Announcement

Collapse
No announcement yet.

Help with a variation on Pete S. 3Way Peerless Design

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

  • Help with a variation on Pete S. 3Way Peerless Design

    I am working on a variation of Pete Schumacher's 3 Way Peerless design from 2010. His design can be found here: Tech Talk Peerless 6 woofer 3 Way. I am building mine with the Peerless 830874 woofers and a pair of ESS AMT5 tweeters that were given to me, in the same box as he designed. Thought this would be a neat project to put them together. I recreated the overall crossover targets in PCD using the data provided by peerless and my own measurements of the ESS AMT5 tweeter via Dayton WT3 and the Omnimic.

    [Side note, I measured the ESS tweeter at 2v (4ohm tweeter) on a big flat baffle from 4" away. It turned out super smooth, but with a rising response, which I know is accurate from reading about these tweeters]

    So I am using basically the same target slopes and overall freq response as Pete and I was able to recreate the target pretty well with the different drivers. Pete modeled his in PCD without including the box response and built the speakers and liked them. I tried to model the box response using UniBox and Response Modeler as a way to verify the merits of the design. But, that totally screwed up my results. I tried it several times in case my settings were wrong, but it always failed. I went from a smooth well behaved model in PCD to a lumpy mess that would never mesh with the ESS tweeter no matter what I tried.

    So, I am wondering what went wrong? Is this common? Did I completely mess up response modeler? Does SPLTrace based data never work in RM? Should I be considering RM to be accurate and I need to go back to the drawing board and find a new tweeter? Or, should I just build it, considering Pete's design worked out well for him and my PCD response with my drivers looks just like his with his drivers?

    -Steve

  • #2
    IMHO, you first work up the box model for the woofer(s) using Unibox or WinISD (I like Unibox the best), that will give you the box size,
    the lower F3 point, and the shape of general response for 30Hz - 200 Hz. If you want, you can "splice" that into your Omnimic measurement
    of the woofer for frequencies 200-2000. But it's not really necessary to do that.
    The box size that you get from Unibox means that you can now figure out box width, height, and depth.
    Build a box (maybe a test box), install your woofers, measure FR (with phase) in Omnimic, then import that into PCD (or Xsim) and you are good to go.
    The ESS tweeter should be totally awesome with those Peerless HDS woofers. Keep it!
    Hope this helps.

    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


    • #3
      What did you use for your Z offset of the woofer's acoustic center in relation to the tweeter's acoustic center? If you don't get that right then your simulations in PCD, WinPCD, or Xsim are not going to match the final results. Some guys on here offer up crossover designs for others but they just take a SWAG at the Z offset.

      Did you add in the baffle step loss and diffraction in Response Modeler?
      Craig

      The lowest possible F3 box alignment is not always the best alignment.

      Designing and building speaker projects are like playing with adult Lego Blocks for me.

      Comment


      • #4
        Trace data works after you add in the baffle and the box. The Response Blender is a bit easier than Response modeler but you need Response Modeler for the impedance file. Kind of hard to guess what went wrong without seeing your steps/ data. The Peerless HDS looks like it can go to 4K, So like Don said build, measure, and design the x-o if you have omnimic.
        John H

        Synergy Horn, SLS-85, BMR-3L, Mini-TL, BR-2, Titan OB, B452, Udique, Vultus, Latus1, Seriatim, Aperivox,Pencil Tower

        Comment


        • #5
          Originally posted by shudson105 View Post
          I am working on a variation of Pete Schumacher's 3 Way Peerless design from 2010. His design can be found here: Tech Talk Peerless 6 woofer 3 Way. I am building mine with the Peerless 830874 woofers and a pair of ESS AMT5 tweeters that were given to me, in the same box as he designed. Thought this would be a neat project to put them together. I recreated the overall crossover targets in PCD using the data provided by peerless and my own measurements of the ESS AMT5 tweeter via Dayton WT3 and the Omnimic.

          [Side note, I measured the ESS tweeter at 2v (4ohm tweeter) on a big flat baffle from 4" away. It turned out super smooth, but with a rising response, which I know is accurate from reading about these tweeters]

          So I am using basically the same target slopes and overall freq response as Pete and I was able to recreate the target pretty well with the different drivers. Pete modeled his in PCD without including the box response and built the speakers and liked them. I tried to model the box response using UniBox and Response Modeler as a way to verify the merits of the design. But, that totally screwed up my results. I tried it several times in case my settings were wrong, but it always failed. I went from a smooth well behaved model in PCD to a lumpy mess that would never mesh with the ESS tweeter no matter what I tried.

          So, I am wondering what went wrong? Is this common? Did I completely mess up response modeler? Does SPLTrace based data never work in RM? Should I be considering RM to be accurate and I need to go back to the drawing board and find a new tweeter? Or, should I just build it, considering Pete's design worked out well for him and my PCD response with my drivers looks just like his with his drivers?

          -Steve
          Since you can measure, the x-over sim doesn't need to be perfect, or even good. You can fix anything x-over related after you have a box ready for measurements. The 830874 is an easy driver to work with, and you can cross it almost anywhere needed. I have lots of practice with this driver, and enjoy playing with sims, so you have help available from me and others on the x-over. I'm pretty sure I've even simmed a similar design. I have some other sim tricks that might help as well.

          Post your PCD sim results, and let me see what you came up with.

          Comment


          • #6
            Originally posted by shudson105 View Post
            I am working on a variation of Pete Schumacher's 3 Way Peerless design from 2010. His design can be found here: Tech Talk Peerless 6 woofer 3 Way. I am building mine with the Peerless 830874 woofers and a pair of ESS AMT5 tweeters that were given to me, in the same box as he designed. Thought this would be a neat project to put them together. I recreated the overall crossover targets in PCD using the data provided by peerless and my own measurements of the ESS AMT5 tweeter via Dayton WT3 and the Omnimic.

            [Side note, I measured the ESS tweeter at 2v (4ohm tweeter) on a big flat baffle from 4" away. It turned out super smooth, but with a rising response, which I know is accurate from reading about these tweeters]

            So I am using basically the same target slopes and overall freq response as Pete and I was able to recreate the target pretty well with the different drivers. Pete modeled his in PCD without including the box response and built the speakers and liked them. I tried to model the box response using UniBox and Response Modeler as a way to verify the merits of the design. But, that totally screwed up my results. I tried it several times in case my settings were wrong, but it always failed. I went from a smooth well behaved model in PCD to a lumpy mess that would never mesh with the ESS tweeter no matter what I tried.

            So, I am wondering what went wrong? Is this common? Did I completely mess up response modeler? Does SPLTrace based data never work in RM? Should I be considering RM to be accurate and I need to go back to the drawing board and find a new tweeter? Or, should I just build it, considering Pete's design worked out well for him and my PCD response with my drivers looks just like his with his drivers?

            -Steve
            Since you can measure, the x-over sim doesn't need to be perfect, or even good. You can fix anything x-over related after you have a box ready for measurements. The 830874 is an easy to work with driver, and you can cross it almost anywhere needed. I have lots of practice with this driver, and enjoy playing with sims, so you have help from me and others on the x-over available. I'm pretty sure I've even simmed a similar design. I have some other sim tricks that might help as well.

            Post your PCD sim results, and let me see what you came up with.

            Comment


            • #7
              If you know Pete's tweeter and high-pass filter, all you should have to do is to take your AMT files files and match his tweeter level & rolloff (phasing issues aside).
              If you can't get the phase close enough by flipping the AMT polarity, your next recourse would be to (probably) up IT's pass order a notch and try again.

              Comment


              • #8
                For some reason, my post was deleted by a moderator as spam. Any moderators want to take a shot at explaining?

                Comment


                • #9
                  The 830874 is easy to work with. I've used it, and made many sims with it, including designs like what you are planning. There's no reason to be concerned about getting a proper x-over.

                  Please post some of the sims that you have been working on.

                  Comment


                  • #10
                    I uploaded the PCD images of my work without the box model which closely mirrors Pete's project. Then there are 3 variations of crossovers that use a modeled box. The box is exactly the same dimensions as that used by Pete. 38" tall, 8" wide. A .25cf sealed midrange enclosure and a 1.5 cf woofer box ported at 45hz. As you can see in the box model version, no version I tried could get the tweeter to blend with the midrange. Don't know why. The more I tweak the crossover with the box included, the worse the upper midrange appears. At no point could I get a smooth transition regardless of order or part values, or even polarity.

                    Also, Craig, I did include BSC and diffraction in my box model.

                    The steps I performed in RM included inputting the TS parameters from the Peerless data sheet, inputting the box from Pete's design, splicing it into the model then modifying the model to account for diffraction based on individual speaker location, then auto extracting the phase then saving it separately for each speaker.
                    Attached Files

                    Comment


                    • #11
                      Here is the overall variations in PCD results using the box model
                      Attached Files

                      Comment


                      • #12
                        I think I see the problem. The low-pass on the mid appears wrong. Sometimes PCD acts up, and the dialed in x-over order isn't doing what it is supposed to be doing. I say this because the mid response is running out to about 5k, just like it does raw. Try switching orders a few times, and see if that helps. Also, I've seen lots of in box measurements of the 830874. I'd like to see the raw response as it appears in PCD. Did your target for the mid get changed to 5k by accident?

                        I like the work you did in the first sim.

                        Comment


                        • #13
                          I see similar weirdness on the tweeter. Maybe post the mid section of the sim, where it's filtered response is shown on the same graph with the raw response, and the filter values are on the left.

                          Comment


                          • #14
                            Dumb question...anytime you change the FR splicing in data etc...do you re-extract minimum phase?

                            Comment


                            • #15
                              Originally posted by Dave Bullet View Post
                              Dumb question...anytime you change the FR splicing in data etc...do you re-extract minimum phase?
                              Yes
                              John H

                              Synergy Horn, SLS-85, BMR-3L, Mini-TL, BR-2, Titan OB, B452, Udique, Vultus, Latus1, Seriatim, Aperivox,Pencil Tower

                              Comment

                              Working...
                              X