So what's important when matching speakers together? How do you choose a speaker, and how do you determine their limits? This one covers midbass/midrange drivers, design theories, tradeoffs. The midrange region is the heart of the frequency spectrum. The critical portion of audio spectrum is considered to be 300hz-3khz, so it's important to get this one right.
There are three things you have to balance with speaker cones and their surrounding soft parts.
1: Stiffness. Stiffness is critical for quality audio reproduction, as it resists torsion and flexing. In the real world and it's limitations, the rigidity of a driver determines where resonances occur.
2: Damping. Anyone who has sound deadened a vehicle can tell you the importance of damping. It's purpose is to prevent fatiguing resonances and breakups.
3: Sensitivity. Not really critical to performance, just the fact that you don't want a midrange driver requiring 1kw to move it.
These three properties, unfortunately, are exclusive. Striking a balance means that as we add to one, we're taking from another.
Rigid drivers: Examples are metal cone drivers (aluminum, magnesium, beryllium), Kevlar, Carbon Fiber, Ceramic, Nomex, and other proprietary alloys and composites developed by many different manufacturers. These drivers are capable of near pistonic motion throughout their bandwidth, resulting in the lowest distortion and the lowest energy storage within their intended range. Why aren't they all over the place then? The resonance has to occur somewhere. In the case of rigid drivers (especially metal drivers), this occurs in upper midrange or lower treble, severely limiting their bandwidth and requiring a clever crossover designer.
Damped (soft cone) drivers: The most prominent are paper and polyprophylene. Past that there are plenty of variations and tons of proprietary composites from given manufacturers. The resonances of these drivers are well damped, showing no nasty peaks anywhere. They don't typically require a lot of crossover magic to get them to integrate well with a tweeter, and aren't unbearable if you play the driver where resonances occur. These typically have a wider bandwidth than a rigid driver, making them much more flexible with potential combinations. The bad? They don't perform as well throughout the passband. Higher distortion, higher energy storage.
How do I decide a suitable range for a driver
Where do resonances occur? How bad are they? What will I have to do to compensate? These are questions you will (hopefully) be asking yourself when choosing a driver. We, the audiophiles, must depend on both manufacturer specifications as well as independant testing data.
Let's look at a rigid cone driver:
Technical - Seas Reference
This is a technical overview of the SEAS Lotus driver. I chose it to keep you from having to download PDF spec sheets. Load the FR chart and the harmonic distortion graph. See that breakup at 4.5k, and the resulting distortion spike? We'd have to take that into account. More specifically, we'd deliver a precision strike to ~4.5k down with a notch filter
"Not too bad" some may say.
Not so fast. Look at the distortion chart again. What are those peaks in distortion BELOW 4.5khz? Harmonics. 2nd and 3rd order harmonics, to be more specific. Why? Harmonic distortion can still excite this resonance peak, which requires us to cross the driver lower. A general rule of thumb is to cross a driver over roughly 2 octaves below any breakup peaks. Assuming that the breakup is at 4.5khz, abiding by the 2 octave rule means we will be crossing the driver over at about 1.1khz. Yikes, I pity the poor tweeter that would have to take over duty there. So what they end up doing is crossing the driver so that acoustic rolloff occurs at roughly 1.7khz, which is roughly 1.5 octaves down from 4.5khz. 3rd order harmonics affect a breakup 1.5 octaves above that, so it works pretty well. Desireable? No. Tolerable? Sure. But what tweeter to mate with it? It takes a beast of a tweeter to deliver with that low a x-over frequency. More on that later.
It takes a good motor to make a good driver (this applies to any driver). Repeat that to yourself as many times as needed . With rigid drivers this is moreso. The lower distortion, the less their peak is excited. Ever wondered why a lot of metal dome tweeters can be so harsh? Especially the cutesy palm in your hand neo compact car audio domes? Turn the volume down. Chances are they aren't too bad assuming all is set up correctly. But the more and more the volume turns up, they get harsher, don't they? And by the time you're at their limits, they're screaming at you, aren't they? Your typical 1" metal tweeter has a breakup around 25khz. We can't hear 25khz. But if we have a breakup at 25khz, that means that 2nd order distortion can excite this peak, which means 12.5khz would be able to excite this peak, and this breakup can bleed into the audible region. And 3rd order distortion means that ~8.3khz is capable of exciting this breakup. This means you need a motor with extremely low distortion.
So why not do a damped driver? It doesn't have these limitations of crossover right?
Sort of. Let's look at a "soft cone" driver.
The Adire Audio Extremis:
Distortion is low, no crazy breakups in FR. Should be easy to mate with a tweeter. But look at the waterfall:
See that energy storage below 2khz? In case you've never seen a waterfall, it's that big peak around 2khz-it's the one that's sticking out in the open. So we have pinpointed where resonance is at it's worst with this driver. The bad news is that it's a lot of energy storage, and it's in a very audible frequency range. It's also in a lower frequency range than the above driver. The good news is that it doesn't cause any crazy spikes in distortion or FR. Because it is damped, we don't have to take the same measures that we did with the above driver. But the breakups of both drivers are very much audible, we just treat them differently. Remember what I just said above:
So that SEAS Lotus midwoofer shouldn't be crossed past 1.7khz at the most, but the Extremis? It's damped enough that you could run it above 3khz if you wanted to. Maybe even higher, your choice, but it will partially be determined by my next point:
Drivers are only so good off axis. What is this relative to? The size and shape of the cone. If the short high frequency waves radiated from one side of the cone meet the same sound waves from the other side of
the driver, those sound waves will interfere to cancel and reinforce each
other. The result: Uneven frequency response and poor off-axis performance.
Sometimes this is countered with phase plugs, but they are only good to an extent. What's important to remember? Once the wavelength becomes shorter than the diameter of the speaker cone, "beaming" will occur. In other words, off axis dispersion suffers. If pushed high enough, dispersion literally narrows to a beam.
For a 7" midbass driver (or "oversized 6.5"), this typically occurs around 2khz. Meaning your off axis dispersion will suffer past that. Just take a look at a FR graph and notice where the off axis plots fall off at. Cone shape can aid to an extent, but like phase plugs they are not a cure.
I found this chart a while back:
Speaker Diameter (inches)
Theoretical Maximum Frequency Before Beaming (Hz)
Just keep the word theoretical in mind.
So we've covered theory of soft parts. Now the most important part: the motor. It doesn't matter what kind of speaker you're making, the motor is the backbone of a good speaker driver.
Is it about crazy x-max? Not always. With a midbass/midrange, what's most important is the consistency. What consistency? A flat BL curve and a flat inductance curve Low inductance is important, too, but it must also be consistent over the range of a driver's excursion. The good news is that motor design has improved substantially with the aid of computers-primarily the ability to analyze motors via FEA, Klippel and other tricks. What are some good examples of speaker motors? Some of the best I've seen and dealt with come from Peerless, Scan-Speak, SEAS, and the XBL^2 units. A couple of examples:
I will further delve into motor design later. This is my first stopping point. If it seems incomplete, it is. There is more on the way, it's that I've posted a lot of graphs, charts, said a lot of "lingo", and made a lot of points in a really long post.
There are many motor topologies and many different ways to make them perform well. Motors tend to have certain characteristics, but it's important not to stick to generalization. If you looked at the drivers linked above, you'd swear they were underhung or XBL^2 by looking at the BL curve. And all except that CSS mid use overhung motors. Bottom line is that there are excellent and poor implementations of any motor topology, and you won't be able to determine this unless you look at testing of a specific driver. 'Nuff said.
What's #1 when referring to motors? IMO-quality control. Engineering gives you an idea of what a driver is supposed to do. QC determines what will arrive on your doorstep. It's too easy to screw up a speaker. Peerless is an example of a company that has some of the best QC in the industry. Every Peerless driver I've seen tested has performed VERY close to the specifications.
How much excursion do you REALLY need? Good question. For every octave lower, you have to increase excursion 4 times. Repeat that, 4 times. Chances are if you think that extra millimeter of x-max in a midwoofer is doing something-not really. If you have a 3mm x-max midwoofer, you'll need a 12mm x-max woofer to go an octave lower at the same volume level. But do you want your driver producing delicate midrange and 50hz simultaneously? This brings my next point:
We've all heard it, typically considered doppler distortion. Muddiness. Cluttering of the frequency spectrum. Loss of resolution. Fatigue. Harsh, screechy midrange.
Intermodulation is the interference within speaker components. As you can imagine, this increases as more power is applied. Excursion at low frequencies tends to modulate middle and higher frequencies. And at the top end of the spectrum, the chaotic resonances and vibrations that occur within a speaker cone can cause lower frequencies to sound harsh.
How to prevent this? There is a commonly accepted theory that a speaker should produce no more than 3 octaves. If you have a midbass driver covering 80hz on up, this would mean that it would hand off to a dedicated midrange driver at 640hz. So on and so forth.
As you're probably thinking, this is difficult to do in a car. 2 ways are generally the best thing for most vehicles. My recommendation: If you can't do a dedicated midbass driver, then don't try to get a midwoofer doing 50hz. Cross it as high as possible without destroying the imaging and upfront bass. With some tweaking you can make this work. Remember, 100hz is an octave higher than 50, meaning it requires 1/4 the excursion that 50hz does. This cuts down distortion dramatically.
So, to summarize.
When choosing midbass drivers you want to make sure that:
1: You know where resonances lie, if they need specific attention, and if a driver performs well enough to make your list.
2: Consider the dispersion capabilities of the driver, especially if you are going to be installing in a location that is far off axis.
3: Choose a driver of good motor design. Also choose a driver of good quality control.
4: Don't make a driver do too much. Using a single driver over a wide range is rarely as desirable as using a driver intended for a specific bandwidth. The most attention should be put into not trying to make a midbass play too low for your intended listening level.
Part 3 will discuss choosing drivers based on the limitations of the car environment.
Before we take the car environment into consideration, I'm going to give a tutorial upon the human ear. It is imperative that you understand how we percieve sound in order to achieve good results within a vehicle. This is going to be a VERY long post, for your warning.
Human perception of sound
The backbone of human hearing is due to stream formation. Primary sound events MUST be separated from secondary sounds. Primary sound would be the source signal, secondary being reflections, reverberations, etc. Primary sounds would be in streams-being in a common direction, a common timbre. A set of sound events from a single source becomes a sound stream.
Intelligibility of a sound is dependant upon the separation of these sound streams from secondary sources. This means we have to minimize reflected energy and noise. Fluctuations in the level within a given time period are percieved as a sense of distance and "air". These fluctuations are caused by reflections. Time periods are crucial here, though. If a reverberation occurs quickly (within ~50ms of the original signal), we percieve that as having space around the source. If longer than that, we perceive these reflections as space around the listener. This is referred to as envelopment. Reflections are how we percieve distance as well. If the reflected energy occurs within around 150ms of the original signal, we percieve that as distance.
Now, the first set of problems. Reflections up to ~50ms do not impair intelligibility. Reflections from 50-150ms add to the perception of distance, but they also create muddy sound
Primary sound stream: The major portion of the sound. In music, this would be the band, a vocalist. Intelligibility is the indicator for this one.
Secondary sound stream: An example would be background noise. Sounds that occurs a long time after the end of a primary sound stream. Reverberation is the key to this one.
Spatial impression: This would be an early reflection that occurs very shortly after the primary sound. The perception is the presence of the music ("air", distance, acoustic space).
Last one relates to secondary reflections. Late reflections of the original source. We percieve this as distance and/or muddy sound (loss of intelligibility).
Imaging and Localization
Localization is related to our above points. For localization, we depend upon:
Interaural Intensity Difference. The difference in the intensity of sound arriving at the two ears. Volume level, basically.
Interaural Time Difference. The difference in the time of arrival of a sound wave at the two ears. This is primarily dependant upon pathlengths, but in a car we also have to take reflections into account. More on that in the next tutorial
The outer ear. When sound comes in contact with the outer ear, its frequency characteristics are modified. These modifications vary depending on the position of the sound source and provide an important directional cue.
Those are the basic things to take into account for imaging. But wait, there's more:
Distance. How do we percieve distance? The brain perceives distance as a function of the relative amount of reverberant or reflected sound to the amount of direct or unreflected sound received by the ear. This quantity is expressed as the reverberant/direct (R/D) ratio. This is critical in acoustic engineering, especially when referring to architecture.
The Haas (precedence) effect. It's also called "the law of the first wave front". This one is VERY related to car audio. The Haas effect describes our perception of imaging when using two or more sources. Note that this is a BIG explanation of situations with differing pathlengths (such as in a vehicle). After hearing a signal, the ears will suppress any subsequent signals for about 25-35 milliseconds. If the difference is greater than 25-35ms, we hear two distinct sounds. Here we're describing the human psychoacoustic phenomena of correctly identifying the direction of a sound source heard in both ears but arriving at different times. Obviously we have two ears spaced apart, which are separated by a barrier. The direct sound from any source first enters the ear closest to the source, then the ear farthest away. The Haas Effect tells us that humans localize a sound source based upon the first arriving sound, if the subsequent arrivals are within 25-35 milliseconds. If the later arrivals are longer than this, then two distinct sounds are heard. This is even true if the later sound is louder than the first (even by 10db-twice as loud).
So what does that mean? It means that if your right speaker is delayed in comparison to the left (within the time window), the total sound will actually seem to come from the left speaker alone. But what's also important: sound arriving at both ears simultaneously is heard as coming from straight ahead, or behind, or within the head. To simply sum this up: You have two sources. Source A, source B. The sound created at source A, which is closer to the listener than source B, arrives first. To the listener, this creates the impression that A is the only source of the sound. If both arrive simultaneously, the sound seems to come from in front of you, behind you, or within your head.
Now, localization after all these terms. Requirements:
- The rise-time of the sound event must be more rapid than the rise-time
of the reverberation
- The IID (Interaural Intensity Difference) and the ITD (Interaural Time Difference) must be unaffected by reflections, and during this period we detect the direction of the sound source. Then the brain holdsthe detected direction during the reverberant part of the sound.
- The conversion between IID and ITD and the perceived direction
is simple in natural hearing, but complex (and unnatural) when
sound is panned between two loudspeakers.
Our perception at different frequencies
Head Related Transfer Function. The HRTF represents the spectral or frequency component filtering that occurs to a sound as it travels from outside the head into the ear canal. HRTF captures all of the physical cues to source localization. This is EXTREMELY complicated and takes four variables into account: three space coordinates (azimuth, elevation & range) and frequency. To make matters even worse, they change from person to person. HRTF is dependant upon YOU. Your head shape, the space between your ears, your shoulders, amongst many different variables. Next point:
Panning of two speakers
The pressure at each ear is the sum of the direct sound pressure from one speaker and the diffracted sound pressure from the other, which interfere with one another and create a highly frequency dependent signal.
A two channel pan is completely different from the localization process in natural hearing.
Localization is dependant upon IID and ITD. Naturally these alone vary due to HRTF. With two channels, IID and ITD vary substantially because of the above described interference. This is physical, not perceptual. It's measurable and able to being calculated. This makes ITD and IID highly frequency dependent.
Hey, this one's easy. We have a visual field or expected location in mind. We also go on a "past history" type of perception. Basically, this can be proven by playing a known source such as speech. We could then alter the bandwidth and/or frequency of the source and we will still percieve this as having the same localization (despite the drastic differences in IID and ITD). On top of that, our thought process tends to guess at localization over the entire stream, and not the localization of the individual streams. This is also in relation to our hearing mechanism being reluctant to change it's perception.
I'll make it short and sweet. Syllabic sources (when one note is related to one syllable in the text) are percieved as being very sharp and narrow. Broadband sources (such as an instrumental section, chorus) are percieved as being much wider than they are due to inconsistencies in IID and ITD.
Applying this in the car
Nothing is perfect. The two properties you need to consider the most will be IID and ITD. I'll keep this simple:
Our perception of imaging is mostly dependant on ITD below 500hz or so, in fact volume level isn't very important below that point at all.
From 500hz to about 1500hz, IID becomes increasingly important. You want to have good performance in both areas.
Above 1500hz, we become less dependant upon ITD and more dependant upon IID for imaging.
At around 6000hz, our heads create an acoustic shadow. IID is the dominant form of percieved imaging-which means pathlengths aren't as important.
More to come in the next tutorial
Part 2-The Human Ear
The above relates mainly to our ability to localize sources in the horizontal plane , which would be a center image, stage width, etc. But how do we localize sources in the vertical plane, which gives us our perception of stage height?
Defined above. It is our heads and ears that allow us to perceive stage height. Consider the anatomy of the human head, though. It is essentially a sphere with two holes in the sides. Which means that sounds of ALL directions, from ANY point source could absolutely NOT be localized vertically by our ear canals alone.
Outer ear, again described in the last tutorial. Our perception of stage height depends solely upon the frequency modifications that the shape of our outer ears create. Look in the mirror. Obviously the shape of your ear would cause frequencies to be modified in different ways from sources of different heights.
This relates to two points above, HRTF and Expectation. Our brains remember how our ears modify given sounds, and due to this memory, it is how we perceive stage height. Meaning that if a sound comes from below you, our brain has predetermined what characteristics a sound from below would have. And the same applies to a sound emanating from above.
What frequencies are important for stage height
The dimensions of the outer ear can only have an effect upon frequencies above ~2kHz. Which means frequencies below that point can be physically low and still be perceived as high. That is assuming the 2khz and up region is high or is perceived as high http://www.caraudiocentral.net/forum...ilies/wink.gif)
The most important point of this post: Our ears do not perceive physical height, they perceive the frequency modifications that our ears make upon sources of different heights.
What does that mean in your car? It means that you have to replicate the characteristics of a high stage.
Second point: Your drivers (even tweeters) do NOT have to be physically high to have stage height
One way to get stage height would be to make sure frequencies above 2khz are physically high. BUT, what about tweets down low, such as doors or kick panels? If you tune your tweeters to replicate the frequency characteristics your ears would have if they were mounted higher, you will perceive just as high a soundstage That's a BIG point. It means that if your tweeters are low, you can tune them to replicate a frequency response associated with a higher source. A lot of times you reap the benefits of not being directly beside a reflective surface as well.
Now to get the ball rolling and dig into the good stuff http://www.caraudiocentral.net/forum...iggrin.gifThis tutorial will start to cover general theory into system design.
Running a lower crossover point
I'm doing this one first because this one benefits those that plan to install in stock spots. This tutorial applies to 2 way applications. A typical vehicle will have 6.5" or 5.25" midbass drivers in a low door location. When you are attempting to design a speaker system that will work from factory locations, the first thing you have to take into account is the extreme off axis nature of a factory door location. The drivers side door is the worse of the two, often being 60 degrees or even more off axis to you. The passenger side mid will usually be off axis, but less so than the driver side.
The main benefits
You are limiting your midbass drivers to a region where they retain good dispersion. This is critical for getting good imaging from off axis locations such as factory door spots. From there the tweeter takes over, which has much better dispersion than a midbass driver would within that intended range.
Car doors tend to create reflections, aka standing waves. Not really just car doors, but enclosures in general. This can especially be a problem in the upper midrange region and a lower crossover point can help combat some of these problems.
The lower the crossover frequency, the less lobing is an issue. Lobing is a fairly complicated phenomenon and is better defined as a form of destructive interference. To keep this part of the tutorial simple, as the crossover frequency becomes higher the problem worsens. The effect of lobing is cancellation at different axises. As a side note, generally the shallower the crossover slope, the more severe the effects of lobing.
The basic summary is that lower crossover points allow you to get better off axis performance from a speaker system, and that is critical when you're talking about installing speakers in factory spots.
In case you're wondering which conventional car audio manufacturers tend to show this design theory, you'll typically find it from mostly the more expensive offerings of high end manufacturers:
DLS (2 ways)
Focal (some cases)
Keep in mind those are just a couple of examples.
The tradeoffs of this design
The limitation really ties in to a statement that I've made numerous times http://www.caraudiocentral.net/forum...s/biggrin.gif:
Most car audio equipment sucks
In this case, the tweeter is really the limiting factor. A wussy compact neo dome isn't going to do any justice with a 2khz acoustic crossover point, much less a 2khz electrical point. They don't have the motors, the rear chamber, or the flange design to cut the mustard. There are some notable compact car audio tweets that do a pretty good job at it, but none I've seen can match the performance of a well built large flange tweeter. It boils down to the fact that it's easy to build a tweeter, but extremely difficult to build a good one. Mounting a 1.5" deep tweeter with a 4" flange isn't on the priority list of most. Even if you can, the lower the crossover frequency, the more you limit the high volume capability of a tweeter. Excursion requirement quadruples with every decreasing octave. Meaning a tweeter crossed at 2.5k will have to move 4x as far as one crossed at 5k. With that comes added distortion, especially at high volumes.
One of the bigger tradeoffs is that if you want a tweeter to perform better down low, that means a larger tweeter. Many of them are 25mm (1") or 28mm (~1 1/8") designs. When you use a tweeter this large and put it off axis, top end dispersion typically suffers.
Another tradeoff is that you're crossing over in a more "critical" frequency range. Our ears are most sensitive in upper midrange and ideally we'd have one driver performing the critical region. Some, including myself, question that argument at times since our ears are most sensitive at 4khz, and most 6.5" midbasses don't sound great running that high to begin with. Or at least they don't sound great compared to what could be covering that region- a good tweeter or midrange.
Using a higher crossover frequency
This applies to 2 ways. Your typical setup will be a 6.5" midbass driver and a compact tweeter.
The benefit of running a higher crossover frequency, especially with car audio equipment such as compact neo domes, is the fact that a midbass driver will retain good performance at higher frequencies (say the 2-5khz region) in regards to distortion and FR. That is of course assuming the midbass driver doesn't have any large resonant peaks, low inductive rolloff, poor frequency response, etc. This area is where a lot of tweeters would struggle. Another benefit is the high crossover frequency allows cleaner performance at higher volumes, at least in that region (though the midbass driver may still suffer in the bass region due to the level of output required). It can also let you use a physically smaller tweeter (say a 3/4" dome), which means you'll typically have better top end extension and detail as well as better off axis performance from said tweeter.
Now, if you remember from the previous tutorials, one point I made is the fact that around 5-6khz or so, IID becomes the dominant form of percieved imaging. What this means is that running a midbass higher will allow you get closer to that point and be able to take advantage of A-pillar mounted tweeters, since pathlengths are of decreasing importance and volume is what mostly determines how your tweeter images at that point. Having the tweeter up top will obviously help solidify your stage height. You'll also notice that the midbass driver is covering nearly all of the "critical range", and there is no crossover point in there or difference in pathlengths. Meaning you'll get uniform sound if your driver is up to snuff.
The downside of using higher crossover frequencies is the fact that you're pretty limited in driver choice. There are only certain drivers that will play that high cleanly, and your options boil down to the damped (paper, poly, etc) variety. Similarly, drivers that will perform well in the upper midrange tend to trade off for less bass extension due to the fact that paper/poly drivers tend to be (intentionally) lightweight , meaning usually a higher Fs and lower Qts, boiling down to a rolloff in the midbass region, perhaps at a higher frequency than you'd prefer. Also, the motors can't have too long a coil in order to avoid an inductive rolloff, especially if you're talking drivers without a plated polepiece or similar. The other big downside is that dispersion from a 6.5" driver is rather poor above 2khz or so, if you're using a 5.25" driver it'll be good up to around 3khz or so. This means you really need to have the midbass drivers be more on-axis to take advantage of their upper midrange performance. Also, since the passenger side tends to be more on-axis than the driver's side, this can mean a side-bias to the right side due to the fact that it's upper midrange will be more audible than the left (driver's) side. This will mess up your imaging as well as the music itself. If you want more directional drivers, it usually means either door pods (which most aren't willing to do) or kick panels. Kick panels open up their own problems, mainly the fact that it doesn't make a good enclosure, and consequently you have to take action to make sure your midbass (which may already be on the thin side) doesn't suffer. Also, lobing is a bigger issue with higher crossover points. Lobing, as described before, is a destructive interference between drivers. Higher crossover slopes tend to help here, but you still have to pay attention to the separation between the drivers. They don't necessarily have to be side by side, it's just that the difference in distance from your head has to be similar (say an A-pillar tweeter is 20" from your head, and a kick panel mounted midbass is 21" from your head, that would image pretty dang good. Note: that scenario would rarely happen). But the downside to having the drivers in such places is that the reflective surfaces are much different, and consequently how the drivers sound will also be different and the system as a whole may not integrate to your preferences both tonally and as far as imaging characteristics. That can basically mean you can tell pretty easily what sounds are coming from which driver. That's not a good thing. Time/phase alignment helps here, too. The last issue is energy storage from the midbass driver. The upper midrange region is where midbasses typically suffer from energy storage, which can create a slightly muddled sound if bad enough. In other words, realism, precision, and clarity isn't going to be the best at that point, certainly not as good as a tweeter in regards to that aspect of performance. But note that the distortion will still be low at that point due to the damped nature of the driver.