Why not 4th order crossovers?

[yepimonfire]

Figured the diy section was more appropriate for this topic. Many modern 2 way speaker designs employ either 2nd order hp/lp or 2nd order lp 3rd order hp (especially when tweeters are xovered at lower points).

Three issues commonly considered when choosing an xover point are keeping the woofer out of its breakup mode, keeping the tweeter from lower frequencies that would cause over excursion or run down into the fs of the driver, and matching directivity at the xover point for even off axis response.

4th order crossovers also have the benefit of a 360 degree phase shift, so there shouldn’t be any problems at the xover point between the woofer and tweeter. Wouldn’t a 4th order xover allow a lower xover point, better matching the tweeter and woofers dispersion, whilst keeping the tweeter from going down too low and keeping the woofer very far away from its breakup point, and avoid all of the problems associated with phase shift at the xover frequency?

Really the biggest issue is insertion loss, but power is cheap and plentiful.

Why aren’t more speakers designed with them?


Sent from my iPhone using Tapatalk

 

[TLS Guy]

Figured the diy section was more appropriate for this topic. Many modern 2 way speaker designs employ either 2nd order hp/lp or 2nd order lp 3rd order hp (especially when tweeters are xovered at lower points).

Three issues commonly considered when choosing an xover point are keeping the woofer out of its breakup mode, keeping the tweeter from lower frequencies that would cause over excursion or run down into the fs of the driver, and matching directivity at the xover point for even off axis response.

4th order crossovers also have the benefit of a 360 degree phase shift, so there shouldn’t be any problems at the xover point between the woofer and tweeter. Wouldn’t a 4th order xover allow a lower xover point, better matching the tweeter and woofers dispersion, whilst keeping the tweeter from going down too low and keeping the woofer very far away from its breakup point, and avoid all of the problems associated with phase shift at the xover frequency?

Really the biggest issue is insertion loss, but power is cheap and plentiful.

Why aren’t more speakers designed with them?


Sent from my iPhone using Tapatalk

Because it is not usually the best solution.

Again you are not understanding that time and phase shift are the same thing. Do not use 360 degree phase shift, express it as time shift which you should. You would not be so cavalier about it then.

But the essence of the problem is that all drivers have inherent roll offs. The trick is to design the crossover to mesh with the driver roll off and or peaks to get the flattest response and provide baffle step compensation.

You also have to pay attention to power to the driver at Fs, Power needs to be well down for mids and tweeters at Fs unless it is a very low Q driver.

I try to accomplish all of this while trying to minimize time shifts.

Finally fourth order electric crossovers are expensive because of a high component count. On the lower end the inductors become large. This means the use of expensive components, otherwise the DC resistance will be high. This not only increases insertion losses but changes the Q of the system. The other downside of the increased insertion loss if that you add to the problem of thermal dynamic compression.

We don't talk about dynamic thermal compression is speakers as much as we should. However it is a major factor impairing realistic reproduction from most speakers. It is pretty much NEVER specified and hard to measure. The skilled listener can easily learn to detect its presence. It is a major factor in usually more expensive and larger speakers being able to provide more accurate reproduction.

 

[everettT]

Here is a speaker that was designed for 2nd order and a 4th order. Note the graphs for comparison.

http://murphyblaster.com/content.php?f=MBOW1.html

TLS covered about all that is needed. But as a note, there more speakers than you think that use 4th order.

 

[TLS Guy]

Here is a speaker that was designed for 2nd order and a 4th order. Note the graphs for comparison.

http://murphyblaster.com/content.php?f=MBOW1.html

TLS covered about all that is needed. But as a note, there more speakers than you think that use 4th order.

But do they really? It is usual in the spec to sum the acoustic and electrical roll off. Unless the consumer can actually look at the crossover circuit, you have no idea what the electrical slopes are.

 

[everettT]

But do they really? It is usual in the spec to sum the acoustic and electrical roll off. Unless the consumer can actually look at the crossover circuit, you have no idea what the electrical slopes are.

I was going on what companies like Salk, Philharmonic, Linkwitz , Legacy etc. state either in literature, spec, or conversation. Was by no means stating that 4th order is highly common.

 

[yepimonfire]

@tlsguy
How big of a factor is power compression in home audio though? I’ve often wondered myself. Outside of somebody trying to fill a huge room with inefficient bookshelf speakers, a typical movie played back at 75dB average, could have peaks as high as 95dB per channel. With a typical 87dB 1w1m speaker at a distance of 10’, that’s less than 1w during quiet scenes and peaks of about 50w which could be sustained for 2-3 seconds.

How likely is it that’s enough power and enough time to heat a voice coil high enough to drop the spl by 3dB?

Lastly, what does power compression sound like?

 

[Swerd]

Three issues commonly considered when choosing an xover point are keeping the woofer out of its breakup mode, keeping the tweeter from lower frequencies that would cause over excursion or run down into the fs of the driver, and matching directivity at the xover point for even off axis response.

This is a good statement of the general principles involved. I often get irritated by you for stating things much too broadly while ignoring the details, but occasionally you surprise me. There’s hope for you . Of course, the devil is in the details.

In general, to perform well, a 2-way speaker with a 2nd order crossover (acoustic slopes roughly 12 dB/octave) must use drivers that have a flat frequency response and behave well for about 2 octaves above and below the crossover frequency. By “behave well” I mean woofers must stay below breakup mode and produce enough SPL roughly 30° off-axis compared to on-axis (within 3 dB), and tweeters must stay above low frequency distortion all within 2 octaves of the crossover frequency. A Linkwitz-Riley 4th order crossover (acoustic slopes roughly 24 dB/octave) requires that drivers must behave well within 1 octave of the crossover frequency. So, yes, if done right, a LR 4th order crossover does have advantages. For example, if you use a metal coned woofer with a loud breakup peak, a 4th order design can allow you to crossover that woofer at a higher frequency than a 2nd order crossover can. I think that may be why most 2nd order designs use doped paper and not metal woofers.

Many modern 2 way speaker designs employ either 2nd order hp/lp or 2nd order lp 3rd order hp (especially when tweeters are xovered at lower points).

When talking about the order of a crossover filter, it’s important to remember whether you are talking about electrical order (defined by design and arrangement of inductors and capacitors and measured by volts vs. frequency), or acoustic order (defined by the slopes of drivers combined with crossover filters and measured by SPL vs. frequency).

But the essence of the problem is that all drivers have inherent roll offs. The trick is to design the crossover to mesh with the driver roll off and or peaks to get the flattest response and provide baffle step compensation.

TLS Guy’s point about taking into account the inherent roll-off of drivers makes reference to the same thing as I did above.

A good design can achieve 4th order acoustic performance while using crossover filters with fewer parts and look like something less than 4th order electrical filters. The Dennis Murphy DIY MBOW1 speaker is a good example of this (@everettT thanks for pointing this out). It is important to remember that this can work if you select the proper drivers, as Dennis Murphy does, but it doesn’t always work for all drivers. Most of his designs, whether DIY, Salk, or Philharmonic Audio, make use of Linkwitz-Riley 4th order crossovers, where crossover and drivers achieve 4th order roll-off slopes as measured acoustically. FWIW, I listen to two sets of speakers both designed by Dennis Murphy, one with 2nd order crossovers and the other with LR 4th order. I can tell the difference between those two speakers, but I cannot say I can tell the difference because of the crossovers.

4th order crossovers also have the benefit of a 360 degree phase shift, so there shouldn’t be any problems at the xover point between the woofer and tweeter. Wouldn’t a 4th order xover allow a lower xover point, better matching the tweeter and woofers dispersion, whilst keeping the tweeter from going down too low and keeping the woofer very far away from its breakup point, and avoid all of the problems associated with phase shift at the xover frequency?

Again you are not understanding that time and phase shift are the same thing. Do not use 360 degree phase shift, express it as time shift which you should. You would not be so cavalier about it then.

I try to accomplish all of this while trying to minimize time shifts.

To directly answer YIOF, yes, 4th order (electrical) crossover do have a 360° phase shift. (2nd order electrical crossovers have a 180° phase shift. This can easily be corrected by wiring the tweeter or the woofer in opposing polarity.)

In a 2-way speaker, the woofer and tweeter would be in phase with each other (360° out of phase = 0° out of phase = in phase). For the same reason, they would be one cycle off with regard to time.

In the past, speaker junkies loudly debated the merits of speakers that were ‘time and phase correct’. It never was clearly decided because no one could scientifically establish whether listeners could hear a difference between time correct and time incorrect speakers. Some other individuals maintained they could clearly hear a difference between them. Although this debate was never decided, it, fortunately, has grown quiet.

One of the major problems with building ‘time correct’ speakers is the expense and effort of building the physical offset between drivers so their acoustic origins are the same distance from a listener. This however ignores the real problem that even speakers with such physical alignment will still be ‘time correct’ over a narrow listening distance and angle, and over a narrow range of frequencies. At roughly 2000 Hz, one wavelength is under 2 inches. If a listener moved just a bit, all efforts at correct time alignment go out the window.

 

[Irvrobinson]

@tlsguy
How big of a factor is power compression in home audio though? I’ve often wondered myself. Outside of somebody trying to fill a huge room with inefficient bookshelf speakers, a typical movie played back at 75dB average, could have peaks as high as 95dB per channel. With a typical 87dB 1w1m speaker at a distance of 10’, that’s less than 1w during quiet scenes and peaks of about 50w which could be sustained for 2-3 seconds.

How likely is it that’s enough power and enough time to heat a voice coil high enough to drop the spl by 3dB?

Lastly, what does power compression sound like?

Mark said thermal compression (often called power compression) was a factor in "realistic reproduction", which often is quite different from the modest levels most people use. For example, a saxophone or a trumpet, reproduced realistically, is very loud. And unless you play them back at a realistic loudness they literally sound different. For a trumpet during thermal compression, it loses its "bite" - the highest frequencies are rolled off compared to the real thing, while the lower frequencies aren't (or to a lesser degree). At an audio show years and years ago a speaker designer (he was at ADS at the time) said that it was often due to voice coil conductor heating in cheaply-designed tweeters causing an increase in the driver's resistance. ADS was using all custom drivers back then, and as I remember the discussion it was to highlight the design considerations of their supposedly well-designed custom tweeters, as opposed to cheaper and more common OEM designs. I also suspect that power compression is also one reason why some speakers seem to cause more listening fatigue in extended sessions than other speakers do.

 

[shadyJ]

Mark said thermal compression (often called power compression) was a factor in "realistic reproduction", which often is quite different than the modest levels most people use. For example, a saxophone or a trumpet, reproduced realistically, is very loud. And unless you play them back at a realistic loudness they literally sound different. For a trumpet during thermal compression, it loses its "bite" - the highest frequencies are rolled off compared to the real thing, while the lower frequencies aren't (or to a lesser degree). At an audio show years and years ago a speaker designer (he was at ADS at the time) said that it was often due to voice coil conductor heating in cheaply-designed tweeters causing an increase in the driver's resistance. ADS was using all custom drivers back then, and as I remember the discussion it was to highlight the design considerations of their supposedly well-designed custom tweeters, as opposed to cheaper and more common OEM designs. I also suspect that power compression is also one reason why some speakers seem to cause more listening fatigue in extended sessions than other speakers do.

Some that we can, and should, do to combat thermal compression is to switch to all plasma drivers. The ionized gas of the plasma medium is much less subject to changes in resistance than a copper coil, in fact it depends on the electrical resistance of air. This way we can listen at realistic levels always without fear that amplitude levels might differ slightly from that of the recording. It can be argued that no solid matter can ever produce truly high-fidelity sound. In fact, I scoff at your solid matter transducers, and I very much doubt that you have ever bore witness to a recording as it was intended to be heard! The toxic gas emission and high-voltage discharge hazards are a small price to pay for excellent sound, and I will never again subject my ears to the excruciating dullness of solid matter transducers that passes for loudspeakers in your pedestrian audio world!

 

[Pogre]

Some that we can, and should, do to combat thermal compression is to switch to all plasma drivers. The ionized gas of the plasma medium is much less subject to changes in resistance than a copper coil, in fact it depends on the electrical resistance of air. This way we can listen at realistic levels always without fear that amplitude levels might differ slightly from that of the recording. It can be argued that no solid matter can ever produce truly high-fidelity sound. In fact, I scoff at your solid matter transducers, and I very much doubt that you have ever bore witness to a recording as it was intended to be heard! The toxic gas emission and high-voltage discharge hazards are a small price to pay for excellent sound, and I will never again subject my ears to the excruciating dullness of solid matter transducers that passes for loudspeakers in your pedestrian audio world!

Ha! Don't forget your tank of helium!

I posted about plasma drivers in another thread not too long ago! I've heard them and they really do sound amazing, but they're SO impractical (as you comically pointed out. ). When speaking with Dr Hill about his plasma speakers he said the same thing you did about solid matter transducers. The plasma driver has no moving parts and is technically a perfect driver for sound reproduction.

 

[Swerd]

At an audio show years and years ago a speaker designer (he was at ADS at the time) said that it (thermal compression) was often due to voice coil conductor heating in cheaply-designed tweeters causing an increase in the driver's resistance.

On a more practical level than using plasma drivers, what can a speaker builder do to minimize or alleviate thermal compression?

• If cheap tweeters are the problem, would it be better to use tweeters with more robust voice coils? Does a tweeter with a low Fs, such as 500 Hz instead of 1500 Hz, mean that the voice coil/motor assembly is more robust and therefore less prone to thermal compression?
  
  
• Do ribbon or planar tweeters avoid this problem? They lack voice coils as seen in typical dome tweeters – their diaphragm also functions as the voice coil.
  
  
• If thermal compression varies with insertion loss, would it help to use inducters of lower DCR (larger gauge wire or iron/steel core)? Usually, this seems to be done only for large inducters for a woofer network. If the ADS guy was right, this may not matter.
  
  
• Would using an active crossover avoid thermal compression? Or is it still a problem due to driver voice coils?

 

[Irvrobinson]

Some that we can, and should, do to combat thermal compression is to switch to all plasma drivers. The ionized gas of the plasma medium is much less subject to changes in resistance than a copper coil, in fact it depends on the electrical resistance of air. This way we can listen at realistic levels always without fear that amplitude levels might differ slightly from that of the recording. It can be argued that no solid matter can ever produce truly high-fidelity sound. In fact, I scoff at your solid matter transducers, and I very much doubt that you have ever bore witness to a recording as it was intended to be heard! The toxic gas emission and high-voltage discharge hazards are a small price to pay for excellent sound, and I will never again subject my ears to the excruciating dullness of solid matter transducers that passes for loudspeakers in your pedestrian audio world!

I'll send Gene a private message. Someone obviously stole James' account and is posting while stoned.

 

[Irvrobinson]

On a more practical level than using plasma drivers, what can a speaker builder do to minimize or alleviate thermal compression?

• If cheap tweeters are the problem, would it be better to use tweeters with more robust voice coils? Does a tweeter with a low Fs, such as 500 Hz instead of 1500 Hz, mean that the voice coil/motor assembly is more robust and therefore less prone to thermal compression?
  
  
• Do ribbon or planar tweeters avoid this problem? They lack voice coils as seen in typical dome tweeters – their diaphragm also functions as the voice coil.
  
  
• If thermal compression varies with insertion loss, would it help to use inducters of lower DCR (larger gauge wire or iron/steel core)? Usually, this seems to be done only for large inducters for a woofer network. If the ADS guy was right, this may not matter.
  
  
• Would using an active crossover avoid thermal compression? Or is it still a problem due to driver voice coils?

I'm not technical enough with drivers and crossovers to have expert answers to any of these questions, but if you do any live recording with acoustic instruments, the first thing that hits you is how loud some of them are in small venues (like almost any home), and how few audio systems sound accurate at realistic volumes. I've spent a lot of money on audio equipment, but it can't accurately reproduce my wife's large "rock band" drum kit in a residential room. It does pretty well on her much smaller "jazz" kit, and jazz doesn't typically involve rim shots, which are very difficult for realism. A flute sounds absolutely real, but a sax doesn't always. I can't be in the same room with an aggressively played trumpet; I wonder if all trumpet players go deaf. ;-) Our upright piano sounds nearly realistic (I think I'm limited by my mikes). A violin is surprisingly difficult to get right close up (and surprisingly loud when played aggressively). I have no idea what's compressing or distorting, but I'd bet on the tweeters. (The Salon2s have heroic-looking passive crossovers, which tend to impress me as a non-expert, I'd guess, but it is difficult to believe anything so impressive overloads before the tweeter and midrange units.)

IMHO, until you've recorded acoustic instruments in your own home or studio you really don't have a good idea of what accuracy is. (And you need decent mikes that can handle the loudness too.)

In my experience ribbons are not the answer, but I'm anxious to hear the RAALs Dennis uses some day. I still haven't had the opportunity.

 

[Swerd]

I'm not technical enough with drivers and crossovers to have expert answers to any of these questions…

Neither am I, but I can always ask.

I do know enough to realize there usually are two opposite considerations involved. Should a voice coil/motor be heavy, robust enough to avoid overheating? Or should it be as light as possible to allow to more rapid starts and stops in movement?

Those heroic looking passive crossovers you described are analog electronic filters, which at audio frequencies do tend to get large. In a 3-way speaker, some parts need to be large because of the low frequencies involved. All those parts must handle voltages some what higher than in typical electronic circuits. Capacitors rated to withstand 50 or 100 volts DC are smaller than caps built to handle 250 or 400 volts. And I bet Revel didn't skimp on the wire gauges of the inductors. So the two opposite considerations are low cost vs. insertion loss & voltage handling.

I know nothing about live recording.

It's a good thing the trumpet player is behind the trumpet's bell and not in front . It's all those other musicians in front of the trumpets that should worry. Rock drummers, unless they religiously wear ear protection, seem to go deaf before trumpet players.

You are always welcome to come hear my speakers. You only have to travel from California to DC. Many of the information technology nouveau riche (such as Jeff Bezos) come to DC to tell us they have all the answers to our country's problems . You could charge the travel as a business expense .

 

[shadyJ]

On a more practical level than using plasma drivers, what can a speaker builder do to minimize or alleviate thermal compression?

• If cheap tweeters are the problem, would it be better to use tweeters with more robust voice coils? Does a tweeter with a low Fs, such as 500 Hz instead of 1500 Hz, mean that the voice coil/motor assembly is more robust and therefore less prone to thermal compression?
  
  
• Do ribbon or planar tweeters avoid this problem? They lack voice coils as seen in typical dome tweeters – their diaphragm also functions as the voice coil.
  
  
• If thermal compression varies with insertion loss, would it help to use inducters of lower DCR (larger gauge wire or iron/steel core)? Usually, this seems to be done only for large inducters for a woofer network. If the ADS guy was right, this may not matter.
  
  
• Would using an active crossover avoid thermal compression? Or is it still a problem due to driver voice coils?

Seems to me like most of the problems would occur in the coil, since the magnetic field strength reduction is the root cause. Most of the progress made in this field is done primarily for commercial sound or public address systems like Cinemas, concerts, live sound reinforcement, etc, since those speakers are driven so hard for so long. JBL has some interesting application notes, here and here. Stereophile has an interesting article here. If you want to get your hardcore nerd on, Dougles Button of JBL published one of the major papers in the JAES in the mid 80's that can be read here.

From what engineers have told me, one thing to do is use ferrite magnets instead of Neodymium. I guess ferrite iron cools better due to more surface are. Also use aluminium formers attached to aluminium cones. the formers and cones can act as heat conductors that use the large surface of the cone to radiate heat. Vented backplates or vented baskets, and controlling the air flow therein. I have read that the best thing to do is use one or the other but not both. I have also read that smaller vents on backplates can be more effective than larger vents. Ferrofluid is used in tweeters for its thermal dissipation. You can use a lot of coil in order to handle lots of current, but that presents its own set of problems, like lower sensitivity due to the added mass not to mention greater inductance which further impacts sensitivity and linearity. I would guess that JBL's differential drive helps reduce thermal compression a lot.

I wonder how much cooling technology for CPUs can be used for things like compression drivers? The lengths people have gone to in order to keep their CPUs cool is pretty amazing.

 

[Irvrobinson]

Those heroic looking passive crossovers you described are analog electronic filters, which at audio frequencies do tend to get large. In a 3-way speaker, some parts need to be large because of the low frequencies involved. All those parts must handle voltages some what higher than in typical electronic circuits. Capacitors rated to withstand 50 or 100 volts DC are smaller than caps built to handle 250 or 400 volts. And I bet Revel didn't skimp on the wire gauges of the inductors. So the two opposite considerations are low cost vs. insertion loss & voltage handling.

That much I know. I also know that Mark despises the Salon2 crossover design because (like some other high-end towers) it has a low frequency (in this case, 120Hz) low-pass filter.

I know nothing about live recording.

I strongly recommend ordering a low-cost hand-held digital recorder, and trying a few recordings, even of non-musical sounds, yourself. In fact, next time you're at Dennis's place have him drag out his fiddle and play a lively tune for you, recording it from about 3-6 feet away. I suspect you'll find it an enlightening experience at playback time.

 

[Swerd]

That much I know. I also know that Mark despises the Salon2 crossover design because (like some other high-end towers) it has a low frequency (in this case, 120Hz) low-pass filter.

FWIW, Dennis also says he avoids a woofer-to-mid crossover below roughly 400 Hz. The much larger size of the caps & coils involved for the woofer network significantly add to the cost. He also said, without specifying, it introduces problems that he'd rather avoid.

 

[Irvrobinson]

FWIW, Dennis also says he avoids a woofer-to-mid crossover below roughly 400 Hz. The much larger size of the caps & coils involved for the woofer network significantly add to the cost. He also said, without specifying, it introduces problems that he'd rather avoid.

I get that, but two of my all-time favorite speakers, the Salon2 and the KEF 207/2, both use 120Hz low-pass frequencies. Perhaps I just like the sound of problematic designs.

 

[ski2xblack]

On a more practical level than using plasma drivers, what can a speaker builder do to minimize or alleviate thermal compression?

• If cheap tweeters are the problem, would it be better to use tweeters with more robust voice coils? Does a tweeter with a low Fs, such as 500 Hz instead of 1500 Hz, mean that the voice coil/motor assembly is more robust and therefore less prone to thermal compression?
  
  
• Do ribbon or planar tweeters avoid this problem? They lack voice coils as seen in typical dome tweeters – their diaphragm also functions as the voice coil.
  
  
• If thermal compression varies with insertion loss, would it help to use inducters of lower DCR (larger gauge wire or iron/steel core)? Usually, this seems to be done only for large inducters for a woofer network. If the ADS guy was right, this may not matter.
  
  
• Would using an active crossover avoid thermal compression? Or is it still a problem due to driver voice coils?

One method to address thermal compression is to employ a transconductance amp for the tweet. It's kind of like placing the speaker load in the feedback loop of the amp...or something. (I'm educated in the lowly biological sciences, sorry.) Anywho, the source impedance is far, far higher than the driver load in this case. This arrangement is apparently immune to thermal compression. It also requires upstream correction for the impedance related nonlinearities inherent to the true-current-source approach, so it does necessarily add complexity. It's used by at least one brand who sells active studio monitors from across the pond.

 

[Swerd]

One method to address thermal compression is to employ a transconductance amp for the tweet. It's kind of like placing the speaker load in the feedback loop of the amp...or something. Anywho, the source impedance is far, far higher than the driver load in this case. This arrangement is apparently immune to thermal compression. It also requires upstream correction for the impedance related nonlinearities inherent to the true-current-source approach, so it does necessarily add complexity. It's used by at least one brand who sells active studio monitors from across the pond.

Thanks. I never heard of a transconductance amp until now. That's something new to learn.

I'm educated in the lowly biological sciences…

So am I, PhD in Biochemistry with a career in cancer research. Never call the biological sciences lowly. It is so complex that it is only just now being understood at a high level.

What I've learned about speakers and what not were strictly self-taught at the hobby level. There's a lot of electronics I don't pretend to understand. For that matter, some other highly regarded AH posters are also educated in other similarly low sciences. Dennis Murphy has a PhD in Economics and TLS Guy is a physician.