white noise to learn, aim, and adjust speakers
| AIMING and "LEARNING"
Here is a method for "aiming"
(and learning) speakers --- for determining the splay angle AND the "lobe"
of the sound coming out of [any] speaker. This method is far more accurate
than ANYTHING else, even lasers. Doing
it with your ears with this procedure takes EVERYTHING into account,
including interchannel digital delays or anomalies, circuit group delay,
phase shifts, if any, anywhere in the chain, including passive or active
crossovers, and it also takes into account mechanical interchannel timing
issues (often somewhat incorrectly called time alignment) caused by the
simple fact that the speakers or the drivers are different distances from
The physically larger your
speakers are, such as with 7 foot tall floorstanders or large electrostatic
panels, the more sensitive and revealing this test is. However this test
is still very useful with smaller speakers such as 2-way "bookshelf"
speakers, and VERY useful with very large systems such as clubs, discos,
line arrays, halls, etc.
The procedure is this:
Use the WHITE (not pink) Noise Track #25 on the TEST CD. The
track is now 4 MINUTES and "stereo" (Both channels are sample-accurate
/ identical, in-phase) and are recorded at -20dBFS.
Please note that -20dBFS (20 decibels below Full Scale Digital) is
so-called "REFERENCE LEVEL".
Please be aware that OTHER tracks
on the CD are recorded 19dB HOTTER, at -1dBFS.
Set your CD player to NOT continue to the next track(s) !
However, if you play the
white noise at 75dB, then even if the other tracks play they
will not be louder than 95dB, so no damage will be done.
Ideally, use a separate CD (or DVD) player with ANALOG OUTPUTS. Connect
ONE analog output (the LEFT CHANNEL) to a number of "Y"
cords. You SHOULD do it this way . Do NOT attempt to use "mono"
switches in the receiver or preamp. The setup should look like this:
Fig 1. Method of using Y-cord
adapters for multiple feeds.
This is the same thing as "daisy chaining"*
and also the same thing as "connecting all the inputs in parallel".
"daisy chaining" is a ridiculously untechnical phrase. Please
do not use it.
If you have a CD player where you CAN use the analog outputs (especially
for a 2-channel setup) then use TRACK 25 (which has the L and R IN
phase) for in-phase tests.
If you have a CD player where you CAN NOT use the analog outputs
then use an HDMI connection (or digital coax) and use TRACK 26 which
has the L and R OUT of phase with each other when doing the cancellation
test(s) (steps 15 + 16)
Disconnect the sub or turn its power off.
White Noise Page 1 of 4
Make sure the TREBLE controls for each channel are set at "0".
If your receiver or preamp has level trim adjustments, make sure they are
all set the same, preferably to unity gain.
If you have the ability, I suggest turning all the BASS levels all
the way down. If you are attempting to learn the splay pattern of speakers
which are a flat panel, such as electrostatics (Quad, Sound Lab, Magnepan,
Martin Logan, etc) then you might want do a separate test with the bass
at "0" (or higher) in order to learn how the lower frequencies
leave the dipole and bounce off the wall behind the speaker. But for now,
please do the test with the bass turned down.
Since the wavelengths at high frequencies are
so small, by turning the bass down you are simply removing the longer wavelengths
from your auditory test. Please see my frequency-to-wavelength chart here:
If you have channel "delays" (sometimes called "distance
settings" in a Home Theater receiver), MAKE SURE they are set OFF or
to the minimum, and ALL THE SAME. I suggest setting these settings for the
ALL the TOP speakers (L C R Ls Rs) at 7 feet as the entire concept
and in many cases the execution (bass management pickoff spot) is flawed.
However, you CAN use this concept to "fix" the inherent group
delay in a powered, sealed subwoofer, by adding an equivalent delay to all
the top speakers to match the sub's delay. Please see my white paper here:
Make sure that the noise floor of the room is quieter than perhaps
50 to 55 dBA Slow weighted C. Turn off fans and air conditioners. The signal
you will be listening to should be perhaps 18 - 20 dB louder than the noise
floor of the room. The objective is to get the test signal loud enough to
understand but not so loud that you get a headache, and not so loud that
you excite room modes, which is another whole topic of discussion.
Plug in the LEFT channel only (for example, use the AUX ins, sometimes
listed as "analog 5.1 inputs".) Advance the main volume control
to give a MODERATE level in the room. My suggestion is to use about 70 -
75 dB SPL (use Slow weighted C), not higher than 85 dB. If you are using
an HDMI / coax connection you will have to disconnect one speaker.
Start by sitting in your "sweet spot" chair. Now stand up.
Try and discern the difference in the splay lobe from your SEATED to STANDING
POSITION. You might hear a frequency change; you might hear some combing,
especially if you have a tall line array. Notice that even if you DO hear
a combing effect, you cannot hear it unless your body is moving! This is
one reason why no one complains about multiple-driver combing in real-life
use: you are not usually getting up and sitting down while listening critically.
Now walk around the L speaker in an arc and try to discern
the high frequency splay or lobe pattern of the speaker. Learn the sound
of the speaker from far away, to closer, until you get to the real near
field, say, closer than 1 meter (3 feet). Get a feel for BOTH how the speaker
is sending the waves out and HOW THEY ARE REACTING WITH THE ROOM.
You should be able to discern the splay pattern of the speaker and get a
mental picture of the sound almost as if it were a "theatrical flood"
or "spotlight". Cup your ears so they are directional and face
the back wall of the room and try to determine what sound, if any, is reflecting
from back there.
Sit in the sweet spot and cup your ears and try and discern what is
bouncing off the 1st reflection point on the side wall. Typically this is
the most important spot to have a wideband absorber. Since the path length
from each speaker to the side wall is different than the path length from
the speaker to your face, the summation of these out-of-time (and therefore
out-of phase) signals will often produce comb-filtering anomalies. There
are also reflection areas on the floor between you and the speaker, and
on the ceiling as well. You might be able to hear the localized reflection(s)
by cupping your ears and directing your attention to the area in question.
Another method to determine side reflections is to have a 2nd person hold
a mirror flat on the wall on the sides until you can see the front of the
speaker in the mirror when you are sitting in your chair. That is therefore
the "main" spot to apply absorbtive treatment; specifically a
Turn OFF the L and turn ON the R. Do the same
test again with the Right channel.
Now turn on BOTH the L and R. Plug the Y cord into the
L and R inputs. Assuming for the moment that the L
speaker is already positioned "where it belongs" then have another
person move ONE speaker (for example the R) while you are listening
in the sweet spot. When the R speaker is aimed into the room correctly
so it matches the L speaker, the high frequency signal should SNAP
TO A "DOT" in the center. You should perceive a small "dot"
of sound --- NOT a large diffused ball or indeterminate globule of noise
that seems to be everywhere. If you cannot get the sound to become this
"dot" then something is wrong. It could be the wiring, the receiver,
and of course even the speakers, but typically it is a combination of aiming
and reflections which diffuses the focus. Once this focus is achieved your
imaging should be better, if not uncanny.
Noise Page 2 of 4
The test above has the speakers set up "normally", that
is, wired correctly, IN POLARITY with each other; (often incorrectly called
IN PHASE) Now we are going to try a MUCH more critical test.
Reverse the POLARITY of one of the speakers. If you are calling the
LEFT channel the "reference channel" as far as positioning goes,
then reverse the wiring to the RIGHT speaker, in case you accidently move
it a little bit. Now the speakers are OUT OF POLARITY with each other.
Since the speakers are OUT OF POLARITY with each other, when you play
the white noise through BOTH you should hear a NULL, i.e. "nothing".
The more accurately you perform this test, the more the 2 sources will cancel
out. If you still hear a loud diffused glob of sound then something else
is wrong - I have been surprised many times that in a speaker that is a
line array, ONE driver might be wired incorrectly. Leaving the LEFT channel
in its reference position, have the other person adjust the RIGHT speaker.
As the right speaker is pivoted and tilted, you should be able to mechanically
"tune" it until the null becomes the sharpest. We can only assume
(or hope) that whatever the internal [passive] crossover in your main speaker
is and that entire cabinet is an electro-mechanical and physical
entity the manufacturer has correctly set it up!
When you SUM 2 "exactly the same" signals IN phase
(in polarity) they algebraically sum so that the net result
is 6dB louder, or twice the voltage or Sound Pressure Level.
When you sum 2 signals OUT OF POLARITY they cancel completely,
which would literally be 50 or 60 dB (or more) weaker. That
is why it is so much easier to hear a NULL rather
than a PEAK.
As a further corollary, when you
add 2 speakers together in a room, because of the typically
de-correlated signals AND the fact that the speakers are spaced
apart and have slightly differing coupling modes to the room,
do NOT expect a 6dB increase in the room; expect a 4 or 5 dB
increase. This is one reason why you are using y-cords; so there
is absolute correlation to start with, which then passes through
all the circuitry and anomalies in your system.
If you have electrostatic panels, this is where the tilt adjustment,
both vertically and toe-in become most critical. Since the surfaces are
FLAT and since the same signal is emanating from everywhere on the surface,
this test becomes remarkably sensitive and you should be able to discern
cancellation changes on the order of 1/4". It may take some time to
learn this phenomena. IF your speakers are on a carpet then this is the
ONLY time I would ever suggest using (and adjusting) spikes. In any event,
you don't want the springback of the carpet to interfere with your tests.
You want both speakers to be rock steady, and to stay where you put them
Also, with flat panel / electrostatic speakers, since they are dipoles
there is an equal sound coming off the back, and hitting the wall behind
the speaker, then bouncing around forward. When you turned the bass down
earlier you are only listening to the higher frequencies coming off the
panel. With flat panel speakers only, you might want to experiment by turning
the treble all the way down, and turn the bass all the way up; this will
enable you to hear and experiment with the back wave. I suggest (read my
SUBS white paper again...) that you completely absorb the back wave from
a flat panel dipole and only attempt to match a sub up with the front wave.
Put the polarity of the Right Channel back where it belongs.
If you have a Home Theater 5.1 setup, after you are FINISHED determining
the L and R signals, listen to the C channel by itself.
Then have someone else change back and forth between both the L and
R wires and the C wire only. Now you are listening to determine
how the REAL C sounds relative to how the PHANTOM C sounds.
This part of the test is extremely critical. It will immediately
point out room and acoustic issues which might smear the sound such as early
reflections from the side walls and other reflections. Please note that
in movie DVD's, the L and R are for Music & Effects and
the C is for dialog. There is NO panning correlation that takes place
between L C and R. Music is panned LR. This is another
reason why attempting to set the speaker distances to real-room measurements
is wrong; typically the signals coming from the LR, and C
and Ls Rs have very little to do with each other. It is only the
LR that are phase correlated.
Remember that you are going to try to get a number of perceptions:
a) The direct signal from the L
b) The direct signal from the R
c) The direct signal from the C channel
d) The PHANTOM CENTER channel image from the acoustic summation of
the L and R
e) Early reflections, such as off the side walls.
f) The reverberant field
g) The later echo field, including flutter echo and reflection(s)
off the back wall.
|Using White Noise
Page 3 of 4
Do not be surprised if you think you are getting results from this
test that you may consider odd. This test is probably the MOST sensitive
test you can ever do where your hearing is part of the measuring equipment.
When switching between the C only
and L and R only, this is a VERY sensitive way to adjust
the inter-channel balance.
23) If you have a Home Theater setup, when you are finished with the
LCR part of the system, it is helpful and educational to learn how the Ls
Rs are splaying into the room as well. One interesting test is to have
someone hold the speaker at your ear level while you are sitting down and
they then move in an arc from 90 degrees to 165 degrees when measured from
the Center channel line, as in the diagram below. Note the Rs is
shown at 110 degrees of arc from the C.
You will usually find that the best place for
the Ls Rs is when they are in the psychological null of the listener
which roughly corresponds to the acoustic null of the listener as well.
What does this mean? If and when sounds are from directly behind you, they
tend to draw your attention AWAY from the movie and may be frightening to
young children and elderly people, who are not prepared to expect loud noises
from behind. The other extreme is sounds coming from directly to your left
and right (90 degrees) which again, make you turn your head and turn your
attention AWAY from the movie. But at an angle of about 110 degrees, something
amazing happens: those sounds are no longer frightening and no longer divert
the attention away from where it belongs, but they are integrated smoothly
into the theatrical experience, and they integrate into the "surround
Using this white noise test, you are now able to discern the splay pattern
of your surrounds and how they are integrating into the room. If you have
surrounds which may be switched or changed between front-firing to dipole
mode to Tripole mode (such as M&K's) then you will find this part of
the test to be exceptionally interesting.
These series of tests are not
only very accurate but a rather inexpensive way (i.e. NO test equipment
is required, other than your ears and some Y cords...) to get VERY precise
results. Good luck with your acoustics tests!
|Using White Noise
Page 4 of 4
page was last updated on
January 5, 2012