Ameliorating the diffraction problem completely, home audio

[thadman]

I was contemplating how to make the perfect baffle (ie no diffraction effects) and began reading the information on Linkwitz' website and found this quote.

The benefits of edge rounding come into play only when the radius is greater than 1/8th wavelength. Thus a typical 1/2 inch radius begins to diffuse the diffracted wave at frequencies above 3.4 kHz, but will decrease in relevance at higher frequencies, when the driver illuminates less of the edge due to its increasing directivity. With most speaker cabinets the radius or chamfer is acoustically too small and is primarily cosmetic.

How effective is a baffle with no edges (sides completely rounded over) at removing diffraction artifacts? I want to incorporate front and rear firing ribbon tweeters and dome midranges similar to what Linkwitz has done with his Orion setup and the front/rear firing Seas Milleniums. For reference, the Aurum Cantus G2si/NeoCD3 are <3.5" deep and the Dayton RS52 is 5.125" wide, so for simplicities sake lets entertain the idea of a baffle with a depth of 8 inches and effective "flat" width of 6 inches. For the roundover to completely consume that edge, a radius of 4" and a 4" roundover bit would be required. We now have an effective 14" wide by 8" deep baffle (4" radius + 6" baffle front + 4" radius=14")

radius=1/8 lambda, 4=1/8 lambda, lambda = 32". 13,397in/s divided by 32in=418.65hz. Thus the 4" roundover will begin to diffuse diffraction at 418.65hz.

Baffle width=14", 13,397in/s divided by 14"=956.93hz. Thus the panel will switch from 2pi to 4pi radiation ~956.93hz, ~1.3 octaves above where the effects of diffraction are being diffused by the roundover.

Here is a cross-section of the baffle (top view)

An externally hosted image should be here but it was not working when we last tested it.

Opinions?

(I'm aware that a 4" roundover bit is impractical and I've already thought beyond that. I could use a jigsaw to cut the specific baffle shape several times...ok maybe more than several times and layer them so that they ascend to a certain height)

 

[DSP_Geek]

Would this help?

http://www.rockler.com/product.cfm?Offerings_ID=10569&TabSelect=Details

 

[Aengus]

thadman said:

I'm aware that a 4" roundover bit is impractical

If you are going to stack-laminate it ("layer them so that they ascend to a certain height") then cut it very carefully once, then:
- cut the next layer roughly to shape and slightly oversize (~1/8")
- glue or otherwise fasten it on
- use a router with flush-trim bit to cut it to the shape of the template (i.e., the first carefully-cut one).

This technique works well for stack-laminated plywood or MDF.

Regards.

Aengus

 

[Andy G]

http://members.optusnet.com.au/~gradds55/ARGOS/Stacks.htm

 

[liasom]

I saw Norm do a thing on New Yankee Workshop to turn two identical half-round ornaments on a lathe to frame a mirror. It involves gluing up a wood/paper/wood sandwich. However I'm completely uncertain if it will scale up to what you need to accomplish.

If you have access to a wood lathe that will handle the large diameter you might could glue 2 rectangular , e.g. 4.05" x 8.1", cross-section pieces together to make a 8.1" square cross section piece. Use regular yellow wood glue and some heavy kraft paper sandwiched all in between them down the length dimension (heightwise in baffle terms) of the piece. Then, after easing over the corners with a table saw, use the lathe to turn an 8" diameter cylinder. After turning, the two halves could be separated by prying them apart, thus tearing the paper, and you'd have two half-round cross section pieces with 4" radius and whatever length the lathe could turn. Sand the paper off before gluing them to either side of your rectangular baffle.

 

[CeramicMan]

Hi Thadman,

Yeah I think it's a great idea! I had a similar thing in mind here, whereby stacked layers enable you to create an intricate internal layout as well.

[edit: the image didn't want to paste, but it's that link about halfway down the page.

In terms of timber consumption it would be quite wasteful and time consuming - especially if you want to give the box a compound curve rather than just a cylindrical shape. But since the design isn't constrained to simple straight lines, you might as well make something more complicated.

 

[croat47]

Or...

http://aitwood.com/StoreFront.Asp?WoodType=POPLAR&CATID=10&Section=HALFCYL&wDesc=Half Cylinders (180%20degrees)

I think there is a similar idea for a large radius edge on an OB floated in the "Beyond the Ariel" Thread...can't find it just now.

 

[Andy G]

http://members.optusnet.com.au/~gradds55/curved_loud.htm

 

[Svante]

Rounding the edges of a baffle does not eliminate the reflection at the baffle edge. A rounded edge can be seen as very many smaller edges, and each of these will result in a small reflection. The sum of these reflections is equivalent to the single reflection of a single edge, but it is smeared in time (which is good).

The only way to eliminate baffle diffraction is to put the loudspeaker "in-wall". Or to put it in vacuum...

 

[planet10]

affle width=14", 13,397in/s divided by 14"=956.93hz. Thus the panel will switch from 2pi to 4pi radiation ~956.93hz, ~1.3 octaves above where the effects of diffraction are being diffused by the roundover.

That may be where BS starts but we usually consider the -3dB point, 4560/14= 326 Hz.

dave

 

[thadman]

Rounding the edges of a baffle does not eliminate the reflection at the baffle edge. A rounded edge can be seen as very many smaller edges, and each of these will result in a small reflection. The sum of these reflections is equivalent to the single reflection of a single edge, but it is smeared in time (which is good).

The only way to eliminate baffle diffraction is to put the loudspeaker "in-wall". Or to put it in vacuum...

Are the negative effects of diffraction removed? How bad would the frequency response aberations be? Is it as pronounced as a flat baffle and requires asymmetrical driver positioning?

 

[forr]

Hi,
Have a look at Art Ludwig's solutions :
http://www.silcom.com/~aludwig/images/roomint.jpg
http://www.silcom.com/~aludwig/Loudspeaker_construction.html#Diffraction

 

[CeramicMan]

That reminds me, I'm pretty sure there is one particular baffle resonance with spherical and spheroidal speakers. When the wavefront has diffracted 180 degrees at the back of the loudspeaker, it collides with itself and gets reflected back to the driver.

The focusing effect of a sphere and the absence of other exterior resonances could make that reflection quite prominent. An improvement over a sphere might be a kind of "animated teardrop" shape instead, where the wave cannot recombine behind the speaker due to a long point shaped like an exponential cone.

 

[Svante]

Are the negative effects of diffraction removed? How bad would the frequency response aberations be? Is it as pronounced as a flat baffle and requires asymmetrical driver positioning?

You mean by rounding the edges?

Well, diffraction is certainly not removed. It is not nessecarily so that diffraction effects are negative (in any case, not just this), though, they are simply part of the system design.

I have not done simulations of rounded edges, but I have a pretty clear view of what happens. As I said, the result of rounding the edges is to smear the impulse response of the diffraction time-wise. This, in turn results in less high-frequency content in the edge pulse, and this results in a smoother response for frequencies above the baffle step. This is nice, but maybe not as important as choosing a smart placement of the driver on the baffle. By making the distance from the driver to the baffle edge vary depending on direction, the same smearing of the edge reflection is acheived. And since the variation in distance that is possible by driver placement typically is larger than the size of the rounding of the edges, placement on the baffle is typically more important.

But then again, there is no reason to avoid rounding of the edges, other than practical.

If you have not played around with "The Edge" in my signature, you may want to give it a shot. It can be rather informative to see the effects of driver placement there IMHO. It cannot simulate rounded edges, though, but the effects are simular to having an asymmetrical placement of the driver on the baffle.

 

[Svante]

That reminds me, I'm pretty sure there is one particular baffle resonance with spherical and spheroidal speakers. When the wavefront has diffracted 180 degrees at the back of the loudspeaker, it collides with itself and gets reflected back to the driver.

The focusing effect of a sphere and the absence of other exterior resonances could make that reflection quite prominent. An improvement over a sphere might be a kind of "animated teardrop" shape instead, where the wave cannot recombine behind the speaker due to a long point shaped like an exponential cone.

Well, the baffle step for a sphere is pretty smooth actually. However, the directional pattern has a lobe in the straight backwards direction, and this lobe gets narrower for higher frequencies.

There is a story, which I don't know if it is true in relation to this. This will be slightly OT, please forgive me. There was some old time scientists that were arguing about diffraction around spheres. One of them had a theory, and the other was very sceptic, and his main argument was that if the theory was true, there would be a bright spot in the middle of the shadow of a sphere (the 180 degree lobe). Everyone knew that this was not the case. He thought. When the experiment was done, the spot was there, but given the short wavelength of light the spot was very small, ie the back lobe was very narrow.

Another problem with spherical boxes is the resonances that appear inside of them, they can be rather prominent. Anyone who has bounced a basketball have heard them ringing.

 

[thadman]

OT from the focused discussion, but will their be negative effects as a result of diffraction from the rear radiating drivers? The sound radiating from the rear will undergo several reflections before it reaches the listening position...is it worth spending time to ameliorate diffractions from the rear?

 

[CeramicMan]

Another problem with spherical boxes is the resonances that appear inside of them, they can be rather prominent. Anyone who has bounced a basketball have heard them ringing.

Exactly! Many people seem to give up on boxed speakers and conclude that the "boxy colourations" are somehow inherent in the design. Well they are - that's how they built it! Even if the box is extremely strong, the weak point is the cone itself. It's only a thin, light diaphragm, so it can't be expected to attenuate undamped resonances.

 

[thadman]

OT from the focused discussion, but will their be negative effects as a result of diffraction from the rear radiating drivers? The sound radiating from the rear will undergo several reflections before it reaches the listening position...is it worth spending time to ameliorate diffractions from the rear?

?

 

[CeramicMan]

It's hard for me to say actually. Imagining sound going "around" a curved surface isn't that easy.

Speaker datasheets sometimes have directivity plots, with e.g.: -15dB response at +-90 degrees. So, extrapolating that, it suggests that the sensitivity at 180 degrees would be around -30dB, but it's hard to say for sure without proper calculations or looking at simulations. And it would depend on the size of the baffle.

 

[thadman]

Rounding the edges of a baffle does not eliminate the reflection at the baffle edge. A rounded edge can be seen as very many smaller edges, and each of these will result in a small reflection. The sum of these reflections is equivalent to the single reflection of a single edge, but it is smeared in time (which is good).

The only way to eliminate baffle diffraction is to put the loudspeaker "in-wall". Or to put it in vacuum...

Diffraction may not necessarily be as bad as I previously imagined. I was reading Lynn Olson's "Beyond the Ariel" thread where he mentions that the ear uses 0-1ms for localization, 1-25ms for ambience, and beyond that we receive an echo.

If we could perhaps extend the length of time the waves need to reach the edge of the baffle, we could achieve "ambience" vs localization smears.

13,397in/s times 2ms (ie .002 seconds) we get 26.794 inches

If we wanted a 2ms delay for diffraction, would it be 27" from edge to edge of baffle or 27" from center of diaphragm to edge?

How do we calculate the time delay of a curved baffle? Do the single diffractions sum to the total diffraction as the effective width of the baffle or something else?