Multi Subs- How to time align with different distances - Page 3 - AVS Forum
Forum Jump: 
Reply
 
Thread Tools
post #61 of 86 Old 02-07-2013, 04:21 PM
Senior Member
 
rock_bottom's Avatar
 
Join Date: Apr 2004
Posts: 430
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 14
Agreed, but it's also possible to consider it exclusively as a frequency-domain problem. In Earl's video, he does not use any delay at all, but just gets the required phase shift for proper integration via the phase shifts of the per-sub filters. The filters will have a group delay that depends on frequency, and since the delay isn't constant with frequency, its effect on the time domain isn't straightforward to determine. But it doesn't need to be determined, since the only thing that matters to him is the flatness of the combined responses of the subs and mains. The thing I like about his approach is that it could be implemented entirely with analog filters if desired, with no need for a digital delay. So then one wouldn't be locked in to an AVR or AV pre/pro, but could use an analog-only preamp (because there's no need for a "sub delay" function).

Although I can see if you were using a linear-phase FIR-based crossover, then phase and delay are much more intimately connected, so one could consider it as a time-domain problem in that case.
rock_bottom is offline  
Sponsored Links
Advertisement
 
post #62 of 86 Old 02-07-2013, 05:28 PM
Advanced Member
 
terry j's Avatar
 
Join Date: Aug 2006
Posts: 866
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 2 Post(s)
Liked: 14
well, that (actually) is why I initially looked at the thread (because of the title). I want to learn more 'technically'. Ie, I want to see/learn something new that I can then go and experiment with and see if it gives me better results, or easier, whatever.

It seemed to come to a stop with 'at sub frequencies it don't matter much'.

The video, is that the marching woofer and tweeter society one? I think I watched it all, sure an outline of philosophy but no real mechanical how to's.

Unless you are referring to something else?

Agreed, but it's also possible to consider it exclusively as a frequency-domain problem
. Well yeah, if you ignore the question of 'it is audible' which I am not so sure about, and I have had direct discussions with earle about, this forum I think, and you are willing to use amp power to brute force your way to that end goal.

I guess that is fine if you don't have the means to delay things, from that view it could be 'thankgoodness I can still get to the end point'.

I, at least, am firmly convinced there are audible differences between the same FR yet two different starting points, subs 'as they are yet brute forced' vs 'time aligned first then eq'd'. Obviously not everyone agrees with me there.

So if I have been looking at the 'wrong earle vid' could you direct me to the correct one, or if you are able could you describe a methodology different from mine, that of measuring impulse arrival times from all the bass sources and aligning them but using your idea of looking at phase shift?

That would be great, thanks.
terry j is offline  
post #63 of 86 Old 02-07-2013, 07:37 PM
Senior Member
 
autox320's Avatar
 
Join Date: Jan 2006
Location: NC
Posts: 464
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 4 Post(s)
Liked: 35
Not entirely related but a week ago I stumbled somehow into these videos and watched all three parts. Yes it's pro audio but I like the part about time delay and coverage pattern explanation. Yes pre warning it's dry listening but you could say Dave Rat knows a thing or two(check out his page for background info).

Can't remember which part but using phase to join subs caught my attention showing the sine waves.

https://www.youtube.com/watch?v=VwLH7zP6Lwo

"I should really see what dB levels I'm pushing. Long as it can't foam my beer during a movie we are ok "
autox320 is offline  
post #64 of 86 Old 02-07-2013, 08:03 PM
Senior Member
 
rock_bottom's Avatar
 
Join Date: Apr 2004
Posts: 430
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 14
Quote:
Originally Posted by terry j View Post

well, that (actually) is why I initially looked at the thread (because of the title). I want to learn more 'technically'. Ie, I want to see/learn something new that I can then go and experiment with and see if it gives me better results, or easier, whatever.

It seemed to come to a stop with 'at sub frequencies it don't matter much'.

I assume you're referring to the Geddes video referenced earlier in the thread, and his assertion that anything less than 30 msec delay in the sub(s) doesn't matter. I've found that Earl tends to make sweeping statements, but when pressed will concede that what he meant to say was really much more specific than what he did say. At the risk of misinterpreting him, I think what he meant to say was that two otherwise identical bass signals that are delayed from each other cannot be distinguished until the delay is about 30 msec or more. He did acknowledge in the video when asked by an audience participant that such a delay affects sub/mains integration, but that he takes care of that in the individual sub filters.
Quote:
Originally Posted by terry j View Post

The video, is that the marching woofer and tweeter society one? I think I watched it all, sure an outline of philosophy but no real mechanical how to's.

Unless you are referring to something else?

Yes, that's the one
Quote:
Originally Posted by terry j View Post

(Me) Agreed, but it's also possible to consider it exclusively as a frequency-domain problem (end Me).

Well yeah, if you ignore the question of 'it is audible' which I am not so sure about, and I have had direct discussions with earle about, this forum I think, and you are willing to use amp power to brute force your way to that end goal.

Considering the problem only in the frequency domain does not in any way imply boosting your way out of the problem. His summation of filtered sub outputs is a vector summation - that is, it takes into account both amplitude and phase, both of the individual subs and the mains. It's possible to match the phase of the summed/filtered subs and that of the mains at the crossover frequency, but this does not necessarily imply matching the corresponding delays at that frequency. The phase is just the phase, while the delay is proportional to the negative of the derivative (slope) of the phase with respect to frequency.
Quote:
Originally Posted by terry j View Post

I guess that is fine if you don't have the means to delay things, from that view it could be 'thankgoodness I can still get to the end point'.

I, at least, am firmly convinced there are audible differences between the same FR yet two different starting points, subs 'as they are yet brute forced' vs 'time aligned first then eq'd'. Obviously not everyone agrees with me there.

A reasonable view, but not everybody has a DEQX with a nifty multi-way FIR digital crossover ahead of their system DAC. Some of us have access only to analog signals after the DAC and crossover, and would like to stick to affordable solutions, the majority of which have IIR filters and involve A/D->D/A conversion.
Quote:
Originally Posted by terry j View Post

So if I have been looking at the 'wrong earle vid' could you direct me to the correct one, or if you are able could you describe a methodology different from mine, that of measuring impulse arrival times from all the bass sources and aligning them but using your idea of looking at phase shift?

Earl does not spell out the algorithm used by his software, presumably because he wants to keep it proprietary.

I've been trying to figure out how Earl does this, and looking into open-source optimization software to help. But that's a longer-term project.

Also, just matching the impulse response peaks is not sufficient as I'm sure you know - frequency domain considerations such as crossover frequency, slope, etc. need to be chosen judiciously too.
rock_bottom is offline  
post #65 of 86 Old 02-07-2013, 09:22 PM
Advanced Member
 
terry j's Avatar
 
Join Date: Aug 2006
Posts: 866
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 2 Post(s)
Liked: 14
Quote:
Originally Posted by rock_bottom View Post

I assume you're referring to the Geddes video referenced earlier in the thread,

you answered your own question! biggrin.gif
Quote:
Yes, that's the one



Quote:
A reasonable view, but not everybody has a DEQX with a nifty multi-way FIR digital crossover ahead of their system DAC.

Was that necessary? I agree they don't, but it seems a bit of a sidetrack. It is not only deqx that can delay things, minidsp mentioned earlier can (tho not very much afaik) behringer etc etc.
Quote:
Some of us have access only to analog signals after the DAC and crossover, and would like to stick to affordable solutions, the majority of which have IIR filters and involve A/D->D/A conversion.
Earl does not spell out the algorithm used by his software, presumably because he wants to keep it proprietary.



I've been trying to figure out how Earl does this, and looking into open-source optimization software to help. But that's a longer-term project.

Oh well, at least I did not miss anything obvious in the video. That can easily happen heh heh.
Quote:
Also, just matching the impulse response peaks is not sufficient as I'm sure you know - frequency domain considerations such as crossover frequency, slope, etc. need to be chosen judiciously too.
Well yeah sure, I just assumed that fixing the bass FR is the final step, all else has been done.
terry j is offline  
post #66 of 86 Old 02-07-2013, 09:49 PM
Senior Member
 
rock_bottom's Avatar
 
Join Date: Apr 2004
Posts: 430
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 14
Quote:
Originally Posted by terry j View Post

Quote:
A reasonable view, but not everybody has a DEQX with a nifty multi-way FIR digital crossover ahead of their system DAC.

Was that necessary? I agree they don't, but it seems a bit of a sidetrack. It is not only deqx that can delay things, minidsp mentioned earlier can (tho not very much afaik) behringer etc etc.

Necessary? Maybe not. Pertinent? Yes. With complementary linear-phase FIR crossover filters, matching delay also matches electrical phase over frequency. That's not necessarily true with IIR filters due to the frequency-dependent nature of their delay.

I was a bit puzzled that Earl would exclude individual sub delays too. While these can be compensated by other filter phase shifts to achieve flat combined response, it seems to me that, all other things being equal, this would result in more complex filters than if appropriate delays were used to accomplish the same. It makes me wonder if this may be a bug in his software that causes it to have problems when delay is introduced. This is just speculation of course.

OTOH, no delays means the IIR filters can be transformed to equivalent analog filters and used as-is. The audiophile in me loves that, as it eliminates all kinds of potential dynamic range issues, at the expense of severe lack of flexibility. But I plan to get it right, then leave it, so the lack of flexibility doesn't matter that much to me.
rock_bottom is offline  
post #67 of 86 Old 02-07-2013, 11:11 PM
Advanced Member
 
terry j's Avatar
 
Join Date: Aug 2006
Posts: 866
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 2 Post(s)
Liked: 14
This is gonna expose my ignorance big time! but hey, we are here to learn.

(oh, btw, I only asked about pertinent cause it would have been a lucky guess to just use deqx as an example, so I assumed you knew or remembered vaguely that I use one. I am often the 'enemy' haha around here, it could have been taken as a sneer of some sort. If not and purely an example fine, if a dig then 'ok')

Because I have not got my head wrapped around all of this, at least looking at the 'new' data you introduced, the question probably won't even make sense.

It seems to be a contradictory circular argument. As you say phase and time are inextricably linked. And that 'delay is not audible'. Well, if you gotta address it from a phase angle, as it affects FR which clearly is audible, else why correct it...how can you say delay is inaudible? ( I am not saying it is YOU that is saying that, but that seems the story)

In other words, delay (directly related to phase) is inaudible yet phase is (if only for what it does to FR)

We fix the FR by (first) fixing the phase, well that has to do something for the delay too?

You might be able to work out where I am mixed up on this by the way I phrased the question.
terry j is offline  
post #68 of 86 Old 02-08-2013, 12:16 AM
Senior Member
 
Antripodean's Avatar
 
Join Date: Apr 2009
Location: The Antripodes (OZ)
Posts: 328
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 12
As Autox wrote, the pro audio guys know it's important to be in phase and time aligned to get the most out of any system. Even the 'lowly' DCX has this as an auto correct feature. It's not there just to add bells and whistes. I often think Earle's statements are misleading or even leading to condescending. Suppose that all has been said because hi-fi people are OK with a suboptimal system, not caring if they have to add a few more 100 watts to get the desired SPL. Whereas the pro-audio operator needs to squeeze the most of every last watt. It doesn't answer your question but I wanted to say that believing verbatim what Earle says often leads to misunderstandings. Audibility is a tough question but I believe time aligned and in phase is important for system optimisation.

explore the music
Antripodean is offline  
post #69 of 86 Old 02-08-2013, 02:35 AM
AVS Special Member
 
Martycool007's Avatar
 
Join Date: Jun 2011
Posts: 3,551
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 189 Post(s)
Liked: 154
Can someone tell me what exactly phase is in audio, (speaker & subwoofer) terms? Also, how does phase affect sound in a theater setting, and more importantly, how does one go about properly setting the phase when setting up a 5.2 or 5.4 home theater system?
Martycool007 is offline  
post #70 of 86 Old 02-08-2013, 05:25 AM
AVS Special Member
 
Bill Fitzmaurice's Avatar
 
Join Date: Nov 2009
Location: New Hampshire
Posts: 9,679
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 1352
Quote:
Originally Posted by Antripodean View Post

As Autox wrote, the pro audio guys know it's important to be in phase and time aligned to get the most out of any system.
Pro audio guys don't have their systems in rooms that average 12 x 16 feet, where room modes and boundary reflections dominate low frequency response and time/phase issues. If HT had the same requirements as pro sound we'd all be stacking our subs front and center, to avoid creating a power-alley. Not that there aren't some aspects of pro-sound that apply to HT, but for the most part there's very little in common between the genres.

Bill Fitzmaurice Loudspeaker Design

The Laws of Physics aren't swayed by opinion.
Bill Fitzmaurice is offline  
post #71 of 86 Old 02-08-2013, 06:32 AM
AVS Special Member
 
coctostan's Avatar
 
Join Date: Oct 2005
Location: Indy
Posts: 1,960
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 128
Quote:
Originally Posted by Bill Fitzmaurice View Post

Pro audio guys don't have their systems in rooms that average 12 x 16 feet, where room modes and boundary reflections dominate low frequency response and time/phase issues. If HT had the same requirements as pro sound we'd all be stacking our subs front and center, to avoid creating a power-alley. Not that there aren't some aspects of pro-sound that apply to HT, but for the most part there's very little in common between the genres.

+1. Once you are in the low bass region, there is nothing in common between small room home audio and pro sound (unless also in a very small room). Part of the reason bass typically sounds great in large installations is because the bass is not in the modal region outdoors. Pulling off that same feat indoors is quite different.

It is not that time and phase are meaningless in a small room (certainly important at higher frequencies), but that setting delays for subwoofers doesn't have the effect that people assume it does. Phase/delay are a tool that may be used, but the idea that there is some "optimal" delay setting for a small room is false.

Geddes has spent the last 30+ years studying and researching this specific area. I believe he wrote his PhD dissertation on the subject. It works.
coctostan is offline  
post #72 of 86 Old 02-08-2013, 08:01 AM
Senior Member
 
rock_bottom's Avatar
 
Join Date: Apr 2004
Posts: 430
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 14
Quote:
Originally Posted by terry j View Post

This is gonna expose my ignorance big time! but hey, we are here to learn.

(oh, btw, I only asked about pertinent cause it would have been a lucky guess to just use deqx as an example, so I assumed you knew or remembered vaguely that I use one. I am often the 'enemy' haha around here, it could have been taken as a sneer of some sort. If not and purely an example fine, if a dig then 'ok')

Not a dig at all. I'd love to have a DEQX, but their previous lowest-cost device seems to have been discontinued and replaced with one that's out of my price range.
Quote:
Originally Posted by terry j View Post

It seems to be a contradictory circular argument. As you say phase and time are inextricably linked.

I am not saying exactly that. But the relationship is more straightforward for linear-phase FIR filters than for IIR, so let's talk about them first. For such filters, the group delay, which is proportional to the negative of the slope of the phase vs. frequency, is constant for all frequencies (because phase shift vs. frequency is a straight line, so its slope is constant for all frequencies). If you were to look at the impulse response of a linear-phase FIR low-pass filter having a delay of, say, 1 msec, its peak will occur 1 msec after the impulse is applied.

Now consider an IIR or analog low-pass filter. Its phase vs. frequency is not linear, so its slope of phase vs. frequency is not constant with frequency. Therefore its group delay is not constant with frequency either. The group delay is just a mathematical construct though, and it doesn't always match up with intuition. Sometimes it has a clear interpretation in terms of the impulse response (such as the linear-phase FIR low-pass filter example of the previous paragraph). Sometimes that interpretation in terms of the impulse response is not so straightforward. If the delay depends on frequency, where is the peak of the impulse response? Which delay should be used? The one at very low frequencies, very high frequencies, mid frequencies, or some other one? They're all different, so which is the right one? What is the time domain interpretation, in terms of the impulse response, of a group delay that varies, sometimes considerably, with frequency? There is no straightforward answer to that. The upshot is that dealing with IIR and analog filters in the time domain in the context of crossovers may not always be productive.
Quote:
Originally Posted by terry j View Post

And that 'delay is not audible'.

I am not saying that either.
Quote:
Originally Posted by terry j View Post

Well, if you gotta address it from a phase angle, as it affects FR which clearly is audible, else why correct it...how can you say delay is inaudible? ( I am not saying it is YOU that is saying that, but that seems the story). In other words, delay (directly related to phase) is inaudible yet phase is (if only for what it does to FR)

As mentioned before, there are at least two entirely different contexts to be considered. Consider two low-pass filtered subwoofers only, no mains, with one sub delayed with respect to the other. Apply some kind of transient signal to the subs and vary the delay of one relative to the other. At which delay value do the sounds appear distinct between the two? I don't know, but maybe Earl's 30 msec figure is right.

The other context is integration of mains and subs, which is an entirely different problem. A statement like "delay is inaudible" has no meaning without context.

I am not saying that "delay is inaudible" in either context BTW. What I am saying is that the most productive use of relative delay of mains and subs, and/or multiple subs with respect to one another, is to optimize frequency response flatness.
Quote:
Originally Posted by terry j View Post

We fix the FR by (first) fixing the phase, well that has to do something for the delay too?

Yes.
rock_bottom is offline  
post #73 of 86 Old 02-08-2013, 09:18 AM
FOH
AVS Special Member
 
FOH's Avatar
 
Join Date: Apr 2010
Location: Midwest
Posts: 4,749
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 16 Post(s)
Liked: 211
Nice, interesting discussion gentlemen.
Quote:
Originally Posted by rock_bottom View Post


What I am saying is that the most productive use of relative delay of mains and subs, and/or multiple subs with respect to one another, is to optimize frequency response flatness.
Yes.

^ ^ ^

+1


From my first post in this thread, back on page one;
Quote:
Just as it's clearly obvious how important the alignment of a sub system, to the remaining speakers in a bass managed/crossover system are, it's similarly important to align each sub source within that group. It works.

Right?

confused.gif

I've not yet seen the Geddes vid, nor do I understand his assertions wrt this issue. But ...

This we know;
We're quite easily fooled psychoacoustically.
Measurements rule the day.

Flat response, a house curve, or tilted up bass response, call it what you will, but we know a smooth FR is prefferrable.

So, if FR smoothing is affected via manipulation of time/signal alignment between the mains and sub(s), or sub to sub, then why would it not be prudent to pursue individual time alignment in a multi-sub approach?


Thanks

------------------------------------
Flat, Deep, Clean, Linear, and Loud
------------------------------------
Active 16.8kw, 7.3 system
(3)Seaton Cat12C up front, (4)QSC K8 sides/rears
(2)Seaton SubM-HP, (4)18" IB
FOH is offline  
post #74 of 86 Old 02-08-2013, 09:50 AM - Thread Starter
Advanced Member
 
randyc1's Avatar
 
Join Date: Dec 2009
Location: Montreal
Posts: 651
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 67 Post(s)
Liked: 19
Wish there was a video on how to time align or phase align ( are they the same) ? Or is it simply done by Trial and Error to achieve flattest FR around the Crossover Freq,...or Overall flattest FR ????
randyc1 is offline  
post #75 of 86 Old 02-08-2013, 10:05 AM
Senior Member
 
rock_bottom's Avatar
 
Join Date: Apr 2004
Posts: 430
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 14
Quote:
Originally Posted by FOH View Post

Quote:
Originally Posted by FOH 
Just as it's clearly obvious how important the alignment of a sub system, to the remaining speakers in a bass managed/crossover system are, it's similarly important to align each sub source within that group. It works.

Right?

confused.gif

Sounds right to me!
Quote:
Originally Posted by FOH View Post

So, if FR smoothing is affected via manipulation of time/signal alignment between the mains and sub(s), or sub to sub, then why would it not be prudent to pursue individual time alignment in a multi-sub approach?

I think it is prudent, assuming the adjustment is available, such as in a miniDSP or similar.

As to why Geddes avoids using delays in his proprietary multi-sub setup software, I'm baffled about that too. I can only make a wild guess that maybe doing so caused bugs in his software? Dunno.
rock_bottom is offline  
post #76 of 86 Old 02-08-2013, 10:18 AM
Senior Member
 
rock_bottom's Avatar
 
Join Date: Apr 2004
Posts: 430
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 14
Quote:
Originally Posted by randyc1 View Post

Wish there was a video on how to time align or phase align ( are they the same) ? Or is it simply done by Trial and Error to achieve flattest FR around the Crossover Freq,...or Overall flattest FR ????

That's how Geddes recommends doing it - trial and error to get the flattest overall response. He has proprietary software that does it all with just some strange-looking filters and no delays at all. His proprietary software approach is different from what he tells people to do when trying to optimize experimentally. smile.gif There's a couple of articles of his about it on DS-21's blog. He has changed the approach numerous times, so it may not make sense to follow that approach to the letter. For example, he uses no high-pass filter on the mains at all, but that might be problematic for some, maybe even the majority of people.
rock_bottom is offline  
post #77 of 86 Old 02-08-2013, 11:46 AM
FOH
AVS Special Member
 
FOH's Avatar
 
Join Date: Apr 2010
Location: Midwest
Posts: 4,749
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 16 Post(s)
Liked: 211
Quote:
Originally Posted by rock_bottom View Post

For example, he uses no high-pass filter on the mains at all, but that might be problematic for some, maybe even the majority of people.

An often underutilized approach, the more physical sources/locations the better. Being mindful of not losing any re-directed energy.

I've read unnecessary concerns over modulating effects. However, one can only imagine the amount of radiating area assigned to cover the upper bass "punch" region in such a system, .. all the subs, all the LCR LF drivers. One needn't be overly concerned for modulating the upper range of the LCR's bass driver, since the combined Sd would be enormous!

In my system it'd be fourteen 12", 15", and 18" drivers, covering the range. eek.gifcool.gif

------------------------------------
Flat, Deep, Clean, Linear, and Loud
------------------------------------
Active 16.8kw, 7.3 system
(3)Seaton Cat12C up front, (4)QSC K8 sides/rears
(2)Seaton SubM-HP, (4)18" IB
FOH is offline  
post #78 of 86 Old 02-08-2013, 11:54 AM
AVS Special Member
 
zheka's Avatar
 
Join Date: Jan 2007
Location: Chicago burbs
Posts: 1,049
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 16
Quote:
Wish there was a video on how to time align or phase align ( are they the same) ? Or is it simply done by Trial and Error to achieve flattest FR around the Crossover Freq,...or Overall flattest FR ????
Quote:
That's how Geddes recommends doing it - trial and error to get the flattest overall response. He has proprietary software that does it all with just some strange-looking filters and no delays at all. His proprietary software approach is different from what he tells people to do when trying to optimize experimentally. smile.gif


Here is a version of the Geddes multi sub how-to I've been following with good results.

http://mehlau.net/audio/multisub_geddes/

phase control is one of the tools used. My supplementary subs only have 0 or 180 degree phase switch. I would not have been able to get good result without it, at least not in the spots I placed the subs. I cannot imagine why a continuously variable phase/delay control would not offer even more flexibility.
zheka is offline  
post #79 of 86 Old 02-08-2013, 12:23 PM
AVS Special Member
 
zheka's Avatar
 
Join Date: Jan 2007
Location: Chicago burbs
Posts: 1,049
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 16
Quote:
Originally Posted by FOH View Post

An often underutilized approach, the more physical sources/locations the better. Being mindful of not losing any re-directed energy.

I've read unnecessary concerns over modulating effects. However, one can only imagine the amount of radiating area assigned to cover the upper bass "punch" region in such a system, .. all the subs, all the LCR LF drivers. One needn't be overly concerned for modulating the upper range of the LCR's bass driver, since the combined Sd would be enormous!

In my system it'd be fourteen 12", 15", and 18" drivers, covering the range. eek.gifcool.gif

I've been doing this (mains full range, blending with subs in the upper bass region) until recently. It looks good with test signal. However I am not so sure it is without negative side effects especially in multi channel systems. The problem I am seeing is that the mains contribute only with part of the LF signal. They never get full summed subwoofer signal that includes info from other channels and LFE. So if the system relies on the mains for good FR what happens when the contribution from the mains is absent? One way to get an idea is to compare the system with the mains full range vs high passed. IME one of the two would always measure worse, depending on which scenario the system was optimized for. So to me high passing seems like a safer, more consistent choice.

What do you think?
rock_bottom likes this.
zheka is offline  
post #80 of 86 Old 02-08-2013, 12:57 PM
The Truth Hz
 
beastaudio's Avatar
 
Join Date: Dec 2010
Location: Western NC
Posts: 7,413
Mentioned: 9 Post(s)
Tagged: 0 Thread(s)
Quoted: 452 Post(s)
Liked: 718
For multi-channel movie viewing I would have to agree with you, but for stereo music listening, I think it is a little different. I have just tried dozens of different setups and formats and I always end up back at stereo with no dynamic junk on. Pure and simple, full range with or without subs smile.gif

(European models do not accept banana plugs.)

 

"If you done it, it ain't bragging." ~ Walt Whitman

 


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.

beastaudio is offline  
post #81 of 86 Old 02-08-2013, 01:05 PM
AVS Special Member
 
zheka's Avatar
 
Join Date: Jan 2007
Location: Chicago burbs
Posts: 1,049
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 16
Quote:
Originally Posted by beastaudio View Post

For multi-channel movie viewing I would have to agree with you, but for stereo music listening, I think it is a little different. I have just tried dozens of different setups and formats and I always end up back at stereo with no dynamic junk on. Pure and simple, full range with or without subs smile.gif
I agree.
zheka is offline  
post #82 of 86 Old 02-08-2013, 02:32 PM
AVS Special Member
 
Mark Seaton's Avatar
 
Join Date: Jul 2000
Location: Chicago, IL, USA
Posts: 6,037
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 62 Post(s)
Liked: 253
Quote:
Originally Posted by rock_bottom View Post


Considering the problem only in the frequency domain does not in any way imply boosting your way out of the problem. His summation of filtered sub outputs is a vector summation - that is, it takes into account both amplitude and phase, both of the individual subs and the mains. It's possible to match the phase of the summed/filtered subs and that of the mains at the crossover frequency, but this does not necessarily imply matching the corresponding delays at that frequency. The phase is just the phase, while the delay is proportional to the negative of the derivative (slope) of the phase with respect to frequency.

While a minor difference, if any at all, I would suggest the term "complex summation" rather than "vector summation" might be a little more appropriate and come with the correct connotations.
Quote:
Originally Posted by rock_bottom View Post


I was a bit puzzled that Earl would exclude individual sub delays too. While these can be compensated by other filter phase shifts to achieve flat combined response, it seems to me that, all other things being equal, this would result in more complex filters than if appropriate delays were used to accomplish the same. It makes me wonder if this may be a bug in his software that causes it to have problems when delay is introduced. This is just speculation of course.

OTOH, no delays means the IIR filters can be transformed to equivalent analog filters and used as-is. The audiophile in me loves that, as it eliminates all kinds of potential dynamic range issues, at the expense of severe lack of flexibility. But I plan to get it right, then leave it, so the lack of flexibility doesn't matter that much to me.

Phase adjustment is relatively straight forward to account for in computation and optimization software. Delay isn't so simple as for the magnitude evaluation (yes, it certainly can be done). I suspect any curve fit or optimization routines might have more trouble with the changes from delay adjustments. I know this can be a big pain to add to a CFD model.

To be clear, phase and fixed delay adjustments are simply 2 different types of group delay adjustments vs. frequency, just as a high pass and a shelf filter have different magnitude effects. I feel delay is more appropriate to the most common problems faced, but I'm sure there are cases where the opposite might be true. In a simple comparison, a straight delay has much less relative change at lowering frequencies vs. higher frequencies, where phase has a more constant change to magnitude vs. frequency. Most often the subs combine just fine with very little adjustment below 25Hz. It is at higher frequencies we need more room to manipulate without causing cancellation at VLF.

When ever reading Earl's approaches, be sure to remember that from his perspective, there should never be any concerns of sufficient VLF output. I suspect he'd be shaking his head at the expectations most here have for their subwoofer systems.

Mark Seaton
Seaton Sound, Inc.
"Make no little plans; they have no magic to stir men's blood..." Daniel H. Burnham
Mark Seaton is offline  
post #83 of 86 Old 02-08-2013, 02:39 PM
AVS Special Member
 
Mark Seaton's Avatar
 
Join Date: Jul 2000
Location: Chicago, IL, USA
Posts: 6,037
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 62 Post(s)
Liked: 253
Quote:
Originally Posted by randyc1 View Post

Wish there was a video on how to time align or phase align ( are they the same) ? Or is it simply done by Trial and Error to achieve flattest FR around the Crossover Freq,...or Overall flattest FR ????

I've written it out here and on my forum many times. Assuming measurement capabilities like REW or OmniMic, I would call it guided or intelligent trial and error. You are also co-mingling 2 calibration steps.

First you get the subs blended together to drive the room efficiently. This comes from some or partial combination of placement, EQ, delay and relative phase adjustment. Part B of this first step is to then smooth the frequency response to a desired target.

The second step is to combine the subwoofer with the main speakers. For any system prioritizing movie/multi-channel playback, this should start with the center channel followed by a best compromise with the L/R speakers. In this step you only adjust the subwoofer *system* as a whole, not individually.

Mark Seaton
Seaton Sound, Inc.
"Make no little plans; they have no magic to stir men's blood..." Daniel H. Burnham
Mark Seaton is offline  
post #84 of 86 Old 02-08-2013, 03:41 PM
Advanced Member
 
terry j's Avatar
 
Join Date: Aug 2006
Posts: 866
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 2 Post(s)
Liked: 14
Quote:
Originally Posted by rock_bottom View Post

Not a dig at all. I'd love to have a DEQX, but their previous lowest-cost device seems to have been discontinued and replaced with one that's out of my price range.
I am not saying exactly that. But the relationship is more straightforward for linear-phase FIR filters than for IIR, so let's talk about them first. For such filters, the group delay, which is proportional to the negative of the slope of the phase vs. frequency, is constant for all frequencies (because phase shift vs. frequency is a straight line, so its slope is constant for all frequencies). If you were to look at the impulse response of a linear-phase FIR low-pass filter having a delay of, say, 1 msec, its peak will occur 1 msec after the impulse is applied.

Now consider an IIR or analog low-pass filter. Its phase vs. frequency is not linear, so its slope of phase vs. frequency is not constant with frequency. Therefore its group delay is not constant with frequency either. The group delay is just a mathematical construct though, and it doesn't always match up with intuition. Sometimes it has a clear interpretation in terms of the impulse response (such as the linear-phase FIR low-pass filter example of the previous paragraph). Sometimes that interpretation in terms of the impulse response is not so straightforward. If the delay depends on frequency, where is the peak of the impulse response? Which delay should be used? The one at very low frequencies, very high frequencies, mid frequencies, or some other one? They're all different, so which is the right one? What is the time domain interpretation, in terms of the impulse response, of a group delay that varies, sometimes considerably, with frequency? There is no straightforward answer to that. The upshot is that dealing with IIR and analog filters in the time domain in the context of crossovers may not always be productive.
I am not saying that either.
As mentioned before, there are at least two entirely different contexts to be considered. Consider two low-pass filtered subwoofers only, no mains, with one sub delayed with respect to the other. Apply some kind of transient signal to the subs and vary the delay of one relative to the other. At which delay value do the sounds appear distinct between the two? I don't know, but maybe Earl's 30 msec figure is right.

The other context is integration of mains and subs, which is an entirely different problem. A statement like "delay is inaudible" has no meaning without context.

I am not saying that "delay is inaudible" in either context BTW. What I am saying is that the most productive use of relative delay of mains and subs, and/or multiple subs with respect to one another, is to optimize frequency response flatness.
Yes.

thanks indeed for the time you took! (ha, maybe I only view how much effort people put into their posts by my own...I have to correct so many damned typos it takes me forever! You OTOH maybe could type three hundred words/minute without error haha)

I have to admit (which could be frustrating to the more knowledgeable here) that in certain cases I just hunker back down to 'forget the theory, what actual steps do I take?' To be fair the correct counter to that might be 'well, if you understood the theory that will guide you'.

Anyway, I think from some of the responses here that at least some fall into my camp on that. I guess there are just some areas that I am not that willing to take the time to study to come to a 'full intellectual understanding'.

I mean it is all well and good to say 'earle has some proprietary and secret way to overcome all this'. We're (us dummies) are not really any better off. And a further subset of those dummies (me as but one) are either not able or unwilling to delve into what can be viewed as difficult intellectual territory.

Even on a philosophical level I can't quite grasp the 'it is inaudible in sub territory' when it clearly is, if not in terms of 'oh I can hear that one or both of those subs are leading or lagging each other or the mains' but certainly in terms of 'oh, I can hear the non flat response that has resulted'. It IS a simple experiment to (ha, well only if you can of course) do a sweep, then delay/advance-even by a small amount- a sub, or both differently, and do another sweep and see the often widely varying FR that results.

I did manage to remember what photo hosting service I used, but even then my appalling lack of documentation means I have to guess (a few years later) what graph is what!

I know I did the above experiment but don't recall exactly the set up back then, so take these as illustrative only, just to show the point.

(oh, even tho I am quoting you RB, this is just general talk)

Now to remember how to post graphs! wish me luck

These are kinda in a sequence so it is reasonable that they are together to show the point I was making

Firstly, here is the FR of two sweeps, note the differences





Ok, that worked. The crossover point looks about right, so am confident I have got the right pics. If not, just view them as demonstrating the point.

We could see the 'poor' response, decide that we do not need to bother with delays, which DOES seem to be the general thrust, and if the system is capable enough apply eq and correct the response. Of course in doing that if we find it is a room null then we don't boost.

Here are the two measurements of the impulse arrival times that precede that graph, the arrival times 'as is' and then adjusted. They correspond to the two different sweeps just shown











(ah, when I inserted one of them it did have the title 'subs aligned', so am even more confident that these are the correct graphs) I had already aligned the subs with each other, then as a group aligned them to the mains. It would have been even more stark had the two subs not been aligned.

As I said, if you can do it it is a very easy experiment to simply change the delay on a sub and compare the resulting FR.

So even if the ONLY bottom line is to fix the FR, it makes sense to align the bass sources prior to eq.

It actually drives me insane in the bass if things are not aligned. I am aware that I seem to be running counter to the prevailing view that 'different arrival times 'down there' are unimportant' (oh yeah? tell that to your wife haha), yet I am also aware that we can fool ourselves. It is hard to imagine it is all expectation bias or whatever, yet it is equally hard to ignore my reactions to non aligned bass.

Oh, this being a home theatre forum perhaps I'd better also say this is all about 2 ch? I don't have a HT set up.
terry j is offline  
post #85 of 86 Old 02-08-2013, 04:36 PM
FOH
AVS Special Member
 
FOH's Avatar
 
Join Date: Apr 2010
Location: Midwest
Posts: 4,749
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 16 Post(s)
Liked: 211
Quote:
Originally Posted by zheka View Post

I've been doing this (mains full range, blending with subs in the upper bass region) until recently. It looks good with test signal. However I am not so sure it is without negative side effects especially in multi channel systems. The problem I am seeing is that the mains contribute only with part of the LF signal. They never get full summed subwoofer signal that includes info from other channels and LFE. So if the system relies on the mains for good FR what happens when the contribution from the mains is absent? One way to get an idea is to compare the system with the mains full range vs high passed. IME one of the two would always measure worse, depending on which scenario the system was optimized for. So to me high passing seems like a safer, more consistent choice.

What do you think?

Good point, I'm currently operating a more conventional, high passed method.

However, even though double bass scenario sends mains bass to the subs, and doesn't reciprocate, the manner in which films are mixed these days allows for much latitude.

What I mean is many of the big special effects laden films, mix strong amounts of deep bass across the mains. There are those that possess these effects on the mains, and not the LFE channel! Sometimes both,... so generalizations are tough to make wrt LFE, and where it resides. Myself, I don't much care as long as it's well executed, not clipped, not high passed. I mean, .. it's like the ol' west out there.

------------------------------------
Flat, Deep, Clean, Linear, and Loud
------------------------------------
Active 16.8kw, 7.3 system
(3)Seaton Cat12C up front, (4)QSC K8 sides/rears
(2)Seaton SubM-HP, (4)18" IB
FOH is offline  
post #86 of 86 Old 02-08-2013, 04:45 PM
Senior Member
 
rock_bottom's Avatar
 
Join Date: Apr 2004
Posts: 430
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 14
Quote:
Originally Posted by Mark Seaton View Post

When ever reading Earl's approaches, be sure to remember that from his perspective, there should never be any concerns of sufficient VLF output. I suspect he'd be shaking his head at the expectations most here have for their subwoofer systems.

Hi Mark,

Yes, I noticed also that the descriptions of the setup procedure that Earl has been giving for some time now have the property that when measuring, you start with one sub, then add in another, then another, and so on, and each added sub contributes less to the total output than the previous one. As you've also mentioned in the past, that's sub-optimal (if you'll pardon the pun biggrin.gif) from the POV of maximizing total output SPL.

But then a link to a presentation video of his was posted, in which he shows what his proprietary multi-sub software does. After watching it, I realized that what he is doing in this process is different from what he has been telling others to do all this time. This actually does make sense, as humans can't do the extreme number of iterations that optimization software might do. You can also download the PowerPoint presentation used in the video. In slide 29 of that presentation, he shows the output of each sub after individual EQ. It's clear that the green and blue traces are very close to each other in level, at least when averaged over frequency. This strongly suggests that his optimization algorithm does not favor the first sub over the second (and so on) as his past descriptions of how to manually do the setup imply. In fact, according to Earl, his optimization algorithm doesn't "know" anything about sub position or capability, or anything except for what's implied by the frequency response. That suggests that in a worst-case scenario, the algorithm might provide maximum boost to the least capable sub! At any rate, I think using multiple identical subs is probably the safest bet.

I've been looking into open-source global optimization software, and it seems that one might be able to constrain the solution such that the worst-case difference in the average output in dB SPL over some frequency band of any two subs not exceed some number, say 6 dB or something. Anyway, it's just something I've been thinking about.
rock_bottom is offline  
Reply DIY Speakers and Subs

User Tag List

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off