BRACING
Bracing is probably one of the most important things you can do to your
cabinet. No matter how good the drivers are, no matter how good the
crossover blends the drivers, if the cabinet resonates, it will destroy
the clarity and imaging of the speaker you have built. If the panels
resonate, they will produce their own sound which will interfere with
the sound being radiated by the drivers. In discussing
resonance
of panels in this context, what is being discussed is the excitation of
the actual panel; NOT
the standing wave bouncing back and forth between the wall panels.
Excitation of a panel is by three primary means:
Vibration from the vibrating air in the enclosure,
Vibration from the driver, and
Vibration from movement of other panels.
An article by James K. Iverson titled, The Theory of
Loudspeaker Cabinet Resonances in the Journal of the Audio
Engineering Society, April 1973, Volume 21, Number 3, studied panel
resonances. Some of the results are summarized:
Above 1,200 Hz, very little panel movement is observed.
Lesson:
Raise the panel resonant frequency into this range so the panel won't
flex,
Resonance can be raised above 1,000 Hz (where panels are
less
likely to flex) by spacing braces no farther than 10" apart,
From curves of cabinet pressure vs. frequency, there is
very
little air pressure above 500 Hz to force movement of the panels,
A hole in a panel decreases the plate stiffness which
lowers the
fundamental frequency. Likewise, so does suspending a mass on
the
panel. The upshot of this? Brace that baffle!
While absorptive materials may slightly help reduce panel
excitation by absorbing some acoustic energy, it is no where near as
effective as bracing,
Bracing is most effective in the longitudinal direction and
can
raise the resonant frequency by 2 1/2 to 3 times,
He showed results of another experiment by P. W. Tappen on bracing a
metal plate clamped at all edges. In terms of raising the fundamental
frequency, the dowel cross-brace was the least effective, while the
longitudinal brace was the most effective. Here are Tappen's results
summarized in a chart:
As can be seen, the longitudinal brace is the most effective at raising
the resonant frequency. Iverson also tried a 1x2 pine brace with screws
at 3" on center bracing a plywood/particle board enclosure and noted
that it did not appreciably raise the resonant frequency. He believed
it showed that high stiffness braces were preferred. (Of course, it
could also show that he should have glued the brace instead of using
screws.)
At least two different articles (there may be more) from Speaker
Builder/Audio Express also explore Bracing. One article is entitled, Navigating
Speaker
Design: Listening to Walls, by Mark Wheeler (Speaker
Builder
8/99) and is the third
part of a three part series entitled, Navigating
Speaker
Design. In it he covers some background
on panel resonances and sets up a listening study whereby various
commercial loudspeakers were modified with bracing and then auditioned.
Following are some key observations:
Mid-panel front to back or side to side braces
(cross-braces)
were ineffective enough to show no observed change, or at the very
least, they were the least effective,
Vertical braces were more effective than horizontal ones or
even
circumferential ones. He states that a number of designers report the
greater effectiveness of bracing the long direction,
He notes that the type of glue can affect bracing quality
with
white wood glue being least effective, while two part epoxies were best,
High density (hardwood) braces outperformed low density
(softwood) or particle board. Even 1/4" thin hardwood braces maintained
their effectiveness,
Bracing effects were similar in orders of magnitude to
driver
selection,
The second article is by Jim Moriyasu from AudioExpress (Feb. 2002)
entitled, Panel
Damping Studies: Reducing Loudspeaker Enclosure Vibrations.
In
it the author builds a test box with an 8" subwoofer, with a removable
square panel and then clamps panels with various forms of bracing to
the box. He then measures
the vibration using an accelerometer and graphs the frequency and
waterfall response. He first tests blank panels of MDF, particle board,
and 7-ply birch plywood. His tests show that all three are very similar
in the frequency and level of resonance, though MDF shows a 1 to 2 dB
lower level of the primary resonances, but admits there's no major
differences between the three. He decides to use the MDF panel for all
the
bracing tests. He then tests how much air vibration is cut down by
stuffing the enclosure. He notes a 2 to 3 dB reduction in peak levels.
He goes on to test multiple panel layers, applied damping materials,
constrained layer damping, and bracing (as well as woofer isolation).
He tried the following braces: 1 1/8" dowel rod cross-brace, 3/4" thick
x 2 1/2" wide MDF brace (placed across the width of the panel), and
various shelf (window) braces
featuring
square windows, oval windows, and a single large oval. His results
favor the 3/4" x 2 1/2" wide MDF brace placed across the panel. He used
a square panel so there
is no distinction between widthwise or longitudinal. The window braces
were next best with the oval windows outperforming the square
ones. The single oval one was not quite as good as the other
window braces. The dowel rod showed mixed results - some higher modes
were reduced by 10 dB, but lower modes did not change much.
The
third mode was actually 5 dB higher in level than no brace at all. When
I have time, I will chart his results for bracing as well as the other
forms of damping he checked.
From the studies above, it is apparent that the purpose of bracing is
to raise the panel's natural resonant frequency up to a point where it
won't flex the panel. Stuffing and applied damping compounds can then
control any remaining resonances and will keep the panel from
"singing." If you
divide a panel in half, you double it's natural resonant frequency. Of
course we are also adding stiffness, which raises the resonant
frequency even more - for a total of perhaps 2 1/2 to 3 times the
original frequency. If we divide the panel with bracing every
10"
or so, we keep the
resonance above 1,000 Hz where it's less likely to cause the panel to
move. Bracing "damps" panel movement. Note that the other form of
damping is adding a damping compound
of some kind to the enclosure wall. This is the antithesis of bracing,
since adding mass lowers
the resonant frequency. With bracing we are trying to raise the
frequency.
I don't want to propose that you don't add some kind of damping
compound, because I do believe it's also beneficial in stopping the
panel
from transmitting sound. However, be aware that the panel can still
resonate at a particular frequency and color the sound even though you
have damped most other vibrations. A combination of bracing
and damping is best, but if using damping compounds, or if you apply
heavy coats of it, you may want to
apply bracing even more liberally. Perhaps the spacing can be
decreased to 6" or 8" instead of 10".
Other
notes:
Make sure the brace is rigidly
attached. Don't just screw the brace to the panel. It must
be
glued throughout it's length! And clamped tight! Therefore, if you use
screws and glue, drill the hole in the panel slightly larger than the
diameter of the screw threads so the threads won't bite into the panel
itself. Of course the screw hole in the brace must be smaller
than the threads. :-) Then when you put the screw through the panel and
into the brace, the threads will bite into the brace and pull the brace
and panel together providing very high clamping pressure. Otherwise, if
the screw threads were to bite into both the panel and brace, there
will always be a minuscule gap between the brace and the panel that you
can't quite get rid of by tightening the screw harder, and the two
pieces won't bond properly.
As mentioned before, use hardwood braces. I used oak braces
in my subwoofer.
Another point to consider if using a longitudinal or widthwise brace -
Don't put it in the center of the panel! Offset it slightly to the side
(say somewhere between 1/3 and 1/2 the width of the panel). This will
cause the two "subpanels" to resonate at different frequencies instead
of "stacking" the resonant frequencies on top of each other.
I can recall one other Speaker Builder article that I've read and if I
locate it I will add it's conclusions. more
to come...
