2 ch listening room, aiming for best possible.

[terry j]

Originally Posted by collo

Can't see that it would be worth doing. The outer panel is only giving you one true diffusion freq @ 644hz, the inner gives 4068, 6102, 8136. We could call it the "donut" - as in wow, that's a big hole in the middle..

''donut'', that works...as does swapping the vowels..the ''dunot'' as in wow, du not bother.

agree totally, not worth doing (hence showing the example). It workds from all my ';old' ways of looking, ie it fits physically, per the flow of qrdude they overlap yada yada. (that bit in qrdude about effective diffusion frequencies, hadn't noticed it before, useful).

It was the 'old' way that I designed the trifractal...I gotta look at it thru the new way and see if it stacks up!


Funny how your thoughts on a subject can ebb and flow. we have often discussed the relative advantages of 1d, 2d and prd's. I am now (will check once I have another look) back to thinking that even tho you only get diffusion in one plane with a qrd, it really feels like it is so much easier to get an effective range of diffusion with them.

That might change once I have another look at a difractal tho.

But a BIG 2d qrd or prd (let's call the one I was going to build 'big' for sake of argument)...I was using a well size of 35 mm, and even THEN I only got to just over 4.7k..whatever it was!

Ok, instead of one BIG one (prd or 2d qrd) let's use *eight* (say) much smaller ones...remember how much 'angst' there was over the barker code for a qrd...what the hell is the 2d equivalent of the barker code??!!

Oh well, I was really hoping for a breakthrough in terms of speed and ease with the tricks I was going to add to the welled build.

The 'old' method of building is kinda...'ugly', and I'm not talking looks. Brute force, no elegance, that is why I am a bit disappointed. (I think my qrd build trick IS elegant you see...)

[Nordenstam]

Originally Posted by terry j

You are good at this, (explaining in simple terms for simple folk like me) let's say that we have the same number of wells in a 2d or prd, same number of depths, why would the prd be more effective? Is it as simple as no symmetry in the design?? is that all it is?

Semantic: Both QRD's and PRD's can be made in 1D and 2D. I take it that you mean QRD when you write 2D, but it would be less confusing if you could specify.

The main difference is the lack of symmetry, indeed. I'm not math geek enough to explain the more subtle sides of this. Though I can try..

Maximum length sequences (MLS) are the simplest diffusers. They use a binary sequence with only two possible values. The goal of a maximum length sequence is to generate a pseudorandom string of numbers. Typically used to generate "random" in an otherwise non-random digital world. The way to check that it is indeed as random as can be for a given length, is to check for a flat frequency response(fourier transform). The sequence is repetetive in that it can repeat without loosing the flat frequency response in the overlapping region. The on/off nature of such a sequence gives it limited usefulness in acoustics as the bandwidth is severely restricted.

Quadratic Residue Sequences is another way to generate a pseudorandom sequence. It gives more different steps than the on/off MLS sequence and as such the bandwidth is much broader. The frequency response is still flat. One of the quirks of the QRD is that it doesn't use all available numbers and it mirrors in the middle. This makes it less random than the flat frequency response may indicate at first glance, but the non-used numbers can, as we know, be used to optimize depth response.

Primitive Root Diffusers are based on the only public key encryption algorithm (RSA) that have not been broken yet, as far as we now. It basically scrambles a string of numbers in a way that is next to impossible to decrypt. A special feature of PRD's is that they supress specular reflection. The attenuation is 1/n where N is the prime number, ie 1/157=0.00637 = -43dB. Have a hunch this is due to the encryption nature of the device, but dunno. Nonetheless an interesting property.

I think the difference between QRD's and PRD's are best seen in the scalability. Making a large QRD quickly reveals the strickt geometry of those devices. They basicaly looks like this: \_-_/ no matter how small or large they are. The fact that it's so easy to see a pattern in them is a strong indication of the weak randomness inherent in the devices. PRD's, on the other hand, looks random now matter how you flip them. They can be small or large, rectangular or near square, they'll still give off no obvious hint as to any pattern in there. Making a single large rectangular unit close to 2:1 in length/width looks and acts quite the same as having two smaller units side by side. An example would be to cover an entire wall with a single diffuser. Doing so with a QRD would make for a very striking geometric pattern, but the same features that gives it the cool distinctively recognizeable pattern is also what makes the array less effective than it could be. Covering the same wall with a large PRD would make it impossible to see any patttern to it. It would be random stuff all over the place, no matter if you look at a portion of it or the total array.

I know this isn't as well explained as it should be.. Ask me again in a year.

Originally Posted by terry j

remember how much 'angst' there was over the barker code for a qrd...what the hell is the 2d equivalent of the barker code??!!

Barker code for two in a row is 11, so it doesn't matter with two side by side. For three..??? Dunno!

Originally Posted by terry j

The 'old' method of building is kinda...'ugly', and I'm not talking looks. Brute force, no elegance, that is why I am a bit disappointed. (I think my qrd build trick IS elegant you see...)

It is, indeed! Go with what you feel best about building..