I understand that you don't understand. The radial distance to the edges is all that matters.
Try simulating on driver axis:
case 1: a 1mx1m baffle with driver (make it small ~2.5cm) centered
case 2: a 1mx1m baffle with driver 5cm from 2 edges (so 0.95 from other 2)
case 3: a 10cmx10cm baffle with driver centered
case 4: a 1.9x1.9m baffle with driver centered (1.9=0.95x2)
case 2 and case 3 look interestingly similar at HF
case 2 looks a lot like a combination of case 3 and case 4 doesn't it?
almost looks like 2 "baffle steps"So there must be 2 "launch areas", to use your (faulty) analogy.
If i understand you correctly then you are just confirming my hypothesis. There are two sources of disturbance created from the baffle. The baffle itself launching (case 4) and the edge diffracting (case 3).
I found the launching analogy in the loudspeaker cookbook i mentioned earlier.
Last edited:
I have played a lot with Edge and BDS. RonE is right. But even that is simplified because usually the baffle has 4 sides! It is fascinating to watch Edge response when we move the driver on the baffle! Tolvan Data
I am not really fond of modern narrow baffles on monopole loudspekers! This is close to acoustically ideal baffle in my yes.
What is the directivity as a function of frequency for such a wide baffle? Not constant for sure?
//
If the chamfer radius is similar to the wavelength of the relevant frequencies, the chamfer will help to smear out the ripples. However, the chamfers you see on most speakers are much smaller than this and therefore don't do a lot to the ripples in the frequency response. Siegfried Linkwitz has a lot of good explanations about this (and much more) on his website.
The thing is that the wavefront, when pictured as rays, encounters the baffle edge at a variety of angles. The further the wavefront travels before encountering the baffle edge, the larger an apparent radius it will see- because it's traversing the radius or chamfer over a longer length. Imagine a piece of string nailed to the center of the driver mount location, and dragging that string stretched along the roundover- the length of string touching the roundover increases as the length of string before encountering the roundover increases.
So smaller roundovers can be effective at a much lower frequency than one might expect.
Of course, I use 4" roundovers in my mains because I take the matter rather seriously (it's not really hard to correct for this known issue at all, why not solve for easily solved performance issues?)
No. Each radial distance is a discrete (delayed) source of sound that adds to the direct sound. It is the combination of all of them at the observation point that gives the response. The shape of the baffle and driver location only determines how much delay and how spread out in time the individual returns are. Chamfers and roundovers slightly modify the response by spreading out the diffracted energy over a longer time.
The driver near the corner in my example acts like it is on a small baffle and a big baffle, not because it is on two baffles, but because it has near and far baffle edges and the difference between the two distances is great enough that it can be seen (sort of) separately.
Ok, now i understand your example, thank you for clarifying!
But we are now talking about diffraction as a whole, the entire effect the baffle has on the end result. I want to know what happens if we take the edge out of the equation, and focus on how the face of the baffle is contributing.
If that were true then an infinite baffle would behave identical to a spherical baffle. Because neither of them have edges.
A dome tweeter has a flat faceplate limiting the design beamwidth to no greater than 180 degrees, at least at the top end, unless the unit is oriented vertically such as with an omnidirectional design.
It has been suggested that a sphere is the ideal concept baffle for a direct radiator and baffles run the gamut from this, to narrower as with a waveguide. There is a start, middle and end (throat, body and termination). The aim is to fix directivity (either constant or widening with frequency), and to guide it gradually from it's start value to its ultimate value. (Also no reason then that the driver should be placed anywhere but the origin)
A flat baffle is one point on the continuum where the start and middle can coincidently be flat, however there is no theoretical justification for holding the baffle walls at a constant angle like this unless constant directivity is the goal, and if it is, they should either be sized to maintain this over the design bandwidth, or compromise at some lower frequency and terminate gracefully.
Sometimes a box shape can make a good speaker, not all diffraction will ruin a speaker, it is a matter of preference. But is there justification for putting up BDS plots other than boxes are in common use. I think it misses the point of the thread.
It has been suggested that a sphere is the ideal concept baffle for a direct radiator and baffles run the gamut from this, to narrower as with a waveguide. There is a start, middle and end (throat, body and termination). The aim is to fix directivity (either constant or widening with frequency), and to guide it gradually from it's start value to its ultimate value. (Also no reason then that the driver should be placed anywhere but the origin)
A flat baffle is one point on the continuum where the start and middle can coincidently be flat, however there is no theoretical justification for holding the baffle walls at a constant angle like this unless constant directivity is the goal, and if it is, they should either be sized to maintain this over the design bandwidth, or compromise at some lower frequency and terminate gracefully.
Sometimes a box shape can make a good speaker, not all diffraction will ruin a speaker, it is a matter of preference. But is there justification for putting up BDS plots other than boxes are in common use. I think it misses the point of the thread.
Again you are wide of the mark. A sphere or a rounded edge is radiating continuously as sound wraps around it. It is not a no-diffraction shape, it is a distributed diffraction shape.
I tire of this whack-a-mole with your misconceptions. Have fun with your learnings.
Sorry if i am oversimplifying things, but i am trying to create a hypothetical queston that you keep whack-a-moling. This is not a battle for wits.
On an infinite baffle you will not loose sensitivity because of baffle-step. No ripples.
On a spherical baffle you will start loosing sensitivity at a higher frequency point then any other shape. No ripples.
On a 1m*1m you will start loosing sensitivity from a certain frequency. But you will also have ripples because of sudden edges.
Now picture a 1m*1m baffle, added a 1m radius roundover, and with the driver centered. Will it it start loosing sensitivity at a lower frequency point then with the spherical baffle? Will there be ripples?
Good explanation, i havent thought of comparing it to a waveguide, which is basically what a baffle it is. The termination will be the edges then?
Baffle size and shape do not affect directivity of a driver (except for horns). The diameter of the driver and radiating surface topology (dome, planar, concave) make that.
In general, I don't know if there is a simulator for round/bended/multiform baffle contour. BDS can not use other than rectangular baffle (4 corners) and Edge an not use rounded edges.
Something worth knowing, By Dr. Wolfgang Klippel https://www.klippel.de/
https://www.klippel.de/fileadmin/kl...ure/Papers/KLIPPEL_Sound_Radiation_Poster.pdf
https://www.klippel.de/fileadmin/kl...ture/Papers/KLIPPEL_Cone_Vibration_Poster.pdf
https://www.klippel.de/fileadmin/kl...ure/Papers/KLIPPEL_Sound_Radiation_Poster.pdf
https://www.klippel.de/fileadmin/kl...ture/Papers/KLIPPEL_Cone_Vibration_Poster.pdf
PSD-Lite uses the same algorithm for rounded edges as the original BDS program written by Paul Verdone. Here is a link to the BDS manual that explains the weights used by the ray-tracing algorithm for the rounded edges. Paul's assumptions for those weights make some sense, and they could be modified for chamfers--that's left as an exercise for the reader. PSD-Lite includes a simple drawing program that allows up to 20 corners for both the front view and side views. So it is between BDS and Edge in capability, but I would like to think it is easier to use than either of those programs.
These ray-tracing programs have been shown to be accurate for the simple baffle case and they all give comparable results at the same summation point. However, to my knowledge the roundover algorithm hasn't been extensively tested, so I would not assume the model is fully validated. I believe some testing was done on the roundover results, but I've never seen any data on the model accuracy for that feature.
PSD-Lite is available here: http://www.audiodevelopers.com/Software/PSD_Lite/setup.exe. It requires an older version of the .NET framework and it uses a Microsoft database for the driver T-S parameters, so it might take a while to install. But it is all "safe" stuff, and it will uninstall properly.
This isn't correct. The radiation from the edges will cause interference that will affect the radiation pattern and the loudspeaker directivity. However, it's difficult to see how the baffle affects the directivity because it is usually not easy to move the summation point and understand the nulls and peaks as a function of listener location. And that's where these simple ray-tracing algorithms for baffle diffraction are very limited: they only show the summed amplitude at a single point in space.
The complete 3D solution requires a more robust CFD algorithm that isn't available in the freeware modeling programs. I've been tempted to try extending the ray-tracing program to show multiple points in a 3D view, but the number of calculations increases dramatically for the 3D solution, and there are probably better approaches that would be more computationally efficient.
Last edited:
Termination is about addressing the gap between where something is and where it is going. Edges and curves are kind of the same thing in different amounts. It depends on what you are trying to do, eg. how far is it to full space, how well is a given frequency already supported by the baffle and is it in a hearing sensitive range.
Additionally for what it's worth, the LeCleach horn is interesting as an example of where termination begins at the driver an happens gradually over the entire length.
Last edited:
^Thank you Neil for remarks and info! Lack or inadequacy of simulations of complex baffle forms is actully just more likely a challenge and quest to make a prototype and do measurements of it! A turntable is needed for off-axis measurements.
To skogs, the edge of a baffle is not "dead ending", think of it as a curve that shoots some of the energy to diffractions. This is wavelength-dependent, high frequency "drives faster" and throws more dirt to bushes!
Rally Finland: VW race Kamaz over Ouninpohja stage jump
To skogs, the edge of a baffle is not "dead ending", think of it as a curve that shoots some of the energy to diffractions. This is wavelength-dependent, high frequency "drives faster" and throws more dirt to bushes!
Rally Finland: VW race Kamaz over Ouninpohja stage jump
An externally hosted image should be here but it was not working when we last tested it.
I consider a roundover a quarter round, and you can't give more than a 0.5m roundover on a 1m wide baffle. Even a 1mx1m baffle cut out from a 2m diameter sphere will still have edges.
Look at olson's original measurements of a number of shapes. You will find ripples on all of them. Be careful not to put too much importance on them, because almost all of them are a very small driver on square or circular baffles with driver in geometric center, so worst case.
I think it would be useful if, rather than coming up with a bunch of random ideas, you read the references I gave in post 7. The pdf Neil links also has some good info. There is a limit on what can be explained here without specific questions. If you formulate a hypothesis without knowing the background, it probably won't be worth much.
ANY baffle has diffraction; whether the response is smooth or not is another story. You don't even have to get very complex with baffle shape (like the 1x1m punched out of a 2m diameter sphere I mention above) before the response is non-trivial to calculate. If your driver is in the geometric center, it will have ripples even in that. At 1m wide, they will start around 300Hz....
I've got copies of most of those references, but I found the simple explanation by Andy Unruh to be most useful when I wrote the Baffle Diffraction module for PSD-Lite: Understanding Cabinet Edge Diffraction
Also, there is a help file for that module in PSD-Lite that explains the math I used. There's enough info there to get you started writing your own model.
Here is a direct link to the online help file: PASSIVE SPEAKER DESIGNER . The math really isn't that complicated, but there is a lot of it, because something like 4 million rays are calculated and summed for a typical woofer, every time the driver is moved or box dimensions change.BTW, one of the most difficult challenges I had in PSD-Lite was figuring out the interior corner points of an enclosed polygon as a function of wall thickness. It's not as easy as you might think! You need that algorithm to calculate the edge points for roundovers and chamfers. If you hit the "show interior" button and move around some corners or change the wall thickness you can see that algorithm in action. Unfortunately, the program doesn't actually show the edge points used for the roundover calculations--I should probably add that in someday.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.