Announcement

Collapse
No announcement yet.

Testing baffle edge treatments for tweeter diffraction.

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #46
    Originally posted by Chris Roemer View Post
    Most usually, the staggered distances to baffle edges will smooth out "on-axis" anomalies. A lot of designs where they're not staggered might not be that great on-axis, but will probably smooth out off-axis a bit.
    So, if staggered - you can toe 'em "in" (toward your listening position), but if they'e not, they often should be pointed straight out (perpendicular to the wall behind).
    If you want YOUR version to match the one you think you liked enough to build (from reviews), I'd NOT alter the placement of the drivers on the baffle.
    Very informative. I have read many reviews in the past where the reviewer mentions that a certain speaker works toed in etc. I had not even given thought to this.

    Thank you very much.

    Comment


    • #47
      Originally posted by dlr View Post
      As far as I know, felt for diffraction control was first used on the LS3/5A and yes, it is effective. To what degree with the amount they used, I don't know. Thicker and broader is usually better with a point of diminishing returns. It was especially helpful for the LS3/5A since the front baffle was actually recessed to allow for a front grill. However, I have no doubt that more than they used would have improved the response, but as a commercial product, reducing cost is always a top consideration.

      dlr
      Thank you dlr. I have to look at wool and or other materials and what frequencies they are best may help attenuate. I have a fair amount of wool. I am dumb enough to try to make a "functional grill” using wool. I could cover all the baffle area with wool (covered in speaker cloth.)

      Comment


      • #48
        Originally posted by DaveFred View Post
        Do you have drop box? The data file is 36 megs, I am happy to send it to you.
        I missed this the other day. No, I don't have that. 36M sounds like a lot for frd files (text only). Should be a lot smaller, especially when zipped. The text frd files are all that I want. I've never used REW, but I would expect that it provides a means to export frd (txt) files.

        Can you put a zip file into the Onedrive folder created by ergo?

        dlr
        WinPCD - Windows .NET Passive Crossover Designer

        Dave's Speaker Pages

        Comment


        • #49
          REW stores all the measurements as a sweep (ie WAV or something like that internally) into the file, so it would include all necessary to process the IR... I could extract the FRD out of it. I do have REW and have used it to do that too.

          Comment


          • #50
            Originally posted by ergo View Post
            Very interesting study DaveFred - Thank you for sharing.

            I created a OneDrive link with write access.
            https://1drv.ms/u/s!AvVPh3wl37Nzm4Qu...ktCxQ?e=YJUFZD
            Would be cool if you could copy the REW files there. Will make it easier to select the type of interest for further study.
            REW raw data file uploaded.

            Comment


            • #51
              Thank you DaveFred. I've extracted the txt files out with space and tab delimited format and added a zip file to the same folder above. VituixCAD for example is able to open both without changes

              Comment


              • #52
                One alternate view for each baffle type and offset used for each angle for clarity. 0deg highilighted
                Attached Files

                Comment


                • #53
                  Thanks for the additional data. Lots of good information here. Another interesting test would be to mount this particular tweeter (SB Acoustics SB26ST-C000-5) on an IEC-268-5 test baffle and run up the same set of curves. This would give us a better control reference against which to judge how the various baffle board choices affect performance.
                  SideTowers: http://techtalk.parts-express.com/fo...corundum-build
                  Totally Flat: http://techtalk.parts-express.com/fo...5-totally-flat
                  Plumber's Delight: http://techtalk.parts-express.com/fo...notech-winners
                  Linehopper: http://techtalk.parts-express.com/sh...Esoteric-build

                  Comment


                  • #54
                    Thanks for posted the "txt" files. I used Omnimic and created normalized polar graphs of the square box and big bevel measurements. Clear improvement.
                    Click image for larger version

Name:	normalized-square-box.jpg
Views:	217
Size:	59.3 KB
ID:	1427417 Click image for larger version

Name:	big-bevel-normalized.jpg
Views:	205
Size:	54.1 KB
ID:	1427418

                    I think I hear a difference - wow, it's amazing!" Ethan Winer: audio myths
                    "As God is my witness I'll never be without a good pair of speakers!" Scarlett O'Hara

                    High value, high quality RS150/TB28-537SH bookshelf - TARGAS NLA!
                    SB13/Vifa BC25SC06 MTM DCR Galeons-SB13-MTM
                    My Voxel min sub Yet-another-Voxel-build

                    Tangband W6-sub

                    Comment


                    • #55
                      Originally posted by ergo View Post
                      Thank you DaveFred. I've extracted the txt files out with space and tab delimited format and added a zip file to the same folder above. VituixCAD for example is able to open both without changes
                      Any chance you could export those with FFT of maybe 500 instead of almost 54000? WinPCD could import that, but I have a program that doesn't re-map down, so working with 54000 data points is a problem. Plus, it's reading all of the files as well, so I'll have to modify it to re-sample and I don't know when I could get around to that. I suppose I should do that in any case. I'll have to review the code to see if I may also have a bug.

                      dlr
                      WinPCD - Windows .NET Passive Crossover Designer

                      Dave's Speaker Pages

                      Comment


                      • #56
                        All of the files, space or tab delimited, seem to load and display OK in OmniMic.

                        I was curious as to how these curves would compare to the published specs on an IEC baffle, so I traced the data sheet curves into an FRD. My tracing skills are not the best, so please take the attached comparison curve with a grain of salt. The bulge and dip from 1-2kHz is obviously the baffle step, which flattens out on the IEC. It is interesting to note that all baffles seem to depress the 8-12kHz area compared to the IEC baffle.

                        Click image for larger version

Name:	On-axis comparison to IEC baffle.jpg
Views:	201
Size:	130.6 KB
ID:	1427537
                        SideTowers: http://techtalk.parts-express.com/fo...corundum-build
                        Totally Flat: http://techtalk.parts-express.com/fo...5-totally-flat
                        Plumber's Delight: http://techtalk.parts-express.com/fo...notech-winners
                        Linehopper: http://techtalk.parts-express.com/sh...Esoteric-build

                        Comment


                        • #57
                          I think that I can re-map them myself. In the past I've used the free version of Praxis quite a bit for some post-processing, but it didn't seem to work in Win10 last I tried. I just tried it on my Win10 laptop, seems OK.

                          dlr
                          WinPCD - Windows .NET Passive Crossover Designer

                          Dave's Speaker Pages

                          Comment


                          • #58
                            Originally posted by 4thtry View Post
                            It is interesting to note that all baffles seem to depress the 8-12kHz area compared to the IEC baffle.
                            It's more likely that the raw driver varies from the published specs in that area. That kind of variance happens more often than not.

                            dlr

                            WinPCD - Windows .NET Passive Crossover Designer

                            Dave's Speaker Pages

                            Comment


                            • #59
                              Originally posted by donradick View Post
                              Thanks for posted the "txt" files. I used Omnimic and created normalized polar graphs of the square box and big bevel measurements. Clear improvement.
                              Click image for larger version  Name:	normalized-square-box.jpg Views:	39 Size:	59.3 KB ID:	1427417
                              For what it's worth, here's the result of a diffraction simulator I wrote. I can't do arbitrary shapes, but I think I can do roundovers and square edges. My square edge result is attached. I didn't import the frequency response or anything, so my 1" tweeter has perfectly flat on axis response. Still, I see a lot of similarity. I suspect that either I'm over-estimating the dispersion of the tweeter, or else some detail has been missed in using measurements taken every 15 degrees. More measurements would be interesting.

                              Comment


                              • #60
                                Originally posted by benb View Post

                                For what it's worth, here's the result of a diffraction simulator I wrote. I can't do arbitrary shapes, but I think I can do roundovers and square edges. My square edge result is attached. I didn't import the frequency response or anything, so my 1" tweeter has perfectly flat on axis response. Still, I see a lot of similarity. I suspect that either I'm over-estimating the dispersion of the tweeter, or else some detail has been missed in using measurements taken every 15 degrees. More measurements would be interesting.
                                Very impressive work, dude!
                                I'm not competent to discuss the mathematics of your simulation, but I think some other folks here can get into deep discusssions.
                                The most sophisticated software I've seen is "the edge", you might want to compare your results with those simulations.

                                I think I hear a difference - wow, it's amazing!" Ethan Winer: audio myths
                                "As God is my witness I'll never be without a good pair of speakers!" Scarlett O'Hara

                                High value, high quality RS150/TB28-537SH bookshelf - TARGAS NLA!
                                SB13/Vifa BC25SC06 MTM DCR Galeons-SB13-MTM
                                My Voxel min sub Yet-another-Voxel-build

                                Tangband W6-sub

                                Comment

                                Working...
                                X