Understanding sub phase/group delay for placement

[FargateOne]

I found an error in the SPL chart above. I accidentally summed the FR and FL instead of averaging them. That made the SPL support look a little better than it actually was. Below is the correct SPL chart. This has no impact on the timing adjustment, or the other 3 charts.

Yé! I wasn't sure about this chart but would never questioned it! Thanks!.

Here the same mdat but with the 2 missing measurements (I added also center channel positive polarity just in case) for the 100Hz xo measures.

 

[jtalden]

How does this timing look? Not measured with timing reference. Also, no delays were used. Just phase/polarity settings.

There is not enough information in that chart to make any comment regarding delay timing between SW and mains. If you feel you may have a problem with SPL in the XO range and want some to confirm the delay timing, please start a new thread. Provide the basic info on your setup and attach an mdat containing 3 measurements.

Measurement Setup:

• AVR XO active
• Full range sweeps
• REW acoustic timing activated
• Mic at LP

Measurements:

1. FL (alone - no SWs)
2. FR (alone - no SWs)
3. SWs (All SWs together) [Measure this using the L channel with the left main speaker disabled thus the redirected bass is measured]

Be sure you know or record all the AVR/MiniDSP distance/delay settings used for these measurements. My recommendations will be adjustments from those settings. You can leave any SW or mains EQ active if you have made those settings.

I will provide 1 or 2 delay recommendations, resulting charts and answer any questions you have.

 

[jtalden]

Yé! I wasn't sure about this chart but would never questioned it! Thanks!.

Here the same mdat but with the 2 missing measurements (I added also center channel positive polarity just in case) for the 100Hz xo measures.

Would you really consider this option? Probably not as the best XO range SPL support will be about the same probably. The low bass response <60Hz will be that same weak response we got using the 80 Hz XO that resulted with that SW and Mic locations in the room.

 

[thexder]

1. SWs (All SWs together) [Measure this using the L channel with the left main speaker disabled thus the redirected bass is measured]

Okay, thank you very much for your kind help. I will do that and open a new thread.
Just to clarify: when measuring SWs, L or R channel outputs the full range signal. In this case subs will play whatever they can?
What redirection are we talking about when LFE is chosen, if we assume AVR is not doing any crossover (I believe 7-CH Input is proposed when doing sweeps via HDMI).
Since I have a delay question as well, would you care to check this thread of mine?

 

[jtalden]

Okay, thank you very much for your kind help. I will do that and open a new thread.

Or, if you prefer, just use the one you already linked to.

Just to clarify: when measuring SWs, L or R channel outputs the full range signal. In this case subs will play whatever they can?

Yes, the L or R will be disconnected so only the 'redirected bass' will play through the SW.
You may be wondering why there are no measurements of SW+L and SW+R requested. This is because they can be generated within REW using the arithmetic tool or the alignment tool. You can provide those 2 extra measurements as well if you like, but I don't really need them unless you have issues getting the REW acoustic timing feature to work properly.

What redirection are we talking about when LFE is chosen, if we assume AVR is not doing any crossover (I believe 7-CH Input is proposed when doing sweeps via HDMI).

I don't understand this question. HDMI 4 is normally the LFE. We are not using the LFE channel as there is no XO active for the signal in that channel.

We are trying to determine the delay timing (SW distance setting in the AVR) to provide a smooth handoff between the main channels and the SW so the L and R channels sound as smooth as possible even though the low bass is coming from the SW. The AVR XO must be active for this determination. That way the L or R channel signal below the XO frequency is redirected to the SW.

Possibly you are instead still trying determine the XO frequency to use, or try to answer another question? Speaker and SW location and XO frequency must be determined before this delay setting is determined. Some EQ work can be done before or after this effort. Some automated EQ setup routines in AVRs may reset the SW and speaker distances and not allow changes so that would make this effort moot.

Since I have a delay question as well, would you care to check this thread of mine?

??, I didn't understand that there was a SW delay timing question there already. I am not an expert concerning room acoustic (reflection delay) questions so have no helpful input on those issues.

I have experience with one method to integrate the SW into a room setup. There are many other methods and most are easier and effective in achieving favorable SPL support in the XO range. The reason I use this particular method is that is provides a more insight into the phase relationship of the XO handoff. Some of us are more comfortable with a setup when we better understand the tradeoffs between other good delay settings.

 

[thexder]

Or, if you prefer, just use the one you already linked to.

I will make new one while I did some gear shifts.

Yes, the L or R will be disconnected so only the 'redirected bass' will play through the SW.
You may be wondering why there are no measurements of SW+L and SW+R requested. This is because they can be generated within REW using the arithmetic tool or the alignment tool. You can provide those 2 extra measurements as well if you like, but I don't really need them unless you have issues getting the REW acoustic timing feature to work properly.

OK, I understand.

I don't understand this question. HDMI 4 is normally the LFE. We are not using the LFE channel as there is no XO active for the signal in that channel.

OK, that makes sense, point taken.

We are trying to determine the delay timing (SW distance setting in the AVR) to provide a smooth handoff between the main channels and the SW so the L and R channels sound as smooth as possible even though the low bass is coming from the SW. The AVR XO must be active for this determination. That way the L or R channel signal below the XO frequency is redirected to the SW.

I was thinking this way: you put a full range signal to L+R+SW with no XO. With L+R disabled, SW is playing whatever it can (most are limited to 120 Hz).
That being said I don't always see I need for XO when integrating subs with mains, right?
Or do you have better experience with integration when using XO?

Possibly you are instead still trying determine the XO frequency to use, or try to answer another question? Speaker and SW location and XO frequency must be determined before this delay setting is determined. Some EQ work can be done before or after this effort. Some automated EQ setup routines in AVRs may reset the SW and speaker distances and not allow changes so that would make this effort moot.

I was once able to get a good integration without crossover so I wonder if it is necessary for a good blend-in?

??, I didn't understand that there was a SW delay timing question there already. I am not an expert concerning room acoustic (reflection delay) questions so have no helpful input on those issues.

Not really. It's a delay question for main speakers. I found out adding distance to one speaker improved the curve and would like to get a confirmation.

 

[jtalden]

From the questions you may be thinking that you want to either:
1. Set the mains to large so the SW only plays to LFE channel?
2. Set double bass where both mains and SW overlap throughout the bass range?

I was assuming you wanted the more common situation of the mains set to small and thus redirected bass from the mains goes to the SW. With this option the XO should be active for the measurements as the XO filters impact the delay timing, i.e., there is a difference in the delay timing between XO active vs no XO. Choosing a different XO frequency also will change delay timing. The significance of the delay timing difference in any given case is difficult to predict particularly given the room effects. It's best to measure in the way the setup will be used.

 

[JStewart]

80 Hz XO analysis:

• The SPL support was not very favorable when delay timing was adjusted based on the direct sound. For this reason, I decided to try to find better timing using the unwindowed LP response.
• There is a significant difference between the unwindowed phase response between the FL and FR speakers in the 80 - 120 Hz range. For this reason, the FL and FR were aligned and averaged to find the best delay timing for SPL support in the XO range. This led to strong SPL support across the XO range at the expense of a few milliseconds of additional sub delay. The resulting increase in group delay would not be expected to have any significant negative impact. This is thus likely to be the best tradeoff.

I was wondering if you could show us the windowed phase response (i.e. Direct) you were referring to in this post and provide some guidance on when you might use it or not vs the unwindowed response.

Regarding aligning the FL and FR speakers due to differing phase response... Can you determine from the measurements why they were not aligned?

 

[jtalden]

I tried a FDW of 4 to get a view of the direct sound phase and then adjusted the phase to track as well as possible. Below is the phase tracking with the SWs polarity inverted and the delay reduced by 2.54 ms (option 1). It's ideal below 115 Hz and then departs rapidly above that. It looks pretty good overall, but as we will see below the SPL support is not favorable.

[I also tried inverting the SW and finding the best tracking. That tracks better above 115 Hz and depart rapidly below (not shown).] This happens when the acoustic rolloff slopes are not complementary.

Below is the phase tracking using the original SWs polarity and the delay increased by 2.4 ms. (option 2) This was my recommendation for the OP. The phase tracking using the same FDW of 4 looks poor across the entire XO range.

Below is the phase tracking with this same delay setting, but with the FDW removed.

We can now see why this timing was chosen. The phase tracking is now very good from 65 - 200 Hz. This is almost all of the XO range.

Below is the resulting SPL comparison of the 2 options. There is significantly better SPL support in the 115 - 150 Hz range.

Other timing options tried but none were more favorable than this one.

Below is the relative timing of the SW impulse for options 1 and 2 and also for the OP's original position in the middle of the 2 options.

All these lag the mains impulse more that is usually necessary, but I was not able to identify a favorable delay timing closer to 0 ms. It was judged more important to help mitigate the 115 - 150 Hz SPL sag.

 

[thexder]

From the questions you may be thinking that you want to either:
1. Set the mains to large so the SW only plays to LFE channel?
2. Set double bass where both mains and SW overlap throughout the bass range?

I was assuming you wanted the more common situation of the mains set to small and thus redirected bass from the mains goes to the SW. With this option the XO should be active for the measurements as the XO filters impact the delay timing, i.e., there is a difference in the delay timing between XO active vs no XO. Choosing a different XO frequency also will change delay timing. The significance of the delay timing difference in any given case is difficult to predict particularly given the room effects. It's best to measure in the way the setup will be used.

I would like to be sure if XO is needed for an optimal sub integration. Some say main should remain full range. What's your though on that?
Note: I'm leaning to "no crossover" option only for stereo listening while I believe XO for movies seems more reasonable choice.
Besides that I'm wondering if closing the ports or changing the delay of L or R speaker improves the situation. That's why I was asking for an opinion in the other thread.

 

[jtalden]

Which of the 3 general approaches above will work out better? I would expect that answer is very much situational. As you know, there are also numerous setup options within each of these approaches. Determining the relative favorability of the approach or any given setup option can only be answered with appropriate measurements and listening sessions. I am not in a position help sort through all those possibilities. There is general advice/guidance from experts as well as strong specific opinions from other sources easily available. I tend to wary of specific recommendations when the specific setup situation is not considered.

Often room effects make the SPL in the XO range look like there may be something wrong with the delay timing between the SW and mains. I offer to help confirm if the timing is favorable for those that have a concern in that regard. I expect that offer to be limited to situations where all the other setup decisions have been made. If you have decided on a different setup approach for movies and music and have established the other setup conditions then I can help confirm favorable delay timing for each of those.

 

[FargateOne]

and that the second sub position may be more favorable to the initial position.

Would you really consider this option? Probably not as the best XO range SPL support will be about the same probably. The low bass response <60Hz will be that same weak response we got using the 80 Hz XO that resulted with that SW and Mic locations in the room.

Hello,
a follow up just to let you know how I appreciate yours advices.

For amateur like me, having the means (Rew) but not the knowledge forces to be humble and not to pretend to do what others can realy do. But, children learns imitating their parents (for good or worse is not the topic here!).

Taking your advices in # 3 and #28 (and as Christmas gift, my wife allowed me to do some changes in the room), I decided to put back the sub in front wall but at "Bahut" position (option 2 in #3) the place that I didn't choose at the beginning of this thread.

In order to learn by myself, I tried to follow your steps to find by myself the smoothest delay at XO frequencies .

Sub bahut position is very nearby position 1 that I choosed in post #4 .
To explain, position 1 was the sub on the front wall between FR and center and position called sub bahut is the sub on the front wall on the right side of the FR speaker. A difference of less than 60cm away to the right.

So I choose to let delay settings in Rotel at 240cm which were option 2 that you suggested in #10.
As you will se I hit the wall when I tried to use the alignment tool. But I tried and I concluded that the 240cm delay was good for this final position of my sub!

I may add that I did also many trials and errors delays tests the usual way ( changing delays in AVR by step of 30cm , measure, and continue) and found that the smoothest response for the center channel was xo100 and 360cm delay. Here is where my ignorance buried me alive.

 

[jtalden]

Your doing very well. I analyzed only the 80 Hz XO options.

I agree with your current distance setting for the FL and FR mains. It is the same setting I would have chosen.

There are many options for the CC; none with a clear cut advantage. It may be best left at the current distance. I might try reducing the CC distance by about 85 cm. This setting is a compromise. It allows the impulse location to stray a little from the location of FL and FR, but it appeared to help mitigate a sharp SPL dip at about 250 Hz. It gives up a little SPL in the middle of the XO range, but the smoother frequency response may be easier to EQ. This 250 Hz dip may not be a real advantage however as it is very high in the XO range. It well may reappear if the mic location changes slightly. That is, the apparent improvement there may not be real. Only additional measurements in the room with the changes will tell.

Don't hesitate to use a 100 Hz XO on the CC if you find a better response that way. I do suggest you don't stray too far afield from the location of the FL and FR impulse location however.

Rather than posting a bunch of charts, attached is my mdat. The first 4 measurement are the same as your first 4 measurements. Measurement 5 is the average of the FL and FR after their impulses were aligned. Measurement 6 is (FL+FR)+SW. Again, all these have the same timing that you established. Measurement 7 is the -85 cm adjusted CC position and 8 is the CC+SW frequency response.

 

[FargateOne]

I might try reducing the CC distance by about 85 cm

Thanks again. Do you mean to change the delay of the CC only (CC in receiver is 255cm to 170cm) or the delay of the sub for all the satellites (240cm - 85cm )?

 

[jtalden]

CC only.

 

[JStewart]

I tried a FDW of 4 to get a view of the direct sound phase and then adjusted the phase to track as well as possible. Below is the phase tracking with the SWs polarity inverted and the delay reduced by 2.54 ms (option 1). It's ideal below 115 Hz and then departs rapidly above that. It looks pretty good overall, but as we will see below the SPL support is not favorable.
View attachment 28043

[I also tried inverting the SW and finding the best tracking. That tracks better above 115 Hz and depart rapidly below (not shown).] This happens when the acoustic rolloff slopes are not complementary.

Below is the phase tracking using the original SWs polarity and the delay increased by 2.4 ms. (option 2) This was my recommendation for the OP. The phase tracking using the same FDW of 4 looks poor across the entire XO range.
View attachment 28044

Below is the phase tracking with this same delay setting, but with the FDW removed.
View attachment 28045

We can now see why this timing was chosen. The phase tracking is now very good from 65 - 200 Hz. This is almost all of the XO range.

Below is the resulting SPL comparison of the 2 options. There is significantly better SPL support in the 115 - 150 Hz range.
View attachment 28046

Other timing options tried but none were more favorable than this one.

Below is the relative timing of the SW impulse for options 1 and 2 and also for the OP's original position in the middle of the 2 options.
View attachment 28047

All these lag the mains impulse more that is usually necessary, but I was not able to identify a favorable delay timing closer to 0 ms. It was judged more important to help mitigate the 115 - 150 Hz SPL sag.

Thanks for such a detailed reply. I’ve read it several times since it was posted.
I’m suffering a mental block and perhaps my lack of understanding will reveal itself in this question... when, if ever, would one choose to set the alignment using the windowed responses?

 

[jtalden]

when, if ever, would one choose to set the alignment using the windowed responses?

It's a good place to start.
Close phase tracking allows the direct sound to arrive at the listening position at the same time from the mains and SWs throughout the XO range. This is the popular target timing for modeling software for XOs in main speakers. In maybe 70-80% of the room setups I analyzed it also results in favorable SPL support for the SW to mains XO. These are normally the better setups that to a reasonable extent have followed good practices with room setup and have selected a favorable XO frequency thus avoiding the worst of the room effects. In those cases there are other good timings that also provide similar SPL support with possibly different sound quality. A user may prefer the sound quality of one of those.

For those setups where the SPL support is unfavorable at the target timing, and setup changes are not considered, then different timing makes good sense. Timing that provides significantly more favorable SPL support is almost always a good idea. I do tend to suggest that the timing not stray too far away from the original target however, so it is good to identify what that target timing is. There is nothing wrong with using another one of the easier setup methods to finding a delay timing that provides good SPL support. Some of us just want to understand how that timing relates to the target timing.

 

[JStewart]

It's a good place to start.
Close phase tracking allows the direct sound to arrive at the listening position at the same time from the mains and SWs throughout the XO range. This is the popular target timing for modeling software for XOs in main speakers. In maybe 70-80% of the room setups I analyzed it also results in favorable SPL support for the SW to mains XO. These are normally the better setups that to a reasonable extent have followed good practices with room setup and have selected a favorable XO frequency thus avoiding the worst of the room effects. In those cases there are other good timings that also provide similar SPL support with possibly different sound quality. A user may prefer the sound quality of one of those.

For those setups where the SPL support is unfavorable at the target timing, and setup changes are not considered, then different timing makes good sense. Timing that provides significantly more favorable SPL support is almost always a good idea. I do tend to suggest that the timing not stray too far away from the original target however, so it is good to identify what that target timing is. There is nothing wrong with using another one of the easier setup methods to finding a delay timing that provides good SPL support. Some of us just want to understand how that timing relates to the target timing.

So to check my understanding... First check phase alignment using a more direct to reflected sound ratio with a frequency dependent window (FDW) of 4 cycles applied to the un-windowed measurement taken at the MLP. If a phase alignment timing can be found using this FDW that results in good SPL support through the XO range then use it. If not then room interactions are interfering with the direct sound to an extent that you have to use the un-windowed response.
Do I have that right?

Also, I've noticed with most of these phase alignments, the peak of the sub impulse response usually follows the peak of the Mains impulse response by several milliseconds. Might you know if is there a limit to how long after the mains impulse peak the sub impulse peak can be and it not be perceived as detrimental? Or as a practical matter would it always work out to be "close enough"?

Thank you for the invaluable help!

 

[jtalden]

No magic to 4 cycles - use what ever is needed to get a good feel for the phase of the direct sound. 5 cycles is my common first choice. I often adjust it lower or higher to confirm it holds up because sometimes a particular setting can be misleading. I then remove the FDW and view the impact with full room effects. If SPL doesn't look good that way, I adjust timing as needed. [Note - It is necessary to reselect one of the traces in the REW alignment tool to activate the FDW change in the traces - no need to exit the tool.]

We are working with IIR XO's intended to approach LR-24 XO shapes acoustically. If we magically achieved that in our setup without any speaker or room effects then the initial rise of the 2 impulses would align at the start of the of the impulse rise. That means the peak of the SW impulse will trail the peak of the main speaker impulse by several ms. The amount of peak lag depends on the XO frequency. The lower the frequency the greater the lag. I attached an mdat where I modeled this ideal condition at XO of 80 Hz. We can see the peak of the LP filter trails the peak of the HP filter by about 5.7 ms. You can also see the phase tracking and SPL is ideal with that setting. It's interesting to review the resulting group delay, step response and spectrogram. For real setups the speaker rolloff rates, EQ settings and room characteristics distort this ideal condition. The best delay timing and other characteristics thus departs significantly from the ideal. We are thus looking for a good compromise timing.

 

[jtalden]

It's interesting to see what happens to all the charts when the delay timing changed to other settings. This is easily investigated using this file.

 

[JStewart]

It's interesting to review the resulting group delay, step response and spectrogram.

It took me a second looking at the group delay to realize the delay was the effect of the High & Low pass filters. I was expecting a straight line at 0. Now I know better. Same with the spectrogram. I need to do some reading for a definition and very basic understanding of a step response. Clueless at the moment.

For real setups the speaker rolloff rates, EQ settings and room characteristics distort this ideal condition. The best delay timing and other characteristics thus departs significantly from the ideal. We are thus looking for a good compromise timing.

That understanding helps a lot.

It's interesting to see what happens to all the charts when the delay timing changed to other settings. This is easily investigated using this file.

Thanks, it was interesting to see the degree to which they are affected when the High passed and low passed signals are no longer aligned.

IIRC somewhere i read you had not had a chance to look at what measurements might look like post Dirac. If that's the case, here's my mdat for your viewing pleasure. I found it interesting that Dirac aligned the sub(s) (3 subs - single output) very well. I've made no changes to the delay settings post Dirac. (Prior to running Dirac the subs were phase aligned as best as possible using un-windowed responses at the MLP and REWs alignment tool.) Whether this alignment was Dirac's magic or luck I can't say for sure. I'm leaning towards Dirac magic though as it seems the odds of it being luck would be pretty high.

Can't thank you enough for all your help and learnings!

 

[jtalden]

Thanks for the file. Are you sure Dirac was active when the mains were measured? Most all Dirac charts I have see posted show a near ruler flat SPL response for the mains.

 

[JStewart]

Thanks for the file. Are you sure Dirac was active when the mains were measured? Most all Dirac charts I have see posted show a near ruler flat SPL response for the mains.

100%.Dirac On. That ruler flatness has been pointed out before. I can't say with certainty but I believe Dirac looks to correct "common" problems among the measurements and perhaps there is not much in common with the measurements in the "chair" pattern group in my (bad) room.

Here's an experiment with all 9 Dirac measurements taken in the same mic position in an attempt to "trick" Dirac into correcting it as compared to my usual "chair" pattern and with Dirac off. The measurements are separated below the overlay to see the patterns more clearly.

The 10 measurements in the same spot (BLUE) is much flatter than the other two and Dirac off (GREEN) is not too dissimilar from my usual "chair" pattern (RED)

Here's the impulse response with the Chair Patter (RED) and Dirac off (Green)

I'd be grateful if you have any observations or ideas.

Thanks!

 

[thexder]

Which of the 3 general approaches above will work out better? I would expect that answer is very much situational. As you know, there are also numerous setup options within each of these approaches. Determining the relative favorability of the approach or any given setup option can only be answered with appropriate measurements and listening sessions. I am not in a position help sort through all those possibilities. There is general advice/guidance from experts as well as strong specific opinions from other sources easily available. I tend to wary of specific recommendations when the specific setup situation is not considered.

Often room effects make the SPL in the XO range look like there may be something wrong with the delay timing between the SW and mains. I offer to help confirm if the timing is favorable for those that have a concern in that regard. I expect that offer to be limited to situations where all the other setup decisions have been made. If you have decided on a different setup approach for movies and music and have established the other setup conditions then I can help confirm favorable delay timing for each of those.

Okay, thank you for your explanaton. I'm already feeling guilty for hijacking someone else's thread.
I would be grateful if you could share your thoughs about L/R speaker delay but I'm going to open a new thread.
I find it very interesting that (even minimal) L/R delay can make such a difference in almost symmetrical environment.

 

[FargateOne]

I'm already feeling guilty for hijacking someone else's thread.
(...) but I'm going to open a new thread.

Thank you. I had some specific questions (about how to interpret group delays) too but wasn't sure to continue.
I think it is difficult to follow 2 discussions on 2 subjects sometimes less related.
Thanks again.