These numbers are very different from the ones in Cox/`D`Antonio "acoustic absorbers and diffusors"4 periods semicylinders, width 60cm, radius 30cm(very good book by the way):
0.51 100hz
0.43 125hz
0.35 160hz
0.30 200hz
0.24 250hz
0.20 315hz
0.25 400hz
0.62 500hz
0.37 630hz
0.21 800hz
0.26 1000hz
0.39 1250z
0.32 1600hz
0.48 2000hz
0.44 2500hz
0.42 3150hz
0.43 4000hz
0.54 5000hz
Here is coeffecients for 2 periods
0.66 100 hz
0.66 125
0.64 160
0.57 200
0.47 250
0.46 315
0.71 400
0.70 500
0.76 630
0.51 800
0.55 1000
0.74 1250
0.66 1600
0.78 2000
0.77 2500
0.73 3150
0.77 4000
0.80 5000
These coefficients is without any variation to the pattern (different width/debth)
Both mathematical and geometric diffusors work well when implemented in a good way.
Regarding the rpg paper about polys, the test results showing combfiltering from a poly, it looks similar to one in the book "acoustic diffusors and absorbers", figure 10.18 10.19, the width of the poly was 1 meter, would be interesting to know distance.
These numbers are very different from the ones in Cox/`D`Antonio "acoustic absorbers and diffusors"4 periods semicylinders, width 60cm, radius 30cm(very good book by the way):
I say they correlate well considering the different depth and the fact that I used 2,4 meter total period for that rendering instead of the usual 3,6 meter total period (as in AAaD) used when extracting diffusion coefficients (and the reason for me using 2,4 meter instead of 3,6 is stated in the original post): QRD and Skyline Well Dividers?
Jens, in the original thread I mentioned 2-3 diffusors.
There are several options one can use for optimising the effect of poly diffusors when used in arrays:
- if the diffusors have same width/debth one can introduce randomness by varying spacing,
-one can use absorptive surfaces between diffusors,
-use difusors with different width/debth
Another method is to have the diffusors "free standing" on the floor with varying distance, introducing good temporal diffusion, spreading the arrival time of the attenuated reflections.
When poly diffusors are placed in arrays, there will be temporal diffusion introduced
Last edited by hsal; 2nd October 2011 at 07:45 PM..
Reason: edited, added info reffering to 2-3 diffusors having good coeffecient
I never said polys are useless, just not as good as “real” diffusers but certainly easier to build (if DIY at least).
Both mathematical and geometrical diffusors are valuable tools, there are great sounding rooms being build using poly diffusors where it is not a result of having to settle with a "inferrior " design ,some examples are Galaxy studios (Belgium), and the opera house in Oslo.
Last edited by hsal; 3rd October 2011 at 06:32 AM..
Reason: morning coffe kicks in
Living here in Indonesia, I get to observe the extremes of human behavior & I find is every so curious how they don't care how things work or sound as long as it looks good. My point is, don't build something because you are impressed at how it looks, build what you need for your room.
NOTE: Very interesting statement from the Measuring Diffusivity document linked by Jeff; "These measurements demonstrate that corrugated diffusors exhibit erratic behaviour per frequency and are only modestly effective at accomplishing the re-direction of impinging sound waves across widely varied angles of reflection. The curved surface of the polycylindrical device in comparison presents markedly superior smoothness of reflection response across the frequency band and more consistent angular dispersion of reflected sound regardless of the angle of incidence or observation." - don't 'ya love it!?
Cheers,
John
__________________ John H. Brandt
Recording Studio, Performance Hall & Architectural Acoustics Consultants
(American) in Jakarta, Indonesia
AFMG reflex modeled diffusion coefficient for 9 poly diffusors, total width 520cm:
from around 320hz and up coefficient is 0.38-0.5 , in other words good even diffusion across the frequensy spectrum, (calculation stopped at 5000hz in order to shorten rendering time, this calculation took 20 minutes).
from around 320hz and up coefficient is 0.38-0.5 , in other words good even diffusion across the frequensy spectrum, (calculation stopped at 5000hz in order to shorten rendering time, this calculation took 20 minutes).
If one can spare 363 mm of depth and non-repeating periods, it is possible to build diffusers using polys that work ok. Can you render in higher resolution 1/12 or 1/24 or perhaps post the Reflex model so I can do it. Sometimes it looks super using 1/3 oct resolution but not so great when using higher.
I have attached a file with data, 1/12 resolution, stopped at 4k in order to shorten rendering time..
Thanks!
Quote:
Originally Posted by hsal
36cm debth is at one point, the rest is appr 20-25cm out from the wall.
but still, the total depth is 363 mm and if you have that amount of depth to play with, it’s not too hard finding good shapes that offers good diffusion coefficients.
Here's two more scenarios/coefficents.
The first with just dual polys, total width 2m.
The second one is a combination of 5 polys. Theres a small gap between them (5cm), the width is 3,6 meters, max depth is 30cm but only 10-15% of the array is deeper than 20 cm as you can see.
PS. I can understand how the polar plot is useful for predicting behavior and usage in the studio treatment but I never understood the usage for the diffusion coefficient for other than perhaps comparing devices. And I'm not really clear on that either.
__________________
:: Christian Borg
"It's the sound itself that's important and the ability to work with it, not where it comes from. They [musicians] are manipulators of sound, not purists." - Bob Moog
Unless you have the time to look at all the combinations of incidence and receiving angle of the polar plots, the diffusion coefficient is what matter since this represents all the polar plots compressed into one number. If one or more plots are “bad” (not evenly spread energy in all directions) the diffusion coefficient goes down. The scattering coefficient only tells you that the surface redirects energy from the geometric path, not how evenly the energy is spread in the operational plane. A flat plane, slightly angled will show a high scattering coefficient for certain angels but it is naturally not a good diffuser since it does not spread the energy in all directions.
Also, just so no one misses this; a low number of ploys will not offer any significant temporal scattering and two will offer none (only one additional reflection can hardly be called temporal scattering).
I think one reason might be that often rather shallow polys are used (easier to bend, less stress on the structure), as seen in the picture below, shallow diffusors side by side will make for a very uneven polar plot, in some directions there will scattered more high frequency content , which can create a unwanted high pitched effect
red =500hz pink=4000hz
when you add spacing, the polar plot is much more even
With a combination of three deep polys and one shallow, the polar plot is better for polys adjacent to eachother.
another possible reason that polys traditionally have had a spacing between them, is that they often are identical, as shown in the data some posts above, identical polys without spacing have very low diffusion coefficient in larger arrays.
I wrote:
Quote:
yes, obvious benefits of avoiding repetition in larger arrays with diffusors side by side, these numbers are comparable to good qrd designs.
Quote:
I can add that these numbers only show spatial diffusion coefficient.
having worked with afmg a couple of weeks, it is appearant how much work Jens Eklund has put into the design of his diffusor, kudos to him.
One aspect not being mentioned, is that polys have a very low % of surface area compared to mathematical diffusors, and because of this should maintain a higher magnitude of reflection. This can come in handy when trying to make a smaller room livelier.
__________________
phantom power doesn't make your voice sound spooky
Here's what I do. Free if you like, pay if you REALLY like:
One aspect not being mentioned, is that polys have a very low % of surface area compared to mathematical diffusors, and because of this should maintain a higher magnitude of reflection.
That would depend on a lot of factors:
* Reflectiveness of the material used. * Angle of incidence. * Frequency range considered. * Type of diffuser compared to.
having worked with afmg a couple of weeks, it is appearant how much work Jens Eklund has put into the design of his diffusor, kudos to him.
Thanks!
Yes, I have spent way too much time with Reflex ...
I’m thinking of perhaps devloping a budget version of the Optiffuser, about half as deep and only 600 x 600 mm per panel, but considering the time it takes to find the best possible stepped shape using positive and a negative panels in different configurations to reach a total of 3,6 meter panel width (the standard width when comparing diffusion performance), we’ll see if it happens anytime soon …
What is the significance of your graphs? They are for adjacent polydiffusers. Not a real world scenario.
This post is for the benefit of users who are not familiar with your continuing diatribe against polydiffusers.
Andre
You just volunteered to give us an example of a real world design of a poly diffuser I can model in Reflex so that we can compare the results.
Please make it 3,6 meter wide in total (since this is the standard for comparing diffusers) and a total of 220 mm effective depth (because I already have a lot of data for different models using this depth and it is also a good balance between space and performance).
If you are not willing to provide me with an example design using polys, I suggest that you keep your opinions to yourself.
EDIT:
Oh, and I have nothing against polys as long people understand the difference between them and normal diffusers.
EDIT 2:
For those who wants to learn about the limitations of polys (and other types of diffusers), I strongly recommend the book “Acoustic Absorbers and Diffusers, SE”.
Above shows coefficients only for -45 degree incidence (and polar for -68 deg). Can you post the data for random incident (or post the Reflex model so I can render it)?
You just volunteered to give us an example of a real world design of a poly diffuser I can model in Reflex so that we can compare the results.
Please make it 3,6 meter wide in total (since this is the standard for comparing diffusers) and a total of 220 mm effective depth (because I already have a lot of data for different models using this depth and it is also a good balance between space and performance).
The restrictions are limiting in the "total effective depth" for polys.
4 440 mm chord polys with an included angle of 90° with 613 mm between the polys.
4 440 mm chord polys with an included angle of 180° with 460 mm between the polys and 230 mm at the ends of flat surface.
3 440 mm chord polys with an included angle of 180° and 760 mm between them
3 762 mm chord polys with an included angle of 120° and 657 mm between them
Those are all 220 mm deep polys, if I did my calculations right.
Quote:
If you are not willing to provide me with an example design using polys, I suggest that you keep your opinions to yourself.
The ball is in your court. You set the rules, now please play by them.
hasty post,here is random incidence for diffusion coefficient and 0 degree incidence for polar
I will post afmg report later (rendering time).
As can be seen, good spatial diffusion coefficient from 400hz and up (coefficient mostly above 0.4).
I am currently testing out different designs,
The first polys I created did not work for the task at hand, placement at sidewalls, they where to shallow to have a good effect, current designs work very well, also in settings with live instruments.
New Reply 
My point is, don't build something because you are impressed at how it looks, build what you need for your room. 
