The great low-frequency debate

The great low-frequency debate
sweet spot
The great low-frequency
deception
‘Fantastic, true sounding bass: these small monitors tell you exactly what is on the recording’ – a statement typical of what you read in
many advertisements, but it is often far from the truth. In fact it cannot be true says PHILIP NEWELL. At realistic monitoring levels, the low
frequency response of small loudspeakers cannot be as accurate in terms of frequency response and transient response as a good large
monitor system, flush-mounted in the front wall of a well-controlled room.
M
OVING COIL LOUDSPEAKERS in boxes
are volume-velocity sources. The acoustic
output is the product of the area and the
velocity of the diaphragm, so for any given output
either a large volume of air can be moved slowly, or a
small volume of air can be moved quickly. For a given
frequency and SPL a large diaphragm does not need to
move as fast or as far as a small one. The smaller
loudspeakers need longer throw, faster-moving cones,
but the restricted volume of the air inside the smaller
boxes experiences a much greater pressure difference
between the extremes of the cone excursion than
would be the case in a larger box.
Let us say that the diaphragm of a 15-inch woofer
in a 500-litre box moved 2mm peak to peak. With
diaphragm radius of 6.5-inches, or 160mm, the
radiation area would be 80,000 mm2. Moving 2mm
peak to peak means moving 1mm from rest to the
peak in either direction, so the unidirectional
displacement would be 80,000mm2 x 1mm or
80,000mm3. This is equal to 0.08 litres, so the static
pressure in a 500-litre box would be compressed (if
the cone went inwards) by 0.08 litres, or by one part
in 6250 of the original volume.
For the same SPL, a 6-inch (150mm) loudspeaker
in a 10-litre box would still need to move the same
amount of air. However, with an effective piston radius
of 2.5-inches, or 65mm, the cone travel would need to
be about 12mm peak to peak, so the cone would also
need to travel six times faster than the cone of the 15inch loudspeaker. What is more, the displacement of
80,000mm3 (0.08 litres) in a box of only 10 litres
would represent a pressure change in the box of one
part in 125 of the original volume. The air
compression inside the box would therefore be 50
times greater than that in the 500-litre box, and there
are several consequences of these differences.
Anybody who has tried to compress the air in a
bicycle pump with a finger over the outlet will realise
that air makes an effective spring. They will also know
that the more the air is compressed, the more it resists
the pressure. The force needed to compress the air by
each subsequent cubic centimetre increases with the
compression, so the process is not linear.
In the case of the 15-inch and 6.5-inch cones, the
small cone in the small box would have a much harder
job to compress the air by 1/125th of its volume than
the large cone in the large box, which only needs to
compress the air by a 1/6250th part in our previous
example. Large boxes therefore tend to produce lower
distortion at low frequencies, because the non-linear
air compression is proportionally less. The concept is
48
shown diagrammatically in Figure 1.
The non-linearity of the air spring can be better
understood when you consider that it would take an
infinite force to compress 1 litre of air to zero volume,
yet it would take only a moderate force in the opposite
direction to rarefy it to 2 litres. The forces needed for a
given change in air volume (in this case +/- 1 litre) are
thus not equal, so the restoring forces applied by the
air on the compression and rarefaction half cycles of
the cone movement are also not equal.
a
Fig. 1. Each pressure change of 10 newtons produces
progressively less change in the volume of the gas. It’s not
linear and can give rise to harmonic distortion.
b
c
Fig. 2. Waterfall plot of a small, sealed box loudspeaker: the
NS10M.
Fig. 4. Step function responses of the loudspeakers measured
in Figures 2 and 3. Plot c) shows the electrical input signal to
which the loudspeakers are responding. Note how rapidly the
NS10 returns to a flat line on the zero amplitude axis.
Fig. 3. Waterfall plot of a small reflex (ported) enclosure of
similar size to that in Fig. 2.
resolution
May/June 2003
sweet spot
The non-linear air-spring forces thus vary with the
degree of displacement and also with the direction of
the displacement. Changing air temperature inside the
boxes also adds more complications with waste heat
from the voice coils.
The main controlling factor for the extension of the
low frequency response of a loudspeaker is its
resonant frequency, because the low frequency
response of any conventional loudspeaker system will
begin to fall off quite rapidly below the resonant
frequency. The resonance is a function of the stiffness
of the air spring formed by the air inside the box,
coupled with the moving mass of the loudspeaker
cone/coil assembly. The fact that air inside a small box
presents a stiffer spring than air in a larger box, for any
given air displacement, means that it will raise the
resonant frequency of any driver mounted in it;
compared to the same driver in a larger box (i.e.
loaded by a softer spring).
The only way to counter this effect, and to lower the
resonant frequency to that of the same driver in a
larger box, is to increase the mass of the cone/coil
assembly. (Imagine a guitar string; if it is tightened, the
pitch will increase. Maintaining the same tension, the
only way to lower the note is to thicken the string, i.e.
make it heavier.)
To move the heavier cone to displace it by the same
amount as a lighter cone in a larger box, more work
must be done, so more power will be needed from the
amplifier. The sensitivity of a heavy cone in a small
box is therefore less (for the same resonant frequency
and bass extension) than for a lighter cone in a larger
box. Therefore, for any given drive unit, as the box
size decreases, the bass extension must also decrease.
As previously stated, increasing the mass of the
May/June 2003
moving parts can restore the bass extension, but the
sensitivity will reduce. There is currently no way out
of this dilemma.
However, larger boxes often tend to use larger drive
units and a large diaphragm will tend to be heavier
than a smaller one and the diaphragm may also need
to be heavier to maintain its rigidity. This would
suggest a lower sensitivity in free air, but a larger
magnet system can easily restore the sensitivity.
In a small box, the greater pressure changes may
also require a heavier cone (so as not to deform under
high-pressure loads) and the efficiency can again
reduce. A bigger magnet could be an answer but it
may reduce the internal volume of the box, hence
stiffening the spring more and raising the resonant
frequency, which may, or may not, be offset by the
extra weight of the cone.
Let us consider two loudspeakers of similar
frequency range but very different size. A large
loudspeaker, such as the double 15-inch woofered Urei
815 driven by 1 watt would give the same SPL at 1
metre as a small loudspeaker, such as the ATC SCM10,
driven by almost 200 watts. There is a currently
unbreakable connection between box size, low
frequency extension, and sensitivity. Reducing the
box size demands that the low frequency response will
be reduced or the sensitivity will be reduced. If the
sensitivity is to be increased, then the box size must be
increased or the low frequency extension must be
reduced. High sensitivity and good low frequency
extension can only be achieved in large boxes.
If the ATC seeks to achieve a good low frequency
extension in a small box, then the sensitivity must be
low. The ATC SCM10 has a box volume of about 10
litres; the Urei 815 contains almost 500 litres. Given
resolution
that they cover the same frequency range, a sensitivity
difference of about 22dB is the result.
When small cones move far and fast, they also tend
to produce more Doppler distortion (or frequency
modulation), and this problem is often exacerbated by
the small woofers being used up to higher frequencies.
Long cone excursions also mean more movement in the
cone suspension systems and the restoring forces are
rarely uniform with distance travelled. This tends to give
rise to higher levels of intermodulation and harmonic
distortion than would be experienced with a larger cone,
of similar quality, moving over shorter distances.
The larger movements also require greater
movement through the static magnetic field of the
magnet system, which tends to give rise to greater flux
distortion and even more audible non-linear Bl profile
distortions. Furthermore, the reduced sensitivity of the
smaller boxes means that more heat is expended in
the voice coils compared to that produced in the voice
coil of larger loudspeakers for the same output SPLs.
This problem is aggravated by the fact that the smaller
loudspeakers have greater problems in dissipating the
heat. The hotter the voice coil gets, the more its
resistance increases and the less power it can draw
from the amplifier for any given output voltage. The
resulting power compression produces yet more
distortion products.
It can clearly be seen that the distortion
mechanisms acting on small loudspeakers are far
greater than those acting on similarly engineered large
loudspeakers.
However, the market demands more output of a
wider bandwidth from ever-smaller boxes, so
manufactures try to rise to the challenge. One example
of a technique used to augment the low frequency
49
sweet spot
output is to use a reflex-loaded cabinet, with one or
more tuning ports. In these systems, the mass of air
inside the ports resonates with the spring that is
created by the air trapped within the cabinet. If the
resonant frequency is chosen to be just below where
the driver response begins to roll-off, then the overall
response can be extended. The resonance in the
tuning port(s) takes over where the driver begins to
lose its output.
This effective extension of the low frequency
response also increases the loading on the rear of the
driver as resonance is approached. This helps to limit
the cone movement and to protect the drivers from
overload. Unfortunately, once the frequencies pass
below resonance the air merely pumps in and out
through the ports and all control of the cone
movement by the air in the cabinet is lost. In many
active systems, electrical filters are used to sharply
reduce the input power to the driver below the cabinet
resonance frequency. This enables higher acoustic
output from the loudspeaker systems, within their
intended bandwidth of use, without the risk of
overload and mechanical failure due to high levels of
programme below the resonance. By such means, a
flat response can be obtained to a lower frequency
than with a sealed box of the same size, and the
maximum SPL can be increased without risking drive
unit failure. There is a price to be paid for these gains.
It must be understood that a resonant system can’t
start or stop instantly. The time response of reflex
loaded loudspeakers therefore tends to be longer than
that of similar sealed box versions. This means that
transients will be smeared in time. The impulse
response will be longer. Moreover, the effect of the
electrical high-pass filters is to further extend the
impulse response, because the electrical filters are also
tuned, resonant circuits. In general, the steeper the filter
slope for any given frequency, the longer it will ring.
More effective protection therefore tends to lead to
greater transient smearing. Figure 2 shows the low
frequency decay of a sealed box loudspeaker, with its
attendant low frequency roll off. Figure 3 shows the low
frequency response of an electrically protected reflex
cabinet of somewhat similar size. Clearly the response
shown in Figure 3 is flatter until a lower frequency, but
a flat frequency response is not the be-all and end-all of
loudspeaker performance. Note how the response
between 20Hz and 100Hz has been caused to ring on,
long after the higher frequencies have decayed.
Figure 4 shows the corresponding step-function
responses, and Figure 5 the acoustic source plots.
These plots clearly show the time response of the
reflex cabinets to be significantly inferior to the sealed
boxes. The low frequencies from the reflex enclosure
arrive later, and take longer to decay, which
compromises the ‘punch’ in the low frequency sound.
A sealed box cabinet will exhibit a 12dB per octave
roll-off below resonance, but a reflex enclosure will
exhibit a 24dB/octave roll-off, as the port output
becomes out of phase with the driver output. As the
system roll-offs are often further steepened by the
addition of electrical protection filters below the system
resonance, sixth, and even eighth order roll-offs (36dB
and 48dB per octave, respectively) are quite common.
With such protection, some small systems can produce
high output SPLs at relatively low frequencies, but the
time (transient) accuracy of the responses may be
very poor.
Inevitably, the different resonances of the different
systems will produce musical colourations of different
characters and such inconsistency of colouration does
little to help the confidence of the users in studios. If a
mix sounds different when played on each system,
then how do you know which loudspeaker is most
right or when the balance of the mix is correct?
The tendency is that well-designed sealed-box
cabinets sound more similar to each other than do
small reflex-loaded boxes. The resonances of the
sealed boxes tend to be better controlled, and are
usually much more highly damped than their reflexloaded counterparts. This leaves the magnitude of the
frequency response of sealed boxes as their
predominating audible characteristic, but it is usually
the time responses of reflex enclosures that give rise to
their different sonic characters.
There is considerable evidence to suggest that the
use of Auratones and Yamaha NS10Ms has been due
to their rapid response decays. A roll-off in the low
frequency response of a loudspeaker used for mixing
is in itself not a great problem, because any wrong
decisions can usually be corrected by equalisation at a
later date, such as during mastering.
An error in the time response, such as that added by
tuning port and filter resonances, can lead to
misjudgements, especially between percussive and
tonal low frequency instruments, which cannot be
adjusted once they have been mixed together.
A problem therefore exits in terms of how we can
achieve flat, uncoloured, wide-band monitoring at
relatively high sound pressure levels from 10-litre
boxes. At the moment, the answer is that we cannot
do it. Just as there is a trade-off between low frequency
extension, low frequency SPL and box size; there is
also a trade-off between low frequency SPL, bass
extension, and transient accuracy if bass reflex loading
and electrical protection are resorted to in an attempt to
defeat the box size limitations.
In fact, at low SPLs, good low frequency extension
can be achieved from small boxes, but the nonlinearity of the internal air-spring leads to high
distortion when the cone excursions, and hence high
degrees of internal pressure changes, become
significant. Also, in a small sealed box, there is the
problem of how to get rid of the heat from the voice
coil. Thermal overload and burnout is always a
problem at high SPLs due to the high power necessary
to overcome the problem of the poor system efficiency.
This leads to the thermal compression that limits the
output dynamics from accurately following the
dynamics of the input signal.
From the waterfall plots of Figures 2 and 3 it can be
seen that the decay is never instantaneous, and that
there is a slope to the time representation (although at
low frequencies some of this can be due to the time
response of the measurement filters). The question has
often been asked whether a flattening of the low
frequency response would inevitably lengthen the
time response, even with the sealed boxes. In truth,
the tendency is for the flattening of the amplitude
response to shorten the time response, by means of its
correction of the phase response errors that are
associated with the roll-off.
This means that a large or small sealed box,
equalised or not, would still exhibit a much faster time
response than a reflex enclosure. Figures 6 and 7
show the comparative effect. What is very significant
is that an enormous number of music recording
professionals, by opting for NS10s, Auratones and
others of similar response characteristics, have
indicated their preference for time response accuracy
over absolute frequency response accuracy.
sweet spot
a
b
Fig. 6. Waterfall plot showing the effect on the time response
of electrically flattening the response of an Auratone.
Although the response is flatter down to a lower frequency
than the reflex loaded enclosure shown in Figure 3, the time
response is still much quicker - the decay is much more rapid.
Such equalisation is not a practical solution because the
loudspeaker would overload badly even at low SPLs.
Fig. 5. Acoustic source plots corresponding to the stepfunction plots of Figure 4. The low frequency response delay
is shown in terms of from how many metres behind the
loudspeaker the low frequencies are apparently emanating.
As each metre corresponds to about 3ms, it can be
appreciated how the low frequencies from the sealed box, a),
arrive much more 'tightly' with the rest of the frequencies
than from the reflex loaded box.
Fig. 7. Acoustic source plot for a 700 litre, wide-range, flushmounted, studio monitor system. Note how the NS10M’s
response (Figure 5a) mimics that of this large monitor system.
Many mastering engineers concur with this choice,
citing low distortion and good transient accuracy as
being more important to them than absolute
frequency response flatness. (Strictly speaking, we
should be speaking about the pressure amplitude
response, or the modulus of the frequency response,
because the term ‘frequency response’ technically
also includes the phase response, but I will stick to the
popular term here.)
May/June 2003
The whole close-field concept of monitoring seems
to have been borne out of a recognition that the
monitoring of the direct sound has been more reliable
than monitoring the direct/room-sound combination
from the wider-range large monitors. Many studio
designers now aim for highly absorbent control rooms,
which can maintain the direct sound from the main,
flush-mounted monitors, all the way to the mixing
console and beyond. Contrary to a popular fallacy,
resolution
these rooms are not oppressive to be in, because
reflective surfaces are positioned in the rooms in places
where they cannot affect the monitoring, but they give
life to the speech of the occupants. This is probably the
only way to deliver flat, full-frequency range, fast
time-response monitoring, because current technology
cannot supply this from small boxes.
There are those who say that very fast time
responses are not necessary from small loudspeakers,
because their decay times are still shorter than many
of the rooms in which most of them will be used.
However, what they fail to realise is that the small
loudspeakers are usually being used in the close-field,
which is normally considered to be within the critical
distance where the direct sound and room sound are
equal in level. It therefore follows that if one is
listening in the close-field, then the responses of the
loudspeakers will predominate in the total response.
Indeed, this is the principal reason for the use of closefield monitoring.
A great deal of illogical thinking, lack of
awareness of the facts, and a legacy of continuing to
look at traditionally measurable aspects of
loudspeaker design, has led to the manufacture of
loudspeakers that seek to respond to the traditional
response norms. This is despite the fact that many
recording professionals have opted for loudspeakers
whose responses did not comply with the perceived
technical requirements. They have chosen to use
loudspeakers that they found reliable to work with,
either despite or in total ignorance of their published
response plots.
It must be added that much ignorance of the facts
also exists within many loudspeaker manufacturing
companies, where the people responsible for defining
what is produced are not loudspeaker engineers or
sound engineers. In many cases they are simply
business people. So, with no clear signals from the
recording industry about what it needs, the
businesses produce what they think they can sell
most of. If this means a battle to improve some largely
irrelevant specifications, then this is the path they
pursue. This has led to the state of affairs in which
chaos rules in the low frequency responses of current
small ‘monitor’ loudspeakers. Improving the time
responses of many small monitor loudspeakers is a
long overdue priority. ■
51
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement