Announcement

Collapse
No announcement yet.

Designing a better baffle for power response / off axis tweeter response

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

  • Designing a better baffle for power response / off axis tweeter response

    Hi all!

    Thanks to the power of VirtuixCAD and REW my interest in building speakers has come back. At least I'm understanding how important off axis and power response is for speaker deisgn.

    I hit a roadblock, though. I built 3 baffles. All of them have a 45° chanfer all around. Every driver is countersunk. But responses vary wildly. And some of them are better than others for integrating with a woofer.

    All of them have issues at around 2 KHz. This means I have to add a "BBC dip" for responses to look correct when integrating with a woofer. Some of them are worse than others.

    I am reaching the following conclusions...

    1) Tweeters shouldn't be crossed over that low. When doing that, you get a peakiness at 2-4 KHz, the presence area. Which is really bothersome. I know people love very capable tweeters crossing over really low. But I don't see a way to do so while keeping power response at a downward slope.

    2) Waveguides are amazing at helping integrating woofers with tweeters.

    3) Narrow baffles are better than wider baffles.



    I'd like to understand what the best design is for a baffle when considering off axis responses... any help will be greatly appreciated!

    Example 1: The waveguide.
    Baffle size: 29 cms wide, 48 cms tall.
    Drivers: Dayton RS-28 on a Visaton WG-148 Waveguide.

    Click image for larger version

Name:	WaveguideOffAxis.png
Views:	231
Size:	45.7 KB
ID:	1476353

    This looks amazing! But I have an intersting observation. At around 2 KHz, FR looks identical from 0 to 30 degrees. So power response is peaky at that area. A bit of a dip is needed around that frequency (I'm crossing over at around 1.1 KHz to a 10" woofer).

    Example 2: The narrow baffle.
    Baffle size: 18 cms wide, 40 cms tall
    Drivers: Dayton ND-20FA

    Click image for larger version

Name:	OffAxisNarrowbaffle.png
Views:	209
Size:	39.8 KB
ID:	1476354

    Looks kind of bad. I don't know why. Perhaps it was the close proximity to the woofers. But all in all, it was really simple to integrate with a pair of Zaph Audio ZA14's, and cross them over at 2.6 KHz with a reasonable looking power response, with a very light BBC dip.

    Example 3: The wide baffle.
    Baffle size: 27 cms. wide, 1 meter tall
    Drivers: Peerless DA32TX Corundum

    Click image for larger version

Name:	PeerlessConundrumOffAxis.png
Views:	206
Size:	36.9 KB
ID:	1476355

    Now, this is the head scratcher. I have a peak at 1.1 KHz on axis. But at 2-5 KHz, the off axis response is stronger than on-axis. In order to integrate this tweeter with an SB Acoustics 8" woofer I needed to create a very wide dip from 2 to 4.5 KHz.

    Out of all the designs, this is the one I still can't figure out how it came so wrong.

    Any ideas will be greatly appreciated!
    Line Array: IDS-25 Clone, FE-83.
    2-2.5 Way:
    Zaph Audio's winning entry: ZA5+SB29. - Microliths: RS125+RS28. - Small Bangs: TB W4-1658SB+SEAS 27TBFC/G. - Monoliths: Peerless 830884+SEAS 27TBFC/G.
    3-3.5 Way:Miniliths: SEAS P21/CA21REX+Neo8 PDR+Neo3 PDR. - Megaliths: 2xDayton RS270+2xT-B W4-1337SB+SB29. - ZDT3.5 +: 2xDayton RS180+Dayton RS52+Vifa DQ25. Reflexos: OB 4xDayton RS150 + Neo3 PDR.

  • #2
    What are the dimensions of the chamfers? If they are 3/4", you will have minimal effect on diffraction and radiation. You might consider looking into baffles that are faceted or have very large round overs in addition to using a waveguide.

    Also, there are three extremely different tweeter systems used in your test - hardly a methodology that will lead to understanding what is happening here.
    Don't listen to me - I have not sold any $150,000 speakers.

    Comment


    • #3
      Originally posted by johnnyrichards View Post
      What are the dimensions of the chamfers? If they are 3/4", you will have minimal effect on diffraction and radiation. You might consider looking into baffles that are faceted or have very large round overs in addition to using a waveguide.

      Also, there are three extremely different tweeter systems used in your test - hardly a methodology that will lead to understanding what is happening here.
      It's a 1" chamfer.

      Yep. I tried 3 different tweeters to try and understand the impact on them. So far, all I can understand is that waveguides are really the best for mating with a woofer, followed by narrow baffles.
      Line Array: IDS-25 Clone, FE-83.
      2-2.5 Way:
      Zaph Audio's winning entry: ZA5+SB29. - Microliths: RS125+RS28. - Small Bangs: TB W4-1658SB+SEAS 27TBFC/G. - Monoliths: Peerless 830884+SEAS 27TBFC/G.
      3-3.5 Way:Miniliths: SEAS P21/CA21REX+Neo8 PDR+Neo3 PDR. - Megaliths: 2xDayton RS270+2xT-B W4-1337SB+SB29. - ZDT3.5 +: 2xDayton RS180+Dayton RS52+Vifa DQ25. Reflexos: OB 4xDayton RS150 + Neo3 PDR.

      Comment


      • #4
        All the three tests show is the Dayton/Visiton combo works somewhat better on a 29cm baffle than the other two combinations attempted.

        A shallow guide on a 20mm tweeter on a baffle wide enough to drive diffraction ripples well below anticipated xo point will be stellar.

        Are you familiar with Edge? Diffraction modeling software. Even the basic modeler included in one of Jeff's older spreadsheets can really get you started. Diffraction is only one part of what you are attempting to understand, though. Off-axis behavior is the other part.

        Unfortunately, off-axis predictions in software are not to be blindly trusted - so testing is important. But if you want to understand baffle geometry and it's effect on power response, you need to set a constant.

        There are ways to solve it, it has been solved actually. Keep up the testing, I personally enjoy arriving at a conclusion empirically that matches well with what the guys at places like Harmon have determined.
        Don't listen to me - I have not sold any $150,000 speakers.

        Comment


        • #5
          Example 3: The wide baffle looks the smoothest overall and has dramatically less variation between angles. Why is it not the favored example?

          Comment


          • billfitzmaurice
            billfitzmaurice commented
            Editing a comment
            I agree. If anything the waveguide is the worst of the three IMO. Also, I'd remove the traces for wider than 40 degrees. All they do is confuse the issue.

          • fjhuerta
            fjhuerta commented
            Editing a comment
            The waveguide provides the most controlled off axis dispersion, and even at 90°, off axis curves between woofer and tweeter look the best.

          • ernperkins
            ernperkins commented
            Editing a comment
            IMO the RS28 in a waveguide is the best of the three. The added boost in the low end is a waveguide featue. It is easily flattened in the crossover and the added gain reduces the required tweeter drive level. That reduces the tweeter distortion and optionally let's you cross a bit lower. And the off axis plots show the best constant directivity response. For an example, see the Revelation Two – M4 WG (https://www.audioexcite.com/?page_id=6845). It uses a different tweeter (ScanSpeak D2905/950000), but its raw on-baffle response is very similar to your RS28A. Notice how simple the tweeter crossover is to acheive a LR2 response @ 2 KHz.

        • #6
          Excellent work by the OP, but I agree with Johnny.
          Better empirical results would be attained if you kept the tweeter under test the same, and varied waveguide and baffle edge designs.
          It would be awesome to see a simulation followed by a measurement of the same.
          I have a hard time really understanding the multiple curve graph for off axis, I prefer the "heat" graph, IE:
          Click image for larger version

Name:	polar-response-0-to-50-degrees- 2021-09-13.png
Views:	185
Size:	295.0 KB
ID:	1476374

          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


          • #7
            Silly question here: I've seen at least a couple of people mentioning how angles above 45° don't really matter that much. I thought from a power response perspective, they did (not so much from a listening window perspective).

            At 45°, the best performing tweeter is, indeed, the wide baffle one.

            What is your perspective on this?
            Line Array: IDS-25 Clone, FE-83.
            2-2.5 Way:
            Zaph Audio's winning entry: ZA5+SB29. - Microliths: RS125+RS28. - Small Bangs: TB W4-1658SB+SEAS 27TBFC/G. - Monoliths: Peerless 830884+SEAS 27TBFC/G.
            3-3.5 Way:Miniliths: SEAS P21/CA21REX+Neo8 PDR+Neo3 PDR. - Megaliths: 2xDayton RS270+2xT-B W4-1337SB+SB29. - ZDT3.5 +: 2xDayton RS180+Dayton RS52+Vifa DQ25. Reflexos: OB 4xDayton RS150 + Neo3 PDR.

            Comment


            • billfitzmaurice
              billfitzmaurice commented
              Editing a comment
              The same as my first post. What it does at 90 degrees only matters if you listen at 90 degrees. When's the last time you did that?

            • fjhuerta
              fjhuerta commented
              Editing a comment
              Obviously I never listen at 90°, but reflections are what worry me at this point. MOre to come.

            • billfitzmaurice
              billfitzmaurice commented
              Editing a comment
              If reflections are a concern you don't want wider dispersion, you want narrower. BTW, most of what you're seeing from the wave guide has nothing to do with diffraction. It's because the wave guide is a band pass device. That's why the high end droops. The bit of gain in the low end is where the throat is small enough to load the driver. This band pass effect is typical of horns, especially constant directivity horns. Many pro-sound power amps employ what's called CD EQ, which boosts the high end to compensate for the high frequency droop of CD horns, which looks very much like what you have.

          • #8
            Before I address a few points, I'll point out an issue that is largely ignored in most discussions on diffraction (although I see you considered it some) and likely one (possibly significant) reason for some of the variation with angle. With the exception of a waveguide on a tweeter, there will be diffraction of the tweeter from any nearby driver, the larger that second driver, the bigger the impact. You could have a baffle with perfectly controlled edge diffraction, yet there will be variations with angle due to the nearby driver. My first baffle diffraction control is always placing a strip of felt between drivers, at least 1/2" deep. I test various widths and heights to find an optimum or at least acceptable. There is a secondary effect in the baffle edges that are then occluded from direct tweeter waves that also have reduced diffraction impact.

            Originally posted by fjhuerta View Post
            All of them have issues at around 2 KHz. This means I have to add a "BBC dip" for responses to look correct when integrating with a woofer. Some of them are worse than others.

            I am reaching the following conclusions...

            1) Tweeters shouldn't be crossed over that low. When doing that, you get a peakiness at 2-4 KHz, the presence area. Which is really bothersome. I know people love very capable tweeters crossing over really low. But I don't see a way to do so while keeping power response at a downward slope.
            This can be greatly ameliorated with additional diffraction control, such as using felt, even if you have chamfers. You also have to consider what the tweeter will be paired with along with the crossover order and Fc. There isn't any single rule that fits every situation. I use the Ultimate Equalizer (UE) on a smallish baffle with diffraction control, despite using the DXT with limited horn response crossed LR8 @1200Hz. Granted you wouldn't do this passively, but I would still probably be below 2KHz passively having the diffraction control added. It can greatly reduce the peaking in the 2-4KHz area with the added benefit of reducing the variations off-axis.

            Due to using the UE that guaranteed an absolutely flat on-axis response no matter what crossover order and Fc I tested I found that a smooth off-axis (hence smooth power response) required low Fc and high order due to the midwoofer used coupled with the baffle dimensions. Of course that was due to extensive diffraction control as well. Remember that the baffle dimensions affect more than the tweeter. In my case added consideration was using the midwoofer in dipole mode.

            2) Waveguides are amazing at helping integrating woofers with tweeters.
            Depending on the woofer and tweeter pairing.

            3) Narrow baffles are better than wider baffles.
            For my purposes I prefer narrow baffles, but again, I don't see any single rule that fits all situations.

            Looks kind of bad. I don't know why. Perhaps it was the close proximity to the woofers. But all in all, it was really simple to integrate with a pair of Zaph Audio ZA14's, and cross them over at 2.6 KHz with a reasonable looking power response, with a very light BBC dip.
            The smaller the tweeter, the more diffraction effects from a nearby driver. And the smaller diameter the nearby driver, usually the higher it can be crossed with smooth off-axis due to the different driver directionality.

            Now, this is the head scratcher. I have a peak at 1.1 KHz on axis. But at 2-5 KHz, the off axis response is stronger than on-axis. In order to integrate this tweeter with an SB Acoustics 8" woofer I needed to create a very wide dip from 2 to 4.5 KHz.
            I would at least test with felt between drivers only, then add some to the edges to investigate where the diffraction problems are located, even if you don't want it on permanently. It's much easier than trying to guess where the problems are.

            dlr
            WinPCD - Windows .NET Passive Crossover Designer

            Dave's Speaker Pages

            Comment


            • #9
              As always, thanks all for your most well thought answers.

              I assumed it was diffraction from the nearby woofers and edges. I just had never gone beyond 30°, and as such, my measurements suffered.

              Due to using the UE that guaranteed an absolutely flat on-axis response no matter what crossover order and Fc I tested I found that a smooth off-axis (hence smooth power response) required low Fc and high order due to the midwoofer used coupled with the baffle dimensions. Of course that was due to extensive diffraction control as well. Remember that the baffle dimensions affect more than the tweeter. In my case added consideration was using the midwoofer in dipole mode.
              This is exactly how I crossed over a 10" woofer to my waveguide-modded Dayton RS-28, and it works pretty OK.

              The smaller the tweeter, the more diffraction effects from a nearby driver. And the smaller diameter the nearby driver, usually the higher it can be crossed with smooth off-axis due to the different driver directionality
              Thank you... this is the kind of stuff I'm aching to learn.

              I would at least test with felt between drivers only, then add some to the edges to investigate where the diffraction problems are located, even if you don't want it on permanently. It's much easier than trying to guess where the problems are.

              I have wool felt from McMasterCarr, but I guess it's too thin to be of any use (real wool felt, 1/4x1/2). I'll place it semi permanently.

              Regarding power response, in room response vs. listening window, I simulated my crossover using my MiniDSP. The "flat on axis, flat listening window", performed terribly and was shouty at all levels.

              The version where I optimized for a slightly dropping power response:

              Click image for larger version  Name:	PowerDI.png Views:	0 Size:	29.1 KB ID:	1476397

              Click image for larger version  Name:	Offaxishor.png Views:	0 Size:	42.8 KB ID:	1476398

              Actually was completely spot on with real life measurements (in-room, at about 3m listening distance).

              Click image for larger version  Name:	WhatsApp Image 2021-09-21 at 15.47.32.jpeg Views:	0 Size:	119.0 KB ID:	1476399


              The on-axis response was understandably weird to begin with because it was never a good idea to mate a 1.25" tweeter with an 8" woofer, but as a proof of concept, it was great. The speaker actually sounds very good, indeed.
              Line Array: IDS-25 Clone, FE-83.
              2-2.5 Way:
              Zaph Audio's winning entry: ZA5+SB29. - Microliths: RS125+RS28. - Small Bangs: TB W4-1658SB+SEAS 27TBFC/G. - Monoliths: Peerless 830884+SEAS 27TBFC/G.
              3-3.5 Way:Miniliths: SEAS P21/CA21REX+Neo8 PDR+Neo3 PDR. - Megaliths: 2xDayton RS270+2xT-B W4-1337SB+SB29. - ZDT3.5 +: 2xDayton RS180+Dayton RS52+Vifa DQ25. Reflexos: OB 4xDayton RS150 + Neo3 PDR.

              Comment


              • #10
                Originally posted by fjhuerta View Post
                I have wool felt from McMasterCarr, but I guess it's too thin to be of any use (real wool felt, 1/4x1/2). I'll place it semi permanently.
                You can layer felt. It isn't easy to make one layer stick well to another, but it will function about as well as a single piece of the same thickness. I would test two layers. I'm assuming that you mean 1/4" thick, so you'll have 1/2" thick by 1/2" width by maybe 6" long. That's a good test between drivers. Probably good enough for baffle edge testing as well. For edges, place it maybe half way between tweeter and edge, that is far enough away to have minimal reflection issues and provide more occlusion of the edge overall than placing it directly at the edge.

                dlr
                WinPCD - Windows .NET Passive Crossover Designer

                Dave's Speaker Pages

                Comment

                Working...
                X