USM-2 DIY Speakers (ND91-4 + ND16FA-6)

Discussion in 'DIY' started by ultrabike, Nov 26, 2017.

  1. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,962
    Trophy Points:
    113
    Location:
    Irvine CA
    Background

    I've been intrigued by the ND91-4 for quite a while (years). It's a 3 1/2" driver with an Fs of 74 Hz, a Va of 0.05 cubic feet, and a peak Xmax of 25 mm. With a sensitivity of 85.6 dB 2.83V/1m, it is not necessarily the last word of sensitivity. But I think it's competitive in the 3 1/2" full range speaker driver market.

    It's very similar to the Aurasound NS3-193-8A, but lower impedance (more sensitivity) and with a smaller optimal compliance volume. It has higher Xmax as well. In fact, the ND91-4 is used in a variety of relatively high end line array applications (which at the moment do not interest me that much).

    Given the mentioned characteristics, it is possible to make a very small speaker with optimal response down to 70 Hz. This makes it perfect for a small 2.1 or home theater system. Which has haunted me for a while ever since I bought the Mirage nanosats a long time ago. Those are great, but given the design choice to make them fully omni-directional/uber-small, they lack a port and roll off around 200 and something Hz. The best choice I know that compare to what one can do with an ND91-4 are the Energy Classic Take 5's, but those also roll of around 100 something Hz, not quite hitting the < 80 Hz goal. There are actually other somewhat similarish speakers than can go down to 50 Hz, but those are active and use digital cross overs.

    Note: The term cross over is actually a small pet peeve of mine. They are analog equalizers, they do hell of a lot more than "cross over". In fact, one has to tailor things for a particular set of drivers given driver impedance, response of the driver once mounted on the baffle, and all kinds of weirdness, i.e. one equalizes the damned thing.

    Anyhow, I wanted to do this as a hobby. A real f'ing hobby. So I bought some power tools: DeWalt Circular Saw, DeWalt 2 1/4 HP router and some guides. I might make more shit with those as well. I downloaded PSD-Lite and got busy.

    The nice thing about Dayton Audio stuff and PSD-Lite is that one can get FRZ and ZMA files and mock around with the computer to get some cross over. I'm fully aware that there already is an online speaker called the "Helium" using the exact same drivers I'm using. In fact, originally I was going to get the ND20 tweeter. But the Helium encourage me to get the ND16FA-6 instead. I could just buy the Parts-Express 0.05 cubic feet flat pack and just build the Helium:

    http://techtalk.parts-express.com/forum/speaker-project-gallery/64062-helium-a-true-micromonitor

    But I wanted to do this from scratch, as a hobby and to learn. I made my own flat packs and re-did the cross over. When I modeled the Helium cross over, to me it seemed that the results would be a tad V-shapped. And in-fact, somewhere in the thread the designer mentions that he had some trouble with the tweeters, and once that was sorted out, in fact, they did seem a little bright.

    Preliminary Construction

    Here are some pictures of my flat packs. The sides and the top shown glued. Below is front:

    IMG_6167.JPG

    I'm using 0.5" MDF and I'm trying to hit the optimal volume using the following external measurements: 7" (H) X 5" (W) x 5.5" (D). I routed 1/4" on the sides and the back to allow for easy fit. I was a little off when I glued one of the speakers (about 1/16"), but I think I'll let it slide. Below is back:

    IMG_6168.JPG

    Baffle

    I had to redo the baffle quite a few times. The first time because instead of cutting 5", I got 4 15/16" since I did not consider the circular saw teeth's width.

    Then because I did not like the way things look and decided it was time to get a router improve looks and functionality,

    And yet one more time because in an effort to place the tweeter fully flush I made the holes a little to large. These is a press on tweeter and in the end I decided to forget about the flush deal.

    Crossover

    Anyhow, here is the cross over for the woofer:

    xover_woofer.png

    For the tweeter I'm doing this:

    xover_tweeter.png

    The response I expected is this (before the port, with the port things should really extend down to 70 Hz):

    fr_target.PNG

    With this impedance:

    impedance.png

    I was targeting a 1" (Diameter) x 7" (Depth) port for a 70 Hz tuning. This was successful.

    The response I got was this:

    FR.png

    Above is +30, 0 and -30 degrees. I've done 45 degrees and no suck-outs, meaning the sweet spot is wide and graceful.

    Note these are relatively flat down to 70Hz, and have output down to 60 or so Hz! I don't know many speakers about these size that can hit that low w/o DSP or active crossover. The horizontal off-axis performance is very good.

    Here is how the look finished on my old 46" TV:

    IMG_6639_SCALED.JPG

    Making 3 more to complete my 5.1 system :)
     
    Last edited: Oct 2, 2018
  2. fraggler

    fraggler A Happy & Busy Life

    Pyrate
    Joined:
    Oct 1, 2015
    Likes Received:
    5,183
    Trophy Points:
    113
    Location:
    Chicago, IL
    Good stuff! I recently built some similarly sized satellites for my rear channel speakers. They use the ND90-8 full range, though. Basically, the Carmody Sprite in separate, sealed boxes. Sounded like tinny, twangy ass til I stuffed them full of polyfill and added a notch filter. Probably did all sorts of stuff wrong, but they now sound pretty good, and certainly good enough for my purpose.
     
  3. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,962
    Trophy Points:
    113
    Location:
    Irvine CA
    I think the ND91-4 has a break up above 10 kHz. Did you put the notch around there?

    EDIT: 15 kHz, that's where it's at in the ND91-4. Dunno if the ND90 is the same. Will check later.
     
  4. fraggler

    fraggler A Happy & Busy Life

    Pyrate
    Joined:
    Oct 1, 2015
    Likes Received:
    5,183
    Trophy Points:
    113
    Location:
    Chicago, IL
    I just used the filter that Paul Carmody had on his site to see how it sounded.
    https://sites.google.com/site/undefinition/sprite

    I know the filter probably doesn't do exactly what it was supposed to since I have a different box type, but it works. I do plan to make a ported set to compare once I get a chance.
     
  5. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,962
    Trophy Points:
    113
    Location:
    Irvine CA
    LOL! The ND90 does not exhibit the same 15 kHz break up, so it may be better suited for full range applications. They can go lower too, but one would need a larger box. If shooting for a smaller box then the ND91 might be the way to go.

    Paul Carmody brings up really nice points regarding full range operation for the ND90:

    "In the other hand, the 4 Ohm woofer is more efficient--almost to a fault. It goes LOUD, like "I can't hear myself think" loud; but past about halfway on the dial, and you become intimately aware of what Intermodulation Distortion is. But what do you expect when you are asking a driver to not only play bass down into the 40 Hz range (with 4 mm of xmax!), and simultaneously play treble up past 16,000 Hz? So horses for courses, I guess. For most practical applications, especially in smaller rooms, I'd use the 8 Ohm version. But if your application involves larger rooms or crowds of people and you want them to shut the hell up and listen to the music, then the 4 Ohm would do you right."

    For full range (single driver) probably a larger driver will have less intermodulation distortion, at the expense of sweet spot width. I think he is right. If not going super loud, the 8 ohm option is really nice.

    The 8 ohm filter indeed basically makes the system less bright and extend all the way to 15 kHz well behaved. It's a sweet and simple true full range that should work great for home theater as well IMO.

    I really wanted to screw around with 2-way, but this would be an awesome easy 1-way which I may give a shot later. I mean, about 40 Hz to 16 kHz with a simple solution like this is really appealing. The box is not huge either.

    BTW, I think one could do something similar with the Aurasound.
     
    Last edited: Nov 26, 2017
  6. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,962
    Trophy Points:
    113
    Location:
    Irvine CA
    LOL! Man I'm finishing this, but the ND90 full range project is really appealing. Will definitively take a look a this one later.

    Thanks for pointing this out @fraggler. I saw this before, but believe it or not the brilliance of it just hit me right now. Such is life.
     
    Last edited: Nov 26, 2017
  7. Priidik

    Priidik MOT: Estelon

    Pyrate BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    2,257
    Trophy Points:
    113
    Location:
    Estonia
    I see that the resistor network for woofer is to damp down the resonance peak.
    Maybe better sound can be had with larger parallel cap or even parallel LCR (12ohms+0.39mH+10uF).
    At least no loss in sensitivity this way, theoretically tighter bass too.
     
  8. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,962
    Trophy Points:
    113
    Location:
    Irvine CA
    Yup. Problem is number of parts in such a small enclosure. Inductors are sort of biggish. Will give the suggestion a wiggle :)

    BTW, I also gave Paul Carmody's Sprite a sim with PSD-Lite. Sort of a sanity check for things that I should have done before but didn't.

    LOL! That thing is a boom-box for sure. It is tilted a bit and it will indeed give you all the way to 63 Hz. If you change the port to PVC 1.25" and length of 6.8" it can hit 55 Hz and little less boom. But the RL network there will still give you a tilted response. It's not totally crazy though, and I'm sure it will work awesomely. It's perfect for home theater.

    From what I gather, it's hard to get flat because the treble does roll off and so the mids are equalized to make the most of the driver.

    It needs a 4 litter enclosure and one could do maybe 5" wide, 8.5" deep, and 11.5" tall. So it still may work fine. One might get a little peaky around 7 kHz. This is all simulation, but it does seem to jibe well with Paul's impressions.

    BTW, this is another interesting driver for full range:

    https://www.parts-express.com/dayto...point-source-full-range-driver-8-ohm--295-349
     
  9. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,962
    Trophy Points:
    113
    Location:
    Irvine CA
    So @Priidik, tried the suggested mod in the sim world. This indeed reduces the resonances in a similar fashion, giving back about 2 or so dB (I still want the 2 ohm there to reduce boominess). But it comes at an increase of parts. The concern would not be $ obviously. The concern is space inside the box.

    I think a similar mod on the parallel resistor can be applied to the tweeter (or a small reduction on the resistor after the cap), but the same concern applies. Otherwise it's a great suggestion.

    LOL! This is great. If a cross over was just a "cross over" then I would put a Butterworth here and there and done. Which means one could just buy it. Problem is, the speaker impedance and the box are a bitch. It's not like in active analog design where one can isolate things with op-amps or active discretes. That was an eye opener to me (I knew this before, but I guess I forgot or more like didn't think about it in this context).
     
    Last edited: Nov 27, 2017
  10. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,962
    Trophy Points:
    113
    Location:
    Irvine CA
    Will update later with my baffle cutting adventures (and crossover nonsense and stuff). Need to buy more MDF (screwed up a batch that was 4 15/16" instead of a good 5 1/16") and wife was not happy with my dust and noise. Had to vacuum and put Christmas three to make happy again.
     
  11. Priidik

    Priidik MOT: Estelon

    Pyrate BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    2,257
    Trophy Points:
    113
    Location:
    Estonia
    That is why it's good idea to make some room for LCR networks. There is always something the sim won't account for.
    Fortunately you don't need low R coils for this, so they can be tiny.
    Also it think lytic caps do fine there.
    Wow, deja vu. :D
     
  12. Cspirou

    Cspirou They call me Sparky

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    8,319
    Trophy Points:
    113
    Location:
    Northwest France
  13. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,962
    Trophy Points:
    113
    Location:
    Irvine CA
    The port elbow is indeed a must. Originally I was going to go for the optimal 1.25" diameter which meant 11" deep. That meant not one but two elbows (90 degree bend) going from one corner of the box to the other one almost touching the edges of the box. It was not going to work. I'm also scared of large ports that may resonate or something. Just as I'm scared of too small diameter ports due to noise.

    So I'm shooting for the minimal diameter of 1" for 7" deep which may be accomplished with only one elbow.

    I also will go with a larger cut out for the baffle and trim with the router. I bought a set of bits including a 1/2 round edge one a few weeks ago. It's going to be interesting because I need to route with the baffle set in the box so the trim is precise (the box is not perfect). This is harder to do than trim it before installation given the small size of the cabinet. But we will see. If I f**k up, I'll just make another one.

    It's all part of the plan :)
     
  14. Cspirou

    Cspirou They call me Sparky

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    8,319
    Trophy Points:
    113
    Location:
    Northwest France
    You can always do an 11" chimney.

    Also realized that Wood Magazine sounds like a gay porn mag
     
  15. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,962
    Trophy Points:
    113
    Location:
    Irvine CA
    I think even a chimney would not work well with a 11" long port in an enclosure with 6.5" (H) x 4.5" (D) x 4.5" (W) internal dimensions (I'm actually going to do 7" with 1" diameter so that relaxes things a bit). The elbows and the PVC pipe itself is a little thicker than 1.25" which creates all sorts of problems. I was going to do what these guys did:

    http://noaudiophile.com/IK_Multimedia_iLoudMM/

    Unlike noaudiophile, I like front ports. And I don't think their port approach was "port going full retard". That comment made no sense to me. That's what I had in mind given size limitations. I like some of the stuff noaudiophile has done. But I don't think he hit the target in some of his comments.

    In a nut shell, I think a shorter port will get you a bunch of nothing in that 50 Hz frequency vicinity so DSP is not going to work. A longer port like the one used by the iLoud will likely give you 50 Hz. But it will not be flat. Likely it will down slope and have some resonant peak at the corner frequency. DSP will help you there, because there is something to work with in the first place. Below 50 Hz that thing should tank down to nothingness fast. Probably no need for digital filtering, but it doesn't hurt. Maybe they tuned it lower and removed the peak and everything else with DSP.

    I think that speaker is all kinds of awesome. For the size, I don't think you can't really put the power inside. It will compress in the bass if you push it to hard, and it will run out of Xmax at some point to. But that how life is for small speakers. No free lunch. Chuffing would be a concern, but the port is really not that narrow relative to the woofer diameter. Another concern would probably be port resonance given length, but I don't think this speaker is intended for PA applications. In the end noaudiophile did recognize this was pretty awesome. So that's good.

    Like I said before, in the end, I may not need two elbows. Just one with 1" diameter. Front port me thinks. We'll see. Two elbows like the iLoud is still a possibility if needed.

    EDIT: I also love the narrowness of the iLoud cabinet. I would have chosen a smaller tweeter for better off-axis response. But as it is, that thing probably images nicely.
     
    Last edited: Nov 27, 2017
  16. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,962
    Trophy Points:
    113
    Location:
    Irvine CA
    Crap. Forgot to include baffle. LOL, may have to rethink crossover.

    EDIT: Impedance will drop to 3 ohms minimum, but if baffle is included move 2 ohm resistor in ND91-4 crossover in series with the 8 uF cap.

    Like so:

    ND91-4_crossover_change.png

    No changes to tweeter crossover.
     
    Last edited: Nov 28, 2017
  17. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,962
    Trophy Points:
    113
    Location:
    Irvine CA
  18. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,962
    Trophy Points:
    113
    Location:
    Irvine CA
    Updated drawing after making holes for the tweeter. Press-on drivers are a bitch. I used a Dremel 4000 to route the wholes for the tweeter and the woofer (less powerful Dremel does not have enough juice to do this in a reasonable time). Too small for my 2 1/4 HP router. Actually did not want to spend $55 on a Jasper 240 jig (and sort of kicking myself for not doing so - nah, I'm too f'ing cheap, maybe I'' get it later).


    BoxDrawing.png

    Will post pictures of the cut outs in a bit. They were not as awesome as they could have been with the Jasper jig.

    Actual dimensions of my particular cut out mess from caliper:

    Woofer: 2 31/32"
    Tweeter: 1 7/32"

    Did not route port because it's on it's way to me.
     
    Last edited: Dec 3, 2017
  19. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,962
    Trophy Points:
    113
    Location:
    Irvine CA
    Pictures of my mess...

    IMG_6170.JPG

    Need to make the holes for the port (after the port arrives), and then may route the edges for a prettier look.

    Baffles are not glued yet.

    There is some weirdness in the woofer cut out. It turns out the ND91-4 connections protrude the 3" diameter and some massaging was needed there to let the speaker settle down. The speaker was not designed for recessed installation. It mounts on top. Which some folks (like me) don't like that much.
     
  20. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,962
    Trophy Points:
    113
    Location:
    Irvine CA
    LOL! I'm actually thinking of redoing the holes with the deWalt + Jasper. Dammit!

    EDIT: Nah, we move forward, the holes would have been fine if not for that stupid extra work I had to do for the ND91-4 connector that interferes with the 3" hole. It's covered by the driver mounting. Measured the tweeter and the diameter is with in 1/32" as far as I can tell. Baffle edge routing will hopefully bring the goods.
     
    Last edited: Dec 3, 2017

Share This Page