DIY speaker troubleshooting

Discussion in 'DIY' started by JustAnotherRando, Jul 29, 2019.

  1. JustAnotherRando

    JustAnotherRando My other bike is a Ferrari

    Contributor
    Joined:
    Aug 15, 2017
    Likes Received:
    1,011
    Dislikes Received:
    2
    Trophy Points:
    93
    Location:
    Hong Kong
    My second DIY speaker attempt (after an Overnight Sensations kit) has been causing me some confusion.

    I've been using a Mark Audio driver that only became available in the past month- the Pluvia 7.2 HD, which is a replacement for the Alpair 7.3.

    Reasons for selecting these drivers were:
    • Full range to avoid thinking about crossovers
    • Availability of established plans online
    • Mark Audio is located locally (ironically, ordering them was very complicated)
    Plans are located on https://www.markaudio.com/Plans/itemlist/category/22-Alpair-7-Gold. So far I'm building enclosures out of foam core board, before deciding on a design for wood. Stuffing follows the recommend amount of polyfill plus some additional damping behind the drivers.

    My chain is RPi + SPDIF HAT -> Modi Mulitbit -> Loki -> Sys (for volume control) -> Akitika GT-102 -> Speakers.
    Measurements were done with a MiniDSP u-mik and REW, with measurements gated at 4.5ms (so anything under 200Hz should be ignored). 1/12 octave smoothing has been applied.

    The initial 8L bass reflex box which I tried sounded so anemic that I moved on to the Pensil MLTL design which sounds a lot better, but has too much treble. The treble reminds me of listening to an HD800S, and I can solve it exactly the same way- using a Loki with the treble knob turned down to the 10 o'clock position.

    The weird thing is, I tried measuring both the normal and the equalised output, and the graphs look practically the same. The same thing happened with the HD800S when I measured on the EARS- the sound is different, but the graphs are virtually indistinguishable. Green line is no EQ, red is EQ. So for starters, I am not quite sure what's going on here.

    [​IMG]

    Just to check my own sanity, I did an extreme EQ test with the treble knob turned all the way down, just to confirm that the Loki really was doing something. It's the orange line below:

    [​IMG]

    Can anyone suggest what to do about the treble? I'd prefer not to remain dependent on external or software EQ, so I'm wondering if modifications can be made to the enclosure design, to the electronics (perhaps figuring out a notch filter around that massive 13KHz peak?) or whether I have to look at different drivers entirely.

    Thanks.

    Edit: Any other advice is also welcome, this has been a a fun learning experience so far.
     
    Last edited: Jul 29, 2019
  2. dmckean44

    dmckean44 In a Sherwood S6040CP relationship

    Friend
    Joined:
    Nov 17, 2016
    Likes Received:
    1,209
    Dislikes Received:
    7
    Trophy Points:
    93
    Location:
    San Diego
    The Loki appears to be doing it's job. Setting to 8khz knob to 10'oclock will only lower the signal by 1 db or two. Lowering it all the way should only bring the signal down 12db.
     
  3. JustAnotherRando

    JustAnotherRando My other bike is a Ferrari

    Contributor
    Joined:
    Aug 15, 2017
    Likes Received:
    1,011
    Dislikes Received:
    2
    Trophy Points:
    93
    Location:
    Hong Kong
    Yeah, I was just expecting to see something more dramatic on the graph that separates "ouch" from "this sounds okay".

    Overall, I am still less than enamoured of the results that I'm getting. Imaging and staging is really impressive to this speaker noob, but overall tonality even with EQ via the Loki isn't as pleasing as the previous OSMT kit build. I guess this is a good excuse to start looking into the electronics side of things with filtering, but I suspect I won't keep these drivers long.
     
  4. peef

    peef Acquaintance

    Joined:
    Oct 5, 2015
    Likes Received:
    92
    Dislikes Received:
    0
    Trophy Points:
    18
    The peak shows up in the datasheet, too, so it is likely the driver's main breakup mode and not something to do with the enclosure. The easiest fix is probably to give them a listen off-axis as they're probably quite directional above a few kHz.

    Since you're running REW, I'd also peak at the HD/CSD plots for anything weird around 3-4k. Usually, you'll see a H3 peak at 1/3 of the main breakup frequency, and that might be more objectionable than the 13k peak itself.

    If you'd like to pursue the notch route, note that efficiency will take a hit. There are some interesting comments about the use of notch filters in Zaph's L18/27TBFC article-- it will probably involve trial and error. You might also want to experiment with a small resistor (<2R) in series with the driver to reduce Qes.
     
    JustAnotherRando likes this.
  5. JustAnotherRando

    JustAnotherRando My other bike is a Ferrari

    Contributor
    Joined:
    Aug 15, 2017
    Likes Received:
    1,011
    Dislikes Received:
    2
    Trophy Points:
    93
    Location:
    Hong Kong
    Interestingly, I've been spending the evening listening to them about 20 degrees off axis. The treble problems pretty much go away, but at the same time imaging takes a big hit. I wasn't expecting this secondary effect at all.
     

Share This Page