Need help interpreting measurements

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi To All,
I finally took the plunge and bought a measurement mic (Umik1) and after some setup headaches, I was finally able to capture my very first measurement:). Everything is set up using Room EQ Wizard. My lack of understanding is where do I gate the IR given the attached screenshots. All the examples I have seen, there is a logical spot to gate the measurement around 4-5 ms. Any explanations will help in understanding what I'm looking at or what I should be looking for.

Best Regards,
Rich
 

Attachments

  • ss18w-8434FR-screenshot.jpg
    ss18w-8434FR-screenshot.jpg
    83.4 KB · Views: 482
  • SS18W-8434IR-screenshot.jpg
    SS18W-8434IR-screenshot.jpg
    56.8 KB · Views: 473
That is actually a really clean impulse! :)

The goal is really to get the most USEFUL FR plot. :) If you are designing a 2-way, you want to gate low enough that your smoothing is below the expected crossover point.

As D'Appolito points out, we can kind of play with this by adjusting the distance as well. I mean, we normally use 1m, but there's no law that forces us to do this. It's a convention. If you have too much noise still, get closer! :)

Best,


E
 
In "impulse" graph, click the "limits" icon and set "Left s" to -0.002 and "Right s" to 0.008 (typical).

You may also want to set the "limits" in the SPL graph to 20 - 20.000 Hz and phase to 720 deg (top) and -180 deg (bottom), this to give you a more standardized look.

And, in "IR Windows" try this: check the "Add frequency dependent window" and select 6 cycles, and then hit "Apply windows". Select 1/48 smoothing. This will smooth the measured curve but still with plenty of useful data.

You can always revert back to whatever your current settings are.
 
Last edited:
Thanks Erik,
I have been following Jeff Bagby's paper on 'Accurate In-Room Response Measurements Down To 10 Hz.' His method is to get close as possible and still remain in far field. Thus 3-5 times the radiating diameter of driver being measured and at list twice the baffle width away in order to pick up baffle step correctly. So, In my case baffle width is 10 inches and radiating diameter of driver is 5.25 inches. Armed with that information, I setup up mic 22 inches from driver on tweeter axis (2 way floor stander) and elevated to 66.5 inches from floor. I'm measuring in a very large room with vaulted ceiling. Let me know if I'm overlooking anything.:) My issues right now are trying to get my head around REW. The Umik1 mic is plug and play but the software is not.:D

Best,
Rich
 
Thanks Draki,
That is exactly the kind of expert information I'm looking for.:D Anything that moves the learning curve ahead is much appreciated. I will re-visit measurements tomorrow using the settings that you recommended. I found that zooming in close to the IR helps to see the arrival of early reflections. Do you also set a gate time in the right window along with the 6 cycle setting?

Best,
Rich
 
Just another Moderator
Joined 2003
Paid Member
As Eric said it's a very clean impulse! Usually what I look for is the first prominent bump (indicating first reflection) in the IR

Below shows different gating for the same impulse (with the fourth being 1/8th octave smoothed.

The measurement was taken at 66cm distance outside.

You can see the first reflection around 4.5ms, the first graph shows the gate set a bit before this which gives the cleanest high frequency response. The next three show moving the gating out and subsequently increasing "hairiness" of the resulting measurement. I'm not sure if the software you are using allows you to move the gate window in real time (I'm using holm impulse and it does). I normally look for the first obvious anomily in the response and then play around + or - around that.

The last graph is using 1/8th octave smoothing for comparison.

Tony.
 

Attachments

  • one_eighth.png
    one_eighth.png
    55.2 KB · Views: 149
  • gate4.png
    gate4.png
    58 KB · Views: 143
  • gate3.png
    gate3.png
    56.6 KB · Views: 405
  • gate2.png
    gate2.png
    56.3 KB · Views: 419
  • gate1.png
    gate1.png
    56.1 KB · Views: 433
Thanks Tony,
Your detailed explanation helps immensely. The graphs you attached are very helpful. Question: can a person just use smoothing techniques rather then gating if near field measurements will be spliced in anyway? Is their advantages to one method or the other?

Best Regards,
Rich
 
AllenB,
Thanks so much for pointing me in the right direction!:D I did notice if I zoomed in on the IR there is brief space that is blank before the first reflection arrives. That I would assume is the cabinet effect you mentioned. I'm sure there will be many more questions along this journey.:) Fun stuff though.

Best,
Rich
 
Just another Moderator
Joined 2003
Paid Member
As Allen said the smoothing would be questionable. I've tried it and didn't get good results, I went back to gated with spliced in nearfield.

If you go down the splicing route do make sure you apply a baffle step function to the nearfield result to get the most accurate results. Some of Jeff Bagby's spreadsheets will help you out there :) I can't remember off the top of my head which one, as he has two that I have used.

Tony.
 
Do you also set a gate time in the right window along with the 6 cycle setting?

Best,
Rich

Further to Tony's response: below is one same measurement shown in REW' default left/right window setting of -125 (left) and 500 (right) in IR Windows, and then with right set to 2.3 ms, just before the aprx 2.5 bump visible on the step impulse (due to a very near reflection, to show the point).
When set to 2.3 ms, you can see the dotted line showing the window opening, and its' corresponding FR - no bumps below cca 450 Hz as the range below is smoothed, but 450 and up is correct and a good "far field" data.
Shown also the same in Holm (like Tony), the 2.5 ms peak and the window setting at 2.3 ms visible. The gating mark positioned automatically at abt 450 Hz.
Note: different scales, REW 40dB, Holm 70 dB, thence HOLM looks "better" ..
 

Attachments

  • REW 500.JPG
    REW 500.JPG
    116.9 KB · Views: 226
  • REW FR 500.JPG
    REW FR 500.JPG
    74.5 KB · Views: 214
  • REW 2.3.JPG
    REW 2.3.JPG
    114.5 KB · Views: 202
  • REW FR 2.3.JPG
    REW FR 2.3.JPG
    71 KB · Views: 203
  • REW HOLM 2.3.JPG
    REW HOLM 2.3.JPG
    173.3 KB · Views: 203
Last edited:
Just another Moderator
Joined 2003
Paid Member
Thanks for posting those REW screen shots with the settings shown, especially the %fs one!! I've never been able to get meaningful impulse responses in REW before today! :)

The first pic is after changing to % FS the second is with the default of dB FS. You can see why I thought it was useless as the 2nd one is what I always got!

Tony.
 

Attachments

  • IR1.jpg
    IR1.jpg
    54.2 KB · Views: 121
  • IR2.jpg
    IR2.jpg
    100.4 KB · Views: 123
Thanks for posting those REW screen shots with the settings shown, especially the %fs one!! I've never been able to get meaningful impulse responses in REW before today! :)

The first pic is after changing to % FS the second is with the default of dB FS. You can see why I thought it was useless as the 2nd one is what I always got!

Tony.

Nice clean step! 3-way, mid reversed I guess?
 
Just another Moderator
Joined 2003
Paid Member
Thanks :) I'm pretty sure it is just my MTM without the 10" woofers as the date on the measurement is Aug 2013 and I don't think I had my active crossover (to add in the woofers) going till 2016.

The Drivers should be close to time aligned, crossover is Bessel 4th order acoustic.

Tony.
 
Hi guys,
Thanks for the the Awesome tips! I will post a full array of measurements with new settings in place later today if you don't mind taking a look and making sure I'm on the right track before proceeding to the next step. Also, concerning baffle step what is the rule of thumb for amount to include for a small room? I would assume there is some art involved in the science of it all. With regard to room gain, it would seem the accepted 6db of baffle step is to much for small room. What are your guys thoughts?

Best Regards,
Rich
 
diyAudio Moderator
Joined 2008
Paid Member
on the IR there is brief space that is blank before the first reflection arrives. That I would assume is the cabinet effect you mentioned.
That's it. There should be a gap between the speaker and room. Measure your distances (the obvious first reflections in your room) to confirm it on the time scale. The cabinet edges are usually closer to the direct sound but sometimes it isn't clear.
Also, concerning baffle step what is the rule of thumb for amount to include for a small room? I would assume there is some art involved in the science of it all.
I'd say there is. If you must include a baffle step in your design then it depends on the choices you make in room placement.
 
Guys,
Got some measurements to critique. Hopefully, I'm on the right track and can start on putting files together for xover design.:) Another question I have is it correct to unwrap phase in REW before export in .frd file? The attached screenshots are with phase unwrapped.

Best,
Rich
 

Attachments

  • woofer+tweeter_FR+phase.jpg
    woofer+tweeter_FR+phase.jpg
    58.2 KB · Views: 73
  • woof+tweeter_IR.jpg
    woof+tweeter_IR.jpg
    52.1 KB · Views: 89
  • sSD2604_fr+phase.jpg
    sSD2604_fr+phase.jpg
    55.5 KB · Views: 89
  • SSd2604-83300_IR.jpg
    SSd2604-83300_IR.jpg
    49.9 KB · Views: 97
  • SPL+phase_SS18W-8434.jpg
    SPL+phase_SS18W-8434.jpg
    57.9 KB · Views: 99
  • impulseresponse_ss18w-8434.jpg
    impulseresponse_ss18w-8434.jpg
    55.8 KB · Views: 109
^Here is the beginning of xover design from measurements above. I used PCD7 to find correct offsets then imported summed response into Xsim and adjusted delay until lines overlayed perfect (1.71 in.). Hope I'm going about this the proper way. I welcome any comments that will help me learn at little more about measurements and xover design.:D

Best,
Rich
 

Attachments

  • 2way_xover.jpg
    2way_xover.jpg
    34.9 KB · Views: 107
  • 2way_FR.jpg
    2way_FR.jpg
    101.3 KB · Views: 111
  • 2way_Z+phase.jpg
    2way_Z+phase.jpg
    85.2 KB · Views: 100
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.