Quantcast
DIY Sound Diffusers—Free Blueprints—Slim, Optimized DIY Diffuser Designs (+Fractals) - Page 19 - Gearspace.com
The No.1 Website for Pro Audio
DIY Sound Diffusers—Free Blueprints—Slim, Optimized DIY Diffuser Designs (+Fractals)
Old 23rd January 2015 | Show parent
  #541
Gear Addict
 
Arqen's Avatar
 
🎧 5 years
Stepped vs PRD vs QRD diffuser subjective comparison

Quote:
Originally Posted by proV ➡️
I have read all the pages in this topic but it seems like most people are only interested in form, in "how to build" the diffuser like it is some kind of furniture and not in its main function, the sound itself. So, how does it SOUND? What difference did it make?

Good question.
Here's a one sentence summary since this is a long post: If you were to compare it to a QRD diffuser of equal size and complexity, I'd expect the Leanfuser to perform better and be easier to build (and IMO look better). This is just a hypothesis, but see the last section of this post for my reasoning.

You're right that the most important thing is how they sound. Several people have done listening tests and emailed me expressing happy results.
Schaap posted a with / without audio comparison here, but he did not have room to test a complete, 7-panel diffuser array.

Here are a few subjective listening comments:

“I've build dozens of all kind diffusers esp. PRD(skylines) which I like very much and there's not much space left, guess I'm addicted to diffusion. and I like to do experiments .
One thing that was remarkable IMO with your stepped diffuser was the good transient or attack response(though quite normal with diffusion) despite the small dimensions... example in the piano samples. “
- Schaap

“Another recording... Diffusion diminishes the 'smearing' and 'glassiness'” - Schaap

“The flutter echo in my studio was instantly transformed into a much more pleasing reverberation.” - Chris Rude

“These covered an otherwise open section of concrete wall at the first reflection points of the loudspeakers. I was expecting an improvement to sound, but not this degree. It really made a great difference to the sound of the room. Previously it had been over-damped. I had filled in the beams in the (too low) ceiling above with absorbent covered with burlap. With the diffusers added, the sound became both more lush and much more detailed.” –bwaslo in the Audiocircle forum

Quote:
Originally Posted by proV ➡️
Did you compare it to QRD (with fins) or PRD diffusor?
For a scientific performance comparison of stepped vs finned diffusers, see the section below.
I've not seen a direct listening comparison of stepped vs PRD vs QRD diffusers. But I agree, it would be great if someone did those tests.

Just curious, have you seen any listening comparisons of PRD vs QRD diffusers? I did a quick Google search for this and did not find anything conclusive.
For a fair comparison you would need diffusers of each type with similar depth and width. Otherwise you’re comparing apples to oranges.

There are many commercial diffusers on the market which don’t even post their diffusion coefficients, and people buy them on faith that they work. I know for a fact that certain commercial designs are based on no real science… they are just random shapes that look like diffusers. And that’s fine, provided that they actually work and someone has verified the performance before selling them.

Still, I see commercial diffusers that present no scientific or tangible evidence. No diffusion coefficients. And of course, no split listening tests comparing them to other diffusers. This may be because it's difficult to do a fair listening comparison due to reasons discussed below.


Quote:
Originally Posted by proV ➡️
Did you calculate its depth right or it didn't perform as you have expected? Do you think that you needed to build more of them to make the same difference as with classical PRD diffusors? Or you didn't even noticed any difference in sound vs. bare wall at all?
It actually performs better than I expected, considering how simple and low profile it is.

I wrote a program to find the depth sequence. I basically told the program to look for the best diffusing surfaces with 7 wells and a maximum depth of 16 cm. The code combines a physical model that simulates sound scattering + an optimization algorithm that simulates natural selection to search for the best diffusing surface.

It spat out a few good designs, but the Leanfuser was the cream of the crop. The Reflex test results show the diffusion coefficients, and I have compared those against a couple other optimized stepped diffusers that are presented by Trevor Cox in this paper.

(Cox compared optimized stepped diffusers vs QRD diffusers and finned optimized diffusers, and the stepped diffusers outperformed both).

You can see one of those comparisons at the end of my diffuser design paper, where I compared the performance of an array of 5 Leanfuser panels vs 5 periods of Cox's N=7 optimized stepped diffuser, on the basis that both configurations have similar build complexity and depth (the Leanfuser won because it can be arranged to remove periodicity, greatly increasing performance while still being shallower than the competitor). Of course the comparison is not truly fair, because I'm comparing a modulated design vs a periodic one.

It's hard to fairly compare two diffuser designs that are completely different. Unlike larger stepped diffusers, the Leanfuser is easy to construct and uses modular assembly (profiled modulations) to achieved depth.
  • Assuming equal depth, width and well width, I expect one giant 35 well optimized stepped diffuser would perform better than an array of 7 Leanfuser panels. But the Leanfuser array would be easier to build.
  • Assuming equal depth, width and well width, I expect one giant N=37 QRD or PRD would perform better than an array of 7 Leanfuser panels. But the Leanfuser array would be easier to build (and IMO look better).
  • Assuming equal total depth, 5 modules of the Leanfuser, arranged in Profiled Modulation 1, performs much better than 5 periodic modules of a deep optimized stepped diffuser with similar complexity (see the end of my paper).
  • I expect X shallow Leanfuser modules, arranged in a profiled modulation, will perform much better than X periodic modules of a deeper, N=7 finned QRD or PRD (assuming both arrays have the same total depth).
  • Since Cox's optimized stepped diffusers outperformed equivalent Schroeder diffusers (including QRD diffusers), and I compared the Leanfuser against his designs, I expect that the Leanfuser will outperform a QRD diffuser of equal size and complexity.

I don't know how the Leanfuser system would compare to aperiodic modulated QRD or PRD diffusers, but it would be interesting to test that.

Also, we have the fractal version, the Leanfractal. I don't know how it would compare to the RPG Diffractal if both were designed to have equal depth, but it would be cool to test that too.

If we start comparing diffusers with different depths, widths, well widths and prime numbers, it becomes a balancing act. Complexity, aesthetics, cost, size, 1D vs 2D diffusion... all these variables are important. It's like comparing a racing bike with a fixed gear with a mountain bike. Which one is better?

What I do know is that the Leanfuser kicks ass at achieving good diffusion performance with simple construction. And that's exactly what it's designed to do.

Last edited by Arqen; 25th January 2015 at 08:08 AM..
Old 24th January 2015
  #542
Gear Head
 
🎧 5 years
Hi Tim,

I've made some progress with absorption in my room, getting to constructing the diffuser now.

I'm not completely settled with the locations of the modules yet, however I've got several ideas, would love to hear what you think.

One option would be to mount 3 modules in a profiled modulation on the back wall. The wall is 2.5m long and 2.46m high. It's got a door on one of its sides. In fact it's two "sliding" doors, and they are built into a niche which makes the geometry somewhat interesting. This niche's dimensions are 1.45m (width)*1.95m (height)*22cm (depth). To clarify, this niche goes 22cm into the wall, so the door is 22cm deeper than the wall's surface. I wonder how much will this affect left-right symmetry?

In order to make some space for proper acoustic treatment on the back wall, I decided to give up on one of the door's halves and only use the other half for entrance, the one that is close to the corner.

So if I'm going for the 3-module option, the central module will be mounted in front of the vertical edge of the aforementioned niche, the other one in front of the wall, and third one in front of one of the halves of the door.

I've made some calculations with AFMG Reflex, trying out several modulations, and I've come up with this modulation that seems to perform better than others I've tried.

Unfortunately, the trial version would only allow me to use up to 20 elements, so I had to let go of the 2 zero-depth elements on the sides. Not sure whether this compromises the modelling. For every model, I created 2 pdf files - first with my parameters, second with parameters taken from here:
https://gearspace.com/board/attachme...n-settings.jpg

With my custom parameters, I used 1/6 octave resolution for Coefficients, and 1/3 octave frequency averaging for Spatial Response. For some reason, Reflex does not mention this in the generated pdf file.
You'll notice I have -9.5 degree angle plotted everywhere. That's because this is approximately the angle from my speakers to the center of the diffuser.

Since 3 full-width modules won't fit (3*42=126cm, while I can only use about 118cm), I decided to have the side modules' bases 38cm wide. I calculated the different options, and it looks like 6cm gap should be left alone, so the bases will be only 2cm on the "outer" end while keeping uniform 6cm widths throughout.

Now if I go with the 3-module option, this is my current idea for how to mount it:
Since the first reflection from the diffuser will only travel about 5.2m before it reaches my head, it's only about 12ms behind the direct signal. So perhaps it's better to splay the diffuser, in order to make the ISD gap longer. I'm not sure how would it be better to splay the modules, though. All parallel? The upper halves upwards, the lower halves downwards? Or maybe both slay vertically (all of them) and horizontally (all except for the central modules)? Probably I need to direct as much sound as possible to the side walls, sides of the ceiling and to the upper sides of the front wall (otherwise the diffused waves will end up in absorbers).

The other way around would be to treat first reflection spot with absorption and mount the diffusers as standalone in the corners of the back wall. Since one of the corners has a door to it, I'm thinking to mount only one side of the diffuser to the side wall and make it rotatable, to keep the entrance useable.

Third thing I'm thinking of is to mount the modulated diffuser splayed, and add some absorption behind it to catch the low frequencies. Not sure how efficient this absorber can be though, since it won't have any air gap to the wall and to the diffuser. Also this way I'd have to mount the diffuser even closer to my listening position (without absorber, it would be 2.2m away).

In short, any comments or suggestions greatly appreciated!

P.s. My room's dimensions are detailed here:
https://gearspace.com/board/10663938-post533.html


EDIT: wrong files, sorry. See next post.
Attached Files
File Type: pdf MyDiffuser4-StandardPar.pdf (382.0 KB, 306 views) File Type: pdf MyDiffuser4-CustomPar.pdf (657.4 KB, 296 views)

Last edited by tangerine; 24th January 2015 at 10:58 AM..
Old 24th January 2015 | Show parent
  #543
Gear Head
 
🎧 5 years
That was just a little test with widths variation.
Here's the right model.
Attached Files
File Type: pdf MyDiffuser2-Standard.pdf (379.4 KB, 263 views) File Type: pdf MyDiffuser2-CustomPar.pdf (656.1 KB, 220 views)
Old 25th January 2015 | Show parent
  #544
Gear Addict
 
Arqen's Avatar
 
🎧 5 years
Hi Tangerine,

Looks like you've explored a lot of good potential options.

Quote:
Originally Posted by tangerine ➡️
I had to let go of the 2 zero-depth elements on the sides. Not sure whether this compromises the modelling.
Alas, that does compromise the modelling because it will treat those zero depth elements as absorptive instead of reflective.

I still think your configuration will work, because we've tested several 3-module configurations and moving the center module out from the wall never did anything to harm the performance (it either increased performance, or had a negligible effect).

Quote:
Originally Posted by tangerine ➡️
Now if I go with the 3-module option, this is my current idea for how to mount it:

Since the first reflection from the diffuser will only travel about 5.2m before it reaches my head, it's only about 12ms behind the direct signal. So perhaps it's better to splay the diffuser, in order to make the ISD gap longer. I'm not sure how would it be better to splay the modules, though. All parallel? The upper halves upwards, the lower halves downwards? Or maybe both slay vertically (all of them) and horizontally (all except for the central modules)? Probably I need to direct as much sound as possible to the side walls, sides of the ceiling and to the upper sides of the front wall (otherwise the diffused waves will end up in absorbers).

The other way around would be to treat first reflection spot with absorption and mount the diffusers as standalone in the corners of the back wall. Since one of the corners has a door to it, I'm thinking to mount only one side of the diffuser to the side wall and make it rotatable, to keep the entrance useable.

Third thing I'm thinking of is to mount the modulated diffuser splayed, and add some absorption behind it to catch the low frequencies. Not sure how efficient this absorber can be though, since it won't have any air gap to the wall and to the diffuser. Also this way I'd have to mount the diffuser even closer to my listening position (without absorber, it would be 2.2m away).

In short, any comments or suggestions greatly appreciated!

P.s. My room's dimensions are detailed here:
https://gearspace.com/board/10663938-post533.html
I'm not sure I completely understand all these options. You could tilt the entire array upward by 5 or 10 degrees. It might look funny and be hard to do, but it's worth considering.

If by splayed you mean mounting it in a triangular fashion like this, /\ (but not as steep), that's an option but it would only work properly for an even number of diffuser panels. You could also mount the three panels like this, /---\. But I can only speculate as to how these splayed options will actually perform.

Mounting absorption directly behind you and diffusion on either sides a good option.

An ISD gap of only 12 ms is okay if the reflections are highly diffused and at least 10 dB below the direct signal. I think you could get away with mounting your modulated array as you've shown, but it depends on your listening goals. I can't guarantee that the rear wall first reflections will be reduced by 10 dB, so if clarity is your top priority absorption directly behind you is the safest option.

But if you're willing to experiment, I'd say try whatever option above sound most appealing to you.

Let me know how it goes.
Old 26th January 2015 | Show parent
  #545
Here for the gear
 
🎧 10 years
Thank you for your response Tim, much info and I see you are very confident in performance of Leanfusers.

I already decided to build a stepped diffusor so your post is encouraging. I am thinking on 2 smaller 3-panel Leanfusers in profiled modulation and 1 bigger (around 2m width) one. Although for this one I haven't decided yet which design to take. I almost went with unqlenol's 35mm well width low profile diffusor but then I decided to install Reflex myself and try to come with something on my own. As I already have some 18mm cut timber I went with 18mm width. I couldn't get your design to work with just simply scaling the well width to such an extreme so I modded your Leanfuser and Trevor J. Cox's 36step diffuser with fractal like move and came with some (preliminary) results I am unsure they would perform well in real world. Would you be willing to take a look at it and tell me your opinion? Or perhaps point to a better alternative? The diffusion coefficient curve is quite good I think but I am especially unsure about 0°Angle graph and perhaps some other details my unskilled eyes don't see.
Attached Thumbnails
DIY Sound Diffusers—Free Blueprints—Slim, Optimized DIY Diffuser Designs (+Fractals)-leanfuser-modm1.jpg   DIY Sound Diffusers—Free Blueprints—Slim, Optimized DIY Diffuser Designs (+Fractals)-cox-modm1.jpg  
Attached Files
File Type: pdf Cox modM1.pdf (454.7 KB, 234 views) File Type: pdf Leanfuser modM1.pdf (446.5 KB, 245 views)
Old 27th January 2015 | Show parent
  #546
Gear Addict
 
Arqen's Avatar
 
🎧 5 years
Diffusers with narrow wells

Quote:
Originally Posted by proV ➡️
Thank you for your response Tim, much info and I see you are very confident in performance of Leanfusers.

I already decided to build a stepped diffusor so your post is encouraging. I am thinking on 2 smaller 3-panel Leanfusers in profiled modulation and 1 bigger (around 2m width) one. Although for this one I haven't decided yet which design to take. I almost went with unqlenol's 35mm well width low profile diffusor but then I decided to install Reflex myself and try to come with something on my own. As I already have some 18mm cut timber I went with 18mm width. I couldn't get your design to work with just simply scaling the well width to such an extreme so I modded your Leanfuser and Trevor J. Cox's 36step diffuser with fractal like move and came with some (preliminary) results I am unsure they would perform well in real world. Would you be willing to take a look at it and tell me your opinion? Or perhaps point to a better alternative? The diffusion coefficient curve is quite good I think but I am especially unsure about 0°Angle graph and perhaps some other details my unskilled eyes don't see.
Wow this looks very cool! They have similar diffuse field performance, but the big dip at 2 kHz in the zero degree graph of your modified Leanfuser version is something I would try to avoid. Bringing the center panel out further from the wall might help this. I would give that a try, or try the 7-panel modulation (which is based on fractal self symmetry).

Since you're working with a narrow well width I think you could get better diffusion performance by building a diffuser specifically designed around those narrow wells. But I would not recommend this in most cases because deep, narrow wells will produce a lot of absorption (because with with deep, narrow wells, the viscous boundary layer becomes significant compared to the well width).

You can actually visualize how deep narrow wells would increase absorption in this video I made (just for fun). Notice how much of the energy is lost inside the diffuser's wells rather than scattered back.


Last edited by Arqen; 27th January 2015 at 02:04 AM..
Old 31st January 2015
  #547
Gear Head
 
🎧 5 years
I've been exploring a bit more what can be achieved with stepped diffusers, and I have come up with several interesting shapes that seem to produce good predictions in AFMG Reflex. Since I can't use more than 20 steps (is it only me? I see proV had been able to use many more steps...?), I attempted to go with 19 steps. Using 6cm well width, this gives 114cm, which actually fits perfectly into the space I have available.
Some of the best predictions are with shapes simulating an arc. So I also tried to simply create a round element 102cm wide and 16cm deep (Reflex allows this), and the predictions also look fairly good. The resulting spatial response charts look very good up to at least 5000Hz, predicting very even dispersion and very little lobes. Normalized diffusion coefficient from random incidence is about 0.5 from 500Hz up. From 9.5 degrees, it's about 0.6 from 500Hz up.
The latter fact is a little surprising, considering that Trevor Cox claims in his paper that half-cylinders should produce no temporal diffusion at all.
http://www.rpginc.com/docs%5CTechnol...The%20Ugly.pdf

Someone is in mistake here, so I'm wondering if it's Cox, or is it AFMG Reflex that produces wrong predictions? Or am I reading the charts wrong?
I will post some reports soon, just wanted to hear if anyone here has any idea about this and about arc-shaped diffusers in general. Cox wrote that there had been unverified reports that half-cylindrical diffusers sound in an unpleasant way... I wonder if it's considered a fact, and if this makes stepped diffusers approximating an arc a bad choice also?

Last edited by tangerine; 31st January 2015 at 12:08 AM..
Old 31st January 2015 | Show parent
  #548
Lives for gear
 
Jens Eklund's Avatar
 
🎧 10 years
Reflex can only simulate spatial diffusion and scattering, not temporal. If only concerned with spatial diffusion/scattering; a single big poly would be the optimum, but it would be terrible when it comes to temporal diffusion/scattering since a single poly would provide none of it.

More on polys here: Poly coefficients (cont from another thread)

And some relevant OT here: https://gearspace.com/board/8160618-post43.html
Old 31st January 2015 | Show parent
  #549
Gear Head
 
🎧 5 years
Quote:
Originally Posted by Jens Eklund ➡️
Reflex can only simulate spatial diffusion and scattering, not temporal. If only concerned with spatial diffusion/scattering; a single big poly would be the optimum, but it would be terrible when it comes to temporal diffusion/scattering since a single poly would provide none of it.

More on polys here: Poly coefficients (cont from another thread)

And some relevant OT here: https://gearspace.com/board/8160618-post43.html
Thanks for pointing me to that thread, Jens! I'm reading it.
I thought that by "diffusion coefficient" Reflex means temporal diffusion, that's where I'd been wrong. So how can we evaluate temporal diffusion? I understand that FDTD modeling is necessary... Is there a handy tool out there to do this for someone who doesn't want to learn to code?

Reg. sample width, what I make of it is if we just take 5 modules and place them side by side, we get a very low diffusion coefficient due to repetition. If we modulate them, diffusion coefficient is very good. Correct?
However I don't quite understand why you say that modeling of a single module is meaningless... I saw that you reference Cox's book where they use 3.6m width. Here, the module is way less. However Reflex seems to predict good results regardless?
Let's say I only have space for 3 modules, and I model them in Reflex in a fairly deep modulation (say 14-16cm total depth), should I expect results to be incorrect to the "real world"?
Old 31st January 2015 | Show parent
  #550
Old 31st January 2015 | Show parent
  #551
Gear Head
 
🎧 5 years
Quote:
Originally Posted by Jens Eklund ➡️
I saw this post and this is in fact what I had been referring to. Basically what is it we are trying to predict? 1 period's performance is meaningless if trying to predict performance of arbitrary number of periods. It makes perfect sense to require enough width or enough modules to make a good prediction. However what if we model 3 periods and we want to predict behavior of 3 periods? Why would there be a requirement to model 16 periods for that?
Old 31st January 2015 | Show parent
  #552
Lives for gear
 
Jens Eklund's Avatar
 
🎧 10 years
Quote:
Originally Posted by tangerine ➡️
I saw this post and this is in fact what I had been referring to. Basically what is it we are trying to predict? 1 period's performance is meaningless if trying to predict performance of arbitrary number of periods. It makes perfect sense to require enough width or enough modules to make a good prediction. However what if we model 3 periods and we want to predict behavior of 3 periods? Why would there be a requirement to model 16 periods for that?
If you want to know how a small number of panels will perform in free field, then the prediction from Reflex is valid.

However; if the panels are to be mounted on a surface (assuming flat); the prediction from Reflex will be way off. I guess you could model the panels with the rest of the wall on either sides of the panel included in your model.
Old 31st January 2015 | Show parent
  #553
Here for the gear
 
🎧 10 years
Quote:
Originally Posted by Arqen ➡️
Wow this looks very cool! They have similar diffuse field performance, but the big dip at 2 kHz in the zero degree graph of your modified Leanfuser version is something I would try to avoid. Bringing the center panel out further from the wall might help this. I would give that a try, or try the 7-panel modulation (which is based on fractal self symmetry).
7-panel modulation would end up to big to fit on the wall although it was my favourite from the start. Anyway I have got rid of that 2 kHz dip so the zero degree graph curve looks much nicer now. The diffuser is already half built. I went with it instead of chosing modded Cox's diffuser because I found it easier to build.


Quote:
Originally Posted by tangerine ➡️
is it only me? I see proV had been able to use many more steps...?
I don't seem to have limit on number of steps. The limitations I have is I can't save, can't get it not to crash when reaching past 1.5GB ram and the thing I miss the most, the undo button. I think you downloaded the Basic version, which is limited to 20 elements, I downloaded Standard trial version.
Old 31st January 2015 | Show parent
  #554
Gear Head
 
🎧 5 years
Quote:
Originally Posted by proV ➡️
I don't seem to have limit on number of steps. The limitations I have is I can't save, can't get it not to crash when reaching past 1.5GB ram and the thing I miss the most, the undo button. I think you downloaded the Basic version, which is limited to 20 elements, I downloaded Standard trial version.
Yes! You're right. They have this small radio button to choose between Standard and Basic, which I did not notice... Thanks for that info!
Old 31st January 2015 | Show parent
  #555
Gear Head
 
🎧 5 years
Quote:
Originally Posted by Jens Eklund ➡️
If you want to know how a small number of panels will perform in free field, then the prediction from Reflex is valid.

However; if the panels are to be mounted on a surface (assuming flat); the prediction from Reflex will be way off. I guess you could model the panels with the rest of the wall on either sides of the panel included in your model.
Yes, I have tried this with several shapes, and every time I add a "wall" to the system, performance gets reduced. Isn't it what will happen with any diffuser in general?
So our only option to get a good diffusion from the back wall is to cover the entire wall length with diffusers and optionally absorbers? Otherwise it's not worth the trouble?
But then we also have the side walls, perhaps they should be also taken into account...
Old 31st January 2015 | Show parent
  #556
Lives for gear
 
Jens Eklund's Avatar
 
🎧 10 years
Quote:
Originally Posted by tangerine ➡️
Yes, I have tried this with several shapes, and every time I add a "wall" to the system, performance gets reduced. Isn't it what will happen with any diffuser in general?
So our only option to get a good diffusion from the back wall is to cover the entire wall length with diffusers and optionally absorbers? Otherwise it's not worth the trouble?
But then we also have the side walls, perhaps they should be also taken into account...
That’s the common problem. People often look at the result from Reflex, thinking it looks super, but the problem is that the panel width is often not big enough so the coefficient values get beefed and mean nothing. I always simulate at 3,6 meter total period width to avoid getting fooled by values that mean nothing. And as you have noticed, if you simulate a short period width and add flat surfaces on either side of it, the performance will drop, and this is logical since only a section of the entire surface is scattering incoming energy.

The only time I would say it’s useful to model a smaller than 3,6 meter wide section is if your back wall is narrower than this, and you´re trying to find a custom shape for a specific room based on the available surface area (or at least width). But it´s then important to remember that the numbers generated by Reflex are only valid for that specific width and cannot be compared to other diffusion/scattering coefficients (assuming properly gathered … unfortunately, not all diffuser manufactures are “aware” of these things and it´s naturally more fun to present a graph with higher values …).

https://gearspace.com/board/studio-b...ml#post8443037
Old 1st February 2015 | Show parent
  #557
Gear Addict
 
Arqen's Avatar
 
🎧 5 years
Diffusion coefficient standards

Quote:
Originally Posted by Jens Eklund ➡️
Reflex can only simulate spatial diffusion and scattering, not temporal. If only concerned with spatial diffusion/scattering; a single big poly would be the optimum, but it would be terrible when it comes to temporal diffusion/scattering since a single poly would provide none of it.

More on polys here: Poly coefficients (cont from another thread)

And some relevant OT here: https://gearspace.com/board/8160618-post43.html
+1

Quote:
Originally Posted by tangerine ➡️
I thought that by "diffusion coefficient" Reflex means temporal diffusion, that's where I'd been wrong. So how can we evaluate temporal diffusion? I understand that FDTD modeling is necessary... Is there a handy tool out there to do this for someone who doesn't want to learn to code?
Good question.

You're right that we need to first capture the diffusion data in the time domain. One way is modelling using FDTD, or more advanced methods that are faster like the k-space pseudo-spectral method. Common side effects of trying to program this stuff include sleepless nights and hair loss.

You can also get time domain data using physical measurements, but that requires an accurate measurement environment and diffusers that are already built.

Unfortunately there's not a handy modelling tool I know of, partly because the interpretation of temporal diffusion data is not well established. It's a research topic that people are still exploring.

Also, FDTD modelling has only recently become computationally viable, so research in that area has lagged behind.

Last edited by Arqen; 1st February 2015 at 07:55 AM..
Old 1st February 2015
  #558
Gear Addict
 
Arqen's Avatar
 
🎧 5 years
Diffusion coefficient standards

Quote:
Originally Posted by Jens Eklund ➡️
That’s the common problem. People often look at the result from Reflex, thinking it looks super, but the problem is that the panel width is often not big enough so the coefficient values get beefed and mean nothing. I always simulate at 3,6 meter total period width to avoid getting fooled by values that mean nothing. And as you have noticed, if you simulate a short period width and add flat surfaces on either side of it, the performance will drop, and this is logical since only a section of the entire surface is scattering incoming energy.

The only time I would say it’s useful to model a smaller than 3,6 meter wide section is if your back wall is narrower than this, and you´re trying to find a custom shape for a specific room based on the available surface area (or at least width). But it´s then important to remember that the numbers generated by Reflex are only valid for that specific width and cannot be compared to other diffusion/scattering coefficients (assuming properly gathered … unfortunately, not all diffuser manufactures are “aware” of these things and it´s naturally more fun to present a graph with higher values …).

https://gearspace.com/board/studio-b...ml#post8443037
It's good that you're bringing this up.

From the photo in the attached PDF, it looks like even RPG may use single diffuser modules to express diffusion coefficients.

And RPG researchers helped develop the most recent coefficient standards.

Since the big companies are expressing diffusion this way, I presume it's more useful to the end user to compare diffusers like this (and assume the single module is surrounded by absorption), vs to not compare them at all. As a way to compare all diffusers, baffled measurements (with one module on a wall) are problematic too, because diffusers don't all have the same width.

So I agree that the best way to compare diffusers is to compare them using wide arrays (e.g. 3.6 meter array vs 3.6 meter array).

Ideally each company would post diffusion coefficients as measured in an array. Of course, this is still not a perfect comparison because some diffusers are designed to be used in a large array and some are designed to be used in isolation (e.g., a 2.5 meter wide, symmetrical diffuser).

In this case you would expect an optimized, 3.6 meter wide diffuser to perform best, because it can take advantage of the measurement standard. It's not periodic and it fits perfectly into the 3.6 meter width. But the 2.5 meter wide diffuser will get penalized by the standard because to meet the 3.6 m requirement you need to double it, so it becomes becomes 5 meters wide, and periodic. It's being tested in a configuration it's not designed for, and in this setup it's at a disadvantage to the 3.6 meter, non periodic design.

Even if the array width approaches infinity, the comparison is not fair because periodic designs will get a performance boost (due to lobes clustering infinitely close together), where in real world applications the lobing actually hurts the performance.

At the end of the day, the most important thing is the performance for the scenario in which the diffuser will be used. In many scenarios, it will be flanked by broadband absorption. Ideally there would be multiple plots for each diffuser to show performance for various scenarios .
Attached Files
File Type: pdf FlutterFree_Acoustical Data.pdf (103.2 KB, 140 views)

Last edited by Arqen; 1st February 2015 at 05:27 PM..
Old 1st February 2015 | Show parent
  #559
Gear Addict
 
Arqen's Avatar
 
🎧 5 years
Quote:
Originally Posted by proV ➡️
I don't seem to have limit on number of steps. The limitations I have is I can't save, can't get it not to crash when reaching past 1.5GB ram and the thing I miss the most, the undo button. I think you downloaded the Basic version, which is limited to 20 elements, I downloaded Standard trial version.
Thanks for clearing that up! I was wondering why a couple people reported the 20 element limit.
Old 1st February 2015 | Show parent
  #560
Lives for gear
 
Jens Eklund's Avatar
 
🎧 10 years
Quote:
Originally Posted by Arqen ➡️
It's good that you're bringing this up.

It looks like even RPG uses single diffuser modules to express their diffusion coefficients in marketing materials (see attached for an example).

And RPG researchers helped develop the most recent coefficient standards.

Since the big companies are expressing diffusion this way, I presume it's more useful to the end user to compare diffusers like this (and assume the single module is surrounded by absorption), vs to not compare them at all. As a way to compare all diffusers, baffled measurements (with one module on a wall) are problematic too, because diffusers don't all have the same width.

So I agree that the best way to compare diffusers is to compare them using wide arrays (e.g. 3.6 meter array vs 3.6 meter array).

Ideally each company would post diffusion coefficients as measured in an array. Of course, this is still not a perfect comparison because some diffusers are designed to be used in a large array and some are designed to be used in isolation (e.g., a 2.5 meter wide, symmetrical diffuser).

In this case you would expect an optimized, 3.6 meter wide diffuser to perform best, because it can take advantage of the measurement standard. It's not periodic and it fits perfectly into the 3.6 meter width. But the 2.5 meter wide diffuser will get penalized by the standard because to meet the 3.6 m requirement you need to double it, so it becomes becomes 5 meters wide, and periodic. It's being tested in a configuration it's not designed for, and in this setup it's at a disadvantage to the 3.6 meter, non periodic design.

Even if the array width approaches infinity, the comparison is not fair because periodic designs will get a performance boost (due to lobes clustering infinitely close together), where in real world applications the lobing actually hurts the performance.

At the end of the day, the most important thing is the performance for the scenario in which the diffuser will be used. In many scenarios, it will be flanked by broadband absorption. Ideally there would be multiple plots for each diffuser to show performance for various scenarios .
http://www.rpginc.com/docs%5CTechnol...efficients.pdf

Se page 38. I don't think RPG is cheating (using data gathered by measuring/simulating a single panel), but I think they sometimes are not that picky on what photos they use to illustrate something.

About different sample widths; I think that as long as you make the sample at least 16 times wider that it is deep, you're fine.

Last edited by Jens Eklund; 1st February 2015 at 04:06 PM..
Old 1st February 2015 | Show parent
  #561
Gear Addict
 
Arqen's Avatar
 
🎧 5 years
Quote:
Originally Posted by Jens Eklund ➡️
http://www.rpginc.com/docs%5CTechnol...efficients.pdf

Se page 38. I don't think RPG is cheating (using data gathered by measuring/simulating a single panel), but I think they sometimes are not that picky on what photos they use to illustrate something.

About different sample widths; I think that as long as you make the sample at least 16 times wider that it is deep, you're fine.
Yes it would make sense that they practice what they preach. I definitely trust and respect RPG, and have learned much from them.

But I must say, that photo of the measurement setup is misleading.

It's in the book Acoustic Absorbers and Diffusers (Figure 4.2), it's in product spec sheets beside the coefficient spectra plots, and it's even shown four times (Figures 1.3, 5.1, 5.7 and 6.2) in the paper on state of the art measurement standards.

I know it's easier to show a photo of just one period, but there is nothing in the captions telling readers that they should not use the setup shown in the photograph. Without a clear note telling them otherwise, the average person scanning this literature is going to take the photos literally.

Last edited by Arqen; 3rd February 2015 at 08:35 AM..
Old 3rd March 2015 | Show parent
  #562
Here for the gear
 
🎧 5 years
Exclamation

I have been reading this thread for a while and even built my 7 panels, but I have run into a question I can't seem to find an answer to here or anywhere else.

Bare with me I'm new at posting and acoustics.

My plan was to place the 7 panels, at different depths, on the back wall. The problem is that about 30 cm off the midpoint of the back wall there is a 140 cm window that about 20 cm deep (the wall is 60 cm thick). It didn't appear out of nowhere, I knew it was there ;-) but I hadn't thought about any potential off axis effects before commencing fabrication.

My first question is.

Is it worth while to place the diffuser array on the half that doesn't have a window? More generally speaking, how much of the diffusion effect will be lost at the listening position when placing a diffuser off center on a wall?

Second:
If the answer to the first question is that it is a terrible idea that will ruin balance of the room etc. Will curving the 7 panel array mitigate the problems caused by the off center placement.
Old 3rd March 2015 | Show parent
  #563
Lives for snowflakes
 
12ax7's Avatar
 
🎧 10 years
.
Hmmm...

'EA SWE's post got me to thinking:

What about making diffusers of glass (or plexi)?
.
Old 3rd March 2015 | Show parent
  #564
Lives for gear
 
Jens Eklund's Avatar
 
🎧 10 years
Quote:
Originally Posted by 12ax7 ➡️
.
Hmmm...

'EA SWE's post got me to thinking:

What about making diffusers of glass (or plexi)?
.
https://gearspace.com/board/7851341-post42.html
Old 3rd March 2015 | Show parent
  #565
Here for the gear
 
🎧 5 years
I read this thread and I think this may have been mentioned but I can't seem to find it. I've mounted 2 of the 3 diffusers, but I haven't put the middle one up yet because in order to make it stick further out from the wall I just screwed some wood to the back in the middle, which would leave the sides open. Will it affect the function if there's open space between the edges? Should I cover them?

Not a great pic but it gives you an idea:

http://s21.postimg.org/hh3e2oo7r/IMG_0063_2.jpg


Thanks!
Old 4th March 2015 | Show parent
  #566
Gear Addict
 
Arqen's Avatar
 
🎧 5 years
Quote:
Originally Posted by EA SWE ➡️
I have been reading this thread for a while and even built my 7 panels, but I have run into a question I can't seem to find an answer to here or anywhere else.

Bare with me I'm new at posting and acoustics.

My plan was to place the 7 panels, at different depths, on the back wall. The problem is that about 30 cm off the midpoint of the back wall there is a 140 cm window that about 20 cm deep (the wall is 60 cm thick). It didn't appear out of nowhere, I knew it was there ;-) but I hadn't thought about any potential off axis effects before commencing fabrication.

My first question is.

Is it worth while to place the diffuser array on the half that doesn't have a window? More generally speaking, how much of the diffusion effect will be lost at the listening position when placing a diffuser off center on a wall?

Second:
If the answer to the first question is that it is a terrible idea that will ruin balance of the room etc. Will curving the 7 panel array mitigate the problems caused by the off center placement.
How far is your back wall from your listening position?

Sounds like that window might be a problematic first reflection point. If so I would consider mounting the diffusers (or absorption, depending on your distance to the window) on a mobile stand so they cover part of the window.

I would strive for left-right symmetry on your back wall, especially if it's within 10 feet of your listening position.

A convex curve is generally a good thing in acoustics, but curving the diffuser array won't make up for lack of symmetry, or the fact that there is a window behind you. The potential problem I see is the contrast between first reflections hitting the window on one side vs the diffuser on the other side.

If you post a sketch of your room layout (with speaker placement, listening position and distance to the rear wall) we can figure out if that window is likely to create a noticeable problem.

Last edited by Arqen; 4th March 2015 at 12:06 AM..
Old 4th March 2015 | Show parent
  #567
Gear Addict
 
Arqen's Avatar
 
🎧 5 years
Quote:
Originally Posted by 12ax7 ➡️
.
Hmmm...

'EA SWE's post got me to thinking:

What about making diffusers of glass (or plexi)?
.
Certainly. You can make diffusers out of any reflective material. I believe at least one person built the Leanfuser out of polycarbonate, but I've not seen photos.

Stained glass anyone? (Control room at Manifold Recording Studio, designed by Wes Lachot)

Last edited by Arqen; 4th March 2015 at 12:39 AM..
Old 4th March 2015 | Show parent
  #568
Gear Addict
 
Arqen's Avatar
 
🎧 5 years
Quote:
Originally Posted by Nick_B ➡️
I read this thread and I think this may have been mentioned but I can't seem to find it. I've mounted 2 of the 3 diffusers, but I haven't put the middle one up yet because in order to make it stick further out from the wall I just screwed some wood to the back in the middle, which would leave the sides open. Will it affect the function if there's open space between the edges? Should I cover them?

Not a great pic but it gives you an idea:

http://s21.postimg.org/hh3e2oo7r/IMG_0063_2.jpg


Thanks!

Yes. I suggest you cover those gaps. They will affect the performance.

What type of wood / finish did you use? I like the look.
Old 4th March 2015 | Show parent
  #569
Here for the gear
 
🎧 5 years
It's just sanded plywood with veneer on the edges. I stained it espresso and polyurethaned it. I'll post finished pictures as soon as I finish the sides of the middle piece.
Old 4th March 2015 | Show parent
  #570
Here for the gear
 
🎧 5 years
Arqen

You will have to excuse the sketch. It is not really to scale and the windows are a little off in their positioning and scale.

Hopefully it is evident that there are 4 "niches" spread over two walls.

The distance from the listening position to the wall is roughly 2 meters, but it is negotiable since there is no other furniture at the moment. The room is waiting to get renovated and I'm trying out different acoustic ideas since I can just fasten things to ceiling and walls without worry.

Thanks for your help!
Attached Files
File Type: pdf RoomLayout.pdf (36.3 KB, 175 views)
📝 Reply

Similar Threads

Thread / Thread Starter Replies / Views Last Post
replies: 1366 views: 1114825
Avatar for chrispire
chrispire 26th March 2021
replies: 2619 views: 475406
Avatar for S21
S21 22nd March 2021
replies: 75 views: 31916
Avatar for massimomucci
massimomucci 11th June 2020
replies: 88 views: 5860
Avatar for TobyB
TobyB 1 day ago
Topic:
Post Reply

Welcome to the Gearspace 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…

Forum Jump
Forum Jump