Once again, thanks -- every post, new information revealed -- how did you come by all this (forgive my ignorance).
I'll comment on your 'old school' formulation:
Quote:
Originally Posted by collo  In that case, go old school and have a look at 3 N23's and one inverse (I23)
The depths are different, but if you go for the +2 design, you get the best mix. For your 135mm depth, your design frequency will be 1162hz
Whilst this gets the lowest design frequency, you give up any zero-depth wells, meaning there is no point in side-shifting. |
Can I ask, why 'old' and 'school'? Just curious...
On the way to work today I realized something. I'd like to return to your
post #15 where you have an N11 in a half-frame (46.75cm (1.5') long x 93.5cm H). It's true, building half-frames takes more time and materials, but I'm planning to build half-frames of N11 laid horizontally over the 4 vertical panels, in any case. It can't hurt to build two more.
In Post #15 you write:
"N23 N11 N23 N23 is still a valid Barker sequence (you can run the sequence backwards) ... To get the benefits of multiple panels of the same order adding together to give more effective diffusion, you would want them all to be the same. This means all the N23's should have the same amount of left shift as each other, and none of them reversed."
Thus the need for half-frames of 11N, as all N23 must have a zero-well (choose one) hard left or hard right -- they cannot span frames as suggested (by me) before. I follow the
Barker code for 4 arrays of two types as:
+1 +1 −1 +1
+1 −1 +1 +1 (reversed)
or
+1 +1 +1 −1
−1 +1 +1 +1 (reversed)
Which looks like:
(N23 +, N11 = -)
N23 N23 N11 N23
N23 N11 N23 N23 (reversed, as suggested)
or
N23 N23 N23 N11
N11 N23 N23 N23 (reversed)
But I am short one N11 panel I need a minimum of @380cm length coverage -- where can it go? A Barker Code for 5 is needed:
+1 +1 +1 −1 +1
+1 −1 +1 +1 +1 (reversed)
And the problem is that the code calls for 4 N23 and 1 N11 (or the opposite). A possible solution, keeping in mind my wall is 500cm and I want corner traps,
I can build:
N23 N23 N23 N11 N23 - 437.75cm
This means 4 pcs. 93.5cm frames (plus left-side spacer) for the N23, and an additional half-frame for the N11, which will have its zero-well on the left side.
This would be the perfect arrangement, but it's too long. For the corner trap I need
at least 45cm (1.5') on each side of the diffusor -- and my plan calls for 60cm on each side, nominal. So I can't go beyond 410cm in length and would prefer 380cm of diffusion.
This problem encourages me to void the Barker code. In terms of dimension, the wall works perfectly with 3 N23 + 2 N11, so to void the Barker code, by taking the code for 5 and altering the last digit:
from
+1 −1 +1 +1 +1
to
+1 −1 +1 +1 −1
= N23 N11 N23 N23 N11
I'm not sure how badly the Barker code alteration will affect dispersion; the pattern is non-symmetrical -- it seems a compromise I have to make, as I don't want to alter my well depths or widths (and the relatively low LF of this-type N23 doesn't hurt. Also, the 4 horizontal half-frames will fit perfectly overhead the vertical frames.) In this design, all zero wells will be on the left side of each panel (and each N).
Not perfect ...
