The No.1 Website for Pro Audio
Diffuser Video Simulations Shred Sound To Death—optimized sound diffusers—for thesis.
Old 26th July 2012
  #31
Gear Addict
 
Arqen's Avatar
 

Acoustic of Stonehenge

Since the theme of this thread is sound diffusion simulations, I thought you guys should check out this video from Jonathan Sheaffer at the University of Salford. (thanks to andrebrito for sending me to his site):

The acoustics of stonehenge!



He's using the same simulation technique I used (finite difference time domain).

Go check it out and let him know if you dig it: the sound of stonehenge.


Tim
Old 27th July 2012
  #32
Gear Maniac
 
cborg's Avatar
 

Thanks for the tip! Already subscribed to Jons chan. Sometimes wish I had the time, resource and knowledge to try FDTD out myself.
Old 27th July 2012
  #33
Gear Addict
 
Arqen's Avatar
 

Quote:
Originally Posted by cborg View Post
Thanks for the tip! Already subscribed to Jons chan. Sometimes wish I had the time, resource and knowledge to try FDTD out myself.
I'm not going to lie . . . scientific computing with numerical methods (like FDTD) is really intense and challenging for most people (including me). Some people have this analytical, super logical mindset that makes them natural programmers. I'm not one of them (even after going through engineering school, this stuff does not come easy to me).

But, of the various physical modelling methods, FDTD is one of the most useful, most intuitive and easiest to code.

If you know how to program, you can play with FDTD using the open source software Octave (It's basically a free clone of the popular software for scientific computing, "Matlab"):

GNU Octave

There is also this acoustics toolbox for Matlab: k-Wave: A MATLAB toolbox for the time domain simulation of acoustic wave fields, which I made great use of. It might work in Octave as well, but I don't know. It's a simpler way to play with FDTD because you can abstract yourself from the actual wave propagation equations.

But, if you have no urge to program, and no natural knack for it . . . you're probably a healthy human . It feels like some of my classmates were robots, or had Asperger's syndrome. I generally have to consume a lot of caffeine to focus while programming, and after spending hours and hours working with pure logic and numbers it starts to chew at my soul ...

But that's just me .
Old 27th July 2012
  #34
Gear Addict
 
Arqen's Avatar
 

Optimized Stepped Diffuser

New diffuser simulation video. This one shows my simplest, optimized stepped diffuser (a periodic array of 5 modules).

Even though it's periodic, it's a very efficient diffuser for it's depth... and ridiculously easy to DIY build.

I was amazed how shallow and simple this design came to be. In a bit I'll post a video of a single module of this diffuser, so you can see how the sound behaves as it hits the isolated module (the sound wave almost seems to leap off of the iscolated diffuser module).

Old 28th July 2012
  #35
Gear Maniac
 
cborg's Avatar
 

Thanks Arqen for sharing those resources.

Quote:
Originally Posted by Arqen View Post
But, if you have no urge to program, and no natural knack for it . . . you're probably a healthy human .
Unfortunately that's not the case.
And the greater the challange the greater the urge. But I will try hard to stick to actually finish some of my current projects. (It has happened once, or twice ..... )

Quote:
and after spending hours and hours working with pure logic and numbers it starts to chew at my soul ...
Amen to that.
Old 28th July 2012
  #36
Gear Addict
 
Arqen's Avatar
 

Quote:
Originally Posted by cborg View Post
Thanks Arqen for sharing those resources.


Unfortunately that's not the case.
And the greater the challange the greater the urge. But I will try hard to stick to actually finish some of my current projects. (It has happened once, or twice ..... )


Amen to that.
Haha, you're the inventor breed!

Loves challenges. Lots of projects on the go. Projects are an essential part of life. I like to think I'm similar in that respect, and I'm soooooo very lucky that ma lady tolerates the various esoteric passions and gadgets that invade the living room :
Old 29th July 2012
  #37
Gear Addict
 
Arqen's Avatar
 

Optimized stepped diffuser module

New video! This one shows a single module of an optimized stepped diffuser.

I like this simple simulation because you can see clearly how the shape sends sound scattering.
(for best effect, watch it in full screen).

Old 31st July 2012
  #38
Gear Addict
 
Arqen's Avatar
 

Efficient Numerical Simulation of Sound Propagation

While this next video is scientific, it's still has cool animations, and I think it's important to highlight for any researchers visiting this thread.

The video shows a numerical model of sound propagation that's much more efficient than standard FDTD (finite difference time domain). This simulation method could be used to enhance my work, making optimization faster.

Old 1st August 2012
  #39
Here for the gear
 

I'm something of a fan of Boggy's MyRoom concept, and I'm wondering how these units would work with gaps between, or perhaps aranged in a MLS type sequence? Do you know of any examples or anyone who is working on this type of arrangement?
Old 1st August 2012
  #40
Gear Addict
 
Arqen's Avatar
 

Quote:
Originally Posted by smaggers View Post
I'm something of a fan of Boggy's MyRoom concept, and I'm wondering how these units would work with gaps between, or perhaps aranged in a MLS type sequence? Do you know of any examples or anyone who is working on this type of arrangement?
Good question.

That's another potential way to modulate the periodic diffuser. But, bear in mind that these diffusers were optimized to work side by side in an array.
Without simulating it, I don't know if an MSL type arrangement will offer improved performance, or decreased performance.

It will certainly still diffuse sound, but it will no longer function optimally.

However, if you had a huge room and you arranged the entire array according to an mls sequence (while each "1" in the sequence = 5 diffuser modules), then it would work very well. This would be a good way to implement the diffuser in a large room.

Cheers,

Tim
Old 3rd August 2012
  #41
Gear Addict
 
Arqen's Avatar
 

Stonehenge Diffusion

Here's another cool finite difference time domain (FDTD) simulation by Jonathan Sheaffer.

Visualize the scattering produced by Stonehenge.

Old 4th August 2012
  #42
Gear Addict
 
Arqen's Avatar
 

Sound Diffusion - Wave Scattering off a Wall of Spikes

Okay!

So here's that video I mentioned that shows sound diffusion chaos as a wave hits a wall of spikes...





Some more funky acoustic diffuser videos are in the making (very psychedelic ones).
If you're curious about what's to come, you might want to subscribe to my youtube channel...

Enjoy!

Tim

Last edited by Arqen; 5th August 2012 at 12:29 AM.. Reason: Revision on first page. The first post of the thread has been updated to accommodate this video.
Old 13th August 2012
  #43
Lives for gear
 
Jens Eklund's Avatar
Quote:
Originally Posted by Arqen View Post
In case you guys are interested in performance data, here are the diffusion coefficients for the two fractal diffuser designs. The plot shows the autocorrelation diffuser coefficient for the chunkier fractal diffuser (B2-Fractal) and the lower profile fractal diffuser (A1-Fractal).

Just so you know... this data does not come from real-world measurements. It comes from a finite difference time domain scattering simulation (a physical model). Also, these plots assume that the incident sound wave hits the diffuser at a 0 degree angle (meaning, head on).

FYI: I created these plots by coding up some equations in the book "Acoustic Absorbers and Diffusers" by Trevor Cox at the University of Salford and Peter D’Antonio of RPG Diffusor Systems.




Tim

Hi Tim


First of all; I admire your and efforts and ambition to share your results with us. Your optimization routine seems to work for two reasons; the resultant shape is close to a series of polys (the shape you often end up with after unrestricted optimization if temporal scattering performance is not taken into consideration) and the coefficient curve (normalized diffusion coefficient for random incidence, 8 periods, rendered by Reflex) is very even.

Your algorithm for extracting diffusion coefficients on the other hand seems a tad optimistic. Looking at the A1-Fractal data in post 20, your graphs suggests a normalized diffusion coefficient (the normalized diffusion coefficient is after all what matters when evaluating diffuser performance) of about 0,4 at 200 Hz, which is impossible for a panel measuring only 74,25 mm deep to achieve. If you look at the appendix in AAaD on the first and second page (appendix C with diffusion coefficients); not even a 300 mm deep panel of polys can reach those numbers (the normalized diffusion coefficients at 200 Hz is 0,00 for random incidence and 0,01 for 0 degrees incidence for the 300 mm deep panel).

Below you´ll find some simulations of the A1-Fractal using AFMG Reflex:
Diffuser Video Simulations Shred Sound To Death—optimized sound diffusers—for thesis.-a1-fractal-random-incidence-8-periods.gifDiffuser Video Simulations Shred Sound To Death—optimized sound diffusers—for thesis.-a1-fractal-normal-incidence-8-periods.gif


Even if using only 5 periods (you state 5-10 panels on your page) resulting in a total period with of only 2,1 meters, the result is still far from the coefficients you´ve provided:

Diffuser Video Simulations Shred Sound To Death—optimized sound diffusers—for thesis.-a1-fractal-random-incidence-5-periods.gifDiffuser Video Simulations Shred Sound To Death—optimized sound diffusers—for thesis.-a1-fractal-normal-incidence-5-periods.gif


A FDTD model should render very similar (if not close to identical) results as a BEM model (but with the benefit of being able to model temporal scattering as well unlike BEM) so I´m afraid something is not right with your model regarding diffusion coefficients. I haven’t modeled the other models but I might if I can find the time (random incident coefficients at 1/24 octave resolution from 100 Hz to 10 kHz takes some time to render as you probably know).


Sincerely Jens Eklund
Old 14th August 2012
  #44
Gear Addict
 
Arqen's Avatar
 

Hi Jens,

I appreciate you taking the time to check my results.

First off, I'll need to clear up something in my explanation: 5-10 modules should be 5 OR 10 modules, depending on the length of the module. If the modules are full-height (4' long), then only 5 modules would be used. If the modules are half-height (2'), then 10 would be used.

I optimized the 2D shape in an array of 5 modules, so optimal results are only expected when there are 5 repetitions (not 6, 7, 8, 9 or 10).

My bad for lack of clarity.

My diffusion measurements and calculations have been coded from scratch, and there are several things that could have resulted in an inaccurate measurement. I was a bit hesitant to include the coefficient measurements in my thesis, because I could not do a direct comparison with existing diffuser designs (other known designs don't fit on my simulation grid). However, I did not expect the measurement to be that far off. I wonder if I'm using the same definitions as Reflex for the autocorrelation diffusion coefficient.

As mentioned in the thesis, my scattering model has not yet been validated, but it needs to be. So I hope that this is a diffusion coefficient measurement error, and not a modelling error in the scattering simulation.

To validate the scattering simulation, I'll need to compare the scattered polar distribution plots that I've measured with the plots produced in AFMG Reflex. I.e., if the scattering polar plots are almost the same for both my FDTD model and the reflex model, that would suggest that the model and optimization results are sound, but the diffusion coefficient calculation if off.

I'm heading away on a backpacking trip this week, but I'll look into that when I get back.

I'm also very curious what results my better-performing diffuser (B2 frac) has using Reflex.

Thanks for pointing this out to me!
Tim
Old 19th November 2012
  #45
Here for the gear
 

Quote:
Originally Posted by Arqen View Post
To validate the scattering simulation, I'll need to compare the scattered polar distribution plots that I've measured with the plots produced in AFMG Reflex. I.e., if the scattering polar plots are almost the same for both my FDTD model and the reflex model, that would suggest that the model and optimization results are sound, but the diffusion coefficient calculation if off.

I'm heading away on a backpacking trip this week, but I'll look into that when I get back.

I'm also very curious what results my better-performing diffuser (B2 frac) has using Reflex.
Hi Tim,

I just downloaded the plans for your modules, which I am planning to build soon. thanks for sharing this!

Where do we stand regarding the mesurements "check"? It does look too good to be true!

thanks

Lolo
Old 19th November 2012
  #47
Gear Maniac
 

Oh no. So having built but not yet installed an array of 5 A1-LF modules, are we now saying they don't work??? Because if that simulated performance is way off, it seems it would be doing very little diffusion whatsoever. This is very concerning.
Old 19th November 2012
  #48
Lives for gear
 
Jens Eklund's Avatar
Quote:
Originally Posted by Radial185 View Post
Oh no. So having built but not yet installed an array of 5 A1-LF modules, are we now saying they don't work??? Because if that simulated performance is way off, it seems it would be doing very little diffusion whatsoever. This is very concerning.
They work, but one has to understand that a diffuser only 74 mm deep will never perform miracles. A diffusion coefficient of between 0,1 and 0,18 (2,1 meter total panel) from about 500 Hz is not bad considering the shallow depth.
Old 20th November 2012
  #49
Gear Addict
 
Arqen's Avatar
 

Quote:
Originally Posted by Radial185 View Post
Oh no. So having built but not yet installed an array of 5 A1-LF modules, are we now saying they don't work??? Because if that simulated performance is way off, it seems it would be doing very little diffusion whatsoever. This is very concerning.
Don't worry... as Jens mentioned, they still work! And they perform particularly well if you mount them using a "profiled modulation", as shown on the last page of the current version of the blueprints. This reduces the effects of periodicity in the array.

I.e., mount modules with varying depths as follows: [0 cm, 5 cm, 6 cm, 5 cm, 0 cm]

Below are the diffusion coefficients for an array of 5 modules, mounted using this profiled modulation. As you can see, this gives a big performance boost!



There are many other useful modulations which can be used to improve the performance. E.g., if you plan to mount an array of 7 modules, a low frequency fractal modulation would well (7 modules would be mounted at depths proportional to the 7 wells depths of the original diffuser... creating a nested diffuser structure).
Old 21st November 2012
  #50
Here for the gear
 

thanks, will give it a go around Christmas time.
Old 23rd November 2012
  #51
Gear Addict
 
Arqen's Avatar
 

Quote:
Originally Posted by loloo View Post
thanks, will give it a go around Christmas time.
Awesome. Let us know how it goes, and if you have any questions regarding the blueprints (or if you take any photos) you can post them here:

DIY Sound Diffusers Thread

Take care,
Tim
Old 23rd November 2012
  #52
Here for the gear
 

thanks for your help Tim, my father is a carpenter so it's going to be an easy one for me.
Old 3rd December 2012
  #53
Gear Addict
 
Arqen's Avatar
 

Quote:
Originally Posted by loloo View Post
thanks for your help Tim, my father is a carpenter so it's going to be an easy one for me.
Hey, no problem!

Also... since your father is a carpenter I'm curious: what material are you guys planning to use?
Topic:
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump