Custom made diffusors?

[Jeffrey Hedback]

Great stuff Jens in both data and presentation! +1

Diffusors.com: since you've mentioned my name in association with your product I'd like to expand my experience. The Procella Demo room at CEDIA 2012 was (to me) THE BEST in show...beyond a doubt. I did take notice of the diffusors and their implementation neither of which being typical. The CD design of the Procella speakers is equisite. All I can truly say to your diffusors is that if they were poorly designed, that application would have exposed them...now PLEASE note that this is not an endorsement. How they may work in a situation with lesser speaker is not an assumption I would be willing to make. I look forward to testing data of your units which is an expected obligation.

BWO- apologies offered on this side note to your thread.

[Glenn Kuras]

An option could be to use QRDude and get a carpenter to build them. But I would feel safer having someone who's well acquainted with building diffusors and has measured the end result

If the carpenter follows the build correctly you will be fine, but you're right, can't tell you how many studios I have visited and the diffusion was built wrong (not just talking about depth either).

[bwo]

Originally Posted by Glenn Kuras

If the carpenter follows the build correctly you will be fine, but you're right, can't tell you how many studios I have visited and the diffusion was built wrong (not just talking about depth either).

Yes, if they are built correctly. I wouldn't bet on it, when a carpenter hasn't done something like that before.

Makes me think about those who have glued everything by hand and nothing looks like it's straight. Sure,it diffuses, but with how much lobing issues?

[Glenn Kuras]

Not to brag (ok I am bragging ) but when we build our Q7D it is build around a physical guide system. Basically if it does not fit in the guide you can't build it. we can build a 1000 of them and everyone of them will be exactly the same.

[bwo]

Originally Posted by Glenn Kuras

Not to brag (ok I am bragging ) but when we build our Q7D it is build around a physical guide system. Basically if it does not fit in the guide you can't build it. we can build a 1000 of them and everyone of them will be exactly the same.

Out of curiousity. Is there a spesific reason why you built it so narrow? Freight cost?

[Glenn Kuras]

Originally Posted by bwo

Out of curiousity. Is there a spesific reason why you built it so narrow? Freight cost?

Basically yes. Have thought about offering a 2'x2' one though

[Nordenstam]

Hei Bjørn!

Having seen your nicely built frames, I have no doubt that you can build your own diffusers as well. Especially these ones!

Tooting my horn here, to push one of my design onto you. For free.. I'd be delighted to see if they perform to expectations both in ease of build and desired room response. The primary design goal for this project was a diffuser that I could provide for the relatively low budget projects I'm involved in. Don't want price to be a factor stopping people from having lots of diffusion in their rooms.



It was "simplefuser" in my mind for a while. It's really dead simple to construct. If you can cut the planks to the desired length and use a drill to lock it all up, you can build this! Though the simplefuser name didn't quite have a fancy enough ring to it. So out of the aether popped the acronym NERD; Nordenstams Evolved Root Diffuser. Aside from the obvious punage on PRDs and the fact that there are some traces of primitive root in there, the name also plays on the evolution from math sequences to computer optimized designs.

One of the big differences between a diffuser as this one and the math sequences is that this is an object, as in a singular entity. It's meant to perform on its own, it's not meant to be part of a larger array. (and since these ones are asymmetrical - use them in symmetrical pairs in the room to get even overall response) - There's also a fundamental difference in that the optimized diffusers shown here are not scalable, each design is made to suit a particular size. An inch or two to or from doesn't change much, but you can not take a 2 feet wide design and scale it up to 3 feet.

The computer optimization works as wonderful here as it does in other engineering fields. It gives unpreceded possibilities in designing for certain criteria. The chances of making a good diffuser using random numbers are pretty slim. Chances of making a good diffuser using certain design goals which have nothing to do with acoustics is even slimmer. As in the goals leading to the diffuser design shown above. It is built around practical concerns, not theoretical ideals which are hard to implement in real life. Using the excellent Reflex simulator by AFMG have given the possibility to design bricks that fly, so to speak. Just like modern military airframes are aerodynamically unstable and flies against common sense, these ones diffuses despite looking way too simple.

Hope this doesn't come across as too spammy. It's hard to write something like this without sounding like a spambot. The Reflex program is available for anyone.. It gives the freedom to custom design (1D) diffusion for just about any purpose. It's a pretty neat evolution in diffusion. I don't have a design, it's wide range of starting points, which is easy to adapt to any particular local materials and any particular task. For example, the diffuser gets sound from different incoming angles when it's placed flat on the rear wall, compared to a diffuser on an angled rear wall or sidewall. So I make different ones to suit the different tasks! Find it to pretty amazing to have such fine tuning options available. In addition to the more obvious frequency and distance related customizing, and making it suit whatever local materials that are cheap and handy.

The low frequency range of the diffuser depends on size too, not only the depth. You'll need to go beyond the standard 2 feet width to get lower diffusion. I think the main point of the diffuse return in LEDE is to get a wide frequency range high energy blast. Supporting diffusion down to the Schroeder crossover zone makes a lot of sense to me. These larger diffusers do. The subjective impression is fullness and the oh so cliché "warmth".

The 3x distance thumb of rule is perhaps more important for diffusers with dividers. For these dividerless designs, I'm utterly unable to hear any strange effects unless being silly close to the diffuser. Have found the same to be true for the (dividerless) 2D diffusers.


Have made a bunch of simulations to show that these brick are indeed able to fly. Have made too many, rather than too few, as I know these graphs may be interesting to some readers. As can be seen, all QRD's are not created equal. Ditto with PRD's, despite the theory treating any such sequence as having equal performance. It's also interesting to note the divider vs dividerless behaviour.

All models and simulations are 1D, ie, the object is assumed to be of infinite length. The height of the diffuser needs to be at least the width. Using same pattern, same width and different heights (in excess of width) gives different sounding diffusers. (!) None of these simulations accounts for effects like that.

Here are some larger size diffusers that may suit Bjørns stated goals;











and since I was curious, here's the same NERD diffuser as above, with 3mm dividers inserted between each piece;



I'll continue in another post as this is going to be very long..

[Nordenstam]

..

Had already made quite a few graphs for a nearly 2 feet wide design. For easy comparison to the established standard. As Jens knows all too well, it takes for ever and ages to generate these graphs! So I'll stick to the ~2 feet designs below.














The graphs above are diffuse field simulations. Hence they have the same size related boost in the low mids as is seen in the reverb chamber measurements of bass traps.


Here's an empty box to show this point; 1x22cm endpiece, 56cm backing plate, 1x22cm endpiece:




The diffuse field is the response across all angles. It's not particularly relevant to say rear wall use where most of the sound will come at around minus to plus 30 degrees. Checking what's up at some particular angle of incidence gives different stories than the graphs above tells.


Here are the same diffusers at zero degree incidence:












The empty box at zero degree incidence:



Then again, this doesn't mean that much, as the response changes at plus and minus 20 degrees. And so forth. It takes both a multitude of graphs and a familiarity with the expected values to evaluate potential performance using the simulations (and real measurements). (as usual)

Here are the same diffusers at 20 degrees incidence:












The empty box at twenty degree incidence:


Let me again stress that these graphs are not giving the full picture.


Have included quite a few comparisons with and without dividers. It's an interesting topic. I believe the dividerless designs have an asset in fewer losses, providing stronger return energy. Smoother shapes are probably even better in that regard. But that would defeat the ease of build goal in this particular design.

Hope these graphs may be interesting to some. It took ages to generate!

..

The main point is that the answer to the question in the original post is: yes.

Though if you want to try this route, Bjørn, you'll have to build them yourself. Unless you can wait for me to have a studio build going where producing some for you comes natural. It doesn't seem to make time/money sense to set up a wood workshop dedicated to this purpose. Maybe.. Perhaps.. But not right now. Am silly busy these days.

It may also be that you think they are too ugly. That's fine too. There are of course other options if you want to pay more for the build. Can provide plans for that too.


Cheers from your local diffuser NERD,

Andreas Nordenstam

[Jens Eklund]

Wonderful posts Andreas!


I’ve had similar plans, setting up a database for diffuser performance but since the current version of Reflex only can utilize 2 Gb of RAM, it’s not possible to render high resolution plots of repeated periods consisting of higher order QRDs. Next release will.

Originally Posted by Nordenstam

For easy comparison to the established standard.

Originally Posted by Nordenstam

Let me again stress that these graphs are not giving the full picture.

This is a very important point: A single period can look impressive in terms of performance but in order to evaluate the actual performance of a diffuser, the total panel width needs to be 3,6 meters in order to be able to compare to other products. One single panel is not going to do much in terms of redistribution energy if mounted on an otherwise flat surface. We’ve discussed this previously:

http://www.gearslutz.com/board/6014298-post14.html
(The 2,4 meter total period with chosen in the post above was in order to directly compare our panels directly to one another.)

Compare the single QRD N7 to the panel consisting of 4 of them (2,4 meter total). The diffusion coefficient drops significantly.

I know that you know all this but others might not, and it is essential to understand the difference in order to make an educated comparison of diffuser performance.

I would love to see some graphs of your model if using 6 panels (totaling up at 3,48 meters total panel width).


Sincerely Jens Eklund

[bwo]

Originally Posted by Nordenstam

One of the big differences between a diffuser as this one and the math sequences is that this is an object, as in a singular entity. It's meant to perform on its own, it's not meant to be part of a larger array. (and since these ones are asymmetrical - use them in symmetrical pairs in the room to get even overall response) - There's also a fundamental difference in that the optimized diffusers shown here are not scalable, each design is made to suit a particular size. An inch or two to or from doesn't change much, but you can not take a 2 feet wide design and scale it up to 3 feet.

The computer optimization works as wonderful here as it does in other engineering fields. It gives unpreceded possibilities in designing for certain criteria. The chances of making a good diffuser using random numbers are pretty slim. Chances of making a good diffuser using certain design goals which have nothing to do with acoustics is even slimmer. As in the goals leading to the diffuser design shown above. It is built around practical concerns, not theoretical ideals which are hard to implement in real life. Using the excellent Reflex simulator by AFMG have given the possibility to design bricks that fly, so to speak. Just like modern military airframes are aerodynamically unstable and flies against common sense, these ones diffuses despite looking way too simple.

Thanks for chiming in.

I have a few questions. What exactly do you mean by being built for practical concerns? I remember you also saying that a calculator like QRDude has nothing to do with how a diffuser performs in real life. Can you please elaborate on this?

If these are meant to work on their own and not in an array, how exactly do you place them in a room? How would go about about finding the right positions? Not sure what you meant by symmetrical pair.

Also, do they have any diffusion from 8k to 20k?

[John White]

Absolutely fascinating Andreas. Very interested in how this progresses. Perhaps a dedicated thread-

[Nordenstam]

Hello!

Originally Posted by bwo

What exactly do you mean by being built for practical concerns?

First generations diffusers where based on math sequences. The shape of the design is set by the math, it is then later sought practical ways to actually build the structure. Modern diffusers based on computer aided designs may start out as an aesthetic ideal which sets the constraints for the search. The NERDs started with an idea for an easy and affordable way to build big wood diffusers.

The computer aided searching for good response can be done by brute force, simply trying random combinations and wait for one that satisfy the set criterias. It can take ages that way. Having some way to guide an automated search in the right direction speeds it up significantly. These diffusers are made by using an excel sheet to get some decent starting points and from there on, it's manual tweaking. It ends up being a system that is relatively easy to use for making 1D diffusers to any size.

Originally Posted by bwo

I remember you also saying that a calculator like QRDude has nothing to do with how a diffuser performs in real life. Can you please elaborate on this?

As I started on above; QRDs are made in the realm of math. Namely, in the frequency domain. The underlying shape is presumed to repeat an infinite number of times. As is always the case with frequency domain analysis. So they are first and foremost made to work well as repetetive sequences, not as stand alone units. The theories for repetitive sequence diffusers concentrate on a few particular frequencies. All other frequencies are what they accidentally happen to be. That is significant in some cases. Practical tests shows there is a sorta good enough correlation between the math theories and the actual results. Good enough to work endlessly much better than the previous alternative, which was none.

The correlation between theory and reality is stronger if actually attempting to model how acoustical waves behaves when they strike the object. (Of course given an accurate enough model.) The computer models used are not giving 1:1 results compared to reality and acoustics is particularly hard to model. Yet the simulations are giving a wastly better idea of what's actually going to happen compared to the math theories. The most obvious benefit is that it frees the design from the constraint of the repetetive sequences that comes from working in the frequency domain. One can concentrate on the behaviour of the shape itself, ignoring the response as an array with many repeats. Or concentrate on how it behaves as a modulated array with a few particular units involved. It also means that all frequencies are created equal as there are no particular design frequencies.

Originally Posted by bwo

If these are meant to work on their own and not in an array, how exactly do you place them in a room? How would go about about finding the right positions? Not sure what you meant by symmetrical pair.

They will work fine in a modulated array too, if the array is a good sort of array. Though I'd make it a bit different if a continous array (without gaps, angles etc) is the sole purpose. The sort of arrays I usually like to do can not be modeled in Reflex as they feature gaps, angles and commonly combinations of different diffusers. Like 2D. As far as I know, there is no computer model good enough to simulate highly complex structures such as a (relatively) small room setup. Creating an array with several units is pretty simple if each diffuser have good performance on its own.

These are custom made for (relatively) small room acoustics. They are not meant to be used in repeated/periodic arrays. Periodic structures are dominated by the resulting periodicity lobes. Good periodic diffusers will have a good spread of energy within in the periodicity lobes. There are many situations where periodic performance does not matter, especially so for small room acoustics. They'll still work fine if used in modulated arrrays and somewhat decent enough as a repetetive device. They're not that bad. Yet there are some finer points that are tweaked in certain directions if it is intended to be used specifically for modulated or periodic arrays.

Each of the asymmetrical diffusers will in itself have a fairly even response. Though they are indeed asymmetrical. Putting them in a mirrored/symmetrical pair will guarantee stereo symmetrical response in a listening room. So I presume these to come in pairs for most listening room applications.

Placement is dependent on application. The bigger deeper units works nicely in applications calling for a strong return, such as the rear wall and/or rear sidewalls in a LEDE room. Diffusers can also be used on front wall and ceiling in typical listening rooms. Or just about anywhere in a tracking room or a highly diffuse room (reflection rich zone or whatever it's called). It all depends on what you want to achieve.

Originally Posted by bwo

Also, do they have any diffusion from 8k to 20k?

Yes. At such complexity as these structures provides, it's nearly impossible not to have ample diffusion from 8k to 20k. I think it's reasonable to assume the trend as seen for three octaves from 1.25-10K holds true for the last octave from 10-20k as well. Although I do not know specificially what happens up there, as Reflex stops at 10k.

[Nordenstam]

Originally Posted by Jens Eklund

This is a very important point: A single period can look impressive in terms of performance but in order to evaluate the actual performance of a diffuser, the total panel width needs to be 3,6 meters in order to be able to compare to other products. One single panel is not going to do much in terms of redistribution energy if mounted on an otherwise flat surface. We’ve discussed this previously:

http://www.gearslutz.com/board/6014298-post14.html
(The 2,4 meter total period with chosen in the post above was in order to directly compare our panels directly to one another.)

Compare the single QRD N7 to the panel consisting of 4 of them (2,4 meter total). The diffusion coefficient drops significantly.

I know that you know all this but others might not, and it is essential to understand the difference in order to make an educated comparison of diffuser performance.

I would love to see some graphs of your model if using 6 panels (totaling up at 3,48 meters total panel width).

This is a bit of a big topic.

The result of repeating a diffuser will be dominated by the periodicity lobes created by repetetition and bears no relation to how the diffuser itself behaves. For any given comparison, an aperiodic surface with no repeats will show the best performance. Anything featuring repeats will show lower levels in the diffusion coefficients and the lowest levels will always be seen for a purely repeating structure.


This is 6 repeats of the 58cm unit:



This is the same 6 (asymmetrical) units, with two of them reversed in direction to create a modulated array:


Notice that those graphs shows the 6 exact same units. The only difference is the composition of the array, purely periodic or modulated. This is a topic of its own which I'd rather stay away from as there are endless possibilites for modulated arrays while there is only one possible graph for the diffuser itself.

This is in any case irrelevant. I don't mass produce 58cm wide units. It was just an example a possible setup. I would never recommend repeating it like that. If someone wants a 360cm wide setup and can live with an asymmetrical array, then I'll make a 360cm wide diffuser for them.

Like this 360cm wide NERD setup:


Which is the only graph representative of what a NERD may look like at exactly 360cm.

As this is a big topic and deserves consideration (and lots of graps! ) I'll make another thread for that when theres some free time available.

very interesting posts!
Thank you Andreas