(The design of the speaker network circuit from an electric circuit theoretic
point of view) |
|
|
|
|
The speaker network circuit is the constitution that is very simple for an electric circuit, but the influence to |
give the quality of the speaker system is big. And there are some points that I feel not a little doubt in the |
present conditions of the speaker system, which include current design thought and which is made by famous |
manufacturer also. It may be thought too particular about trifles,
and it seems to be one of few fields left |
behind in the modern advanced electric industry by an electric theory. It is
the one of these causes to feel doubt in |
calculating the impedance of the speaker as pure resistance, and to downplay
a theory. As for another one, the sound |
is often in some situation so that a theory is ignored because prominent
expertise of the listening is given priority and |
is considered to be the field of the art. However, when you lost the
judgement of a theoretical route, you should not |
have given priority to a theory in all situation, there can be that solving
by auditory capability is a shortcut. However,
|
I do not need to say daringly, and the electric circuit theory is a tool that
it seems to be like the car navigation system
|
which guides to the neighborhood of the destination for the field of the
audio system, and it is not the thing which
|
should never make light of it. I compiled the main point
about the design of
the speaker network which I felt from |
the
electric theoretical standpoint. The contents of this
sentence extracted the
purpose of the attached document. |
|
|
1. Alternating current resistance, not direct current
resistance is to be used
for calculation of speaker network. |
|
|
|
The impedance of the speaker is mentioned in a catalogue so
that the maker says
4Ά or 8Ά as nominal impedance. |
If the impedance of the speaker is not pure
resistance in calculating formula,
a calculation of impedance becomes |
complicated, and a calculation becomes difficult, and the
network circuit in a speaker system is
not desirable because |
the unevenness of the frequency characteristic grows big.
Therefore, the reproduction range must be chosen |
without contradiction as a frequency range in which
the impedance is similar with pure resistance.
But this is a |
case of the multi-way speaker and it is used without minding
even if impedance is disturbed in the case of the full-range |
speaker if an output response is good. Pure resistance to say here
is not direct current resistance by alternating |
current resistance. It is a too unconcerned story that the
alternating current
resistance substitutes by the direct |
current resistance because alternating current resistance are
10-30% big values than direct current resistance. |
|
Emission resistance (while it is installed in a speaker box), eddy
current loss are included in alternating current |
|
resistance other than direct current resistance. Because it is a
formula that (impedance = alternating current |
|
resistance + reactance composed by LC), in the case that reactance
composed by LC equal
to zero, the alternating |
current resistance and the impedance become equal. Therefore, you
may regard minimum impedance (nominal |
|
impedance) as alternating current resistance. Such a virtual alternating
current resistance is called effective resistance. |
|
|
2. Total output voltage is important at the low frequency,
and the output power
is important at the high frequency. |
|
|
A full-range speaker has no responsive difference between
the voltage and the power, but in the case of a multiway |
speaker, a phase change occurs to the voltage of each
speaker, and a difference between response of electric |
power and response of the total voltage may occur.
Because a wavelength is long in the case of the low tone,
a phase |
difference caused by the position in the sound field
becomes small. Therefore, it is important that a total
output voltage- |
frequency curve is flat through low frequency band
so that it is thought that the total voltage is in
proportion to sound |
pressure. Because a wavelength is short in the case of
the high tone, a phase difference caused by the position
in the |
sound field becomes big. In the case of total output
voltage-frequency curve being flat, it is only an angle
on the axis of |
the speakers that sound pressure-frequency curve becomes
flat, and if the angle is different, a phase difference
will |
change, and sound pressure will change. Therefore, the
output level of the sound field will depend on output
power. |
You need not to do the useless effort that you move the
installation of the tweeter 1cm to match the phase. |
It is surely desirable for both the total voltage curve
and the electric power curve against the frequency to be
flat, for the |
output response of the network. |
|
|
|
3. It is important that the input impedance of the network
circuit must be equal to speaker resistance. |
|
|
|
The output voltage of the audio amplifier is constant
voltage regardless of frequency when input is given
constant voltage. |
|
Generally it is called constant voltage source. This is
a digression, and in case of constant current source,
internal |
resistance is nearly infinite, because dumping of speaker
is very bad, sound quality of constant current source is
not good. |
|
The constant output power is produced when constant
resistance is connected to the constant voltage source. |
When the input impedance of the network changes by
frequency, both the output power and the input electric
current |
change, and the output power curve for the frequency is
not flat. |
|
|
|
4. Both the phase distortion and cost of parts increase
when the crossover of the network is set to low frequency. |
|
|
The fixed number of the network parts becomes big when
crossover frequency is taken low. When they make the part |
fixed number big, a cost rises, a phase change becomes big,
and phase distortion increase. Therefore, it is not good to
|
|
lower the crossover frequency of the network more than
required. |
|
|
|
5. The coil without core is good for speaker networks. |
|
|
|
The coil with iron core make reducing size and lowering the
direct current resistance, but is not good for sound
quality |
because distortion increase by the non-straightness of the
B-H curve. The coil with iron core forming closed magnetic |
circuit (a toroidal core, an E-I core, an U-U core and
e.t.c.) is especially unsuitable for a network part because |
|
distortion will be big. On the other hand, coil with
large-scale ferrite core forming open magnetic circuit
will be available |
because structure is near to coil without core.
I recommend you to make the coil without core if you have |
means to measure inductance. It is not necessary to wind
coil in rows, and there is no changing in performance
between |
a coil being wound in rows and a coil being wound at
random even if it was formed around one hand. However,
I think |
that you had better to practice tape winding and varnish
processing or adhesive fixing because stability of the
shape is |
necessary. |
|
|
|
6. A film condenser is suitable for a condenser for speaker
networks. |
|
|
|
Except a super high-cost speaker system, aluminum
electrolysis condensers and coils with silicon steel
sheets forming |
open magnetic circuit are usually used. However,
a frequency characteristic of the aluminum electrolysis
condensers is |
bad, and the function as the condenser is not so sure at
more than 1kHz. When they become the low temperature, they |
have not good reliability at all by influence of a
capacitance deviation and an inferior frequency
characteristic. The use |
of the aluminum electrolysis condenser is limited to only
low frequency, and at more than an middle frequency the use |
of the film condenser is recommended. The film condenser
can expect stable characteristics till the frequency more |
than 100kHz. |
|
|
|
7. The 6dB/oct and 18dB/oct network circuits have good
performance. |
|
|
|
If the characteristic of the speaker is good, 6dB/oct
network circuit is very suitable and superior. There is no |
phase distortion in 6dB/oct circuit, and it
communicate the wave form of the transient phenomenon
faithfully. |
Even if a corrective circuit may be added, amount of the
phase distortion can be limitted to small, but the |
separation of the midrange is not clear, and you will hear
an middle tone from the speaker of the low tone and the |
speaker of the high tone in the case of 3 - 4way. When the
change of the characteristic of the speaker is big, |
and
the output in stop band is big, though phase distortion increases,
a 18dB/oct network circuit is suitable. |
|
Instead of that the separation of the midrange becomes
clear, phase distortion will increase, and it does not |
|
communicate the wave form of the transient phenomenon
faithfully. However, the network circuit will be chosen |
by interest because it is low level of phase distortion
that is collateed with a CCIR standard of transmission of |
high quality, and is estimated to be no problem. |
|
|
|
8. When the phase difference is 90 degrees at the crossover frequency,
the characteristic becomes the best. |
|
|
|
In 18dB/oct and 24dB/oct network circuit there are about
270 degrees phase difference of the speaker voltage, |
|
and can obtain a almost level curve of frequency response
without trouble by the same aspect connection. |
|
However, the phase distortion can be considerably decreased
by making one of speakers a reverse aspect |
|
connection and adjusting the phase difference to about 90
degrees like a level curve of frequency response. |
|
This fact can be confirmed in the Excel design sheets for
speaker network. In "SpeakerCorrection1" of excel |
|
sheet all the cell data of "Lower cross over" and "Higher
cross over" are cleared, and data of the "Polarity |
|
after correction"is input. In case of 2way woofer only
might be selected for the reverse aspect and in case of |
|
3way sqawker only might be selected. When you select
a sheet of "18dB2wayType" or "24dB2wayType" or |
|
"18dB3wayType", and you fill the data of "1" to the
cell of the "Sp correction1-4 add", you should be able to |
|
confirm the decrease of phase distortion. |
|
|
|
9. Phase distortion increases when you use a speaker
corrective circuit. |
|
|
|
Phase distortion increases when you use a 6dB/oct
corrective circuit or a 12dB/oct corrective circuit. |
|
Because circuit development and the theory analysis about
the corrective circuit are the poor at present, they |
|
have not watched the article in which calculated influence
after the correction either. However, it is a fact that |
|
phase changes after correction, and connection polarity of
speaker can turn over depending on a case. |
|
|
|
10. For the calculation in the resonance type network
circuit must be used the formula of the band pass filter. |
|
|
|
You can not set attenuation at the crossover frequency by
the general calculating formula precisely, but you can |
set crossover in -3dB precisely by using the formula of the
band pass filter. |
|
|
|
11. Even if several stacks Zin constant type networks
repeat it, Zin becomes constant. |
|
|
|
Even if the network circuit of the Zin constant type that I
devised repeats the number of the arbitrary stacks, you |
can design the network circuit of the Zin constant in 6way
or even 11way much less 4way freely, because Zin |
|
becomes constant. It is the different theme of course whether such
large number of speakers are necessary. |
|
The Zin constant circuit devised by the present has two
kinds, 6db/oct and 12dB/oct, and there are series type |
and parallel type in each. Because a characteristic is
good, as for the Zin constant type of 6dB/oct in
particular, |
without a limit of number of speakers freely, you can
design the network circuit of the good characteristic that has |
flat curves of total voltage and an electric power
vs. the frequency, and has zero phase distortion. |
|
In addition, you can use modifying the network circuit of
the Zin constant type as a speaker corrective circuit for |
the good point of the impedance characteristic. |
|