Home Theater Forum and Systems - HomeTheaterShack.com

Home Theater Forum and Systems - HomeTheaterShack.com (https://www.hometheatershack.com/forums/)
-   REW Forum (https://www.hometheatershack.com/forums/rew-forum/)
-   -   Aligning Driver Phase - REW v5 Example (https://www.hometheatershack.com/forums/rew-forum/38617-aligning-driver-phase-rew-v5-example.html)

jtalden 01-25-11 06:25 PM

Aligning Driver Phase - REW v5 Example
 

19 Attachment(s)
I saw a case study that provided good information regarding this subject for aligning the SW to Main speakers. The method I use for aligning a MR (aka MW) to a TW is somewhat different so here it is.

This example uses a DCX2496 for XO and driver delays, but the process will be similar using other units such as the MiniDSP. The XO between MR and TW drivers is set at 2 kHz. This method works because a near field mic location can be used. [See notes regarding aligning SW to Main Speakers below.]

Pic 1 Speaker/Mic Setup:
Attachment 27254

Mic Settings:
> Mic Distance: (about 1 m Ė not critical) Close enough to avoid multipath in the XO range and far enough back to get mic alignment reasonably accurate.
> Mic Alignment: The mic should be on the line of sight from the LP (main Listening Position) to the midpoint between the centers of the MR and TW.
> Mic Calibration: Generally not important because the SPL is not the issue and the phase of most mics is pretty flat in the range of most XOís.

[The XO filters types and slopes in this example are; TW is Butterworth 12 dB/Oct, and MR is Bessel 24 dB/Oct. Both are set to 2 kHz. There is no best choice for XO filter settings as there is no agreement among the experts. Iíve tried dozens of XO filter combinations. This one is my current setup as it provides moderately steep XO filters and can still provide good phase alignment.]

REW Setting:
> Setup and Activate ďUse Loopback as Timing ReferenceĒ

WE HAVE TO START SOMEPLACE
I arbitrarily set both TW and MR delay to 1000 mm (2.91 ms) to start. The MR is on output channel 3 is and the TW is on output channel 4.
[Weíre only after the relative delay between the two drivers so this provides plenty of room make either positive or negative delay adjustments.]

PROCESS
> Mute the MR then measure the TW

Fig 1 DCX Remote screenshot below:
Attachment 27255

> Mute the TW then measure the MR

Fig 2 SPL Overlay of MR and TW:
Attachment 27256

Note that the XO range extends from about 800 - 5k Hz to provide -25 to -30 SPL reduction.
The acoustic XO is about 1.8k Hz even though the electrical XO is set at 2 kHz.

Fig 3 Impulse Overlay of TW (red) and MR (green):
Attachment 27257

The impulse peaks do not align closely with the delays set equally in the DCX. The MR is delayed about 0.27 ms (90 mm).

For this example I will assume that we are looking to minimize the total phase rotation while keeping the phase differential in the XO band within 90 deg.

> So, letís start by reducing the delay of the MR by 0.27 ms (90 mm), as found above, to match the IR peak of the TW and take another MR measurement.

Fig 4 Adjusted Delay Ė MR to 910 mm
Attachment 27258

Fig 5 Adjusted MR Impulse (blue) now aligns with the TW peak:
Attachment 27259

The phase cannot be viewed yet because the delays of the impulses are almost 68 ms. The impulse needs to be near 0 ms to see the minimum phase rotation. I strongly recommend that the TW is taken as reference driver for identifying the exact timing offset to use.

Fig 6 Delay of the TW - REW reports it to be 67.81 ms:
Attachment 27260

> We must maintain the same relative timing of MR and TW to view the relative phase so we need to offset both drivers by the same 67.81 ms (important).

Fig 7 Impulse of TW and MR both offset by 67.81 ms (now near 0 ms).
Attachment 27261

Fig 8 Wrapped Phase:
Attachment 27262

> Now letís unwrap the phase. The XO setting is 2 kHz so it is necessary to place the cursor near that range. Anyplace between 1.8k and 5k will actually work in this case. We must avoid over 7k where the MR phase is lost because the MR measurement in that range is buried in the noise floor and both the SPL and phase look chaotic. There is a similar issue with the TW below 600 Hz for the same reason. Also, note the phase wrap that occurs around 1.4k to 1.8k, it is necessary to stay above this level as well, otherwise the relative phase of the MR will be offset 360 deg when the unwrap occurs.

Fig 9 Unwrapped Phase:
Attachment 27263

We can now more easily see that the MR and TW phases do not align well through the XO region, 800-5k Hz. There is about a 170 deg mismatch at 2k. If the timing/phase alignment was left this way there would be major SPL cancelation in the XO range. We could also say that the acoustic phase XO is near 5k when it needs to be around 2k.

> Since the phase error at the target XO is almost 180 deg, letís first just invert the polarity of the MR to see what happens. I donít see a facility do this in REW for an existing measurement so letís use the DCX to invert the driver signal and then measure again.

Fig 10 DCX Invert the MR polarity.
Attachment 27264

Fig 11 Impulse (again after applying the 67.81 ms offset):
[It would be convenient if REW provided the means to set a fixed offset for new measurements as this is an iterative process requiring manual offset of each new measurement.]
Attachment 27265

Fig 12 Phase (after unwrapping):
Attachment 27266

This looks much better now with acoustic phase alignment at about 1.8 kHz.

[Note that the phase curves cross such that the phase difference is about 180 deg at 5k (just at the extreme of the XO region where the MR SPL is about -30 dB). At 800 the difference is about 90 deg. We could further fine tune the acoustic phase XO point a little higher by adjusting the DCX MR delay. The purpose would be to better equalize the phase error at the extremes of the XO range. Moving the acoustic phase XO point up to maybe 2.5k would make the relative error at 800 and 5k a little more balanced. [I will not show this adjustment as you should now know how to do it. Again, always remember to apply the 67.81 ms IR offset to any new MR measurement.]

Fig 13 SPL response of TW, MR and TW+MR (smoothed to 1/24 for clarity):
Attachment 27267

Note the expected good reinforcement of SPL around 2k and the expected minor very reduction in SPL from 3.5k to 5k where the phase differential exceeds 90 deg.

Fig 14 Phase response of TW, MR and TW+MR (smoothed to 1/24 for clarity):
Attachment 27268

Fig 15 Impulse response TW+MR:
Attachment 27269

Fig 16 Step Response TW+MR:
Attachment 27270

Fig 17 Group Delay TW+MR:
Attachment 27271

Notes:
> The alignment as it stands here is pretty good and generally meets the target for SPL reinforcement and phase smoothness, but it does not meet the stated alignment criteria of retaining the relative phase within 90deg at the extents of the XO range. Changing the MR delay to get the acoustic phase XO to around 2.5 kHz will balance the top and bottom of the range a little better, but it would still not meet the 90 deg criteria.

> To provide better relative phase agreement you will discover it is necessary to delay the MR about 1/2 WL (about 0.28 ms) and change the MR polarity back to positive. This actually puts the setting back to where we started (just happenstance). The phase will now will now be more parallel and will closely track throughout the XO. The tradeoff is more total phase rotation of the system. The total phase rotation of the example alignment is about 360 deg from 100 Hz (my SW XO point) to 22 kHz. With the MR polarity set to positive and delay increase 0.28 ms the total phase rotation will be about 450 deg over this same range. Again, there is differing opinions as to the relative merits of these optional alignments.

> SW to mains phase alignment can sometimes use the same process.
- The mic should be at the main LP
- For multiple SWs: Delay the nearer SW(s) to match the delay of the most distant SW prior to starting this phase alignment process.
[Also note; there is much greater opportunity for multipath in mid/far field measurements that can make it very difficult to interpret the phase response correctly. The reflections can result in REW showing numerous rapid 360 phase rotations that obscure the direct phase. The severity of this issue depends greatly on your particular situation. I am not able to use this method from my LP using REW.]

> The default IR Window works fine for near field mic setups like this example and they do not help with LP mic setups.

> We may be able to align the SW to Mains by using the Group Delay feature. I looked at this and it does show some promise. However, it is not completely clear to me how to set the criteria and avoid misleading results. I have not yet found this method discussed in detail.

> If you want to use this example method and canít get a clean phase response at the LP using REW you may want to try HOLMImpulse as it provides a cleaner phase signal in the presence of multipath. Smoothing also improves the situation using HOLMImpulse, but makes it worse using REW. I much prefer REW in for all my EQ and other work, but still use HOLMImpulse occasionally for LP phase alignment work.

Fig 18 Phase Response SW, MR and TW, Mic at the LP, Front Right Speaker, HOLMImpulse:
[The target was to aligned phase closly through the 100 and 2k Hz XOís (at the expense of greater system phase rotation)]
[These curves were smoothed; 1/6 Oct for SW, and 1/4 Oct for both MR and TW.]
Attachment 27272

I hope you find this information helpful.

Jason_Nolan 01-25-11 08:07 PM

Re: Aligning Driver Phase - REW v5 Example
 

Very interesting.

A little above what I can understand at the moment, but I'm going to bookmark this so I can look at it more when I have more time.

laser188139 01-25-11 08:29 PM

Re: Aligning Driver Phase - REW v5 Example
 

Hi John,

It looks as if you are having as much fun as I did looking into how to set the sub distance/timing. It is interesting to see how, as I found with the sub, the limited frequency response of the MR affects the shape of the impulse curve.

With my front speakers, I replaced the original crossovers with better ones, but they were still passive. As in your situation, I found that the acoustic crossover was not necessarily in the same place as the designed electrical crossover.

You mentioned that you wanted to try inverting the phase, but did not find where this option was in REW. In the Impulse view Controls, there is an option Invert Impulse. Checking / unchecking this box inverts the impulse that REW uses, and so inverts the displayed Phase. (You can see this by opening the Phase plot in the Overlay window, while manipulating Invert Impulse in the impulse Controls.)

I was curious how the MR and TW separate impulse curves overlaid, after you made the 0.28ms adjustment in the MR timing. You initially set the two curves such that the peaks were at the same time; I am wondering how the leading edges of the curves align after your final adjustment.

Regards,
Bill

jtalden 01-25-11 10:13 PM

Re: Aligning Driver Phase - REW v5 Example
 

1 Attachment(s)
Quote:

laser188139 wrote: (Post 341752)
It looks as if you are having as much fun as I did looking into how to set the sub distance.

Yes, that thread helped me decide to finally put this one together. I had been thinking about it for awhile.
It looked to me like your final alignment would be very good as your final IR curves overlayed as I would expect to get a good handoff. I do like to see the phase curves though to be sure. I need go through that thread again more thoroughly as I think there were several points that I missed.

Quote:

It is interesting to see how, as I found with the sub, the limited frequency response of the MR affects the shape of the impulse curve.
I don't know if I follow your thought unless it is just that the IR shape is dependent on the FR of the driver. That is true.

Quote:

You mentioned that you wanted to try inverting the phase, but did not find where this option was in REW. In the Impulse view Controls, there is an option Invert Impulse.
Thanks! I thought I had seen this option before, but couldn't find it when I needed it. I guess it was too obvious. :R

Quote:

I was curious how the MR and TW separate impulse curves overlaid, after you made the 0.28ms adjustment in the MR timing. You initially set the two curves such that the peaks were at the same time; I am wondering how the leading edges of the curves align after your final adjustment.
The initial alignment (see Fig 3 of my original post) is very close to my current or "final" alignment. My current MR DCX delay is only 0.01 ms shorter than the DCX delay for the TW for both my FL and FR speakers (the CC MR delay is slightly shorter than that at 0.05 ms less than the TW because that it is about 8" lower). Below is the resulting Phase handoff for that alignment. The phase overlays well throughout the 800 - 5k XO range. Of course a different choice of XO filters results in a differ timing requirement so it is only happenstance that the correct offset in my case is so near 0.

Attachment 27277

The REW phase handoff above matchs up well with the HOLMImpulse measurement shown as Fig 18 above.

paulspencer 01-26-11 05:53 AM

Re: Aligning Driver Phase - REW v5 Example
 

It would be interesting to see how auto align compares to this manual process.

jtalden 01-26-11 09:22 AM

Re: Aligning Driver Phase - REW v5 Example
 

Quote:

laser188139 wrote: (Post 341752)
Hi John,

It looks as if you are having as much fun as I did looking into how to set the sub distance/timing. It is interesting to see how, as I found with the sub, the limited frequency response of the MR affects the shape of the impulse curve.

Bill, I just reviewed your referenced thread again and agree that is the best method to align SW to Mains using REW. I just tested that general procedure (as I understand it) and it agrees closely with my current alignment in which I have great confidence that it is set correctly. I will put together and summarize my findings in that thread.

jtalden 01-26-11 09:50 AM

Re: Aligning Driver Phase - REW v5 Example
 

Quote:

paulspencer wrote: (Post 341813)
It would be interesting to see how auto align compares to this manual process.

There is a lot of info there. Are you referring to the DCX auto alignment process that is referenced there or possibly to another more manual method I did not spot in my quick look?
I did try the DCX auto alignment method early on and think it is adequate. The numbers jumped around quite a bit in my trials, but if I remember correctly the variation was probably not too critical from a practical perspective. I would have to investigate it again to be sure. Being a hobbyist, I wanted to optimize the alignment in a repeatable way and to be assured that it is the best that can be done within the given equipment and tools I have available. I would definitely recommend the DCX procedure for those that donít look forward to diving into the complexities of REW. My guess would be that the DCX auto alignment works best for the lower XO frequencies where there is more room for timing error without major impact to the system.

paulspencer 01-26-11 11:09 AM

Re: Aligning Driver Phase - REW v5 Example
 

Yes, auto align. The funny thing about it is that it does it differently every time, even if you change nothing. It will invert polarities one time and not do it the next, but the time delays are also different. Most likely two different ways of getting a very similar result. I tend to use auto align because it's quick and I don't often have time to use a longer method. However, when I get more time I plan to do it the manual way. It wouldn't be too hard to compare that result to auto align. Great tutorial BTW!

jtalden 01-26-11 03:07 PM

Re: Aligning Driver Phase - REW v5 Example
 

Quote:

paulspencer wrote: (Post 341867)
It will invert polarities one time and not do it the next, but the time delays are also different. Most likely two different ways of getting a very similar result.

That's a good thought. It is possible to align phase at the XO in 1/2 wavelength increments if the polarity is inverted. It is sometimes a toss up as to which setting to use even when its set manually. I am focused on other work now, but I will try to keep this in mind when I am looking for something new to investigate.

atledreier 01-27-11 02:05 AM

Re: Aligning Driver Phase - REW v5 Example
 

Wow.. I have been looking for something like this for a while, since I introduced midbass units to my system. I need to study your post more closely and see if I can make heads or tails of it at all.. :p

Thank you for the time and effort making the write-up!


All times are GMT -5. The time now is 10:35 AM.

Powered by vBulletin
Copyright ©2000 - 2019, Jelsoft Enterprises Ltd.

Copyright ©2006 - 2019, Home Theater Shack, LLC.


vBulletin Security provided by vBSecurity v2.2.2 (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
User Alert System provided by Advanced User Tagging (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.

 


For the best viewing experience please update your browser to Google Chrome