Midwest Audio Fest

It’s that time audio enthusiasts! Registration for the 2019 Speaker Design Competition is now open! Visit for details and to list your speaker project. We are excited to see all returning participants, and look forward to meeting some new designers this year, as well! Be sure your plans include a visit to the Parts Express Tent Sale for the lowest prices of the year, and the Audio Swap Meet where you can buy and trade with other audio fans. We hope to see you this summer! Vivian and Jill
See more
See less

Minimum Phase Question For Xover Sim

  • Filter
  • Time
  • Show
Clear All
new posts

  • Minimum Phase Question For Xover Sim

    If you are designing a xover for a speaker that is going to be mounted to an infinite baffle, like an in-wall, can the PE .frd & .zma files be used as they are posted or do you still need to extract minimum phase with a program like Frequency Response Modeler or Blender?

    I am just looking to put together some quick and easy speakers that will be mounted in the wall of my shop.


  • #2
    I have used the (10-15 yr) OLD files that way (the ones made by Darren K., who's been gone a long, long, time). I have NO IDEA how the new files are generated. Obviously, if they contain "measured phase", then they'd all have to be taken at exactly the same baffle distance to be of any use. Maybe someone @ PE can shed some light on that process?


    • #3
      Per the read me file with the measurements,
      "All data measured at 2.83Vrms at a distance of 1 meter. All FRD data files include near field measurement data below 450 Hz. Smoothing is 1/24 octave."


      • #4
        The text files say the low frequency response has been spliced, so I have always assumed minimum phase. They look like minimum phase. You will still need to derive an offset for a 2-way speaker system.
        John H

        Synergy Horn, SLS-85, BMR-3L, Mini-TL, BR-2, Titan OB, B452, Udique, Vultus, Latus1, Seriatim, Aperivox,Pencil Tower


        • #5
          Taken at the same distance also needs to specify the phase conditions, since phase is provided. Either it's excess-phase, meaning measured phase at the measurement point, or it's minimum-phase. If it's measured phase, then to be directly used, all measurements must have had the same start time marker if it's a feedback probe, 2-channel system. If not, they would need to be minimum-phase.

          Even for an infinite baffle system, the acoustic offset will be needed for minimum-phase responses. For measured phase, the time reference is essential. However, given that there are differences in the acoustic center of the drivers, it's likely that the start time marker (if a 2-channel setup) is not the same. The reason for this is that to get the longest impulse length, it's best to set the start time marker just before the impulse, which has different time of arrival due the differing acoustic center distances between drivers.

          If a constant time marker is used, then it should to be set to be just before the impulse rise of the driver with the closet acoustic center to the mic, usually a tweeter. Other drivers will then have a time maker some time short of the start of their impulse, meaning the FFT for those drivers will have pre-impulse data in the FFT, which tends to provide unrealistic smoothing and lower frequency limits due to an artificially longer impulse in the window used for the FFT. Small differences aren't significant, so details count.

          I note that John mentioned minimum-phase appearance of the files and the offset required. Full agreement here if that's the case.

          WinPCD - Windows .NET Passive Crossover Designer

          Dave's Speaker Pages


          • #6
            I believe they are using omnimic which should output minimum phase


            • #7
              The FRDs are not derived minimum, but are as measured from 1m measured from the baffle mounting surface, and then phase delay adjusted to be as if the mic could be at that plane. So they should be usable as provided, assuming all will be mounted on the same vertical baffle.

              Omnimic does NOT by default output minimum phase, it measures actual phase (important in cases when you might measure a non-minimum phase system such as nearly any multiway system or an array). You can derive minumum phase using OM (or Xsim), but what it gives when you measure at first is the actual phase curve -- though unless you do delay matching steps there will be an unknown fixed delay involved (since OM doesn't know when the signal left the speaker) -- Jeff Bagby has a writeup somewhere here about how to do delay matching. The guys at Dayton Audio go through an equivalent procedure so that delay offsets due to to driver depths or mountings are accounted for already. Phase is for distance =0 from the baffle (note: the baffle, not from the voice coil!).

              I wouldn't suggest assuming that a "minimum phase" derivation will always be the best to use. Individual 1-way drivers are indeed minimum phase by nature, but minimum phase derivations work from measured data curves which are almost never complete -- the responses have noise due to dynamic range out past the rolloffs (which affect the min phase derivation in the passband also) and likely have outright missing data due to limited measurement bandwidths or windowing. Results can be pretty good (and are good for use with baffle simulation models), but you need to fix up the rolloff tails to be at the right rates in noise corrupted or data-missing areas if within about two octaves of where you need the derivation to be accurate. And then, of course, you need to make delays compatible between drivers again if you do that, since minimum phase derivation disposes of any 'time of flight' delays that might have been in the measured data!
              Free & Free-form simulator/designer for Passive Crossovers
              SynergyCalc 5: design spreadsheet for Wooden horns and DIY Synergy Waveguides
              Super easy and cheap to make high performance sound diffusers