3D printing a QRD N53 diffuser

[Esprit]

How much do you save compared to a product like this?

Diffusore Quadratico Gotham N23 5" (14cm) - GIK ACOUSTICS ITALY

Lo strabiliante diffusore quadratico skyline Gotham N23 5" offre ineguagliate perfomances nel controllare le riflessioni mantenendo allo stesso tempo presenza e vitalità nelle alte frequenze facendo suonare una stanza come se fosse più grande.

gikacoustics.it

 

[restorer-john]

For the next step, I'm investing in a nicer printer than I have now, the Bambu Labs P1S, which is apparently set to go on sale tomorrow. It prints about 4x faster.

This entire project is an excuse to buy a better 3D printer isn't it??

 

[restorer-john]

I'm printing this to test it out.

Print this instead:

 

[Curvature]

I definitely will, although I am not sure how visible a single large diffuser will be in typical measurements. I don't *think* my room has super-strong resonances in the frequencies it's going to affect, and I don't know that it would affect overall RT60 much?

To the group: Does anyone know exactly what type of measurement will tend to best reveal any differences from a diffuser? Will be happy to do them as long as it just takes a UMIK...

Probably won't affect much, but standard ETC, RT60 or variants, and FR measurements will show whatever there is to see.

You can also point the microphone at the diffuser vs. a flat wall to measure the reflected wave (using gating), at different angles.

There is also this room texture analyzer program: https://alejandrobidondo.wordpress.com/

 

[KSTR]

Does anyone know exactly what type of measurement will tend to best reveal any differences from a diffuser?

I'd try to provoke a dominant strong flutter echo and look at the impact of the diffusor, both in the time domain (step/impulse response) and corresponding (windowed/gated) frequency response.
I would expect a large enough diffusor to really even out and spread the strong specular reflection, within a limited bandwidth at least.

The repeated reflection (without diffusor) should be very strong and long-lasting compared to other reflections/modes which should make comparison easier, notably when looking at tail of the flutter echo.

Set up would be something like this:

Panel centered on front wall (short wall).
A small speaker pretty close (0.5m), a bit below 1/2 room height
Mic flush on the rear wall, at bit above 1/2 wall height so that the speaker cabinet is a bit out of the way for the first reflection path.

 

[kemmler3D]

Print this instead:

View attachment 374316

That's a JPG, john, it's not that kind of printer. It takes 3D files.

This entire project is an excuse to buy a better 3D printer isn't it??

"no"

How much do you save compared to a product like this?

Diffusore Quadratico Gotham N23 5" (14cm) - GIK ACOUSTICS ITALY

Lo strabiliante diffusore quadratico skyline Gotham N23 5" offre ineguagliate perfomances nel controllare le riflessioni mantenendo allo stesso tempo presenza e vitalità nelle alte frequenze facendo suonare una stanza come se fosse più grande.

gikacoustics.it

The GIK diffuser is 17.5″ x 17.5″ x 5.5″ or 445mm x 445mm x 140mm roughly, and costs $199 US.

The diffuser I am printing is about 1300mm x 1300mm x 250mm and will cost me maybe $400 in materials including paint and whatever.

If we ignore performance/design frequencies and only consider cost per volume (not just area) of diffuser, ( 27,723 cc vs. 422,500 cc) I am saving about $2600. So if you look at it like this, the project pays for itself many times over.

If you only consider area then the savings is only $1300 or so. But still, not bad.

Not many places sell large diffusers off the shelf. I found an N23 at 35" x 35" for $475 (actually not bad) but it's not deep and the wells are wide, so the low and high frequency performance will be limited: https://www.etsy.com/listing/613710045/huge-35-diy-acoustic-n23-skyline-sound

If you want to get something a little smaller than what I am building with a N67 layout, it will cost you $4800 and does not ship to the US. https://www.etsy.com/listing/1105506216/handmade-wood-acoustic-wall-art-sound?variation0=4189457103

If you want something a little bigger than what I am building, it will cost you a cool $12,000 https://rdacoustic.cz/en/high-end-audio/acoustic-diffuser-qrd67/.

All in all I don't think it's hard to argue this project is cost-effective, anyway.

I'd try to provoke a dominant strong flutter echo and look at the impact of the diffusor, both in the time domain (step/impulse response) and corresponding (windowed/gated) frequency response.
I would expect a large enough diffusor to really even out and spread the strong specular reflection, within a limited bandwidth at least.

The repeated reflection (without diffusor) should be very strong and long-lasting compared to other reflections/modes which should make comparison easier, notably when looking at tail of the flutter echo.

Set up would be something like this:
View attachment 374351
Panel centered on front wall (short wall).
A small speaker pretty close (0.5m), a bit below 1/2 room height
Mic flush on the rear wall, at bit above 1/2 wall height so that the speaker cabinet is a bit out of the way for the first reflection path.

This tactic seems like it could reveal a difference, I may have too much crap in my room to produce a really obvious flutter echo, but it's definitely worth a shot.

 

[kemmler3D]

In your own room you can measure the impulse response before and after and try and analyse it, you might see some changes as diffusers are also absorbers. It's not so easy in room to measure the relative phase differences at high frequencies which is why it can "sound" more different than the measurements might suggest, for good or bad depending on where you put the panel.

Indeed, if you showed me a waterfall or impulse response and asked me to say how much diffusion was in the room, it would be very hard unless the room was extremely diffuse or extremely reflective.

Gotta be the quickest assembly method, still very light.

Gluing up 2800 pieces can't be that quick, can it? If measuring, cutting, and gluing each piece took 1 minute, it would take a full work week. I agree foam would be quick / easy compared to wood, but probably not compared to printing. Plus it's all hands-on time, while most of the 3D printing is hands-off time.

Curious how this works out. If you make the shape too hollow without enough infill, it will sound like a cowbell when struck.

I've been really mulling over the infill and shells... I'm going to print some test pieces of a few squares at realistic scale to get a sense of whether there's a likely issue of internal resonances. However, the time and filament per piece is already pretty high, some of them are >24h and >1kg already. I'm hoping the angled orientation of the infill walls prevents that kind of thing.

 

[kemmler3D]

So, finally some actual audio content! This is an audio forum after all.

I printed 3 small, 50% scale sections of the diffuser with different print settings.

One with heavier (8%) gyroid infill (large) 2 outer layers, and 3 top and bottom layers. One with light gyroid infill (2%), 2 outer layers, and 2 top and bottom layers, and one (small) with light cubic infill (hardly any actually came out), and only 1 outer layer, but still 2 top/bottom.

Note: I used the worst filament I have on hand, some random PLA+ that had been sitting without any kind of moisture protection for months. Even with really challenging slicing settings (1 shell! basically no infill!) the surface quality on all 3 was quite decent. All 3 of these printed in under 2 hours. Looks better IRL. I am truly impressed with this printer so far.

I then did a highly scientific resonance analysis by holding them next to my UMIK and hitting them with a hex wrench while recording the noise.

Here's the overview:

I hit them on all sides a few times. It's not great news in terms of resonances, but at a glance, it looks like the "medium" version has the highest resonance frequencies, and the highest minimum resonance frequency, although I guess they all share some due to having identical exterior dimensions. I *think* this means that's my best bet. (higher frequency resonances tend to get excited less because of the natural power spectrum of music.)

The internal structure seems to matter somehow, and actually the full-size print will have the same infill pattern and size... but it will also have lower resonances because each piece will be twice as large on a side. This means I can maybe expect some resonances down in the 750hz range, which is not ideal... but I am not sure changing anything about the internal structure will help with that. The lowest frequencies are probably coming from the largest structures, which are of course the outer faces.

Surprisingly, the most robust, "large" print settings doesn't have an obvious advantage over the much lighter "medium" print setting. Which is good news because I didn't buy enough filament to do that one.

If anyone is interested enough to do more analysis on this, I've attached the WAV files here.

Also, if anyone happens to have a block of wood of similar dimensions (thin part 12.5mm, thick part ~37mm) and wants to compare by hitting it with an allen wrench, that would be really interesting. I do not happen to have one handy.

 

[Curvature]

So, finally some actual audio content! This is an audio forum after all.

I printed 3 small, 50% scale sections of the diffuser with different print settings.

One with heavier (8%) gyroid infill (large) 2 outer layers, and 3 top and bottom layers. One with light gyroid infill (2%), 2 outer layers, and 2 top and bottom layers, and one (small) with light cubic infill (hardly any actually came out), and only 1 outer layer, but still 2 top/bottom.

Note: I used the worst filament I have on hand, some random PLA+ that had been sitting without any kind of moisture protection for months. Even with really challenging slicing settings (1 shell! basically no infill!) the surface quality on all 3 was quite decent. All 3 of these printed in under 2 hours. Looks better IRL. I am truly impressed with this printer so far.

I then did a highly scientific resonance analysis by holding them next to my UMIK and hitting them with a hex wrench while recording the noise.

Here's the overview:

View attachment 374717
I hit them on all sides a few times. It's not great news in terms of resonances, but at a glance, it looks like the "medium" version has the highest resonance frequencies, and the highest minimum resonance frequency, although I guess they all share some due to having identical exterior dimensions. I *think* this means that's my best bet. (higher frequency resonances tend to get excited less because of the natural power spectrum of music.)

The internal structure seems to matter somehow, and actually the full-size print will have the same infill pattern and size... but it will also have lower resonances because each piece will be twice as large on a side. This means I can maybe expect some resonances down in the 750hz range, which is not ideal... but I am not sure changing anything about the internal structure will help with that. The lowest frequencies are probably coming from the largest structures, which are of course the outer faces.

Surprisingly, the most robust, "large" print settings doesn't have an obvious advantage over the much lighter "medium" print setting. Which is good news because I didn't buy enough filament to do that one.

If anyone is interested enough to do more analysis on this, I've attached the WAV files here.

Also, if anyone happens to have a block of wood of similar dimensions (thin part 12.5mm, thick part ~37mm) and wants to compare by hitting it with an allen wrench, that would be really interesting. I do not happen to have one handy.

Diffusers have two underappreciated aspects: the spectrum of the reflected wave is uneven and they act as absorbers.

I suggest that the resonant behavior will contribute to both.

 

[kemmler3D]

Diffusers have two underappreciated aspects: the spectrum of the reflected wave is uneven and they act as absorbers.

I suggest that the resonant behavior will contribute to both.

Well, I know the spectrum is uneven at a certain distance (hopefully I am sitting far enough) and there is some absorption, although hopefully by keeping the wells at 25mm width it won't be severe. And I suppose the resonances will contribute to all of that, sure.

Is this a lot or a little resonance compared to wood or other materials? I don't think I can answer that by only testing pieces of plastic.

Acoustic diffusers (not usually QRDs, but others) are often made from a single layer of hollow plastic from injection molding / vacuum forming. This suggests to me that maybe resonances within the structures aren't a big concern in general.

Anyway, I think I have enough information to at least choose an option among those available, I don't know if this test could really tell us anything about the suitability of 3D printing either way.

 

[Curvature]

Acoustic diffusers (not usually QRDs, but others) are often made from a single layer of hollow plastic from injection molding / vacuum forming. This suggests to me that maybe resonances within the structures aren't a big concern in general.

Resonances are important (think Helmholtz). The manufacturing process, market and ideal purpose of treatments are better established than the science. I recommend reading Trevor Cox, who someone else in this thread also mentioned, for details on the complications of treatment construction. The problem as such is that the reflected wave is different and inconsistent across the spectrum in when looking at the normal measurements: scattering, diffusion, absorption. The effect of this inconsistency overall, in terms of what we hear, is unclear because there are few psychoacoustic studies delving into this kind of detail. Cox mentions none in his book. Acoustic treatment construction in general aims at certain physical targets, and the only reason to study treatment from an engineering perspective is that inconsistency of the reflected waves makes those targets more difficult to meet.

We only recently have good 3D measurements of the reflected wave: https://nwaalabs.ipower.com/Acoustics First Page 1.html They aren't pretty. We certainly would not accept, with the data we have now, speakers with radiation patterns this uneven:

For comparison, the Genelec 8351B at the same frequency: https://www.genelec.com/8351b#section-downloads

You're in a kind of no-man's land when it comes to what matters and what doesn't. If I were interested in getting answers about the viability of 3D printing and its effects, I would look at metamaterial studies. All the really new and exciting stuff about acoustic treatment is coming out of that field as far as I know.

 

[Curvature]

For comparison, the horizontal maps of the same diffuser and speaker above:

 

[kemmler3D]

I would look at metamaterial studies. All the really new and exciting stuff about acoustic treatment is coming out of that field as far as I know.

Indeed, I've read some of those papers (Zhu) and the results are exciting, even if the math is a steep climb for me (at best). However, even though I've asked smarter people than myself (one of them on this forum, although haven't seen him recently) there's no way I've found as a non-scientist non-engineer to design such a panel myself. I've waited almost 2 years for something I could actually use to come along in that space, no dice, so I'm back to 1970s tech.

If you know how, let's forget this QRD crap and design some really effective diffsorbers to print up. I'd be happy to print a set and mail them to you if you (or anyone) could handle the design.

Otherwise this a QRD is my 1st or 2nd most viable option (and again, I do think they look cool, so the decorative factor is tipping the scales a bit too.)

Believe me, I'd much rather be printing a 100mm panel that absorbs / diffuses evenly down to 80hz or something, but at the moment nobody is sharing designs or calculators for those.

As for the unevenness of the reflection - from every simulation I've seen it's less bad for QRDs than it is for many types of diffuser.

The globe plot you posted is for this:

Whatever this is, I am not sure I'd expect good results from it, certainly there's reason to believe a real QRD will do better. And if you want to be generous (based on the horizontals) it looks like it spreads out the reflection quite well in the critical midrange, so I might still take it over a bare wall.

If it sounds like garbage in my room... well, I'll sell the QRD for $1K and the printer will have paid for itself.

At some point after this is done, I will probably start a second project to print up some Binary Amplitude Diffuser facings for rockwool panels (should be way easier to fabricate) and put them on the ceiling. From what I've read online, the theory on that is also a little shaky, but the anecdotes are almost universally good.

 

[Curvature]

Indeed, I've read some of those papers (Zhu) and the results are exciting, even if the math is a steep climb for me (at best). However, even though I've asked smarter people than myself (one of them on this forum, although haven't seen him recently) there's no way I've found as a non-scientist non-engineer to design such a panel myself. I've waited almost 2 years for something I could actually use to come along in that space, no dice, so I'm back to 1970s tech.

If you know how, let's forget this QRD crap and design some really effective diffsorbers to print up. I'd be happy to print a set and mail them to you if you (or anyone) could handle the design.

Otherwise this a QRD is my 1st or 2nd most viable option (and again, I do think they look cool, so the decorative factor is tipping the scales a bit too.)

Believe me, I'd much rather be printing a 100mm panel that absorbs / diffuses evenly down to 80hz or something, but at the moment nobody is sharing designs or calculators for those.

As for the unevenness of the reflection - from every simulation I've seen it's less bad for QRDs than it is for many types of diffuser.

The globe plot you posted is for this:

Whatever this is, I am not sure I'd expect good results from it, certainly there's reason to believe a real QRD will do better. And if you want to be generous (based on the horizontals) it looks like it spreads out the reflection quite well in the critical midrange, so I might still take it over a bare wall.

If it sounds like garbage in my room... well, I'll sell the QRD for $1K and the printer will have paid for itself.

At some point after this is done, I will probably start a second project to print up some Binary Amplitude Diffuser facings for rockwool panels (should be way easier to fabricate) and put them on the ceiling. From what I've read online, the theory on that is also a little shaky, but the anecdotes are almost universally good.

Cox includes limited 3D plots of QRD and other diffusers in his book. They are not much better IIRC.

It really is a question of goals. I would rather preserve the spectral balance of reflections instead of breaking them up semi-unpredictably through phase grating. If there is use to these kinds of treatments I would think it would solely for late, low-energy reflections. That at least is consistent with what I know of hearing and how I interpret speaker radiation behavior. On the other hand, I may be wrong because we may be much more tolerant of of spectral inconsistency for indirect sound. I don't know of research on this and my own experience hasn't allowed meaningful testing.

I don't mean to discourage you at all. It is an interesting project regardless.

 

[kemmler3D]

I'm up against the same problem we all are - we have to make use of the space we have and the spaces we have are rarely even within shouting distance of ideal.

It really is a question of goals. I would rather preserve the spectral balance of reflections instead of breaking them up semi-unpredictably through phase grating.

I'm starting with the assumption that the reflections are too specular/regular and I'm getting unwanted cancellations/reinforcement at my listening position, (that's how it sounds to me, anyway) so breaking it up randomly even at the cost of a little deviation in overall response, seems like a way to help with that. It's not like I have a wonderful room response now and this is just going to be a turd in a punchbowl.

The QRD is going to go almost directly behind me, which is currently just bare wall and a metal filing cabinet.

Plus it will make for a good zoom background on my work calls.

Anyway, I don't think your points are wrong, I just don't think I have much better options for diffusion.

FWIW I did design this thing to get the widest bandwidth of "proper diffusion" I could in the space I have.

 

[dualazmak]

Interesting thread, indeed!

Even not a large diffuser/disperser for wide Fq coverage, I am currently testing hard-and-heavy random-surface Bohemian crystal-cut-glass salad bowl for wide-3D reflective dispersion of metal-horn super-tweeter sound (narrow directivity, mainly 8.8 kHz to 25 kHz) (ref. here and posts thereafter) in my SP setup (so far so nice, subjectively) where I have nice sound deadening space (actually rooms) between the L&R SPs and also behind my head sitting at the listening position (ref. here).

Even though my tests/experiments with the "hard-and-heavy random-surface Bohemian crystal-cut-glass salad bowl" are just very primitive fun/play subjective listening evaluations, I wonder and hope if someone could do some theoretical simulation/calculation for the 3D reflective dispersion (no absorption, I hope) using random surface hard-heavy material for super-tweeter high Fq sound.

 

[fluid]

At some point after this is done, I will probably start a second project to print up some Binary Amplitude Diffuser facings for rockwool panels (should be way easier to fabricate) and put them on the ceiling. From what I've read online, the theory on that is also a little shaky, but the anecdotes are almost universally good.

If you haven't already seen it this video is full of good information but is very long, not everyone has 4 hours to watch youtube. At about 2:08:00 There is a section on BAD panels and how they work. Interesting that the best function is when the panel is made of stainless steel so the hole edges are sharpest.

 

[kemmler3D]

Interesting that the best function is when the panel is made of stainless steel so the hole edges are sharpest.

That IS interesting. Somehow it depends on diffraction, then? I thought it was just something to do with reflection vs. absorption, but to be honest I don't find QRDs terribly intuitive and BADs are less so.

I almost never sit through long YT videos so thanks for the tip on what part to watch

 

[fluid]

That IS interesting. Somehow it depends on diffraction, then? I thought it was just something to do with reflection vs. absorption, but to be honest I don't find QRDs terribly intuitive and BADs are less so.

I almost never sit through long YT videos so thanks for the tip on what part to watch

A new term

 

[theREALdotnet]

I almost never sit through long YT videos so thanks for the tip on what part to watch

I think this is one of the few long ones I actually watched in its entirety. They do go off on tangents and chew the fat on all things audio, but there are many interesting nuggets of insight and information in there. One point Sauro made, that I found particularly fascinating, was that in-room measurements show that velocity absorbers work far deeper down than they have any right to, because there are several other effects at work (besides friction between air molecules and fibres).