Designing to account for baffle step remains a very elusive subject for me, particularly with regard to unconventional enclosure shapes (extremely narrow enclosures, spherical shapes, teardrop shapes, etc...). As far as I can see, we take response measurements of the design once it's been built, and try to compensate from there, post-build. However, is it not possible to accurately or somewhat accurately predict and model what the low frequency drop will be according to enclosure shape, dimensions, and mounting of the driver? I'm completely missing locating such information. Perhaps others can direct.
Hi,
There are tools that will predict baffle step and ripple. Narrow
certainly but I don't think too good with curves and teardrops.
http://audio.claub.net/Simple Loudspeaker Design ver2.pdf
FRD Consortium tools guide
Designing Crossovers with Software Only
rgds, sreten.
Zaph|Audio - ZA-SR71
Baffle diffraction sim, bassmid and centered tweeter vs offset
There are tools that will predict baffle step and ripple. Narrow
certainly but I don't think too good with curves and teardrops.
http://audio.claub.net/Simple Loudspeaker Design ver2.pdf
FRD Consortium tools guide
Designing Crossovers with Software Only
rgds, sreten.
Zaph|Audio - ZA-SR71
Baffle diffraction sim, bassmid and centered tweeter vs offset
Thank you sreten. Until software designers take it to the next level and cease to assume only box shapes are relevant, I'm thinking there must certainly be calculations around that can be made which account for how a typical low frequency loss for a given baffle size might transfer to a perfect sphere. Front mounting or recessed might also result in more dramatic changes on the sphere, I'm guessing.
I started downloading every free computer program that I could find that deals with enclosures, crossovers, baffle step calculators, T/S parameters, passive radiator calculators, etc... and didn't come close to opening half of them, let alone understand those that I did look. Keep it simple. Build a basic crossover for whatever enclosure you want/like. Then fine tune.
May not be scientific, but, you choose the shape, not a computer. Your ears can hear something good that your eyes also like. There's just no magic formula. Enjoy both senses.
May not be scientific, but, you choose the shape, not a computer. Your ears can hear something good that your eyes also like. There's just no magic formula. Enjoy both senses.
Baffle step involves wavelengths, which vary with frequency, proximity to room boundaries, and the size of the front panel of the speakers. Only the latter is perdictable. Speakers placed very close to walls will have a very different need for correction than speakers placed a few feet away from walls. If you know where they will be placed in a room, you can take a guess at what correction would be optimal. It's rarely an exact science. Personally, I rely on my four section Baxandall tone circuit to help me fine tune it in by ear.
You require a program that solves the 3D acoustic wave equation for external 3D geometries. There are plenty of commercial ones and one or two freely available ones that usually require a bit of knowledge to use.
sreten, yes, if you're referring to those early diffraction studies by Harry F. Olson, then I recall the diagrams of large planetary-like orbs with relatively small drivers in comparison. Indeed I am only interested in the more 'truncated' shapes, i.e. the size of the driver relative to the enclosure size being much greater, similar to, for example, the spherical section of B&W 800 speakers. I believe here, they're using the resulting baffle step of this form to integrate the mid unit physically rather than purely through the crossover, or to assist in that somewhat.
I use minidsp, infinitely more flexible and easier than tweaking hardwired x-over units. Ideally four settings would be offered, somewhat similar to Genelic monitors: wall, corner, desk, free. Such a feature for speakers with digital crossovers might as well become a standard.
andy19191, what are these programs you refer to, even though user-friendly they may not be?
The baffle step and baffle step compensation is really not something which I find presented in an intuitive easy-to-grasp fashion anywhere really. In the designs that I'm attempting, it's particularly relevant, but for most rectangular designs, following roughly a gold ratio-esque construction, it seems to be much less of an issue, particularly with flush mounting.
A spherical design with a large driver relative to enclosure is perhaps not at all an ideal arrangement in this department. Am I seeing this correctly - more low frequency response goes omni at a higher frequency, and thus low frequency response appears to drop off dramatically in the listening position. We then try to compensate for this by choking down some of the mids and higher frequencies to restore a flatter, more balanced response in the listening position. This involves losing a great deal of total efficiency, and drawing more power.
I use minidsp, infinitely more flexible and easier than tweaking hardwired x-over units. Ideally four settings would be offered, somewhat similar to Genelic monitors: wall, corner, desk, free. Such a feature for speakers with digital crossovers might as well become a standard.
andy19191, what are these programs you refer to, even though user-friendly they may not be?
The baffle step and baffle step compensation is really not something which I find presented in an intuitive easy-to-grasp fashion anywhere really. In the designs that I'm attempting, it's particularly relevant, but for most rectangular designs, following roughly a gold ratio-esque construction, it seems to be much less of an issue, particularly with flush mounting.
A spherical design with a large driver relative to enclosure is perhaps not at all an ideal arrangement in this department. Am I seeing this correctly - more low frequency response goes omni at a higher frequency, and thus low frequency response appears to drop off dramatically in the listening position. We then try to compensate for this by choking down some of the mids and higher frequencies to restore a flatter, more balanced response in the listening position. This involves losing a great deal of total efficiency, and drawing more power.
The computer programs engineers have used to simulate sound fields since the 1960s. Commercial ones tend to work smoothly. Freely available ones are mainly derived from research codes and usually require knowledge about grid generation, boundary conditions, file formats and the like. Here is an example but there are quite a few others that may be better or worse depending on your knowledge.
Basta is a relatively cheap simulation program. Edge is a simple freeware baffle diffraction (eg. baffle step) Simulator. Homepage here Tolvan Data
Last edited:
My understanding is that a sphere is no different than any other shape in terms of baffle step loss - all of them exhibit a 6dB loss in free space starting at a point that is dependent on their dimensions (width or radius). Where they differ is in the amount the baffle affects the response above the baffle step. A sphere produces the least diffraction effects above that frequency compared to any other shape. See examples below. So regardless of whether a sphere of a given size has a large or small driver in it, the baffle step loss will still be the same.
There are 3 free diffraction programs that are commonly used that I am aware of. sreten has linked you to 2 of them. The other is The Edge - Tolvan Data. Each one can handle slightly different parameters but none of them work much beyond 2 dimensional shapes, meaning that none of them can model a sphere. The Edge will do a circular baffle (or just about any other 2-D shape you can come up with) but if you give it a try, you'll see that the diffraction simulated won't be anything like that of the sphere below. For that, I really do think that you are stuck with building and then measuring to get accurate results. Ballpark though, I would think the baffle step loss frequency for a sphere with the same diameter as a box's width should be pretty close (but not the rest of it).
Plus don't forget that even for something like the B&W spheres, the mid's diffraction signature will also be affected by the rest of the cabinet as well.
There are 3 free diffraction programs that are commonly used that I am aware of. sreten has linked you to 2 of them. The other is The Edge - Tolvan Data. Each one can handle slightly different parameters but none of them work much beyond 2 dimensional shapes, meaning that none of them can model a sphere. The Edge will do a circular baffle (or just about any other 2-D shape you can come up with) but if you give it a try, you'll see that the diffraction simulated won't be anything like that of the sphere below. For that, I really do think that you are stuck with building and then measuring to get accurate results. Ballpark though, I would think the baffle step loss frequency for a sphere with the same diameter as a box's width should be pretty close (but not the rest of it).
Plus don't forget that even for something like the B&W spheres, the mid's diffraction signature will also be affected by the rest of the cabinet as well.
Attachments
Great resources to know of, thank you. I have built the enclosure both in spherical form and one of the same volume in conventional rectangular form. Comparing them side by side, they perform very differently, and that is quite audible without measurements. There is a great loss of low frequency response with the spherical one. It's diameter matches the height of the rectangular baffle, and exceeds its width. It was hoped that such dimensions would prevent loss due to baffle step. Perhaps complicating matters is the fact that I'm unable as of yet to recess the driver on the spherical and it indeed protrudes a bit far mounted on an additional flange affixed to the sphere (or rather more teardrop shape, to be a bit more exact).
"My understanding is that a sphere is no different than any other shape in terms of baffle step loss - all of them exhibit a 6dB loss in free space starting at a point that is dependent on their dimensions (width or radius)"
It is the 6dB loss "starting at a point that is dependent on dimensions" that is of interest to me, hoping to better understand and define that point and the dimensions, along with the impact of recessed or flush or top mounting, which my senses are telling me, can have an even more dramatic impact on these relatively small truncated spheres than along a flat baffle. The dimensions of a spherical baffle do make quite a difference in terms of low frequency loss it seems, and where it is generally known what the dimensions of that baffle need be in rectangular box for a given driver to avoid a major loss of low frequency at the listening position, for a sphere, this is less clear. Using an example like Olson's, we get minimal loss, but those massive spheres relative to driver size are not the target design.
It's probably quite clear now, I'm not concerned about diffraction signatures in this case, but rather the loss of low end.
"My understanding is that a sphere is no different than any other shape in terms of baffle step loss - all of them exhibit a 6dB loss in free space starting at a point that is dependent on their dimensions (width or radius)"
It is the 6dB loss "starting at a point that is dependent on dimensions" that is of interest to me, hoping to better understand and define that point and the dimensions, along with the impact of recessed or flush or top mounting, which my senses are telling me, can have an even more dramatic impact on these relatively small truncated spheres than along a flat baffle. The dimensions of a spherical baffle do make quite a difference in terms of low frequency loss it seems, and where it is generally known what the dimensions of that baffle need be in rectangular box for a given driver to avoid a major loss of low frequency at the listening position, for a sphere, this is less clear. Using an example like Olson's, we get minimal loss, but those massive spheres relative to driver size are not the target design.
It's probably quite clear now, I'm not concerned about diffraction signatures in this case, but rather the loss of low end.
Last edited:
I get the impression that the you might not be clear on the mechanics behind BS.
What is happening is as the baffle reduces in size, its ability to support lower frequencies is reduced. BSC is an attempt to lower the spl of the frequencies (wave lengths) that the baffle can support. These are radiated forward with a perceived increase in spl as the frequency rises. As the frequency rises, more of the wave length is supported "on baffle" ... so you hear a greater proportion of those frequencies. The lower frequencies aren't "lost" ... it's just that they are not fully directed forward to your ears at the listening position. To regain that "wrap around the cabinet" loss of spl you need to increase the baffle dimensions such that the baffle can hold/support longer wave lengths.
Diffraction is a relative to how kind a given baffle shape is to acoustic wave lengths. It only to do with the period where those wave length begin to leave the baffle and "diffract" off its edges.
What is happening is as the baffle reduces in size, its ability to support lower frequencies is reduced. BSC is an attempt to lower the spl of the frequencies (wave lengths) that the baffle can support. These are radiated forward with a perceived increase in spl as the frequency rises. As the frequency rises, more of the wave length is supported "on baffle" ... so you hear a greater proportion of those frequencies. The lower frequencies aren't "lost" ... it's just that they are not fully directed forward to your ears at the listening position. To regain that "wrap around the cabinet" loss of spl you need to increase the baffle dimensions such that the baffle can hold/support longer wave lengths.
Diffraction is a relative to how kind a given baffle shape is to acoustic wave lengths. It only to do with the period where those wave length begin to leave the baffle and "diffract" off its edges.
There is no way to prevent baffle step loss in free space (4pi) regardless of shape. The only way to do so is to return to 2pi space which is something like an in-wall or infinite baffle speaker.
However the loss in 4pi can be mitigated by effectively reducing that space by using very wide baffles, putting the woofer closer to the floor, and/or moving the speaker closer to the side and back walls. Jeff B's diffraction program can show you these effects.
But still the loss for any shape will be limited to 6dB. So to say that you are getting more loss from a sphere than a rectangle doesn't make sense unless:
1. it is a subjective effect of a larger relative difference between the bass and the mid/highs, or
2. the box is getting some boundary reinforcement that the sphere isn't.
I'm assuming you are comparing the exact same driver.
Can you not do what B&W does and use the sphere for the mid and something else for the bass? Cross over near the mid's baffle step loss frequency and it will help the xo function. Use 2 woofers in parallel for the bottom end if you can and you'll get the 6dB loss back with the 2nd woofer (although I suspect you already know that).
However the loss in 4pi can be mitigated by effectively reducing that space by using very wide baffles, putting the woofer closer to the floor, and/or moving the speaker closer to the side and back walls. Jeff B's diffraction program can show you these effects.
But still the loss for any shape will be limited to 6dB. So to say that you are getting more loss from a sphere than a rectangle doesn't make sense unless:
1. it is a subjective effect of a larger relative difference between the bass and the mid/highs, or
2. the box is getting some boundary reinforcement that the sphere isn't.
I'm assuming you are comparing the exact same driver.
Can you not do what B&W does and use the sphere for the mid and something else for the bass? Cross over near the mid's baffle step loss frequency and it will help the xo function. Use 2 woofers in parallel for the bottom end if you can and you'll get the 6dB loss back with the 2nd woofer (although I suspect you already know that).
That's an excellent description by puppet. I also found the following helpful when I was starting out:
True Audio Speaker Topics: Spatial Loading
True Audio TechTopics: Diffraction Loss
True Audio Speaker Topics: Spatial Loading
True Audio TechTopics: Diffraction Loss
Hi,
You will inevitably get ripple above and beyond
just baffle loss with a driver in a reasonably
sized sphere, i.e. truncated or on a flange.
Such arrangements will raise the transition frequency,
and cause higher apparent bass loss, though the low
bass is unaffected, compared to a simple box.
Worst is the driver on the end of a same diameter pipe.
rgds, sreten.
You will inevitably get ripple above and beyond
just baffle loss with a driver in a reasonably
sized sphere, i.e. truncated or on a flange.
Such arrangements will raise the transition frequency,
and cause higher apparent bass loss, though the low
bass is unaffected, compared to a simple box.
Worst is the driver on the end of a same diameter pipe.
rgds, sreten.
Thanks all. jReave, I indeed assumed that when the driver is the same in both enclosures, and the position of the speakers the same, the answer why the two identical volume enclosures sound different in the listening position is, in your answer #1, the "subjective effect of a larger relative difference between the bass and the mid/highs", due to the baffle step. That's what wish to understand more fully. I suppose it was inaccurate to speak of lost bass when what we really mean is the relative difference of mid/highs to bass, as heard in the listening position. I did not mean to imply an understanding that it was lost in any absolute sense, but rather simply not heard in the listening position, as it was diverted to the rear by the absence of a larger baffle.
The amount of perceived loss in the listening position would depend on the size of the baffle, which is variable, and therefore the perceived decrease in the listening position would be a range of anywhere from 0dB–6dB, would it not?
The speakers I'm working with are not to be ideal B&W 800 copies. I wish to retain as much low end as possible within a small form factor, specifically the same volume as the rectangular enclosure. And indeed the spherical one is the same volume. So, given all that, it appears that it is the size of the sphere's baffle that is leading to the relative difference of mid/hights to bass in the listening position. Might that be correct?
The amount of perceived loss in the listening position would depend on the size of the baffle, which is variable, and therefore the perceived decrease in the listening position would be a range of anywhere from 0dB–6dB, would it not?
The speakers I'm working with are not to be ideal B&W 800 copies. I wish to retain as much low end as possible within a small form factor, specifically the same volume as the rectangular enclosure. And indeed the spherical one is the same volume. So, given all that, it appears that it is the size of the sphere's baffle that is leading to the relative difference of mid/hights to bass in the listening position. Might that be correct?
Since passive baffle step EQ would reduce the overall efficiency of the speaker by as much as 10dB (depending on a bunch of variables), it could be argued that you'd only want to use passive baffle step EQ on a speaker that was going to be placed very close to a wall so you wouldn't need much correction. In the case of tower speakers that will be out from any walls by a few feet or more, especially where you have a very narrow baffle for the sake of better stereo imaging, it seems like you'd want to do the baffle step correction actively, ahead of the power amps, and once you've decided to build up a chassis and power supply to do that, you might as well do the crossover actively as well.
Hi,
No, and your lack of ability to listen is getting very tedious.
Clearly for two speakers, both with indentical 6dB baffle step,
but one centred on 500Hz, and the other centred on 250Hz
the latter will sound like it has more bass, not that it matters
really if you correct baffle step correctly for both cases.
A low diffraction / low ripple design is always going to
have a high transition frequency and there is nothing
you can do about it, there is no bass you can save.
rgds, sreten.
Right, so I stand corrected about a sphere and box of same diameter and width (respectively) having about the same baffle loss transition frequency. The box has a much larger diffraction peak in the midrange that moves the transition frequency lower when compared to the low ripple sphere diffraction which has a higher baffle step -3dB point. So the box still doesn't have any more bass but it does have more midrange. See graphs above again.
But I'm pretty sure that just like increasing the width of a box lowers the baffle step -3dB point, so will increasing the sphere's diameter do the same. But sreten is still right, it doesn't really matter - in both cases (well all cases), the mids/highs still need to be brought down to match the same 6dB baffle step loss in the bass in all these situations (unless you're getting some boundary or baffle reinforcement and you only need 3 or 4 dB compensation).
Again, I think you're going to have to measure to get the xo right. But personally, if I was trying to do something similar to the B&W 800's in a smaller size (which strangely enough, I am actually doing at the moment), I would still use the sphere for just the mid (smaller mind you) and still use a smaller cab for the woofers. That's just me of course.
But I'm pretty sure that just like increasing the width of a box lowers the baffle step -3dB point, so will increasing the sphere's diameter do the same. But sreten is still right, it doesn't really matter - in both cases (well all cases), the mids/highs still need to be brought down to match the same 6dB baffle step loss in the bass in all these situations (unless you're getting some boundary or baffle reinforcement and you only need 3 or 4 dB compensation).
Again, I think you're going to have to measure to get the xo right. But personally, if I was trying to do something similar to the B&W 800's in a smaller size (which strangely enough, I am actually doing at the moment), I would still use the sphere for just the mid (smaller mind you) and still use a smaller cab for the woofers. That's just me of course.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.