No announcement yet.

Skylark Flying Towers: Nested Array Speakers in Denovo Cabinet

  • Filter
  • Time
  • Show
Clear All
new posts

  • Skylark Flying Towers: Nested Array Speakers in Denovo Cabinet

    I've been working on a speaker build I'm referring to as the Skylark Flying Towers. The design concept is to create a speaker with significantly reduced floor and ceiling reflections, but wide horizontal dispersion. The vertical beamwidth was kept as wide as possible and as consistent as possible at different frequencies while still drastically limiting the energy in the first vertical reflections, which typically happen at about 30 degrees for the floor, and 40-45 degrees for the ceiling.

    This has been achieved using a “nested array” approach, which nests the higher frequency arrays inside the lower frequency arrays. In this case, the high frequency array consists of a single Fountek Neo X 2.0. The mid frequency array consists of 2 x TC9FD18 full range drivers. The low frequency array consists of 4 x RS150T woofers.

    The design utilizes the Denovo knock down tower speaker cabinet, but requires the speaker to be lifted off the floor to place the tweeter at ear level. Thus the “Flying” towers. It also requires building an enclosure inside to separate the mids from the woofers. The nice thing is that the internal enclosure is completely hidden, so it doesn't have to be made to look nice. The braces also need modification, but all 3 can still be used. All the externals come from the Denovo cabinet and are precision cut.

    Being a “nested” array and using “flying” towers led me to name this after a bird. I decided on the Skylark because BT made a song called Skylarking, and skylarks supposedly have a melodious song.

    This design is intended to be used with a subwoofer (or multiple subwoofers) to fill in the bass below 80 Hz or so. It uses a sealed alignment for easy subwoofer integration. I see this as the ideal way of dealing with bass frequencies in home audio.

    All the drivers used are good value, low distortion drivers, and the result is a 6 ohm speaker (in my estimation) with about 89 or 90 dB sensitivity and full baffle step compensation that should be able to play quite loud. The total cost for a pair is about $750, due in part to the complicated crossover required to achieve the target slopes that produce the consistent vertical response across frequency.

    This is a relatively inexpensive and simplified version of this nested array concept, so some compromises were made. For reference, my previous build was a 4 way with 15 drivers in each one, standing over 5 feet tall. Still, this one is generally effective at achieving the design goals.

    Compared to other arrays, there are many benefits to the nested array concept:

    No comb filtering.

    No EQ required (some arrays require a 3 dB per octave EQ).

    No DSP to insert delays on some drivers (required by many straight arrays.)

    No curved baffles to build.

    You just plug them into a single amplifier channel and they work like any regular speaker.

    This build represents a very simple, inexpensive way to try an array concept with very little down side (unless you like vertical reflections).

    At the moment I only have 1 so I can't comment too much on the sound. (I'd be somewhat reluctant to do so anyway since everyone thinks their design sounds awesome making the subjective assessment rather meaningless). I can say that the measurements are quite nice, and the sound is extremely detailed (which is the point of limiting the vertical reflections).
    Attached Files

  • #2
    It's really the tweeter crossover that got complicated, with the mids and woofers using relatively simple circuits. In many places I used multiple resistors to achieve a target value in order to increase power handling. I also used an air core 2.7 mH inductor to avoid saturation. It was worth it to me to spend a few extra dollars to make sure the crossovers weren't the limiting factor in power handling.

    The woofers are numbered 1,2,3,4 from top to bottom, so the top two woofers are wired in parallel, and the bottom 2 woofers are wired in parallel. Those parallel combinations are then wired in series.

    Note that the mids are wired with reverse polarity compared to the woofers and tweeter (typical of 2nd order crossovers).
    Attached Files


    • #3
      Here's a quick look at the vertical dispersion of the speaker. My apologies for any narrow band blips caused by my laptop fan. This plot shows the frequency response at three vertical angles: 0 degrees (on axis in blue). 15 degrees off axis (in red). 45 degrees off axis (in green). As you can see, out to 15 degrees the response simply shelves down a couple of dB at higher frequencies. That was the design goal, and +/- 15 degrees gives a fairly large (vertically) listening area.
      Beyond 15 degrees the level really starts to attenuate, and you can see that at 45 degrees off axis the response is down at least 10 dB at all frequencies above 450 Hz, but more typically 15 or even 20 dB. That should make for pretty good reduction of ceiling reflections without the need for any treatment (for typical speaker height, listening distance, and ceiling heights).

      Edit: When I moved the mic to capture the different angles, the reflections came in with less delay, and I had to reduce the gating. That makes the bass response here appear to suffer. I only included it to show that I was maintaining the proper relative levels between the measurements (they all line up at low frequencies). The actual response goes deeper than this would indicate.
      Attached Files
      Last edited by benb; 02-23-2020, 01:54 PM.


      • #4
        Here's a look at the horizontal dispersion of the speaker. This shows the same on axis response as the above (in blue). The 45 degree horizontal response is shown in green (some diffraction is evident, but nothing major). The 90 degree horizontal response is shown in red. For best results, these should probably be listened to on-axis (toed to the listening position). I have used a 40 dB dynamic range, which is smaller than a lot of people use, so that needs to be considered when assessing the consistency. The grid lines are only 2 dB apart vertically. I have used gating only, and no smoothing at all. I also had the speakers measured by a more well known speaker designer to verify the calibration, which was spot-on.
        Attached Files


        • #5
          Pretty cool, nice work


          • #6
            Looks great so far. How do you plan to suspend it between floor and ceiling?

            The PEDS 2.1 mini system
            My A7 Project - another small desktop speaker
            The B3 Hybrid Dipole - thread incomplete and outdated


            • #7
              Originally posted by BOBinGA View Post
              Looks great so far. How do you plan to suspend it between floor and ceiling?
              Thanks for your kind words. I've been trying to get the speaker to fly as advertised. It hasn't responded, even when I've yelled "yip yip". It looks like I'm just going to have to put them up on stands. Any sturdy stand about 18-28 inches tall will do. I just weighed one at 46.6 lbs.

              This speaker does have some similarity in form factor to the Status Acoustics 8T, so that might be an option:

              Personally I'd put the subwoofers in whatever location sounds best, and not necessarily right under the satellites (I guess something 34 inches tall with 7 drivers weighing 47 lbs can be a satellite.)


              • #8
                I have a stereo pair of skylarks now (see attached picture). I haven't had time to integrate with my subwoofers yet, but the imaging, detail, and dynamics are really impressive. My first impression when I wired them up and played something was that I had mistakenly left my on-wall towers playing. The sound didn't seem to be coming from the skylark speakers. As it turns out, that appears to be a feature of their sound that is similar to what my on-wall towers can do: they play music, but the music seems to exist on its own with no audible connection to the speakers (unless something is panned full left or right). I wasn't sure if this would work for a speaker off the wall, but it does. They still manage to give a "you are there" type of presentation, providing a convincing localization for anywhere between them and behind them. They are exceeding my already high expectations, and I am perfectly satisfied with the Neo X 2.0 tweeter. I made these with the intention of lending them out to people to audition, in order to show the merits of this type of design (which is part of why they are on tables instead of stands), and I never thought it would be hard to let them go. To quote my wife "we don't need another pair of speakers". We don't, but these are pretty special. Hopefully I get a chance to integrate them with my subwoofers soon.

                Click image for larger version

Name:	IMG_9146 - Crop_small.jpg
Views:	1751
Size:	570.0 KB
ID:	1435981


                • #9
                  The speakers look awesome. I posted a question in the projects gallery, repeating here now. Can you please elaborate on how this design reduces comb filtering? For the mids with crossover is at 3KHz, the wavelength is about 4.5″, mid center spacing is about 8.5″. So they are 2 lamba separated, which should lead to some comb filtering, right? What did you do to reduce the comb filtering?

                  Also do you have a thread or site where you share info about your on-wall speakers? I am interested in building a on-wall setup, due to WAF.



                  • #10
                    Originally posted by Brij View Post
                    The speakers look awesome. I posted a question in the projects gallery, repeating here now. Can you please elaborate on how this design reduces comb filtering? For the mids with crossover is at 3KHz, the wavelength is about 4.5″, mid center spacing is about 8.5″. So they are 2 lamba separated, which should lead to some comb filtering, right? What did you do to reduce the comb filtering?

                    Also do you have a thread or site where you share info about your on-wall speakers? I am interested in building a on-wall setup, due to WAF.

                    I answered your question in the project gallery. It took a while to post but it's there now. I figured I'd add some more information over here. The separation is a little less than you calculated (closer to 7.75 inches). Also, the tweeter is responsible for half the output at the crossover to the mids, while each mid is only generating 25%. That makes for a little shaded array [0.25 0.5 0.25]. Shading widens the mainlobe and reduces sidelobes. At extreme angles where sidelobing can and does occur, things get a little more complicated because the directivity of the drivers themselves comes into play. The mids and tweeter exhibit some vertical directionality at 3 kHz, even if they were used alone and not in an array.

                    I generated a simulation of the vertical polar response as part of the design process. It's not as useful as actual measurements, which I may be persuaded to take more of, especially since I realized I can set this speaker on its side. The simulation attempts to take into account the phase and magnitude of the filtered response coming out of each speaker, as well as the directionality of the speaker drivers and the diffraction of the baffle. I have attached the simulated response to this post.

                    Regarding the on-wall speakers, I have not posted the design information, and I don't really intend to. They are incredibly complicated. I am working on an on-wall variant of this simplified version. This would again be a WWMTMWW. I'm intending to use RS125 woofers along with TC9FD18 mids and the Aurum Cantus 2560.
                    Attached Files


                    • #11
                      Thanks for sharing the information. I am looking forward to your on-wall project. I have been thinking of on-wall speaker build for sometime, just have not pulled the trigger yet. I am leaning towards modifying Zaph's ZDT3.5 into on-wall setup.


                      • #12
                        I have made a version of the skylark speaker with an altered crossover, for a more polite voicing that tilts down a bit at high frequencies. This allows me to listen to music that isn't necessarily recorded perfectly, without causing fatigue. There's still tremendous clarity. If there's any interest, I will post this crossover as well. Alternatively, similar results are achievable with the original crossover and a bit of EQ.

                        I also took some far off-axis measurements (vertically). This was actually pretty complicated to get right, since I measure in my room. The measurement software looks for an impulse, and at extreme angles, there's so much attenuation of the direct sound that the reflections are more impulsive... causing the software to center on a reflection. I eventually caught on to what was happening and made the proper adjustments.

                        Compared to my simulation, the measured far off-axis angles of 60 and 90 degrees are a little different. The lobe near 3 kHz isn't as pronounced as in the simulation (which is a good thing), but there is a lobe at 60 degrees that comes within 5 dB of the direct sound at about 1600 or 1700 Hz (I'd prefer that were lower). So it's kind of a wash.

                        The first file shows the response at 0 (blue), 15 (red), and 45 (green).
                        The second files shows the response at 0 (blue), 60 (red), and 90 (green).
                        Attached Files