QRD Diffuser Design

[Octra]

I am going to be building a QRD Diffuser hopefully and while researching designs, I came across something at SoundScapes web site that looked intriguing.

SoundScapeS WaveWeaver -- QRD-diffusers.

Does anyone know anything about if this design will work well as a diffuser? It looks easier to build than the typical model diffusers I have been seeing.

Here is a piture of of the website as well.

[Ethan Winer]

I have no idea if that will really be useful, but I have to admit it doesn't look like it will do much. Real diffusors have angles or curves to scatter sound, or different depths to stagger the reflection times. This seems to have neither, other than the shallower part at each end.

--Ethan

[Tom H]

I don't know, was looking at it the other day myself... aren't there really three different depths in this design? One is moving around the corner, would this still work in theory?

[avare]

A mod 7 QRD would have well depth pattern of 1,4,2,2,4,1. That has 1,3,2,2,3,1. It will diffuse, but not as a QRD.

Andre

[Octra]

The fact that there was no difference of depth exept the outer two parts had me wondering that as well. I thought that maybe the L shape design on the outside was refracting sound somehow so that it did not have to rely different depths, I have no clue.
Thanks for the knowledge though everyone. I appreciate it very much.

[Youn]

*bump*

Any other opinions on the performance of these things as compared to regular qrd diffusors? Are the claims on their site just bs, or what? I'd love to build these things, seems like much less work, less space and lighter...

http://www.soundscapes.nu/ww/m-3.gif

[Amioa]

folded well diffusers.

when made correctly they *theoretically* have the same performance bandwidth of QRD of 'normal', at least thats what whitemark say...

[Torea]

Somebody else was talking about making a folded N7 QRD here. I saw their sketchup, it looked similar, but I can't remember which thread/person it was in. It was one of the really long diffuser threads (that should narrow it down)

[localhost127]

Originally Posted by Torea

Somebody else was talking about making a folded N7 QRD here. I saw their sketchup, it looked similar, but I can't remember which thread/person it was in. It was one of the really long diffuser threads (that should narrow it down)

DIY Diffusors to the Max

[Torea]

Originally Posted by localhost127

DIY Diffusors to the Max

Thanks! I thought that was it but didn't have time to find it.

I was referring to this post: http://www.gearslutz.com/board/6062490-post360.html

[SAC]

Yes. Folded QRDs work well.

That design, however, does not follow Schroeder/QRD mathematics.

[Youn]

Here's my design based on this... I found a good ratio that resulted in easy-to-measure cuts. The height is 4 feet, plus the top and bottom frame. Also note there is no back, so this should be lighter. I like how the wall shows through too...







...again, not sure if it's as effective as a normal qrd, I'm guessing not, but if it's really negligible then I might consider building one myself. Any more thoughts?

[T1M0N]

http://www.gearslutz.com/board/5278945-post4.html

This post tells us, this is covered in the diffusor book by cox.
unfortunately i don't yet have a copy of it, but i think the folds will reach their limits at a certain point.

i made a quick sketch up of a folded 13+4 qrd structure (1-3-7-0-8-5-4-5-8-0-7-3-1), does anyone who has read the book any comment on how far you can take this idea and how much efficiency will get lost? or would someone be willing to share a scan of those 2 pages for educational purpose?

Thanks

[Jens Eklund]

Wow, the idea might be good but I would think the losses caused by diffraction would cause too much absorption to be acceptable. Normal fined QRD if not wide wells already absorb quite a bit. This is one reason to use stepped sequence (assuming the shape is adjusted so it works, just removing the fins from a QRD is not an option if even scattering is required).

[Youn]

Originally Posted by T1M0N

how far you can take this idea and how much efficiency will get lost?

I'd also like to know

I also took it one step further then my previous post, but only two layers of folds, not four like you did. I'm assuming it's a loss in efficiency because the first reflection @ around 90 degrees does not diffuse the sound according to the QRD specs. In other words, it's not possible to expect the same results when it's folded because the ray tracings responds differently.

Taking it as far as you have with that many folds, while it looks awesome and would be really fun to build, I fear might result in too much of a difference. But, yeah... if the smart folks here say it's cool, then it's still worth considering IMO

[T1M0N]

Yes that's what I think. For me it's really hard to imagine, that this could work, especially with the depth 7-channel. But I found it hard to imagine how diffusors work anyway, so perhaps someone can tell me the folded wll logic and I am enlighted. Just not smart enough to figure it out myself. And to build it and see what happens is difficult in my opinion, because I wouldn't know how to measure and rate the results.

I would really like to read what Cox et al have written about this subject, I think I will buy the book.

My design was only a fast idea, perhaps not the best that's possible (why limit yourself to 90° angles, with 45° angles there are even more possibilities), just to demonstrate what I think is above the limits of this design, and to see if someone can tell where the limits are and what the drawbacks are or how the math changes.

[Sebg]

For frequencies that are large enough I would expect that it would treat each well as just a simple well. It is only for higher frequencies that reflected sound becomes an issue due to the bends.

[Jens Eklund]

Originally Posted by T1M0N

My design was only a fast idea, perhaps not the best that's possible (why limit yourself to 90° angles, with 45° angles there are even more possibilities), just to demonstrate what I think is above the limits of this design, and to see if someone can tell where the limits are and what the drawbacks are or how the math changes.

If you want to play around with other shapes (angles, polys etc.) you should have a look at AFMG "Reflex" (BEM-modeling of diffusers) but it does not do folded shapes (... yet, I’ve asked for it when I was in the beta team, perhaps in next version but the demand seems very slim).

I would again like to express my fears about excessive absorption in a folded design, especially if as complex as the above design. If you buy Reflex, you can find good stepped shapes instead thus minimizing absorption:

http://www.gearslutz.com/board/6732004-post3.html
http://www.gearslutz.com/board/6732069-post5.html

[localhost127]

Originally Posted by Jens Eklund

If you want to play around with other shapes (angles, polys etc.) you should have a look at AFMG "Reflex" (BEM-modeling of diffusers) but it does not do folded shapes (... yet, I’ve asked for it when I was in the beta team, perhaps in next version but the demand seems very slim).

I would again like to express my fears about excessive absorption in a folded design, especially if as complex as the above design. If you buy Reflex, you can find good stepped shapes instead thus minimizing absorption:

http://www.gearslutz.com/board/6732004-post3.html
http://www.gearslutz.com/board/6732069-post5.html

hi, jens - for curiosity-sake, did you by chance do a Reflex measurement of your design with the wells? would be curious to see the differences if you felt comfortable sharing. thanks!

[Jens Eklund]

Originally Posted by localhost127

hi, jens - for curiosity-sake, did you by chance do a Reflex measurement of your design with the wells? would be curious to see the differences if you felt comfortable sharing. thanks!

No, I have not tried to add fins to this design since the whole idea was to find a shape that performed well without dividers. I wouldn’t be surprised if the performance decreased if one was to add dividers to it.

[localhost127]

Originally Posted by Jens Eklund

No, I have not tried to add fins to this design since the whole idea was to find a shape that performed well without dividers. I wouldn’t be surprised if the performance decreased if one was to add dividers to it.

thanks!
by the way, your design isn't symmetrical (QRD) - are you using PRD or a home-brew design that you've developed on your own and verified via Reflex modeling?

[Jens Eklund]

Originally Posted by localhost127

thanks!
by the way, your design isn't symmetrical (QRD) - are you using PRD or a home-brew design that you've developed on your own and verified via Reflex modeling?

The Optiffuser is the result of countless hours of searching for the perfect shape offering the highest possible diffusion coefficients using Reflex. It is not a QRD (not only due to the lack of dividers that all QRDs have), it’s an optimized stepped diffuser and even though it’s not symmetric, the scattering is very even as you can see on the polar plots (in the manual).

[localhost127]

Originally Posted by Jens Eklund

The Optiffuser is the result of countless hours of searching for the perfect shape offering the highest possible diffusion coefficients using Reflex. It is not a QRD (not only due to the lack of dividers that all QRDs have), it’s an optimized stepped diffuser and even though it’s not symmetric, the scattering is very even as you can see on the polar plots (in the manual).

that's what i was assuming (home-brew design altered until you got the measured (simulated) performance via your design requirements) - but just wanted to confirm. very very cool

p.s. - you should look into amazon's EC3 cloud services (high performance computing) - maybe you could rent an array for a month and have cloud services do the computational rendering from Reflex

[Jens Eklund]

Originally Posted by localhost127

that's what i was assuming (home-brew design altered until you got the measured (simulated) performance via your design requirements) - but just wanted to confirm. very very cool

p.s. - you should look into amazon's EC3 cloud services (high performance computing) - maybe you could rent an array for a month and have cloud services do the computational rendering from Reflex

As soon as Reflex supports automated scripting, I will!