Announcement

Collapse
No announcement yet.

Speaker Design Question #2 - Measuring Driver Offset (z-axis) - WinPCP

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

  • #16
    Originally posted by wogg View Post
    Really? I always guesstimated to the woofer dust cap, figuring that as frequency goes up the radiation becomes more prominent there. I could have completely made that up in my own head though.
    Yeah, the somewhere in between the peak of the dustcap and the point where the voicecoil meets the cone, depending on the driver and the size of the dustcap. In any case, it isn't the middle of the top plate.
    Don't waste your money on a new set of speakers, you get more mileage from a cheap pair of sneakers. Next phase, new wave, dance craze, anyways it's still rock and roll to me!

    Comment


    • #17
      Let me suggest reading my article on finding relative offset. It's different than Dickason's approach, largely because the software measurement tools have advanced since then. I have never used a cap on the tweeter because it's not necessary to use high power for this. MLS measurements have sufficient S/N ratio at low power for this, but others may use a cap.

      I'll add a few comments to the discussion. As Pete pointed out, you need to be working with minimum-phase files. Jeff's spreadsheet is a good way to go.

      With that said, you need to be sure to set highpass/lowpass slopes prior to generating minimum phase. For the tweeter, the highpass usually only needs extending at a 12db/oct slope, but the lowpass, the top end, can be problematic if it goes above the measurement system limits. However, the primary concern is that once those are set, don't change them.

      For the woofer, the highpass (box range) should be added first if you are going to add that to the curve. I'm not going to cover the various issues related to that, only say that whatever you do for the highpass should be done prior to generating minimum phase. Once you do this, you have what you need to determine relative acoustic offset.

      I frequently point this out, that once you've determined minimum phase, do not alter the driver files. This is because any further change, such as for the box highpass, will affect the minimum phase if you re-generate it. This will in turn require finding offset again. Remember that offset is relative, that is, it's a value required so that the two minimum-phase files sum to the measured sum. If you change either file, the relative offset needs to be updated to account for the change.

      The absolute value of the offset isn't really important. All it amounts to is the difference between the two measurements that "aligns" the time delays between the measurements to correspond to the difference required for the generated phase of each driver file. There is no "correct" offset, only the offset needed for the measurement phase as generated for the SPL of the files.

      For years I never added a lowpass for the tweeter top end. The offset will be different for that vs. one with a lowpass added. The only significant difference is that the offset needed for the flat lowpass vs. an added lowpass is different. The summed response with the woofer will be the same as long as the offset is changed to accommodate the difference in the generated phase. That means using the modified to determine relative acoustic offset again to take the difference in the now re-generated file into account.

      dlr
      WinPCD - Windows .NET Passive Crossover Designer

      Dave's Speaker Pages

      Comment


      • #18
        Your post dlr was quite confusing to me until I realized you are speaking of a HBT to tail the response. Yes, you will get better results by providing tails to the measurements, to remove the noise floor and provide data well beyond the 20kHz measured. This is needed for the minimum phase calculations to work accurately. Using the terms lowpass and highpass here is really confusing, since the intention is not to alter the response, but to provide additional data.

        Using the FRD blender as I did in the minimum phase files I posted above, the HBT is completed for you with some adjustable settings in the spreadsheet, making the whole "extract minimum phase accurately" process pretty dang simple.

        I've posted this many times before, but truly the best/easiest method to obtain accurate acoustic offsets it to ditch the USB mic and move to a 2-channel measurement setup. Here you have a reference input from the amplifier which allows you to easily determine time of flight from the speaker to the mic.

        In any case, with the files provided by the OP the offset is zero with the filters in place. My brain isn't working well today, but I think this may be expected for a design where the drivers are in-phase on the tweeter axis.
        Last edited by dcibel; 09-29-2017, 02:23 PM.
        Don't waste your money on a new set of speakers, you get more mileage from a cheap pair of sneakers. Next phase, new wave, dance craze, anyways it's still rock and roll to me!

        Comment


        • #19
          I tend to use driver lowpass and highpass since a driver is essentially an acoustic bandpass device, but I can see where that could be confusing. I guess that I don't think of tailing partly because for the woofer or a closed box midrange you use a measured or calculated response that's not an assumed results as tailing to a slope tends to be.

          With respect to a 2-channel device, it's not necessary to be able to measure phase to get minimum-phase files since re-generating phase for minimum-phase essentially throws out the measured phase. What's important for the crossover is that the files are minimum-phase and are adjusted for offset. 2-channel systems make it much easier to get the maximum valid data with a time impulse window and in the case of MLS to improve S/N, though the latter isn't an issue for the offset aspect.

          dlr
          WinPCD - Windows .NET Passive Crossover Designer

          Dave's Speaker Pages

          Comment


          • #20
            Here's another good article by David: LINK. As an aside, REW can be setup as a two channel system, that's how I use it.
            "Everything is nothing without a high sound quality." (Sure Electronics)

            Comment


            • #21
              Originally posted by dlr View Post
              With respect to a 2-channel device, it's not necessary to be able to measure phase to get minimum-phase files since re-generating phase for minimum-phase essentially throws out the measured phase.
              I'm not sure what you are trying to say here. Maybe I wasn't clear myself, my point in using a 2 channel measurement is that you don't have to go through this process of measuring each driver, then both, then fiddling with the offset to determine the offset between the drivers. You can simply take your on-axis measurements and determine the distance from the speaker to mic from the time difference between the impulse (mic to reference). Subtract the two distances and there's the offset. Here you can even reference each driver to the baffle, rather than to each other, making it once step easier to swap drivers if you want to. The key here is to use a proper mic stand so that you can raise and lower the mic while keeping the distance to the baffle unchanged.

              SoundEasy, ARTA, and REW can all be used as a 2-channel measurement system, but you'll need to ditch the USB mic. I also use a USB omnimic, but not for designing speakers. It is great for setting up EQ, taking room measurements, etc. because it's a lot more plug and play type of a setup.
              Don't waste your money on a new set of speakers, you get more mileage from a cheap pair of sneakers. Next phase, new wave, dance craze, anyways it's still rock and roll to me!

              Comment


              • #22
                Originally posted by dcibel View Post
                I'm not sure what you are trying to say here. Maybe I wasn't clear myself, my point in using a 2 channel measurement is that you don't have to go through this process of measuring each driver, then both, then fiddling with the offset to determine the offset between the drivers. You can simply take your on-axis measurements and determine the distance from the speaker to mic from the time difference between the impulse (mic to reference). Subtract the two distances and there's the offset. Here you can even reference each driver to the baffle, rather than to each other, making it once step easier to swap drivers if you want to. The key here is to use a proper mic stand so that you can raise and lower the mic while keeping the distance to the baffle unchanged.
                This is not a very precise way to determine offset because an impulse response for a woofer doesn't have a well-defined rise as does a tweeter. In addition, that offset doesn't account for alteration to phase that is introduced by any tailing that may be done such as adding a box highpass to a woofer or midrange if that is made or tailing of a tweeter top end (lowpass). Remember that you must be working with minimum-phase files if you are going to determine an offset that allows for off-axis response prediction in software and that requires re-generating phase. The three-measurement scheme is probably the most accurate and precise way to set up a system for design that is for more than a single on-axis point.

                dlr
                WinPCD - Windows .NET Passive Crossover Designer

                Dave's Speaker Pages

                Comment


                • #23
                  I certainly haven't had any problems with accuracy, and I've used both methods. This is following the method outlined in this instruction for SoundEasy. Acoustic Center Evaluation.pdf

                  Once you apply the HBT, you must apply a small offset and make the phase overlap the original measurement.The result is distance + minimum phase, ready to simulate off-axis.
                  Attached Files
                  Don't waste your money on a new set of speakers, you get more mileage from a cheap pair of sneakers. Next phase, new wave, dance craze, anyways it's still rock and roll to me!

                  Comment


                  • #24
                    I've used Soundeasy and understand that, but it's still adjusting for relative acoustic offset, although many don't have the option of doing it that way and I prefer to have design measurements made at the design point since it already includes the accurate off-axis response of woofer and/or midrange in the case of measuring on the tweeter axis. I've never used SE measurements, I still use LAUD, even for those times when I was using SE since LAUD is a calibrated system and measurements taken years apart produce the same absolute SPL result, but I digress. My point was more that measured phase is not required, meaning a 2-channel measurement system is not required. I used LMS first, years ago, and used its built-in tailing function. It only produced minimum-phase results due to the type of measurement, a stepped sine wave. There was no impulse response. Any measurement system can be used to get a set of SPL files with an accurate offset using the three measurement technique. That's the beauty of it.

                    dlr
                    WinPCD - Windows .NET Passive Crossover Designer

                    Dave's Speaker Pages

                    Comment


                    • #25
                      So - I tried the following with a recently acquired pair of Swan 2.2A Kit (NO crossover installed):

                      1) Utilizing REW - took a full-range measurement of the speaker - On tweeter axis - Distance of 22" - 5.00ms Gating

                      2) Utilizing REW - took Woofer only measurements - Still on Tweeter Axis - Distance of 22" from speakers baffle - 5.00ms Gating

                      3) Utilizing REW - took Tweeter only measurements - On Tweeter Axis - Distance of 22" from speaker baffle - 5.00ms Gating

                      4) Took all measurements and converted in "FRD" files (which contained frequency, amplitude, and phase data)

                      5) Uploaded the files into "Acoustic Offset tab of WinPCD

                      6) Set measurements distance to 0.558 meters (because 22" = 0.558 meters)

                      7) Set "Driver 1" (aka the Tweeter) - X-offset = 0, and Y offset = 0 (because tweeter was used as the reference point)

                      8) Set "Driver 2" (aka the Woofer) - X offset = 0, and Y offset = -0.160 (because the woofer is mounted 6.3" = 0.160 meters below tweeter)

                      9) I believe the next step is to adjust Driver 2 (the Woofer) Z-offset until the mathematical summed response of Step 2 and Step 3 above match the measurement in Step 1 above.

                      The result was 0.0234m per WinPCD. Now I'm not sure whether that number is right or wrong, but I did read article by Jeff B stating that it is not uncommon for 1" T/6.5" W combo to have approximately a little over .0200+/- off set.

                      However, the reason why I am still skeptical about the result is because I did not use minimum phase files. I did try using Jeff B's Frequency Response Blender spreadsheet to extract minimum phase, but I don't think I'm getting accurate results because I'm likely not specifying the tails correctly. When I use the minimum phase files I'm generating, and utilizing them in WinPCD Acoustic Offset tab - I'm getting nonsensical results.

                      One question: once driver off set has been determined, are these files discarded? Or are these the sames files one uses when developing the crossover.

                      I apologize is this is too elementary of a question. I'm trying to understand whether the sole purpose of the files created above is to determine driver offset.

                      When developing crossovers, is the general practice to first measure each driver on axis with the driver (i.e., tweeter @ tweeter axis, mid @ mid axis, and woofer @ woofer axis), or are all measurements taken on the design axis (e.g., tweeter axis). If its the former, I'm guessing one must specify the location of each driver on the baffle, along with the piston diameter, etc so the software can approximate what the summed response would be at various horizontal/vertical axis?
                      Click image for larger version

Name:	swans - driver offset.jpg
Views:	1
Size:	399.2 KB
ID:	1347157



                      Comment


                      • #26
                        Hi Dhar,

                        I've also tried working with your files ( including creating my own Minimum Phase versions within Blender which were only marginally different from the ones that "dcibel" posted ) & I now think there's something wrong with your original data captures.

                        Are you confident they are both the same polarity ?

                        I really don't know why they're acting so wrong but I just can't get them to offer a normal offset reading ( unless I give the woofer a positive offset that effectively puts it out in front of the tweeter ).

                        Perhaps your original captures were done too close ( at 20" ) > I'd try doubling the distance . If that doesn't work ( after trying more "curve-fitting" using new MP files ), keep the 40" distance and then lose the 5 ms gating.

                        Comment


                        • #27
                          Thanks Earl. In my last post (#25) - I was actually using a speaker with no crossover. In the initial post, I was using the Bamberg SS 2TM kit and crossover was included in the signal path. In the last post, no crossover was included. The drivers were wired in parallel when I was taking the summed response. When utilizing those files in WinPCD, I was getting a z-offset of approximately -0.023mm, which still makes sense. I wasn't sure if I can rely on that estimate if the files utilized did not contain minimum phase. Sent from my iPhone using Tapatalk

                          Comment


                          • #28
                            Originally posted by dkalsi View Post
                            Thanks Earl. In my last post (#25) - I was actually using a speaker with no crossover. In the initial post, I was using the Bamberg SS 2TM kit and crossover was included in the signal path. In the last post, no crossover was included. The drivers were wired in parallel when I was taking the summed response. When utilizing those files in WinPCD, I was getting a z-offset of approximately -0.023mm, which still makes sense. I wasn't sure if I can rely on that estimate if the files utilized did not contain minimum phase. Sent from my iPhone using Tapatalk

                            Hello again

                            Thanks for the clarification of the differences between your two file types ( I'm thinking that explains the unexpected results that I got from those files ).

                            ( Moving on ), I would encourage you to solve the issues you have creating minimum phase files. One wants to design filters using the most accepted process ( ie; "best-practice" )

                            IOW, I personally don't think it's worth the effort exploring just how much error is baked into a design when one foregoes designing with minimum phase files ( & essentially, that's becoming a direction behind some of your questions ).


                            0.023mm (?) is a small distance.



                            PS: The link that dlr posted ( in post #17 ) states ( that for final filter design purposes ) he uses a different set of files ( from those used to determine acoustic offset.
                            - The final design files are driver files that are averages made from data collected from within a 30 deg window ( & taken from his preferred listening axis, I assume ).
                            - While it's not stated if those averaged files are created from ( plus & minus ) horizontal or vertical positions , I'll assume its horizontal.
                            Last edited by EarlK; 10-02-2017, 09:36 AM.

                            Comment


                            • #29
                              Originally posted by EarlK View Post
                              PS: The link that dlr posted ( in post #17 ) states ( that for final filter design purposes ) he uses a different set of files ( from those used to determine acoustic offset.
                              - The final design files are driver files that are averages made from data collected from within a 30 deg window ( & taken from his preferred listening axis, I assume ).
                              - While it's not stated if those averaged files are created from ( plus & minus ) horizontal or vertical positions , I'll assume its horizontal.
                              You can design with the same files used to determine offset, I do that most of the time now, but that's because I now use an offset tweeter coupled with significant felt for diffraction reduction. Back then I wasn't aware of how to handle the diffraction issue so a window average can be better than a single measurement if there's no diffraction control because the diffraction influence varies so much with position.

                              If an average is used for the design part after finding offset, just be sure to use the same tweeter tailing at the top end and the woofer tailings (box response and/or top end). A tweeter average is probably the only one really necessary because it's the one most significantly affected by diffraction, the woofer sees little other than step response which is nearly the same on all axes in the front hemisphere.

                              dlr
                              WinPCD - Windows .NET Passive Crossover Designer

                              Dave's Speaker Pages

                              Comment


                              • #30
                                Originally posted by EarlK View Post

                                0.023mm (?) is a small distance.

                                Earl,

                                I'm sorry - I meant 0.023 meters, or 23 mm, or approximately 0.91"

                                Again - that result was obtained without using files with minimum phase.

                                The following files were utilized to obtain the 23mm offset distance indicated above:

                                Swan - Full Range - @ Tweeter Axis - at 22" = "SW-FR-TX-22.Frd"
                                Swan - Woofer - @ Tweeter Axis - at 22" = "SW-W-TX-22.Frd"
                                Swan - Tweeter - @ Tweeter Axis - at 22" = "SW-T-TX-22.Frd"

                                I also have a near-field measurements of the woofer: "SW-W-NF-0.25.FRD"

                                When designing a simple two way, I should:

                                1) Measure Tweeter, Woofer, Tweeter + Woofer at the design axis
                                2) Convert all files to include minimum phase (need to add "tails" to data to extract minimum phase)
                                3) Using WinPCD or Jeff B's PCD spreadsheet, determine driver offset

                                Next,

                                4) Take nearfield measurements of the woofer (and port if applicable)
                                5) Apply baffle diffraction to the nearfield measurements taken (using "FRD Response Blender") in Step 4
                                6) Blend nearfield measurement (modified with diffraction data) (i.e. the result of Step 6) and blend with far-field measurement obtained in Step 1
                                7) Add "tails" to woofer response.
                                8) Extract Minimum Phase
                                9) Save the "blended" woofer file results to FRD file
                                10) Load woofer file derived in Step 9 above and Tweeter file in Step 1 above into your choice of x-over designer (e.g PCD or WinPCD)


                                Attached Files

                                Comment

                                Working...
                                X