Announcement

Collapse
No announcement yet.

Crossover help for a Newbie

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Crossover help for a Newbie

    Good morning,

    A few days ago I started a build thread on my first design. It's a DC28F in SEOS waveguide matched to a Goldwood 8" poly woofer in a sealed cabinet to use outdoors. With the recent plug for the program and its ability to calculate polar maps, I'm using VituixCad to work on the corssovers. I have measured each driver in box separately with 10 degree increments to 60 degrees off-axis. I blended the woofer response for the direct measurement using the Unibox model because it was late when I did the measurements and forgot to measure the nearfield with the same amplification level.

    I've had pretty good success at manipulating the woofer response. However, I'm having the hardest time getting a flat response out of the tweeter after working on it for two full evenings. The response has a hump around 1.2-1.4k before it drops off quickly (almost identical to the RS28 measurements in the current thread on low tweeter crossover points) as well as a quickly declining response at around 12k, which is a little early to drop off. My best attempts at a relatively flat response result in about 8-10 parts to the tweeter section. That seems a bit high. Are there any tips for flattening a waveguide loaded tweeter response? I'm targeting LR 4 response and have tried everywhere from 1.2k to 2k for a crossover point. My best attempt so far uses two parallel LCR circuits before the crossover with a 2nd order electrical and a resistor on the inductor.

    Any tips would be appreciated. What's the best approach in a scenario like this? Based on a graph of all the SPL curves in REW, it looks like around 1.5k is the best spot for an even power response. I'm willing to move up or down as necessary to make the response work.

    Thanks,

    Brian

  • #2
    Give us SOMEthing!
    Your measured .frd and .zma files would be great.

    Comment


    • #3
      Here are frd and zma files for the on axis measurement. I wantwa sure if there's a general approach usually followed when using waveguides. Thanks for being willing to take a look for me.
      Attached Files

      Comment


      • #4
        Do you have a combined frd for finding the offset?
        John H

        Synergy Horn, SLS-85, BMR-3L, Mini-TL, BR-2, Titan OB, B452, Udique, Vultus, Latus1, Seriatim, Aperivox,Pencil Tower

        Comment


        • #5
          wrt the dc28 a simple 3 order electr gets you flat to 12K which would be good enough for most
          John H

          Synergy Horn, SLS-85, BMR-3L, Mini-TL, BR-2, Titan OB, B452, Udique, Vultus, Latus1, Seriatim, Aperivox,Pencil Tower

          Comment


          • #6
            Do you have any parallel resistance across the DC28? (10 ohms should help).

            Comment


            • #7
              I think you could tilt the response a bit but did not want to get into the whole phase issue as the files did not look like min phase. The point of my post is to say after 12K, it's not important to add parts to boost the response.
              John H

              Synergy Horn, SLS-85, BMR-3L, Mini-TL, BR-2, Titan OB, B452, Udique, Vultus, Latus1, Seriatim, Aperivox,Pencil Tower

              Comment


              • #8
                Originally posted by jhollander View Post
                I think you could tilt the response a bit but did not want to get into the whole phase issue as the files did not look like min phase. The point of my post is to say after 12K, it's not important to add parts to boost the response.
                Thanks. This is probably where I was running into trouble trying to shape the response to extend to about 15-17k before rolling off.

                What do you mean by the files not being minimum phase? They are measurements if frequency with phase.

                Comment


                • #9
                  The DC28 response and phase flat lines after 500 Hz then wraps at 10K so the phase slope doesn't look like minimum phase. If you add the tails and extract you end up with a different phase.
                  John H

                  Synergy Horn, SLS-85, BMR-3L, Mini-TL, BR-2, Titan OB, B452, Udique, Vultus, Latus1, Seriatim, Aperivox,Pencil Tower

                  Comment


                  • #10
                    The measurement stops at 500 because I only did a 500hz-20khz sweep on the measurement. I knew below 500hz wasn't going to add anything significant, and I didn't want to damage the tweeter.

                    Comment


                    • #11
                      I THINK bj's using measured phase (which SHOULD be OK - if modeled w/no offsets).

                      Comment


                      • #12
                        Originally posted by Chris Roemer View Post
                        I THINK bj's using measured phase (which SHOULD be OK - if modeled w/no offsets).
                        It would be if the GW also used measured phase, but it looks like it was spliced and phase extracted. Both files need to be the same, either measured or extracted.
                        John H

                        Synergy Horn, SLS-85, BMR-3L, Mini-TL, BR-2, Titan OB, B452, Udique, Vultus, Latus1, Seriatim, Aperivox,Pencil Tower

                        Comment


                        • #13
                          Sometimes simpler is better. Try a 2nd order filter: 5.6 cap in series, 0.27 coil in parallel with a 4 ohm resister before the crossover. The reference curve is 1500 Hz - LR4 - 88 db. If you change the resistor to match the level, you will need to adjust the cap and coil to suit. A 4.7 cap with an 8 ohm resistor will drop the level approx. 3 db.

                          The attached response curve looks pretty excellent to me!
                          Last edited by Billet; 08-04-2018, 11:22 AM.

                          Comment


                          • #14
                            Originally posted by jhollander View Post

                            It would be if the GW also used measured phase, but it looks like it was spliced and phase extracted. Both files need to be the same, either measured or extracted.
                            The GWs are measured down to 200. That response is merged with the unibox output plus a diffraction and boundary simulation based on where they will be placed on the patio. Is this not how to do it?

                            I'm getting close and should have a model to post tonight. Thanks for the suggestions. Billet's response looks nice. Adding a resistor to bring it down in line with the woofer doesn't result reduce the volume by the same amount at all frequencies. Based on the woofer response, I'm targeting 85-86db.

                            Comment


                            • #15
                              Yes, you are correct, its unibox sim plus baffle, merged or spliced to the high frequency. I've used boundary effect/ correction but it doesn't really improve the blending match unless the patio is where you might have measured them. If you are measuring in 4 pi space and trying to convert to 2 pi my advice would be to measure in 2 pi, gate at a higher frequency, and then blend in only the near field plus baffle.

                              It does looks like you added tails and generated minimum phase on the GW. My previous comment was that you shouldn't mix measured phase with minimum phase.
                              John H

                              Synergy Horn, SLS-85, BMR-3L, Mini-TL, BR-2, Titan OB, B452, Udique, Vultus, Latus1, Seriatim, Aperivox,Pencil Tower

                              Comment

                              Working...
                              X