No announcement yet.

Fully active edition: Quarks & Voxel

  • Filter
  • Time
  • Show
Clear All
new posts

  • Fully active edition: Quarks & Voxel

    Dear Forum

    We are writing you from the beautiful lake of Zurich. This project started 15 months ago. Our goal is to build a 2.1 loudspeaker set based on the famous Quarks and Voxel speakers. Instead of using traditional discrete passive crossover networks, we decided to use the ADAU1701 DSP instead.

    What we achieved so far:
    - We've built the speakers from scratch
    - Installed the electronic components into a nice enclosure and connected them

    Now we are trying to find the "ideal" setting for the configuration in Sigma Studio and are overwhelmed by all the possibilities. We would be very grateful for some pointers in the following areas:
    - Which types of crossover-networks to use (2 vs 3 way)
    - Which filters (Bessel vs Butterworth), gain setting, filter steepness and crossover frequencies (for the Quarks and the Voxels)
    - Would you recommend to use an equalizer? If so where and how?
    - Any other helpful suggestions?

    Whenever we switch on/off the PSU, the loudspeakers make a loud and ugly sucking / popping noise. How can we get rid of that?

    Thank you all for your much-appreciated feedback & kind Regards

  • #2
    Nice build so far.

    The original quarks used a 1st order LP on the Fountek and a 2nd order HP on the Dayton Tweeter. Given the expertise of the designer and the relatively flat response of his model, I start by trying to mimic the original passive XO design.

    For the "Pop", I'd start by determining if it's the amp(s) or the DSP, though it's likely the amps. These amp have inputs for mute and standby functions. Does the DSP module have any I/O that can be programmed to hold the amp in mute/standby for a time until the amp settles down after power up?


    • #3
      Hi Mike

      Thank you for your feedback. Do you have an suggestions for the following settings. The software allows to select the following types of filters:

      - Linkwitz-Riley
      - Butterworth
      - Bessel

      Does one sound better than the other in this setup?


      • #4
        Lake side labs.


        • #5
          If my crude reverse calculation of the XOs is in the ballpark, the FE83 has a 1st order LP at ~1275 Hz. That would be Butterworth by default. The ND20FA has a 2nd order HP circa 6900 Hz - not sure the slope on that (LR2 BW, etc.) Does that make sense? It would if the roll off of the woofer is gradual enough and includes BSC before it meets the tweeter. The only way to be sure would be to sim the existing design using the drivers' FR and impedance graphs. One thing for sure, you don't want to HP the ND20FA anywhere below 4K - 5K Hz. It'll blow.


          • #6
            Lakesidelabs, I know this is not specific to the Quarks/Voxel part of your thread, but what ADAU1701 DSP board are you using? Is the screenshot in post #3 Sigma Studio? How easy/hard is it to work with?


            • #7
              Hi a4eaduio

              We use the following board:

              I would say that it requires quite some time and dedication get get it running. Get the original Sure In-Circuit Programmer and the Interface Extension Kit. That will make things a bit easier. We fiddled around with the Free USBi but its not worth the hassle. Also be careful about the pin-layout of the ADAU1701 and double check the printing on the circuit board. The schematic on the Sure website does not always match the configuration on the board! The installation of the drivers for the programmer is a bit of a pain in the neck, because they are unsigned drivers.

              We found this site to be very helpful, but it is in German:

              Sigma Studio is not too complicated to use. The challenge is rather the almost endless functionality of the available filters. High, low, equalizer, gain, bass boost ... Within each of them you'll find a dazzling array of settings. Our initial idea was to build a crossover network using Sigma Studio, which would be superior to a passive one. While this is definitely possible in theory, the practical implementation requires a lot of theoretical know-how and testing.

              Nevertheless .. it has been great fun and we certainty were able to extend our knowledge while drinking non-alcoholic beer and talking s*** to each other! :-)
              Last edited by lakesidelabs; 07-02-2018, 05:06 AM.