Any thoughts on sloped baffles and/or time delay between drivers?

[JonW]

I’m reading about how the acoustic centers of different drivers are often not in the same place. So take a regular flat front baffle with a tweeter and a woofer. The sound from the tweeter will reach the listener before that of the woofer. And so the sound may not appear quite as precise. One solution is to place a time delay for the tweeter into the crossover. Another solution is to have the front baffle sloped (say 10 degrees) such that the woofer is closer to the listener than the tweeter. In the Loudspeaker Design Cookbook, Dickason discusses the pros/cons of each approach to dealing with the problem and he prefers using a time delay. I’ve noticed that some commercial speakers I’ve liked have sloped front baffles. Two examples here:

http://www.sumikoaudio.net/sonus/prod_grandpiano_domus.htm

http://www.vonschweikert.com/zvr4jr.htm

So I’m thinking about sloped baffles and/or time delays for my next project. I’m wondering if any of you folks have any thoughts on the matter. Some questions and thoughts that I have for now:

-For determining the proper slope for a baffle, the book says to make a straight line from the back plate of the woofer to the back plate of the tweeter. Is this precise enough or does one need to do something fancier?

-How would one add a time delay (say ~300 microseconds for the tweeter) into a crossover? I haven’t seen too much explanation of that. And how would you determine the proper amount of delay to use without super fancy measuring gear?

Thanks!

-Jon

[Rolex]

No comments on the time delays for the crossover. I'll leave that to the experts. Thiel, and audio physic are big on slanted front baffles. Apparently so am I. Here is the design I'm currently finishing.

Image not available

[chasw98]

So I’m thinking about sloped baffles and/or time delays for my next project. I’m wondering if any of you folks have any thoughts on the matter.
-Jon

Just look at the Avalon designs and at the M8ta and the upcoming Isiris, I believe. They are slanted and faceted!

Chuck

[JonMarsh]

I’m reading about how the acoustic centers of different drivers are often not in the same place. So take a regular flat front baffle with a tweeter and a woofer. The sound from the tweeter will reach the listener before that of the woofer. And so the sound may not appear quite as precise. One solution is to place a time delay for the tweeter into the crossover. Another solution is to have the front baffle sloped (say 10 degrees) such that the woofer is closer to the listener than the tweeter. In the Loudspeaker Design Cookbook, Dickason discusses the pros/cons of each approach to dealing with the problem and he prefers using a time delay. I’ve noticed that some commercial speakers I’ve liked have sloped front baffles. Two examples here:

http://www.sumikoaudio.net/sonus/prod_grandpiano_domus.htm

http://www.vonschweikert.com/zvr4jr.htm

So I’m thinking about sloped baffles and/or time delays for my next project. I’m wondering if any of you folks have any thoughts on the matter. Some questions and thoughts that I have for now:

-For determining the proper slope for a baffle, the book says to make a straight line from the back plate of the woofer to the back plate of the tweeter. Is this precise enough or does one need to do something fancier?

-How would one add a time delay (say ~300 microseconds for the tweeter) into a crossover? I haven’t seen too much explanation of that. And how would you determine the proper amount of delay to use without super fancy measuring gear?

Thanks!

-Jon

You mean cabinet designs like this?



Nearly aligned or aligned cabinet designs can make life easier for the crossover design, no question. In general, though, you have to understand the phase tracking characteristics of different type of crossovers and their tendencies for lobing off the main axis, too. All pass networks, which presume a complemtary phase transistion of the drivers through the transition region can be optimized for noncoincident and nonaligned applications by adjustment of the filter phase response in the transistion region.

All pass ladder networks can be used to add time delay without changing the nominal frequency response, but I tend to view them as a last resort, due to the number of components required and the potenial impact of small additional errors in frequency response due to component tolerances, and just plain old sonic footprint of capacitors.

Some software like LspCAD has prebuilt blocks for all pass delay elements.

Jon, what references have you picked up so far? With review of the right basic references, I think you'll get up to speed pretty quickly.

~Jon

[Mark Seaton]

Whenever talking about time or physical offset with loudspeakers, remember to keep some idea of relative scale to frequency. While there can be/often are benefits from having a tweeter further than the woofer, it's only partially due to the physical location of the voice coil, but more dependent on the crossover behavior. If you start working out the math, an added 90-180deg phase shift near a mid-tweeter crossover can have a much bigger impact on delay than you're ever going to account for in a lightly sloped baffle. There are various benefits and justifications for a sloped or stepped baffle, but keep in mind relative listening axis of the tweeter and where the listening axis between the drivers ends up.

[JonW]

Nearly aligned or aligned cabinet designs can make life easier for the crossover design, no question. In general, though, you have to understand the phase tracking characteristics of different type of crossovers and their tendencies for lobing off the main axis, too. All pass networks, which presume a complemtary phase transistion of the drivers through the transition region can be optimized for noncoincident and nonaligned applications by adjustment of the filter phase response in the transistion region.

All pass ladder networks can be used to add time delay without changing the nominal frequency response, but I tend to view them as a last resort, due to the number of components required and the potenial impact of small additional errors in frequency response due to component tolerances, and just plain old sonic footprint of capacitors.

Some software like LspCAD has prebuilt blocks for all pass delay elements.

Jon, what references have you picked up so far? With review of the right basic references, I think you'll get up to speed pretty quickly.

~Jon

Jon-
Thanks for chiming in. Right, I need to start thinking about phase as well. Although I’m not far enough along to understand the phase issue too well. (But I’m working on it.)

Excellent to know that LspCAD can deal with this. And interesting that you dislike all pass ladder networks up front. I'll keep that in mind.

So far, I’ve got the standards: “Loudpseaker Design Cookbook” and “Speaker Building 201.” I read them both earlier, just for getting a general idea of how things work. I’m now rereading them with an eye toward what I’ll want to incorporate into a design. I’ve also got “Loudspeaker Testing” but haven’t cracked it open yet.



Rolex-
Yup, just what I’m talking about. Those look super, by the way.



Chuck-
Yeah, once I started thinking about the slopes and facets, I’m noticing them pop up in more commercial designs. That second link I have up there has small facets- it’s not just Avalons.



Mark-
Thanks. I think you’re saying something pretty similar to Jon. I’ve got to do some more reading to understand the implications of phase shift (or no shift) between the drivers around the crossover point.


-=-=-
I’m fine with using facets and/or slopes in the design if it will make things easier in the long run. I also see that some speakers have the woofers closer to the listener, on a separate face than the tweeter. Like here:

http://www.wilsonaudio.com/product/alexandria/

But I’d think you’re then asking for lots of trouble with crazy diffraction issues.

[Mark Seaton]

Many hi-fi companys love to create hurdles and problems in a design and then pass off the effort to minimize the faults they designed into the speaker for real engineering. There are some great designs and ideas out there, but don't blindly assume there was a good acoustic reason for a design choice in a hi-fi speaker.

[JonW]

Many hi-fi companys love to create hurdles and problems in a design and then pass off the effort to minimize the faults they designed into the speaker for real engineering. There are some great designs and ideas out there, but don't blindly assume there was a good acoustic reason for a design choice in a hi-fi speaker.

An excellent point. And one that I need to keep in mind more. Thanks.

[jkrutke]

So far nobody's mentioned the most common way of dealing with acoustic center offsets - asymmetrical slopes. I've used all of the methods many times, but it's the one I use the most often because it's usually the simplest solution. Early on in my speaker design hobby, I used asymmetrical crossovers without even knowing what they were or why they worked.

I think there was an article written by Jeff Bagby somewhere that describes how it works. I can't recall where at the moment but maybe someone can point it out.

[JonMarsh]

Did that a few times myself in the late 80's and 90's. Hawksford has a good paper on it, too.

[dlr]

The solution should also consider other design details

What to do to take into account the relative acoustic offset will depend on a number of factors. If it's pure 2-way, it will likely be more important since most midwoofers have a fair amount of offset. If it's a 3-way with a smaller midrange unit than, say 6.5", then the offset may not be as significant.

BUT, then there's the question of the crossover frequency and order. If the crossover Fc is high, then the offset importance will increase. If it's a low slope, then the same applies.

You also have to decide how much slope is acceptable and how much vertical offset is acceptable for whatever slope is used, since they have to be chosen together. It doesn't have to be absolutely precise, since some control can be effected through modest alteration of the crossover, usually relaxing the slope a bit in the lowpass. This reduces the delay inherent in the lowpass so that it appears as though the lower unit (woofer, midrange) is not as far back.

-For determining the proper slope for a baffle, the book says to make a straight line from the back plate of the woofer to the back plate of the tweeter. Is this precise enough or does one need to do something fancier?

I have to disagree with that one. Usually the suggestion is the front plate, but that's still only a rough approximation. If you can't measure, then I would suggest using the location where the former attaches to the diaphragm.

You may find a couple of articles at my site to be helpful. See the ones on Relative Acoustic Offset and on Phase, the first two. Start here:

Dave's Speaker Pages

-How would one add a time delay (say ~300 microseconds for the tweeter) into a crossover? I haven’t seen too much explanation of that. And how would you determine the proper amount of delay to use without super fancy measuring gear?

I think that the article on Phase will provide the answer for the first question. You really can't determine the delay without measurements and be precise at all. In that case, only rules of thumb as the one that I suggested earlier can be used as a guide.

Personally I prefer to use offset drivers, similar to those of Dunlavy. However, this pretty much requires the use of felt for diffraction control as it's usually more prominent than for a sloped baffle. If you have the audioXpress from June 2005, you can see an example of a stepped baffle with felt diffraction control applied. A sloped baffle will have much less need for diffraction control, though it will still exist.

[TurboFC3S]

Personally I prefer to use offset drivers, similar to those of Dunlavy. However, this pretty much requires the use of felt for diffraction control as it's usually more prominent than for a sloped baffle. If you have the audioXpress from June 2005, you can see an example of a stepped baffle with felt diffraction control applied. A sloped baffle will have much less need for diffraction control, though it will still exist.

Vandersteen uses this approach as well ...

http://www.vandersteen.com/pages/Model5alit1.html

And now that I've found a family member with access to a full machine shop and laser cutter, I'm going to try and replicate something similar myself.

[---k---]

Jon,
So it sounds like you must have been impressed enough with what Thomas had around his place that you want to continue the DIY adventure. I was starting to think you were lost to the world of snooty retail speakers. I'm looking forward to seeing your projects progress.

Did you post any thoughts on what you heard at Thomas's? Would love to know what you heard and what you thought.

[JonW]

So far nobody's mentioned the most common way of dealing with acoustic center offsets - asymmetrical slopes. I've used all of the methods many times, but it's the one I use the most often because it's usually the simplest solution. Early on in my speaker design hobby, I used asymmetrical crossovers without even knowing what they were or why they worked.

I think there was an article written by Jeff Bagby somewhere that describes how it works. I can't recall where at the moment but maybe someone can point it out.

Might this be what you mean?

http://www.t-linespeakers.org/tech/filters/Jeff/index.html

I gave it a read and it looks to be. They had measurement data to know that the two drivers were 70 microseconds apart. That must be helpful to know, although I don’t know if amateurs have access to the needed equipment for that type of measurement.

If I read it correctly, he used a 3rd order slope for the tweeter, but a 2nd order slope for the woofer. Because the higher order tweeter slope has inherent group delay. Interesting! (Can you see the light bulb that just went off?)

Although I do not understand how he knew how much delay he gets from a given slope or crossover. And how he compensated exactly for that 70 microseconds.

Hmmm... might one want to combine this delay with a sloped baffle or only use the delay or...? Hmmm... I've got some more thinkin' and readin' and learnin' to do.

[JonW]

Thanks for all the input fellas.

I just ordered myself a Behringer DCX2496 to play around with. This thing has everything. In addition to the fully adjustable crossovers and a DAC, it also has a programmable delay between channels. So I can play around with adding in delays between, say, the tweeter and woofer. Neat, eh? I’ll see if that helps anything.

Vandersteen uses this approach as well ...

http://www.vandersteen.com/pages/Model5alit1.html

Man, those are some ugly looking speakers. No wonder they put a paper bag over the head.

Seriously, though, I'm open to odd shapes if it will help the sound. I'll worry about the asthetics later. The Vandersteens look to be using various aspects: stepped baffle with the tweeter farthest back, narrower baffles around each driver to control diffraction, and felt as well. Interesting. Maybe the felt is needed to deal with all the diffraction issues from the stepped baffle.

[cjd]

IMHO

Start with a basic box

Understand what you can do to manipulate phase

then start worrying about the rest.

Of course, building a multi-box system means you can use the same drivers for all the different attempts.

C

[JonW]

Jon,
So it sounds like you must have been impressed enough with what Thomas had around his place that you want to continue the DIY adventure. I was starting to think you were lost to the world of snooty retail speakers. I'm looking forward to seeing your projects progress.

Did you post any thoughts on what you heard at Thomas's? Would love to know what you heard and what you thought.

Hi Ryan,

I'll drop you an email because I'm curious about what your next project will be. But the short answer is that as very nice as Thomas' speakers, IB, etc. are, we spent more time chatting than listening to his gear. It was more his perspectives that are egging me on at this point. And with all my future questions, I'm pretty sure he will soon grow to regret his encouragement.

[JonW]

Hi Dave,

Thanks for the excellent info. :T I hadn’t seen your website before- very nice. Nice article on acoustic offset. (I still need to read the phase one more carefully.) Yet another approach to dealing with the offset issue. I guess I don’t understand how you actually measured the phase data for the drivers. Or how practical it is to put each driver in the same spot for measuring when you’ve already build the cabinet. Maybe just do each one in flat baffle.

… some control can be effected through modest alteration of the crossover, usually relaxing the slope a bit in the lowpass. This reduces the delay inherent in the lowpass so that it appears as though the lower unit (woofer, midrange) is not as far back.

That makes sense, in general. But I’m not sure how you tweak the crossover exactly to compensate for a specific amount of delay. Maybe once I model my first crossover in some software it will be obvious.

I have to disagree with that one. Usually the suggestion is the front plate, but that's still only a rough approximation. If you can't measure, then I would suggest using the location where the former attaches to the diaphragm.

My mistake. You are correct. It’s the front plate. So are regular measurements with a ruler OK for this?

Personally I prefer to use offset drivers, similar to those of Dunlavy. However, this pretty much requires the use of felt for diffraction control as it's usually more prominent than for a sloped baffle. If you have the audioXpress from June 2005, you can see an example of a stepped baffle with felt diffraction control applied. A sloped baffle will have much less need for diffraction control, though it will still exist.

I understand. The offset driver approach creates such edges on the baffle, I do wonder if felt is enough to compensate.


-=-=-
So in general, people here are proposing various solutions to the acoustic offset issue. Dare I try to address it in the crossover AND the physical baffle slope/offset- or just pick one? This is all pretty interesting.

[JonW]

IMHO

Start with a basic box

Understand what you can do to manipulate phase

then start worrying about the rest.

Of course, building a multi-box system means you can use the same drivers for all the different attempts.

C

There is definitely wisdom in starting simple and getting more complex later. But Idunno, if you know there are issues you ignored from the very start...

By multi-box system, do you mean like a top MT box and then bass bins?

[cjd]

Well, I mean each driver in its own box so you can line 'em all up straight or line 'em up so the acoustic centers are aligned, etc.

You ALWAYS choose to ignore something or two or three. That's part of the misery - choosing what compromises you'll work with this time.

C

[JonW]

Well, I mean each driver in its own box so you can line 'em all up straight or line 'em up so the acoustic centers are aligned, etc.

Got it. Thanks.
I'm liking the ability to play with delay in the DCX2496. Heck, maybe I won't even hear any difference.

You ALWAYS choose to ignore something or two or three. That's part of the misery - choosing what compromises you'll work with this time.

C

Misery, indeed. Harumph!

[CraigJ]

I feel like we are watching the birth of a Mad Scientist Go JonW, Go.

[Dennis H]

I guess I don’t understand how you actually measured the phase data for the drivers. Or how practical it is to put each driver in the same spot for measuring when you’ve already build the cabinet.

Assuming you're using LspCAD.... when measuring put the mic centered on each driver at a fixed distance from the baffle. If you want, you can subtract a 'time of flight' distance, from the mic to the baffle, from the measurements. It's not really necessary but it will make the phase picture clearer because the HF phase won't wrap so many times and the picture won't look so confusing. The measured phase will include any acoustic center offsets relative to the baffle. When you set up the whole speaker in LspCAD, set the X and Y distances to where the drivers are on the baffle and set the Z distance to 0 for a flat baffle because the measurements already include the acoustic Z offset. When you optimize the crossover, you want the phase of both drivers to match through the crossover region and you do that by adjusting XO components.

Clear as mud?

[JonMarsh]

Oh, and while you're doing what Dennis describes above, always use the same window, even for swept sine chirp measurements, because you align them using an MLS like impulse function. Best to align the impulse start on the tweeter, than measure everything else without changing your setup. That's what works for me.

Then, different crossover types have different phase tracking characteristics in the transition region- the easiest ones to figure out if you're on the money are those that specifically have deep nulls if the drivers are reversed.

Why? Because you can use that as a test to see if you're managing the transistion region well, by reversing the phase connection.

Now, you can get what looks like a sort of acceptable summing on axis even when things aren't tracking and reversing the connection only makes a small difference in the amplitude in the transistion reigon. But THEN, you don't have an all pass crossover, you have something with possibly the phase in quadrature, and destructive interference on axis, and a big lobe off the design axis.

Now, this is what happens in some of the "classic" crossover types like 3rd order Butterworth, which can be connected with drivers in either phase, and sums well on axis. BUT, the power response is a bit goofy, and the speaker can sound weird in the transition region off axis, where the response is peaking. This also happens with the classic 1st order Butterworth, one reason I abandoned those in the late 70s. I could make a speaker that reproduced good pulse waveforms on axis, but the room power response didn't really sound right.

[dlr]

I'm a bit puzzled

Now, you can get what looks like a sort of acceptable summing on axis even when things aren't tracking and reversing the connection only makes a small difference in the amplitude in the transistion reigon. But THEN, you don't have an all pass crossover, you have something with possibly the phase in quadrature, and destructive interference on axis, and a big lobe off the design axis.

Jon, I don't follow this, unless you're referring to the case of non-aligned acoustic centers. In that case, whether the XO is an all-pass or constant-power, there will be some amount of destructive interference, since the phase relationships won't be ideal in either case.

Now, this is what happens in some of the "classic" crossover types like 3rd order Butterworth, which can be connected with drivers in either phase, and sums well on axis. BUT, the power response is a bit goofy, and the speaker can sound weird in the transition region off axis, where the response is peaking. This also happens with the classic 1st order Butterworth, one reason I abandoned those in the late 70s. I could make a speaker that reproduced good pulse waveforms on axis, but the room power response didn't really sound right.

This also is puzzling and I think incorrect on the physics, though the inroom response at the listener position will vary due to room conditions.

Have you seen the web pages by John K in which he demonstrates the power response of various types and orders of XOs for MT and MTM layouts? In those he uses the program he created that shows the full, true integrated power reponse. The odd order Butterworth types all have, in fact, the smoothest power response of the various crossover types. For a typical 2-way he shows the 1st, 3rd and 7th to demonstrate vs. L-R up to 8th order.

John K's power response pages

Later pages go into non-point-source drivers on finite baffles with baffle step response included. In the end, for MT units, with no room-influence issues, the power response for Butterworth crossovers is superior to L-R types of any sort. The BW7 has a 3db dip in power whereas the LR8 has a 6db dip in it at Fc. The BW1 has a much broader dip, but with adjustment for the drop, it could be made to be nearly perfectly flat in power response.

MTMs are also clearly not as good in the power response vs. an MT.

The lobing issues certainly will come into play for the direct response, but when well-implemented, the odd-order BW is closest to a flat power response. So I'm also a bit puzzled about what the source may be for your comment that "the power response is a bit goofy".

John's not a BW1 guy, though, so rather than paraphrase John, I find this paragraph to be most interesting:

"First, the Butterworth crossovers no longer have flat power response which is primarily a result of the increasing directionality of the woofer as the frequency rises. Second, the Butterworth crossovers still have smoother power response than the Linkwitz types, regardless of order. Third, as the crossover order increases the width of the depression in the power response becomes narrower. It is not clear whether this is a favorable result. Since the ratio of direct to reflected sound will be greater in the region where there
is reduced total power, and since a higher ratio of direct sound generally implies greater clarity and localization, it is possibly that narrower, more abrupt variation in the power response of a higher order crossover may result in some apparent emphasis of frequencies in this range. This is speculative, but somewhat born out by my observations over the years that the common LR4 crossover has a clearly identifiable signature to it and I find LR2 rossovers generally, when well executed, sound more natural."

[JonMarsh]

The problem is that in the real world we have non conincident drivers, unless it's a coaxial like AJ favors or Gradient uses.

I'd refer you to Linkwitz's paper "Active Crossover Networks for Nonconincident Drivers", particularly the 4th page.

The issue of audibility of crossover transfer function is a very interesting one, in the relationship and tradeoffs between axial response and power response in typical rooms and room placements.

Hawksford's 1993 paper on All-Pass Crossover Alignments is good reading, as well as Joseph D'Appolito's "A Geometric Approach to Eliminating Lobing Error in Multiway Loudspeakers (AES 200).

[TurboFC3S]

Man, those are some ugly looking speakers. No wonder they put a paper bag over the head.

Exactly what I've always thought!

Seriously, though, I'm open to odd shapes if it will help the sound. I'll worry about the asthetics later. The Vandersteens look to be using various aspects: stepped baffle with the tweeter farthest back, narrower baffles around each driver to control diffraction, and felt as well. Interesting. Maybe the felt is needed to deal with all the diffraction issues from the stepped baffle.

Lots of good ideas in their design, I might even have one or two of my own

I'm curious if their midrange and mid-bass drivers have separate enclosures, I assume so.

[dlr]

I've had those on my "wish" list of reading...

The problem is that in the real world we have non conincident drivers, unless it's a coaxial like AJ favors or Gradient uses.

I'd refer you to Linkwitz's paper "Active Crossover Networks for Nonconincident Drivers", particularly the 4th page.

The issue of audibility of crossover transfer function is a very interesting one, in the relationship and tradeoffs between axial response and power response in typical rooms and room placements.

Hawksford's 1993 paper on All-Pass Crossover Alignments is good reading, as well as Joseph D'Appolito's "A Geometric Approach to Eliminating Lobing Error in Multiway Loudspeakers (AES 200).

but I've never made the effort to order copies.

WRT non-coincident drivers, the graphs on John's pages are for non-coincident, non-point-source driver pairs. That's what's most interesting.

The point about coincident sources is interesting as well. When I was working with the Insignia, it dawned upon me that excluding the directionality differences that may exist in non-ideal examples, the power response from these will be perfectly flat for any crossover that sums to a flat frequency response. They're unique in that respect.

Non-constant directionality from either driver will alter the power response, of course, as will the out-of-band driver rolloff influence, though the latter may be minimal in impact. The alteration should be primarily in the tweeter directionality at the upper end and be seen as a tapered power response.

Outside of this, though, in comparing well implemented versions of various crossovers, the power response by default will be closest to flat with odd-order Butterworth. Issues of perception and its nuances have their affects, of course, but I see that as a separate issue from the objective performance of each kind of crossover.

[dlr]

Some more details

Thanks for the excellent info. :T I hadn’t seen your website before- very nice. Nice article on acoustic offset. (I still need to read the phase one more carefully.) Yet another approach to dealing with the offset issue. I guess I don’t understand how you actually measured the phase data for the drivers. Or how practical it is to put each driver in the same spot for measuring when you’ve already build the cabinet. Maybe just do each one in flat baffle.

The measurements were made with the drivers mounted on the final baffle and without moving the microphone. By using three measurements, W, T and WT combined, it's possible to determine the exact amount of offset required for the modeled minimum-phase response of the drivers (that does not include the excess-phase, meaning the time delay from mic to driver) for that particular model of the drivers.

The latter part is key. If you assume a somewhat different slope for the driver above the measured area, then that change requires readjusting the offset.

That makes sense, in general. But I’m not sure how you tweak the crossover exactly to compensate for a specific amount of delay. Maybe once I model my first crossover in some software it will be obvious.

Yes, you'll see that in the CAD software. Change the slope and the phase response changes.

My mistake. You are correct. It’s the front plate. So are regular measurements with a ruler OK for this?

For this method, yes.

I understand. The offset driver approach creates such edges on the baffle, I do wonder if felt is enough to compensate.

It eliminates the majority of it.

So in general, people here are proposing various solutions to the acoustic offset issue. Dare I try to address it in the crossover AND the physical baffle slope/offset- or just pick one? This is all pretty interesting.

You can do both, yes. Adjusting in the crossover with assymetric slopes only is limited in range of adjustment.

[JonW]

Dave, Dennis, and John-

That’s all quite helpful. Thanks for the great discussion. I don’t understand 100% of what you’re all saying, but much more so than even a week ago. I’m reading, learning, and getting there.

LspCAD is being mentioned often enough I’m starting to think I should spring for that and start using that right away, rather than some other software.

My current thoughts are to build a sloped baffle, with a slope angle determined by measuring the drivers from front plate to front plate. With good measurement data of the drivers in that cabinet, the crossover design may go a little smoother down the road. If the time delay and diffraction issues can be handled somewhat by the cabinet design, I might as well work them in now, not try to force the issue later in the electronics.

And also incorporate some faceting and/or rounding over of edges into the cabinet to deal with diffraction issues. (Plus felt later.) I’m thinking the faceting might be done relatively easily with a handheld jig saw like this one shown. It has a base that can be adjusted to various angles. (Bonus: I already own it.) Clamp a straight edge as a guide. Not nearly as easy as buying a premade cabinet, but the woodworking doesn’t look too daunting. (Naïve? Danger Will Robinson!)



I’ve downloaded Edge and BDS to play around with. I’ve got various ideas of baffle shapes, widths, faceting, rounding over, etc. in mind. I’ll see what models best and go from there. I’m hoping to have time to try that this weekend, but we’ll see. Then slope the baffle. Pick the drivers, put them in the cabinet, measure them… Heck, it’s almost done.

[ThomasW]

Jon,


It's way to difficult to control a jigsaw, you'll never get a straight even bevel.

http://mywebpages.comcast.net/kidder/Audio/4eleven/411.htm

http://www.diyaudio.com/forums/showthread.php?s=&threadid=72574

This is a slide show

http://homepage.mac.com/marc.heijligers/audio/milestones/cabinet/construction/construction.html

[chasw98]

http://mywebpages.comcast.net/kidder/Audio/4eleven/411.htm

I like the "Box of Doom" shown here : http://mywebpages.comcast.net/kidder/Audio/BoD/BoD.htm

That would keep Jon busy for a while! :B

Chuck

[ThomasW]

I've have a stack of capacitor and resistor 'decade boxes' Jon and Bill Kennedy used for prototyping back in the 1980's. They just gather dust in the basement, so I should give them away to some deserving soul.

Keith's a funny fellow, for several years he'd stop by when his wife made her annual trips to Denver for medical conferences.

These days I think the DCX2496 is the best choice for prototyping.

[JonMarsh]

LspCAD Pro--- dial in the changes on the fly, and listen to the output from one computer (a lot less trouble to buy an Maudio 24/96 family sound card than build that monster! Impressive, but so 20th century!

[JonW]

Hi Fellas,

For facets, I was thinking maybe I could rotate the base of the jig saw to an angle, say 45 degrees. Then use some sort of straight edge guide to hold the jig saw against. Kind of like what this guy did using a circular saw (Pics below, from Thomas’ first link). Or heck, I could do what he did with a circular saw. At any rate, cutting facets is a separate topic. If I need help with that, I’ll start a new thread.

In the mean time, I have to figure out what the optimal cabinet will look like. I guess I’ll try a sloped baffle to handle a little of the delay issue, to make the actual crossover need to to less in this regard. Then something for the diffractions. I did a wee bit of modeling last night with Baffle Diffraction Simulator- *very* cool program. One question I was wondering about is if it would be better to make a very narrow cabinet. Or add facets, but that necessitates a wider cabinet. A narrow, straight edged cabinet modeled pretty close to a wider one with facets. Not surprising, I guess. But that was a quick test and I really need to take some time modeling a bunch of different options I have in mind. Roundover edges, a baffle that is narrower by the tweeter versus the woofer, etc.

Impressive, but so 20th century!

I’m feeling pretty lucky that I decided to try designing a speaker after the Behringer DCX2496 exists. It looks like that thing will really speed up the process.



Image not available


Image not available

[augerpro]

JonW- I've found jigsaws to be totally worthless for precision straight cuts, even using a metal guide. OTOH A circular with a good fence, a high tooth blade, metal cut guide, and some skill could probably be workable. I finally gave in two days and bought a Ridgid 12" miter saw that has the extension arm. Now this thing can make precision cuts at any angle baby!

[cjd]

The problem with a jigsaw is the loose blade tip likes to wander. So you may get a perfectly straight line up by the boot, but it goes downhill from there - the tip-end can warble and wander like mad. A hand saw will give you better results. Yeah, it can be a pain, but get a quality saw with a medium to fine tooth (yes, it takes longer, but it gives you better control) and some patience. It may be worth starting the cut with a backsaw as far as it'll go - once you've started it nice and straight, the cut itself will continue to guide the saw.

C

[chasw98]

Jons got a table saw and I sent him a link to a taper jig last night. I told him to look at Colorado Tom's M8ta thread where it shows a 'sled' he built for cutting facets. Much more precise and accurate and repeatable.

Chuck

[JonW]

Hi Guys,

Thanks for the help. I haven’t used my jig saw enough to realize the blade warbles. No problem, I can use my table saw or circular saw instead. Here is the setup that ColoradoTom used, post 112:

https://www.htguide.com/forum/showthread.php?t=19193&page=4&pp=35

If I understand it correctly, the sled allows for 45 degree cuts (call it vertical cuts in the case of Tom’s cabinet) and the taper jig allows for variable angles, up to 15 degrees, for the other angle (call it horizontal). Not sure how you get angles larger than 15 degrees with the taper jig, though. (And ColoradoTom sure has some skills!) But it generally looks pretty straightforward and easy to make.

But way before I get to all that, I need to do much more computer modeling and reading to see what the optimal cabinet design will be. My current thoughts are a double thick (1.5”) cabinet all around to reduce vibrations. With a 10 degree front baffle slope (or whatever angle is best for the drivers I pick) to help with the acoustic delays. Then facets (maybe with some sanding of the sharp edges?) to help with diffraction. And the facets taper to come closer to the tweeter than the woofer, to really try and minimize tweeter diffraction. So far that sounds close to Jon’s M8ta design, but the woofer and tweeter will probably want to be closer together. Not sure if driver offsets from the center will help. And all that spiced with some wool felt. At any rate, these are just ideas. We’ll see what the modeling says.

And to think that I almost was going to start with a premade cabinet... h: