DIY Binary Amplitude Diffuser anyone?

**jayfrigo** · 6th June 2009

These work on amplitude rather than phase (amplitude gratings rather than phase gratings), and scatter above 1 kHz. They are advantageous for some applications because of their limited depth.

They are absorptive below 1 kHz, and based on the depth of absorption behind the grating panel (usually 1-4") they can absorb some bass as well while minimizing absorption above 1 kHz. Size of holes (usually 1/2" or 5/8") affects higher frequency attenuation. These surfaces can also be curved in various ways for some applications.

The idea is to use a binary sequence for the holes that tries to keep the number of ones and zeros even which helps more uniform distribution of the diffuse reflected sound. Hole spacing was originally inspired by Angus' work on MLS (Maximum Length Sequences). Optical sequences usually result in too few 1s in long seuences, and some calculations are likely to result in too many 1s. Generating a familiy of sequences and looking for optimized strings with even numbers of 1s and 0s is necessary.

Some trial and error will be required. Cox/D'antonio suggest that computer searches are a manageable way to do this for sequences up to 20, but finding every possible combination for sequences above 20 becomes unwieldy. For a sequence with 24 1s and 24 0s (N=48), there are 10^13 combinations to search! Wrapping a 1D sequence into a 2D array requires Chinese remainder theorem.

Curved hybrid surfaces, and variations in absorption, including Helmholtz properties and possible hole spacing variations, such as reducing open hole area to increase bass absorption, are the next things to investigate when designing binary amplitude diffusion panels. These things are complicated, time-consuming to design, and it's probably better for sanity's sake to simply buy them from RPG!

jrubbernek · 8th June 2009

Quote:

Originally Posted by jayfrigo

These work on amplitude rather than phase (amplitude gratings rather than phase gratings), and scatter above 1 kHz. They are advantageous for some applications because of their limited depth.

They are absorptive below 1 kHz, and based on the depth of absorption behind the grating panel (usually 1-4") they can absorb some bass as well while minimizing absorption above 1 kHz. Size of holes (usually 1/2" or 5/8") affects higher frequency attenuation. These surfaces can also be curved in various ways for some applications.

The idea is to use a binary sequence for the holes that tries to keep the number of ones and zeros even which helps more uniform distribution of the diffuse reflected sound. Hole spacing was originally inspired by Angus' work on MLS (Maximum Length Sequences). Optical sequences usually result in too few 1s in long seuences, and some calculations are likely to result in too many 1s. Generating a familiy of sequences and looking for optimized strings with even numbers of 1s and 0s is necessary.

Some trial and error will be required. Cox/D'antonio suggest that computer searches are a manageable way to do this for sequences up to 20, but finding every possible combination for sequences above 20 becomes unwieldy. For a sequence with 24 1s and 24 0s (N=48), there are 10^13 combinations to search! Wrapping a 1D sequence into a 2D array requires Chinese remainder theorem.

Curved hybrid surfaces, and variations in absorption, including Helmholtz properties and possible hole spacing variations, such as reducing open hole area to increase bass absorption, are the next things to investigate when designing binary amplitude diffusion panels. These things are complicated, time-consuming to design, and it's probably better for sanity's sake to simply buy them from RPG!

Yikes, so for a rooky like me then I should stay away from these diffusers and just stick with RPGs or QRDs?

I'd love to figure out how to make these though...any tips?

**jayfrigo** · 8th June 2009

The build is fairly straightforward, if tedious making all the holes just right. It's the design that's a pain, laying out the amplitude grating. It may be easier to just copy the sequence of an existing one, but it will be hard to find actual plans online I think. I remember one person trying to do a DIY one, but I'm not sure how he arrived at his sequence. You'd have to double check it to make sure it's not just some random spacing or a poor attempt at a copy, or it won't be very effective.

Nordenstam · 10th June 2009

Quote:

Originally Posted by jrubbernek

I'd love to figure out how to make these though...any tips?

Cox and D'Antonio's book.

boggy · 29th July 2011

Quote:

Originally Posted by jayfrigo

.........I remember one person trying to do a DIY one, but I'm not sure how he arrived at his sequence.........

DanDan ask me to post binary sequence on Gearslutz forum in thread REW Analysis, then, instead of starting a new thread, I'll use this old one that has a nice explanation of binary diffusers functioning (thanks to jayfrigo

), for posting a part of MLS pseudorandom sequence.

Here are a first 127 members of pseudorandom sequence (MLS) that may be used for building binary diffusers. I hope that this is enough for most applications. If someone need more, please ask, and we will generate more...

Sequence:

Code:

position:126 ***** 1
position:125 ***** 0
position:124 ***** 0
position:123 ***** 0
position:122 ***** 0
position:121 ***** 0
position:120 ***** 0
position:119 ***** 1
position:118 ***** 0
position:117 ***** 0
position:116 ***** 0
position:115 ***** 0
position:114 ***** 0
position:113 ***** 1
position:112 ***** 1
position:111 ***** 0
position:110 ***** 0
position:109 ***** 0
position:108 ***** 0
position:107 ***** 1
position:106 ***** 0
position:105 ***** 1
position:104 ***** 0
position:103 ***** 0
position:102 ***** 0
position:101 ***** 1
position:100 ***** 1
position:099 ***** 1
position:098 ***** 1
position:097 ***** 0
position:096 ***** 0
position:095 ***** 1
position:094 ***** 0
position:093 ***** 0
position:092 ***** 0
position:091 ***** 1
position:090 ***** 0
position:089 ***** 1
position:088 ***** 1
position:087 ***** 0
position:086 ***** 0
position:085 ***** 1
position:084 ***** 1
position:083 ***** 1
position:082 ***** 0
position:081 ***** 1
position:080 ***** 0
position:079 ***** 1
position:078 ***** 0
position:077 ***** 0
position:076 ***** 1
position:075 ***** 1
position:074 ***** 1
position:073 ***** 1
position:072 ***** 1
position:071 ***** 0
position:070 ***** 1
position:069 ***** 0
position:068 ***** 0
position:067 ***** 0
position:066 ***** 0
position:065 ***** 1
position:064 ***** 1
position:063 ***** 1
position:062 ***** 0
position:061 ***** 0
position:060 ***** 0
position:059 ***** 1
position:058 ***** 0
position:057 ***** 0
position:056 ***** 1
position:055 ***** 0
position:054 ***** 0
position:053 ***** 1
position:052 ***** 1
position:051 ***** 0
position:050 ***** 1
position:049 ***** 1
position:048 ***** 0
position:047 ***** 1
position:046 ***** 0
position:045 ***** 1
position:044 ***** 1
position:043 ***** 0
position:042 ***** 1
position:041 ***** 1
position:040 ***** 1
position:039 ***** 1
position:038 ***** 0
position:037 ***** 1
position:036 ***** 1
position:035 ***** 0
position:034 ***** 0
position:033 ***** 0
position:032 ***** 1
position:031 ***** 1
position:030 ***** 0
position:029 ***** 1
position:028 ***** 0
position:027 ***** 0
position:026 ***** 1
position:025 ***** 0
position:024 ***** 1
position:023 ***** 1
position:022 ***** 1
position:021 ***** 0
position:020 ***** 1
position:019 ***** 1
position:018 ***** 1
position:017 ***** 0
position:016 ***** 0
position:015 ***** 1
position:014 ***** 1
position:013 ***** 0
position:012 ***** 0
position:011 ***** 1
position:010 ***** 0
position:009 ***** 1
position:008 ***** 0
position:007 ***** 1
position:006 ***** 0
position:005 ***** 1
position:004 ***** 1
position:003 ***** 1
position:002 ***** 1
position:001 ***** 1
position:000 ***** 1

here are some examples:

Good luck!

Boggy

DanDan · 30th July 2011

Thanks Boggy.
DD

PaulP · 30th July 2011

The easiest for us DIY'ers is to copy somebody else's patent. Just don't try to sell them :-)

RPG Bad Panel -- Detailed Plans

technocolour v2 · 2nd August 2012

Quote:

Optical sequences usually result in too few 1s in long seuences, and some calculations are likely to result in too many 1s. Generating a familiy of sequences and looking for optimized strings with even numbers of 1s and 0s is necessary.

Some trial and error will be required. Cox/D'antonio suggest that computer searches are a manageable way to do this for sequences up to 20, but finding every possible combination for sequences above 20 becomes unwieldy. For a sequence with 24 1s and 24 0s (N=48), there are 10^13 combinations to search!

Sorry to resurrect an old thread, but surely the most efficient means of generating sequences would be to start off with an even number of 1's and 0's (eg 12 of each) and randomise their position in the sequence rather than attempting to find a contiguous length of evenly distributed 1's and 0's from a random sequence?

OpusOfTrolls · 2nd August 2012

The important part is "spectrally flat" (silly word). Although the layout could probably be optimized to take early reflections into account, which would be an intensive experience in mathematics.

Syncamorea · 31st August 2012

Quote:

Originally Posted by boggy

here are some examples:

Boggy, those are beautiful treatments. Did you do the design and build of them? Are you happy with their performance? My interest is peaked by the fact that those rooms are smallish and I'm trying to understand more about options for diffusion in small rooms. What can you tell me about the construction of those features? Behind the black fabric - is that just an everyday ordinary broadband absorber?

boggy · 31st August 2012

Quote: