37 replies to this topic

[br85]

I have the feeling that as old as this site is, and as many topics have been raised, there hasn't been a thread that explains what to look for in a good driver (or speaker i.e. component set).

Anyone feel free to jump in and bring something to the table, but be warned if you talk sh!t you WILL be called on it (by me, thematt, muzzy66, Pulse-R, etc).

So much bad advice out there that waste's people's hard earned money when they simply do not know what to look for. Because of this, many people lose faith in measurements and start "trusting their ears only". This is not because the measurements have failed them, but because interpreting them is not as simple as looking at a graph and making instant judgments on how a speaker will perform. Choosing a driver cannot be taught the standard A+B=C method, it requires a lot of interpretation and thinking. There isn't a spoon-feeding method, so I couldn't even if I wanted to.

Anyway, enough babble. First things first.

Distortion.

Distortion can be divided into 2 categories, linear and nonlinear. Linear distortion is simple, have a look at a frequency response graph and where it is not flat, this is linear distortion. However, this is usually not a big deal because you should never look at a driver's performance from 20hz to 20khz. Consider the appropriate use of the driver, and then imagine a big rolloff around the ideal crossover points.

Nonlinear distortion is not so simple. The most important part (and only part that I will discuss in this post) is THD (Total Harmonic Distortion). Sometimes this is listed as a spec, but this is very misleading, a graph will tell the story a lot better (still not great though unless you know how to visualize a graph changing in your brain). When a speaker plays a 200hz frequency, a certain amount of 400hz(2nd order), 600hz(3rd order), 800hz(4th order), 1000hz (etc etc) will output from the driver as well (harmonics). There is usually less and less of each harmonic (i.e. 2nd order is generally the loudest). Looking at the graph gives you a *rough* idea of the frequency range that you should not play the driver. When you see much over 1% distortion of any order, this is a good indication that at 10Watts, or 100watts, it's going to be well over 1% and very audible.

Breakup modes (I will explain later) can often be spotted in Harmonic distortion graphs as 3rd of 4th order actually being higher than 2nd, or other similar anomalies. Point is, you want to steer well clear of breakup modes because even 30db below the main level, they can still be heard as something of a "ringing" or "hanging" sound. Since most of us don't (or don't know how to) use notch filters in the car, best just to cross over a long way away from them.

Distortion is probably the most important factor in choosing a driver for your needs. Harmonic distortion especially, and then Linear distortion plays a part as well. If there is a BIG problem in the frequency response within the target passband, maybe give it a miss. (Having said that, if it's a dip or peak of relatively shallow nature, the baffle, waveguide, or position of install may help to smooth it out, it needs to be considered)

Beaming, Polar and Power response, Off Axis

Ugh. This stuff is very poorly understood if at all by most car audio nuts, and I may lose a few here because the terms used cannot be avoided without writing an entire textbook. Anyway.

Beaming is pretty much what it sounds like. At low frequencies, the driver is spraying out sound pretty evenly in a sphere (hemisphere when we separate the back-wave), but as we get higher, the driver gradually starts outputting most of the sound directly in front of it. When you look at off axis Frequency response measurements, you will see the effects of beaming. Basically at low frequencies, it's like a floodlight, and up higher, it's like a spotlight (or laser beam). This is almost always dependent on the size of the driver's moving mass.

Polar response is a better way to look at beaming, because it tells you everything from on axis to 180 degrees instead of just 30/60. Problem is, polar response graphs are rarely available, and they only tell you the info at a few select frequencies.

Power response is very difficult to get your head around. Basically, it is a way of looking at the speaker's actual output levels with beaming/polar response considered. Eg. If a driver is 90db on axis at 2khz and 80db at 30 degrees, this is a lower power response at 2khz than a driver that is 87db on axis AND at 30 degrees. Maybe ~thematt~ can do a better job explaing this concept than I can.

Anyway, off to work, more later.

Edited by br85, 03 September 2009 - 09:45 AM.

[Matt VIP]

keep it going mang!

[muzzy66]

Many people often ignore frequency response and respond with the whole "just EQ the s**t out of it later" concept.

I've noticed in my own installs that while EQ in the right places and improve the tonality of a system, it often seems to impact begatively on imaging accuracy, staging, etc I think this might be to due to the impacts of EQ on phase response, and it's why my PEQ has been set to 'off' in my last 2 or 3 systems.

I know this isn't directly related to speaker choice, but there are tons of people out there who use EQ's the way women eat chocolate. I don't have the greatest technical understanding of EQ and why it affects signals the way it does (I just understand that it does), so if anyone can describe that concept in greater detail it might help others understand why it's so important to choose speakers carefully, and to use EQ sparingly.

Edited by muzzy66, 03 September 2009 - 11:26 AM.

[fuddbutter]

I generally look for cone, magnets, sometimes pole pieces.. you know, speaker parts

[pyr0maniac]

This thread came up in a google search the other day, I found it pretty interesting. It's to do with T/S specs for midbass application.

http://www.mobileelectronics.com.au/forums...php/t79323.html

[~thematt~]

Energy Storage
A Driver is a converter of energy. It takes electrical energy, and through interaction with magnetic fields, creates a motion which results in pressurization/depressurization of air particles and produces sound. However, only about 0.5-1% of the energy that goes into a standard driver, actually results in sound. Almost 99% comes out as heat.

The behaviour and design of a driver becomes very complex, very quickly, so I'll only skate over the basic information relating to energy storage. In the perfect world, the kinetic energy of motion is converted into acoustic energy (or heat) instantly, such that at any snapshot in time the energy that exists is either acoustic, thermal or electrical.

However, in the real world, the driver is made up of flexible parts. All materials have some degree of strain when stress is applied (short of glass and its partners) and therefore all materials have the ability to store, in snapshots in time, energy within its own particles. For a driver, this energy storage occurs with the spider, the cone and the suspension.

If we were to simply look at a drivers response through individual snapshots in time, we would see energy being 'lost' and energy being 'found' all the time. The storage of this energy is given off slowly by the driver at a certain time after its storage. This arrives to us as acoustic distortion, specifically non-linear distortion.

You may, or may not, have heard of the use of the term 'piston' or 'pistonic'. What this term refers to is the ability to resist the storage of energy through a particular range of frequencies. Drivers with hard cones, such as aluminium or magnesium, resist energy storage through a bandwidth, and therefore enable the driver to have lower non-linear distortion. Drivers with wood, or pulp cones, can store energy quite well, and therefore have higher non-linear distortion. This is why the Seas Excel Mags were always considered very very good drivers. Through their bandwidth, they stored very little energy (due to the magnesium cone), and were therefore a naturally 'clearer' driver with lower distortion. Its also why the Scanspeak Revelators have the 'slit cone' design, to eliminate as much energy storage as possible.

Another example is the current Pioneer TS range. The cone has been designed to store very little energy, and the suspension even less so. This makes these speakers an absolutely wonderful and very clever design indeed.

**Edit**
Polar Response is a fairly complex subject, and I'm not confident I can even partially explain it in simple terms in this thread (because I dont fully understand it myself). That being said, its basically the measure of all energy, both on and off-axis, throughout a drivers bandwidth. Its not the amplitude of the energy, but its radiation response that makes all the difference. The most important use of Polar Response comes into teaming two drivers together through a crossover point. Its obviously only important to people who realise that a crossover point is more then just a mere frequency cut and slope.

Edited by ~thematt~, 04 September 2009 - 12:11 PM.

[br85]

~thematt~ is right on the money about energy storage. However, it is very important to realise that cone geometry plays a much bigger part in how it stores energy and what range it can be used than cone material. They both work together, so it's pretty hard to separate them, but it might make a person wonder why the paper cones on drivers like the SLS8 and the exclusive 4's can perform very well pistonically considering the stiffness of material used. Let's just say these drivers have taken a compromised approach where the geometry is designed for maximum 'pistonic' accuracy within a certain range, and the material is a soft material (but somewhat stiffened to make it acceptable) to keep breakup nodes from requiring sharp crossover slopes. Geometry (with the slit cone design) is also the reason the revelators don't have to be made of aluminum or magnesium to perform well.

Lets have a look at more stuff.

XMAX, Power handling, Sensitivity, radiating cone area

At some point in time you're going to encounter these 4 specs, and you can't really ignore any of them. XMAX is to do with the driver's ability to move in and out before dynamic compression becomes an issue, Power handling is a thermal "meltdown" point which MANY factors can affect so it needs to be taken with a grain of salt, Sensitivity is a spec (or better than a spec, a graph) which tells us the spl we can expect from 1 watt of power, and radiating cone area is going to help us interpret how we can effectively use xmax and sensitivity to decide whether a driver is worth it, and what to do with it.

Let's have a look at the most critical driver, the midrange:
What we are after in a midrange is going to be a bit different for everybody, but all of us should work out the dynamic or spl target we're chasing in this range. It might be 90db for some and 115db for others, but it should always be the first. (I am talking about 3 way fronts here, not 2, conventional 2 way fronts are a serious waste of time if you're going to this much effort). 160hz/6db slope is the absolute lowest, most insane crossover I'd recommend for ANY mid in a car, so 160hz, 200hz, or 250hz might be a good place to start doing calculations at.

If you want a midrange that's got some real dynamics (i.e. 105db or higher), you'll soon find out just how much the "no free lunch" saying applies. You CAN'T have a midrange that can do all of the following:

*play at 105db wothout reaching xmax
*play down to 160hz
*play up to 3khz off axis or in a 2-seat setup

So you're stuck with a difficult choice:
*Live with poor dynamic output (poor spl capability)
*Get a big driver that can play nice and low (lower than 200hz) but needs a very low playing tweeter, or sacrifice the ability to build a 2-seater
*Get a smaller driver, not ask it to play very low, put it somewhat on-axis and try and fix up the low mids in the midbass driver
*double up on the number of mids (but likely still have to make it a one seater)

I didn't go into specifics of the actual specs because I figured most of ou would know what they are by now. However, no question is too dumb or too menial to ask, so if you're missing any info, feel free to ask.

[br85]

It should also be kept in mind that every driver in the world (okay, there may be some exceptions, and dome midranges might be one such exception) is designed to be used On axis, or at least within 30 degrees (stereo triangle without any toe-in). No driver is *meant* to be used off axis, especially not upwards of 60 degrees like factory locations in cars.

In the home, using drivers off axis results in a the main energy of the driver becoming a reflection, and a weaker signal being the direct sound to your ears. Why is that a bad thing? Think.

In factory locations in a car, we've generally got carpet nearby so we might get away with a bit more, but it's still a problem. (Gear sticks, dash, center console etc) And reflections aren't the ONLY problem with off-axis installation. We've also go to deal with higher amounts of linear distortion unless the driver is crossed well before it's beaming point . With midbass, this means the need for 3 way if you're going to install off axis and still have decent bass, and with mids, this means using horns or very low playing tweeters if you absolutely MUST install off axis.

(Why do we even TRY to make traditional 2 way sound good in a car? The more I think about, the more I can't figure out why people say it's easier to get right. It's damn near IMPOSSIBLE to "get right")

[eclipse_DYN]

This is a great write up guys, I've read every single word of it. Thanks for taking the time.

One thing that might be handy to me (and probably other novice's) would be to describe what these different things sound like. ie what does beaming sound like? how do i know when my speaker has hit xmax? what does distorsion sound like?

Br85 - When you said that dome midranges may be the exception to playing on-axis, are you referring to the Dynaudio Midranges such as my own? If this is the case, does that mean that the placement of the midrange is less important as it is not specifically designed to be on-axis?

[~thematt~]

br85, on Sep 5 2009, 12:01 AM, said:

~thematt~ is right on the money about energy storage. However, it is very important to realise that cone geometry plays a much bigger part in how it stores energy and what range it can be used than cone material. They both work together, so it's pretty hard to separate them, but it might make a person wonder why the paper cones on drivers like the SLS8 and the exclusive 4's can perform very well pistonically considering the stiffness of material used. Let's just say these drivers have taken a compromised approach where the geometry is designed for maximum 'pistonic' accuracy within a certain range, and the material is a soft material (but somewhat stiffened to make it acceptable) to keep breakup nodes from requiring sharp crossover slopes. Geometry (with the slit cone design) is also the reason the revelators don't have to be made of aluminum or magnesium to perform well.

Oi! I was getting to geometry! Started simple and was going to extend it outward to the benefits of steering arrays! Now you've just ruined my 'flow'....

XMAX
I want to extend a bit more on Xmax. Understanding of what it is, and why lots of it is bad.

In the most basic of terms, Xmax is just a measurement, in millimetres, of the amount of extension the voicecoil travels from rest to a particular point. Most of the time, this point is chosen when BL is 70% of its maximum value. What this means is, Xmax is a measure of the amount of throw capable by a voicecoil until a point where the non-linear behaviours of the system begin to increase in their influence.

Does this mean that a higher Xmax results in lower distortion, because the motor is still 'linear' within a given range? Well, no. All things in life have compromises, and giving any motor the ability to throw further tends to result in some very very large compromises.

Lets look at it in simple terms. More throw means more distance to cover. This means larger surrounds that can still operate over the larger throw, larger spiders that can extend out that much further, and longer coils to operate in the same range. These result in less surface area for the cone (as larger surrounds eat into cone area), stronger motors to overcome the resistance provided by the spider, larger coils to take the power required to move the entire lot, and higher inductance because of the change in coil parameters.

Overall what does this all mean? It means that a woofer designed for higher xmax is now a more complicated system with a lot more things in it that can (and do) operate in a non-linear fashion, and therefore produces a higher distortion output.

If we buy amplifiers with enough power to always maintain a linear delivery (inclusive of crest factor), why not also buy a speaker with enough 'output' to always maintain a linear delivery? How do we do this? Well, that's the easy part. Output alone is governed by one factor. Volume Displaced (Vd). This is a makeup of cone area and Xmax. If we want to keep distortion down low, and linear, then Xmax needs to be low. This means that Cone area needs to be large. This is why top cars competing overseas have 15" and 18" subbass, 10" and 12" midbasses, 5" to 6.5" midranges or horns. Distortion.

The lower you keep the movement of your drivers, the more accurate and linear they will reproduce the input signal. This however, has tradeoffs (as do all things) which br85 has already touched upon. One being Power response, another is basic staging and two seat listening in a car......

Acoustic Loading
Note to BR85. I'm leading towards Baffles and waveguides, as well as steering arrays here Feel free to pick up and continue!

Acoustical loading is a term given to the 'load' placed upon the cone of the woofer by the air it is agitating. There is a massive difference in density (more then two thousand times in difference) between the cone and air, therefore a lot of energy is wasted in the transfer. If we decrease the density difference, we increase the efficiency of energy transfer. In the most basic of terms, what this means is that if we increase the density of the air attached to the woofer, then we will increase the efficiency of the transfer. This then results in getting louder off the same power (in SPL terms) or using less effort for the same result (in SQ terms).

So how is this achieved? We cant just be expected to drive around in a car filled with water now, can we? This change in density is referred to as modifying our Acoustic Loading. Its not the density itself we start to modify, but the interaction between the molecules which changes their behaviour. In short, we squeeze those molecules more, by restricting their movements. The easiest way is through a baffle. A flat baffle will increase your loading by a factor of two. This is known as 2pi steradian (steradian is the angular measurement of a solid angle). In order to be considered a flat baffle though, all frequencies that are output by the driver must have half wavelengths shorter then the distance between the driver and the end of the baffle. This means that if your midrange plays down to 345hz (for ease of measurements), your baffle must be 0.5 meter long from cone to edge to get full 2pi steradian loading.

For this 2pi steradian loading, your density is increased within this field, and therefore your acoustic loading is increased. The result? 3dB louder!

But what happens when we cut this solid angle in half again? We get pi steradians, and another 3dB in one space! Thats a total of 6dB (or four times the power drop) increase by firing from 'a corner'.

As you get more and more tight on the angular cuts your making, you begin to emulate a 'horn'. Just like a Trumpet. The cone will now agitate only a small fraction of air which, giving its acoustical loading, will now change its density much more, and therefore the efficiency goes way way up.

I now open the door to waveguides and steering......

[muzzy66]

br85, on Sep 5 2009, 10:12 AM, said:

Why do we even TRY to make traditional 2 way sound good in a car?

More often then not, it's due to limitations rather then people actually expecting better performance. Limitations in cost (3way means more amp channels, more drivers, or expensive passives), space (where do you put everything - my car is a good example here!!), complexity (designing a 3way system takes a lot more thought and planning), and extra install cost / effort (building custom FG enclosures for midranges, running extra cables. etc) make 3way a little too much for a lot of people.

For many others, I think they just don't fully understand how much you really do compromise with a 2way setup (from a sound quality point of view). It's increadibly difficult to get exceptional performance from a 2way setup in a car.

Anywhoo's - keep it coming guys...I think this thread would be very useful to many people (if they bother to read it)

Edited by muzzy66, 06 September 2009 - 08:20 PM.

[~thematt~]

muzzy66, on Sep 6 2009, 06:06 PM, said:

It's incredibly difficult to get exceptional performance from a 2way setup in a car.

Thats the thing. Its not difficult. Its impossible.

Polar Response differences. Power Response anomalies. High linear and non-linear distortion. Power Compression. These are only the start.

You can spend millions of hours and tens of thousands of dollars on getting a 2-way setup done well, and I'll still come along with a pair of $10 bookshelves and blow you away.

What I'm adding here isn't just some knowledge/experience that I've got/learned along the way, its hopefully the realization that if you continue to play with Sand, you'll never build your castle.

[br85]

As far as dome mids are concerned, the onset of beaming is very differently to cones. It should be pretty obvious by looking at a dome that it's going to have pretty good off-axis response compared to a similar sized cone driver. The reason it may well even sound better off axis than on axis is to do with how a dome propagates sound. Abmolech on the topic:

abmolech said:

The reason for dome mid range unpopularity is there increased distortion, lack of frequency band width, and the requirement to have them off axis.

Unless your completely stuck, a cone 4 1/2 to 5/14 " mid range will be a superior option.

The reason for this is the way with which each system deals with resonance. For the dome, it must use a self dampening material, that dampens the harmonics as the travel to the centre. (The reason why they must be off axis, this is particularly painful to listen to) It also suffers from inertia distortion because of the material being so lossy, IE the centre "collapses in-wards", that is, it lags the rest of the dome. This limits its useful excursion, and hence the frequency bandwidth.

Cones have a dedicated resonance system (the surround), and this allows a far greater choice of cone materials and geometry. It isn't perfect, because the surround it self has its own resonant frequency (about 2 -3 KHz) and this causes a "surround step frequency response". This can't be too audible because the "golden eared ones" (people that can hear cables, power amplifier differences) seem to recommend crossovers above this frequency. So either their opinion is right, or they can't hear as well as they pretend.

Anyway, avoid domes for a mid range if possible. (The tweeter may be acceptable because of the frequency range to use a dome, however it usually reflects a design towards a budget rather than best possible.)

eclipse_DYN:
"Beaming" is an effect you've probably heard before. Ever walked past a loudspeaker and suddenly experienced pain or a lot of high frequency sound while you are right in front of it? That's beaming. The high frequency information is not able project itself out as much as the lower stuff can.
Distortion sounds like anything that shouldn't be there, i.e. descriptions like "shrilly", "tinny", "muddy", "strained", "harsh" etc etc are all likely to be because a certain type of distortion is present.

Now that we know about xmax, I should probably expand on Radiating cone area (sd). Up to the point where beaming begins, the output spl (volume, sound pressure level, loudness) is all determined by the amount of displacement the driver is undergoing at a given frequency (displacement vd=xmax X sd). For example, a driver with an sd of 200cm^2 (about 8") with 3mm xmax is capable of the same output as a driver with an sd of 50cm^2 (about 4") and 12mm xmax. Remember that a bigger driver is great (9 times out of 10 they've got greater output and for less power) but you've got to watch out for 3 things that stop you from playing them up high: Breakup nodes, beaming (off axis dropoff) and inductance. All of these are likely to show up one way or another in the frequency response plot, but at 1 watt often they are masked, so it's good to know about them.

Waveguides is a big topic, I've got a thread about them already if anyone is confused, but if anyone wants a good read which summarise some of the big advantages they offer,the marketing department of sp tech (now aether audio) has put together a very informative article which is probably a little biased toward their design, but nonetheless scientifically accurate and easy to understand. The article is now floating around somewhere, lost the link, but here is a small excerpt I could find:

sp tech said:

"ACOUSTIC WAVEGUIDE TECHNOLOGY
Waveguides are similar to horns in that they change the air load that the driver "sees". The difference lies in the extremes. Our waveguides are characterized by the fact that they utilize relatively low compression ratios. At higher frequencies the air load is virtually the same as a standard "baffle" mounted driver. This eliminates the associated distortion that is common at high frequencies with traditional horns. The relatively shallow design does not provide the higher loading characteristics of traditional horns and hence, none of the distortion either. Acoustic loading is only increased at the lower frequency end of the drivers operating range. This provides added efficiency where it is needed most. The low frequency range of any dynamic driver is where its mechanical limits dominate and ultimately degrade its performance."

"Our waveguides should be considered as a hybrid union of technologies, bridging the gap between high efficiency and high linearity. Compression ratios are high enough to greatly extend the low frequency performance of the driver as well as ensure that its mechanical system remains within linear operational limits under all conditions and drive levels. At the same time, the decreasing compression ratio with increasing frequency characteristic guarantees avoidance of the non-linear regions of the air medium. Sort of the "best of both worlds" -- if you will."

The other advantage of a waveguide is the ability to control dispersion (steer sound away from space it is not wanted, or simply does not need to be). Constant directivity waveguides are designed to have the same frequency response whether on axis or off (alhough after a certain point the off axis response changes dramatically). The only thing that changes within the listening area is the actual intensity. I wouldn't bother with this aspect too much in a car, because I think chasing proper stereophonic reproduction is a waste of time. Only thing I am attempting is to turn some volatile reflections into the heart of a waveguide.

Arrays are the next frontier. Arrays basically takes everything you know about drivers and puts your knowledge to the test when you try to implement them. More on arrays later, it's a big one, and I need sleep.

Edited by br85, 08 September 2009 - 07:25 AM.

[bobo333]

a very informative and interesting thread, keep it up guys!

[~thematt~]

br85, on Sep 6 2009, 10:04 PM, said:

"Beaming" is an effect you've probably heard before. Ever walked past a loudspeaker and suddenly experienced pain or a lot of high frequency sound while you are right in front of it? That's beaming. The high frequency information is not able project itself out as much as the lower stuff can. Distortion sounds like anything that shouldn't be there, i.e. descriptions like "shrilly", "tinny", "muddy", "strained", "harsh" etc etc are all likely to be because a certain type of distortion is present.

You've got your terms mixed up there br85!

Beaming is the term given to 'torchlighting', as you previously mentioned. Its when the dispersion pattern (or power response) of a particular speaker results in the tightening of the projection of intensity. Basically, a large speaker will begin to narrow its dispersion as the frequencies increase. The high frequencies will beam more broadly when replayed from a smaller diameter speaker.

This is of large interest when crossing over two speakers of different sizes. Most people think that its all as simple as selecting a crossover point (which hopefully they're slowly accepting now, is not so true). If you cross a large midrange too high to a tweeter, the tweeter will have a great frequency response off-axis, but the midrange will be very poor off-axis in intensity. Given that we listen to our speakers off-axis, this should be given more thought!

The increase in high frequency information when you walk past a set of speakers has a lot more to do with the directivity pattern of the waveguides being used, as well as the amount of energy contained within the acoustic signature. Physics dictates that for the same intensity, there is less energy in higher frequencies. Therefore, these are dampened by the air quite a lot sooner, in fact its an exponential relationship. The less power, the easier to damp. Off axis signatures have little power, and are therefore damped very quickly.

Edited by ~thematt~, 07 September 2009 - 11:13 AM.