Announcement

Collapse
No announcement yet.

Budget Duelund Three-way

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

  • Budget Duelund Three-way

    I’ve been testing drivers in this other thread. I thought it was finally time to share what for.

    Reading Jon Marsh’s adventures over at htguide has definitely inspired me here. All credit to Jon. Nothing I’m trying here is an attempt to do anything but piggy-back off his work on the cheap. I really wanted to see if I could design a Duelund three-way for under $200, which admittedly is a fairly arbitrary price point. But I chose it because I think it’s the budget cut-off for lots of the competitions at the get-togethers.

    I’m not qualified to expound on the benefits of a Duelund network, although Jon really explains it well if you care to read up. Rather this should probably be classified more as an intellectual exercise for me with the side benefit being, I get to play with and build speakers.

    I initially considered two different values of aleph in the Duelund equations, but through various modeling and experimentation settled on aleph = 2√2 (blue), but I then shifted the center frequency down to about 1300 Hz, lowering the crossover points.

    Click image for larger version

Name:	Duelund Transfer Functions.gif
Views:	2037
Size:	75.9 KB
ID:	1444114

    Through my driver testing, I settled on the DA215-8 for the woofer, the TG9FD10 for the midrange, and the DX20BF00-04 for the tweeter. The selection of the right midrange is critical here, and the performance of the TG9FD10 is the only reason this is possible.

    Click image for larger version

Name:	PCD Rev08.jpg
Views:	1940
Size:	228.2 KB
ID:	1444115

    The midrange driver needs to be small enough to allow for close tweeter placement since the mid-to-high crossover point is fairly high and we don’t want too narrow of a vertical lobe, but also robust enough to play cleanly down to around 250 Hz. The driver should also be as small as possible to achieve the greatest off-axis performance possible since it covers so much of the upper octaves with the tweeter (it is still contributing a couple dB at 6 kHz). Similarly, it is expected to contribute a couple dB at 200 Hz. It basically must be a full-range driver with solid HD performance across its entire bandwidth with no break-up. It is a pretty tall order.

    Initially I thought I was going to select the PC105-8 based on my HD testing. I declared it my winner based on its ability to play cleaner, lower. Then I put it in PCD and was slapped in the face by the cone break-up centered around 9 kHz. It made hitting the target low-pass slope damn near impossible without an excessive component count. It really highlighted how smooth an FR response is needed in order to keep the crossover network in line with the budget sensibility.

    Note*
    This model and testing was with the TG9FD10-04 which drops the impedance to just above 4 ohms at the 150 Hz and 800 Hz dip. I have the TG9FD10-08 on order with the hopes that it measures as well or better and allows a higher minimum impedance at these frequencies so that this can be classified more as an 8 ohm nominal design that any old receiver can power comfortably.


    The woofer cannot be any old woofer either. It must be able to play cleanly up to around 1 kHz and not have any excessive break-up until around 3—4 kHz. Even still, a bottomless notch was required to keep everything in line. I had really hoped that the DA215-8 would test well enough in the HD department to make it the right candidate, since it is one of the cheapest 8” drivers out there and integral to hitting the target price point. I’m happy enough with the testing to declare it just that.

    I also wanted to do this in a sealed monitor. Form factor is important and floor-standers are not practical for me anymore. I personally feel going below 100 Hz is not as important with most people running subs anyhow. The DA215-8 models reasonably well in around 11 liters which is about all I can afford in my targeted 16” x 10.5” x 11.25” enclosure.

    I don’t feel like there is too much to say on the DX20BF00-04 other than the fact that HiFiCompass tested it and it looks fantastic. It has a small form-factor, is easy to mount, and is even recessed a little bit to more closely align the acoustic centers between it and the midrange.

    So anyhow, the PCD model above has thirteen (13) components. Four (4) on the woofer, four (4) on the tweeter, and five (5) on the midrange. If you assume an electrolytic capacitor for the woofer shunt and the midrange high-pass only because they are large values, but the rest standard Polypropylene capacitors, the drivers and crossover come in just under $200.

    More to come.
    Cheers!


  • #2
    Cool - very ambitious project. As Jon says "Duelund crossovers are not for the faint of heart nor light of wallet". To pull one in for $200 is pretty remarkable, as is hitting one with relatively low parts count. I find the Duelund approach interesting, but haven't dipped my toes in just yet. Jon posted a different alignment that I intend to try with drivers on hand - RS225/SB15NBAC/DA25Tx - obviously this alignment offloads the mid and places more burden on the W/T elements, but I think the drivers can handle it. I'd be looking at an active xover to start, though.

    The little TC9 is a pretty cool driver. If you wanted to stretch the budget a bit, I think the RS100p would also be able to handle the mid duty. The Scan 10f is another one that might work, but might not handle the low end as you have it set up so you might have to tweak the alignment.

    The only question I have is that the Duelund alignment should have 360 degrees of phase shift over the passband, but your graph seems to show that you overshoot that a bit - an extra 60 degrees or so. Is this due to compromising on the target curves?

    Comment


    • #3
      Thanks for the post - I've wondered about this type of crossover for a long time.

      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


      • #4
        Originally posted by dwk123 View Post
        The little TC9 is a pretty cool driver. If you wanted to stretch the budget a bit, I think the RS100p would also be able to handle the mid duty. The Scan 10f is another one that might work, but might not handle the low end as you have it set up so you might have to tweak the alignment.
        Yeah, for sure after I go through the learning process on this budget build I plan on and am already formulating the higher end version using the RS225P, RS100P, and the XT25 or some variation.
        I had to really stop myself from going ahead and purchasing both the RS225P and RS100P to compare to my budget drivers, just to grasp the order of magnitude difference in HD.

        Originally posted by dwk123 View Post
        The only question I have is that the Duelund alignment should have 360 degrees of phase shift over the passband, but your graph seems to show that you overshoot that a bit - an extra 60 degrees or so. Is this due to compromising on the target curves?
        You're right!

        I have no idea.

        I played a little bit with the model today to see if I could pinpoint the extra phase shift. What I posted had realistic driver offsets in the x, y, & z direction, albeit estimated and planned and not measured. The woofer's z-offset will definitely be behind the midrange and tweeter, and I've already completed the design study on the vertical lobe shift that occurs. This is the mechanism that shifts the phase wrap down from the center frequency (where the woofer and tweeter roll-off intersect). But this doesn't seem to affect the phase significantly. The woofer's response also does deviate from the target roll-off slightly, but when I adjusted it back, the phase wasn't significantly altered either.

        But honestly, I was just elated that what I put together came even close to the theoretical Duelund behavior.

        It could also just be sloppy modeling.
        The tweeter .frd & .zma were traced from the HiFiCompass data and I have run into transcription errors when using these simple tools.
        I'm also using all of Jeff B.'s free tools to model, and it is possible in deriving minimum phase that I didn't approximate the tails quite right which can throw the phase off.

        Cheers!

        Comment


        • #5
          I think maybe it's the woofer roll-off adding the extra phase.

          Comment


          • #6
            So making some progress on the cabinets.
            I always inevitably plan on creating something that is far more ambitious from a construction perspective than it has any right to be.
            Here is the concept model:

            Click image for larger version

Name:	Concept Sketch.jpg
Views:	1679
Size:	38.7 KB
ID:	1447290

            I'm a fan of the Klipsch aesthetic and I'm going to duplicate that here. Solid oak panels with baltic birch reinforcement.

            Click image for larger version

Name:	Klipsch Forte III.jpg
Views:	1658
Size:	33.4 KB
ID:	1447291

            Yes, yes; I am aware of the difficulties in working with solid wood. Believe me, it was a terrible choice. I'd be done with these by now if I had just used BB plywood and butt joints everywhere. I just hate finishing speakers and figured using solid wood would allow me to short-cut that process.

            Click image for larger version

Name:	IMG_0787 (Medium).JPG
Views:	1686
Size:	231.8 KB
ID:	1447292
            Click image for larger version

Name:	IMG_0792 (Medium).JPG
Views:	1676
Size:	156.0 KB
ID:	1447293

            Comment


            • #7
              A little bit more on the design as well.
              Here is a vertical polar plot of the anticipated effect from a non-coincident woofer z-offset.

              Click image for larger version

Name:	Polar Plot Woofer z-offset.png
Views:	1698
Size:	379.2 KB
ID:	1447295

              It may be possible yet to deviate from the target roll off to align the lobe in execution, but even if I can't, it is only a 0.5 dB tilt across a pretty wide angle.

              I similarly looked at the vertical lobes created between the midrange and the tweeter resulting from such a high crossover point and significant driver overlap.
              First we start with a 3" C-C distance at 2,900 Hz.

              Click image for larger version

Name:	3 inch C-C 2900Hz Fc.png
Views:	1702
Size:	432.2 KB
ID:	1447296

              Okay, not bad.
              So what happens if we take the crossover frequency up to 3,875 Hz?

              Click image for larger version

Name:	3 inch C-C 3875Hz Fc.png
Views:	1699
Size:	404.2 KB
ID:	1447297

              This is a little problematic in my eyes.
              So what if we keep the crossover frequency at 3,875 Hz, but decrease the C-C distance to 2".

              Click image for larger version

Name:	2 inch C-C 3875Hz Fc.png
Views:	1690
Size:	428.0 KB
ID:	1447298

              Much better! Maybe not constructable though.
              Finally, how bad does it get if we keep the 3,875 Hz crossover point, but increase the C-C distance to 5"?

              Click image for larger version

Name:	5 inch C-C 3875Hz Fc.png
Views:	1690
Size:	416.3 KB
ID:	1447299

              Pretty textbook combfiltering.

              Comment


              • #8
                Great work, Greg!
                Wolf
                "Wolf, you shall now be known as "King of the Zip ties." -Pete00t
                "Wolf and speakers equivalent to Picasso and 'Blue'" -dantheman
                "He is a true ambassador for this forum and speaker DIY in general." -Ed Froste
                "We're all in this together, so keep your stick on the ice!" - Red Green aka Steve Smith

                *InDIYana event website*

                Photobucket pages:
                https://app.photobucket.com/u/wolf_teeth_speaker

                My blog/writeups/thoughts here at PE:
                http://techtalk.parts-express.com/blog.php?u=4102

                Comment


                • #9
                  Is that green tube going to be your mid range enclosure? Is the tweeter going to be mounted in the tube also?

                  Comment


                  • #10
                    The cabinets look nice. Assuming you like the color of raw oak, the most shortcut finish I can think of would be paste wax and 0000 steel wool.

                    Comment


                    • #11
                      Thanks guys.
                      I’ll have to take a screen shot of the baffle with the driver cutouts to show the arrangement, but yes the tweeter fits in the midrange tube as well. It’s 6” PVC. I had it hidden in the model. Barely fits the TG9 and tweeter once the chamfer is applied to the backside. It will work for this budget version, but also fits an RS100P and XT25.

                      The cabinets are still taking forever!
                      Cheers,

                      Comment


                      • #12
                        If you like the lightest Oak finish, you can use rattle can Lacquer. For a little darker finish, I really like Tried & True wipe on oil. I've used oak a lot lately.

                        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


                        • #13
                          Making a little progress on the first cab.
                          Almost ready to close up the other side.
                          Having trouble embedding pictures in the middle of this post now?

                          Also, here is the render of the budget and high-performance baffle and backside cutout to show how the drivers will be arranged.

                          And finally; a bit more on the design and benefits.
                          Here is the predicted horizontal off-axis plot. Notice how the directivity roll-off extends fairly low in frequency, closely aligned to the ideal behavior as described by Geddes.
                          The vertical plots basically replicate what has already been shown above in the polar plots and exhibit the off-axis nulls and lobes resulting from the driver spacing and crossover frequency.

                          Cheers!





                          Comment


                          • #14
                            Making some progress...

                            Both cabinets are together.
                            Overall I'm pretty happy with the joints, although a lot of epoxy filler was used on joints that weren't perfect. A detailed look reveals that the panels warped pretty obviously, making for corner joints that aren't perfect (and bow) and required a decent amount of trim with the flush bit. A table saw trim of the front and back faces make for a very professional look though. The rear panel with the terminals on one side look very authentic to me.

                            Hours upon hours have been spent on the baffle.
                            I dare say I will never use a driver that is not a circular cut-out again. I have almost finally completed my jig for cutting the recess of the frame of the TG9. Then on to trimming the tweeter flange.

                            Finally, I'll work on the grill. It will have cutouts made with a 45 deg. bit and be deep enough that it sits flush with the sides of the cabinet that current protrude past the baffle. Diffraction should not be as big an issue as it looks presently.

                            I cannot wait to take some in-box measurements and start prototyping the crossover.

                            Cheers!

                            Comment


                            • #15
                              I think those mids (after the HighPass is added) might be able to work similarly well with the PVC tube shortened and capped to ~2.5"-long if it isn't being left full-length for rigidity or some other reason.....if you think the woofer would benefit from an additional 1L-1.5L of room. I don't think the added space helps noticeably though unless you're aiming for the absolute smallest outer dimensions and wanted to save maybe 3/4" of depth, lol.

                              That box looks like it could support a house, and looks very nice as well.
                              My first 2way build

                              Comment

                              Working...
                              X