Tutorial: Why Passive Crossovers Always Suck

[Flint]

This started as a reply to a question from @TitaniumTroy in another thread about speaker size:

http://www.theaudioannex.com/forum/...the-job-at-hand-size-versus-bass.12784/page-2

The primary function of a crossover is to filter a range of audio frequencies starting at a specific frequency so the only signal which passes to the driver are in the acceptable pass band, or the range where the driver is expected to operate. Crossovers can also act as equalizers and even dynamic filters which apply power filtering at higher voltages to protect the speaker from overload.

So, assuming two speakers are being used together, one for higher frequencies and the other for lower frequencies with a crossover on each to facilitate the filtering of the signal going to each driver, then the issues are as follows:

CHARACTERISTIC DRIVER IMPEDANCE
Drivers don't have linear 8 Ohm or 4 Ohm resistance across the frequency range they operate in. With few exceptions, nearly all drivers have non-linear impedance curves, some ranging from 3.2 Ohms to over 50 Ohms in the operating range. So, tuning the crossover to achieve a perfect slope with a -3dB point at the desired frequency and a roll-off slope beyond that frequency of exactly 12dB (2nd order), 18dB (3rd order), or 24dB (4th order) per octave is darn near impossible. So, some sort of compromise much be made, either more passive components can be added to the crossover to give a more linear impedance load on the crossover filter portion of the circuit, or tuning of the filter components can be experimented with to reduce the negative affects of the varying impedance to a level which gets results which are acceptable.

DYNAMICALLY CHANGING DRIVER IMPEDANCE
Just about all loudspeaker drivers will exhibit varying impedance based on the input voltage. Variations are caused by heat, the position of the voice coil or diaphragm in the magnetic gap, the resistance of the suspension based on how flexed (stretched) it is, pressure in the enclosure, and several other electrical, magnetic, and mechanical causes. Thus, a passive crossover behaves differently with 1 watt of input than it does with 10 watts of input. It also behaves differently after an hour of high voltage utilization than it does when the signal is first applied after a long period of no signal. It can behave differently over time as the driver relaxes with age. It can perform differently if the driver is set horizontally versus vertically. It even acts differently in a very real dynamic way in that high dynamic peaks will have a different signal than the adjacent average voltage levels.

IMPEDANCE VARIATION AND TUNING
The two issues above should appear obvious before I get into this little detail, but check this out - as the impedance shifts away from the ideal for which the crossover was tuned, inductors go out of ideal in one direction while capacitors shift out of ideal in the opposite direction. In simplistic terms, if the crossover is tuned perfectly for a driver with a 6 Ohm load and in use the driver's impedance shift upwards to a 7 Ohm load, the inductor(s) in the crossover are now too small for the target ideal filter and the capacitor(s) in the crossover are now too large for the target ideal filter. So it isn't like a slightly higher impedance simply raises the crossover frequency slightly higher. Instead, a slightly higher impedance screws everything up where the shape of the filter curve shift away from the target in every direction. And, I am referring to amplitude frequency response, but phase response is equally critical (if not more so) and a shift in driver impedance will completely toss out any attempt to achieve an ideal phase response. It is a nightmare scenario.

PASSIVE COMPONENT FAILINGS
Passive crossovers are generally made from capacitors which store energy and release it over time (phase shift and filtering bass), inductors which are magnetic coils which resonate and absorb high frequencies, and resistors which resist current passing through them.
Capacitors have inherent issues with dynamic response, behaving linearly across their operating range and at all appropriate voltages, and can actually resonate and "ring" in the operating range. As such, there are high end capacitors out there which can cost as much as $150 for a mid-frequency application which do, indeed, perform audibly better than lesser capacitors.
Inductors are basically resonating magnetic coils which can saturate at high voltages and create magnetic fields extending way beyond their physical size. These magnetic fields can interact with other inductors' magnetic fields and cause crosstalk between separate sections of a crossover. They can also be large, heavy, and inconsistent in how they perform at different voltages and loads. Like capacitors, you can get high end inductors that cost more than $300 for a size appropriate for a midrange crossover.
Resistors often struggle not to introduce at least a little inductance, but they are often not a real problem unless they get hot enough to
experience a shift in their resistive value due to high voltages passing through them.

PASSIVE CROSSOVERS AND DECOUPLING
We all know the importance of an amplifier's damping factor. Among other things, it reflects an amp's output impedance (lower in better), current capacity, and ability to control the motion of a speaker. The latter performance characteristic is very important with traditional voice coil driven drivers. Since the common cone/dome and coil is basically a suspended diaphragm on a spring-loaded suspension, once it gets excited tends to resonate until it the energy is lost. While resonating, the voice coil generates voltage, known as "Back EMF." That electrical energy will sustain the resonance beyond what the suspension alone would cause if it isn't removed. The easiest way to remove that energy is to shunt it to ground, which an amp with a high damping factor does. In fact, if directly coupled, a high damping factor amp will very precisely control the position of the voice coil in the magnetic gap and absorb resonance created spurious voltages from the speaker to actually prevent some ringing and mechanical limitations and force the cone to move exactly to the voltage. However, a passive crossover decouples the speaker from the amp and the "Back EMF" will then feed the resonate circuit of the crossover and forceful control of the speaker by the amp is vastly reduced. This isolation, in my opinion, is one of the greatest drawbacks of passive crossovers, and it only takes one passive component between the amp and the speaker to start the isolation process. I've witnessed firsthand the way a simple crossover completely changes the waterfall plots of a speaker's performance versus directly coupled speakers. The greater the voltage, the more extreme these resonances will be, especially with woofers, and as such the performance at very low levels will be completely different at high SPLs. Generally, high levels will cause much greater impact of resonances. This is a huge issue.
​

 

[Flint]

SPL VS. TARGET TIMBER
Passive crossovers are pretty much fixed, with all the limitations I shared above, most of which are affected by voltage dynamics. As such, a designer has to decide what the target listening level for a speaker model is going to be in order to design a crossover which creates the characteristic timber for the brand. At higher levels the bass and the treble are likely to sound louder. Likewise, issues with saturation, passive component interaction, increased resonances, limitations of less than ideal passive components, and more will make the speakers sound stressed, compressed, fuzzier, and much less pleasing than at lower levels. Meanwhile, at listening levels below the target will result in the speakers sounding anemic, dull, and lifeless. This is one inherent flaw of passive crossovers. So, when you listening to an old-school pair of top of the range PA cabinets at home listening levels, they sound dull and lifeless, yet when those same speakers are put in a huge room or outdoors and you move back 25 yards, and get the exact same SPL at your ears (meaning a massively higher input power for the speakers) they will sound amazing, punchy, solid, and clear. The listening level at your ears may be the same, but the speaker system as a whole was designed to be working hard and thus sounds good when pushed. (side note, this is the main reason I laugh at people who buy huge PA speakers or large Cinema speakers to use in their home theaters. A huge set of Meyer Sound cinema speakers might be state of the art when driven hard by a 1,000 watt commercial amp, but in your living room with about 1/100th of the power being used to get the exact same SPL at the listening position, they will sound dull, harsh, and pretty crappy).

NUMBER OF NEGATIVE COMPONENTS
I mentioned, above, the drawbacks of the different types of passive components commonly used in crossovers. But to get a really good amplitude frequency response characteristic there are often more and more components added to a crossover well beyond just the crossover filters. For a 12dB (2nd order) crossover filter, all you need is one inductor and one capacitor. However, you may decide you need to reduce the impedance rise as the frequency increases, caused by the inductance of the voice coil, so a Zobel network might be inserted into the circuit, which adds another capacitor and a resistor. Then you may decide a resonance compensation network is needed, so another inductor, capacitor, and resistor will be added. After all that you might measure a clearly audible peak in the output over an octave and want to mellow that a bit, so another inductor, capacitor, and resistor are necessary in the crossover. That now means there are 4 capacitors, 3 inductors, and 3 resistors. But wait, what if the driver is also 3.6dB SPL more efficient than the other driver in the system - you then have to add at least one more resistor, if not two (L-Pad) or three (T-Pad). Basically, even the most basic crossovers often end up with way more passive components than you'd think - and each one of those components has its own detrimental characteristics, the worst of which I have listed above.


Basically, passive crossovers are an orgy of compromises. Removing the passive crossovers eliminates ALL of those negative characteristics and improves how the amplifier controls the driver. ​

 

[Flint]

The original question was how much of a difference does it make? I cannot put a percentage on it. But I can say that beyond merely having a pleasing timber and balance, it makes all the difference. A system with passive crossovers can have a generally flat response and fill a room with balanced sound. But trying to listen to the details in the music, attempting to create a convincing illusion of real instruments being played, getting identical detail and clarity at all SPLs, and so on cannot be achieved on any level to the degree that an active crossover system can.

PLUS!!!! With active crossover you can apply more tuning in the form of amplitude EQ, phase EQ, and delays which can accomplish a more ideal amplitude response, phase response, impulse response, decay response, and more. You can also do things like use a S.E.T. vacuum tube amp on your tweeters because you love that form of added even harmonic distortion in the treble, then put a 1,000W behemoth ultra current amp on bass woofers to get the power, impact, and control that those things can deliver, and then use an extremely low distortion, low noise, sonically invisible amp on the midrange driver to make the critical midrange as perfect and transparent as possible. You cannot really accomplish this with most passive speakers, and those which do allow for bi-wiring or tri-wiring only marginally improve the performance when mis-matching amps as I described.

So, I will never recommend passive crossovers for high-end, low distortion, high accuracy speaker system intended for the best critical listening experience.

For casual listening background speakers, the simplicity of passive crossover might make sense where a fullrange single driver speaker isn't appropriate. But if a pair of active studio monitors can be used for the same application, that'd be better. ​

 

[Flint]

To prove some level of comparison...

With an active crossover, whether analog of digital, the target ideal crossover slope characteristic can be perfect. The -3dB point will be exactly what you intended and not shift in any way due to driver variation, temp, signal level, or any other factor. Same for the shape of the slope, a Linkwitz-Riley 24dB (4th order) slope will retain the amplitude and phase response exactly as the theoretical design demands to get the benefits from the circuit. I mentioned the ability to add phase control to the circuit with no or very little detrimental affects. You can apply a delay to the signal of one driver to perfectly align the acoustical arrival time to the listener and thus create a perfect impulse response and natural phase response. You can add EQ to one driver without affecting another, or add EQ with extreme precision. With digital crossovers you can sometimes even add a phase EQ to drivers or the entire system to get a smooth or even flat Phase Response.

All of this stuff adds up to accomplishing the basic goals of a crossover without the extreme issues that passive crossovers inherently contribute to the sound. Sure, there is still a need to balance it all out and over processing is definitely something to fear. But everything a passive crossover can do and is needed to do an active crossover does a thousand times better.

 

[TitaniumTroy]

Wow, quite a read there. But I think I got the gist, passive crossovers are pain in the ass. Free up your speakers and amps and go active.

 

[Flint]

This issue of how badly all passive crossover suck forced me to reevaluate the priorities in a top end system. With the negative effects of passive crossovers holding back nearly all traditional multiway speakers, the advantages of a really amazing amplifier become more beneficial. A huge, low noise, ultra high current amp can improve a bit on the variations in the coupling between the amp and the speakers. It can improve the control of the driver, force the crossover to behave more linearly, and so on. These benefits are not massive, but they can be audible - one of the real small ticket items in the diminishing returns category. So, if you think you love your speakers and want to get more from them because you know they should sound better, yes, that $15,000 power amp will likely make a difference which is slightly audible and reduces some of the bad characteristics of the speaker system, mostly caused by the passive crossover.

However, if you can get the same voicing, timber, and power response from the same speakers without passive crossovers and use multiple amps, the performance will be so vastly superior that the clearly audible benefit of going from a quality $750 stereo amp to a $15,000 amp becomes much less important, if even desirable. In fact, I've argued that good digital amps can handle most of the drivers in a multiway system with an active crossover and most people wouldn't know the difference.

You also don't need nearly as large an amp when the individual speaker drivers are directly coupled to their own amplifier. I used to claim that you can cut the amp power output in half and still experience the same SPLs with lower distortion in an active system. But in recent years I am thinking you can cut it to a quarter and never feel the need for a larger amp. So, instead of buying a 250 watt stereo amp for $1,000, you can buy two 65 watt stereo amps for $350 each, saving $300 on amps which pays for the crossover and extra cables. In my current rig I am running just that, small Parasound Z-Amps for the tweeters and midrange speakers, and a larger 250 W mono amp for the woofers. The benefits of a smaller amp are a much lower noise floor (so you cannot hear in the room whether the amps are on or off) and lower THD in the primary operating range (important in solid state amps). It also consumes less power, the low power amps are smaller and less bulky in the room, and everything is easier to power. I love it.

So, instead of buying a $5,000 pair of speakers and a $2,000 stereo amplifier, you could get a $4,000 pair of similar speakers, an active crossover, two or three smaller amps, and the extra cabling, all for similar or less money and the resulting sound will be vastly superior in every possible way.

Of course, if you have an unlimited budget you could still purchase very high end amps and, yes, you'll still get that slight improvement from the costlier amps, but it won't matter as much to the actual listening experience.

 

[TitaniumTroy]

This guy disagree's that active is in the near future, but he says nothing about whether active is better or not. Just audiophile's wanting this or that sound out of their amps. While I do love the look of my used Pass Labs XL250, I would prefer the sonic benefits of active. Also since I like the choice's my speaker designer made (Greg Timbers) given the parameters JBL gave him. Why wouldn't I like his choice in amplifiers and if I wanted to customize their sound, I could do that via the speaker itself.

Plus it would cut down on the amount of cables I have, in my system which is a good thing.

How do you think this Genlec speaker would sound in a living room, Flint?
https://www.genelec.com/studio-monitors/sam-studio-monitors/1237a-sam-studio-monitor

 

[Flint]

How do you think this Genlec speaker would sound in a living room, Flint?
https://www.genelec.com/studio-monitors/sam-studio-monitors/1237a-sam-studio-monitor

Genelec are stunning speakers and offer as close to ideal performance for their intended applications. Basically, if you buy the model designed for how you are going to use it in your space, I believe it will sound incredible. But I feel the same way about most high end studio monitor brands like Dynaudio, Focal, Neumann, JBL and many others.

The problem with audiophiles is they refuse to depart from the past. When the CD makers took away their ability to tweak the source, audiophiles invented stupid ass nonsense, like the green sharpie or freezing discs, with mystical abilities to improve the sound. When the digital interface guys eliminated issues related to jitter, phase shifts & ringing at the brick wall filter, clocking mismatches, least significant bit issues, and such, Audiophiles kept on paying too much to address problems which have been real for over a decade, or they insist high resolution is inherently superior without any evidence to support it other than "I know what I can hear". When the high end amp makers started making digital class amps made from the same silicon used in $500 powered monitors, they insisted they were different because of something like filtering or weight. Basically, they just ignore very real, measurable, and easily perceptual improvements and differences in order to hang on to something they decided was real before they ever tried it out - like amp rolling.

 

[Flint]

Oh, and I've made it pretty damn clear what I think of that dick in the video.

 

[Flint]

So, to give a real world example... as a woofer cone moves in and out, the voice coil driving that woofer is moving in and out of the magnetic gap - that's how the motor of a traditional electromechanical cone woofer works. The voice coil is a wound coil of wire on a bobbin, which by its very nature is a basic inductor placed in the magnet gap. When we measure the inductance of the woofer, standard industry procedures say it should be measured at a specific voltage so we can compare different drivers tested in different labs. The inductance increases resistance from the input voltage to the speaker as the frequency is increased, which is one of the many reasons big woofers with huge high power voice coils do incredibly bad at attempting to reproduce midrange signals.

But here's the thing, that characteristic inductance from the voice coil on a woofer will fluctuate depending on where the voice coil is positioned in the magnetic gap. When the voice coil is pushed outward, the inductance goes down, when it is pushed inward it goes up. This is a critical characteristic of a cone speaker, so industry labs measure this variation using an analyzer from a company called Kippel. It is called an "L(X) Curve."

This is the L(X) Curve of a good quality mid-market mid/woofer from Dayton Audio:

Chart 1: Inductance Shown in a Kippel Analyzer L(X) Curve for a mid-market 4-1/2 Inch Midrange driver

This driver has an effect Xmax (or the physical effective limit on the distance the voice coil can move and the speaker remain functionally linear) of about 3.5mm. So, at maximum loudness the inductance from waveform peak to waveform peak varies from 0.115 mH to 0.200 mH. That a swing of 74%!!!

In normal use and loudness levels one would not be pushing this speaker to its maximum levels, but even in the +/- 1mm range the inductance varies by as much as 22% - that is not an insignificant swing in inductance. And, with a cone midrange driver, impact of the inductance is right in the middle of the crossover region for the transition to the tweeter.

So, the performance of all the components in a crossover is determined by the electrical load of the speaker, yet during operation those electrical load characteristics vary widely. Basically the electrical system is a bunch of resonating circuits working together to get the best results the limited technology can deliver. All these variations impact the sound and result in some speakers sounding way better at loud levels than others and some speakers sounding better with Tube amps than others and so on. It is a messy, sloppy, insane system.

But, it is possible to remove all of the slop of the passive crossover and replace it with an active crossover and directly coupling the speaker to the amp which, with nearly all solid state amps, has an output impedance of virtually zero, all those resonating circuits of the speaker drivers themselves are negated by the direct coupling of the amps low output impedance. Back EMF is essentially eliminated as an issue. The variations in the drivers' impedance, inductance, capacitance, resonance, and so on is negated within the operating limits of the drivers. And the crossover itself will always operate exactly to design at all SPLs, temperatures, and regardless of the load of the speakers.

This is why the clarity is so stunningly high with active speakers. It is also why manufacturers can produce shockingly revealing studio monitors for less than $250 a pair which utilize very low cost drivers, cabinets, and electronics. By recategorizing the product to be an "active nearfield studio monitor" allows consumers to accept the vastly superior design philosophy of active crossovers and smaller amps.

Go one step further than take the extreme engineering of ultra high end components, cabinets, and amplifiers and apply a state of the art DSP using cutting edge algorithms to generate the crossover filters and compensate for other unavoidable limitations of physical speaker systems, and you can get performance with is so revealing you'd swear that recording of Pablo Sanchez is really happening in your room when you listen to it.

 

[Randy]

Oh, and I've made it pretty damn clear what I think of that dick in the video.

I really wish you would quit sugar coating shit and beating around the bush about your opinions @Flint . I think it would be rather refreshing for all of us to actually know where you stand on a subject or individual.