Acoustic phase Measurements

Gmauro · Feb 22, 2024

Hi to all.
I need your help to make a proper measurments of acoustic phase.
Currently my 3-way system allows me to move all the speakers back and forth individually, but what interests me now is to be able to measure and evaluate the acoustic phase between mid and tweeter. I work like this:
- Dayton UMM6 USB microphone
- microphone at medium height between mid and tw (approximately 1 m), at a distance of 1 m.
- measurement with active "use acoustic timing reference"
- I only measure the mid (the beep you hear at the start of the swipe is obviously at a low level, but still recognized by the microphone and rew)
- I access the "impulse" window and see the first peak after the ref peak after about 3-3.5 ms
- in the window management panel I set "right windows" to the value just before the peak of the first reflection, therefore 3-3.5 ms and apply, thus obtaining the windowed measurement and the acoustic phase.
- I detach the mid and attach the tw, repeat the measurements as above, obtaining the data of only the tw. I go into the impulse, establish the windowing time and apply, I find the tw phase.
- in this way I believe (I thought) I had found the acoustic phases of the 2 individual drivers, and by moving the tw back mm by mm find the position of the best phase condition between the 2 drivers.
- Instead, if I repeat the measurement of the tweeter several times (without moving anything, but only by repeating the measurement), I never get the same result, but the phase also varies by 80-120°
What am I doing wrong? what's wrong? can you please help me? I can't sleep at night on this anymore....
Thank you all!

John Mulcahy · Feb 22, 2024

You need to send the timing reference to a speaker that has a tweeter, the timing reference output is independent of the measurement output.

When you have set up a stable timing reference attach an mdat with some example measurements.

Edit: Also make sure to enable "Adjust clock with acoustic ref" in the Analysis preferences.

Gmauro · Feb 22, 2024

Thanks John for your 1st reply! do you mean that the good way is to measure first the tweeter (to keep good and valid timing reference), and use the same value also for mid measurements and windows timing setup?
This evening i made some new test, and i post some mdat file.

John Mulcahy · Feb 22, 2024

No, use a separate speaker for the timing reference, placed nearby. It doesn't need to be anything special, just to have a tweeter.

Gmauro · Feb 22, 2024

John Mulcahy said:
No, use a separate speaker for the timing reference, placed nearby. It doesn't need to be anything special, just to have a tweeter.

Ah.... it's ok also only one tweeter placed near to the measured speakers?

John Mulcahy · Feb 22, 2024

Gmauro · Feb 22, 2024

John Mulcahy said:
Yes.

Ok John, this is first infos about my new setup:
- R channel connected only with a tweeter for timing reference
- L channel connected to mid-hig speaker need to measure
- Attached pics, mdat file, screen of my setup
Note. tweeter used for timing reference is 20 cm higher than midrange and 30 cm backward, also not on axis to microphone
I note that the delay is negative (see screen 4), and now i'm in doubt about what is the right value to put into right and left windows or other data insertion.
Please tell me the right way and where i'm wrong, i've completly lost my night sleeping for this!!
Thanks a lot
Mauro

John Mulcahy · Feb 22, 2024

I added a 1/21 octave frequency dependent window to each measurement to clean up the responses, then used the Alignment tool in impulse alignment mode and aligned at 3 kHz which produces the result attached. That says to invert the tweeter and delay it by 0.24 ms (82 mm), which perfectly aligns their phases through the transition region between them.

Gmauro · Feb 23, 2024

Wow!! this is very interesting, could you help me understand step by step how obtain this results? Of course i will try by myself this evening, when come back to home, but for the moment, this function si completly new for me!
I will update you soon with new test and attached!
Thanks John!

John Mulcahy · Feb 23, 2024

There isn't much to it:

Select the "Add FDW" box on the IR windows dialog for each measurement with a setting of 1/21 octaves or 15 cycles, whichever mode is displayed
Open the All SPL graph, click the Actions button, then click the Alignment tool button
Select the mid measurement and the tweeter measurement in the drop-down boxes
Select the mode as "Impulse alignment" at the top right
Put the graph cursor at 3 kHz
Click the "Align IRs at cursor" button

Gmauro · Feb 26, 2024

John Mulcahy said:
There isn't much to it:

Select the "Add FDW" box on the IR windows dialog for each measurement with a setting of 1/21 octaves or 15 cycles, whichever mode is displayed

Open the All SPL graph, click the Actions button, then click the Alignment tool button

Select the mid measurement and the tweeter measurement in the drop-down boxes

Select the mode as "Impulse alignment" at the top right

Put the graph cursor at 3 kHz

Click the "Align IRs at cursor" button

Good morning John, during this weekend I did several measurement tests, I optimized the use of the additional tweeter for the acoustic reference by placing it on a slimmer support and closer to the speakers to be measured. Speakers to measure on the L channel, tweeter for ref timing on the R channel.
I still have the problem, because the measurements are not repetitive nor do they correspond to the variations made. if I repeat 5-6 measurements of the tweeter and 5-6 of the midrange without moving anything, but only by repeating the measurement, the results of the acoustic phase also vary greatly. If I go to the impulse window, the impulse is always well repeated, but the results are not.
So even with the "alignment tool" function, the results are always different, even by several tens of mm.
Furthermore, even the step response does not seem correct, in fact it never changes, whether you move all the speakers forward or backward, no variations are presented. (obviously my 3-way system allows the movement of each individual driver)
Do you have any advice for me about this problem? What could it be? the microphone not ideal? computer latencies or sound card? any windows settings? I do not know what to think.
If you thin are useful, i will attach some dat file of this measurments.
Thanks again for the support

John Mulcahy · Feb 26, 2024

Gmauro said:
If you thin are useful, i will attach some dat file of this measurments

Better do that, no way to comment otherwise.

Gmauro · Feb 26, 2024

John Mulcahy said:
Better do that, no way to comment otherwise.

file recovered by remote. Thanks

John Mulcahy · Feb 26, 2024

That's concerning. Your tests were with UMM-6. I also did a series of measurements with UMIK-1 and UMIK-2. UMIK-2 differs from UMIK-1 (and, I believe, UMM-6) by having an internal clock source rather than deriving its clock from the USB clock.

Here are your UMM-6 measurements of the tweeter:

UMIK-2, by contrast, delivers extremely stable results:

As the UMIK-1 gets its clock from the USB hub I tried two ports on two different hubs, first one of the PC's front panel ports, with similar results to yours:

I then tried another hub and got results that showed smaller spreads but clustered into 3 groups, 5 with one delay value, 3 with a slightly higher delay (+3 mm) and 2 with the highest delay (+13 mm):

I think we can conclude from this that USB mics that use an internal clock, like UMIK-2, deliver very stable timing. USB mics that rely on the USB clock, like UMIK-1 and presumably UMM-6, are at the mercy of the stability of the USB clock.

sm52 · Feb 26, 2024

A very interesting study, John. Does UMIK-2 have an internal clock generator?

Gmauro · Feb 26, 2024

John Mulcahy said:
That's concerning. Your tests were with UMM-6. I also did a series of measurements with UMIK-1 and UMIK-2. UMIK-2 differs from UMIK-1 (and, I believe, UMM-6) by having an internal clock source rather than deriving its clock from the USB clock.

Here are your UMM-6 measurements of the tweeter:

View attachment 68718

UMIK-2, by contrast, delivers extremely stable results:

View attachment 68717

As the UMIK-1 gets its clock from the USB hub I tried two ports on two different hubs, first one of the PC's front panel ports, with similar results to yours:

View attachment 68720

I then tried another hub and got results that showed smaller spreads but clustered into 3 groups, 5 with one delay value, 3 with a slightly higher delay (+3 mm) and 2 with the highest delay (+13 mm):

View attachment 68721

I think we can conclude from this that USB mics that use an internal clock, like UMIK-2, deliver very stable timing. USB mics that rely on the USB clock, like UMIK-1 and presumably UMM-6, are at the mercy of the stability of the USB clock.

...! this is not good, but found this test results is a very useful thing to keep right way to fix a problem. Thank again John!

John Mulcahy · Feb 26, 2024

sm52 said:
Does UMIK-2 have an internal clock generator?

Yes.

Gmauro · Feb 27, 2024

John Mulcahy said:
That's concerning. Your tests were with UMM-6. I also did a series of measurements with UMIK-1 and UMIK-2. UMIK-2 differs from UMIK-1 (and, I believe, UMM-6) by having an internal clock source rather than deriving its clock from the USB clock.

Here are your UMM-6 measurements of the tweeter:

View attachment 68718

UMIK-2, by contrast, delivers extremely stable results:

View attachment 68717

As the UMIK-1 gets its clock from the USB hub I tried two ports on two different hubs, first one of the PC's front panel ports, with similar results to yours:

View attachment 68720

I then tried another hub and got results that showed smaller spreads but clustered into 3 groups, 5 with one delay value, 3 with a slightly higher delay (+3 mm) and 2 with the highest delay (+13 mm):

View attachment 68721

I think we can conclude from this that USB mics that use an internal clock, like UMIK-2, deliver very stable timing. USB mics that rely on the USB clock, like UMIK-1 and presumably UMM-6, are at the mercy of the stability of the USB clock.

Do you know if exist other brand for USB microphone with inernal clock, usable with rew? Any suggestion?

John Mulcahy · Feb 27, 2024

I only know that is the case for UMIK-2, I'm not familiar with the USB microphone market as a whole.

Gmauro · Mar 8, 2024

John Mulcahy said:
I only know that is the case for UMIK-2, I'm not familiar with the USB microphone market as a whole.

Finally i've received my UMIK-2 to do more test (hoping with a good results). But i've a doubt, UMIK-2 require some special setup in rew or windows? or is plug and play, of course with is calibration file?

John Mulcahy · Mar 8, 2024

It is plug and play.

Gmauro · Mar 15, 2024

John Mulcahy said:
It is plug and play.

HI John, some trouble again. After many test with new umik-2 microphone, i confirm that now all delay measurements are the same, they are repetitive if made without move anything, BUT if i made measurments to find delay between mid and tw and i find how many mm (ms) need to move backward or forward and i move this amount of mm, the new measurements are completly wrong.
For example:
i find tha my tw need 24 mm of movement, i move it 24 mm and i dont find the correct alignment (or 48 mm if moved in wrong direction), but i find completly stange value, for example -73 mm and to invert phase. I attach some file made without move anything.
Please aask me if you need to see some or different data, or i i need to made particular test
Thanks for all support.
MAuro

John Mulcahy · Mar 16, 2024

Hard to comment much since you just provided repeated measurements of the same positioning. You should not use FDW when your impulse window reference time is not at the IR peak. The 1/6 smoothing is just hiding the ragged nature of the measurements. For the mid a right side window of about 3.3 ms (with the ref at the peak) excludes the first reflection causing much of the response irregularity. For the tweeter that reflection can be excluded with about 5 ms right window. Tweeter polarity is inverted so probably worth swapping around the wiring on that. Try measuring at some different tweeter positions (10 mm increments, say) to see how the suggested delays compare. The reference speaker, mid and microphone must stay in the same place.

serko70 · Sep 21, 2024

John Mulcahy said:
Yes.

Would you recommend unticking "adjust clock with acoustic reference" with Umik-2?

John Mulcahy · Sep 21, 2024

No. It has a more stable clock but it is still likely to differ from the output clock.

Acoustic phase Measurements

Registered

REW Author

Registered

REW Author

Registered

REW Author

Registered

Attachments

REW Author

Registered

REW Author

Registered

REW Author

Registered

Attachments

REW Author

Member

Registered

REW Author

Registered

REW Author

Registered

REW Author

Registered

Attachments

REW Author

Member

REW Author

Popular tags