L/R speaker delay

[thexder]

Hi,

I would like to have a good (stereo) starting point before integrating the subs and I'm testing whether delaying one of the main speakers can help.
I'm not too keen when it comes to reading the measurements but my ears are telling me delaying left speaker sounds better.
I did some measurements, I hope we can discuss these plots together.

Thank you!

 

[jtalden]

Delaying L vs R at the LP is of no value.

Measuring a mono signal in stereo is problematic at high frequencies (HF). It is okay only if the mic is perfectly centered and you want to see the vector sum at the head of the mic. This is not the best measure of the HF SPL response that you will hear using 2 ears so it is of little value. It is okay for low frequencies (LF) and mid frequencies where a small mic offset has minimal impact on SPL. The phase impact at those lower frequencies can then be seen.

In general:

• It is best to measure the L and R separately.
• If you want to see the combined SPL impact of 2 together with a mono signal at the higher frequencies the better option is to use a simple average of the 2 measurements.
• If you want to see the vector average of them (includes the phase impact) at LF then it is best to first align the 2 impulses and then use REW (A + B) / 2 trace arithmetic.
• For EQ purposes it is best to use a simple average of several measurements around the LP and match that to your house curve. The experts most often only recommend EQ below 200 Hz. They do normally allow for gentle EQ shaping to the house curve at the higher frequencies.

 

[jtalden]

In the attached file I used your last 2 measurements as an example.
I aligned the L and R closely at 0 ms even though they were already closely aligned. I then created a simple average and then a vector average.
You can use these to compare to with the mistimed L + R mono measurements to see the impact at LF and HF.

 

[thexder]

Hello,
Thank you very much for your reply.

Delaying L vs R at the LP is of no value.

Care to explain more?
Because I think I have come to the same conclusion. At first it might have looked better but then I took a closer look at higher frequencies (up to 500 Hz) and decided no delay looks better altough my ears are telling me otherwise. But since I plan to add the subs this probably doesn't matter.

Measuring a mono signal in stereo is problematic at high frequencies (HF). It is okay only if the mic is perfectly centered and you want to see the vector sum at the head of the mic. This is not the best measure of the HF SPL response that you will hear using 2 ears so it is of little value. It is okay for low frequencies (LF) and mid frequencies where a small mic offset has minimal impact on SPL. The phase impact at those lower frequencies can then be seen.

Very informative, thank you. I was going to address HF later when LF will be done. Hopefully!

In general:

• It is best to measure the L and R separately.
• If you want to see the combined SPL impact of 2 together with a mono signal at the higher frequencies the better option is to use a simple average of the 2 measurements.
• If you want to see the vector average of them (includes the phase impact) at LF then it is best to first align the 2 impulses and then use REW (A + B) / 2 trace arithmetic.
• For EQ purposes it is best to use a simple average of several measurements around the LP and match that to your house curve. The experts most often only recommend EQ below 200 Hz. They do normally allow for gentle EQ shaping to the house curve at the higher frequencies.

Okay, I assume you did this for me in the next thread so let me take a look and ask questions later. These functions are new to me and I'm looking forward to learn new stuff!

 

[jtalden]

Care to explain more?

Delaying L vs R in small increment is maybe like turning our head to listen. I haven't ever read anyone touting that as a way to improve the sound quality. I would expect larger timing differences to move the sound toward the leading speaker and thus disrupt the sound stage.

I assume you did this for me in the next thread so let me take a look and ask questions later.

It's all related to explaining why measuring mono signals in stereo is not normally helpful.

 

[thexder]

Hi.
Sorry for long response, I had to update my REW to view your measurements.

Delaying L vs R in small increment is maybe like turning our head to listen. I haven't ever read anyone touting that as a way to improve the sound quality. I would expect larger timing differences to move the sound toward the leading speaker and thus disrupt the sound stage.

Thank you for clarification. I got mislead by looking at 20-200 area which seemed better with the delay but it turned out it caused more problems at higher frequencies.
Plus, I thought delay is used for sub integration it might help in odd shaped rooms.
I assume delay can mess with imaging.

It's all related to explaining why measuring mono signals in stereo is not normally helpful.

OK, to sum it up - what is the best way when doing sub integration? To feed both, L and R speaker with the same signal and then average both measurements? You said mono signal is supposedly okay for LF measurements and I've heard similar things before: that one should make mains mono and treat as a single source.
By the way, what is the point of averaging?
I'm sorry for these basic questions, I'm learning a lot here.

Looking at the .mdat file you have kindly made for me:
I see 4 equal plots for left speaker and 4 equal plots for right speaker.
Then there are 4 equal "A, B average" plots and 4 equal "Average 1" plots.
Am I seeing this right or did I miss something?
Where can I see mistimed L+R mono measurements?

 

[jtalden]

We are at a bit of an impasse. I am having trouble understanding your concerns and specific questions and have no idea of your equipment. You are not following the resulting, possibly overly detailed, attempts to answer my guess as what your questions were driving at. Let me try to simplify the 2 different subjects that I now realize are being intermixed here. Then you can reread my previous posts if you like to see if they make more sense. Regardless, my providing more detail on either process would not seem to help.

The 2 different subjects being discussed:
1. The relative delay between left and right channels is to create equal arrival time at the LP of the sound from the 2 channels. This is a trivial process that is usually not discussed because the LP is normally equidistant from both speakers. The distance setting in an AVR is thus set correctly when both L and R distances are set equal. Automated AVR setup routines do this automatically. If necessary we can just measure the distance with a tape measure and set it manually.
2. There is also a distance setting for in the AVR for the SW that is also automatically set at the same time. Some automated systems occasionally get this distance wrong. I offer to confirm/suggest a favorable setting for those who think they may have problem in this regard. My process is more difficult to explain than most of the other method commonly used, thus, I will provide 1 or 2 favorable distance settings if you provide the 3 needed measurements. I have recently detailed what those 3 measurements were in the other thread.

 

[thexder]

Thank you for your explanation and your patience. I'm aware I have many non related questions and am probably jumping topics. I really want to do it right this time.
I was asking if L/R delay is useful and learned that it's not. Okay, moving on.
I'm looking at your .mdat file and I'm sorry for the confussion earlier. I don't know why I saw 4 equal plots (12 altogether) yesterday. I must have opened the file three times. I'm seeing 4 today.
The way I see it, time aligned plots have zero impact on LF+HF, right?
I, too, did simple average and Vector average of my non-aligned plots and see no difference when comparing them to your average plots.

I'm however curious, why aren't my two (L+R) measurements time-aligned? Mic not exactly at equidistant position?

Then you explained measuring mono signal in stereo is problematic which I also understand. Makes total sense.
My final question before adding the subs and ask you to check for delay timing between SW and mains is:
Can I measure both, L+R speakers (using L+R signal in REW) when adding subs?

 

[jtalden]

Before I answer these questions and we move forward, I ask again (3rd time) that you first list your speakers, AVR and any other audio equipment you are using. Are you intending to use some type of manual or automated setup of EQ? Is this a dedicated room with room treatment and speaker/LP positions finalized? It is difficult to answer questions and suggest things clearly if I don't know the basics of your equipment and setup intensions. It can also be confusing to you if, for example, I refer to delay settings and you use AVR distance settings instead.

 

[thexder]

Uh, yes, you did ask for the equipment. I'm sorry I have missed that.
It is a dedicated room with some treatment. Sadly, not enough to tame LF issues. There are mostly 8" inch absorbers covered with diffusive surfaces.
Speaker positions can be changed but the distance between speakers must not because it would block the screen.
Sub positions are flexible as well. However, they cannot be placed at the rear corners.
AVR: old Marantz with Audyssey XT32 (as a preamp). I will use it for movies and probably add some gain on the subs when Audyssey will do its job. So far I've been using crossover at 80 Hz.

For stereo listening I'd like to integrate the subs with mains manually. I was going to use phase/polarity settings first. In case it would not work I'm prepared to use the delay (my previous attempts were not successful).
In case EQ will be needed it would be added below 200 Hz (no external EQ is available).
Speakers: even older JBL Ti5000. I can provide factory plots if needed.
Subs: 4xVelodyne DD-15 with 8-band PEQ

 

[DanDan]

I'm a bit lost here as to what is the question, so I will try some generic answers.
I believe there is a problem with delaying one side of stereo. With Minidsp hardware running Dirac Live it is not possible to bypass such delays. In highly treated spaces, pro listening rooms, quite a few of us have found sound delays, even tiny ones, to be blatantly audible.
This may be a cousin of the widening techniques we using in Mixing. But given the tiny delays involved it is hard to believe.
That is fact, and now to an opinion or query. I can imagine if subs are near and the speakers are far, e.g. ground subs and a hanging array in a live scenario, delaying the subs makes sense, for the people in the first few rows at least. But even 50Hz takes 20mS to form it's first single wave, and when do we begin to hear it as a tone? Furthermore there is only a small overlap around the crossover point, literally no bass from the speakers in general, and vice versa. So what are we trying to time align?

 

[jtalden]

Thanks for the helpful info on your setup and intensions.

The way I see it, time aligned plots have zero impact on LF+HF, right?

Correct, for the AV setup using an 80 Hz XO, the distance setting and the SWs phase settings impacts only the XO range - maybe 40-160 Hz or a bit more. In the stereo setup idea of using no XO, it will impact the entire overlap range of the SWs and mains.

Since your AV setup and stereo setup are a different, the timing may be enough different that a different SW distance setting may be helpful/necessary in the AVR. Is that something you don't mind changing for each AV or stereo use? The option is to look at each of the 2 methods for integrating the SWs into the mains and if it is different just use the one that works better for both AV and stereo use.

I'm however curious, why aren't my two (L+R) measurements time-aligned? Mic not exactly at equidistant position?

They were very closely aligned so the mic was almost perfectly centered. The very minor adjusted I made was trivial. I just wanted you to know I did it and that it is good practice before combining them in REW using the vector average method. It's not needed for a simple average.

[/QUOTE]Can I measure both, L+R speakers (using L+R signal in REW) when adding subs?[/QUOTE]

Yes, that is commonly done and is a minor simplifying step. I would expect it to still result in a favorable setting.

 

[jtalden]

For stereo listening I'd like to integrate the subs with mains manually. I was going to use phase/polarity settings first. In case it would not work I'm prepared to use the delay (my previous attempts were not successful).
In case EQ will be needed it would be added below 200 Hz (no external EQ is available).
Speakers: even older JBL Ti5000. I can provide factory plots if needed.
Subs: 4xVelodyne DD-15 with 8-band PEQ

Okay, I think I am finally understanding. The AV situation is not really involved in any of this. The stereo setup timing is the only question here so:
Since the Audyssey setup will now be inactive hopefully the SWs still combine well together. They looked very good as measured above. If they still look okay then the timing between the SWs and mains them is the only question. So the test conditions would be as stated before (with acoustic timing on) and the only required 2 measurements are L+R and SWs (no XO).

 

[thexder]

I'm a bit lost here as to what is the question, so I will try some generic answers.
I believe there is a problem with delaying one side of stereo. With Minidsp hardware running Dirac Live it is not possible to bypass such delays. In highly treated spaces, pro listening rooms, quite a few of us have found sound delays, even tiny ones, to be blatantly audible.
This may be a cousin of the widening techniques we using in Mixing. But given the tiny delays involved it is hard to believe.
That is fact, and now to an opinion or query. I can imagine if subs are near and the speakers are far, e.g. ground subs and a hanging array in a live scenario, delaying the subs makes sense, for the people in the first few rows at least. But even 50Hz takes 20mS to form it's first single wave, and when do we begin to hear it as a tone? Furthermore there is only a small overlap around the crossover point, literally no bass from the speakers in general, and vice versa. So what are we trying to time align?

The qestion was if delaying one of the speaker could be helpful.
Turned out it's not and @jtalden deftly lead me to figure that out by myself.
By the way, many good AVRs can store different profiles for music, movies, etc.
Sadly, mine can't. But there is a workaround - one can memorize the settings and set them manually.

 

[thexder]

Thanks for the helpful info on your setup and intensions.

Thank you for learning lessons.
I forgot to mention I have only one LP so no attention to other seats is needed.

Correct, for the AV setup using an 80 Hz XO, the distance setting and the SWs phase settings impacts only the XO range - maybe 40-160 Hz or a bit more. In the stereo setup idea of using no XO, it will impact the entire overlap range of the SWs and mains.

True. One method I'd like to try is integrating subs at random locations with no XO. It is preferred subs have activated LP filters - different at each sub. I'm very curious how it would turn out.

Since your AV setup and stereo setup are a different, the timing may be enough different that a different SW distance setting may be helpful/necessary in the AVR. Is that something you don't mind changing for each AV or stereo use? The option is to look at each of the 2 methods for integrating the SWs into the mains and if it is different just use the one that works better for both AV and stereo use.

The plan is to set the SW delays to zero and integrate the subs with phase/polarity settings. Subs have 6 presets, so one of these will be activated when listening to stereo.
AV setup: Audyssey will of course change the delays (which I will memorize) and it requires phase and polarity set to 0+. This setting will be stored at preset 2 on SW.
When listening to stereo, the preset 1 at the subs will be activated and delays set back to zero.
That's until I will be able to get some decent preamp/processor with listening presets. I know Anthem has this, for example.

They were very closely aligned so the mic was almost perfectly centered. The very minor adjusted I made was trivial. I just wanted you to know I did it and that it is good practice before combining them in REW using the vector average method. It's not needed for a simple average.

What is the primary point of averaging? I always thought this is needed for multiple seats and PEQ tweaking.

Yes, that is commonly done and is a minor simplifying step. I would expect it to still result in a favorable setting.

Great, I hope I will have a productive weekend! Thank you, sir!

 

[thexder]

Okay, I think I am finally understanding. The AV situation is not really involved in any of this. The stereo setup timing is the only question here so:
Since the Audyssey setup will now be inactive hopefully the SWs still combine well together. They looked very good as measured above. If they still look okay then the timing between the SWs and mains them is the only question. So the test conditions would be as stated before (with acoustic timing on) and the only required 2 measurements are L+R and SWs (no XO).

Correct.
I described my plan in my post above.
Just to clarify: none of the measurements in this thread are containing subwoofers. Just stereo speakers. So I'm curious what do you think looked very good above?

 

[jtalden]

Ah yes, sorry, I looked at a SW response that in the other thread that @FargateOne posted by mistake.

 

[thexder]

Ah yes, sorry, I looked at a SW response that in the other thread that @FargateOne posted by mistake.

No problem. I will post mains+SW soon!
One more thing. Can I play the same signal through both L+R mains in REW?

 

[jtalden]

Yes, in the measurement dialog popup you should see an output field that allow you to select L, R or L+R.

 

[thexder]

Hm, I'm seeing this:
Both options are playing sound only from one speaker.
If I select both - then I'm sending stereo signal to both speakers, correct?

 

[jtalden]

I'm no expert in all the possible problems in this regard. It's best to assign the output device for REW use. Using the default output is sometimes an issue, but I suspect that is not the problem here.

In the first jpeg the output device box greyed out, but REW must have found the default device okay if one channel works. If both L and R outputs work correctly to the FL and FR speakers then I don't know why 'both' doesn't work. Is your cabling correct - stereo output from default device to stereo input on the receiver? If the L and R outputs play the same speaker then the cabling is wrong.

In the second jpeg the FL is chosen for both of the 2 REW outputs. Try selecting the FR for the second REW output.

 

[thexder]

Uh, perhaps my explanation wasn't clear enough. Again!
Both speakers are working fine. I'm using HDMI connection, so wiring is not the issue.
I want to feed both speakers with left OR right signal. I assume L+R sound means is stereo signal?
If both speakers have the same signal that might simplify the sub integration.

 

[jtalden]

Check the Preferences/Soundcard panel for output device. Is the HDMI output device selected? It doesn't appear to be from the jpegs of the measurement dialog you posted. I have never used HDMI output and no longer use that older version of REW so...

 

[thexder]

Here's new version of REW.
I can select L, R or L+R.
The question is if the signal in case of selecting L+R is stereo or mono?
I assume measuring device (Umik-1 in my case) is mono as well?

 

[jtalden]

The same REW measurement sweep signal is sent to both L and R outputs. The timing signal also is sent to the timing channel.