Announcement

Collapse
No announcement yet.

X over help

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

  • #16
    Modeling Xovers is like data entering about sex, generally modeling is used to come up with a rough starting point for multiple designs.
    Those designs get tweaked with listening and measurements until one design sounds/measures the best and "wins".
    The winning design then gets tweaked further after many hours of listening and measuring and more tweaking!
    (and often the finished product barely resembles the original model)

    I have designed many very successful speaker systems without ever modeling anything. As a matter of fact I once entered
    my most successful design into a well know crossover design modeling program for fun and it was so wrong I had to
    get a second opinion from a friend using a different modeling program, and the outcome was also wrong in a different way!

    This is not to say modeling has no useful purpose, but I will say measurement equipment and empirical science is far more
    important. Also don't be discouraged from using online calculators where they are very useful in regard to cap/inductor
    ratios and knee shapes. Even though the impedance of a driver is not ruler flat you can look at the impedance at the
    desired filter frequency as a starting point. If you do play around with online calculators keep in mind things like baffle
    size, impedance changes, frequency response nonlinearities make textbook filters i.e. both drivers in a 2-way being
    crossed over at the same frequency, often you will need to "spread" the low and high pass sections to compensate.
    Forums can be a positive haven for the hungry beginner but take advice with
    a grain of salt and never get discouraged!
    Last edited by xmax; 10-10-2017, 11:01 PM.

    Comment


    • #17
      OOPs (I think I mixed up this post w/another one - sorry). Look at isaeagle's XO pic (about 10 posts back) - from THAT, make these alterations:
      Tweeter circuit: make 6.8uF cap a 15uF, make 0.50mH coil a 0.30mH. On L-pad: change 5n (ohm) resistor to 1.2n, and 10n to 5.6n.
      For YOUR HiVi woofer: use 2.5mH coil (as on pic), make tiny 0.33uF cap a slightly less tiny 0.68uF cap. The shunt leg (to ground) should just be a 4uF cap (no resistor).

      OR (from YOUR schematic - which you apparently did not model - but I did):
      LP (HiVi): Drop the coil down to 2.5mH (use a low DCR "iron-core" - I modeled w/a DCR of 0.4ohms), and the cap to 4uF. This lifts the woofer XO up so much that a "notch" (tank) filter gets to be helpful. All that needs to be is a (tiny) 0.68uF cap across (in parallel with) the coil.
      HP (XT25): raised cap up to 15uF, and dropped coil to 0.30mH. This changes the shape of the "knee" (or shoulder) of the rolloff. I've effectively raised your Fc from 1.6k to 1.8k - but I'm not sure this tweeter can cross that low (regardless of a rather low Fs). L-pad seems good.

      Parts you could use (if you like):
      2 coils:
      0.30mH 257-028
      2.5mH 257-556 (Dayton iron-core)
      3 caps:
      15uF 027-119 (Audyn) - you COULD use a cheap npe cap for this larger value > 027-344 ($ave about $8 ??)
      4uF 027-421 (Dayton)
      0.68uF (tiny) 027-408 (Dayton)
      2 resistors:
      1.3 ohm (use 1.2n 016-1.2)
      5.8 ohm (use 5.6n 016-5.6)

      Comment


      • #18
        Originally posted by xmax View Post
        Modeling Xovers is like data entering about sex, generally modeling is used to come up with a rough starting point for multiple designs.
        Those designs get tweaked with listening and measurements until one design sounds/measures the best and "wins".
        The winning design then gets tweaked further after many hours of listening and measuring and more tweaking!
        (and often the finished product barely resembles the original model)

        I have designed many very successful speaker systems without ever modeling anything. As a matter of fact I once entered
        my most successful design into a well know crossover design modeling program for fun and it was so wrong I had to
        get a second opinion from a friend using a different modeling program, and the outcome was also wrong in a different way!

        This is not to say modeling has no useful purpose, but I will say measurement equipment and empirical science is far more
        important. Also don't be discouraged from using online calculators where they are very useful in regard to cap/inductor
        ratios and knee shapes. Even though the impedance of a driver is not ruler flat you can look at the impedance at the
        desired filter frequency as a starting point. If you do play around with online calculators keep in mind things like baffle
        size, impedance changes, frequency response nonlinearities make textbook filters i.e. both drivers in a 2-way being
        crossed over at the same frequency, often you will need to "spread" the low and high pass sections to compensate.
        Forums can be a positive haven for the hungry beginner but take advice with
        a grain of salt and never get discouraged!
        Then I would say that either A) The modeling programs were not used properly or B) said programs were just plain junk.

        From known experience, and that of many others, a good modeling program, like Jeff Bagby's, WinPCD, Xsim, is invaluable to a good design. It is much more than just a base to start with. My Sims are by and large spot on to the measurements. There have been plenty of builds based solely on good sim work that performed not only as modeled, but were exceptional as well.
        https://www.facebook.com/Mosaic-Audi...7373763888294/

        Comment


        • #19
          Jeff Bagby's software is a work of art as far as I am concerned. It just is not enough for what is expected from my designs,
          ​hence all the empirical science.

          Comment


          • #20
            Hello there.

            Chris many thanks for your input.

            From the tables I can find online to keep the x over at Around the 2.2k mark I should be using 10uf and 0.5mh on the tweeter and 5.5uf and 2.75mh on the woofer for 2nd order 12db.

            I added you filter in to the LP and came up with this. Do you think this might be a better solution? It seems that the topography of this x over is very popular so may be I could buy the different value caps and coils and experiment? Did you model the FR/Phase graphs? I would be interested to see.

            Many thanks as always

            P

            Attached Files

            Comment


            • #21
              Paul, forget those on line calculators. They are essentially useless. There are simply too many variables in impedance, freq response, phase, etc that they do not account for.
              https://www.facebook.com/Mosaic-Audi...7373763888294/

              Comment


              • #22
                Originally posted by isaeagle4031 View Post

                Then I would say that either A) The modeling programs were not used properly or B) said programs were just plain junk.

                From known experience, and that of many others, a good modeling program, like Jeff Bagby's, WinPCD, Xsim, is invaluable to a good design. It is much more than just a base to start with. My Sims are by and large spot on to the measurements. There have been plenty of builds based solely on good sim work that performed not only as modeled, but were exceptional as well.
                Diffraction

                Power response

                Impedance /back emf of the amplifier

                DA of capacitors

                Litz, foil, P-core, laminate, skin effect of chokes

                Boundary effects of a mixing consoles etc.

                Complex acoustical and electrical phase relations between driver and electrical phase of filter.

                this list goes on and on.

                how simulators are not real good with certain aspects. They are still fun to play with.
                I would love to hear the most celebrated speaker designed with sims only.

                Comment


                • #23
                  Difraction can be modeled, DA of caps is largely irrelevant, as is the skin effect of a decent inductor, boundary effects of the console are in the mix so also irrelevant, acoustical and electrical phase are part of any good design and are part of the better simulation software.

                  I know of a few designers that have built speakers purely from sim and the measurements taken after were dead on to the sun. On a couple of my own, I've taken in box measurements, simmed the crossover and implemented it. Results were so close that any deviation was simply not audible in anyway.
                  https://www.facebook.com/Mosaic-Audi...7373763888294/

                  Comment


                  • #24
                    Originally posted by isaeagle4031 View Post
                    Difraction can be modeled, DA of caps is largely irrelevant, as is the skin effect of a decent inductor, boundary effects of the console are in the mix so also irrelevant, acoustical and electrical phase are part of any good design and are part of the better simulation software.

                    I know of a few designers that have built speakers purely from sim and the measurements taken after were dead on to the sun. On a couple of my own, I've taken in box measurements, simmed the crossover and implemented it. Results were so close that any deviation was simply not audible in anyway.

                    People might still like to design a speaker with less mixing desk interaction.
                    Diffraction modeling is not accurate.
                    DA of caps can make or break the sound of high component designs
                    Chokes not so much.
                    Phase predictions are ok. but overall with lots or parts and drivers and their "time alignment"
                    The sims are wack.
                    Don't forget about back emf from the speaker motors and amp interactions...

                    Comment


                    • #25
                      Hi issaeagle I hear you but don't have an alternative at the mo. I'm touring and won't be home til mid Nov then I can measure FR curves etc but I need a place to start.

                      Chris seemed to suggest his values had changes the x over point and was unsure the tweeter could be crossed so low. I was attempting to push the crossover point to around 2/2.2k.

                      If you think that's a better place?

                      In any thing I've built I've always had a plan and then adapted along the way to get things sounding/working as I like.I don't imagine a X over will be any different. I'm on expecting perfection on 1st attempt but would like to get as close.

                      Thanks again for your input all.

                      P

                      Comment


                      • #26
                        Originally posted by xmax View Post


                        People might still like to design a speaker with less mixing desk interaction.
                        Diffraction modeling is not accurate.
                        DA of caps can make or break the sound of high component designs
                        Chokes not so much.
                        Phase predictions are ok. but overall with lots or parts and drivers and their "time alignment"
                        The sims are wack.
                        Don't forget about back emf from the speaker motors and amp interactions...
                        Difraction modeling can be very accurate. Good measurements in=accuracy and that is where many fail. It's not the fault of sim, but the user input.
                        I will simply agree to disagree with the da on a cap, same as an inductor.
                        Not quite sure of your quotation on time alignment is meant to be. With proper acoustic center measurements, phase alignment is really not hard to achieve and again is very accurate.
                        I also don't believe that back emf is nearly as important as you are wanting to make it out to be. It's minute in contrast and unless it's a very poorly designed amp and speaker, not audible.
                        https://www.facebook.com/Mosaic-Audi...7373763888294/

                        Comment


                        • #27
                          Originally posted by paulspencer16 View Post
                          Hi issaeagle I hear you but don't have an alternative at the mo. I'm touring and won't be home til mid Nov then I can measure FR curves etc but I need a place to start.

                          Chris seemed to suggest his values had changes the x over point and was unsure the tweeter could be crossed so low. I was attempting to push the crossover point to around 2/2.2k.

                          If you think that's a better place?

                          In any thing I've built I've always had a plan and then adapted along the way to get things sounding/working as I like.I don't imagine a X over will be any different. I'm on expecting perfection on 1st attempt but would like to get as close.

                          Thanks again for your input all.

                          P
                          Though Xmax will vehemently disagree with me, you can use a program such as spl trace to get the Frd and zma files. I've heard some very good speakers that were built solely off Sims that were incredibly accurate once measured.
                          https://www.facebook.com/Mosaic-Audi...7373763888294/

                          Comment


                          • #28
                            I've had simulation software give me really bad results before. Then I realized I gave the software really bad inputs :D Sent from my SGH-I337M using Tapatalk
                            https://www.youtube.com/channel/UCm2...oSKdB448TTVEnQ

                            Comment


                            • #29
                              Again I think it is a valuable tool, I also think people should not be afraid to "free ball" it
                              once in a while, they could learn a thing or 3. And for beginners to not be discouraged from
                              diving into the hobby with some books and measurement gear (mic, interface, software)
                              I'm guessing some beginners are overwhelmed with the idea such complicated software.

                              Also my random thoughts of where simulations might fail were not the best examples,
                              perhaps I should have said take 2 different sims that look great on paper, but one
                              sounds way better and different than the other...

                              Comment


                              • #30
                                Hey there

                                So I think I'm gonna wait til I'm back home and set up some active X over/ eq networks from Pro Tools and tweak them til I get something I like the sound of . Measure the FR and phase and then buy some inductors, caps, resistors etc and try to recreate the sound.

                                Chris I'm still interested in your modelled response if you have any graphs etc.

                                Thanks for all your help.

                                P

                                Comment

                                Working...
                                X