Sticky Thread Proposal: Xover Design, Voicing, the magic of speaker design

[dlneubec]

I have been visiting this site, among others, for the last year or so, since I first got into DIY Audio. One thing I see lacking is advice/teaching in regards to the critical aspects of crossover design and vocing when doing your own DIY design, not just building others designs.

I know there are a number of folks out there who have a lot to share with the new (or not so new) to the craft and have spent years honing their knowledge. Wouldn’t it be great if more of that knowledge could be passed along on a regular basis or maybe some rules of thumb or tips and tricks that would be beneficial?

I’m doing my first complete DIY design (see the omnidirectional thread) and I have and continue to have lots of questions in regard to crossover design and voicing. I’ve read Testing Loudspeakers and the Loudspeaker Design Cookbook, but the leave me with many questions.

For example, what FR shapes do the magic makers prefer, rising top end, saddle shaped with the mids a bit lower than the highs and lows, ruler flat, etc. and why?

What is the best shape for music, or particular kinds of music, or for HT., etc.?

What crossover slopes do you prefer and why?

When voicing, how do you proceed? Do you listen to a variety of music? Do you listen to the same music over and over to be sure you understand changes you’re trying? Do you concentrate on a particular kind of music, i.e. female vocals, or instrumentals, etc.? If you have a 2-way that crosses to actively controlled stereo subs at 100hz, do you listen to just the two-way when voicing or include the sub? Do you listen to one speaker, one channel only or must two prototypes be built?

I have noticed that certain vocal recordings, especially some female, can get a “hot” sound, particulary at higher volumes. What is the source of this? Is it the recording, the tweeter being overdriven, tweeter crossed too low, not enough attenuation of woofer breakup or on the tweeter circuit, etc.?

It seems to me that many of these questions are asked and some answers passed along, but they are often deep in somewhat unrelated threads that everyone who might benefit is not following. Does anyone else think a sticky thread where one could get advice/help on these types of issues be beneficial to the DIY community in general? Does anyone have other questions along these lines they would like some feedback on?

[JonW]

Hi Dan-

Good idea. I'm thinking about doing what you're doing- trying to design my own speaker for the first time. And, indeed, I have many newbie questions. I just started a related thread:

http://www.htguide.com/forum/showthread.php4?t=23688

So yeah, I certainly could benefit from such a collection of knowledge.
(Even better would be some help in trying to talk myself out of such an endeavor. )

[dlneubec]

Well, I guess that was a dud of an idea! :tired:

[JonW]

Well, I guess that was a dud of an idea! :tired:

I think your idea stinks, man.
:P

[cjd]

You're asking for years and years of learning and experience to be boiled down into a few simple posts!

:P

I think a lot of the information you're asking for IS here, but scattered about. And the scattered nature isn't necessarily a bad thing - you get to see more of the process, the wandering mental illness that drives the end results...

Also, for every designer you're likely to find a completely different set of answers and they'll sometimes be at odds.

C

[JonMarsh]

You're asking for years and years of learning and experience to be boiled down into a few simple posts!

:P

I think a lot of the information you're asking for IS here, but scattered about. And the scattered nature isn't necessarily a bad thing - you get to see more of the process, the wandering mental illness that drives the end results...

Also, for every designer you're likely to find a completely different set of answers and they'll sometimes be at odds.

C

What? You an me at odds? Never!

Now, Evil Twin, that's another matter....

[cjd]

What? You an me at odds? Never!

Now, Evil Twin, that's another matter....

The irony is that philosophically, it's not all that likely. Though Evil Twin is, well... evil.

It is far more likely that any issue you might take with something I do is either: I still have learning to do OR I'm not building something I agree with but for whatever reason fits a different set of requirements. So maybe I'm not as set in my ways just yet.

I'm also still married... and hope to keep it that way. Have a few years difference perhaps, also.

I'm trying to keep up on having (what's left of) my hair the crazy scientist type.

So, just a few of the tricks of the trade are already showing up in this thread!

C

[dlneubec]

You're asking for years and years of learning and experience to be boiled down into a few simple posts!

:P

I think a lot of the information you're asking for IS here, but scattered about. And the scattered nature isn't necessarily a bad thing - you get to see more of the process, the wandering mental illness that drives the end results...

Also, for every designer you're likely to find a completely different set of answers and they'll sometimes be at odds.

C

That's not quite what my intention is. My thinking is that it might be helpful to see these types of questions and their answers in one location/thread. That way, if someone has a specific crossover, voicing, etc. question, they could post it here and know that it has a shot at getting some attention and help others. First, if it is buried in a long thread, it might not be noticed by those with crossover expereince to share, since they may not be reading that thread, and also, when an answer is given, those who are interested in this subject can be sure to see it, rather than missing it deep in another, thread. Also, what about new members who have not been following any threads?

I realize there will be some disagreements, but that is how we all learn, by discovering our assumptions may not be good ones and having them challenged.

FWIW, I think it just has more potential as a teaching and learning tool than the distributed answers available now.

One of the main benefits of this forum over say Madisound or PE, is that threads can carry on forever, so anyone who comes along is not as likely to miss a great educational opportunity.

[dlneubec]

I'll take another shot at getting this going and expose myself to ridicule.

Attached below is my newest, yet to be tried crossover design for my omni project.

Omni-directional project

The omni (boy do I need a name for it) is sort of a WMTM with a 12" downfiring sub on the bottom, 2-RS180's, one upfiring and one downfiring, with an RS28AS situated vertically bewteen them, but front firing. Below is a photo of it with a veneer mock up of what it will look like finished. The RS180's are in sealed 7.5l enclosures and will cross at 100Hz to the sealed sub, which is controlled by an active analog circuit for Q.5 at 20Hz (designed for my NaO Mini's by John Kreskovsky).

The first graph shown is actually the 2 RS180 raw responses and also the RS28AS response all mounted in their final box positions for measurements (with felt added to the RS28 baffle all around). Notice that the performance of the RS180 is quite different from usual, since it is measured at 90º off axis do to the upfiring/downfiring arrangement. They have -6db added to the plot as compared to the RS28, so they will sum properly in SoundEasy.

The second graph shows my newest crossover design. I have borrowed from the Modula MTM for the RS180 section. The tweeter has its polarity reversed. The RS28 may need a little more attenuation, but since it's often described as lacking a bit in the top end, I'm starting with it up a little bit.

I'd love to have some critiques of this, my first crossover design. Perhaps other noobs will learn from my mistakes.

Fire Away :a>

[JonMarsh]

That's not quite what my intention is. My thinking is that it might be helpful to see these types of questions and their answers in one location/thread. That way, if someone has a specific crossover, voicing, etc. question, they could post it here and know that it has a shot at getting some attention and help others. First, if it is buried in a long thread, it might not be noticed by those with crossover expereince to share, since they may not be reading that thread, and also, when an answer is given, those who are interested in this subject can be sure to see it, rather than missing it deep in another, thread. Also, what about new members who have not been following any threads?

I realize there will be some disagreements, but that is how we all learn, by discovering our assumptions may not be good ones and having them challenged.

FWIW, I think it just has more potential as a teaching and learning tool than the distributed answers available now.

One of the main benefits of this forum over say Madisound or PE, is that threads can carry on forever, so anyone who comes along is not as likely to miss a great educational opportunity.

Good idea for a thread in concept, but the devil is in the details, and for the most part I have to agree with Chris.

A VERY large per centage of everything on this forum section is about this topic, apart from the contruction details and finishing discussionis.

EVERY decision made in the design process affects the voicing of the system- including where the speaker will be located in the room, what configuration of enclosure or no enclosure, etc. Condensing this into one thread would make for a very, very large thread, and a lot of work to boot. That's why we have all the threads archived back to day one, as well as have the Missions Complete section.

Sometime I hear people talking about voicing as if it's just some polish applied in the final stages of the crossover desgin. And while this does occur, by the time you've gotten to that point, you've made a host of decisions that already have a profound effect on the perceived tonality, character, room interaction, etc. While that doesn't mean you shouldn't fine tune the crossover in the last stages, it also means that at that point it's too late to change the loudspeaker system's character fundamentally.

What are the basics in voicing? In my opinion, the most fundamental aspects in the design are considering how the implementation is going to affect the ratio of axial response to power response (direct sound to room sound), and where the speaker will be located in the room. Additional fundamental configuration choices are such as those between a point source and line array, between a monopole and a dipole, and between a flush mounted wall system vs. a speaker sitting out in the room away from the wall. (i.e., one that can be taken somewhere else).

Within those areas, there are subcategories. I'll VERY briefly discuss some thoughts on the major categories and subcategories, keeping in mind that this is my perspective and your milage and taste may vary...


First choice:
Built in or free standing

In the 70s and 80s I helped design and build some small studios, and became fairly conversent with the issues surrounding recording monitors. A well designed rerding monitor is usally built in flush to a wall, and as such does not have baffle step compenstion. It's axial response and power response, therefore, are fairly similar, within the dispersion properties of the midrange and high frequency drivers. And this can produce a well balanced sound in the critical recording area, especially with a good LEDE impelmentation, though some care has to be taken in monitoring due to combfiltering from the mixing board (near level path difference). that's one of many reasons that some small nearfield monitors are often used at the board also, though many popular ones these days use waveguides with controlled dispersion.

When you have a box speaker out away from the wall (not built in flush), then things get more complicated, because the speaker isn't radiating into the same directional volume at all frequencies. At moderately low frequencies, it acts like it's radiating into a full circular volume (4pi). At very low frequencies, it looks like 2pi again. At some upper range, it transistions back to an effect based on boundary loading and wavelengths that looks like 2 pi. This variation affects how the room farfield response balances out, and the relatively omnidirectional LF radiation going backwards then forwards increases frequency response irregularities and room interaction.

Now, let's consider the size and shape of that monopole speaker sitting out in the room.

At some frequency range related to the baffle size and shape it gradually transistions from 4pi to 2pi radiation for the bass driver. If the baffle is shaped like a circle, this transistion can be rather abrupt. Depending on the size of the baffle, and location of drivers, the midrange or tweeter may be involved in baffle dependent changes in radiation pattern. This may be a conscious choice (Avalon), or an accident/byproduct of the design. Non the less, it's a voicing influence, whether or not consciously made.

Let me illustrate. Three three way speakers. Two built in a tower configuration. Drivers mounted on front. Baffle 14" wide, to accomodate 12' woofer, 7" midwoofer, and tweeter. Say the tower is 36-40" tall, exact height doesn't matter. There wil be a baffle step transision approximately starting at 700 Hz from 2pi radiation to 4pi at approximately 250 Hz. This affects response of midrange and tweeter as well as woofer. Let's say you crossover midrange at 400 Hz, tweeter at 2.5 kHz. Woofer crossover will have to have BSC comp, so will midrange. System will start to transistion from 360 degree horizontal radiation to about 180 degree radiation between 250 Hz and 700 Hz. Crossover design will have to reflect that. no biggie. Midwoofer will need contouring as well, which makes the midrange crossover more complicated. Can try to ignore that, but this will result in a dip in the 375Hz to 700 Hz region. Not an uncommon thing in some three ways. From ~800 Hz and up, radiation will be 2pi, so there will be a bit of a drop in far field reverberent power response, becuase the room is energized in all directions in the lower range, (good or bad), and only in one direction above ~700 Hz. Though the ear weights early arrivals highly (direct sound), it's also sensitve to indirect sound, and does tell the difference in ambient and lower level cues when tonality is changed. Also, all of this off axis radiation tends to mask the acoustic of the actual recording.

Conventional speakrs often fight this by elevating some of the upper end so that the presense and high end range recording acoustic cues are more prominent. Sort of like turning up the edge enhancement in your TV. Seems to help solve a problem at first, but it's still a coloration. But this is a common "voicing" step.

Speaker number two, also monopole box, built with a fixed width for woofer, but a tapering baffle width for midrange and tweeter. What does this do? It changes the baffle step frequency for the midrange and tweeter driver, which increases the off axis SPL, and changes the shape of the room reverberent response curve. More midrange and presence range info in the reverberent decay. On axis doesn't need to be voiced as hot for the same room reverberent response curve. Typical example: Avalon Eidolon, Isis, Opus. Still has a tapering room curve, with a lot of off axis room excitation, but in a larger room may reveal more of the recording acoustic and have a flatter room power curve.

Speaker number three. Tapered diople baffle. Radiation is figure 8 at most frequencies through upper midrange frequencies, side axis null on both sides and top, baffle shape and drivers chosen to try to maintain the same directivity with rising fequency. Waveguide to help control tweeter directivity flare at lower treble frequencies. Far field reverberent response is very similar to on axis response, and is at lower level due to cancellation at sides and top. More opportunity for recording acoustic to be heard compared with playback room acoustic, because the latter is at a lower level, and has less masking in the bass frequencies.

This is voicing through fundamental design concept. Once your'e past this point, the major fundamentals have been chosen, and though driver selection, crossover integration, response balance, BBC dip for commercial programming, etc, all have their role, they can't materially change the basic voicing set by the design concept.

Just my 0.02. YMMV.

Regards,

Jon

[dlneubec]

Hi Jon,

Thanks for taking the time for a detailed response.

What it reminds me of is that any kind of design is essentially problem solving. A series of decisions are made, attempting to reach a goal. There are innumerable paths to reach the goal, but every decision made along the way, especially those made early in the planning and design process, effect the rest of the path and both remove and make possible opportunities.

I'm a Landscape Architect by profession, so I understand the design process quite well, however, I never thought to apply it for speaker design! Thanks for reminding me of the similarities. It gives me a fresh insite into how to move forward with any future designs. Of course, at this point in my speaker design knwolege, I'm in the unconscious incompetant stage. That is, I don't know what I don't know. ops:

That's part of what I was hoping this thread would do. Open the eyes of the novices like me out there and let them begin to discover some of the things they don't know!

One question I have. What do you mean by "tweeter directivity flare at lower treble frequencies"?

[chasw98]

Let me ask a really basic question here, if I may. I have been talking with JonW about his project of wanting to design a 2 way from scratch. He has asked what he might want to buy in terms of "tools of the trade". I have told him about LSPcad and that a computer to run and analyze it is helpful (duh!), but I also told him that it might be helpful to have a DCX2496 digital crossover to "emulate" a design on paper and to test with. I have also heard that you can emulate a crossover design using LSPcad software I believe. I have seen JonMarsh reference the use of "mules" for testing various drivers to see how they work and interact with each other. Now it doesn't seem practical to have cabinets around ranging from .5 cu feet to 2 cubic feet and baffles premade to accomadate everything from 5 inch to 8 inch drivers and associated tweeters.
Just what kind of hardware setup would you use to test drivers where you can mix and match, take measurements, listen to various combinations while making decisions on a design. Or do you have to have a bunch of cabinets, a bunch of baffles, and a bunch of drivers to do some preliminary testing with?

Chuck

[ThomasW]

Personally I think JonW. needs to learn to walk, before he plans his mission to Mars.... :W

What people are seeing when JonMarsh creates a design is almost 40yrs of experience, combined with modern tools of the trade. Buying the tools and not having the knowledge to use them is only going to confuse....the DCX2496 is a prime example. Have JonW download the DCX users manual and give it a read... :B

Most often test mules are built, used for measurements, then tossed. If the decision is a full court press (no test mule), a finished baffle is fabricated, then the measurements are made. This is only after a ton of software sims are done.

For measurements currently Jon's using Praxis, prior to that he used Clio/ClioWin

Freeware software tools like 'The Edge' and 'Baffle Diffraction Simulator' are very helpful with simulations regarding baffle shapes and driver placement.

There are lots of drivers in my basement and Jon's storage area that where purchased, tested and cast aside because better drivers were found.

[JonP]

A few comments... and Thomas beat me to a couple of them.

SoundEasy can emulate a crossover as a digital filter, also. It also can download the filter curve into a DCX2496. But, I'd agree with getting into that after some basic learning

I was going to mention modeling with FRC or other tools like Tom mentioned, though at some point building a box (preliminary mule or not) and dropping drivers in it should be done. Though usually, you've already predicted the overall result fairly closely with the tools.

Something I'm trying to do, is create a multi purpose test box. Current incarnation is a tall rectangular tower, but with a movable inner brace/panel to change volume. One could come up with a set of panel sizes, meeting several common widths, and depths, and join them using screws, foam tape to airtight joints, etc.. and panels for different boxes. A PE style removable front baffle would allow for easily putting different drivers in the same test box.

OTOH, you are working on one design at a time, simplest would be to build your test boxes as above, and save the pieces to possibly mix and match later.

[warnerwh]

A few years ago I decided to build myself a world class speaker. After doing considerable research going to the PE board among others and reading the Loudspeaker design cookbook I learned to not even try it.

The best I can tell is that even very experienced speaker designers often have to go through alot of R&D on the crossover. That means lots of time and money. These guys know what they're doing too.

I ended up buying speakers instead. I know enough now that tells me you're asking for too much. I think the best bet would be to buy a digital crossover that can be manipulated in numerous different configurations. It will be cheaper in the long run. You obviously also need alot more amplifiers but if I were to try to design my own speakers I'd not even consider trying it any other way.

I recently read that a commercial speaker designer can easily go through a dozen or more different designs of the crossover before they've got it right. So I'd rather buy or copy another design that's proven because I could spend alot of money and end up with a pile of nice parts that sound like crap.

I commend your effort however.

[JonMarsh]

The irony is that philosophically, it's not all that likely. Though Evil Twin is, well... evil.

It is far more likely that any issue you might take with something I do is either: I still have learning to do OR I'm not building something I agree with but for whatever reason fits a different set of requirements. So maybe I'm not as set in my ways just yet.

I'm also still married... and hope to keep it that way. Have a few years difference perhaps, also.

I'm trying to keep up on having (what's left of) my hair the crazy scientist type.

So, just a few of the tricks of the trade are already showing up in this thread!

C

No arguements or disagrement with ANY of that! :B

Staying happliy married is a good thing, and to be given precedence if possible over many other goals in life... (BTW, I was a psych major in school). While being a "failure" in marriage (other than a wonderful daughter and relationship with her) has few upsides, freedom to create speaker systems and enjoy them is one of them.

There are many cases in which pragmatic expeiencey dictates the basics and you just have to work with things from that point on... your skills and ability to strike a chord with the DIY guys are very obvious, Chris- just look at the length of the Dayton RS WMTMW & TMWW thread!

Looking forward to see what evolves next year!

Happy Holidays!


~Jon

[JonMarsh]

Let me ask a really basic question here, if I may. I have been talking with JonW about his project of wanting to design a 2 way from scratch. He has asked what he might want to buy in terms of "tools of the trade". I have told him about LSPcad and that a computer to run and analyze it is helpful (duh!), but I also told him that it might be helpful to have a DCX2496 digital crossover to "emulate" a design on paper and to test with. I have also heard that you can emulate a crossover design using LSPcad software I believe. I have seen JonMarsh reference the use of "mules" for testing various drivers to see how they work and interact with each other. Now it doesn't seem practical to have cabinets around ranging from .5 cu feet to 2 cubic feet and baffles premade to accomadate everything from 5 inch to 8 inch drivers and associated tweeters.
Just what kind of hardware setup would you use to test drivers where you can mix and match, take measurements, listen to various combinations while making decisions on a design. Or do you have to have a bunch of cabinets, a bunch of baffles, and a bunch of drivers to do some preliminary testing with?

Chuck

As Thomas hinted at, by normal DIY or personal hobby standards I am way off the beaten track (polite version), or totally insane (normal middle class version).

Yes, I DO have baffles around pre-cut to handle all kinds of size drivers, at least the kinds I'm interseted in. Last year I built a special subwoofer/woofer test stand for 10's and 12's, tested a lot of drivers, then had to can it because I was running out of space. I have Woodstyle and Dayton cabinets around, but most testing get's done on large baffles with near field measurements to simulate classic IEC measurements; from there it's baffle simulation time, etc, to work out the speaker system parameters. There are good reasons my dipoles don't look like Orions, for example.

The way I approach this hobby is too much like I was actually in the business of devleoping speaker systems for a living; I was very close to doing that at one time (late 70's) in partnership with another guy, and the investment has only increased dramatically since then.

It would be an interesting psych case book study to find out what makes Jon (me) tick in this case, just why I have found this such an important creative outlet. You know, painters (artists) don't stop painting just because they've filled up a couple of walls in their home...

My ex wife couldn't stand my intersest in this area (in fact, she's partially responsible for the very existence of Avalon Acoustics (butterfly effect), as she insisted I give away the last/most high end prototypes I had at home because she couldn't stand their size.) In 1983 they went to my friend Charles Hansen, currently VP Engineering of Ayre Acoustics, founder of Avalon Acoustics and designer of the Ascent and Eclipse. Yeah, these were big heavy guys, with faceted front baffles, very thick dead enclosure walls, four way with dual 10's, 6-1/2 midrange, dome midrange, dome tweeter, to try to keep each driver in it's pistonic range. He kept them from years, didn't have the idea to make a speaker company, but wanted to build bike frames. When he gave up on that, it was back to the speakers and trying to figure out why they sounded so good and differnet from most speakers you could hear, even stuff like B&W. Figured if he could condense it down to a two way or three way, he might have something. Chas was a phsyics major anyway, so the rest is history.

Sometimes people wonder why I'm so laid back at work about politics and jockeying for position or title- the point most don't know or get is that I get my ego jollies away from work- my engineering ego is invested in this hobby, not my day job. That throws people for a loop, until I explain one of my projects, and the reasoning behind it, and these guys are amazed at the background knowledge and theory involved, and their comment is that I seem to approach this in a very professional manner compared with what they read about the real speaker design companies.

Hey, everyone has to have a little passion in their life about something!


~Jon

[JonMarsh]

Hi Jon,

Thanks for taking the time for a detailed response.

What it reminds me of is that any kind of design is essentially problem solving. A series of decisions are made, attempting to reach a goal. There are innumerable paths to reach the goal, but every decision made along the way, especially those made early in the planning and design process, effect the rest of the path and both remove and make possible opportunities.

I'm a Landscape Architect by profession, so I understand the design process quite well, however, I never thought to apply it for speaker design! Thanks for reminding me of the similarities. It gives me a fresh insite into how to move forward with any future designs. Of course, at this point in my speaker design knwolege, I'm in the unconscious incompetant stage. That is, I don't know what I don't know. ops:

That's part of what I was hoping this thread would do. Open the eyes of the novices like me out there and let them begin to discover some of the things they don't know!

One question I have. What do you mean by "tweeter directivity flare at lower treble frequencies"?

If you look at a lot of two way systems, there's a big mismatch between the midwoofer directivity and the tweeter directivity/dispersion in the crossover region. This involves both amplitude and phase. Driver spacing plays a role, too.

Looking at the crossover behavior off axis horizontally, the midwoofer output is dipping, but the tweeter output is flat, and the phase between the two is different than on axis, so as you go down in frequency you get a flare in the output level (from the tweeter) then a dip due to the limitations of the midwoofer.

It's very illustrative to check out a lot of the reviews for speakers, particluarly the measurements, at the Stereophile site. There are, IMO, a lot of mediocre "high end" speakers, which exhibit classic problems in design with regards to the crossover integration and driver behavior.

Remember though, that a commercial design is trying to do the most adequate performance at the least amount of cost, in most cases, and perhaps they're counting on people not making sophisticated measurements or listening tests.

I went to CES 2006, and for the most part, I wasn't particularly impressed by the high end stuff there, with a few exceptions showing good competency in design, but perhaps a few minor flaws or issues with associated components. Nothing I saw or heard sounded as good as a properly constructed Orion system, though many would play louder.

[ChaoticKinesis]

In hopes that this thread will get moving, because I think it's a great idea, I have a question regarding power response.

I see that when Jon Marsh displays his designs the power response looks nearly identical to the on-axis response. However, when I model a 4th order LR with the FRD's PCD I always see a dip in power response of about 5dB in the crossover, followed by a slight peak above, and then a gradual sloping down in the highs. The high end dropping off is understandable due to tweeter dispersion characteristics, but what about the dip in the crossover?

Is this strictly a function of the crossover topology or is it something that can be fixed while maintaining the same type of crossover? If the former is correct then how is this problem normally addressed in loudspeaker design?

Dave

[jdybnis]

It's very illustrative to check out a lot of the reviews for speakers, particluarly the measurements, at the Stereophile site. There are, IMO, a lot of mediocre "high end" speakers, which exhibit classic problems in design with regards to the crossover integration and driver behavior.

Stereophile has also reviewed some "mass-market" hi-fi that are very well engineered. I think KEF was one and Monitor Audio was another that have very good off axis response in the crossover region.

[Dennis H]

I think PCD simulates the power response in a full sphere while Jon's graphs just show a few curves fairly close to on axis, at least in the vertical direction. Getting way off axis vertically causes the big suckout at the XO frequency.

[ChaoticKinesis]

I think PCD simulates the power response in a full sphere while Jon's graphs just show a few curves fairly close to on axis, at least in the vertical direction. Getting way off axis vertically causes the big suckout at the XO frequency.

So does that mean this is something that I should not pay too much attention to, as this is typical of any speaker and cannot ever be fully addressed?

[JonMarsh]

In hopes that this thread will get moving, because I think it's a great idea, I have a question regarding power response.

I see that when Jon Marsh displays his designs the power response looks nearly identical to the on-axis response. However, when I model a 4th order LR with the FRD's PCD I always see a dip in power response of about 5dB in the crossover, followed by a slight peak above, and then a gradual sloping down in the highs. The high end dropping off is understandable due to tweeter dispersion characteristics, but what about the dip in the crossover?

Is this strictly a function of the crossover topology or is it something that can be fixed while maintaining the same type of crossover? If the former is correct then how is this problem normally addressed in loudspeaker design?

Dave

Total power response will never be perfect, though you can improve some things with tricks like coincident drivers (mid and tweeter) or driver loading which modifies the off axis response.

In the first order, one should be looking at the vertical and horizontal curves out to at least 30 degrees, preferably 45 degrees. Vertical is tough unless one uses a relatively low crossover frequency relative to the driver spacing- one reason some of my more expensive designs use an emulated LR8 from a cauer-elliptic filter, and a 1200 Hz crossover frequency. They than have remarkably good response from about -30 degrees to plus 45.

But many speakers even have fairly poor off axis horizontal response- this is most often caused by high crossover frequency for the midwoofer, so it's beaming at 2.5 kHz, whereas the tweeter is in full flare. Combine that with crossover types which have destructive phase interference in parts of the crossover region, and normally have vertical lobing (BW3, for example), and things get messy.

I prefer all-pass style crossovers where the drivers stay in phase through the crossover, but a net 180 degree transistion may occur- LR8 fullfills this.

If an LR4 is what you like to build, you're better off doing a Bullock compromise; it sounds better in most cases, it's bumped up about 2.5 dB on axis, so the net power response is only down a couple of dB. Even there this assumes nearly coincident drivers relative to wavelength.

So, for example, a worst case spread center to center for 2.5 kHz crossover is 5-1/2", and as you can imagine, that's tough to do. Drop the crossover frequency to 1200-1400 Hz, and things are looking a lot better.

[ChaoticKinesis]

Thank you Jon for that excellent response! It answered some questions I had but left me with a new one. Unfortunately, after doing numerous searches on this "Bullock compromise" the only answer I seem to find would require me to pay the AES $20, which may well be worth the money for the read but that's besides the point. Any of the more knowledgeable members care to clue us in on this?

And once again, many thanks for that Jon.

Dave

[JonMarsh]

It's just an LR-4 with the damping coefficients modified so that instead of being - 6dB on the HP and LP at the corner frequency, it's about -4 to -3.5 dB; the on axis pressure response kicks up a hump about 2.5-3 dB, but the net power response is closer to flat.

Another filter transfer function that sums well for power response and axial response is the 3rd order all pass, which is what the Modula MT and Natalie P are loosely based on. Problem with BW3 is the off axis vertical lobe; kind of helps the power response, but in a weird way. To get flat axial response in the crossover region, a certain amount of destructive interference is required. Not a good thing. Still, it sounds better (usually) than a straight LR4, espcially on pink noise.

Acid test for crossover characteristic testing is comparing two wide band mides with a crossover set in the middle of their operating range (say, 1 kHz). It's not hard to hear differences on music, but it's VERY easy to hear differences with pink noise, compared with the mid reproducing pink noise by itself full range. Try it. It's an interesting test, if you're really curious about the sonic affect of different crossover types.

Somewhere around here I have the Bullock paper in one of my AES anthologies (I'm an AES member). I'll try to dig that up during the holidays.

~Jon

[ChaoticKinesis]

Thanks once again for the excellent response Jon. That Bullock paper would probably be a great read for forum members who are interested in designing their own speakers.

I was wondering if anyone has any opinions on how low would be a safe crossover point for a 27TDFC with a BW3 slope.

Dave

[AJINFLA]

you can improve some things with tricks like coincident drivers (mid and tweeter)

Tricks? :B

The XO will always be the key.
Well, unless of course you have a Fostex or a Lowther or..... :rofl:

cheers,

AJ

p.s. any 'treats' under the christmas tree?