Announcement

Collapse
No announcement yet.

Xsim and Phase Questions

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

  • 4thtry
    replied
    I tried to post the entire thing, but got a message that my post exceeded 10,000 characters. So I moved the TIPS section to this 2nd post:

    TIPS:

    A) Recently, I have begun to use dlr's new winfilters program to generate target curves,
    importing them into XSim using the FR window's “get file” drop down menu.

    B) I watch the impedance curve closely as I make changes to make sure the impedance
    does not drop too low (below 3 ohms or so) with a high negative phase angle. From what
    I have read in the forums, a high negative phase angle with low impedance is considered
    to be a very difficult amplifier load.

    C) As I design, I click on the frequency response curves drop down menu and then load the
    various drivers with phase curves onto the screen. I turn off the system and woofer phase
    curves to remove clutter. I watch the phase curves for the tweeter and midrange and try
    to get the very best "phase tracking" on these drivers. I want the curves to be parallel or
    overlapping throughout the crossover region. If these curves are non-parallel, then the
    crossover model will not sum properly.

    D) If I plan to use the speaker well away from the rear and side walls, then I try to apply the
    full 6dB of baffle step compensation (BSC). If I plan to put the speaker near a back
    wall, then I apply much less BSC. Also, if I place the woofer very close to the floor in my
    design, I have found that much less BSC is needed, due to the floor boundary
    reinforcement effect.

    E) I often repeat step 15 measurements above for other microphone positions. This makes
    it possible to perform horizontal and vertical polar response simulations in addition to
    on-axis simulations. When I do this, I am careful to use the “get file” procedure again
    to set “mod delays” for each microphone position. Every time the microphone is moved,
    the relative distance from mic tip to individual driver VC changes as well.

    F) I have also changed the FRD phase source from “as measured” to “derived” which brings
    up a “define response tails” dialog screen. This replaces the “as measured” phase with
    minimum phase data. I can then export the revised minimum phase FRD files and import
    them into other modeling software packages, such as PCD or WinPCD. This allows me
    to run polar response simulations, once I have the x,y,z driver coordinates set up properly.

    G) I can switch back and forth between “as measured” phase and “derived” phase in XSim,
    but when I do it is necessary to repeat the “get file” matching process to adjust the “mod
    delay” settings each time.
    -----------------------------------------

    Leave a comment:


  • 4thtry
    replied
    Here is the detailed listing (and one attached screen shot). It is fairly long and boring with 35 steps and 7 tips. There are probably several mistakes in this list, so use with caution. Basically, this is my interpretation of how to do things after having read and re-read several PETT threads & white papers over the years. Enjoy!

    -------------------------------------------------

    My XSim crossover development procedures:

    1) I use OmniMic and DATS V2 for all tests.
    2) I build the speaker box and mount all drivers.
    3) I drill a small hole in the back of the speaker and run a pair of wires to each driver and out
    the back. I seal the hole temporarily where the wires come out with a small dab of
    caulking.
    4) I set OmniMic up in my great room, which has a sloped ceiling leading to a loft area.
    All reflective surfaces are at least 5 to 6 feet away from the mic and speaker, except for
    the floor, which is always much closer.
    5) I usually position OmniMic 26” to 39” away from the speaker, directly on axis with the
    tweeter dome. The taller the speaker, the further away I position the mic.
    6) Sometimes I put the speaker on top of my DIY horizontal polar table (17 inches high) and
    sometimes I put the speaker on a stand. If a tower speaker, sometimes I place it directly
    on the floor so that the mic picks up the proper floor boundary gain effect. Remember that
    the amount of BSC applied is greatly influenced by how high you place the speaker during
    measurements!!!
    7) I hook up a good amplifier and CD player and run a pair of wires to the back of the
    speaker. The amp and CD player tone controls are always defeated or set to "FLAT". I
    do not use my internal laptop CD player or small digital amp.
    8) I insert a 33uF or 50uF non-polarized electrolytic protection capacitor in series with the
    tweeter for all tests.
    9) I hook the amplifier to the woofer first, carefully observing + and – polarity.
    10) I insert the OmniMic test CD into my CD player and play track 2, sine sweep.
    11) I set Omnimic to sine sweep, blended mode, 5ms gate, 1/48th octave smoothing,
    phase on.
    12) I adjust the amp volume to a moderate level that seems like it will not damage the
    drivers, especially the tweeter.
    13) VERY IMPORTANT: Once I set the testing volume level, I DO NOT change the volume
    level during any of the tests listed in step 15 below..
    14) If I do decide to change the volume level for some reason, perhaps because it sounds
    too loud on the tweeter, then I go back and re-do all step 15 tests from the beginning
    using the same volume level for all drivers.
    15) I take a FR measurement, then I pause OmniMic, then I pause the CD player, and then I
    click 'save curve" as an FRD file for all driver combos as follows:
    A) Woofer alone.FRD
    B) Midrange alone.FRD
    C) Tweeter alone with 33uF series protection cap.FRD
    D) Woofer+Midrange.FRD (paralleled up connection to amp)
    E) Woofer+Tweeter with 33uF prot cap.FRD (paralleled up connection to amp)
    F) Midrange+Tweeter with 33uF prot cap.FRD (paralleled up connection to amp)
    G) Woofer+Midrange+Tweeter with 33uF prot cap.FRD (paralleled up connection to
    amp)
    16) I am very careful to observe polarity + or - for all step 15 tests.
    17) The number of step 15 tests will be reduced from 7 to 3 for a 2 way build.
    18) I am also very careful not to move or bump the microphone or speaker during all step 15
    tests.
    19) This gives me a complete set of on-axis FRD files that can be used to develop a new
    crossover.
    20) Now I need a complete set of ZMA files for each driver measured "in box"
    21) I use DATS V2 to run an impedance sweep on each driver alone, and then export the
    curve using a zma extension as follows:
    A) Woofer in box.ZMA
    B) Midrange in box.zma
    C) Tweeter in box.zma (without 33uF protection cap)
    22) Note: I do not use a series protection cap on the tweeter for the zma test, as this is a
    very low power, single sweep type test.
    23) I load the XSim program, and, on the schematic CAD screen, create a 3 way connection
    of tweeter, midrange, and woofer to the amp with no crossover parts, just hard wire
    connections between amp and drivers. I do NOT install a 33uF cap in series with the
    tweeter on this schematic screen. See the attached screen shot example of what this
    schematic looks like.
    24) I then click on each driver icon to bring up the driver's input dialog box and enter the
    appropriate "alone.FRD" and "driver.zma" files for the 3 drivers that I created in step 15.
    25) Once I start entering the FRD and ZMA files, a strange looking, composite type,
    frequency response graph starts to show up on the frequency response graph. This
    looks like a bad response, but this is OK and normal.
    26) Once this is done, I set the "mod delay" offsets in XSim using the "Get file"
    procedure as outlined in the next few steps.
    27) On the top of XSim's frequency response graph, I click curves, then “get file”, then load
    the Midrange+Tweeter with 33uF prot cap.FRD file in, which then loads this file on top of
    the modeled graph.
    28) On the CAD schematic screen, I connect the tweeter and midrange icons, but temporarily
    disconnect the woofer icon.
    29) Now the goal is to adjust the "mod delay" of the tweeter until the model and "get file"
    match as closely as possible. Again, since this is a measurement of the tweeter and
    midrange alone, I temporarily disconnect the woofer on the schematic CAD screen.
    30) When the two curves match as closely as possible, the "mod delay" is correct for the
    tweeter driver dialog box.
    31) I repeat this same "get file" matching procedure for all combinations of measurements
    and possible crossover model connection possibilities. When done, I have a "mod
    delay" setting of some value for the tweeter and woofer, but NO "mod delay" setting for
    the midrange. The midrange is considered the anchor driver and does not require a
    "mod delay" adjustment.
    32) Once I have the "mod delays" set up, I am ready to start developing the crossover.
    But before starting, I must establish the SPL of the 100-300Hz region so
    that I can decide on the amount of baffle step compensation (BSC) to apply.
    This is not an easy thing to do, because the FR measurements below 400Hz are
    bouncing up and down due to room gain and boundary reflections.
    33) One method used to establish the 100-300Hz SPL level would be to take near field (NF)
    frequency response measurements of your midrange and woofer drivers and then merge
    them with the far field (FF) measurements using the blender program. If I decide to use
    this procedure, I need to replace all “as measured” phase data in my model with
    minimum phase data.
    34) Another method would be to apply 1/3 octave smoothing to all frequencies below 400Hz
    and then splice this data into the far field curves. With experience in looking at how the
    response rises or falls from 1000Hz to 100Hz, I have found that I can, with reasonable
    accuracy, estimate the amount of baffle step needed in this manner. I also adjust the
    amount of BSC applied based on the amount of floor boundary reinforcement in my
    measurements, which varies with microphone testing height.
    35) I then develop my crossover by inserting capacitors, inductors, and resistors in series
    and/or parallel with my drivers. As I do this, the frequency response and impedance
    curves change instantly, so I can see if I am moving in the right direction.

    Leave a comment:


  • 4thtry
    replied
    Originally posted by skatz View Post
    Bill

    Your technique as outlined in post #3 seems very helpful and I'd like to be able to hang onto it and refer to it. Is there a chance you could write out the steps in detail and put that into a sticky?
    I have been asked a number of times for my step-by-step XSim procedures. I have them written out in detail, but the file is on another computer. I will go get them and post tomorrow. The file is quite long and I think it also includes instructions for running horizontal and vertical polar response simulations as well. XSim does not have target curves, but you can use dlr Winfilters program to generate them and import as necessary to check your slopes.

    Leave a comment:


  • 4thtry
    replied
    Originally posted by Murphy-Pie View Post
    Which means you can never get good phase data from Omnimic. If accurate phase is important to you then using something like ARTA in its semi-dual channel configuration or Holm Impulse and its "time lock" function are better ways to go. But most people find the phase data in Omnimic close enough for government work.
    The phase curve does bounce just a little bit during measurements because of this. However, I have found that this slight inaccuracy has had no impact on my models. The variation is extremely small.

    Leave a comment:


  • Murphy-Pie
    replied
    Originally posted by 4thtry View Post

    I need to correct a mistake I made in the previous post. I incorrectly stated that the raw, as measured, OmniMic phase data consistently "starts" at the beginning of the impulse response. This is incorrect. Per OmniMic's help manual, "At zero delay, the time reference is the instant that the peak of the impulse response arrives at the microphone."

    Bill
    Which means you can never get good phase data from Omnimic. If accurate phase is important to you then using something like ARTA in its semi-dual channel configuration or Holm Impulse and its "time lock" function are better ways to go. But most people find the phase data in Omnimic close enough for government work.

    Leave a comment:


  • skatz
    replied
    Bill

    Your technique as outlined in post #3 seems very helpful and I'd like to be able to hang onto it and refer to it. Is there a chance you could write out the steps in detail and put that into a sticky?

    Leave a comment:


  • Geoff Millar
    replied
    Thanks Bill

    Geoff

    Leave a comment:


  • 4thtry
    replied
    Originally posted by Geoff Millar View Post

    Thank you all for those responses, very helpful indeed.

    I assume the "Hilbert Transform" is calculated by Xsim when you check the 'derived phase' box, or do you have to calculate it yourself? From what I found when I 'googled' it after reading your advice, it looked like rather complicated maths.

    From the above I think it's clear that I need measuring equipment.

    Thank you again

    Geoff
    I need to correct a mistake I made in the previous post. I incorrectly stated that the raw, as measured, OmniMic phase data consistently "starts" at the beginning of the impulse response. This is incorrect. Per OmniMic's help manual, "At zero delay, the time reference is the instant that the peak of the impulse response arrives at the microphone."

    To answer your question, yes, XSim calculates this for you. When you click the "derived" button, XSim brings up dialog box called "define response tails" which then asks you to put "tails" on the upper and lower end of the drivers response curve. You adjust the slope and frequency of the lower tail to "fill in" where you think the driver response would naturally continue below the noise floor. You adjust the slope and frequency of the higher tail to "fill in" where you think the driver response would naturally continue above the 20kHz area. When you hit enter to accept the tails, a small light blue "H" appears next to the driver icon to indicate that this driver is now using "minimum phase data."

    Hope this helps,

    Bill

    Leave a comment:


  • Geoff Millar
    replied
    Originally posted by 4thtry View Post
    I'm an XSim user and I use raw phase data, as measured by OmniMic, for all of my designs. I build the cabinets, mount the drivers, then take measurements of each driver alone. I also take measurements of all possible driver "paralleled up" combinations. I then load the driver alone FRD's into an XSim model that has no crossover components, just wires connecting the drivers to the amplifier. I then set the "mod delays" by using the 'get file" drop down menu to load one of the "paralleled up" combinations into XSim's FR window. When the "get file" matches the model, the "mod delay" is correct. The "mod delays" in XSim are not x,y,z coordinates placing the drivers on the baffle. They are simply the time delay that is necessary to match the two curves.

    If I want to, I can also click the FRD phase source "derived" button and then run the Hilbert transform to replace my raw phase data with minimum phase data. I have done this a number of times and it seems to produce similar results, but it varies a little bit depending on how I attach the tails. One thing to note, however, is that if I replace my OmniMic raw phase data with minimum phase data, I must use the "get file' procedure again to reset the "mod delays" for the minimum phase data. The mod delays are different because the raw phase data starts at a different point in time compared to minimum phase data. As I understand it, the raw OmniMic phase data consistently starts at the beginning of the impulse response. And because it is consistent from measurement to measurement, you can use it to create an accurate, repeatable model.

    Thank you all for those responses, very helpful indeed.

    I assume the "Hilbert Transform" is calculated by Xsim when you check the 'derived phase' box, or do you have to calculate it yourself? From what I found when I 'googled' it after reading your advice, it looked like rather complicated maths.

    From the above I think it's clear that I need measuring equipment.

    Thank you again

    Geoff

    Leave a comment:


  • 4thtry
    replied
    I'm an XSim user and I use raw phase data, as measured by OmniMic, for all of my designs. I build the cabinets, mount the drivers, then take measurements of each driver alone. I also take measurements of all possible driver "paralleled up" combinations. I then load the driver alone FRD's into an XSim model that has no crossover components, just wires connecting the drivers to the amplifier. I then set the "mod delays" by using the 'get file" drop down menu to load one of the "paralleled up" combinations into XSim's FR window. When the "get file" matches the model, the "mod delay" is correct. The "mod delays" in XSim are not x,y,z coordinates placing the drivers on the baffle. They are simply the time delay that is necessary to match the two curves.

    If I want to, I can also click the FRD phase source "derived" button and then run the Hilbert transform to replace my raw phase data with minimum phase data. I have done this a number of times and it seems to produce similar results, but it varies a little bit depending on how I attach the tails. One thing to note, however, is that if I replace my OmniMic raw phase data with minimum phase data, I must use the "get file' procedure again to reset the "mod delays" for the minimum phase data. The mod delays are different because the raw phase data starts at a different point in time compared to minimum phase data. As I understand it, the raw OmniMic phase data consistently starts at the beginning of the impulse response. And because it is consistent from measurement to measurement, you can use it to create an accurate, repeatable model.


    Leave a comment:


  • jhollander
    replied
    XSim can derive minimum phase with the Hilbert transform from the frequency response like other programs. What's not accurate would be the frequency response you are deriving the phase from unless you have measured it. The second inaccuracy is the match between the drivers. Unless you measure the frequency of both drivers (in a 2-way) playing at the same time you don't know the delay or offset between the drivers frequency response. The exception would be capturing the frequency response with as measured phase and a 2 channel system (for a t=0 marker).

    Leave a comment:


  • Geoff Millar
    started a topic Xsim and Phase Questions

    Xsim and Phase Questions

    Hello from Australia

    I use Xsim for mucking around with crossover designs, but am curious about 'phase' and how it's measured, particularly as it's so important in speaker/crossover design. What I've read suggests that you need to use measuring equipment and software to get accurate and useable data.

    Xsim has a check box in which to derive phase from driver frd and zma files when you "Tune" the drivers, but I'm not sure how accurate the phase figure is, given that information about 'acoustic centres' isn't entered.

    Is it possible to measure or assess phase without measuring equipment, to a standard from which you can design a reasonable crossover?

    I have built a speaker using Xsim to design the crossover (with much help from PETT members) but that was an adaptation of an existing commercial DIY design - presumably with correct phase already calculated in the original crossover..

    Your thoughts greatly appreciated!

    Thank you

    Geoff
Working...
X