Announcement

Collapse
No announcement yet.

I could really use some help designing 3 way Crossovers

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

  • I could really use some help designing 3 way Crossovers

    Several months ago I picked up a pair of Rectilinear III Highboy speakers that had already been heavily modified. I saw that the baffle was easily removed and the replacement drivers were in good condition so I figured it would be a fun project to rework them into something new. I have zero experience when it comes to designing and building speakers, but I'm OK with wood and metal, so I figured it would be a great learning experience.

    I really wanted to make a pair of 3-way speakers, mostly because I used to own a pair of JBL 4312a Control Monitors that I bought back in 1985 and I just loved how they sounded. I never should have sold them years ago...

    So I start my project by looking on the internet, and come to realize just how much goes into designing a good crossover. I download WinISD and XSim and really started getting confused. Eventually I come to the realization I'm never going to get started due to newbie analysis paralysis. Using the old adage, when in doubt, do something, I went ahead and built the speakers without a crossover design. I re-used the woofers that came with the speakers and added mids and tweeters that I hope compliments them. Everything is 8 ohm and similar sensitivity 89-90 dB. The Woofer/Mid has more than 5 octaves of crossover and the mid/tweeter has 2 octaves of crossover.

    I ported the speakers with a calculated tuning frequency of 35 Hz. I used a 3" flared port tube, which might be a bit undersized, but the original Rectilinear III speakers had unflared ports that were just under 3" and I couldn't find anyone online complain about them chuffing. Worse case scenario, I can always stuff some socks in the port tubes.

    I calculated the internal speaker volume at 2.4 cu ft or 68 liters after accounting for drivers, bracing and midrange enclosure. The mids are isolated in some very thick plastic flower pots I found on that South American river sight.

    Drivers used are:
    Dayton Audio DC300-8 12" Classic Woofer. 80 Watts RMS
    Dayton Audio DC130B-8 5-1/4" Classic Woofer Speaker. 40 Watts RMS
    Dayton Audio DC28F-8 1-1/8" Silk Dome Tweeter. 50 watts RMS

    From what I've been reading, crossover points should be midway between the frequency response limits, so about 360 Hz for the Woofer/Mid and 2550 Hz for the Mid/Tweeter. The 360 Hz for the woofer seems low to me, historically 3-way speaker woofer crossover were typically 550-1000 Hz. Even the new JBL L100 Classics crossover at 450 Hz. Is this because the woofer can handle more power? Should my woofers crossover up around 500 -600 Hz?


    In hindsight, I should have probably gone with a 2-way Econowave design and kept it simple. But hey, I've got 3-ways on the brain. If anyone would care to help out I would much appreciate it. I do not have measuring equipment so my ears are going to be the judge on how good they sound. I'm not expecting miracles here, I'm just looking for something I can enjoy. I like studio albums, rock, grunge, jazz, bluegrass. Speakers will be going in the basement rec room close to the wall about 10 ft apart. I'm not sure if I will raise the speakers on stands or angle them upward slightly.

    If I'm really off the rails somewhere, please let me know. Any criticism will be understood as totally constructive. Thanks.


    Before:





    After:







    Attached Files

  • #2
    ZMA files
    Attached Files

    Comment


    • #3
      Love those old cabinets that have horrid edge diffraction.

      Anyway, the process is like all others. As you have by default selected your drivers, now you figure out how they perform.
      First is to see how your box tuning is. The WT-2 or other tool that can generate accurate T/S parameters will let you plug into a box modeling program and adjust the tuning to your wishes. Not knowing when those were designed, don't know if they were tuned to best advantage. I would be suspect of any built pre about 1985.

      Once you have the box taken care of, you need to measure the performance of all the drivers IN PLACE.
      You first need to measure and generate the impedance and phase files. I use Woofer Tester 2, but it can be done with Limp or other programs. You measure each driver independently. As this is very low power, you can measure all drivers full range. No crossover.

      Then you need to measure the acoustic output. For this you need a calibrated mic and a program like Holm, Arta, Rew etc. ALWAYS use a blocking cap on the tweeter when doing acoustic measurements.
      With the .zma files and .frd files, you then input them into a crossover simulator. and start playing. I use PSD-Lite.
      Build what you come up with, measure and listen, change what you need. Lather, rinse, repeat.

      I caution you about manufacture published frs and zma files. Good for selecting parts, but usually way too far off to design a crossover. Consider them advertising. They are taken in laboratory conditions, not fitted to your box and baffle, It matters!

      Yes, designing a system is an engineering process. It takes tools, time and techniques.

      Just looking at the drivers, I might guess starting to do the crossovers @ 300 and 3K. I like high order crossovers, but have no idea about the breakup modes of the drivers. They look like doped paper which is usually pretty benign, so you may get away with second order. Just a guess.

      If you can't do the measurements, I might suggest putting Zobels on the drivers and then do textbook crossovers assuming a restive load and see how it sounds. You can use a program like EDGE to give you baffle step compensation without measurements. Zobels do need the inductance and resistance.

      Yes folks, I am big on Zobels. It took me a lot of years of struggle to learn that. They may not always make it sound better, but always make it easier to optimize the crossover which does.

      Can't give you 45 years of how to in just a forum response. Go read the instructions to the freeware tools, Holm, Arta, Limp, Edge, PSD-Lite, and any others you can find and they will teach you a lot.

      Comment


      • #4
        Here you go.

        Tweeter might be crossed a little though than you estimated but the DC130B-8 doesn't appear to be great beyond 2Khz and I dont think you really want to be listening to it. Hence the slightly lower xover point to smooth things out.

        Realistically this is all theoretical if these files aren't from direct measurements in your box. The real response curves can be wildly different to what a manufacturer publishes as there are so many variables that go into the final response.

        If you don't have the capacity to measure then all you can really do is take an approximate design, build it, listen to it and based on your observations tweaks can be made.
        Attached Files

        Comment


        • #5
          I see what you are saying. Decent mids are few and far between. I was more concerned with the bottom end of the tweeter. I would for sure think about higher order on the tweeter then.

          Comment


          • #6
            Dear clang,

            LOOK at dazzers "level" - it says "new member". (28 posts - total) Enough said? (tvr's only got 56)

            daZZer - you really shouldn't be doing XO design for ANYone - at this point.
            You have no XYZ offsets in your sim (Dayton' s zip files (probably) are all "minimum phase" (I don't think you know what that means?) and REQUIRE offsets to get the summing correct. You're crossing the "Silky" (tweeter) at 1.5kHz w/NO attenuation (it may not live long - and will sound pretty bad before it does die). I see NO "BSC" (baffle-step compensation) in your sim. clang might not need it, but he might. You should really do a LOT of reading (SpeakerBuilding 201 is a really good place to start, it's by Ray Alden, PE sells it, and craig doesn't like it when I mention it (sc)).
            Once you build several systems (of your own design) and get a few others to critique them (positively), you can start to offer more help.

            Your port looks like your box is tuned OK to me.
            What's the (internal) volume of your 5"mid's "enclosure" - do you think? It should be stuffed w/either polyfill (pillow stuffing) or fiberglass.

            Comment


            • #7
              Chris, number of posts may not be indicative. I have been building speakers off and on for 45 years. I built enough bad ones to know better now. I thought I was done a few years ago, but new house, new problems, so joined a forum as I have several to build /rebuild. Can't speak for Dez as I do not know the gentleman.

              Clang, You need a place to get started. A way to get hooked before spending a lot of money on test equipment ( and the boxes of passive components we accumulate.) My trick of using Zobels and doing text book crossover into a restive load calculation will get one started. Not the best by a long shot, but usually not terrible. Many a speaker was sold not even that good.

              There are some excellent books on the subject as well as some very good information free on the WEB. Do some homework. Google is your friend. Do be careful though, as there is as much folklore* as truth. Joe De Appolitto Measuring Loudspeakers is a good place to learn a lot as it really helps to know what the numbers really mean. Bullock on Boxes used to be the bible for box design. I actually borrowed a copy of Small's thesis and found it way too hard to read. Even his papers in the AES Loudspeaker Anthologys were a bit complicated. Self on Electronic Crossovers has a lot of good theory. I disagree with his universal amplifier design** but do not have any question about his engineering expertise.

              Measuring the actual acoustic offset @ crossover is not for beginners. Important as we know, and not the top of the voice coil as so many seem to think. One needs to measure it and that takes tools.

              Chris is quite correct, optimal results must deal with the offsets, baffle step, lobing, box internal resonances, edge diffraction, port velocities, and a host of other things. But one needs to get started. And as old timers know, sims are a place to get started, not the end-all. Again, If I were using these drivers, I would stick with 300 and 3K if second order and see how it sounds. We can compensate for quite a bit of frequency variation, but our poor brains do not do well with harmonic distortion. Psycho-acoustics is an entirely different subject and it bothers engineers as it is hard to put a number on how our brain takes clues and fills in the gaps. Musicians do better as they know how many notes they don't play but you think they did. Hope this helps.


              * Folklore: My term for Total BS, usually strongly defended using multiple logic fallacies. " Everyone knows", "The expert said" , personal attack of the opposing viewpoint, etc. ( see POTUS)
              ** Self amp sounded too much like a British "stiff upper lip" Kind of like trying too hard to be perfect. Hard to describe; just not as musical as a Parasound or my own.

              Comment


              • #8
                Thanks Gents,
                Reading some of your responses is like trying to do Calculus while skipping over algebra. I have a lot to learn.

                DaZZer, - thanks for taking a shot it it.

                Chris R - internal volume of the mids is just about 2 liters. They are stuffed with polyfill.

                Comment


                • #9
                  Clang,,,,, listen to Chris!

                  Have Fun! Mark

                  Comment


                  • #10
                    Originally posted by Chris Roemer View Post
                    Dear clang,

                    LOOK at dazzers "level" - it says "new member". (28 posts - total) Enough said? (tvr's only got 56)

                    daZZer - you really shouldn't be doing XO design for ANYone - at this point.
                    You have no XYZ offsets in your sim (Dayton' s zip files (probably) are all "minimum phase" (I don't think you know what that means?) and REQUIRE offsets to get the summing correct. You're crossing the "Silky" (tweeter) at 1.5kHz w/NO attenuation (it may not live long - and will sound pretty bad before it does die). I see NO "BSC" (baffle-step compensation) in your sim. clang might not need it, but he might. You should really do a LOT of reading (SpeakerBuilding 201 is a really good place to start, it's by Ray Alden, PE sells it, and craig doesn't like it when I mention it (sc)).
                    Once you build several systems (of your own design) and get a few others to critique them (positively), you can start to offer more help.

                    Your port looks like your box is tuned OK to me.
                    What's the (internal) volume of your 5"mid's "enclosure" - do you think? It should be stuffed w/either polyfill (pillow stuffing) or fiberglass.
                    Chris you're full of yourself. You are waaaaaaay off here. I've seen some of his work, and he has as much right as anyone to offer a x-over sim. Maybe you are too old to offer sims. I've simmed at least one of your suggested filters, and to me it looked wrong. Asked you about it, and no reply. That was likely 5 years ago. Must be the corona virus or something.

                    In addition... Anyone has a right to offer a sim, even if they are not very good at it. Sims are fun. People can even learn from a bad one sometimes. I've learned more than a few things from people with different approaches.

                    Comment


                    • #11
                      Originally posted by Chris Roemer View Post
                      LOOK at dazzers "level" - it says "new member". (28 posts - total) Enough said? (tvr's only got 56)

                      Once you build several systems (of your own design) and get a few others to critique them (positively), you can start to offer more help
                      If you want to tell Clang to ignore my submission for a series of technical and objective reasons - no problem with me. However try your absolute best next time to leave all of your personal assumptions and judgements about me, my background, my experience and my knowledge out of it. Accumulating posts on a parts express forum isn't exactly everyone's main priority in life!

                      Clang is asking for a crossover based on theoretical response and impedance curves - I've given him a series of simple crossover points - pretty hard to throw more theoretical compensations into the mix but I look forward to hearing what 12,500+ posts has to say about it.

                      Comment


                      • #12
                        Heck, maybe I'll offer one. I'll go for low part count, tilt the speaker, and tweak the treble / bass knob if needed design. Cheap and dirty. Then someone can tell me how wrong it is.


                        Comment


                        • #13
                          The mid and woofer summation on axis can be flat, but if you are 20 degrees above the axis, there's a dip in response at the x-over. If you go with steep filters, the dip is narrow, but, the filter is expensive. Is this something to casually listen to while playing games of some kind? Want good bass, but perfect imaging, and detail not needed?

                          Does your amp handle difficult 4 ohm loads?

                          Comment


                          • #14
                            Thanks Again Gents, I didn't mean for this thread to become a fight.

                            A little more about the use of these speakers -I'm planning on using them in the basement rec room/bar. It will NOT be a Maxell Man situation where I plan in sitting in one perfectly positioned chair and expect to recreate a symphony in front of me. I'll probably have a second pair of smaller speakers on the far side of the room by the bar to more evenly fill the room with background music when needed. The second pair of speakers are JBL 4401 Control Monitors from the mid-80s.

                            Music preference is mostly studio albums, rock up to grunge and alternative. Some Jazz and bluegrass and oldies. Some live stuff, but not my primary focus. If I had a choice between imaging and detail, I'd probably take detail. I don't expect to close my eyes and imagine a perfect stage in front of me.

                            The music sources are a CD player, tablet or phone, and occasionally I pull out my old turntable for nostalgic reasons.

                            My current amp is a Technics SU-V707 Integrated Amp from the mid-80s. I think its 85 watts/channel and handles 4-16 ohm. I prefer 8 ohm speakers since I believe they are easier on the amp. Other amps, receivers of varying vintages and power make the rotation, but most of the stuff is 80s-90s vintage. (I'm probably dating myself with my preferences).

                            I do not run a sub-woofer, so good base from the speakers is important. My gold standard target sound would be those old JBL 4312 Control Monitors I mention in my first post, but to be honest, I'd be completely happy if they sounded something like an old pair of Cerwin Vega D5s my brother has. Think 1980s-1990s 3-way monkey coffins - that's the sound I'm hoping for.

                            Comment


                            • #15
                              clang, for anyone to do the sim you are requesting, you need to post cabinet dimensions and driver positions on the baffle in order for the baffle diffraction simulations to be even remotely accurate.

                              Comment

                              Working...
                              X