Dual Sub Alignment - Affirmation with Procedure

[mattel]

Hi,

I posted this in another thread but did not receive a reply. So, I am trying it in a new thread.


I have relocated one of my subs, so I am trying again to align my two subs (HSU's; one ported and the other sealed). I have an Emotiva UMC-1 processor, with a 5.2 setup and minidsp 2x4HD. I am trying to understand the relationship between good SPL and what is a good phase response. There seems to be multiple delay settings that look good when moving the sliders.
Determining the best one is where I could use some help.

Measurements of the subs, mains and center are attached.

This is what I understand to perform 2 Sub alignment:
1. Use R sub base 200Hz XO and L sub base 200Hz XO (measurements with XO set to max =200Hz).

2. Use the slider to find the smoothest response, preferably with the summed output above the individual subs output. Question: while doing this, what should I be looking for from phase response, all 3 lines to converge and overlay each other?

3. Once I find the correct delay, click on align sum.

4. Re-measure both subs with XO's set to max to verify it and the aligned sum measurement are the same.



These are measurements before delay is applied. Except for the dip between ~35-50 Hz. I thought it looked okay.

3 sec delay set on slider. Smoother response, however SPL is below individual sub measurements at some frequencies. Not sure what the phase is telling me, though.

-16.34 sec delay set on slider. Not as smooth, but good SPL. This seems strange since the right sub is closer to MLP (~7ft. vs. 15.6ft for the left sub. Is the graph indicating the 3 lines are in phase (or close to it, from 40 - 120Hz?

I appreciate any help in sorting this out. Once I get the subs aligned properly, I will move on to aligning mains with subs.

 

[jtalden]

I checked your result using measurements 7 and 8 in the file. The SPL in chart 3 (16.34 ms) looks reasonably good and the phase tracking also looks reasonable. I would have chosen 14.5 ms for the delay of the left SW instead as the impulse alignment and phase tracking up to 100 Hz is slightly improved and the SPL is pretty similar. There is often a tradeoff in how the SPL support is best distributed and it is difficult to say if one setting in that range will work better than the other after the XO has been finalized.

FYI:
I notice there is a 1.8 ms (2 ft.) difference in the arrival time between the FL and FR main speakers. I suggest this difference be adjusted to <0.5 ms before the XO timing is determined.

 

[mattel]

Thanks so much for taking the time to look at this.

Adding delay to the left subwoofer is counterintuitive since it is further from the MLP than the right. So, does that mean the best delay will not always be a function of subwoofer distance?

In choosing your recommended 14.5ms left sub delay you look at a comparison of the impulse response (of the combined subs and the aligned sub measurements?), the SPL and the phase response? And the goal of the phase response it to get the lines in the phase graph to lay on top one another over the greatest FR range?

There is often a tradeoff in how the SPL support is best distributed and it is difficult to say if one setting in that range will work better than the other after the XO has been finalized.

I am not sure what you mean? Is this relative to when I align mains with subs? My XO is 80Hz. So, I may end up with a different XO afterwards?

At the end, the proof is in measuring and listening.

I notice there is a 1.8 ms (2 ft.) difference in the arrival time between the FL and FR main speakers. I suggest this difference be adjusted to <0.5 ms before the XO timing is determined.

Wow!, I didn't even notice. I am not sure why, since FR and FL are symmetric. I will measure again and see if its repeatable. Thanks for the pointing that out to me.

 

[jtalden]

Adding delay to the left subwoofer is counterintuitive since it is further from the MLP than the right. So, does that mean the best delay will not always be a function of subwoofer distance?

It suggests there is a problem with the measuring methodology and should be investigated.

In choosing your recommended 14.5ms left sub delay you look at a comparison of the impulse response (of the combined subs and the aligned sub measurements?), the SPL and the phase response? And the goal of the phase response it to get the lines in the phase graph to lay on top one another over the greatest FR range?

I looked at the overlay chart of the the 2 SW impulses to determine the time offset between them and to confirm that their shape suggests the measurements were good. In the alignment tool I looked for both close phase tracing throughout the XO range and favorable overall SPL in that range. There is more discussion in this thread.

I am not sure what you mean? Is this relative to when I align mains with subs? My XO is 80Hz. So, I may end up with a different XO afterwards?

It is just a recognition that the chosen timing of the 2 SWs may have some impact on the final SWs+FL and SWs+FR SPL/phase responses after the XO timing is determined. We cannot predict which setting of the SW timing may be more favorable to the end result. Occasionally it may be advantageous to misalign the timing of the SWs some small amount in order to achieve a better final response. This can be investigated if the final SPL/phase results do not look very favorable. It is just a DIY thing, so if you are not intending to experiment with many different settings, I should not have mentioned it. Just select a favorable SW timing and move on to the final setup (XO and EQ).

Wow!, I didn't even notice. I am not sure why, since FR and FL are symmetric. I will measure again and see if its repeatable. Thanks for the pointing that out to me.

It was just another way for me to find out if there was possibly a problem with the measurement timing. Repeated measurements of the FL or the FR speaker should show all impulses fall on top of each other when zoomed in on the time scale. If they don't, it suggests that the REW acoustic timing not working properly.

 

[mattel]

It suggests there is a problem with the measuring methodology and should be investigated.

Can you offer some things I might check? I have done several measurements in the past, all in the same manner, but this is the first time for the recommended delay be placed on the closer sub.

There is more discussion in this thread.

Thanks for the link. I will read it to try to educate myself more.

This can be investigated if the final SPL/phase results do not look very favorable. It is just a DIY thing, so if you are not intending to experiment with many different settings, I should not have mentioned it.

No, experimenting is definitely my plan. I am trying to make my system sound as best as it can. I have been concerned about having one sealed and one ported sub and the negative consequences associated with them. So much so, I have been considering buying another sub, probably another HSU ULSII.

Just select a favorable SW timing and move on to the final setup (XO and EQ).

Make sure I have the steps in the right order:
1. Align subs
2. EQ subs
3. Align Eq'd subs with main at favorable XO,
4. Final Eq of speakers and subs in XMC-1 using Dirac.

Is this correct. Also, is XO timing and sub/main alignment the same thing?

It was just another way for me to find out if there was possibly a problem with the measurement timing. Repeated measurements of the FL or the FR speaker should show all impulses fall on top of each other when zoomed in on the time scale. If they don't, it suggests that the REW acoustic timing not working properly.

I will take measurements again with the center and also with FL as the timing reference to see if that will shed some light.

 

[jtalden]

Can you offer some things I might check? I have done several measurements in the past, all in the same manner, but this is the first time for the recommended delay be placed on the closer sub.

I don't know of anything that would cause that large of a discrepency. Are you sure the the wiring is not crossed, i.e., the closer sub is the really playing when it is selected for the measurement?

Make sure I have the steps in the right order:
1. Align subs
2. EQ subs
3. Align Eq'd subs with main at favorable XO,
4. Final Eq of speakers and subs in XMC-1 using Dirac.

2. A preliminary EQ of the subs is probably not needed and possibly not advantageous as Dirac will handle the EQ. Any manual filters applied in the XO range may not be appropriate as the impact of the mains is not yet known.
3. Yes, if Dirac doesn't determine the XO settings. No, if Dirac handles the XO and EQ settings.

Also, is XO timing and sub/main alignment the same thing?

Yes.

I will take measurements again with the center and also with FL as the timing reference to see if that will shed some light.

The CC is fine to use as the timing reference. Take several measurements of the FL to confirm the impulses all fall at the same time. If the FL is 1 ft. more distant than the CC the impulses should all fall at about 1 ms on the overlay chart. You can repeat the SW measurements also opf course, but it is more difficult to judge the repeatability as accurately because the impulse response is much more spread out.

 

[mattel]

The CC is fine to use as the timing reference. Take several measurements of the FL to confirm the impulses all fall at the same time. If the FL is 1 ft. more distant than the CC the impulses should all fall at about 1 ms on the overlay chart. You can repeat the SW measurements also opf course, but it is more difficult to judge the repeatability as accurately because the impulse response is much more spread out.

I have taken new measurements of mains and subs. I took several of FL and FR. There were a few things that stood out to me.
1. The timing references showed negative delays and with some variance. They should always be positive, correct? What tolerance should it be for a speaker, +/- 0.05ms? The FL timings were close but there was one for the FR that was way off. See below.

Also, the timing reference delay,19.123ms, for the Rsub was much larger than expected. However, it is connected to a wireless receiver. You think that is the cause?

I did not look at sub alignment since I wasn't sure if the measurements would prove fruitful. I have attached the mdat file for your review.

Thanks again for all your help. I really appreciate it.

 

[jtalden]

I took several of FL and FR. There were a few things that stood out to me.
The timing references showed negative delays and with some variance. They should always be positive, correct?

When using the CC for acoustic timing the FL and FR delay will be negative if they are closer in distance to the mic than the CC.

What tolerance should it be for a speaker, +/- 0.05ms? The FL timings were close but there was one for the FR that was way off. See below.

FL measurements 1 and 3 fall on top of each other as expected. Measurement 2 is close enough to be insignificant for SW timing purposes. It leads measurements 1 and 3 by about 0.04 ms. My experience is that they stay within 0.02 ms when acoustic timing is working properly. I am using 96k sample rate and use a XLR mic so possibly the 0.04 ms is within normal variation at 48k sample rate with the UMIK-1?
FR measurements 4 and 6 fall on top of each other and 5 and 8 also fall on top of each other. The 2 groups are separated by about 1.5 ms. This is much more than expected and thus questions the validity of all the measurements. It is impossible to know for sure which if any of the these measurements are correct. It may be safe to take the -0.45 (FL) and -0.72 ms (FR) measurements as assume they are correct, as they fall closer to each other as we would expect for a well centered mic. I suggest you spend some time to try to sort out the problem first however. I do occasionally get a outlier so if most repeated measurements are in agreement one of those can be selected for the timing work.

Also, the timing reference delay,19.123ms, for the Rsub was much larger than expected. However, it is connected to a wireless receiver. You think that is the cause?

Yes, I would expect a wireless sub to be delayed more than a wired sub. I don't know how much extra delay is typical, but if the delay is a constant value the relative delay can just be adjusted out as needed using the MiniDSP.

I suggest you use the impulse overlay charts to determine how closely the timing agrees. The measurement notes can occasionally be in wrong if the initial peak is not the largest peak. This was the issue with measurement 4. Below is the FL and FR impulse overlay charts.

 

[mattel]

When using the CC for acoustic timing the FL and FR delay will be negative if they are closer in distance to the mic than the CC.

The speaker distances are:
CC = 17.18ft
FL = 16.63ft
FR = 16.34ft
It agrees with your statement and are also in the ballpark of the timing reference figure. These distances are set in the processor during measurements. Should they not be?

You mentioned a "well centered mic". What do you mean; mic distance to FR and FL should be equal?

Adding delay to the left subwoofer is counterintuitive since it is further from the MLP than the right. So, does that mean the best delay will not always be a function of subwoofer distance?

So, with the wireless (right) sub seen as more distant of the two, would that explain why adding delay to the more distant (left) sub is recommended?

FWIW I am using REW V5.20 beta60.

I will look at the measurement and impulse graphs in more detail as you suggested. Also, will do more measurements of each channel.

Is it possible room reflections will have an effect on the timing?

Thank you for your patience.

 

[jtalden]

The speaker distances are:
CC = 17.18ft
FL = 16.63ft
FR = 16.34ft
It agrees with your statement and are also in the ballpark of the timing reference figure. These distances are set in the processor during measurements. Should they not be?

Entering them into the processor is correct. If they exactly reflected the actual distances to the mic then FL, FR and CC impulses would fall on top of each other. My comment explained what occurs if equal distances are set in the processor. I should have worded that differently since I didn't know your distance settings. I'll try again. You indicated the FL and RR were reading negative delays compared to the CC reference measurement. So, if we wanted to fine tune the processor distance settings, the FL and FR distances would need to be reduced in order to increase their delay. Instead, we could increase the CC distance to reduce its delay. We are limited by the distance resolution of the processor distance settings so small offsets of the 3 impulses are common place.

You mentioned a "well centered mic". What do you mean; mic distance to FR and FL should be equal?

Yes, I should have referred to equal distances instead of centering.

So, with the wireless (right) sub seen as more distant of the two, would that explain why adding delay to the more distant (left) sub is recommended?

That sound correct. It depends on the actual distance difference, the difference in processing delay of the 2 SW and any current delay settings in the Mini DSP. The main point is that we want after reasonably good SPL support and phase agreement throughout as much of the SW range as possible. We can measure the 2 SW with any distance/delay settings we want and then use the REW alignment tool to identify the delay change needed to create that situation. The requirement for that to work properly is that the acoustic timing needs to have good accuracy and repeatability. Your planned investigation will confirm that.

Is it possible room reflections will have an effect on the timing?

Yes. We can first confirm the delay setting needed to time align the direct sound and then determine if a fine tuning adjustment is helpful to account for room effects.

 

[mattel]

Entering them into the processor is correct. If they exactly reflected the actual distances to the mic then FL, FR and CC impulses would fall on top of each other. My comment explained what occurs if equal distances are set in the processor. I should have worded that differently since I didn't know your distance settings. I'll try again. You indicated the FL and RR were reading negative delays compared to the CC reference measurement. So, if we wanted to fine tune the processor distance settings, the FL and FR distances would need to be reduced in order to increase their delay. Instead, we could increase the CC distance to reduce its delay. We are limited by the distance resolution of the processor distance settings so small offsets of the 3 impulses are common place.

Sorry for not responding sooner. I had to sleep on this to try to digest what you were saying. My take on it is: it is essential that FL and FR distances are the same as measured by REW. Equal main distances will permit proper alignment with aligned subs. And, if the CC distance is the same as the mains, it will also permit alignment with the aligned subs. Correct?

So, the first step is to ensure the FL, FR and CC are set precisely the same in the processor as their actual distances and remeasure each several times to confirm timing repeatability.

Back to the 2SW alignment. You agree, with good SPL and phase agreement obtained, adding delay to the closer Lsub is okay?

 

[jtalden]

Sorry for not responding sooner. I had to sleep on this to try to digest what you were saying. My take on it is: it is essential that FL and FR distances are the same as measured by REW. Equal main distances will permit proper alignment with aligned subs. And, if the CC distance is the same as the mains, it will also permit alignment with the aligned subs. Correct?

This was not the message. That comment goes back to your question about how/why the FL and FR were negative delay values. I tried to explain that if the REW reference signal arrives at the mic after the FL and FR then they will both show a negative delay time. I did not make that point clear and you were misunderstanding my point so I tried again. [Maybe I misunderstood the initial question and thus my attempted answers didn't make sense.]

If you understand how some of delays can be negative instead of positive then you do understand the only point I was trying to convey. There will usually be a some reasonable difference in the delay times of the FL, FR and CC as exact distances to the mic are very unlikely.

So, the first step is to ensure the FL, FR and CC are set precisely the same in the processor as their actual distances and remeasure each several times to confirm timing repeatability.

No, we just need to be sure the delays between them are reasonable. Your first data had large differences on some of the measurements that didn't look correct. That is why we questioned the acoustic timing repeatability.

Back to the 2SW alignment. You agree, with good SPL and phase agreement obtained, adding delay to the closer Lsub is okay?

Yes. Just apply the setting change found using the REW alignment tool and then remeasure to confirm that the alignment was met.

 

[mattel]

If you understand how some of delays can be negative instead of positive then you do understand the only point I was trying to convey. There will usually be a some reasonable difference in the delay times of the FL, FR and CC as exact distances to the mic are very unlikely.

Yes, I do understand, now. All measurements are relative to the reference distance.

Ok, I have taken more measurements. I adjusted the FR and FL to the same distance. I took a dozen measurements of each and was still getting a couple of variations in reference timing. I am still at a lost why that is. I changed the reference to one of my surrounds and it did not make a difference. Anyhow, I kept the ones that I believe to be correct.

I looked at the impulse overlay and wondered what causes that second peak? One of them is greater than the first peak(s).

It does show good repeatablity.

I measured the combined sub responses of the Rsub connected wireless and wired. I was surprised to see such a discrepancy. There is a significant decrease in SPL between 30-50Hz.The timing delay for Rsub wired is 4.68ms and for the wireless its 17.65ms. Does this mean the additional delay is causing it; the sub is out of phase more when operating wireless?

When comparing Rsub wired vs wireless, I only see a drop in SPL of approximately 2dB.

My next step is to compare alignment of the subs wired vs wireless. My assumption is with Rsub wired, the recommended delay will be positive (add delay to Rsub; closer sub) and negative (add delay to LSub; distant sub).

Your thoughts are appreciated. Thanks again.

P.S. Is there a limit on file size?

 

[jtalden]

Ok, I have taken more measurements. I adjusted the FR and FL to the same distance. I took a dozen measurements of each and was still getting a couple of variations in reference timing. I am still at a lost why that is. I changed the reference to one of my surrounds and it did not make a difference. Anyhow, I kept the ones that I believe to be correct.

Yes, I looks like you can pick out a correct measurements for the mains using repeated measurements.

I looked at the impulse overlay and wondered what causes that second peak? One of them is greater than the first peak(s).

Yes, I noticed the large secondary peak in all your FL and FR measurements (CC also if I remember correctly). This is usually due to a surface reflection either near the speakers themselves or near the microphone. Bookshelf speakers on the floor, desk surface or very near a wall is a possibility. The back of a leather chair or other surface near the mic is another possibility. These situations are best avoided. Also, I now notice that the offset error on some of your measurements is about the same 1.4 ms difference as the spacing to the second peak. If the CC has the second peak that is near the same size as the initial peak, that may be the cause of the problem. REW may sometimes be picking up the second peak of the reference sound from the CC as the timing reference time instead of the first peak? It's just a thought, but you could measure the CC several times using the CC as both the reference and the measurement to see if this same offset is present in some of the measurements and accounts for the outlying measurements. You probably ruled this out however as the same problem occurred when you used the surround as reference.

It does show good repeatablity.

Measurement 14 lined up correctly as shown in your chart, but note that measurement 13 is again leading by about 1.3 ms. You should have chosen to save a measurement that agreed more closely with the others. I didn't really pay too much attention to the mains+SWgroup combined measurements here as they are not really a part of the effort to determine a favorable SWgroup setting.

I measured the combined sub responses of the Rsub connected wireless and wired. I was surprised to see such a discrepancy. There is a significant decrease in SPL between 30-50Hz.The timing delay for Rsub wired is 4.68ms and for the wireless its 17.65ms. Does this mean the additional delay is causing it; the sub is out of phase more when operating wireless?

Yes, the actual Rsub delay is significantly different in the 2 modes so the DSP delay setting needs to be different to create the same relative timing of the 2 SWs together (SWgroup). You just need to decide if you want to run Rsub wired or wireless and set the Lsub or Rsub DSP delay accordingly.

We previously estimated a 14.2 ms delay needed for the Lsub for good SWgroup timing using wireless mode. You now identified the reduction in delay time for the wired mode of the Rsub to be ~13 ms. The needed Lsub delay for the wired mode is thus reduced that much. We would thus expect the same result in wired mode using a 1.2 ms delay setting on the Lsub. This difference in delay should now agree more closely with the distance difference to LP of the 2 SWs. The SPL and phase should be the same when comparing these two setups. If differences are seen between the 2 setups then there is still a problem with the acoustic timing of measurements.

When comparing Rsub wired vs wireless, I only see a drop in SPL of approximately 2dB.

The amount of the SPL impact is a function of the relative delay timing difference of the 2 SWs. It is best to use the REW alignment tool to assure the DSP timing is well selected for the the chosen mode (wired or wireless).

My next step is to compare alignment of the subs wired vs wireless. My assumption is with Rsub wired, the recommended delay will be positive (add delay to Rsub; closer sub) and negative (add delay to LSub; distant sub).

It's fine to experiment to better understand. The direct approach is:

1. Chose the mode that you want to use.
2. Confirm the DSP delay needed using the REW alignment tool.
3. Apply that DSP delay setting and remeasure to confirm agreement with the REW prediction.

P.S. Is there a limit on file size?

Probably, but I don't know the size. It's best to post only the needed measurements. Now that you can identify the good ones you can discard all the others for the posting.