Manually Aligning Impulses

[Dan Twomey]

This is more of a statement than a question. As highlighted in one of the Rephase tutorials and some posts here. It seems to be a good practice to manually align the impulses for a multi-measurement average of your speakers.
It's well beyond the scope of my knowledge but having tried the practice it does produce very different results. More later.............

Regards,
Dan

 

[Dan Twomey]

Strictly for the sake of conversation....

 

[John Mulcahy]

It isn't very meaningful to post a set of responses without explaining whether a timing reference was used to produce them, if so what kind of timing reference, and how the measurements were made - same mic position, different mic positions, same speaker, different speakers, ...

If the mic is being moved between measurements it makes more sense not to use a timing reference, since a moving reference isn't a reference at all.

 

[Dan Twomey]

It isn't very meaningful to post a set of responses without explaining whether a timing reference was used to produce them, if so what kind of timing reference, and how the measurements were made - same mic position, different mic positions, same speaker, different speakers, ...

If the mic is being moved between measurements it makes more sense not to use a timing reference, since a moving reference isn't a reference at all.

Way beyond the scope of my knowledge but I'll do my best to explain. I was basically following the Rephase Additional Tutorial as seen here. RePhase Additional Tutorial: 9-point measurements

My interim goal was to produce the best averaged response using SwissBear's Tutorial seen here. making average measurements in REW


Regards,
Dan

 

[Dan Twomey]

My measurements attached here.

 

[Dan Twomey]

A pic of what I'm trying improve!

Regards,
Dan

 

[Dan Twomey]

Another interesting note I found in another tutorial is that 'zero' on the impulse graph should be in the middle of the largest peak.

If the largest peak is not on top at the time zero, go to the options menu shown above, and go under “time zero” and select “largest peak” and click “autodetect.” This should center the impulse on the time zero. If all the impulses you measure are upside down, you may need to check the “invert” box; if only some measurements are upside down, leave them be. It's important to get your time-zero lined up or your phase results will be wonky. You should check on this for every measurement (the “auto” usually gets it right, but it's worth confirming).

Taken from here.....Topic: Guide to Speaker/Room Correction Using Freeware and JRiver

Regards,
Dan

 

[John Mulcahy]

Another interesting note I found in another tutorial is that 'zero' on the impulse graph should be in the middle of the largest peak.

The peak isn't the correct zero time from a phase perspective but can act as a reference for measurements of loudspeakers that cover the full frequency range. It would be very wrong for a subwoofer,

 

[Vladimir_1953]

The peak isn't the correct zero time from a phase perspective but can act as a reference for measurements of loudspeakers that cover the full frequency range. It would be very wrong for a subwoofer,

How should the reference plots of the impulse response and phase of a 3-way speaker system with a subwoofer look like?
Please provide examples of reference graphs that you need to strive for.
It will be useful for everyone.
Thanks in advance

 

[jtalden]

Dan,
It's probably best to continue to work with the very skilled diyAudio contributors. Two 2 sources of support may be more confusing than helpful. That said, if you want me to take a first look at your situation also. Please provide the following info.

mdat file with 5 Measurements:

1. L (alone; mute the other speakers)
2. R (alone; mute the other speakers)
3. SW (alone; mute the other speakers)
4. L+SW (left channel)
5. R+SW (right channel)

[The first 3 measurements is all that is needed if the measurements are made with REW acoustic timing or loopback timing active.]

A measurement mic should be placed at the LP. It should not be in front of a high back leather chair. Above a fabric covered couch or low back chair is okay, but it would best to move them aside a little. Point the mic forward toward the TV and use the calibration file for 0° if you have one. The sweep range should be 12-20k Hz for all 3 measurements. How far away will the mic be from the back wall? Is the LP centered on the TV or is it offset from center as the photo is?

 

[Dan Twomey]

Dan,
It's probably best to continue to work with the very skilled diyAudio contributors. Two 2 sources of support may be more confusing than helpful. That said, if you want me to take a first look at your situation also. Please provide the following info.

mdat file with 5 Measurements:

1. L (alone; mute the other speakers)
2. R (alone; mute the other speakers)
3. SW (alone; mute the other speakers)
4. L+SW (left channel)
5. R+SW (right channel)

[The first 3 measurements is all that is needed if the measurements are made with REW acoustic timing or loopback timing active.]

A measurement mic should be placed at the LP. It should not be in front of a high back leather chair. Above a fabric covered couch or low back chair is okay, but it would best to move them aside a little. Point the mic forward toward the TV and use the calibration file for 0° if you have one. The sweep range should be 12-20k Hz for all 3 measurements. How far away will the mic be from the back wall? Is the LP centered on the TV or is it offset from center as the photo is?

"Two 2 sources of support may be more confusing than helpful."
A valid point that reminds me of an old quote, "Beggars can't be choosers!" Very perceptive of you! As a rookie climbing a very steep learning curve I was taking 'guidance' from wherever I could find it.

That aside, I'd like to accept your very generous offer with one modification. I've found it relatively easy to integrate the sub with these two speakers so all my measurements to date have been without the sub.

1.) I can move the furniture to place the mic in an uncluttered measuring position.

2.) I have the 0 degree calibration file.

3.) Sweep from 12hz to 20Khz?

4.) The mic is about 10 feet from the back wall

5.) The LP is perfectly centered (equidistant) between the two speakers.

Regards,
Dan

 

[jtalden]

The SW response is not needed then. The sweep range can just cover the mains range. If the measurements in your chart above already contain individual left and right mains at the LP you could just post an mdat containing those 2 measurements. I just wanted to see if some of the irregular looking impulse is due to L+R measurement interference or if it was in one or both channels individually. There is also a slight possibility that there will be some other hint of a problem in the mdat file.

It is impossible to just look at an impulse chart showing 1 ms alone and determine much of anything. Thus the request for an mdat file.

John pointed out above that it is better to use a simple average of listening area measurement for EQ purposes. This should be done for each channel individually. He also indicated that the impulse position alone cannot be used to accurately determine favorable XO timing. I haven't followed the diyAudio discussion closely, but also noted the impulse shape irregularity and after seeing you speakers and room setup wondered if there was a measurement irregularity of some sort, hence my request for an mdat file.

 

[Dan Twomey]

JT,

While I'm working on your exact request. Attached here is L1 and R1 which are 'close' to what you requested. Also, attached is a very quick sweep of a single speaker I just bought. They

are the Zu Audio Dirty Weekend as seen here...
Omen Dirty Weekend [Mk.II]

 

[Dan Twomey]

The SW response is not needed then. The sweep range can just cover the mains range. If the measurements in your chart above already contain individual left and right mains at the LP you could just post an mdat containing those 2 measurements. I just wanted to see if some of the irregular looking impulse is due to L+R measurement interference or if it was in one or both channels individually. There is also a slight possibility that there will be some other hint of a problem in the mdat file.

It is impossible to just look at an impulse chart showing 1 ms alone and determine much of anything. Thus the request for an mdat file.

John pointed out above that it is better to use a simple average of listening area measurement for EQ purposes. This should be done for each channel individually. He also indicated that the impulse position alone cannot be used to accurately determine favorable XO timing. I haven't followed the diyAudio discussion closely, but also noted the impulse shape irregularity and after seeing you speakers and room setup wondered if there was a measurement irregularity of some sort, hence my request for an mdat file.

The three MDAT files I just posted should help shed some light on what is going on? ;-)

Regards,
Dan

 

[jtalden]

I just looked at the file and additional measurements are not needed. The unusual impulse response is also apparent on the 'Dirty' speaker and that rules out several things.
My initial guess is that there is a mismatch between the 48kHz mic capture rate and the ?? kHz output from the Realtek speaker/headphone output. Check the windows speaker output settings and confirm that it also is set to 48 kHz.
I don't use a USB mic and am thus not familiar with the possible setup issues. If that is not the problem then someone else will need to help troubleshoot.
This issue is not directly related to your initial questions that have been answered, I do expect it may impact the setting of favorable XO timing and it will at least clean up several of the various charts that REW creates from the impulse response.

 

[jtalden]

I should explain that my attention was drawn to the excessive phase rotation in the mid-woofer range of the response. I tended to rule out the line array itself when I saw the same phase characteristic with the 'Dirty' speaker. That made me wonder if a sampling rate difference could cause this.

 

[Dan Twomey]

I should explain that my attention was drawn to the excessive phase rotation in the mid-woofer range of the response. I tended to rule out the line array itself when I saw the same phase characteristic with the 'Dirty' speaker. That made me wonder if a sampling rate difference could cause this.

Will check my setup and report back. Thanks for your input!

Regards,
Dan

 

[Dan Twomey]

I turned off all the 'enhancements' for the audio software on my laptop and checked the sample rates. My laptop audio card suggests 16bit 48000 when I think REW assumes 24 bit 48000? More testing tomorrow.

Regards,
Dan

 

[jtalden]

Sample rate is not the problem then. The bit depth is not an issue. Enhancements should definitely be off, but I don't know if any of them could cause this particular issue.

 

[John Mulcahy]

Clock rate error perhaps, try measuring with the Analysis option "Adjust clock with acoustic ref" selected.

 

[Dan Twomey]

I am puzzled. Everything amplitude and eq related seems to work perfectly. It’s seems that phase is out by a mile and both of you see something off in my impulses.

John,

I’m going to try your suggestion and report back.

Regards,
Dan

 

[Dan Twomey]

Unfortunately, "Adjust clock with acoustic ref" was already selected. I turned off decimate IR and made a couple of casual measurements.
The first is the Zu Audio, the second is the left line array.

 

[Dan Twomey]

Forever curious. I was looking at the MDAT for the left line array posted above. After importing it into Rephase I notice that the phase was well with 'correctable' range for 2khz and above.
This varies greatly from my 'averaged' measurements which were 1400 degrees and greater out of phase.

Hmmmmm............

Regards,
Dan

 

[jtalden]

Today's left array measurement finds the excessive phase rotation, IR and step response appearing much improved. The phase rotation is about 1/2 of what it was. Do you know what changes were made? The SPL looks like the mic position was changed a little? Even though the measurement is much improved, I believe the issue is still not fully resolved as I would expect the line array to have more like maybe 90° phase rotation from 100-5000 Hz. See charts below.

Is all unnecessary processing in the AVR is turned off? Are you using stereo mode? Do you have a direct mode choice that bypasses all AVR processing? Is the power amp one that has a internal processor and if so, is all unnecessary processing turned off? Is the DCX is handling the XO for the towers and the AVR the SW XO, or is the DCX doing both XOs? EQ is off in all boxes for this testing?

I really wouldn't expect mode selection, XOs or EQ to be related to this issue. A careful check of all the other digital processer options in the chain may find some setting that does.

A cable loopback test from the DCX left mid-woofer output back to the laptop Realtek input could rule out any processing issues concerning the soundcard, AVR or DCX.

No need to answer all these question unless you find an issue. It is just my thinking it case it is helpful in reviewing the setup to look for any anomalies.

The option is to see of this improvement is repeatable and if so just move forward to complete the setup.

I just noticed that the mic input level for this new measurement is greatly higher than the previous one (-9 vs -31 dB). The chart SPL is about the same however. I don't understand how that would happen. Was the sound level of the sweep 20 dB louder this time? If it was 20 db different was one of the traces manually offset to match the other after measuring? The mic calibration file should include the mic sensitivity so there should have been a 20+ dB SPL difference. I don't see how a mic or mic connection problem would cause the IR issue however.

I did not normalize the Impulse and step charts to emphasize the difference in signal level.

 

[jtalden]

Forgot to attach SPL