Announcement

Collapse
No announcement yet.

Dual-Woofer Measurements with Omnimic

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

  • Adam_M
    started a topic Dual-Woofer Measurements with Omnimic

    Dual-Woofer Measurements with Omnimic

    I'm starting work on a W-M/T-W center channel (woofers in parallel) and was curious as to how the woofers should be measured. I see a few potential options:
    1. Measure both woofers separately, add both responses to PCD
    2. Measure 1 woofer only (without a connection to the second) and let PCD assume the response and impedance are the same

    Should the woofers ever be wired in parallel and measured as one for the measurements? I assume this would only work farfield. Of course they have to be when its time to wire the crossover or else things just won't work as designed.

  • jhollander
    replied
    I've done the hypotenuse method and got close.

    Leave a comment:


  • Adam_M
    replied
    Thanks. I assumed that the delay matching with raw driver measurement was portable from PCD to XSim, which was apparently a bad assumption. The parts I don't have and can't easily create from what's in the parts drawer are on the way. Now the fun really begins.

    Leave a comment:


  • jhollander
    replied
    Originally posted by Adam_M View Post
    Regarding the power measurements, I'll probably be using 12W mills (I have most of the values I need laying around). When simulating power dissipation in Xsim, the series resistor in the filter mid hit 38ish watts of at 20v at about 2.4k. I'm of course not going to be feeding a 2.4k sine wave to this thing at 20v, so I'm not sure how to interpret that information. Is 1 12w mills OK in that application or should I double it up?
    I use about 1/3 the watts shown in XSim at the volume level for woofer xmax, so a 12 watt mills looks pretty good.

    Originally posted by Adam_M View Post
    Also, how do you get the delays set correctly in Xsim? I tried the z offset from PCD directly as well as the hypotenuse of the triangle formed by the XYZ positions of the drivers as the offset, and neither summed as they do in PCD.
    Delay matching is the same as PCD. Create the circuit for T+M, load your FRD files. In the Frequency Response window, choose Get File and add the Measured T+M file to the window as an overlay. Assuming you measured on the the tweeter axis adjust the M delay (mod delay in the driver tuning window) to match the curve.

    If you want to exactly match PCD export any curve from PCD with your drivers, create the circuit to match what you are exporting then do the delay matching. I like to remove the smoothing from the curves and adjust the vertical step size in the FR window.

    Leave a comment:


  • Chris Roemer
    replied
    I wouldn't think a 4-ohm amp would complain.

    Leave a comment:


  • Adam_M
    replied
    Also, is this impedance particularly nasty? I'm thinking the phase angle briefly exceeding +-30 degrees is ok, but not optimal. This is obviously a 4 ohm speaker, but is that dip to under 3 in the upper midrange problematic? I'd expect the dip to about 3 in the lower midrange would be more problematic, but I don't think there's anything to be done with that - it kind of is what it is when paralleling those 2 woofers.

    Leave a comment:


  • Adam_M
    replied
    Cool, thanks! I fired up Xsim, and was able to simplify the XO a bit. Got rid of the 3rd order inductor and shrunk the big cap in the LP and got a more conventional L-pad in the mid. The HP RC has almost exactly 2dB of lift at 10K, decreasing below that but certainly present. I had to take the measurement rig down for Christmas, so it'll have to go back up to check the peak at 7.5k - hopefully this weekend.


    Regarding the power measurements, I'll probably be using 12W mills (I have most of the values I need laying around). When simulating power dissipation in Xsim, the series resistor in the filter mid hit 38ish watts of at 20v at about 2.4k. I'm of course not going to be feeding a 2.4k sine wave to this thing at 20v, so I'm not sure how to interpret that information. Is 1 12w mills OK in that application or should I double it up?

    Also, how do you get the delays set correctly in Xsim? I tried the z offset from PCD directly as well as the hypotenuse of the triangle formed by the XYZ positions of the drivers as the offset, and neither summed as they do in PCD.

    Leave a comment:


  • jhollander
    replied
    Originally posted by Adam_M View Post
    Woofers (2x W18E001): I can take the 2nd inductor out and adjust the other components, but it causes problems with the M-T crossover because the W-M side of the bandpass crossover needs tweaked. I guess that's a comment as much as a question, but it seems strange to have 3rd order electrical for this. I was also considering steel laminate on the 1mH for the low DCR and slight efficiency bump. Any issues with that?
    If you take out the second inductor on the woofer then lower the order on the mid to keep the phase match. I don’t mind a few lumps below 1,000 Hz as it’s harder to hear them and I like a dip in the 500 to 700 Hz. range. I also like using the steel laminates down to 1 mH, but I might trade that big cap for a larger inductor, kind of depends what your impedance looks like.

    Originally posted by Adam_M View Post
    Mid (MDM55): I have the padding resistor in front of the rest of the XO. I don't see that done often. Is there a reason to not do that, other than power handling? Do I need to double up that resistor (either 2 at .5 value in series or 2 at 2x value in parallel). A traditional l-pad ends up with strange peaking in the response.
    For the mid, now that you have your x-o in PCD, I like to move over to XSim to optimize. You can see the power handling by component. I like to double up or use a higher wattage resistor. You can move the resistor to the other side of the low pass or high pass and get similar results. This is much easier in XSim.

    Originally posted by Adam_M View Post
    Tweeter (OW2): Similar question as the mid regarding the RC in front of the XO. Another topology I don't see much. That cap lifts the top end though. Also, there's a bit of a diffraction peak at 7.5k. I'm going to put some felt on the grill frame - Its probably in the 1.75" from the tweeter range to tame it. DO you think the peak as it exists will be audible?
    For the tweeter, you can move the RC to the other side of the x-o and adjust (XSim). The RC could be a waste unless you get at least a 1-2 dB lift at 10K or lower. The peak at 7.5K might go away off axis, did you happen to look? This would be a something that I think would be audible when voicing.

    If you want to post your files you might get more ideas (you would need to filter the good from the bad). IMO keep tinkering and buy enough parts to test a few ideas.
    Last edited by jhollander; 01-04-2019, 10:58 AM. Reason: spelling

    Leave a comment:


  • Adam_M
    replied
    I'm on to the crossover design of this speaker and would welcome suggestions - it seems like a lot of components. I have a couple of specific questions although general simplification suggestion would be great:

    Woofers (2x W18E001): I can take the 2nd inductor out and adjust the other components, but it causes problems with the M-T crossover because the W-M side of the bandpass crossover needs tweaked. I guess that's a comment as much as a question, but it seems strange to have 3rd order electrical for this. I was also considering steel laminate on the 1mH for the low DCR and slight efficiency bump. Any issues with that?

    Mid (MDM55): I have the padding resistor in front of the rest of the XO. I don't see that done often. Is there a reason to not do that, other than power handling? Do I need to double up that resistor (either 2 at .5 value in series or 2 at 2x value in parallel). A traditional l-pad ends up with strange peaking in the response.

    Tweeter (OW2): Similar question as the mid regarding the RC in front of the XO. Another topology I don't see much. That cap lifts the top end though. Also, there's a bit of a diffraction peak at 7.5k. I'm going to put some felt on the grill frame - Its probably in the 1.75" from the tweeter range to tame it. DO you think the peak as it exists will be audible?


    Leave a comment:


  • jhollander
    replied
    Originally posted by Adam_M View Post

    Is it critical to model the comb filtering?

    If we're considering the woofers as a single unit, does that then change the definition for farfield measurements i.e. does the far-field measurement need to be at ~3x+ the total radiating dimension, or approximately 54" (~18" from the far radiating ends of the woofers) or would a more typical ~24" work, since each woofer independently will be radiating into the far-field at that point?

    A photo of the baffle, with integrated grill but without the magnets and cloth is attached.
    No, you don't need to model. You can model with dummy drivers in PCD if you want to test your x-o off axis. However, I would also measure off axis with either the raw drivers and/ or completed x-o. I want to avoid correcting a FR dip that fills in off axis.

    With Omni mic I'd start with 2x, 1 meter, and move the mic back to see if the peaking around 1,000 Hz changes. I think that 2x would be OK.

    Leave a comment:


  • Adam_M
    replied
    Thanks for all of the information - your point about measuring in the intended location is well made, but not directly applicable in this case. I'm actually building 2 of them, for 2 separate locations with rather different surrounding furniture and video equipment - so I was planning to design with full BSC (the front baffle in both cases will be 2+ feet from the back wall) although one will have a TV to cause "baffle step" where the second will be used in a front projection room and behave normally.

    Is it critical to model the comb filtering? The baffle is built and I've been fairly careful to design the baffle so that the woofer CTC spacing and the M-T spacing are less than 1 wavelength apart at their respective XO frequencies. I modeled with manufacturer data and off axis looked really good until the null showed up at around 45 degrees, which is plenty far enough off axis for these rooms.

    If we're considering the woofers as a single unit, does that then change the definition for farfield measurements i.e. does the far-field measurement need to be at ~3x+ the total radiating dimension, or approximately 54" (~18" from the far radiating ends of the woofers) or would a more typical ~24" work, since each woofer independently will be radiating into the far-field at that point?

    A photo of the baffle, with integrated grill but without the magnets and cloth is attached.

    Leave a comment:


  • daryl
    replied
    Originally posted by jhollander View Post
    Duh...

    Just reread this is a center channel.
    Sheesh...…

    I got the ball and took off in the wrong direction! (like Adam Sandler)

    Disregard what I said!

    I thought we were talking a large vertical WMTW.

    Just measure according to Jeff or jhollander's instructions with the woofers connected in parallel.
    Last edited by daryl; 12-28-2018, 11:12 PM.

    Leave a comment:


  • jhollander
    replied
    Duh...

    Just reread this is a center channel. Here are a few more comments. I like to measure the center channel where its intended to be used. This will get the diffraction correct for the location. It's OK to use the above method with the summation point of the two woofers as a single woofer, but the off axis sim in PCD will not work correctly.

    If you want to see the comb filtering effect use the same measurements as above then in the Response Blender decrement the woofer pair measurement by 3.01 dB, then extract min phase. Now all you need is the single woofer impedance file in half the volume. I use the old Response Modeler to create the impedance file for the single woofer in half the volume with the same tuning.

    Let me know if this makes sense. I may have assumed a few items along the way...

    Leave a comment:


  • jhollander
    replied
    Just assume the two woofers are one when they are equidistant from you measuring axis, usually the tweeter axis. In PCD use the single woofer, the Y (well no for a horizontal CC, X=0) offset is zero. This is the easiest when the woofers share the same cabinet.

    Wire them in parallel if that’s the final wiring and measure. Use the tweeter to set your SPL level. Measure each driver, do not move the mic. Then measure TW plus Mid and TW plus Woofer pair, don’t move the mic. With these measurements you can determine the offsets.

    I then move the mic and change the volume and take a near field measurements of the mid and one woofer. Use the Response Bender program to add the near field to the far field then extract min phase.
    Measure the woofer pair in box for your zma.

    Hope this helps. If this was a center channel and off axis was critical we would derive the impedance file for a single driver and adjust the spl of the driver pair for a single.
    Last edited by jhollander; 12-28-2018, 08:03 PM. Reason: spelling

    Leave a comment:


  • daryl
    replied
    You can do this inside too Adam,

    Just use the biggest room you have, get in the middle, move everything out of the way (no reflective surfaces!) and set your gating for the nearest surface not including the floor (you will be including floor effects in your woofer measurements, the coverings you place on the floor between the speaker and mic should reduce floor interaction at the high end of your woofer measurements).

    For the midrange and tweeter you can lift up the speaker to get the midrange and tweeter as far as possible from ALL room surfaces and measure from as close a distance as possible for the baffle edge diffraction and drivers to still sum close enough to their infinite distance sum for your purposes.

    You know a long time ago I worked at a large hotel and took my equipment into one of the ballrooms in the middle of the night.

    Was able to use nice long gate times and measure low!

    BTW a battery and inverter will allow you to make measurements outside.

    You could even use your car for the battery.

    Run the motor and shut it off during measurements.

    Leave a comment:

Working...
X