Dvärgmusik (aka Thorin II-way)

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • Bear
    Super Senior Member
    • Dec 2008
    • 1038

    Dvärgmusik (aka Thorin II-way)

    Does the world need another 6+1 design? Not in the least. But I need one. Three, in fact. These will be the center channel and surrounds for the next iteration of my home theater. I've designed and built a two-way a bit over two years ago that were then gifted. They featured the ZA14 and the SB29RDNC. Getting back to today, Brandon's/Augerpro's work on waveguides has just about perfected a waveguide for the SB26ADC/CDC tweeter. Combine one of those gems with the SB17NBAC, and you have ... the Bromo? That won't do. Besides, the Bromo doesn't have the time offset built into the cabinet.

    (Hey, Bear. Um... Are you now talking about the SBA-61 from Troels Gravesen?)

    Shut up, disembodied voice. I started this thread with the acknowledgment that the world doesn't need another one of these things. And yes, I'm aware that other people have also mentioned similar pairings with one Brandon's waveguides (including the developer himself). So... I'm not breaking any new ground here. In fact, I'm trying to pave over a cowpath that was turned into a multi-lane superhighway a long time ago. But since I apparently like to make mistakes in full view of the public, here we go.

    What's my hook here? Cutting boards. These are being made out of bamboo cutting boards. I will now pause to accept all of your apologies for not understanding the awesomeness of my creativity.

    (An overpriced IKEA hack, Bear? That is pretty weak.)

    Well, fine. Let's get to it anyway.

    Project name: Dvärgmusik (bad Swedish translation for Dwarf Music)
    Since everything in my household that gets named has a name related to Tolkien, these are going to be the Thorin II-ways. You didn't think that the humor got any better, did you? I'm not going with the Bromo's because I really don't like the white anodizing for the CAC series. Since SBA could just as easily create a darker gray than the NBAC and still use the CAC anodizing process, I just don't dig the aesthetic. I'm not going with the Troels design because a) I'm really just not a fan of the stepped baffles that have become his signature design element, and b) Jantzen only accepts credit cards with some European security protocol that isn't supported by any of my cards. Also, I don't wire anybody money. The dwarf theme was chosen because these are about as short and small as I can reasonably make them given the driver compliment and design goals.

    Since my Helios speakers are up and running, I can spend some time on these. Also, I'm overdue for getting to play with actual crossover design, so let's get to it.

    What are the key design parameters?
    • 105dB at 1m (+20dB of headroom from a 75dB baseline 3m from the source)
    • Low distortion
    • Only needs ~100 watts of power to reach full SPL
    • Specifically designed to cross to a subwoofer (80 Hz for 100W, or around 100Hz for 200W)
    • Clean midrange and good phase transition through the crossover region
    • Keep impedance as close to 4ohms or above as much as possible
    • Eliminating the need for bass allows a much smaller sealed cabinet, even for a driver designed for vented applications


    Target completion: 2Q21 (just as exact as a died-in-the-wool product manager would do it, but without any pesky customers for the sales team to deal with)

    Drivers and Noteworthy Parts:
    • SB17NBAC-4
    • SB26ADC
    • 5" SB26 waveguide, version J (no phase shield, 0.75" depth)
    • External crossovers connected via 4-pole SpeakOn connectors
    • 0.5 pounds of stuffing
    • Sealed enclosure (note the lack of low-end bass)
    • 7 Aptitlig 11x17 cutting boards per cabinet (19mm bamboo is also used)
    • Dimensions: 9" wide x 13.5" high x 9" deep; ~10L gross, 9L net


    Since I do expect to cheat off of the Smart Kid's test paper, at least a bit, I know that I'm probably aiming at about a 2kHz Fc. The Bromo does what all factory two-way designs seem to do, in that it has an Fc at 3kHz. The Purifi model design also has a 3 kHz crossover point. This is also what people used to do with the SEAS Excel line. Since the only real flaw in the SB26 design is that the sensitivity is on the low-end for a full-sized tweeter, the waveguide will help with this.

    Waveguide Benefits:
    • Improved sensitivity at the bottom of the tweeter's operating range
    • The above allows for higher sensitivity and/or a lower crossover point
    • Improved directivity matching to a woofer (TANSTAAFL; that increased sensitivity comes from the energy that is not being radiated outwards)
    • Better Z-axis/Time alignment with a woofer (I'm expecting the SB17 to be around 18mm behind the baffle at 2kHz)
    • Reduced crossover complexity/cost for the tweeter, though I'm not holding out for a "one cap to rule them all" HP filter.


    Click image for larger version

Name:	Thorin II.jpg
Views:	1
Size:	74.6 KB
ID:	872557

    One important note: the dimensions, above, work for 19mm/0.748" bamboo lumber. The cutting boards I used were consistently about 16.3mm according to my digital calipers. That would cause a reduction of about 0.25" in each dimension.
    Last edited by Bear; 17 November 2020, 09:55 Tuesday.
    Welcome to Rivendell, Mr. Anderson.
  • Bear
    Super Senior Member
    • Dec 2008
    • 1038

    #2
    Reserved (crossover)
    Welcome to Rivendell, Mr. Anderson.

    Comment

    • Bear
      Super Senior Member
      • Dec 2008
      • 1038

      #3
      Breaking down 21 cutting boards (plus more than a few that were wasted)

      Click image for larger version

Name:	IMG_9588.jpg
Views:	1
Size:	81.2 KB
ID:	865759

      Initial glue-up

      Click image for larger version

Name:	IMG_0075.jpg
Views:	1
Size:	128.8 KB
ID:	865760

      Just before using the PETG jig and the guide bushing to clean out the through-hole for the waveguide.

      Click image for larger version

Name:	IMG_0102.jpg
Views:	2
Size:	125.7 KB
ID:	865761

      Before and after the flush trim bit. The rear panel and the inner baffle are the critical dimensions. Everything else is trimmed.

      Click image for larger version

Name:	IMG_0161.jpg
Views:	1
Size:	125.2 KB
ID:	865762

      Before sanding vs after a comprehensive pass with 80 grit discs on my 5" Random Orbital Sander.
      Click image for larger version

Name:	IMG_0166.jpg
Views:	1
Size:	121.7 KB
ID:	865763
      Last edited by Bear; 16 November 2020, 17:23 Monday.
      Welcome to Rivendell, Mr. Anderson.

      Comment

      • Bear
        Super Senior Member
        • Dec 2008
        • 1038

        #4
        I went through a number of attempts to make routing the waveguide cut-out relatively painless. That never really happened, as the ragged through-hole in the third picture, above, indicates. However, I did finally figure out a process that should work based upon my mistakes. I made a slightly taller jig than what was used in my first several attempts, and then I would use that, a plunge base, a 0.25" spiral upcut bit, and a 1.25" guide bushing, all to cut the through-hole. After that, a 1/2" rabbeting bit will do the driver recess. Oh, yeah, use double-sided tape to secure the jig for the through-hole passes.

        What didn't work:
        • Top-bearing trim or dado bits. Those tended to slip and nick the jig, making it useless. I also burned a couple of bits with only 3mm passes.
        • Using the guide bushing with a fixed base and manually plunging the bit. Yeah, dangerous and less effective.
        • Having a jig that was shorter than the depth of the guide bushing -- not enough support for the router.


        Here are the "final" routing templates for the 4" and 5" versions.
        SB26 Routing Templates.zip
        Welcome to Rivendell, Mr. Anderson.

        Comment

        • Bear
          Super Senior Member
          • Dec 2008
          • 1038

          #5
          More build photos.

          I added an inch to the inner baffle through-hole, and then applied a half-inch roundover.
          Click image for larger version

Name:	IMG_0115.jpg
Views:	1
Size:	77.4 KB
ID:	865765

          The SpeakOn openings in the rear required a 1" spade bit for the through-hole, and then I came back with the 0.5" rabbeting bit for the recess.

          Click image for larger version

Name:	IMG_0116.jpg
Views:	1
Size:	60.2 KB
ID:	865766


          Click image for larger version

Name:	IMG_0117.jpg
Views:	1
Size:	128.8 KB
ID:	865767
          Welcome to Rivendell, Mr. Anderson.

          Comment

          • augerpro
            Super Senior Member
            • Aug 2006
            • 1866

            #6
            Fun stuff Bear! I'll be curious just how the acoustic centers actually line up.
            ~Brandon 8O
            Please donate to my Waveguides for CNC and 3D Printing Project!!
            Please donate to my Monster Box Construction Methods Project!!
            DriverVault
            Soma Sonus

            Comment

            • Bear
              Super Senior Member
              • Dec 2008
              • 1038

              #7
              Originally posted by augerpro
              Fun stuff Bear! I'll be curious just how the acoustic centers actually line up.
              Thanks! I'm hoping that they will be within a millimeter or two, requiring just a tweak to lock-in the phase transition.
              Welcome to Rivendell, Mr. Anderson.

              Comment

              • augerpro
                Super Senior Member
                • Aug 2006
                • 1866

                #8
                AC's are kind of a moving target. One way I like to do it (because I voice with a miniDSP) is shape the responses to match the target response, say LR4, as exact as possible then flip the polarity to get a null and adjust delay to create the deepest null. You need the actual filters in place to accurately judge the AC is how I understand it.
                ~Brandon 8O
                Please donate to my Waveguides for CNC and 3D Printing Project!!
                Please donate to my Monster Box Construction Methods Project!!
                DriverVault
                Soma Sonus

                Comment

                • Bear
                  Super Senior Member
                  • Dec 2008
                  • 1038

                  #9
                  Originally posted by augerpro
                  AC's are kind of a moving target. One way I like to do it (because I voice with a miniDSP) is shape the responses to match the target response, say LR4, as exact as possible then flip the polarity to get a null and adjust delay to create the deepest null. You need the actual filters in place to accurately judge the AC is how I understand it.
                  Click image for larger version

Name:	RSS210 Reverse Null.png
Views:	1
Size:	61.6 KB
ID:	865773

                  With the design study that I did most recently (a traditional center channel layout), I was not only focused on getting a deep reverse null, but also extending the phase alignment as much as possible on either side of the crossover points (highlighted below). I've been able to get very deep nulls in the past, but they always seemed fragile. The phase locked-in at the Fc, but then deviated wildly just a bit on either side of Fc or with slight changes in part values. That could be an artifact of PCD, but getting the "wings" to also line up took a fair bit of work to get it right in this effort. In this one, I've got two octaves of phase agreement between the RSS-210 and the ZA14, and -2/+1 octaves on the ZA14 to SB29RDCN transition. It's not a Duelund, but it's better than I've done previously. I'm aiming to get even better with this design (which is obviously being built!).

                  Click image for larger version

Name:	SPL + Phase - RSS210.png
Views:	1
Size:	71.4 KB
ID:	865774

                  Of course, I used old measurement files from Zaph for the ZA14 and the RDCN (for a 9" wide baffle), and the posted files for the RSS210 from PE (infinite baffle). That means that this phase alignment is purely fictional, but it was useful as a learning exercise for me. Another thing this effort taught me is how critical a padding resistor is to getting a smooth SPL curve for a midrange driver, but that's a different discussion.
                  Welcome to Rivendell, Mr. Anderson.

                  Comment

                  • Evil Twin
                    Super Senior Member
                    • Nov 2004
                    • 1531

                    #10
                    Originally posted by Bear
                    Breaking down 21 cutting boards (plus more than a few that were wasted)

                    [ATTACH=CONFIG]30821[/ATTACH]

                    Initial glue-up

                    [ATTACH=CONFIG]30822[/ATTACH]

                    Just before using the PETG jig and the guide bushing to clean out the through-hole for the waveguide.

                    [ATTACH=CONFIG]30823[/ATTACH]

                    Before and after the flush trim bit. The rear panel and the inner baffle are the critical dimensions. Everything else is trimmed.

                    [ATTACH=CONFIG]30824[/ATTACH]

                    Before sanding vs after a comprehensive pass with 80 grit discs on my 5" Random Orbital Sander.
                    [ATTACH=CONFIG]30825[/ATTACH]

                    I love the smell of bamboo dust in the morning...
                    DFAL
                    Dark Force Acoustic Labs

                    A wholly owned subsidiary of Palpatine Heavy Industries

                    Comment

                    • draki
                      Member
                      • Oct 2012
                      • 37

                      #11
                      [QUOTE... is how critical a padding resistor is to getting a smooth SPL curve for a midrange driver....[/QUOTE]

                      Agreed

                      Comment

                      • Bear
                        Super Senior Member
                        • Dec 2008
                        • 1038

                        #12
                        Bass Alignment: Chaotic Good

                        I don't have anymore shop time until next Wednesday, so I'm in a bit of a holding pattern, unless I'm going to annoy the neighbors (I'm talking to you, Kid-learning-to-play-trumpet!). However, I'll throw out a few graphs from the design files (the originating data is from Unibox using factory T/S parameters). First up is looking at the Driver/Enclosure-limited SPL vs the Xmax-limited SPL.

                        Click image for larger version

Name:	SPL Limit.png
Views:	1
Size:	40.5 KB
ID:	865784

                        A couple of basic things to review here. First, SPL is first and foremost a function of piston volume (cone area * stroke). As frequencies get lower, a driver needs to move more and more air with each stroke in order to maintain the same SPL. At some point, the physical limit of the driver (Xmax or Xmech) becomes the limiting factor. This is the blue line at 45 degrees, and it is basically a -12dB/octave line. The Purifi and Anarchy woofers all generate lower bass from a smaller driver by significantly increasing the stroke.

                        For the driver SPL, the low-end is controlled by the combination of resonance frequency, input power and Qtc of the box. This is the left side of the driver's response curve, and the Qtc determines the slope of the curve (Qtc = 0.5 represents a -6dB/Octave rolloff; Qtc = 0.707 represents -12dB/octave). The modeled driver Qtc is about 0.64 in 9.2L of net volume, heavily stuffed (~1.4lbs per cubic foot; or a bit more than 22g/l). The right side of the driver's response curve is dominated by the inductance (Le) of the voice coil and motor. This acts just like any other low-pass filter.

                        What I'm looking at with these two curves is at their intersection. At what point does the SPL constraint shift from the broader driver parameters to the basic physics of the piston volume? In this case, it's at 108Hz and 106dB with 100W of input power. I need to throw a couple of qualifiers on that statement, since I'm modeling the response with 0.5 ohms of series resistance to give me a DCR "budget" for the crossover, not just with the untouched factory data.

                        The other major qualifier is that the intersection point actually doesn't matter. From what I understand, a typical subwoofer high pass crossover is often an LR4 slope. While the "4" indicates fourth-order, and a -24dB/octave rolloff, the more important design parameter is actually the "LR". That indicates that the input signal will be reduced by -6dB at the crossover frequency (Fc). This is why the red circle is at 100dB and 76Hz (SPLmax - Fc attenuation = SPL @ Fc; really that's an upper bound, but I'm not Vance Dickason).

                        The net-net / TL,DR: an 80Hz subwoofer crossover should be safe for the driver to absorb 100W peaks for short periods without mechanical damage.

                        The long-term power handling is governed by the driver's motor structure's ability to dissipate heat into the box, so don't think that this will play at 105dB all day. I'm looking at headroom here, not steady-state. Really, 105dB is too loud to be near for extended periods. If you want to increase power to 200W, you want to cross at ~100Hz or higher, but do be careful about dumping that much current into a driver that's not designed for it.

                        So, that gets us through the SPL design calculations (for the moment). Let's look at the bass alignment itself, and the step response. So, in addition to the effects of the box alignment on SPL generation, the box alignment also has a few other effects, with Step Response being one of the more interesting:

                        Click image for larger version

Name:	Step Response.png
Views:	1
Size:	28.2 KB
ID:	865785

                        • Qtc = 0.5, 1/sqrt(4) - Critically damped; minimal overshoot
                        • Qtc = 0.577, 1/sqrt(3) - Bessel; minimal delay
                        • Qtc = 0.707, 1/sqrt(2) - Butterworth; flat response


                        There are other alignments available, but I'm not covering those here. What I'm aiming to achieve is something around a Qtc of 0.707 or below, knowing that my realized Qtc is likely to be higher than whatever parameters I'm plugging into Unibox. Unibox's calculation result relies heavily on the effects of the stuffing, and without that stuffing, the Qtc is calculated at 0.767 (too high).

                        While the Butterworth target provides the flattest response, it also provides the fastest settle time (i.e., the time it takes for the driver's suspension and the box's air spring to return the cone to rest). I am sure that the overshoot provides an audible effect, but settle time is something that I can understand from a mechanical engineering perspective, so I'm aiming to minimize settle time. Yeah, I may have just put a stake in the ground for the "fast bass" debate. Let's ignore those last two sentences and move on...
                        Last edited by Bear; 22 November 2020, 11:14 Sunday. Reason: Attachments are corrupted
                        Welcome to Rivendell, Mr. Anderson.

                        Comment

                        • 5th element
                          Supreme Being Moderator
                          • Sep 2009
                          • 1671

                          #13
                          These are coming along nicely and that 3D printed waveguide measures like a champ!
                          What you screamin' for, every five minutes there's a bomb or something. I'm leavin' Bzzzzzzz!
                          5th Element, otherwise known as Matt.
                          Now with website. www.5een.co.uk Still under construction.

                          Comment

                          • Bear
                            Super Senior Member
                            • Dec 2008
                            • 1038

                            #14
                            Wherein I fight back against kid-learning-to-play-trumpet

                            So, there really is a kid learning to play the trumpet nearby. S/he typically starts practicing in the late afternoon, but sometimes it starts much earlier. I'm really not so much of a jerk to really try to retaliate, but let's not have that stop a good story. Mostly I just did not want to wait until Wednesday to make progress, so I grabbed my router and headed onto my balcony for some work.

                            To get things rolling, I flush-trimmed the the third cabinet.

                            Click image for larger version

Name:	IMG_0229.jpg
Views:	1
Size:	132.1 KB
ID:	865786

                            Then, I swapped from the bottom bearing bit to the spiral upcut bit to finish the through-holes on each of the boxes.

                            Click image for larger version

Name:	IMG_0231.jpg
Views:	1
Size:	170.6 KB
ID:	865787

                            My current template is not only too short, but it also is turning out to be too big. I'll fix the gap with some extra gasket tape, but I didn't want to have to use it on the perimeter of the waveguide. The jig posted earlier in the thread includes the extra height, but it is still oversized too much compared with my latest prints from 3D Hubs.

                            Click image for larger version

Name:	IMG_0232.jpg
Views:	1
Size:	68.7 KB
ID:	865788

                            For comparison, here is a Thorin cabinet on top of a Helios tower.

                            Click image for larger version

Name:	IMG_0235.jpg
Views:	1
Size:	53.5 KB
ID:	865789
                            Last edited by Bear; 22 November 2020, 12:49 Sunday.
                            Welcome to Rivendell, Mr. Anderson.

                            Comment

                            Working...
                            😀
                            😂
                            🥰
                            😘
                            🤢
                            😎
                            😞
                            😡
                            👍
                            👎
                            Searching...Please wait.
                            An unexpected error was returned: 'Your submission could not be processed because you have logged in since the previous page was loaded.

                            Please push the back button and reload the previous window.'
                            An unexpected error was returned: 'Your submission could not be processed because the token has expired.

                            Please push the back button and reload the previous window.'
                            An internal error has occurred and the module cannot be displayed.
                            There are no results that meet this criteria.
                            Search Result for "|||"