home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Night Owl 6
/
Night Owl's Shareware - PDSI-006 - Night Owl Corp (1990).iso
/
040a
/
loudsp30.zip
/
LOUDSP30.DOC
< prev
next >
Wrap
Text File
|
1991-11-28
|
16KB
|
303 lines
*******************************
** LOUDSP30.EXE **
*******************************
WELCOME TO THE WORLD OF LOUDSPEAKER DESIGN! If you are just starting your
first system, or already have several under your belt, you will find this
program a useful addition to your "toolbox".
You are probably aware of many programs on the commercial and shareware
market, which perform speaker design calculations. They are all good, and
in fact, some are capable of performing more functions than LOUDSP30.
Though they may display data in different ways, most arrive at their
results by using various interpretations of the Thiele/Small equations
established some years ago. It is not surprising then, that designs
produced by these programs tend to be similar, if not identical, since the
calculation methods are well established.
So why use LOUDSP30 to get your answers? While doing my own designs, I
tried several different programs. Though they all produced accurate
results, the user interfaces were cumbersome and confusing. Often simple
functions like data storage were far more difficult than necessary.
LOUDSP30 is designed to put most commonly used design functions at your
fingertips, and allow rapid repetition of the major design routines.
Since a properly researched speaker design can easily take a dozens of
trial calculations, the ability to cycle different data through the
program was given high priority. Support functions, like port length, and
crossover design, have been included, eliminating the need to switch to
other programs or formula tables.
A word about crossovers. The very best designs take into account driver
phase response, distances, offsets, and several other factors. Those
designs can some require heavy duty math to manipulate the transfer
functions, and even the experts are not in agreement on the best practical
design methods. The more sophisticated programs have a tremendous
advantage there, and you cannot duplicate their performance using commonly
available formulas and a calculator! If you have a way to measure the
phase response of your drivers, one of the "high end" analysis programs
will be an excellent investment. If, like most of us, you do not, then
the general purpose crossover designs in LOUDSP30 should be adequate
(combined with a bit of listening and fine tuning).
LOUDSP30 has been developed from various sources, including Speaker
Builder Magazine, The Loudspeaker Design Cookbook, Audio, and Weem's How
To Design, Build And Test Complete Speaker Systems. Obtaining any or all
of them is highly suggested to increase your understanding of the many
interactions that occur in speaker design. It is important to get several
viewpoints, as most designers have specific areas of expertise, and tend
to emphasize that area in their designs. Don't look at one design and
then think there is no other way to do the job.
LOUDSP30 will help you achieve a specific low frequency response via the
driver, box size, and tuning chosen. Most of us have been somewhat
brainwashed into believing that a "flat" anechoic response is the best,
that a flat response is the result of a technically perfect design. This
view is reinforced by manufacturers who want to present data in simple
terms to (relatively) unsophisticated customers. It is also encouraged by
design goals for electronics, where flat response generally does represent
good design. The truth is that flat anechoic response is not the best
choice in most real world situations.
Different rooms (and vehicles) will have different volumes, different
proportions, different damping, and a variety of other factors which you
must consider. Like it or not, the room is part of the complete
loudspeaker design. As an example, compare a speaker designed for corner
placement, vs. a speaker designed to stand several feet away from all
walls. Both choices have merit in different applications, but they will
require radically different low frequency characteristics for accurate
bass reproduction.
>>>>>> THINK ABOUT YOUR OVERALL DESIGN GOALS BEFORE RUSHING TO <<<<<<
>>>>>> BUILD THE BOX DESIGN WITH THE FLATTEST RESPONSE CURVE! <<<<<<
So much for philosophy...
In a nutshell, here's how the program works:
You must start with the performance parameters for a given driver. You
will need (as a minimum), the Free Air Resonance (FS), the "Q" (Qts), and
the Compliance (VAS). These numbers should be supplied by any quality
driver manufacturer. If unavailable, they can be measured. Use the
parameter selection in menu #2. These values are then entered into one of
the three major design sections of the program. You can do a ported box,
with the box size and tuning automatically set. You can do a ported box,
with both size and tuning entirely of your own choosing. Or, you can do a
sealed box with your choice of total system Q. Most other program options
are self explanatory, and are support functions for the major box design
routines.
The automatic ported routine will tell you the correct box size and
tuning. You can adjust the box size, and the tuning will be recalculated.
When you are happy with the results, the correct size port to produce this
tuning can be calculated in menu #2. Think of the air in the box as a
spring, and the port volume as a mass hanging on the spring. Changing the
port changes the frequency that the whole thing resonates at, ie: the
tuning frequency. The design routine will tell you the tuning frequency,
then you must go to menu #2 to determine the port size to achieve it.
If you want to try different combinations of box size and tuning, the
modified routine will allow this. It does not produce a data screen, only
a graph or printout.
The easiest design is the sealed box. Common wisdom suggests that the
total system "Q" value (Qtc) be close to 0.707 because that produces the
flattest response. Some even promote 0.577, as that has the best impulse
response. Frankly, these designs sound thin to me. The most satisfying
results have Q values of about 0.9, except for large low frequency
systems, like subwoofers, where the lower values are suggested. 0.9 is
the default value in the program, but feel free to experiment.
A quick "golden rectangle" box dimension routine is included in menu #2.
It will give you an estimation of how big a system you are really dealing
with. Rectangular enclosures are to be avoided, but most speakers end up
with them due to ease of construction. You are much better off with at
least one non-parallel wall, or a curved enclosure. Another common error
is the use of 3/4" thick particle board. It just doesn't have the
strength to produce a solid, resonance free cabinet. This is hard for
most people to accept, until they have listened cabinets made with 1 1/4"
thick material. Again, do your own tests. Also be sure your ports are
large enough to avoid wind noise. 3" diameter works well if the design
can accept the resulting length.
One of your first tasks should be to go to menu #2 and run setup. This
will let you choose colors, video options, and a path for driver files.
You would normally have a subdirectory for LOUDSP30.EXE, LOUDSP.SCR, and
this document, then a subdirectory off of that for the xxxx.DRI files.
The program is supplied configured for AUTO video detection, which usually
is the best method. If the program will run, you can go to menu #2 and
use the configuration selection. If the program will not run, you can
edit the LOUDSP.SCR file with any ASCII editor. Set the first line to the
type of video you are using: "CGA", "HERC", etc. That should allow you to
get to the configuration area.
Here is the format for the SCR file:
"AUTO" > video detection mode
14 > text color (yellow)
1 > text background color (blue)
14 > graph grid color (yellow)
1 > graph background color (blue)
13 > plot color (light magenta)
2 > plot resolution (2 hertz)
"C:\LOUDSP\DRIVERS\" > path to find driver files
The colors are the usual set from 0 to 15: black blue green cyan red
magenta brown white grey lightblue lightgreen lightcyan lightred
lightmagenta yellow brightwhite. Plot resolution will affect both the
speed and smoothness off the graph. If you have a math coprocessor, a
value of 1 hertz is fine. Without it, a value of 2 for VGA, or 5 for CGA
will give better speed, without reducing accuracy significantly. Do not
include comments in the SCR file.
An important goal of this program was to have a very fast user interface.
Most speaker design involves repeating the calculations many times, with
very small changes in the input data. You will notice that the program
keeps default values, so you can step through each routine very rapidly.
Just change what you need to, and use the <CR> key to accept the defaults.
Yes/No questions can also be bypassed with any key other than "Y". This
lets you step very rapidly with the <CR> key or the space bar.
For serious errors, the program should produce an error message and a low
beep, then continue execution after a few seconds. Several error messages
may result, but usually the program will come back to a menu in short
order. Most inputs have minimum and maximum limits on what is accepted,
and will refuse data which is unreasonable. Like anything, you may be
able to break it, but a full crash should be a very rare event!
LOUDSP30 is very similar to it's earlier releases on the surface, however
it is a major re-write in terms of code and program flow. It is hoped
that the improved consistency between screens, and improved video support
will enhance your enjoyment of using it. Some driver files have been
included, but you should consult the latest manufacturers catalogs for
current data.
*********************************
Here is some additional information on both the program, and on speaker
design in general:
---1---
Always run impedance curves of the drivers. If they are not flat, (they
never are!) then consider using simple networks to flatten them. This
should usually be done with the woofer in a two way system, and with both
the woofer and midrange in a three way system. Often a midrange will
require two networks. The first will cancel out the rise in impedance due
to the inductance of the voice coil. The second will cancel out the
resonant peak at the low end. REMEMBER: THE CROSSOVER WILL ONLY WORK
PROPERLY IF THE SPEAKERS IMPEDANCE IS FLAT AROUND THE CROSSOVER FREQUENCY.
---2---
Considering the above, it is usually wise to build a two way system if it
can do the job, as the three way design is quite complicated if it is done
right. Consider an active crossover and separate power amps if the
crossover design is getting out of hand!
---3---
The traditional butterworth crossover formulas have fallen out of favor.
They yield highly unpredictable results, which must then corrected by
trial and error until the system sounds right. Use the crossover designs
from the Loudspeaker Design Cookbook, or from the 1985 issues of Speaker
Builder Magazine. The second order Linkwitz/Riley designs seem to produce
the best results for the least complexity. (That is the design used by
LOUDSP30.)
---4---
LOUDSP30 was written in PowerBASIC, a product of Spectra Publishing Corp.
It is the evolution of Borland's TurboBASIC, written by Robert S. Zale.
Its a great compiler, cheap, and with super manuals. It produces small,
FAST, programs. It is almost fully compatible with brand "M". Try it,
you'll like it!
---5---
The program will support a math coprocesser if one is present. It will
emulate one if not. Error trapping is provided. Whatever is current in
the program is what is stored. IT ALWAYS REMEMBERS THE LAST DESIGN
PARAMETERS ENTERED IN EACH SECTION AND CARRIES THEM TO THE NEXT!! (This
is true of the major design sections, minor routines sometimes keep the
values unchanged to avoid confusion.)
---6---
The parameter section requires the use of test equipment to measure driver
parameters. If you do not have the background or equipment, you may be
able to find a local technician to perform the measurements, or possibly a
high school electronics lab. Weem's book has a good section on
exactly how to make the measurements. If you are really serious, you should
try to get your own signal generator and AC voltmeter. Ask one of your ham
radio friends when the next "hamfest" is. Go to the flea market and look
for working Hewlett-Packard audio frequency generators. The early tube
equipment is rarely more than $50-100, though may require some tubes and a
tune up.
---7---
LOUDSP30 is now far smarter about your video system. In AUTO mode, it
will detect most Hercules, CGA, EGA, and VGA adaptors. Unfortunately,
there are many non-standard video configurations, and it is hard to design
routines which can sort out every one. If you have trouble, you can force
a specific adaptor in the setup area. At worst, use any ASCII editor to
modify the first line of the SCR file to read: "HERC", "CGA", "EGA", or
"VGA", as required.
The VGA graphs are done in mode 12, which is 640 x 480. If you cannot
support this, back down to EGA, which is mode 9, 640 x 350. CGA is mode
2, 640 x 200, no pretty colors. Note that many EGA cards have extended
modes if used with multi-frequency monitors. You may be able to run in
VGA mode if this is the case.
Very little testing has been done on Hercules systems, mostly because
almost no one uses them anymore. Testing has been done with VGA boards,
locked in the emulation mode. The opening screen is active in VGA and EGA
(and runs in 640 x 200 to allow page swapping), but is static in the other
modes. You can hit F or S to go faster or slower if you have the active
opening screen. There is no internal graph printing routine yet, but many
screen dump utilities will do the job. You may wish to load graphics.com,
then set the program to CGA monochrome. This gives a useable result, if
not a work of art!
---8---
There are a few environment variables which can affect the program.
Oddball video hardware can sometimes be forced to work by using these.
You will not often need them, but here is a list:
SET 87=YES SET 87=NO
SET CGASNOWCHK=YES SET CGASNOWCHK=NO
SET HERC=YES SET HERC=NO
SET VGA=YES SET VGA=NO
---9---
This program is offered with no guarantees as to its suitability for any
purpose. The writer assumes no liability for any loss or damage resulting
from its use, misuse, or failure to operate as described or assumed.
This program and source code are hereby released to the public domain.
You may use this program for personal or commercial endeavors, ie: use as a
development tool for commercial products is OK. No fees are requested.
*****************************
Finally, please support quality shareware, when fees ARE requested. You
may not be aware of just how much time it takes, not only to develop an
application program, but to test it on a large number of different
machines, document it, and support it when questions arise. Though not a
large program, LOUDSP30 has several hundred hours of development and
testing invested in it, yet it is stilll far from perfect.
Trademarks:
PowerBASIC Spectra Publishing
TurboBASIC Borland International
>>>>>>>>>>>>>>>>>>>>>>>>>>>> end of file <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<