are a class of electronic filters
designed specifically for use in audio applications, especially hi-fi
. Commonly used loudspeaker
drivers are incapable of covering the entire audio spectrum
with acceptable loudness and lack of distortion
by themselves. Thus, crossovers serve the purpose of splitting the audio signal into separate frequency
bands which can be handled by individual loudspeaker
drivers optimized for those bands. A combination of multiple drivers each catering to a different frequency band is the design pattern for most hi-fi
speaker systems. An audio crossover may also be constructed mechanically and is commonly found in full-range
speakers, portions of whose cones/dust caps/whizzer cones are decoupled at progressively higher frequencies.
Another use of crossovers is multiband processing, in which the audio signal is split into bands, which are adjusted (equalized, compressed, echoed, etc) separately. After the adjustments, the individual bands are mixed together again. Some examples are: multiband dynamics (compression, limiting, de-essing), multiband distortion, bass enhancement, high frequency exciters, noise reduction (for example: Dolby A noise reduction).
The definition of an ideal audio crossover changes relative to the task at hand. If the separate bands are to be mixed back together again (as in multiband processing), then the ideal audio crossover would split the incoming audio signal into separate bands that do not overlap or interact and which result in an output signal unchanged in both frequency, relative levels, and phase response. This behavior cannot be achieved in practice, but can be approximated. Just which is the best approximation is a matter of lively debate. On the other hand, if the audio crossover is performing the task of loudspeaker driver band separation, then there is no longer any requirement for mathematically ideal characteristics within the crossover itself, as the frequency and phase response of the loudspeaker drivers within their mountings will affect the results. The satisfactory output of the complete system comprised of the audio crossover and the loudspeaker drivers in their enclosure(s) must be regarded as the design goal. Such a goal is often achieved using non-ideal, asymmetric crossover filter characteristics.
Many different crossover types are used in audio, but they generally fall under one of the classifications provided below.
Classification based on the number of filter sections
In loudspeaker specifications, one often sees a speaker classified as an "N-way" speaker. N is a positive whole number greater than 1, and it indicates the number of filter sections. A 2-way crossover consists of a low-pass
and a high-pass
filter. A 3-way crossover is constructed as a combination of low-pass
filters (LPF, BPF and HPF respectively). The BPF section is in turn a combination of HPF and LPF sections. 4 (or more) way crossovers are not very common in speaker design, primarily due to the complexity involved, which is not generally justified by better acoustic performance.
An extra HPF section may be present in an "N-way" loudspeaker crossover to protect the lowest-frequency driver from frequencies lower than it can safely handle. Such a crossover would then have a bandpass filter for the lowest-frequency driver. Similarly, the highest-frequency driver may have a protective LPF section to prevent high frequency damage, though this is far less common.
Recently, a number of manufacturers have developed what is often called "N.5-way" crossover techniques for active stereo loudspeaker crossovers. This usually indicates the addition of a single subwoofer with a crossover designed such that it combines stereo input signals into monophonic subwoofer signals and may augment the bass response.
Remark: Filter sections mentioned here is not to be confused with the individual 2-pole filter sections that a higher order filter consists of.
Classification based on components
Crossovers can also be classified based on the design approach; by the type of components used.
A passive crossover is made entirely of passive components, arranged most commonly in a Cauer
topology to achieve a Butterworth filter
. Passive filters use both non reactive resistors
, and reactive components such as capacitors
. Very high performance passive crossovers are likely to be more expensive than active crossovers since individual components capable of good performance at the high currents and voltages at which speaker systems are driven are hard to make and expensive. Polypropylene
and Metalized polyester
foil, and paper-electrolytic
capacitors are common. Inductors may have air cores, powdered metal cores, ferrite
cores, or laminated silicon
steel cores, and most are wound with enamelled copper
wire. Some passive networks include devices such as fuses
, PTC devices, bulbs or circuit breakers
, to protect the loudspeaker drivers from accidental overpowering. Modern passive crossovers are increasingly also including equalization networks (eg, Zobel networks
) that compensate for the changes in impedance with frequency inherent in virtually all speakers. The issue is complex, as part of the change in impedance is due to acoustic loading changes across a driver's passband.
On the negative side,
- Passive networks may be bulky and cause power loss.
- Networks are not only frequency specific, but also impedance specific. This prevents interchangeability with speaker systems of different impedances.
- Ideal crossover filters, including impedance compensation and equalization networks, can be very difficult to design, as the components interact in complex ways. This is an area in which careful computer simulation is very helpful.
An active crossover contains active components (ie, those with gain) in its filters. In recent years, the most commonly used active device is an op-amp
; active crossovers are operated at levels suited to power amplifier inputs in contrast to passive crossovers which operate after the power amplifier's output, at high current
and in some cases high voltage
. On the other hand, all circuits with gain
introduce noise, and in this case the noise
will be amplified by the power amplifiers.
Active crossovers always require the use of power amplifiers for each output band. Thus a 2-way active crossover needs two amplifiers — one each for the woofer and tweeter. This means that an active crossover based system will often cost more than a passive crossover based system, although none of the amplifiers needs to provide output as high as for an equivalent sound level full-frequency, power amplifier, which reduces cost. The cost and complication disadvantages of active crossovers are offset by the following gains:
- a frequency response independent of the dynamic changes in a driver's electrical characteristics.
- typically, the possibility of an easy way to vary or fine tune each frequency band to the specific drivers used. Examples would be crossover slope, filter type (eg, Bessel, Butterworth, etc), relative levels, ...
- isolation of each driver from signals handled by drivers, thus reducing intermodulation distortion and overdriving
- The power amplifiers are directly connected to the speaker drivers, thereby maximizing amplifier damping control of the speaker voice coil, reducing consequences of dynamic changes in driver electrical characteristics, all of which are likely to improve the transient response of the system
- reduction in power amplifier output requirement. With no energy being lost in passive components, amplifier requirements are reduced considerably (up to 1/2 in some cases), reducing costs, and potentially increasing quality.
Active crossovers can be implemented digitally using a DSP
chip or a microprocessor
. They either use digital
approximations to traditional analog
circuits, known as IIR
etc.), or they use Finite impulse response (FIR)
filters. IIR filters have many similarities with analog filters and are relatively undemanding of CPU resources; FIR filters on the other hand usually have a higher order and therefore require more resources for similar characteristics. They can be designed and built so that they have a linear phase
response, which is thought desirable by many involved in sound reproduction. There are drawbacks though - in order to achieve linear phase response, a longer delay time is incurred than would be necessary with an IIR filter. IIR filters, which are by nature recursive have the drawback that if not carefully designed they may enter limit cycles resulting in non-linear distortion.
This crossover type is mechanical and uses the properties of the materials in a driver diaphragm to achieve the necessary filtering. Such crossovers are commonly found in full-range
speakers which are designed to cover as much of the audio band as possible. One such is constructed by coupling the diaphragm of the speaker to the voice coil through a compliant section and directly attaching a small light-weight cone called whizzer
to the voice coil. The compliant section is intended to ensure that the primary full size diaphragm responds only to lower frequencies. The whizzer is directly coupled to the voice coil and responds to all frequencies, but due to its small size only gives a useful level of output at higher frequencies. This combination results in the main diaphragm having an upper cut-off frequency while the size of the whizzer sets the lower limit to the whizzer's response, thereby implementing a crossover action. The choice/weight of materials used for the diaphragm, whizzer and the speaker's suspension determine the crossover frequency and the effectiveness of the crossover. This sort of crossover is much more complex to design, especially if the highest degree of performance is desired. Extensive trial and error is required. Over several years, the compliance of the joint can change, negatively affecting the frequency response of the speaker.
An alternative is to use the dust cap as a high frequency radiating device, also crossed over by mechanical compliance from the primary diaphragm. High frequency dispersion is somewhat different for this approach than for whizzer cones. Another possibility is to build the primary cone with such profile, and of such materials, that the neck area remains rigid, radiating all frequencies, while the outer areas of the cone are selectively decoupled, radiating only at lower frequencies.
Speakers which use these mechanical crossovers have some advantages in sound quality despite the difficulties of designing and manufacturing them, and despite the inevitable output limitations. Full-range drivers have a single acoustic center, and can have relatively modest phase change across the audio spectrum. For best performance at low frequencies, these drivers require careful enclosure design. Their small size (typically 165 to 200mm) requires considerable cone excursion to reproduce bass, but have short voice coils, demanded for reasonable high frequency performance, incapable of the large cone motions required for bass. But within these constraints, cost and complications are reduced, as no crossovers are required.
Those who do not prefer the sound of full-range drivers (e.g., a lack of powerful bass and usually strong highs) sometimes argue that a single diaphragm that must produce both low and high frequencies does neither justice. The fact that nearly all high fidelity speakers are 2 or 3 way lends weight to this view. See full-range speaker for construction details.
Classification based on filter order or slope
Just as filters have different orders, so do crossovers, depending on the filter slope they implement. The final acoustic slope may be completely determined by the electrical filter or may be achieved by combining the electrical filter's slope with the natural characteristics of the driver. In the former case, the only requirement is that each driver has a flat response at least to the point where its signal is approximately -10dB down from the passband. In the latter case, the final acoustic slope is usually steeper than that of the electrical filters used. A third- or fourth-order acoustic crossover often has just a 2nd
order electrical filter. This requires that speaker drivers be well behaved a considerable way from the nominal crossover frequency, and further that the high frequency driver be able to survive a considerable input in a frequency range below its crossover point. This is difficult in actual practice. In the discussion below, the characteristics of the electrical filter order is discussed, followed by a discussion of crossovers having that acoustic slope and their advantages or disadvantages.
Most audio crossovers use first to fourth order electrical filters. Higher orders are not generally implemented in passive crossovers for loudspeakers, but are sometimes found in electronic equipment under circumstances for which their considerable cost and complexity can be justified.
First order crossovers
order filters have a 20 dB/decade
(or 6 dB/octave
) slope. All 1st
order filters have a Butterworth filter
order filters are considered by many audiophiles
to be ideal for crossovers. This is because this filter type is 'transient perfect', meaning it passes both amplitude and phase unchanged across the range of interest. It also uses the fewest parts and has the lowest insertion loss (if passive). A 1st
-order crossover allows more signals of unwanted frequencies to get through in the LPF and HPF sections than do higher order configurations. While woofers can easily take this (aside from generating distortion at frequencies above those they can properly handle), smaller high frequency drivers (especially tweeters) are more likely to be damaged since they are not capable of handling large power inputs at frequencies below their crossovers.
In practice, speaker systems with true first order acoustic slopes are difficult to design because they require large overlapping driver bandwidth, and the shallow slopes mean that non-coincident drivers interfere over a wide frequency range and cause large response shifts off-axis.
Second order crossovers
order filters have a 40 dB/decade (or 12 dB/octave) slope. 2nd
order filters can have a Bessel
characteristic depending on design choices and the components used. This order is commonly used in passive crossovers as it offers a reasonable balance between complexity, response, and higher frequency driver protection. When designed with time aligned physical placement, these crossovers have a symmetrical polar
response, as do all even order crossovers.
It is commonly thought that there will always be a phase difference of 180° between the outputs of a (second order) low-pass filter and a high-pass filter having the same crossover frequency. And so, in a 2-way system, the high-pass section's output is usually connected to the high frequency driver 'inverted', to correct for this phase problem. For passive systems, the tweeter is wired with opposite polarity to the woofer; for active crossovers the high-pass filter's output is inverted. In 3-way systems the mid-range driver or filter is inverted. However, this is generally only true when the speakers have a wide response overlap and the acoustic centers are physically aligned.
Third order crossovers
order filters have a 60 dB/decade (or 18 dB/octave) slope. These crossovers usually have Butterworth filter
characteristics; phase response
is very good, the level sum being flat and in phase quadrature
, similar to a first order crossover. The polar response is asymmetric
. In the original D'Appolito
MTM arrangement, a symmetrical arrangement of drivers is used to create a symmetrical off-axis response when using 3rd-order crossovers.
Third-order acoustic crossovers are often built from first- or second-order filter circuits.
Fourth order crossovers
Fourth-order filters have an 80 dB/decade (or 24 dB/octave) slope. These filters are complex to design in passive form. A 4th
order crossover with −6 dB crossover point and flat summing is also known as a Linkwitz-Riley
crossover (named after its inventors). It can be constructed in active form by cascading two 2nd
order Butterworth filter
sections. The output signals of this crossover order are in phase, thus avoiding phase inversion in driver connections unless the driver acoustic centers are not aligned.
Higher order crossovers
Passive crossovers giving acoustic slopes higher than 4th
order are not common because of cost and complexity. They are sometimes used in active crossover modules.
Mixed order crossovers
Crossovers can also be constructed with mixed order filters. For example, a second order lowpass combined with a third order highpass. These are generally passive and are used for several reasons, often when the component values are found by computer program optimization. A higher order tweeter crossover can sometimes help compensate for the time offset between the woofer and tweeter, caused by non aligned acoustic centers.
Classification based on circuit topology
These are by far the most common. Electrically the filters are in parallel and thus the various filter sections do not interact. This makes them easier to design because the sections can be considered separately, and because component tolerance variations will be isolated.
Crossovers using this topology are almost always passive because it is easiest to construct in passive form. In this topology, the individual filters are connected in series, with a driver or driver combination connected in parallel to each filter. As can be seen in the image, a low-pass filter in shunt
with the tweeter results in a high-pass response for the tweeter, since the lower frequencies are shunted by the LPF via the woofer. Similarly, the HPF in parallel with the woofer shunts away the higher frequencies via the tweeter - a low-pass response for the woofer. One advantage (or disadvantage, depending on how one looks at it) of this crossover, is that the crossover sections interact with each other. Changes in any one component affects both highpass and lowpass sections. To some extent, this makes the crossover somewhat self-balancing - the crossover frequency changes, but the system still sums substantially flat. This characteristic makes them appealing to designers not using sophisticated measuring equipment or simulation software. These crossovers are also more sensitive to component tolerance variations.
These are active crossovers in which one of the crossover responses is derived from the other through the use of a differential amplifier. For example, the difference between the input signal and the output of the high pass section is a low pass response. Thus, when a differential amplifier is used to extract this difference, its output constitutes the low pass filter section. The main advantage of derived filters are that it produces no phase difference between the high pass and low pass sections at any frequency. The disadvantages are either
- (a) that the high pass and low pass sections often have different levels of attenuation in their stop bands, i.e. their slopes are asymmetrical, or
- (b) that the response of one or both sections peaks near the crossover frequency,
In case (a), above, the usual situation is that the derived low pass response attenuates at a much slower rate than the fixed response. This requires the speaker to which it is directed to continue to respond to signals deep into the stopband where it's physical characteristics may not be ideal. In the case of (b), above, both speakers are required to operate at higher volume levels as the signal nears the crossover points. This uses more amplifier power and may drive the speaker cones into non-linearity.