prototype circuit.
Attempt to have
boards fabricated.
Update web page.
Oct. 21: Start work on power
supply. (filter sections will be
working to my satisfaction first)
Update web page.
Dec. 9: FINISH EVERYTHING
Dec. 13: Project Presentation at
school
Dec. 14: (Saturday) Rest!
Parts List: (modified and now accurate)
Filter sections:
Bag
of assorted resistors (500) $5 radio
shack
Bag
of assorted Cer. Caps $3
radio shack
1
MAX264 chips $0
Maxim samples
2
LM347 chips $0
National Semiconductors samples
28
pin dip socket $1
radio shack
1
TTL Oscillator $0
1
CD4060B $0
Assorted
connectors $0
courtesy of Lexmark
8
0.1uF SMD caps $0
courtesy of Lexmark (1206 package)
7
1.0Kohm SMD resistor $0
courtesy of Lexmark (1206 package)
Power Supply:
SMD
24V zener diode $0
courtesy of Lexmark
Assorted
Ele. Caps $0 from
broken computer monitor
PT79SR105
Neg Regulator $0 TI samples
PT5101
Pos Regulator $0 TI
samples
ProtoBoard
solder type $3 radio
shack (used only in prototyping)
6
0.1uF SMD caps $0
courtesy of Lexmark
Relay Board:
4
2N2222 transistor $0
(TO92 package)
4
1N4125 SMD diode $0
courtesy of Lexmark
4
4.7Kohm resistor $0
courtesy of Lexmark (1206 package)
4
12v DPST relay $0
broken computer monitor
MISC:
1
1”x12”x8’ pine board $11
Home Depot (for speaker enclosure)
Assorted
hardware $6
12”
ruler $0.50
References: (I will try
to link some of these later, they can all be found at www.digikey.com )
Data Sheet MAX264 http://rocky.digikey.com/WebLib/Maxim/Web
Data/MAX263,264,267,268.pdf
Data Sheet LF347 http://www.national.com/ds/LF/LF147.pdf
Data Sheet CD4060B not
linkable from TI website, use keyword search on part number
Data Sheet PT5101 not
linkable from TI website, use keyword search on part number
Data Sheet PT79SR105 not
linkable from TI website, use keyword search on part number
Data
sheets for the 3 TI parts are included in allfiles.zip.