Announcement

Collapse
No announcement yet.

Acoustic centers

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

  • Acoustic centers

    I've been using Jeff's method for finding the difference in Z offset between woofers and tweeters in my last three builds. I was surprised which woofers required larger offsets. So far the Anarchy is deepest of all at 44 mm.

    I should probably mention that the tweeters in every case are the RS28A or RS28F.

  • #2
    Re: Acoustic centers

    I've always wondered why we can't develop a standard to specify acoustic centers, it would be tremendously useful. Even just the acoustic center, relative to the surface of a flush mount, at 2000 hZ would make simulation really easy.

    Now that I think of it, though, it seems like we only have methods of finding relative offsets, but again, this does not prevent us from deriving a useful figure relative to baffle.

    Comment


    • #3
      Re: Acoustic centers

      The other thing I find interesting is that if I measure carefully, the frd when measuring both woofer and tweeter together perfectly overlays the woofer and tweeter frds when the woofer z is dialed in.

      Comment


      • #4
        Re: Acoustic centers

        Originally posted by civit View Post
        I've always wondered why we can't develop a standard to specify acoustic centers, it would be tremendously useful. Even just the acoustic center, relative to the surface of a flush mount, at 2000 hZ would make simulation really easy.

        Now that I think of it, though, it seems like we only have methods of finding relative offsets, but again, this does not prevent us from deriving a useful figure relative to baffle.
        That's because there is no accepted method for determining absolute acoustic center. Knowing the absolute value doesn't actually matter because it's the relative offset that if of any importance. If you use direct measurements, it's not a factor at all. Relative acoustic offset is determined by first creating a model of the measured response and then generating phase in which SPL and phase can each be generated from the other. To generate the phase you usually must also extend (tail) the highpass and lowpass because the Hilbert Transform assumes a model that extends from -infinity to +infinity. Fortunately, the impact of the SPL on phase at any given point lessens as the frequency delta increases. The highpass is usually not difficult, but the lowpass almost always requires making an assumption on the lowpass Fc and and/or the slope. The latter is usually assumed to be constant, but it's never so in the real world. We just don't know what's happening above the limits of the measurement system. Change the slope and you've changed the phase, therefore the "absolute" value for that driver.

        It's easier to demonstrate with an example.

        Driver Model Accuracy and its Impact on Phase

        dlr
        WinPCD - Windows .NET Passive Crossover Designer

        Dave's Speaker Pages

        Comment


        • #5
          Re: Acoustic centers

          +1

          Right, it's not a value that's really useful. All we really need to know is the relative phase shift between the two drivers we are using, and different combinations will net different results.
          Click here for Jeff Bagby's Loudspeaker Design Software

          Comment


          • #6
            Re: Acoustic centers

            And FWIW and FYI I have measured the relative accoustic center using an RS28A and an RS28F both using the same woofer, baffle, and mic position... they do not measure exactly the same. If I remember correctly the "F" is 4 mm in front of the "A". I have also swapped and measured the shielded version of the "A" and its AC is identical to the unshielded "A", but it is almost 2 dB more sensitive.
            Craig

            I drive way too fast to worry about cholesterol.

            Comment


            • #7
              Re: Acoustic centers

              Makes sense, the shapes/profiles of the protective grills are quite different between the A and the F.

              Comment


              • #8
                Re: Acoustic centers

                Originally posted by civit View Post
                I've always wondered why we can't develop a standard to specify acoustic centers, it would be tremendously useful. Even just the acoustic center, relative to the surface of a flush mount, at 2000 hZ would make simulation really easy.

                Now that I think of it, though, it seems like we only have methods of finding relative offsets, but again, this does not prevent us from deriving a useful figure relative to baffle.


                Hi Civit,

                Yes, you are correct. As you pointed out, each driver has a “mounting plane”, which is where you bolt the driver to the front baffle, and is therefore an excellent reference plane for acoustic centre. Knowing the acoustic centre of each driver is very useful.

                You could freely swap drivers in your modeling CAD software, and always get the correct phase relationship – well, good enough to continue with CAD modeling. As always, one will need to measure and tweak the final design.

                Finding the acoustic centre is a completely different story. It’s difficult at best and that’s why you see “it does not matter” opinions floating around. And this is strange, because people who claim to have found true “minimum phase” phase response – they have actually determined the acoustic centre at the same time.

                I can offer you an approximated method, which yields good results (not perfect). Well, at least sufficient to be applied across your database of drivers you are working with. The second paper is more recent.

                http://www.bodziosoftware.com.au/Aco...Evaluation.zip
                http://www.bodziosoftware.com.au/Min...riveration.pdf

                Best Regards,
                Bohdan

                Comment


                • #9
                  all cases be far more precise, especially for those using systems that do not provide a feedback probe input. Keep that in mind. As I frequently point out, you can find relative acoustic offset without any measured phase if you use the three-measurement method. I first did it when using LMS that does not measure phase. Do it once and you're done for that design. I did it with CALSOD in meters down to 4 or 5 decimal places in the 90's with CALSOD using DOS. There was no need for that amount of precision, but I always liked to see just how far I could take it. Swapping out drivers can be done with some amount of confidence, but it requires good assumptions be made of all drivers under comparison. Of course measurement repeatability enters in that picture, systems such as HOLM and OmniMic measure with MLS and no feedback probe, but have good repeatability for SPL.

                  Making the claim that one can swap out drivers using some form of estimated AC assumed from drivers responses also totally overlooks the fact that the drivers most sensitive to offset (M/T) are also those most likely to have crossovers highly affected by diffraction in the crossover region. As you know, this can be several db, possibly far exceeding error in AC from any assumptions. Swapping out drivers with confidence would require having a driver model not only of the raw infinite baffle response, but also including diffraction response for the intended baffle.

                  With that in mind, I would not suggest to users here that they can simply swap drivers without having measurements of those drivers on-baffle unless they want to model to get an idea if a driver is suitable for the intended application. If they can measure on-baffle, measure and determine offset.

                  dlr
                  WinPCD - Windows .NET Passive Crossover Designer

                  Dave's Speaker Pages

                  Comment


                  • #10
                    Re: Acoustic centers

                    Hmmmm..... I was going to post a lengthy reply, but I think Dave covered it very well, so no one needs me to say the same thing, except to say I agree completely with dlr.

                    I will only add - Knowing the acoustic center of a driver is not meaningless, it does mean something, it's just not very useful. Let's say this value is published. We'll find that tweeter A has its acoustic center 6 mm behind the flange, tweeter B has it 3 mm behind, and tweeter C has its center 14 mm behind the flange due to its waveguide, then we have the woofer's..... It's so much simpler in crossover simulation to simply find the relative difference between the two drivers, and allow the tweeter (whatever its acoustic center may be) to be treated as the reference plane that we zero from. Absolutes don't benefit us much here.

                    Jeff
                    Click here for Jeff Bagby's Loudspeaker Design Software

                    Comment


                    • #11
                      Re: Acoustic centers

                      Acoustic center is a means to express the how much each frequency is delayed from the electrical signal to the mic. I think the only way to spec it for a driver so that it would be easy to apply in a design is to measure the driver response on axis, and then provide the phase response and a reference observation point for the phase response. eg measure 1m on axis, provide the raw measure and that distance. Even better, take the 1m measure and remove the time of flight to the front of the driver (measured as 90 degree to the center front of the driver but to a distance equivalent to the baffle front). We almost all flush mount the drivers so I think it makes sense to specify it the second way. Then it would be easy to use this in any simulation and different drivers from different manufacturers can be mixed and matched in sim

                      Comment


                      • #12
                        Re: Acoustic centers

                        Originally posted by dlr View Post
                        Making the claim that one can swap out drivers using some form of estimated AC assumed from drivers responses also totally overlooks the fact that the drivers most sensitive to offset (M/T) are also those most likely to have crossovers highly affected by diffraction in the crossover region. As you know, this can be several db, possibly far exceeding error in AC from any assumptions. Swapping out drivers with confidence would require having a driver model not only of the raw infinite baffle response, but also including diffraction response for the intended baffle.

                        With that in mind, I would not suggest to users here that they can simply swap drivers without having measurements of those drivers on-baffle unless they want to model to get an idea if a driver is suitable for the intended application. If they can measure on-baffle, measure and determine offset.

                        dlr


                        Hi Dave,

                        Baffle diffraction is not an issue for swapping drivers. Your particular concern was with midrange and tweeters.

                        Drivers are typically measured on large test baffles. For instance, you can easily show, that diffraction amplitude and phase of large baffle (say 2m x 2m and asymmetrical mounting of drivers, which is recommended) are both essentially flat lines starting from 200Hz, with amplitude sitting at +6dB.

                        Therefore, the diffraction or midrange and tweeter can be simply ignored for such condition if you reduce driver’s SPL by -6dB (amount of diffraction gain).

                        Then, you can simply run your CAD and simulate the diffraction of your individual design baffle – no problem here.

                        Best Regards,
                        Bohdan

                        Comment


                        • #13
                          Re: Acoustic centers

                          Originally posted by Jeff B. View Post
                          I will only add - Knowing the acoustic center of a driver is not meaningless, it does mean something, it's just not very useful. Jeff
                          this is the most useful information of this entire post.
                          craigk

                          " Voicing is often the term used for band aids to cover for initial design/planning errors " - Pallas

                          Comment


                          • #14
                            Re: Acoustic centers

                            I beg to differ. Diffraction is an issue, but not in the way that you represent. Part of that problem is that the diffraction signature will not be uniform to all drivers on a specific baffle/location, since driver directionality is not constant between drivers, therefore the diffraction signature is not constant. Basing differences on diaphragm diameter are not completely reliable, either, before considering differences in materials or in driver type. If someone wishes to attempt to design without direct measurements, they may get a reasonable estimate. But they have no way to know whether or not their AC determination is accurate or close nor how accurate the diffraction simulation is.

                            I'm also well enough aware of large baffle measurements, I have my own 2m x 2m baffle that I've used for years. I think it highly unlikely that manufacturers will ever adopt any "standard" method for providing the AC of drivers. They don't generally provide downloadable files for use and those that are provided,whether in files or graphs, are often not truly representative. You may suggest taking the IEC baffle response and creating a model from it,then attempting to determine AC, but even with that, you do of course have to simulate diffraction and apply that to each driver that you might wish to substitute.

                            The comment made makes me think that the desire was for having a value for each driver that could then be applied to design even for those who can make measurements. It assumes that the user also makes models that provide an accurate minimum-phase response relative to the provided AC.

                            With all of that, if someone has measurement capability, there is no substitute for actually measuring on-baffle. In that case, there is no reason to do anything other than determine offset. Absolute AC becomes moot. It's a simple task and leaves no doubt as to the accuracy.

                            dlr
                            WinPCD - Windows .NET Passive Crossover Designer

                            Dave's Speaker Pages

                            Comment


                            • #15
                              Re: Acoustic centers

                              Originally posted by dlr View Post
                              I beg to differ. Diffraction is an issue, but not in the way that you represent. Part of that problem is that the diffraction signature will not be uniform to all drivers on a specific baffle/location, since driver directionality is not constant between drivers, therefore the diffraction signature is not constant. Basing differences on diaphragm diameter are not completely reliable, either, before considering differences in materials or in driver type. If someone wishes to attempt to design without direct measurements, they may get a reasonable estimate. But they have no way to know whether or not their AC determination is accurate or close nor how accurate the diffraction simulation is.

                              I'm also well enough aware of large baffle measurements, I have my own 2m x 2m baffle that I've used for years. I think it highly unlikely that manufacturers will ever adopt any "standard" method for providing the AC of drivers. They don't generally provide downloadable files for use and those that are provided,whether in files or graphs, are often not truly representative. You may suggest taking the IEC baffle response and creating a model from it,then attempting to determine AC, but even with that, you do of course have to simulate diffraction and apply that to each driver that you might wish to substitute.

                              The comment made makes me think that the desire was for having a value for each driver that could then be applied to design even for those who can make measurements. It assumes that the user also makes models that provide an accurate minimum-phase response relative to the provided AC.

                              With all of that, if someone has measurement capability, there is no substitute for actually measuring on-baffle. In that case, there is no reason to do anything other than determine offset. Absolute AC becomes moot. It's a simple task and leaves no doubt as to the accuracy.

                              dlr



                              Hi Dave,

                              We have to agree to disagree on the diffraction issue – I am OK with that.

                              The other point: to my knowledge, there is no such standard – you are correct.

                              But this should not stop DIY community from developing one.

                              The concept of swapping drivers in not trivial. It extends into a DIY-wide database of properly edited loudspeaker drivers. Imagine, that any DIY designer could take any driver from this database and plug it into his/her current CAD design for evaluation in the system.

                              Almost any other parameter is easily measurable or available one way or another. However, the minimum-phase phase response (or AC location) is still the one, that attracts “too difficult to do” arguments - it does not need to be like this. The accuracy of the parameters in the database only needs to be good enough for a “first-cut” design.

                              Now, measuring the relative phase offsets for each combination of drivers for new design means, that you actually have to purchase the drivers, built the box first and then measure the relative phase. This defeats much of the purpose and attractiveness of using CAD.
                              The CAD was supposed to save you the hard work and money by allowing you to "pick-and-choose" any driver and predict the performance of the system before any money or physical effort is spent.

                              DIY database would allow you to do just that – model your design first and then decide if it’s worth proceeding with the rest of it.

                              From your previous post I understand, that you are concerned, that there is a measurement equipment limitation, because most of the cheap microphones only works to 20kHz. I do not see a problem with woofer or midrange. But, here is the solution – Get 150 people to donate $10 to a nominated member of this forum (could be you), who will purchase a precision microphone going to 50kHz or more. This person will be responsible for measuring an odd tweeter going beyond 20kHz.




                              Best Regards,
                              Bohdan

                              Comment

                              Working...
                              X