Announcement

Collapse
No announcement yet.

miniDSP and ACD experiences..........

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

  • miniDSP and ACD experiences..........

    Attached Files

  • #2
    Re: miniDSP and ACD experiences..........

    The response of the woofer has been spliced to its nearfield at (IIRC) 500Hz and the tweeter output was reduced 10dB by the miniDSP for the measurement process. Using the three-measurment procedure for the delays, I determined that the tweeter “AC” is 0.42 msec behind that of the woofer. Putting that info into the ACD spreadsheets and fiddling for a couple of hours led me to the predicted AND MEASURED responses shown in the previous post.

    The white curve is measured, the yellow and blue the individual driver responses and the dark blue is the predicted summed response. I dropped the measured response (white) 2 dB so that a jumble of four curves becomes a little more tractable to see. Not bad for a first try! The actual ARTA measurement of the system response is the third figure. To compare the predicted response to the measured I also spliced the low frequency response to the measured system response. As is my custom, I failed to save the settings so, evidently, I did it a little differently than I did two weeks earlier when I was splicing it to the woofer-only response.

    Developing a filter in ACD is just common sense…..see a peak you don’t like, put the mouse over the peak to determine the frequency, input the frequency and amount you want to reduce the response, take a guess at the “Q”, press enter and watch the response change. There are also functions other than peaks to keep your mind occupied. Tweak until you’re happy and then go on to the next thing you don’t like. Start to finish on the above crossover took about two hours. The actual crossover design point is 1350 Hz with LR2 slopes.

    The ACD spreadsheets calculate the digital filter coefficients (“miniDSP Biquads”), which then must be input into miniDSP. This is accomplished by simple copy and paste commands. miniDSP shows you what you have changed as soon as you click the “Process” button. The filter transfer functions in ACD and as input from the calculated coefficients into miniDSP are shown for the system above and for the drivers below. As you look at the graphs, note that the filter curves in ACD include the crossover curves, while in miniDSP the crossover curves are separate from the equalization curves.

    To finish the story, these speakers sound great, yes they disappear, there are no objectionable peaks, etc. and, wow! do they sound good with classic rock. Secondly, I will never again do a passive crossover without first listening to it with the crossover done in miniDSP. With the miniDSP I can listen to a dozen different crossovers in an hour if I choose to, all at the click of a mouse. Then build the actual passives to match the filters I decide I like best. And minimize the component count in the passives by selectively turning off various filter elements to leave only those essential to the design. This whole thing has really been fun and educational and I thank miniDSP and Charley Laub for the capabilities they have given us.

    Roger Hill
    Attached Files

    Comment


    • #3
      Re: miniDSP and ACD experiences..........

      Roger, thanks for sharing. I'm on my phone right now so I haven't read it all. I'm currently working through ACD. It's quite involved. I hope to get it figured. Maybe I'll pick up some tips reading through your experience. Thanks.
      https://www.youtube.com/channel/UCm2...oSKdB448TTVEnQ

      Comment


      • #4
        Re: miniDSP and ACD experiences..........

        Thanks Roger.
        Lot to learn but you explain some areas I dint know about miniDSP
        Definitely I need a new Christmas gift.

        Comment


        • #5
          Re: miniDSP and ACD experiences..........

          Thanks for the writeup. I haven't jumped in on the miniDSP yet, but I plan to give it a try soon. Good information like this from detail oriented end users is always helpful.

          Thanks!

          Sandy.

          Comment


          • #6
            Re: miniDSP and ACD experiences..........

            Thanks also. I think a mini-dsp is in my future, and at present I have a "quantum" block about biquads.
            I'm sure Charlie has made it easier with ACD.

            I think I hear a difference - wow, it's amazing!" Ethan Winer: audio myths
            "As God is my witness I'll never be without a good pair of speakers!" Scarlett O'Hara

            High value, high quality RS150/TB28-537SH bookshelf - TARGAS NLA!
            SB13/Vifa BC25SC06 MTM DCR Galeons-SB13-MTM
            My Voxel min sub Yet-another-Voxel-build

            Tangband W6-sub

            Comment


            • #7
              Re: miniDSP and ACD experiences..........

              Roger, how did you determine the tweeter's AC was .42 msec behind that of the woofer?

              What if the tweeter was ahead of the woofer by the same amount? What would you enter for time delay in that case?

              Thank you...

              Comment


              • #8
                Re: miniDSP and ACD experiences..........

                Roger, how did you determine the tweeter's AC was .42 msec behind that of the woofer?

                you do three measurements without moving anything or changing any settings. first, measure the tweeter alone and woofer alone to get their relative output levels if you will be using non-matching amps for woofer and tweeter. I had to reduce the output of my tweeter channels by 10dB in miniDSP to get (approximately) the same spl's out of woofer and tweeter. put the mic on the midpoint of the line that extends from the tweeter hole center to the woofer hole center and then back about five feet from the baffle. measure the output or the woofer alone, then the output of the tweeter alone and, finally, the output of the woofer plus tweeter, both connected to the same input (ordinary alligator clip leads between them is fine). then, measure the nearfield of the woofer. using charlie's response combiner/blender splice the nearfield to the measured woofer response, add the tails necessary to calculate the minimum phase and export and save the minimum phase woofer file. import the tweeter response to the combiner/blender, add the tails and export and save that file. then, using the import/export utility, import the woofer minimum phase file into an ACD driver template, then the tweeter minimum phase file into another copy of the driver response spreadsheet and, finally, the combined (woofer +tweeter) response into the acd system template. charlie gives instructions for how to do all of this.

                now, out at the microphone (five feet away from the baffle) the woofer and tweeter spl's are not in phase. the driver "acoustic centers" are not lined up in the speaker so that sound from one of them passes the baffle after sound from the other. because they are not in phase, they add constructively or destructively at the microphone, and how much coherence or interference there is different at every frequency, and thus, there is a unique pattern of the spl vs. frequency. the ACD system spreadsheet plots this frequency function for you for the woofer alone, tweeter alone, combined drivers and what it calls "predicted" frequency function, which is the complex sum of the woofer plus tweeter signals (amplitude plus minimum phase), with the phase of the furthest driver increased by the number of wavelengths of delay (you enter this number as a time) at a given frequency times 2*pi. you can physically inspect the drivers and see which one will have its "AC" further back, it is close to the voice coil. in the spreadsheet for the driver which is further back, start inserting trial guesses as to the delay and recalculating until the predicted response (system spreadsheet) of the pair matches what you measured. then, in miniDSP, delay the other driver by this amount. in my case (horn tweeter) the tweeter was .425 sec behind the woofer "AC", so in miniDSP i had to add delay of .42 msec to the woofer output signals to align the centers.

                What if the tweeter was ahead of the woofer by the same amount? What would you enter for time delay in that case?
                delay the tweeter output signals by .42 msec

                Comment


                • #9
                  Re: miniDSP and ACD experiences..........

                  Originally posted by Roger Hill View Post
                  Roger, how did you determine the tweeter's AC was .42 msec behind that of the woofer?

                  you do three measurements without moving anything or changing any settings. first, measure the tweeter alone and woofer alone to get their relative output levels if you will be using non-matching amps for woofer and tweeter. I had to reduce the output of my tweeter channels by 10dB in miniDSP to get (approximately) the same spl's out of woofer and tweeter. put the mic on the midpoint of the line that extends from the tweeter hole center to the woofer hole center and then back about five feet from the baffle. measure the output or the woofer alone, then the output of the tweeter alone and, finally, the output of the woofer plus tweeter, both connected to the same input (ordinary alligator clip leads between them is fine). then, measure the nearfield of the woofer. using charlie's response combiner/blender splice the nearfield to the measured woofer response, add the tails necessary to calculate the minimum phase and export and save the minimum phase woofer file. import the tweeter response to the combiner/blender, add the tails and export and save that file. then, using the import/export utility, import the woofer minimum phase file into an ACD driver template, then the tweeter minimum phase file into another copy of the driver response spreadsheet and, finally, the combined (woofer +tweeter) response into the acd system template. charlie gives instructions for how to do all of this.

                  now, out at the microphone (five feet away from the baffle) the woofer and tweeter spl's are not in phase. the driver "acoustic centers" are not lined up in the speaker so that sound from one of them passes the baffle after sound from the other. because they are not in phase, they add constructively or destructively at the microphone, and how much coherence or interference there is different at every frequency, and thus, there is a unique pattern of the spl vs. frequency. the ACD system spreadsheet plots this frequency function for you for the woofer alone, tweeter alone, combined drivers and what it calls "predicted" frequency function, which is the complex sum of the woofer plus tweeter signals (amplitude plus minimum phase), with the phase of the furthest driver increased by the number of wavelengths of delay (you enter this number as a time) at a given frequency times 2*pi. you can physically inspect the drivers and see which one will have its "AC" further back, it is close to the voice coil. in the spreadsheet for the driver which is further back, start inserting trial guesses as to the delay and recalculating until the predicted response (system spreadsheet) of the pair matches what you measured. then, in miniDSP, delay the other driver by this amount. in my case (horn tweeter) the tweeter was .425 sec behind the woofer "AC", so in miniDSP i had to add delay of .42 msec to the woofer output signals to align the centers.

                  What if the tweeter was ahead of the woofer by the same amount? What would you enter for time delay in that case?
                  delay the tweeter output signals by .42 msec
                  I wanted to follow up on Roger's post, above. I don't think that he mentioned it, but the tweeter is a horn loaded compression driver. The horn moves the acoustic center of the driver back compared to a direct radiator tweeter (e.g. your typical dome tweeter). This is why in developing the model of the loudspeaker in ACD, Roger had to enter a "delay" for the tweeter - it's acoustic center is behind the woofer (which is taken as the zero relative delay plane).

                  If the tweeter is "ahead" of the woofer (as is often the case with DIY projects) then you should keep the tweeter delay at zero and set the woofer delay to the appropriate amount as determined in the three measurement procedure that I and other outline for determining driver offsets.

                  It is essential to have accurate phase information when determining driver offsets from minimum phase FRD data. I highly recommend using the FRD blender and splicing on low frequency data to the measured driver data before doing the determination of acoustic offset, especially if the crossover point will be below approximately 2k Hz.

                  -Charlie
                  Charlie's Audio Pages: http://audio.claub.net

                  Comment


                  • #10
                    Re: miniDSP and ACD experiences..........

                    Thank you for the replies.

                    Also, I'm a little puzzled about measuring things from 5 feet away. Don't you typically measure drivers from about 39 or so inches away, except for nearfield?

                    I measure the woofer and tweeter separately, then measure them in parallel to try to determine the "z offset" in PCD.

                    Then I extract minimum phase for the woofer and tweeter using Jeff's Response Modeler.

                    Would I do the same for designing a miniDSP crossover?

                    Comment


                    • #11
                      Re: miniDSP and ACD experiences..........

                      Originally posted by foxfire3 View Post
                      Thank you for the replies.

                      Also, I'm a little puzzled about measuring things from 5 feet away. Don't you typically measure drivers from about 39 or so inches away, except for nearfield?
                      You want to be in the "farfield", which is anything past about 1m for a typical home speaker. Whether you are at 1m or 3m doesn't really matter, as long as you can make a measurement independent of the room (e.g. derive the frequency response from a gated impulse measurement).

                      Originally posted by foxfire3 View Post
                      I measure the woofer and tweeter separately, then measure them in parallel to try to determine the "z offset" in PCD.

                      Then I extract minimum phase for the woofer and tweeter using Jeff's Response Modeler.

                      Would I do the same for designing a miniDSP crossover?
                      Yes, that's the same approach I use, essentially. I have recently released a tool called the Frequency Response Blender and Minimum Phase Extractor for combining various measurements (nearfield, farfield, etc.) and then extracting the minimum phase that serves the same purpose as the Response Modeler. This is what I was referring to in the post above.

                      -Charlie
                      Charlie's Audio Pages: http://audio.claub.net

                      Comment


                      • #12
                        Re: miniDSP and ACD experiences..........

                        Originally posted by charlielaub View Post
                        I wanted to follow up on Roger's post, above. I don't think that he mentioned it, but the tweeter is a horn loaded compression driver. The horn moves the acoustic center of the driver back compared to a direct radiator tweeter (e.g. your typical dome tweeter). This is why in developing the model of the loudspeaker in ACD, Roger had to enter a "delay" for the tweeter - it's acoustic center is behind the woofer (which is taken as the zero relative delay plane).

                        If the tweeter is "ahead" of the woofer (as is often the case with DIY projects) then you should keep the tweeter delay at zero and set the woofer delay to the appropriate amount as determined in the three measurement procedure that I and other outline for determining driver offsets.

                        It is essential to have accurate phase information when determining driver offsets from minimum phase FRD data. I highly recommend using the FRD blender and splicing on low frequency data to the measured driver data before doing the determination of acoustic offset, especially if the crossover point will be below approximately 2k Hz.

                        -Charlie
                        Charlie,

                        You've confused what Roger wrote. He added delay to the woofer, not the tweeter, because the woofer's voice coil is physically ahead of the tweeter's. With typical home audio tweeters you would do the opposite and add delay to the tweeter.

                        Comment


                        • #13
                          Re: miniDSP and ACD experiences..........

                          Another way to find the correct delay is to set matching woofer and tweeter output levels, wire them out of phase, then play a steady tone at the desired crossover frequency. Adjust the delay to the appropriate driver until you hit a minimum SPL. That's the reverse null, and shows that the drivers are perfectly out of phase. When you wire them normally again they will be in phase.

                          Comment


                          • #14
                            Re: miniDSP and ACD experiences..........

                            Originally posted by Jonathan Anspach View Post
                            Charlie,

                            You've confused what Roger wrote. He added delay to the woofer, not the tweeter, because the woofer's voice coil is physically ahead of the tweeter's. With typical home audio tweeters you would do the opposite and add delay to the tweeter.
                            Again, we are talking semantical differences that must be kept straight. I believe Charlie is talking of the estimated delay added to the spreadsheet to get the predictions correct. You tell ACD (driver response spreadsheet) an estimate of how much that driver is delayed relative to the baffle. When you are through with your iterative guesses in ACD, you tell miniDSP how much to delay the signal to the other drivers in order to get the centers aligned. Just keep straight which is which and you'll be fine.

                            Comment


                            • #15
                              Re: miniDSP and ACD experiences..........

                              Kudos to you! I have a lot of experience with the minidsp, but I'm trying to make my way through ACD, having a tough time. The fact you were able to get through it is encouraging, and I'll keep at it.

                              Comment

                              Working...
                              X