Question for OB experts regarding imaging

[madisonears]

I am at the prototype testing stage of my original design OB mids/tweets section of a projected 3 way system. I would appreciate input from anyone experienced with OB implementation.

Two 4" mids and a Peerless HDS tweeter are mounted MTM. Spacing between drivers is very close, with mids slightly overlapping tweeter flange. Baffle width is 15", with the outer 3" of both side edges curved into a 3" radius (essentially "wings" that curve toward the rear). The drivers are offset from centerline by about 2". Xovers sound reasonable; I have no method of testing actual FR, but a CD with test tones sounds pretty smooth from 400 on up, as designed.

My question is about imaging. Mono recordings stay very well centered and have nice 'depth", but on certain stereo recordings, the image seems to jump from one side to the other, depending on frequency. Is this the "problem" that some builders report with OB designs? Is it due to baffle width, driver orientation, or could it be Xover anomalies or inequalities? I did not carefully match components between channels, as this is only the prototype stage, but all components were within 5% or so of label values.

I really like the tonality, dynamics, and spatial qualities of the OB format, but I could never live with this wandering image. Please help me solve this problem.

Peace
Tom E

[Jed]

Did you compare the same recordings that don't image well on both monopole and dipole speakers?

If so, who did your crossover and was it properly designed?

[madisonears]

Jed,

These are familiar recordings that I've listened to on my (currently) primary speakers, B&W 802's with North Creek outboard Xovers, many times, and there is almost no image wandering. One problematic disc is Sonny Rollins "Way Out West" SACD which features old stereo type recording with sax completely on left channel, drums and bass on right. With these OB's, the sax sometimes, and only for a moment in the middle of a run of notes, seems to come from the right speaker! Also, on KOB, Miles' trumpet sometimes wanders off center into one or the other speaker.

My Xovers are somewhat primitive, designed by me carefully using textbook formula, built with Bennic film caps and Sidewinder coils. I did not use the "nominal" impedance, but tried to determine actual impedance at the crossover frequency by using manufacturer's and other's plots. The mids are running full range on the low end right now because I haven't hooked up my active Xover yet, and crossed 4th order at 2400 on the high end. The tweet is crossed 3rd order electrical, 4th order acoustic, at 2400. The frequency response seems pretty even, but I have no way to measure it accurately.

So is it "properly" designed? If by that you mean were the individual driver parameters measured, and then an experienced designer consulted and a complex circuit crafted, and results carefully measured with drivers mounted on baffles, and adjustments made, then, no, they're not properly designed. However, they are certainly not a haphazard mishmash, either. I'm sure they're not perfect, but the music sounds clear, sweet, open, dynamic, tonally realistic with no pronounced hot spots or dips. There is just a hint of brightness, but I firmly believe that is an artifact of listening to only the "top end" with declining bass below 400 and almost none at all below about 150hz until the sub kicks in at 80. You know: like listening with the woofers turned off! It is easy to hear that the Peerless tweeter is far superior to the B&W, and the OB mids sound more spacious and dynamic than their enclosed counterparts.

I just wish Sonny Rollins would stand still when playing that sax!

Peace,
Tom E

[ThomasW]

I just wish Sonny Rollins would stand still when playing that sax!

What do you want to bet a lot of it is a function of the dipole room placement and their interaction with acoustic characteristics of the room?

Also you might try removing the curved sides on the baffle. Magnepan who probably knows more about dipole baffle design than anyone else, uses completely flat baffle with a tapered edge with their higher end designs

[dlneubec]

I agree with Thomas' comments here. The Number 1 place I would look first is room placement. You may very well be getting relfections that are causing these issues.

John Kreskovsky has done a lot of dipole design and research and you will note, if you check out his website, that he does not use wings on his dipole baffles, other than perhaps for dipole bass. I believe he has suggested that baffle edge treatment, from a difraction standpoint, is less important in dipole speakers due to the figure 8 radiation pattern and natural cancellation that already occurs 90º off axis.

[Jed]

I don't think his crossover design considers things like phase alignment, since he was using textbook formulas for the design. Also, there are likely nulls in the FR because the rolloff from the baffle was not modeled into the final response. Just guessing. I can't really know what is going on with his system without a measurement.

Also, there is nothing wrong with using wings that extend off the baffle if it measures better. I recently did an experiment with my Tombstones design and found that a 1.5" extension considerably flattened the area around 1.5k with the SEAS excel, so making the generalization that all small U frames is not as good as a flat baffle needs further investigation. Not saying a flat baffle can't work equally as well, but there are many different ways to go about the open baffle design. In fact, I'm leaving the top end open on the top of the Tombstones, instead of putting an extension there... reason?... because it measured better. There is a considerable amount of experimentation to get this right and measuring equipment is a mandate. Tuning by ear just won't cut or come even close.

[Paul W]

I've built several dipoles and have never noticed the image "jumping" between channels...vague imaging when too close to the front wall yes, but not movement. How close are they to the front and side walls and is wall treatment symmetrical?

Doubt baffle shape is causing it, but it could be something as simple as a crossover wiring error in one channel. A deviation in crossover optimization may cause spatial artifacts. As Jed indicates, measurements are invaluable.

[ThomasW]

I've noticed some instances of what he's describing. The issue was solved with diffraction control felt on the baffle, proper placement, and passive acoustic room treatments.

That said I think he needs to attend to adding in the woofer and using the design tools available on Jeff Bagby's website. These tools are a significant upgrade from a generic XO calculator. Real world measurements are of course the best way to go...

[MuaDibb]

I was researching the bass rolloff on dipoles the other day and found this to be interesting, with regard to wings, width and placement on baffle.

http://www.troelsgravesen.dk/OBS.htm

[dlneubec]

Also, there is nothing wrong with using wings that extend off the baffle if it measures better. I recently did an experiment with my Tombstones design and found that a 1.5" extension considerably flattened the area around 1.5k with the SEAS excel, so making the generalization that all small U frames is not as good as a flat baffle needs further investigation. Not saying a flat baffle can't work equally as well, but there are many different ways to go about the open baffle design. In fact, I'm leaving the top end open on the top of the Tombstones, instead of putting an extension there... reason?... because it measured better. There is a considerable amount of experimentation to get this right and measuring equipment is a mandate. Tuning by ear just won't cut or come even close.

Yes a U baffle can work (consider the Orion). Wings are simply a way of making the baffle effectively wider, while retaining a smaller profile from the front, however you have to concenred about resonances in the back if you get the legs of the U too big. You can do a pretty decent job of simulating the dipole response using The Egde and ABCDipole, but they certainly cannot take the place of measurements. In the BaSSlines project, for a 500 and 2500 crossover, the best dipole baffle width turned out to be 10" a the driver location for a 6.5" driver (probably cone dia of closer to 5" or so). In simulations there was virtually no null in the 1.5khz area, nor was there one in the measurements. Some of that, I believe is driver related and perhaps has to do with it's basket design and possible rear relfections off the basket. One reason I say that is another driver I tried in the exact same baffle, did show a dip at around 1.5khz that the B&C 6md38 does not have.

In terms of generalizations about open baffle/dipoles, I would put John's experience and knowlege in this area up there with anybody and do trust his generalizations as starting points. Of course, I'm pretty certain he would also say you must have measurement capability to do a dipole design correctly. If you don't have or use ABCDipole, you might consider taking a look at it. I think it's a good place to start.

I would guess that the OP's baffle width might be too large for 4" drivers, but that is dependent on where he crosses them over. The wings only make it wider.

[CraigJ]

One problematic disc is Sonny Rollins "Way Out West" SACD which features old stereo type recording with sax completely on left channel, drums and bass on right.

Peace,
Tom E

Hi Tom,
Actually, I'm listening to Sonny play his sax through the right channel of my headphones at this link (left channel off of ear):
https://www.last.fm/music/Sonny+Roll...hand?autostart In stereo, Sonny's sax dominates the left channel, but still plays from the right channel. Possible short somewhere?

Good luck,

Craig
edit: Perhaps you're listening more critically because @2+ minutes into "Way out West", Sonny starts moving, or at a minumum, on occasion, more sax sound comes from the right channel? Who's doing all the talking?

[Jed]

Of course, I'm pretty certain he would also say you must have measurement capability to do a dipole design correctly. If you don't have or use ABCDipole, you might consider taking a look at it. I think it's a good place to start.

Agree, especially with a complex baffle shape. Like curves etc on the top. :T

I have ABCdipole. A nice piece of software to get started for square/rectangular baffle.

Your 10" baffle is not at all like mine so why even think it would have a similar FR at 1.5k? I agree with your point that each driver is going to be different, which is why I stated this is true in my design with X driver. This is all the more reason to measure. Simulations will get you 90 percent there, then the other variables can kick in and throw it off a bit.. like driver basket reflections etc.

And I never suggested that the OP put wings on his design, only to say that the assumption that a flat baffle will be better than one with a small u frame is not always true (post 4). I think he should measure his speaker so we can see the truth. Otherwise quite frankly it's a waste of time to guess what is causing the problems. It's obvious it doesn't have a crossover that is designed to factor in even the linear predictions that can be made about dipole responses. John K's website is probably the best place on the internet to learn about dipoles. And Linkwitz's site. That's where I would start.

[dlneubec]

Jed,

Perhaps there is a better place for this discussion than in this thread and we should start our own thread on it. It doesn't seem like there are many meaningful discussion here these days, just the next build thread of the same designs. I would certainly like to have John K. and maybe Jeff Bagby weigh in, but I'm not sure they frequent here often. I think this warrants further discussion. Maybe Thomas can move this to a new thread. I do know that John does not use U-baffles. I know that Jeff did not in his design for Jim Salk, or in his prototype he made for himself prior to it becoming a design for Jim. Incidently, he used a 6.5" PHL pro mid and his baffle coincidentally 10" wide in the prototype he built for himself. I don't think Jon Marsh was planning to use wings on his higher end open baffle when that design was still active. Still, Linkwitz uses a U-baffle.

Here are my concerns with U-baffles. To my way of thinking they are a compromize done for visual reasons. Otherwise we could simply use a wider flat baffle, which accomplishes the same thing acoustically and I don't think would have the same drawbacks as the U, but may not be as nice aesthetically. Granted, this is a compromise one has to make in some cases.

With a U-shape, it seems to me that you are adding elements that can potentially compromize sound quality, especially if you consider reverberent sound and were to measure what comes off the back of the dipole baffle that eventually reaches the listener as reverberent sound. Each 90º angle of the U trim adds another edge that will cause diffraction. In addition, the inside of the U will most certainly reflect sound back toward the driver and however minute, that can't be a positive thing. the deeper the U, the more siginficant this relfection would be.

It seems to me that if U-baffles were not a comprimize to the sound quality off the back of the speaker (and hence the overall sound of the speaker), then the U shape could just as well be reversed with the inside of the U facing at the listener and have no negative affect. However, none of us would design a speaker that way, because we know it would cause all kinds of problems. Yet we assume it is ok to face the U to the rear because it gives us better looking on axis measurements. Measurements, BTW, which are probably done with a gated window designed to exclude reflected or reverberent sound from polluting the measurement.

To me, the U makes a little more sense where it is necessary for low end dipole extention, but much less in the midrange area above 500hz or so. I believe that is why John uses an active crossover boost at the low end of his dipole midrange drivers in order to avoid a U or a very wide flat baffle.

What are your thoughts?

[Davey]

Dan,

I think you might be inflating the "U"-baffle concept in regards to the Orion mid-baffle. Since the baffle width is extremely wide relative to the side pieces it doesn't fit into the same category as say the U-baffle woofer used by John K. to create a cardioid low frequency response. IMO. The "cavity" created (such as it is) is not for that purpose and doesn't result in that type of behavior.

SL adds the side pieces as a (small) structural and aesthetic feature and then fine tunes in an iterative process in conjunction with the rear radiation of the mid-driver to try an achieve the best polar response transition. The baffle width, driver rear radiation, etc, actually is an attempt to widen the polar response relative to what would be achieved with a closed box radiation so it will match better with the rest of the system. There is no attempt here to eliminate side lobes or create a null directly behind or shoot for a 6db increase on-axis, etc, etc.

If I'm reading between the lines correctly, I think your (Orion) concern is more about higher frequency reflections. I don't see that as being a U-baffle issue because of the frequencies involved here.

I don't see anything wrong with using a completely flat baffle and removing structure on the sides. Optimizing it would be a similar process. I note that John K. uses a completely flat baffle with the NaO's, but I bet he could have achieved just as good a result with a baffle shaped more like the Orion.

My two cents.

Cheers,

Dave.

[pedroskova]

Jed,



With a U-shape, it seems to me that you are adding elements that can potentially compromize sound quality, especially if you consider reverberent sound and were to measure what comes off the back of the dipole baffle that eventually reaches the listener as reverberent sound. Each 90º angle of the U trim adds another edge that will cause diffraction. In addition, the inside of the U will most certainly reflect sound back toward the driver and however minute, that can't be a positive thing. the deeper the U, the more siginficant this relfection would be.

It seems to me that if U-baffles were not a comprimize to the sound quality off the back of the speaker (and hence the overall sound of the speaker), then the U shape could just as well be reversed with the inside of the U facing at the listener and have no negative affect. However, none of us would design a speaker that way, because we know it would cause all kinds of problems. Yet we assume it is ok to face the U to the rear because it gives us better looking on axis measurements. Measurements, BTW, which are probably done with a gated window designed to exclude reflected or reverberent sound from polluting the measurement.

To me, the U makes a little more sense where it is necessary for low end dipole extention, but much less in the midrange area above 500hz or so. I believe that is why John uses an active crossover boost at the low end of his dipole midrange drivers in order to avoid a U or a very wide flat baffle.

What are your thoughts?

I'm no Jed, but I think you're overestimating the importance of the rear wave... fidelity-wise. I've been trying a number of different OB configurations and, while they all sound slightly different, I wouldn't say that a U baffle sounds any worse, or any better than a flat one. The most problematic one that I measured was an 18" flat baffle, with more bumps and peaks than narrower flat baffles and U frame configurations (I took it that the tweeter wasn't too pleased).

Compared to baffle diffraction, the diffraction of the rear wave by the driver's frame will swamp out any comparison with the frontal wave. Add the delayed reflections from the environment, and you have two completely different sources of sound. With the U baffle, the only thing that is a must to deal with is the resonant peak and where it falls.

My 2 cents...so far.

[Dennis H]

Hi Tom,
Actually, I'm listening to Sonny play his sax through the right channel of my headphones at this link (left channel off of ear):
https://www.last.fm/music/Sonny+Roll...hand?autostart In stereo, Sonny's sax dominates the left channel, but still plays from the right channel. Possible short somewhere?

Good luck,

Craig
edit: Perhaps you're listening more critically because @2+ minutes into "Way out West", Sonny starts moving, or at a minumum, on occasion, more sax sound comes from the right channel? Who's doing all the talking?

Bingo, it may be the track rather than (entirely) the speaker. Not saying the speaker is right but some of it may be on the recording.

Dan and Jed, about the flat vs. U thing, keep in mind the AC of the driver. If the AC is way behind the baffle, a bit of U may steer the null to the side where it should be. You can always use angled boards or foam/felt to eliminate resonance between the U wings.

[Undefinition]

I suspect two things:

1. The textbook crossover is causing dips or peaks in the FR--or, more likely-- phase anomalies, which would lead to lobing. To play it safe, you might want to go with 4LR slopes between the mids and tweeter. (This requires simulation) Did you try adjusting the XO point?

2. Disconnect the tweeters and listen to the recordings some more. Does the image "sit still"? I haven't used the HDS tweeter, but I am slightly reminded of some frustrations I had with the Dayton RS28a, which didn't have as good of dispersion as I would have liked. The image wouldn't necessarily "shift," but it did tend to feel lobsided if I sat a few inches to the left or right of center. Switching to a higher-dispersion tweeter fixed that for me.

Also, did you say which type of mids you were using? You don't necessarily NEED measurements; you can trace mfr's FR graphs and use that in simulation and get some decent results (that's how I did the first two versions of the Aethers).

Finally, I hate to be a wet blanket, but using premium XO parts is not going to help the speakers sound their best unless you have the actual XO values optimized first.

[Jed]

Well, it looks like I'm a bit late to the discussion. Good advice in the last posts and I agree with Paul in regards to the query by the OP.

I know what I saw with my measurements, and perhaps it has something to do with the AC, like Dennis mentioned. I should experiment more with some felt in the V of the "wings" to see if it further smooths the response. An interesting thread would be the recording of these results to show the improvements that can be made given X driver, but then as Dan already found out, it's not applicable to all drivers of the same diameter, despite what the model simulation suggests.

Trial and Error is still a part of good OB design. I'm sure it all can be explained in the math though, and for that it would be great to see Johnk from music and design explain or JeffB. However, I think we got some good responses above.

Take care all and Happy New Year!

[madisonears]

Thanks for all of your responses so far. In order to quell some of the debate over flat vs winged baffles, I will clarify the previous description of my design. Following are photos of the protos.

A few words on the drivers and baffles:

Tweeter: Peerless HDS 810921
Mids: Peerless HDS 830854

Baffles are fabricated from PVC sheet: one 1/8" layer and one 1/4" layer with a 1/16" layer of Dynamat laminated between them. The mids are surface mounted with the flanges slightly overlapping the tweeter flange, with another layer of Dynamat under the mid flanges. The backside of the mids cutouts is fully rounded with 1/4" radius. All mounting hardware is nylon, not metal. The tweeter is in a cutout of the top layer of PVC so it is flush with the baffle. In essence, it floats on its own mounting surface, isolated from the mids. There is no vibration transmitted from the mids to the tweeter, and the baffles themselves are nearly inert, with very little vibration from test tones or music. The "wings" are sections of 6" diameter PVC pipe, cut lengthwise, temporarily affixed to the baffle edges with duct tape (!). They will eventually be epoxied onto the baffles, further stiffening the panels.

There are no premium parts in the Xover. I use a 3rd order on the tweeter to achieve 4th order acoustic, but I am considering trying 4th order electrical if it will minimize this image problem. There is some stuff behind the speakers, so rear reflections may be disrupted, but the room itself is pretty well damped. I don't have reflection or image problems with the B&W's.

I have inserted my cheapo active Xover, set at 240. Without the lower frequencies present, the wandering image seems to have declined, although the spaciousness has been killed as well by the additional circuitry. Time to invest in a high quality active Xover, but that's for another thread.

I initially had the mirror image speakers orientated with the drivers close together (narrow edges inward), but I have now reversed the speakers with the drivers farther apart (wide edges inward). That also seems to have lessened the wandering, but there is still some lobing going on.

I also tried reversing polarity of the tweeters, and that definitely smoothed out the FR and and increased spaciousness. It did not affect the wandering image. I am leaving them reversed for now.

I am going to revisit the problematic recording of Sonny out West on my B&W's in order to determine if this wandering is really a distortion caused by the OB's.

Peace,
Tom E

[Paul W]

After seeing the baffles, I'm even more convinced it isn't a problem with the wings. Driver offset looks a little extreme but that probably isn't an issue either.

If it isn't the recording itself, or the physical environment, I'd look at phase tracking in the crossover. To do this well, you need to measure...otherwise you are essentially blindfolded. Active or passive doesn't change the need for measurements.

[dlneubec]

If I'm reading between the lines correctly, I think your (Orion) concern is more about higher frequency reflections. I don't see that as being a U-baffle issue because of the frequencies involved here.

I don't see anything wrong with using a completely flat baffle and removing structure on the sides. Optimizing it would be a similar process. I note that John K. uses a completely flat baffle with the NaO's, but I bet he could have achieved just as good a result with a baffle shaped more like the Orion.

I'm no Jed, but I think you're overestimating the importance of the rear wave... fidelity-wise. I've been trying a number of different OB configurations and, while they all sound slightly different, I wouldn't say that a U baffle sounds any worse, or any better than a flat one. The most problematic one that I measured was an 18" flat baffle, with more bumps and peaks than narrower flat baffles and U frame configurations (I took it that the tweeter wasn't too pleased).

Compared to baffle diffraction, the diffraction of the rear wave by the driver's frame will swamp out any comparison with the frontal wave. Add the delayed reflections from the environment, and you have two completely different sources of sound. With the U baffle, the only thing that is a must to deal with is the resonant peak and where it falls.

My 2 cents...so far.

OK, admittedly, I'm playing a bit of the devils advocate here to facilitate discussion.

It seems to me that one should start from what is theoritically ideal. Perhaps I'm wrong about what that ideal is, but it seems to me that the ideal open baffle dipole result would be a perfect match between front and rear radiation, the rear a mirror image of the front, with a perfect figure 8 radiation pattern and nulls at the side of the baffle. In that case, the reverberent sound would be as closely matched to the front wave as possible, which would seem to be the goal. It seems to me that the closer the reverberent sound matches the direct sound, the more true to the original recording it would be.

With that in mind, you would detract from the mirror image radation pattern with anything that impeeded the drivers rear radiation that was not present on the front baffle. Of course, driver basket and magnet design, etc. makes this theoreticl ideal impossible, but I'm certain that some drivers are much better than others in that regard. Still, introducing wings most certainly could introduce additional changes in the radiation pattern when comparing front to back, not to mention making additional resonances and reflections back at the driver possible. If the wings were not a potential detriment to the mid or tweeter sound quality, radiation pattern and diffraction signature of the rear wave, then it would logically be considered perfectly ok to have wings protruding out the front of the baffle. How many of us would design a speaker with 1-1/2" wings extending out from the front of the baffle, even if it is a mid only baffle? Why is it what we don't?

We know that changes in baffle egdes from square edges to roundovers, chamfers or facets can show up as significant in the FR, their frequency dependent on baffle size. We often go to great ends to reduce or spread out that diffraction signature. Why don't we give it a thought in regards to rear wave in an open baffle dipole, where wings would certainly create much more significant alterations in the rear FR than front baffle edge treatments do?

It also seem to me that changes from the ideal figure 8 pattern could negatively impact imaging and soundstage, which was the subject of the original post. I don't see how it could make the image unstable, like the OP is experiencing, but I can see where the depth, width, possibly height of the soundstage and placement of instruments in space could be altered and consequently degrade the performace.

Has anyone actually tested the same basic design comparing a winged baffle to a flat baffle of equivalent acoustic width to be sure that there is no sinificant effect on the soundstage. imaging and clarity of the speaker? It is easy to say that the driver basket reflections, etc. would swamp out any other issues, but has anyone actually tested that theory?

I'm certain there is ample reason that John chose to use flat baffles in his NaO's. I'd be willing to bet that he believes he can't get as good a speaker with winged baffles. I'll drop him a line and perhaps he will be willing to weigh in here. As some have said, maybe none of this is significant.

[Davey]

Dan,

Yes, I think a perfect figure-8 (symmetrical) radiation would probably be a good objective. Obviously, constructing something like that with conventional drivers would be more involved. Whether it would be worth it or not is the question.

http://www.linkwitzlab.com/models.htm#B (see figure d.)

I was chatting with Siegfried about this a couple of years ago and he feels the concept has real potential, but hasn't pursued it because of practical reasons.

Cheers,

Dave.

[Deward Hastings]

We know that changes in baffle egdes from square edges to roundovers, chamfers or facets can show up as significant in the FR, their frequency dependent on baffle size. We often go to great ends to reduce or spread out that diffraction signature. Why don't we give it a thought in regards to rear wave in an open baffle dipole, where wings would certainly create much more significant alterations in the rear FR than front baffle edge treatments do?

It also seem to me that changes from the ideal figure 8 pattern could negatively impact imaging and soundstage,

The difference (in importance of front v. rear diffraction) probably relates to the path . . . we listen directly to the front wave, and are quite sensitive to its FR at the listening position (and the extent to which it is different at different listening positions). The rear wave arrives through multiple reflections from multiple angles with multiple additional diffractions, both relative to the radiating source and the listener, so rear diffraction effects are mostly "averaged out" as part of the reverberant field.

"Imaging" is determined almost entirely by the front wave, because time-of-arrival and early amplitude differences are established by the front wave. "Soundstage" seems to have more to do with the reverberant field and the sense of "spaciousness" that it produces (particularly in front-wall reflections) . . . but again, the reflective (and delayed) nature of that sound averages out "diffraction signature" to the rear, just as it averages out directional cues.

[Rudolf]

To me the image shift clearly is the result of a wide dispersion dome tweeter mounted on a baffle that is much too wide for its size.
Look at these simulated radiation patterns for a similar 1" dome tweeter:

Image not available

It´s a total mess that must result in a multiple image shift for different frequencies.

[madisonears]

Rudolf,

Thank you for your response. I'm afraid you may have hit the heart of the problem. I am aware that the baffle width is quite large, perhaps excessive. I don't know how you determined it is 15", but that is quite close to the actual dimension. I arrived at that dimension by using the half-wavelength equals baffle width rule.

Can you help me out with this problem? What would you consider the minimum baffle width to achieve -3dB @ 400 hz? I tried narrower baffles and they seemed to attenaute the low frequencies.

Would you offset the drivers?

Would a waveguide help to flatten the FR by making the tweeter more directive?

I intend to raise the Xover point between mids and tweeter to 3k in order to avoid some of the wild response variations shown on your simulation graphs.

I really appreciate your assistance.

Peace,
Tom E

[ThomasW]

We used 1" dome tweeters mounted on the Arvo dipoles with their 15" wide baffles, that speaker had no issues with the image wandering.

[Undefinition]

Can you help me out with this problem? What would you consider the minimum baffle width to achieve -3dB @ 400 hz? I tried narrower baffles and they seemed to attenaute the low frequencies.

Would you offset the drivers?

Would a waveguide help to flatten the FR by making the tweeter more directive?

I HIGHLY recommend you play around with something like Jeff Bagby's Response Modeler. You can experiment with different baffle sizes, and driver locations and see immediate results in the FR graph. I usually do this before I start drawing plans. Occasionally things don't always turn out the way I'd like, but more often than not, by spending some time tweaking driver locations in the simulation first, I can get amazingly-flat tweeter response the first time I measure the raw drivers. (occasionally, I lose, and have to cut again :cry: )

http://audio.claub.net/software/jbagby.html

[Rudolf]

We used 1" dome tweeters mounted on the Arvo dipoles with their 15" wide baffles, that speaker had no issues with the image wandering.

Thomas,
I dont´t want to look through all those Arvo and Isiris threads again. But I´m under the impression that the baffles became narrower and the tweeter waveguides became wider through the development process.

Maybe that Sonny Rollins recording is especially prone to loudspeakers with bad off axis performance. If its directional information is completely pan potted - no time delay or phase shift between the channels - then the image should shift frequency dependent with the SPL ondulations.

[pedroskova]

Tom E., you've probably already tried this, but does swapping speakers side to side change "imaging" qualities? Does the problem with that one recording not change?

[Jed]

My car is veering off to the right. I notice the car has a flat tire. Why is it misaligned? Fix- put air in the tire. My speaker isn't imaging well, I notice it doesn't have an optimized crossover. Fix- design a new crossover.

My point is fix the known problems first and go from there.

[Rudolf]

I'm afraid you may have hit the heart of the problem. I am aware that the baffle width is quite large, perhaps excessive. I don't know how you determined it is 15", but that is quite close to the actual dimension. I arrived at that dimension by using the half-wavelength equals baffle width rule.

You told us about that 15" width in your first post.
Too many people try to START the response range of their baffles at the first dipole peak -avoiding the 6dB dipole roll off below. This will immediately lead to trouble with the off axis response.
Ideally the targeted response range of a baffle should END at the first dipole peak. This would demand EQuing the driver, but would keep the off axis response under control.

Can you help me out with this problem? What would you consider the minimum baffle width to achieve -3dB @ 400 hz? I tried narrower baffles and they seemed to attenaute the low frequencies.

The narrower baffles were perfect BECAUSE they attenuated the low frequencies!

Would you offset the drivers?

Yes. Remove the wing close to the drivers and listen if it makes a positive difference WRT imaging. If it does you may even put the tweeter to the very edge of the baffle (tweeter center 1" inside the baffle edge). For the KE 25 the response would look like this:

Image not available

Obviously you should have the baffle edge close to the drivers next to your ears.

Would a waveguide help to flatten the FR by making the tweeter more directive?

Yes. But at the same time you would get a rather different frequency response.

I intend to raise the Xover point between mids and tweeter to 3k in order to avoid some of the wild response variations shown on your simulation graphs.

Above 2k the midrange probably will only radiate with its center and not with the full cone. So you can´t escape your problem by changing the Xover frequency only.

[Paul W]

Another way to achieve identical left-right off-axis response is to center the drivers in the baffle. This may reveal more warts to be solved in the (sometimes more complex) crossover, but off-axis response is symmetrical. This is the path usually taken by SL and John k.

But in this case I agree with Jed...first things first.

[ThomasW]

Thomas,
I dont´t want to look through all those Arvo and Isiris threads again. But I´m under the impression that the baffles became narrower and the tweeter waveguides became wider through the development process.

You're thinking of the Isiris, it has a tapered baffle like the original Apogees.

The original Arvo was a MTMWW (1" dome no waveguide) it has a 15" wide straight sided baffle. Later Jon created 2 additional baffle designs, those you have posted on your website

[JohnInKY]

Interesting thread, however, having just watched "The Interpreter" 2004 on OB RD-75's on my main HT, which image dramatically with wings, and having listened to 192K mp3's, CD's, etc. of alot of updated '60's and '70's jazz and rock music, I'd have to say I hear much greater variation in "better" / "worse" imaging and spaciality from the source material than I ever hear due to the speaker profile.

Much of the older music images very well across a large "sweet spot", with no image drift, but lacks some of the "envelopment" of more recent work. However, alot of recently (last 10 years or so) music seems to be diffusely mic'd to perhaps "enhance" depth, soundstaging, whatever, but seems to scrifice the image quality of "you are there" I get in spades with NEO 6 surround HT or older "pre 80's" recordings...

Just an observation..

John L.

[dlneubec]

Another way to achieve identical left-right off-axis response is to center the drivers in the baffle. This may reveal more warts to be solved in the (sometimes more complex) crossover, but off-axis response is symmetrical. This is the path usually taken by SL and John k.

But in this case I agree with Jed...first things first.

Paul,

I would like to see a discussion of centering or offsetting drivers in terms of imaging and soundfield. Offsetting drivers get lots of discussion and is often sort of blindly recommended, but this is typically in regards to diffraction reduction, etc. and never about the differences in polar response. The focus is usually on making the on axis FR look better.

I really wonder how we might be negatively affecting imaging and soundstage by offsetting drivers, creating an assymetrical power response and an unbalanced soundfield.

[CraigJ]

Dan,

I have a feeling Rudolf will chime in here. It looks like his website has a lot of details on driver location, as well as simulations and testing. Also fun to see pictures of my speakers (Arvo Type II & III) on his site. 8O If only I could read german or get Babel Fish to work on more than a paragraph at a time.

Craig

[Rudolf]

I would like to lead your attention to some comments of S. Linkwitz regarding on and off-axis response:
First his recommendation for the due order of design steps. He clearly wants to get baffle dimensions and driver positions right BEFORE crossovers and EQ.

Second his thought about narrow baffles and imaging. The essence is in the last 5 lines IMHO. I add the graph of the response he finds best in the Phoenix:



I really believe that my second simulation is more in line with this than the first one I did in this thread.

SL and Kreskovsky both show that too wide baffles will lead to a lobing frequency response. You can quite easily simulate that with EDGE - even for any off-axis angle you want. I find this very important, because too many people restrict themselves to the on axis response. IMHO on-axis response is a quite irrelevant singularity. It´s 5°-30° you need to get right.

Last. I´m not addicted to asymmetrical designs. I would prefer symmetrical ones, but they would need different baffle width for different frequency ranges. Pushing drivers to the edge of a wide baffle is simply a means to achieve a workable radiation pattern at least to ONE side.

[Paul W]

Paul, I would like to see a discussion of centering or offsetting drivers in terms of imaging and soundfield.

Hi Dan,
Assuming symmetrical room reflections, I believe symmetrical polar response should result in more precise imaging. In other words, differing spectral content radiated to the left and right of a speaker is more likely to cause image ambiguity through differing content in room reflections...the degree of ambiguity dependent on the speaker, room, and listener sensitivity.

That said, I can see situations arising where asymmetrical polar response might create a better end result...for example, a speaker too close to a sidewall, narrow bandwidth room treatments, etc. It seems this would be very situation dependent and require careful planning. I always listen on-axis, but there may also be a case for off-axis listening...for example, speakers facing straight out from the front wall with the listener on the centerline (but off-axis from the speakers).

For the overall soundfield, I think it is generally accepted that we want the reflected soundfield to closely track direct radiation, perhaps with a smooth rolloff as frequency increases. I can't see a case for anything other than that, but the result could come from many different ingredients...main speakers, various room treatments, multichannel etc.

That's my 2 cents.
Paul

[Dennis H]

Ah yes, the old even polar response with an OB problem. As Rudolf said, narrow baffles keep the front and back waves correlated -- acting as a true dipole rather than dual monopoles. Rules of thumb distilled from a few threads here.

Cone diameter < 0.8 * shortest wavelength (Source: dipole/monopole power response graph from SL)
Baffle width < 2 * cone diameter (Source: JohnK's ABCDipole)

3K crossover --> max 3.6" cone and 7.2" baffle

With narrow baffles like that it's not necessary or even very possible to offset the drivers.

[dlneubec]

Paul,

That's generally my take on the symmetrical/asymmetrical polar response issue, but I wonder if anyone has done any study of the percieved changes in imaging and soundfield? It would be intersting to see what was percieved in controlled listening tests.

Dennis,

Agreed, that is what I found in doing the BaSSlines design. A smaller baffle gave the best response. John's 2x rule of thumb turned out to be pretty much right on. One of the reasons I decided to use a tapered design, was to allow for a narrower width for the mid than the woofer, for the tweeter than the mid, etc., something that might be of benefit to WMT designs in terms of imaging, that can't as easily be done with a WMTM or WMTMW, unless you use large facets. I also thought it would reduce on axis diffraction since the drivers are centered.

[Rudolf]

Cone diameter < 0.8 * shortest wavelength (Source: dipole/monopole power response graph from SL)
Baffle width < 2 * cone diameter (Source: JohnK's ABCDipole)

Baffle width <=2.2* cone diameter (Source: Ferikides et al. Sorry, German only.

Assuming symmetrical room reflections, I believe symmetrical polar response should result in more precise imaging. In other words, differing spectral content radiated to the left and right of a speaker is more likely to cause image ambiguity through differing content in room reflections...the degree of ambiguity dependent on the speaker, room, and listener sensitivity.

Since you don´t get symmetrical wall reflections from the inside and outside radiations of a speaker in most rooms, a symmetrical polar response would hardly reach your ear anyway. I would concentrate on the constant directivity of the inside radiation instead of symmetry. If your off-axis radiation pattern does not "mimic the on-axis response, but with decreased amplitude as the angle is increased" at least to one side, left-right symmetry is useless.

Just my 2 cents

[Jed]

I also thought it would reduce on axis diffraction since the drivers are centered.

Centered drivers on the baffle will not reduce on axis diffraction, rather it will create a larger ripple in the diffraction pattern on axis versus an optimized non centered layout (in general); however, that ripple with a centered baffle layout is symmetrical, so it can be filtered out with an appropriate crossover, thus resulting in a flatter response on and off axis, which would equate to a smoother polar response.

[Dennis H]

Dan,

Glad to hear your practical experience matches what the experts say. It's tougher trying to match a monopole tweeter to a dipole mid but maybe keeping the baffle as narrow as possible helps the tweeter's power response too.

[Jed]

It's tougher trying to match a monopole tweeter to a dipole mid but maybe keeping the baffle as narrow as possible helps the tweeter's power response too.

Tough to match dipole mids to monopole woofers too.

I like Linkwitz's dipole tweeter approach, where he uses another millennium tweeter firing the opposite way with reverse polarity.

[Davey]

however, that ripple with a centered baffle layout is symmetrical, so it can be filtered out with an appropriate crossover, thus resulting in a flatter response on and off axis, which would equate to a smoother polar response.

I'm not sure I'd agree with that. It's been my experience that if you "smooth" a diffraction effect on-axis it doesn't necessarily make the off-axis response better also. Sometimes it makes it worse.

If the baffle is not properly designed to begin with you can't fix it by EQ'ing.

Cheers,

Dave.