QRD Prime + Period Question (Live Room)

[terry j]

looks wonderful xenon, well done indeed.

I doubt normal styrofoam is any 'different' than what you used, liquid nails etc does not seem to affect normal styrofoam. Funnily enough, I found white glue to not adhere that well to it, maybe I didn't leave it long enough?

[Xenon]

Hi Terry,

Thanks for your comments and for your help.
We hope to apply stain this coming weekend, to create a better look. And I hope to take some better shots, rather than in a dark garage.
I think there may be different white glue formulations? We use a white glue labeled "wood glue" (in Japanese). It smells like the white glue I'm familiar with -- though when dry it seems to create a stronger bond. We are using a lot. The faces lifted at the ends at first, so we put an extra dab under each end. Though we smooth the dab flat it remains a thick smear. On drying, we put spacers over the faces to keep them flush, and lately also run some bungee cord around the frame (can be seen in some of the new pics added today, here.)

After about 18 hours the entire unit is rigid and strong. As to 'normal styrofoam' there is a mixup on the brand name. XPS/EPS is styrofoam. There are beaded and extruded types. "Expanded polystyrene beads" (like a coffee cup) is the classic white styrofoam, which crunches when cut with a knife. This stuff in my experience doesn't white glue as well as "extruded" types like XPS/EPS. Check the wiki "styrofoam".

I think there is an advantage in bonding two flat surfaces with a good surface area/weight ratio. Liquid nails, what type of adhesive is it?

Originally Posted by terry j

looks wonderful xenon, well done indeed. I doubt normal styrofoam is any 'different' than what you used, liquid nails etc does not seem to affect normal styrofoam. Funnily enough, I found white glue to not adhere that well to it, maybe I didn't leave it long enough?

[terry j]

yes, I was referring to the coffee cup type. We call the glue PVA.

liquid nails, you mentioned it yourself. It is construction adhesive.

BTW, I recently found clear construction adhesive (liquid nails) instead of the normal brown stuff.

just a heads up for others..kinda looks like silicone actually, but assume it is more of an adhesive than a sealant.

So maybe just the ticket to help avoid nasty looking gluing errors.

Still, back to the topic, I must say yours are one of the nicest looking I've seen done.

Looking forward to your impressions.

[Schaap]

Great thread!Learned a lot more about the theory, thanks guys.
I've build N7's and to get the well in the right depth between the dividers was the most difficult thing to do. Xenon, great tip to use the styrofoam!

Henk

[Xenon]

Hey,
thanks for your comments.

Terry, wish I could get some more gluing options. Whatever glue is used, it's very difficult to avoid getting some on a fin now and then. So it would be better to stain the materials first -- white glue is pretty innocuous. Not sure how the construction adhesives/liquid nails clean up, but it's a concern.

I'm completely happy with the while glue, now that I see how to use it. We've gone through 6-7 Kg of it, building 8 N23 units.

Henk, yes definitely the XPS styrofoam makes the job a lot easier and the finished result is to spec I think.

I still have a question about "viscous losses" in the wells. Our well width of 40.6mm (actual) is slightly narrow, by this measure. But wouldn't this only affect the two deepest wells? And how much 'flow' is really gained by a 0.8cm wider well (which would put us in the 'OK' range)? I would like to build as narrow as well as possible, to gain bandwidth... I want to return to the RPG QRD-4311, which has a max well depth of @ 40cm (16") with a well width of 2.8cm (1.1"). Wouldn't there be extreme viscous losses, and/or would this have been deemed acceptable? The designers write (1992):

"Physical manufacturing constraints pose a space limitation of approximately 1 in (25.4 mm) on the well width and a depth limitation of approximately 16 in (406.4 mm), after which the units become diaphragmetic and defeat the purpose of increasing the depth further."

The only limit mentioned: "diaphragmetic". Until the philosophy behind QRD-4311 is explained, I question the actual effect of viscous losses. Have experimental measurements been done with actual units which prove this effect? How significant are such losses, what is the problem acoustically and at what point is it seriously going to affect tracking in a smaller room?

Here is quite a wall (click on the image above "La Diffusione"") with "QRD-4311 (120X60X40cm) N43, 210-3.5KHz, embedded in a wall of QRD-734s" and etc. (Input Level Studios, Treviso, Italy).

As diffusor HF bandwidth is increased by narrower wells, with shallower units (like mine at 13.5cm max), bass bandwidth is sacrificed yet HF can be gained with a narrower well. Considering the time and trouble to build, bandwidth must be a major design goal. Can anyone shed more light on this topic?

Wow, searching around I found this page on "flipping" diffusors to create the Barker code. This is from RPG Europe. It seems from the illustration that the 'flipped' unit is simply upside-down/reversed, not an inverse period. There are two illustrations at the bottom of this page titled "Modffusor -- QRD Evolved". ??

[collo]

I don't know that the document you linked to suggested that you could have a 16 inch deepest well AND a 1.1 inch wide well at the same time.
They seemed to be independent limits.

The only recommendation I've seen for width / depth ratio is the original one proposed by Shroeder, as implimented in the other 2 calculators on the net. With QRDude, this is modified slightly, allowing the width to track the deepest well rather than being based on the design frequency, only when the inverse panel is not going to be used. With the N43, the deepest well is almost the full design depth, so the saving is quite small.

I couldn't find detailed specs for the QRD-4311 anywhere.

Modelling an N43 to use 1.1 inch wide wells, (fins 1/8 inch), gives a design frequency of 1600hz. This may be what was built.

Modelling an N43 to use a 16 inch deepest well gives a recommended well width of 4 3/4 inches. If you built this with 1.1 inch wide wells, you would be a very long way from Shroeder's recommendation.

[terry j]

Originally Posted by Xenon

Hey,
thanks for your comments.

Terry, wish I could get some more gluing options. Whatever glue is used, it's very difficult to avoid getting some on a fin now and then. So it would be better to stain the materials first -- white glue is pretty innocuous. Not sure how the construction adhesives/liquid nails clean up, but it's a concern.

Glad you mentioned this, I forgot to comment in my last post.

I had a bit of a shock when you said you were going to stain it after assembly. Been there, done that.

cause after staining comes finishing too.

Made that mistake myself, so 'never again' is my catchcry.

Re the glue, I ws only responding to some comments you had made, was not particularly recommending anything.

Yikes, 7kg of it!!! At least PVA is cheap.

I still have a question about "viscous losses" in the wells. Our well width of 40.6mm (actual) is slightly narrow, by this measure. But wouldn't this only affect the two deepest wells? And how much 'flow' is really gained by a 0.8cm wider well (which would put us in the 'OK' range)? I would like to build as narrow as well as possible, to gain bandwidth... I want to return to the RPG QRD-4311, which has a max well depth of @ 40cm (16") with a well width of 2.8cm (1.1"). Wouldn't there be extreme viscous losses, and/or would this have been deemed acceptable? The designers write (1992):

"Physical manufacturing constraints pose a space limitation of approximately 1 in (25.4 mm) on the well width and a depth limitation of approximately 16 in (406.4 mm), after which the units become diaphragmetic and defeat the purpose of increasing the depth further."

The only limit mentioned: "diaphragmetic". Until the philosophy behind QRD-4311 is explained, I question the actual effect of viscous losses. Have experimental measurements been done with actual units which prove this effect? How significant are such losses, what is the problem acoustically and at what point is it seriously going to affect tracking in a smaller room?

I too noted the description in alton eversts book of the width of those diffusors and the corresponding depth. Mentioned it in my thread actually.

So I kinda agree with you, maybe at times we do get bogged down in the minutiae.

Here is quite a wall

looks kinda a strange way to go about it! A less wide diffusor sitting in front of a bank of 'deeper' diffusors. Heck, just design one to do it all!

like a di or tri fractal haha.

Wow, searching around I found this page on "flipping" diffusors to create the Barker code. This is from RPG Europe. It seems from the illustration that the 'flipped' unit is simply upside-down/reversed, not an inverse period. There are two illustrations at the bottom of this page titled "Modffusor -- QRD Evolved". ??

Good find. Very much backs up Jasons comments on how (or at least one way) to implement the barker code.

That is enough to 'settle' it for me. I accept that you may get even better performance by using a true inverse panel, but that is as close as I could expect as coming from the horses mouth, ie RPG.

Originally Posted by collo

I don't know that the document you linked to suggested that you could have a 16 inch deepest well AND a 1.1 inch wide well at the same time.

Am pretty certain almost that exact wording comes out of alton everest's book. And, as RPG seem to be the consultant in that book on matters diffusion, again a pretty reliable source on the face of it.

[Xenon]

Collo,
you wrote:

Originally Posted by collo

I don't know that the document you linked to suggested that you could have a 16 inch deepest well AND a 1.1 inch wide well at the same time. They seemed to be independent limits.

I thought so too. Yet Alton Everest includes a photo of 2x vertical RPG QRD-4311 + horizontal QRD-1911 array in editions (incl. the latest) of his "Master Handbook of Acoustics." See page 266. Here is the googlebook result. -- he writes "This particular quadratic-residue cluster provides excellent diffusion in the horizontal hemidisc... (p. 265)"

Of course I am struck by the ratio of well width to depth, which is wildly beyond what the calculators recommend. I'm not sure there's more detail to be had for the QRD 4311, but we have the well width and depth and N43 period. Here is a picture. And another of it installed. Appears to be made of wood.

Originally Posted by collo

The only recommendation I've seen for width / depth ratio is the original one proposed by Shroeder, as implemented in the other 2 calculators on the net. With QRDude, this is modified slightly, allowing the width to track the deepest well rather than being based on the design frequency, only when the inverse panel is not going to be used. With the N43, the deepest well is almost the full design depth, so the saving is quite small.

Yes, and RPG did not follow that. They have extensive measurement facilities.

Originally Posted by collo

Modelling an N43 to use 1.1 inch wide wells, (fins 1/8 inch), gives a design frequency of 1600hz. This may be what was built. Modelling an N43 to use a 16 inch deepest well gives a recommended well width of 4 3/4 inches. If you built this with 1.1 inch wide wells, you would be a very long way from Shroeder's recommendation.

Yup. And this strong divergence is what makes the topic intriguing. It's not just any example -- it's considered representative of "best in class" so I hope someone somewhere could shed more light on its theory and function. Or at least, this example may give pause.

[collo]

Originally Posted by terry j

Good find. Very much backs up Jasons comments on how (or at least one way) to implement the barker code.

That is enough to 'settle' it for me. I accept that you may get even better performance by using a true inverse panel, but that is as close as I could expect as coming from the horses mouth, ie RPG.

Comparing apples with oranges I'm afraid.

The modfuser panels being referred to are assymetrical, so of course, rotating them 180 degrees will give a different diffusion pattern.

Standard QRD panels are basically symetrical about the midpoint (apart from one end well) so applying the same trick doesn't give near the result that the modfuser does.

[collo]

Originally Posted by Xenon

I thought so too. Yet Alton Everest includes a photo of 2x vertical RPG QRD-4311 + horizontal QRD-1911 array in editions (incl. the latest) of his "Master Handbook of Acoustics." See page 266. Here is the googlebook result. -- he writes "This particular quadratic-residue cluster provides excellent diffusion in the horizontal hemidisc... (p. 265)"

Of course I am struck by the ratio of well width to depth, which is wildly beyond what the calculators recommend. I'm not sure there's more detail to be had for the QRD 4311, but we have the well width and depth and N43 period. Here is a picture. And another of it installed. Appears to be made of wood.

None of those three references give the depth of the N43 - all the photos are front-on.

[terry j]

Originally Posted by collo

None of those three references give the depth of the N43 - all the photos are front-on.

page 303 of my handbook says

"The uniformity of spatial diffusion is determined by the length of the period. good broad-bandwidth, wide angle diffusion, then, requires a large period with a large number of deep narrow wells. This is why the QRD 4311 has 43 wells of only 1.1 inches width and a maximum well depth of 16 inches"

Originally Posted by collo

Comparing apples with oranges I'm afraid.

The modfuser panels being referred to are assymetrical, so of course, rotating them 180 degrees will give a different diffusion pattern.

Standard QRD panels are basically symetrical about the midpoint (apart from one end well) so applying the same trick doesn't give near the result that the modfuser does.

Yeah, a bit confused actually now that you mention it!

I did recall symmetric and asymmetric being discussed on the soundscapes site, (must be my lucky day!) I managed to find it without too much trouble.

Translation result for http://www.soundscapes.nu/rb_dif.htm

About half way down.

In any case, I counted up the wells in the modfusor in the link, blow me down it's based on a prime six!!! hahaha. (based on the drawing at the bottom of the page, yet in the depiction above we are back to seven...plus half a well width end joint making a prime eight!!! maybe a bit of misdirection to stop sneaky buggers like us???)

AND, do a search of the rpg site with modfusor, no returns.

Ahh, type in modffusor (double f) and you get some stuff. will have a look now.

[collo]

Originally Posted by terry j

page 303 of my handbook says

"The uniformity of spatial diffusion is determined by the length of the period. good broad-bandwidth, wide angle diffusion, then, requires a large period with a large number of deep narrow wells. This is why the QRD 4311 has 43 wells of only 1.1 inches width and a maximum well depth of 16 inches"

That is certainly a long way from Shroeder's thinking. It would be nice to get some more info on it! (which handbook are you referring to?)



The symmetry that soundscapes is talking about is a standard N7 panel, which in their context is slightly asymmetrical - (by one well) vs one built using the half-end well technique (symmetrical)....like this pair:

[Xenon]

Hi Collo,

Thank for pointing out the differences between a (near) symmetrical versus asymmetrical array, in terms of the 'reversing' (versus inverse) trick. It doesn't help that the word "flip" is used for two completely different design concepts!

"which handbook are you referring to?"
It's in the "Master Handbook of Acoustics" (4th ed. 2001) Chap. 14, "The Schroeder Diffusor" by Alton Everest (mentioned above). Page 303:

"The uniformity of spatial diffusion is determined by the length of the period. Good broad bandwidth, wide-angle diffusion, then, requires a large period with a large number of deep, narrow wells. This is why the QRD®-4311 has 43 wells of only 1.1 inches width and a maximum well depth of 16 inches."

The RPG QRD-4311 deserves a closer look and explanation. If only we could pick the authors' brains. On another note, having built my small N23, that N43 at 16" D (total frame depth?) and wells of 1.1" (only 2.8cm) must be near a limit for manufacture. Quite an amazing feat, and certainly that was not an inexpensive item.

If the QRD-4311 actually works well, my 4cm well width (at a max well depth of 13.5cm) should be no problem at all? To be or not to be...

[terry j]

re the symmetric etc, think we are talking very different beasts between qrds and modffusors!

So, in their case, asymmetric would reaally be asymmetric!

bugger, can't cut and paste, so have to type

(re the modffusor)..and relates to the 'prime' six and eight I referred to before

"The asymmetric modffusor contains 7 full width wells and two zero depth half width end wells. The modffusor is no longer bound to contain a prime number of wells as in the number theoretic QRD which contains 6 full width wells and two zero depth, half width end wells"

Ie, the diagram WAS correct with eight wells.

[Xenon]

Terry, Collo,

Doesn't this just mean that they split the zero well between the left and right sides? QRDUDE Calc does the same thing, if you want. So it only appears to have an even number of wells. (PS I'm building an N19 with a split zero well this weekend, it would look like 18 full wells + the last 2 are half width, and include the frame member = 20.) Unless I'm missing something there.

[terry j]

Originally Posted by Xenon

Terry, Collo,

Doesn't this just mean that they split the zero well between the left and right sides?

Not from what I see, have a look and count in the links you made earlier. Yeah there is a half width on each end, but taking that into account there is an even number of wells.

Can we have a prime that is even?? (hahah, of course, two, silly little joke there.)

that was why it 'jumped' out at me when I quoted that little bit earleir,

forgot to give a link

http://www.rpginc.com/commercial/pdf...Comparison.pdf

clearly states 'no longer bound by prime numbers'

[collo]

From Terry's post, the N7 is as you described, ie 6 full wells plus two half-width = 1 more, for a total of 7 wells.
The modfuser has 8 though...

I've been looking for more info on the QRD-4311 to no avail.

I did come across the dimensions for the QRD-734 at RPG.
From the mounting details, the deepest well is 9 inches (assuming a 1/4" backboard). The panel is 2ft wide, meaning the each well+fin combo is 3.4 inches (as per the 734 name). Assume 1/8" for the fins and we get 3.3 inch wells.
The depth / width ratio is within the Shroeder recommendation.
It does fall a bit short on the period width> design wavelength, but not by too much. This is easy to do with the N7.

They may have pushed the envelope with the 4311, but they stayed within the guidelines for this one.

[Xenon]

Terry, you wrote:

Originally Posted by terry j

I had a bit of a shock when you said you were going to stain it after assembly. Been there, done that. cause after staining comes finishing too. Made that mistake myself, so 'never again' is my catchcry.

For sure, if you have the room to spread out about 200 fins, 200 faces, 32 frame members, stain 5 times with spray, while flipping each part, then drying without too much dust, and storing it all (and we do have rain here in the tropics!) ... You see the problem? As a unit, we are compact, easily maneuverable, and portable. That is we avoid the nightmare of hundreds of separate parts. And then, what if they get scratched in handling, what if you need to trim something?

The one issue is that white glue occasionally gets on to a fin, and we're talking a little bit. What I really need is something (paint thinner doesn't work) to remove the glue drop and clean the wood. I'm just too lazy to research it! As we are using a dark stain with multiple coats I'll hope for the best.

Originally Posted by terry j

Re the glue, I ws only responding to some comments you had made, was not particularly recommending anything Yikes, 7kg of it!!! At least PVA is cheap.

Gotcha. But for sure, something like liquid nails in a caulk tube would be perfect for dabbing the top/center/bottom of faces, I think. Yeah 7kg, and we're just a bit over half done. The cost is about $5.50 USD for a 3kg tub. It will pour if you coax it (from the tub to the squeeze bottle). Well it's working out fine. Except for the frame members, which are nailgunned together and the 2.5mm back, staplegunned to supports, the entire unit is glued. Nary a nail or screw to be found. It's all about white glue and $0.30 3cm plastic spreaders.

Originally Posted by terry j

I too noted the description in alton eversts book of the width of those diffusors and the corresponding depth. Mentioned it in my thread actually. [saw that] So I kinda agree with you, maybe at times we do get bogged down in the minutiae. looks kinda a strange way to go about it! A less wide diffusor sitting in front of a bank of 'deeper' diffusors. Heck, just design one to do it all! like a di or tri fractal haha.

Yes, it's interesting to see the various creative solutions. That Italian studio looks to be in the multi-millions of $$$. A few years ago I was in on a mastering session at Capitol Records (in L.A.) with Robert Vosgien -- the room was fairly small, and acoustically perfection (imho). Treatments were complex wall curves and partial stone walls (you can see part of the room here) -- I've seen a number of stoneworks in older rooms (off topic, but on the topic of amazing rooms).

Originally Posted by terry j

Good find. Very much backs up Jasons comments on how (or at least one way) to implement the barker code. That is enough to 'settle' it for me. I accept that you may get even better performance by using a true inverse panel, but that is as close as I could expect as coming from the horses mouth, ie RPG. Am pretty certain almost that exact wording comes out of alton everest's book. And, as RPG seem to be the consultant in that book on matters diffusion, again a pretty reliable source on the face of it.

Though I think Collo's point is that with a near symmetrical period, merely flipping it upside-down isn't going to create the 'opposite' (-1 to the +1) in Barker, right?

[Xenon]

Originally Posted by terry j

Not from what I see, have a look and count in the links you made earlier. Yeah there is a half width on each end, but taking that into account there is an even number of wells.

Duh! I missed that...

Originally Posted by terry j

Can we have a prime that is even?? (hahah, of course, two, silly little joke there.) that was why it 'jumped' out at me when I quoted that little bit earleir, forgot to give a link http://www.rpginc.com/commercial/pdfs/QRD_Modffusor_Comparison.pdf clearly states 'no longer bound by prime numbers'

Thanks for the link. Yeah, I've been reading this and that from the googlebooks limited preview of Acoustic absorbers and diffusers: theory, design, and application By Trevor J. Cox, Peter D'Antonio (2004). There's quite a lot of discussion about optimization. In the PDF you sent: "RPG’s advanced patented QRD, which is called a Modffusor, because it’s design is based on a modulated, optimized diffusor as opposed to the older number theory approach..." Also a 14% decrease in depth, RPG writes. It's my understanding that optimization takes many forms, but involves complex math and also acoustic measurement of prototypes (as with certain informed well designs which have no fins). How does the Modffuser work with 8 wells? Neat.

[terry j]

Originally Posted by Xenon

Terry, you wrote: For sure, if you have the room to spread out about 200 fins, 200 faces, 32 frame members, stain 5 times with spray, while flipping each part, then drying without too much dust, and storing it all (and we do have rain here in the tropics!) ... You see the problem? As a unit, we are compact, easily maneuverable, and portable. That is we avoid the nightmare of hundreds of separate parts. And then, what if they get scratched in handling, what if you need to trim something?

I do understand that, but you rip (what I presume are the equivalent of 8*4) sheets down don't you?

Ie, you don't go out and buy the 40 odd mm strips of ply in bundles.

So, I am not talking about staining 200 fins and 200 faces etc individually, I would stain the 1200 by 600 sheet as a complete piece, even varnish it too, THEN rip it into individual strips.

No nightmare of hundreds of separate parts. Just one big sheet, which you are destined to handle in any case.

It is highly possible that the cut 'face' would need some sort of stain, it is then a simple matter to use the side of a brush dipped in the stain and gently wipe along the edges (usually only one is required).

I mean it's too late now, but when you are trying to get stain inside wells, on faces, not have runs, then do it all again with lacquer....however many times.

An alternative (which for all I know you are going to do) is spray. I have not tried it, but I can just imagine the spray droplets acting like the soundwaves...diffusing and bouncing back out of it in a cloud of mist!!!

Anyway, the die has been cast, onward and forward!

(off topic, but on the topic of amazing rooms). Though I think Collo's point is that with a near symmetrical period, merely flipping it upside-down isn't going to create the 'opposite' (-1 to the +1) in Barker, right?

hah, if anyone can take this thread off topic, surely it's you!!??

I must say, I fully and totally see collos point. I am not arguing against him at all. But it *seems* that the method outlined by jason is, somewhere, somehow, accepted and used as a valid implementation of the barker code.

Then again, I personally am very wary of internet lore, and in this case there does seem to be valid objections???

So, where does the truth lie.

Boy, I must be quite skeptical at heart, re the RPG 'claims'...well there is of course always the advancement of knowledge, no-one is denying that.

But equally, there is also defending one's turf, and there are many ways to do that job. (I feel for them, in the sense that here we are trying to back engineer all *their* work...)

It makes sense to 'advance' the technology from your competitors..

hmm, being cynical can also tie you in knots!!

Let's just sort what WE can eh? And that is forget about secret schroeder advancement voodoo, and just get the twenty year old technology correct first.

[Xenon]

Originally Posted by terry j

I do understand that, but you rip (what I presume are the equivalent of 8*4) sheets down don't you?

Well actually... All the framing and faces are cut at the Home Center. Our 2.5mm panels are 1830 (6') x 921 (3') -- this is the approx. ply sheet size in Japan. It's still too long to carry in our car (for the fins alone we bought 33 sheets of this stuff). So we get the sheets for fins cut in half. To be honest we have little interest in ripping a 6'x 3' 2.5mm sheets of ply on our saw. So we came back from the home Center with about 300 4cm faces, 66 half-panels for fins, and 48 frame members (of 12mm) ply. May as well assemble it all first! (The key to white glue removal involves damp rags.)

Originally Posted by terry j

Ie, you don't go out and buy the 40 odd mm strips of ply in bundles.

Au contraire, monsieur.

Originally Posted by terry j

So, I am not talking about staining 200 fins and 200 faces etc individually, I would stain the 1200 by 600 sheet as a complete piece, even varnish it too, THEN rip it into individual strips.

It makes great sense, if we could manage it that way. I will say though that there is a big issue in ripping 2.5mm plywood. Even with our 100-tooth 10" saw, you always get that natty-ragged-flaky edge, which we sometimes have to lightly sand (with 80 or 120 grit). So this would involve a lot of re-staining, actually for every piece, on both sides, and by hand (versus a sprayer). For such thin wood, pre-staining panels may not be so time-effective as a result.

Originally Posted by terry j

It is highly possible that the cut 'face' would need some sort of stain, it is then a simple matter to use the side of a brush dipped in the stain and gently wipe along the edges (usually only one is required).

Sure, could be done, but you are talking about 2-sided edges of hundreds of pieces, which also need to (re)dry.

Originally Posted by terry j

An alternative (which for all I know you are going to do) is spray. I have not tried it, but I can just imagine the spray droplets acting like the soundwaves...diffusing and bouncing back out of it in a cloud of mist!!!

Ah, you haven't done low-viscosity spraying then. When you get the pressure and droplet-size set (with test pieces) it's really a dream. When outside, in good weather, and wearing a respirator that is. Anyway, perfect for the wells -- like with shooting lacquer, you want a number of fast-drying light coats, without trying to cover every little spot with any one coat ... We have enough room outside to line up all 8 N23 units in a row--by the time you get to the end, you can re-shoot the first (in theory).

Originally Posted by terry j

hah, if anyone can take this thread off topic, surely it's you!!??

Thanks -- I mentioned the Capitol Studios Mastering Room 1, because I've thought a lot about it --

Originally Posted by terry j

I must say, I fully and totally see collos point. I am not arguing against him at all. But it *seems* that the method outlined by jason is, somewhere, somehow, accepted and used as a valid implementation of the barker code.

Really? It seems questionable when near-symmetry between the +1/-1 periods differs by only one well-position. What I wonder about is say, with N23 as an example, if you were to position +1 as a split (left-right) zero well, and the -1, with a zero well in the middle. In other words move the wells over so that there is a 'greatest visual difference' between the two panels (real scientific, huh).

Originally Posted by terry j

Then again, I personally am very wary of internet lore, and in this case there does seem to be valid objections??? So, where does the truth lie.

In measurements in a room... And it doesn't seem easy to get proper readings in rooms with other issues. It does bring up the question of which software and what particular tests would be most helpful. Do you or does anyone have an answer for this?

Originally Posted by terry j

Boy, I must be quite skeptical at heart, re the RPG 'claims'...well there is of course always the advancement of knowledge, no-one is denying that. But equally, there is also defending one's turf, and there are many ways to do that job. (I feel for them, in the sense that here we are trying to back engineer all *their* work...)

I'm in no position to judge. I do have immense respect for the RPG acousticians, and am grateful they post so much specific information online and in textbooks.

Originally Posted by terry j

It makes sense to 'advance' the technology from your competitors.. hmm, being cynical can also tie you in knots!!

Well I have a background in software design and simple electronics (building stompboxes, small amps). Many excellent amps and 'boxes are based on the same circuits (1930s design, with tube amps), and the same for ss circuits. With software, who doesn't use building-blocks of routines? These are in the public domain, and huge chunks of change are made by applying them, every day. Likewise, the math of Schroeder diffusors is free and 'open source,' and various people apply and use it. It's a gift of science to humanity, in this sense (Schroeder's in particular). And if someone can make a buck with my build design, good on 'em. But I hope they'll improve it and post the results here. So I'm all for figuring out all implementations, so I can improve my own design. And I wish I owned all my own genes, but I think this might not be true anymore.

Originally Posted by terry j

Let's just sort what WE can eh? And that is forget about secret schroeder advancement voodoo, and just get the twenty year old technology correct first.

Well topology and boundary mathematics is over my head, but if someone (hint hint) can calculate realworld examples, I'm all for it! It would be cool to make diffusors without wells, and you are working on the fractal insert which is very cool -- so there you go. Probably our end has to do with inexpensive, easy-to-build solutions. One QRD per child, I say.

[terry j]

Xenon, that reminded me, forgot to give my gag about the noble gasses that you obviously have an interest in.

Well I too, strangely enough, am fascinated by these noble gasses, these distinguished beasts that enrich our lives at so many times and at interesting moments.

Meh, but the wife, she doesn't share the same delight and pleasure in the noble gasses as I do, she just says 'stop farting'

...oh dear, now I typed it that's actually pretty lame!

Originally Posted by Xenon

Au contraire, monsieur.

Bloody hell!

Well, in that case, yeah as you say may as well assemble it!!

I will say though that there is a big issue in ripping 2.5mm plywood. Even with our 100-tooth 10" saw, you always get that natty-ragged-flaky edge, which we sometimes have to lightly sand (with 80 or 120 grit). So this would involve a lot of re-staining, actually for every piece, on both sides, and by hand (versus a sprayer). For such thin wood, pre-staining panels may not be so time-effective as a result. Sure, could be done, but you are talking about 2-sided edges of hundreds of pieces, which also need to (re)dry.

This is just 'FYI' for anyone who reads it, not debating or trying to say you're wrong etc. I do get what you're saying about the edge needing a sand, it does. (forgot to mention that bit)

Anyway, that is how I do, and have, done it. Taking a bit of care in setting the fence parallel really helps for a smooth cut. Yes a light rub (one or two sides) with a sanding block, very quickly chamfering the corner as you go, then a quick wipe with the side of the brush.

Making the space couplers that I have been doing involved a lot of that.

As I say, I am not right or wrong, neither are you, just that it does work and each can make their own mind up.

Admittedly I only used a brush, but when I made my bank of diffusors the painting finish took as long as the entire build! That boring (and horribly frustrating) experience was enough for me to say never again!

Ah, you haven't done low-viscosity spraying then. When you get the pressure and droplet-size set (with test pieces) it's really a dream.

This is interesting, but not being a spray painter I am not fully following. Is this similar to these electric stuff you get at hardware stores?? Or does it still require the full kit, compressor lines etc and a special gun.

As I am renovating a house I have the compressor and lines etc, but (depending on price I suppose) am not too keen I going out and buying a set of dedicated spray gear.

Also, as YOU may guess, I would like to keep my build simple and not requiring special stuff!!

Really? It seems questionable when near-symmetry between the +1/-1 periods differs by only one well-position. What I wonder about is say, with N23 as an example, if you were to position +1 as a split (left-right) zero well, and the -1, with a zero well in the middle. In other words move the wells over so that there is a 'greatest visual difference' between the two panels (real scientific, huh).

Well, that IS the $64 question innit.

See, here's the trouble. We ask, and get clarified, what Jason meant by his implementation of the barker code.

We still look at it (with the same 'doubt' as before), and can go no further! Sure, we can 'push' him again, but I don't think the answer will change (and am pretty sure he does not keep a close eye on this or my thread, nor does glenn)...so where do we go???

It seems such a huge hole or gap in the data base that there is no clear definitive answer...unless we HAVE been given it and we just don't accept it!

Aaarrgghh

It's engineering, you get to pick your own compromises.

I'm afraid I would look at a diffusor with a proper inverse panel and think 'what a waste of depth, 80% of the depth is wasted all because ONE panel had to go *twice* (or whatever) as deep as the rest'.

I'd rather choose slightly worse lobing and either save space, or go lower in diffusion across the entire panel.

Others could very well think the opposite.

In measurements in a room... And it doesn't seem easy to get proper readings in rooms with other issues.

I don't think in room will help. Have tried it (but of course I don't know what I am doing) but there was absolutely nothing to see.

On a side note, (maybe it was only me) but I really don't think we have any idea of just how fast these events happen in a room. You read what glen and ethan and the other experts say, and kinda think 'putting a panel on the wall ten feet behind me stops interference with the main wave?' etc etc.

Phrasing it poorly, but you know, it's true that in 'one second the wave has been reflected 500 times' or whatever it is, it IS a complete mishmash of reflections and interference...we just tend to think in our own time scale.

So, all that means is even using a sweep in a room, all the reflections etc completely obscure and mask any measurable result from a qrd.

(I will only find out if I am close to being right when I repeat the measurements outdoors, if I get the same results then I have built it completely wrong! or it was my measurement method that was stupid...time will tell)

It does bring up the question of which software and what particular tests would be most helpful. Do you or does anyone have an answer for this? I'm in no position to judge.

All I tried to do was put a source normal to the qrd, and move the mic in different positions in an arc around it. And similar with a flat panel.

I wanted to see obvious indications of reflections at angles different in one case than the other, start to build up the diffusion map we see often.

Need to try it outdoors at night, prob won't be for a few months tbh.

It would be cool to make diffusors without wells, and you are working on the fractal insert which is very cool -- so there you go. Probably our end has to do with inexpensive, easy-to-build solutions. One QRD per child, I say.

Well, the auralex qfusor does not have wells. It claimed to diffuse down to 800 hz with a depth of three inches.

That claim was met with a great deal of skepticism around here, which I took on board.
'
However, (as I was mucking about with 5 primes at the time, for the trifractal build) when I looked at a five prime, and saw how low the reported diffusion point was, AND imagined it without wells, suddenly the auralex claim did not seem so outlandish to me. (if you have a look at the qfusor, you can kinda see a mixture of seven and five prime in there, if you mentally remove the walls)

And in my thread, you will see lupos demonstration of a seven prime with and without fins...which went a long way (bearing in mind the limitations of the demonstration tool) in allaying fears about removing wells.

(made many 'examples of qfusors' in the ripple tank...all seemed to give very good diffusion too)

Maybe that brings into play the 'greatest good'? Normally it is the COST of these devices that prevent wider use, so if we had a much cheaper alternative like the slutzffusor (tm) without fins, that diffused well anyway, (and being cheaper means you can put more SA in)...well you are at least getting a good (if not perfect) diffusor in place rather than none at all.

All due to much cheaper cost of entry.

You know, demand 'perfectly mathematically rigorous diffusors or NONE at all' and a vanishingly small percentage use them, or allow 'less perfect diffusion yet lot's of them' and due to being cheaper, a much larger percentage get to own and use them.

gee, wandered around a bit myself!!!

[Xenon]

Terry,
Sounds like you've already built quite a number of arrays. Just a note on the sprayer. I was talking about via air compressor. I have two types of sprayers which attach to the air hose. One is often used in automotive work, it shoots lacquer and also polyurethane (and automotive enamels). The other is what I'll try first. I bought it as a kit (two complete spray heds and stainless bottle-holders, and a tire pressure gauge, etc.) for $48. It's designed for house painting among other things. I plan to thin the stain with paint thinner in any case. One of these sprayers will work I think.

Originally Posted by terry j

Xenon, that reminded me, forgot to give my gag about the noble gasses that you obviously have an interest in.

Yes, and anything that's rare, inert, invisible, odorless, and glows when electrically stimulated... Speaking of which, last year I heard a plasma tweeter at the RMAF ("Rocky Mountain Audio Fest," Denver, CO). I don't want one, but the negative ions are a nice benefit.

Originally Posted by terry j

Well, that IS the $64 question innit....so where do we go??? It seems such a huge hole or gap in the data base that there is no clear definitive answer...unless we HAVE been given it and we just don't accept it! ... It's engineering, you get to pick your own compromises. I'm afraid I would look at a diffusor with a proper inverse panel and think 'what a waste of depth, 80% of the depth is wasted all because ONE panel had to go *twice* (or whatever) as deep as the rest'. I'd rather choose slightly worse lobing and either save space, or go lower in diffusion across the entire panel. Others could very well think the opposite.

Can't address everything you wrote here i your last, but other than a labor of love on the part of those with time on their hands, and money, prototyping and testing might be carried out in universities and in private companies, and that data is hard to come by. I'm not in a position to properly test, but my needs are pragmatic. After equipping my studio, I'm moving on to the next thing, and looking forward to having people in the rooms and recording.

I am thinking to build some 2D Skylines out of XPS, for the ceilings...

Love that QRDUDE, and all the talk and information! Sorry to be brief.

[Nordenstam]

Originally Posted by terry j

Hi lupo, and happy ostkaka day. Re the omdiffusor (tm)...normal businesses have the omnidiffusor, we are better than that...would it not make more sense to overlap (in your example) 500 hz downwards rather than above?

It's hard to get more than half to 1 octave below 500 Hz with a practical build depth. Going down from 500 would give a diffuser with, at most, 1 octave range.

Originally Posted by terry j

I did mention to someone that you take breaks every now and then, I said it was because you obviously had a life. That was wrong, per you you are trying to get a life!

Good luck with the gigs and let us know how they go.

Hehe.. Did do a bit of soldering too while I was in the capitol. It takes a lot to give a nerd a life. :D The gigs went very well! Have lifted the live acts to a new level. Many happy people!

Originally Posted by terry j

NOW you mention using qrd wells to extend the effectiveness of a bass trap, you waited till you saw I was committed to the space coupler didn't you!

Well, it's a huge subject! It involves calculations of various impedances, especially important for the covering of the QRD trap. There's also optimizations for the well arrangement that makes it more effective at absorption than a normal QRD sequence. Both of which is beyond my meager math skills at the moment. Unfortunately!

I'm sure the traps you do will work fine as is.

Originally Posted by Xenon

"The large number QRD's need a longer distance to the listening position for all the waves in the diffuser to blend."

Can anyone clarify this point?

What I meant is that you want to be in the far field of the entire sequence. A 2 meter wide sequence needs longer listening distance than a shorter sequence.

PS: Congrats on the build! Looks very good.


Cheers,

Andreas

[Xenon]

Greetings Andreas,
I'm not sure what you're working on, it sounds interesting. Your Styrofoam 2D diffusor thread is one I've read several times. I hope to try something similar. You wrote:

Originally Posted by Lupo

PS: Congrats on the build! Looks very good.

Thanks!
And you wrote:

Originally Posted by Lupo

What I meant is that you want to be in the far field of the entire sequence. A 2 meter wide sequence needs longer listening distance than a shorter sequence.

Yes, what I don't quite understand is how in QRDUDE, the listening distances are arrived at differently. For instance (taking 135mm deepest well as a basis):

1) for N53, Panel width 2279mm: listening distance = 1656mm

2) but N23, Panel width 989mm: listening distance = 2076mm

So your comment and QRDUDE distance results are divergent, I think. Also you wrote in post #47: "The large number QRD's needs a longer distance to the listening position for all the waves in the diffuser to blend." But if you make a smaller well width (WW), a larger N-period is not necessarily longer than a smaller N QRD? But were you thinking of 'larger but maintaining a specific WW, for the sake of comparison'? Or just 'larger N = greater distance in all cases'? I wrote in post #50: "It seems intuitive that a shorter period length will 'cohere' or "blend" in less distance?" And showed these QRDUDE results:

QRDUDE Calc, listening minimum-distances:
N11(c2) = 254cm
N17(c4) = 212cm
N23(c0) = 208cm

Collo explained in post #51: "The recommended minimum distance to the seating position equals 3 times the wavelength of the LF cutoff frequency."

Do you disagree? Anyway, pragmatically, in a given (typical home/small room) application, the LF may change, but the max. well depth (MWD) is critical to the build. Of course we want the best LF we can get as well, but there must again, pragmatically, be a limit to how much room you want the unit to take up off the wall. In the above figures, following QRDUDE here are the LF numbers (MWD = 135mm) for each:

N11(c2) (473mm Length)= 254cm (list. dist.) = 811Hz LF
N17(c4) (731mm Length)= 212cm (list. dist.) = 972Hz LF
N23(c0) (989mm Length)= 208cm (list. dist.) = 994Hz LF

So the N11 with the least length has furthest listening distance (and lowest LF) and N23 with greatest length has the closest listening distance (and highest LF). On this point, between you and Collo, I'm confused, though I'm not confused about how QRDUDE is calculating listening distance.

It seems probable that actual period length must play a role in the time domain -- this issue has some buzz in loudspeaker design (temporal coherence --though here we are looking at incident reflection and phase).

Well I will pursue my earlier comment: following the Barker Code, in terms of difference, what makes a minus (-) to a plus (+)? Take a look at these four images from QRDUDE:
A

B

C

D


Consider this a thought-experiment, playing with the visual approach. For argument, all have the same MWD and WW. A&D have relative similarities (and what about A&B?), where A&C seem less visually similar. Of course D is N19 (and shorter in length than N23, and with a different LF), but my question is, can A&C, which are both N23, be considered as +1/-1, in that the periodic symmetry in either Barker sequence 4 or 5 (see below), we just need one 'different' array. We've discussed 'flipping' in various permutations. Tangentially applying this logic, A&C would appear to some extent to have an asymmetric relation to each other.

Barker sequences for 4 and 5 periods:
1) 4 periods: C as #3 in a set of 4 N23, with all others as A type [+1 +1 −1 +1];

2) 5 periods (as just above, but C as #4 in a set of 5 N23) +1 +1 +1 −1 +1.

At the distance at which all consecutive periods form a united incident diffusive wavefront (in a perfect world with only one wall), there may be issues. But how distant do you need to be from a multi-period QRD array which is at 500cm in length? And considering the Modffusor discussion above, might just moving wells as with A&C above, be sufficient? Advantages are a much more straightforward build, architectural niceties, possibly a benefit to maintaining the same level of diffusive resolution along the wall (in a small room), easier to dimension, less brainwork to build, time-saving -- did I miss anything (yes, RealTraps surrounds rooms with N13s on every wall surface--just to point out)?

I doubt there can be a definite answer without actual experiment, or some challenging math, but I'd be curious what the ripple tank would produce. And if you look here, at Finite Difference (and the links) it seems there may be a mathematical/theoretical method of determination (beyond my ability--and if this is even the correct application). Your thoughts?

[collo]

Related topic: When looking at different arrangements of the same order panel for use in a Barker sequence, it is worth comparing the scatter drawings.

Left shifting alone is not going to give you any substantial change in the diffusion pattern apart from some offset. Is it enough? That's a whole other can 'o worms...

Vertical shifting is the same situation until a well hits the bottom and then flips to become zero depth. You then get a change in the scatter angles. (As in your example D)

On a welcome note, QRDude now remembers your most recent values and settings, plus you can save and load individual projects.

[terry j]

have not got time now xenon, but your post has made the little hampster wheel in my head spin a few revolutions......explaining what I mean tho, well that's the tricky bit!

Bill, good work yet again.

Re the scattering angles in qrdude, I could be completely wrong, but I must confess I always thought they looked a little..primitive??...so kinda have discounted them. But it is yet another little trick we should employ when trying to work this stuff out. What is true about them I guess is that even if 'primitive' they do show that merely side stepping the qrd does not substantively change much....

On a completely different note, just found out that rpg (evidently) sell the bad panel alone, on it's own. Always thought it was the complete assembly etc.

For the life of me I can't see that option on the site (masonite panel on it's lonesome) but according to some bloke he bought them singly and assembled it into the completed unit.

He did mention however that it was perhaps 'a bit expensive for what they were'!! ...like all rpg products it seems.

Must run, and will try and assemble my thoughts into some sort of understandable form.

[Xenon]

Thanks Collo,

QUDUDE has made this discussion possible. And thanks for including a pic of my N23 on your Diffusion explanation page, I'm honored.

In terms of what Terry mentions below, I assumed QRDUDE is showing the phase angle of diffusion of each well, so its simple yet accurate enough to get a basic idea. What's especially beautiful is looking at the higher primes; QRDUDE goes up to N101 at present. I mean just take a gander at N83, that's simply gorgeous.

Check out these two images below, one is flipped so you can better compare the scatter alignments:


You can see the phase positions are quite altered at a given point in the period's length -- I mean that, if you imagine a wall of 4 periods, and period 3 is the above 'N23-shift 18,' the symmetric periodicity seems broken up, in terms of a phase-angle degree x at a given y mm period-length.

This brings up another question for me. We know of two ways to produce the Barker code (-1) variant: N23i (inverse) or using a different N-value period. The inverse seems the best choice (though often problematic in terms of needed MWD). What hasn't been discussed is what are potential negatives in mixing say one N19 within an array of three N23? Might the mixing of resolutions have a possibly unwelcome effect?

Anyway, in a smaller room with a series of 4 periods of say 400cm (13') total length (on a 500cm wall), and the opposite wall about 460cm (15') away, and mic positions being around 1/3-1/2 that distance (200cm-230cm), opposite various points in the array, it's my thought that breaking up the symmetrical phase relationship is perhaps enough. This by the way is not 'flipping' which does almost nothing with an N23. It is rather looking at the position of the scatter angles in QRDUDE and trying to find the best 'mismatch' at the same point in the period, lengthwise (isn't this what the flipping (upside-down) of asymmetric periods actually accomplishes?). As you wrote:

Originally Posted by collo

shifting alone is not going to give you any substantial change in the diffusion pattern apart from some offset. Is it enough? That's a whole other can 'o worms...

Yes, I think it may be to good effect, in a smaller room anyhow. I have only intuition to back this up, plus considering some of the 'flipping' advice. Not as good as building an inverse type, but certainly breaking up perfectly repeating phase 'points'.

Originally Posted by collo

On a welcome note, QRDude now remembers your most recent values and settings, plus you can save and load individual projects.

Awesome, a great plus, thanks!

[terry j]

So I have had twelve hours to gather my thoughts...how successful do you think I was??

So I am just gonna throw some unformed musings out, inspired by what you wrote xenon.

The thing that sparked this was your mention of 500cm long set of diffusors.

(man, that's a big array to pick!!..unless that IS the size of your array, have not gone back and checked)

We seem to be stuck on the barker code in some fashion. But where these ramblings are heading is that I'm beginning to think with a wall of diffusors 5m long, the barker code kinda becomes irrelevant.

I am wondering whether the barker code only really is important when the total width of the set is, gee dunno, 2m or less?? (just pluckin' figures for illustration)

And I vaguely recall in my thread mention of the width of the panel (single in that case, hence is prob part of the qrdude scoresheet) being related to the design frequency. So extending that idea, at some level the width of the entire set of panels must have some bearing.

But using your method of a thought experiment, with a wall 5m long of diffusors, (for ease of discussion 5*1m panels) then what I *see* is that, really, it does not matter too much what is happening at 'section four' as it is NOT (?) interacting with section one. It has got to have something to do with the wavelengths involved, (and I have not bothered to work out what the frequency of a five metre wavelength is...)

It's like some weird audio equivalent of locality in physics, anything happening at 'panel four' is totally divorced from that happening at 'panel one' for a given frequency.

On the *flip* side (barker code joke, geddit?) the wavelengths equivalent to a set of panels that add up to 2m (my arbitrary figure above) are also starting to be in the region of the design frequency (??), so we need to think about lobing and the barker code.

Sorry about the nebulous concepts, he's a tired, old decrepit hampster ok? don't be too hard on him.

[collo]

Xenon: Your idea of visually comparing one pattern above the other is a good one. (Unfortunately you rotated the top image 180 degrees rather than mirroring it vertically - the wells run in the wrong direction)

Here is a comparison of left-shift only versus vertical-plus-left shift.
For the vertically shifted panel, the design frequency was allowed to float to give the same build depth (and thus width)

You can see that the shift-only pattern doesn't introduce as much change as the other one.


I was happy to post a pic of your N23 - am sort of hoping that you would do a detailed writeup that I could link to (you're well on your way!)