No announcement yet.

Need Help with Scott Sehlin's Lithium Build

  • Filter
  • Time
  • Show
Clear All
new posts

  • Need Help with Scott Sehlin's Lithium Build

    (First: I'm new to speaker building/design)

    I'm in the midst of building Scott's Lithium Speakers.

    It's a small speaker and the way I'm currently building the cabinets, I'll only get one try to build the crossover before its permanently glued/fixed in the cabinet.

    So I built the cabinets, double taped the baffle to the enclosure, fished the speaker wires through the port so I can attempt some measurements and external crossover build/test.

    Generally, my response is similar to Scott's measurements (included below) from say 300hz at up.


    Click image for larger version  Name:	Scott's Measurements.JPG Views:	2 Size:	87.8 KB ID:	1400224

    However, I have a MASSIVE bump in my response from 300hz down - and I can't for the life of me figured out why.
    Click image for larger version

Name:	image_77447.jpg
Views:	1
Size:	166.2 KB
ID:	1400225

    Assuming I had messed up in my measurements, I've tried doing the following:
    1) Tooks ND105-4 measurements from Parts Express Website
    - The measurements show a rising low-end response
    2) I modeled low-end response using Jeff Bagby's Frequency Response Modeler
    - The Response Modeler indicated a 1db bump between 100hz to 200hz (due to a small cabinet)
    3) I merged 1 and 2 above and adjusted for diffraction (again using the spreadsheet above).
    4) I then simulated Scott's crossover with the above measurements and - surprisingly - the simulation exactly matched my measured response.

    My next step was to see what the transfer function looks like - I've included that as well.

    Click image for larger version  Name:	Transfer Function.JPG Views:	1 Size:	128.6 KB ID:	1400222
    Click image for larger version

Name:	image_77451.jpg
Views:	1
Size:	151.3 KB
ID:	1400223
    At first, I thought maybe I have a defective woofer - but I compared my measurements with the modeled response using the manufacturer's files and Mr. Bagby's spreadsheet - and the response in reasonably close.

    The comparison included below (blue = mine, red = manu) :
    Click image for larger version  Name:	My Measurements vs Manufracturer's (Diffraction Adjusted).JPG Views:	1 Size:	154.4 KB ID:	1400221

    I know I must be doing something very simple wrong. Any help would be much appreciated.

  • #2
    Happy to upload my measurements files if anyone needs them.


    • #3
      Is your measurement "gated" (doesn't seem like it)?
      Are you measuring "near field", or from at least 20" - 40" away.
      If you measure "far field", with these on stands, away from walls, the rise should go away (if it's from BSC).
      Your XO is most likely OK.


      • #4
        Chris, Measurements were from Far field - gated. I then merged near field response and adjusted for baffle step. When playing music - I can hear the bump (maybe because it’s in my head now from looking at the measurements - but also because it’s 6 dB bump). Per the transfer function - it appears there is nearly 12db attenuation between 150hz and 1,100Hz. Assuming the driver is nearly flat (based on mine and manufacturer’s measurement) between this range - does this amount of attenuation seem reasonable? Sent from my iPhone using Tapatalk


        • #5
          Did you (box) model YOUR woofers' (actual T/S parms) in the Lith cab? You could probably make the vent longer to reduce a hump.
          You need a "few" dB just to flatten the "rise" (from 600 / 1200 or so), then 6dB more (if full BSC), then another -6down at the 2.5k Fc.
          So... doesn't really seem TOO out of line. Like you say, your ears are now possibly over-sensitive to the sim your eyes have seen !

          I'm not real great at doing the 300Hz near-far "merge" myself (when a vent is involved), but I was thinking... plug the port and measure the woofer near, then again w/the port open and compare the box model of closed and vented (with your actual box volume) to help guide you in your port output "splice" at 300 (or whatnot).


          • #6
            After seeing the Nano Neo tread that was just bumped to the first page, I'm revisiting my posts above.

            The Nano Neo use the same mid-woofer and about the same baffle with.

            However, the series inductors utilized in the Nano Neos started off with 0.45mH +/- , and a later version of the crossover utilized 0.85mh+/-

            The Lithiums use s 2.0mH.

            That is a massive difference between the two. I know something boil down to designer's preferred voicing, but the resulting difference in voicing is quite dramatic considering there is nearly 6db different by the time you get to 1,000Hz.

            Again - I'm only asking these questions because I get only one attempt at the crossover before they are permanent glued inside the speakers.


            • #7
              I really don't think you're doing anything particularly wrong here.

              My VERY 1st XO (on a VERY tight "target budget") only used 5 parts. I only had money for 5w resistors, npe caps, and the biggest coil I could buy (on budget) was only 0.44mH, and (being a VERY new designer back then) while I KNEW about baffle-step, that first XO had very little (which isn't that big of a deal if they're not too far out from a wall).

              Later on (after getting them measured), I upped the coil to 0.80mH, giving maybe 3-4 dB of BSC. I used a 2.0mH coil on my MTM (8ohm) version. That would have been equivalent (roughly) to a 1.0mH on the TM model.

              When I model the Lithium, it looks like Scott used (at least) a full 6dBSC (also, using MY files) it looks to ME like a bit of a "smiley-face" curve was used, w/the mid range slightly relaxed, and the top end a little bright. (His FR plots don't show that however.)

              There are other things at play here as well:
              The ND series have had (IMO) a pretty good change to their published T/S parms over time (I allude to this in my orig. post, including a drop in "Q" of maybe 0.20? over a few months time - although that MIGHT have been due to initial prototype drivers VS the actual production run?). Since PE/Dayton came out w/the Omnimic, it SEEMs like the specs get "updated" every other yr or so (not only the T/S parms, but also (now) .frd and .zma files).
              The drivers I measured myself in 2011 showed Qts=.49, Fs=63, and Vas=0.13. One software's "optimum" box shows an F3 near 54Hz in 0.20 cf.
              My last WinISD (box modeling) data (2015) for the ND105-4 has Qts=0.61, Fs=79, and Vas=0.07. The same software gives these parms an F3 of 58/0.16cf.
              It's highly possible that BOTH of those parms sets could yield an F3 around 56Hz in 0.18cf, so...

              Also, when doing your own measurements (given an 8' cubed room w/the tweeter/mic axis 4' from any boundary) a 3.3msec "gate" (should be good to 4' out) is only accurate down to about 300Hz. To get measurements below that, you either go w/the box model, OR take near-field measurements of the woofer (like, < 1/4" from dustcap) and try to splice that data (near 300Hz) to end up w/a full-range measurement. A ported box adds to the mess since near-filed on the woofer's dustcap pretty much drowns out any vent output. You can also measure the vent near-field, but you're left w/the problem of scaling all these measurements (taken at various distances) to try to arrive at a decent overall curve.

              My (orig) box was twice the volume of the Lithium (meaning you'd eXpect more of a "hump" before rolloff in the smaller box). From the 120-160 range down to about 50Hz, my design dropped about -2dB. The Lithium looks more like -4dB (or more?).

              ALL these box/XO designs still have max port output around 40Hz or so, so THAT shouldn't really effect your hump between 100 & 200 Hz. Measuring near-field (woofer only) SHOULD give you an accurate look at how the bottom end (at least in your "hump" range) is shaping up. You'd need a gate around 10msec to get down to 100Hz. You'll still have room (boundary) interactions, but w/the mic so close to the woofer, reflections should get swamped out. Good luck.