Announcement

Collapse
No announcement yet.

ONS MMTMM Crossover Design - is something off?

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

  • #16
    Originally posted by billfitzmaurice View Post
    Only if wired in series. As I understand it the OP has it wired parallel. Please read my post again.I don't believe you understood what I said.


    It only plays a part in so far as the current demand on the amp goes up with a parallel wired resistor. It doesn't affect the tweeter output, which is voltage sourced, not current sourced. The voltage to the tweeter is unaffected by the presence of a parallel wired resistor.
    OK... a tweeter with a 4 ohm, and 8 ohm pad vs the same tweeter with a 4 ohm and 4 ohm pad. Which is louder? The 8 ohm would be in parallel with the tweeter, and changed to 4 ohms in the second sim.

    Comment


    • #17
      Here is my ham-fisted attempt at running the simulation in vituixcad.

      (Note - I am VERY new to this - and some things are likely wrong. For instance the inductors all have the default resistance set.)
      All drivers are at 0,0.

      With the 8.2 ohm resistor I'm getting a better curve - but what I notice is that with 4 drivers I get significantly more SPL before the crossover, which seems wrong. (Or is it? I'm learning here.)

      Also, my woofers that are supposed to be lower, seem almost the same response as the inner ones that are supposed to be mids. I don't have the separation that Paul had in his simulation.

      (I assume the summed total power is the olive green line above the woofers and tweeters?)

      Click image for larger version  Name:	crossover.png Views:	0 Size:	198.9 KB ID:	1472510
      Last edited by Jeremy White; 07-06-2021, 01:16 AM.

      Comment


      • #18
        Here is what Paul had in his writeup for reference...

        Any thoughts are welcome. I'm learning.

        Click image for larger version

Name:	paulreference.png
Views:	151
Size:	136.6 KB
ID:	1472515

        Comment


        • #19
          If I make the impedence flat at 8ohms across the b3n drivers, my graph starts looking better.

          This one also has the spatial separation correct.

          Again, with the 8.2ohm resistor. (bad graph deleted...)
          Last edited by Jeremy White; 07-06-2021, 01:16 AM.

          Comment


          • #20
            The 10uf cap is a 4.7uf in the schematic. I wouldn't think that would be so significant, but maybe. Perhaps there's a problem with the impedance files you used.

            Comment


            • #21
              Originally posted by rpb View Post
              OK... a tweeter with a 4 ohm, and 8 ohm pad vs the same tweeter with a 4 ohm and 4 ohm pad. Which is louder? The 8 ohm would be in parallel with the tweeter, and changed to 4 ohms in the second sim.
              The lower the parallel resistor value is (keeping all other variables the same) the more the tweeter is attenuated. The higher the value, the less it is attenuated. So an 8ohm parallel is louder vs 4ohm.

              4,4
              Click image for larger version

Name:	4and4.png
Views:	107
Size:	27.9 KB
ID:	1472529

              4,8 (+ ~2db)
              Click image for larger version

Name:	4and8.png
Views:	111
Size:	27.9 KB
ID:	1472528
              Constructions: Dayton+SB 2-Way v1 | Dayton+SB 2-Way v2 | Fabios (SB Monitors)
              Refurbs: KLH 2 | Rega Ela Mk1

              Comment


              • #22
                Found the error... the impedence wasn't being treated as logrithmic.

                Here is the final graph (and this one look correct.)

                I think the 8Ohm resistor looks pretty flat for dialog.

                What do you think?

                Click image for larger version

Name:	final crossover graph.png
Views:	138
Size:	137.5 KB
ID:	1472534

                Comment


                • #23
                  I don't have any experience with overnight sensations design but this response doesn't look right to me. The entire midrange is ~6db down. I know some people like the 'smiley face' response but this is a bit extreme. If the response of your speaker really looks like this then its no wonder the voice sounds suppressed. The problem isn't just the tweeter here.

                  Not here to question the design but you have 4 woofers summing output from ~300hz down but a gradual decline in the contribution of two of those woofers above that range resulting in a real imbalance in output through the midrange. I assume all woofers are the same?

                  Again, I don't have any direct experience with Paul's design here but what is the purpose of the secondary low-pass on two of the woofers? I can only see the MTM version on his site which has a much better balanced response than what you're showing here.

                  Try removing that second 3.5 coil. You definitely wont fix voice problems here by playing with the tweeters attenuation.
                  Constructions: Dayton+SB 2-Way v1 | Dayton+SB 2-Way v2 | Fabios (SB Monitors)
                  Refurbs: KLH 2 | Rega Ela Mk1

                  Comment


                  • Jeremy White
                    Jeremy White commented
                    Editing a comment
                    Here is the writeup --- https://www.sites.google.com/site/un...bipolewhatwhen

                    From what I understand the outer drivers act as woofers, with the inner drivers midrange.

                    I would trust his figures much more than my own - this is literally my first time trying to understand a crossover.

                • #24
                  I guess baffle step is the missing factor here. It looks like Paul's design takes advantage of it by summing the response under the baffle step and tapering it off above. I guess the files your simulating with don't have this baffle step factored in.

                  Either way "voice" is the 200hz through 1K range (secondary harmonics etc extend beyond, but the fundamentals lay here). So if the speaker sounds soft here its either the tweeter is really loud for some reason and subjectively overpowering the lower frequencies or perhaps some cancellation/boundary effects are at play depending on where the speaker is positioned. Try moving it around.
                  http://arqen.com/acoustics-101/speak...-interference/

                  Also just double check the speaker setup on your receiver. On mine for instance, you can specifically tell it you don't have a center channel regardless of the listening mode, you can control the relative volume of the center channels and you can set a delay in relation to the other speakers...

                  Constructions: Dayton+SB 2-Way v1 | Dayton+SB 2-Way v2 | Fabios (SB Monitors)
                  Refurbs: KLH 2 | Rega Ela Mk1

                  Comment


                  • #25
                    Originally posted by DeZZar View Post
                    I guess baffle step is the missing factor here. It looks like Paul's design takes advantage of it by summing the response under the baffle step and tapering it off above. I guess the files your simulating with don't have this baffle step factored in.
                    Can you explain what you mean by "baffle step"?

                    The issue I'm having is that dialog sounds less clear than my reference speakers (dayton airs.) The lower range and music sounds great. So I was blaming this on the tweeter.

                    I may be doing my testing in a horrible place. Its essentially in the center of a large tiled room on a coffee table. I'll try moving it and seeing if there is a difference.

                    Comment


                    • #26
                      Changing only the parallel resistor does change the tweeter spl. In the sims below, a 4 ohm is changed to 8 ohm.



                      Click image for larger version  Name:	pad two.jpg Views:	0 Size:	260.7 KB ID:	1472630 Click image for larger version  Name:	pad one.jpg Views:	0 Size:	262.0 KB ID:	1472631

                      Comment


                      • #27
                        Originally posted by Jeremy White View Post

                        Can you explain what you mean by "baffle step"?

                        The issue I'm having is that dialog sounds less clear than my reference speakers (dayton airs.) The lower range and music sounds great. So I was blaming this on the tweeter.

                        I may be doing my testing in a horrible place. Its essentially in the center of a large tiled room on a coffee table. I'll try moving it and seeing if there is a difference.
                        Placing a speaker near large objects can cause tonal balance changes. The tv screen is one such object, and the table top is another. One option is EQ if your receiver has it for the center.

                        Comment


                        • #28
                          I'm not going to get into the technical weeds of this design right now, but I will say a few things.

                          How are you mounting/positioning the speaker? When I designed it, it was meant to be placed kind of close to other reflective surfaces, like a TV. With that said, there's still other variables with regards to positioning.

                          If you feel like you want more treble, we can tweak some values in the filter and bring that up for you. Do you have a desired frequency range you might want boosted?
                          Isn't it about time we started answering rhetorical questions?

                          Paul Carmody's DIY Audio Projects
                          Twitter: @undefinition1

                          Comment


                          • #29
                            TTBOMK, the T20 (HiVi) is only offered as an 8n(ohm) tweeter.
                            In Paul's writeup, he says he didn't do a design (version) for the Dayton ND tweeter (like on the OS kits), just the HiVi drivers.
                            Using an 8n resistor (in parallel) w/the T20 will boost its output (in this design) by approx. +1dB (compared to a 6 or 5.6n resistor - <-THIS diff. is not worth worrying about).

                            You can't really use mfr. data curves (F/Z) to sim this particular speaker in THIS particular case. Paul DOES know what he's doing. Having an MMtMM using an "extra" series coil for a 2nd set of woofers (in parallel w/the 1st set) is quite common for a "2.5-way" speaker design (which this is). The idea is that the furthest (out) woofers boost the bass w/out interfering w/the mid-range, which is only handled by the closer-in drivers (to help reduce/eliminate off-axis "comb filtering" in a design like this). A MAJOR effect on any speaker's sound is its position (in the room, and in relation to other objects in the room). This CC speaker COULD be mounted on a wall (under a wall-mounted LARGE flat-screen), OR it could sit on a large stand (typically right under the screen's bottom edge - bot not necessarily so). It could be tucked under the screen (so the baffle face is flush w/he screen), or it could sit out "proud" of the screen (or the screen might start some distance above the CC). If on a "cabinet" (stand), it MIGHT have the reflective cabinet top projecting out in front of it (reflecting the sound, and probably creating nulls/reinforcements), or it may not. There are just too many variables for CC speaker placement to design a "one-size-fits-all" speaker. Also, YOUR position will have an effect on the sound (in ANY room), obviously.

                            Paul's OSCC is NOT unique in its design that makes it more susceptible to any of these positioning "issues" than any other CC. What it does do is help avoid some common (out of the) "box speaker" problems (like comb-filtering and non-flat FR) while providing a high-value CC (for those who feel the need for one). I still do not have a CC, even though I use my system for 2/ch music as well as for all movie watching (letting the "phantom CC" fill-in where needed).

                            OTOH, now that you've got your CC in position (and have logged several hours listening), there's nothing wrong w/customizing your setup, either by repositioning components, adjusting the FR curve using tone-controls (or EQ), OR by modding your CC to suit your tastes. (You MIGHT want to get some FR measuring equipment before making changes to a "proven" design though - unless you're experienced enough listening to some fairly "flat" systems under your listening environment conditions, or maybe not!) Have fun.

                            [OOPS - Hi Paul !]
                            Last edited by Chris Roemer; 07-07-2021, 05:04 PM. Reason: didn't see Mr.Carmody here !

                            Comment


                            • #30
                              Originally posted by Paul Carmody View Post
                              I'm not going to get into the technical weeds of this design right now, but I will say a few things.

                              How are you mounting/positioning the speaker? When I designed it, it was meant to be placed kind of close to other reflective surfaces, like a TV. With that said, there's still other variables with regards to positioning.

                              If you feel like you want more treble, we can tweak some values in the filter and bring that up for you. Do you have a desired frequency range you might want boosted?
                              Thanks Paul.

                              In the end, it will be mounted on a reflective (glass topped) cabinet, underneath a flat screen TV. I haven't had a chance to test this position yet because the space under the TV is too small and I need to make some risers!

                              I think what I have learned is that I'm probably testing this in the worst possible location. On a pedestal about waist height in the middle of a big room.

                              I'll get back to you once I can figure out how to test it where it will eventually sit.

                              Again, thanks for everyone's help as I learn about speakers.

                              If someone had a recommendation for an ultra-cheap FR measuring setup, please advise.

                              For some eye candy - here is a picture of the speaker in question. It is done in white automotive lacquer with a piano finish of clear coat. The front baffle is solid rosewood.


                              Click image for larger version  Name:	center channel.png Views:	0 Size:	663.6 KB ID:	1472720

                              Comment

                              Working...
                              X