Multi-Sub Optimizer ( a FREE multi-sub optimizer program)

[ttocs]

One really important thing to note about MSO is the importance of how the system is setup and measurements taken by REW.

When getting serious about tweaking my subwoofers, I swap XLR cables around so I can use a Speaker Channel setup as Large so the signal sent to any and all subwoofers is not getting rolled off like what happens if Channel 4 (the LFE channel) in REW is used, or, if one wanted to use a processor channel setup as Small with a 200Hz crossover setting which also starts to rolloff way too soon to be useful when trying to work with EQ apps like MSO.

I ran some MSO runs yesterday and decided to try to use the LFE channel just to see what would happen, and it was a disaster. Setting up MSO requires some guessing as to how to tell MSO to only go so far with the EQ by way of using the Center Frequency setting. Then you're left with some spillover at a certain point, not to mention how drastically MSO cuts the levels both in the PEQ and just the regular Gain cuts. I ended up with a filter that buried everything MSO could find way down at -20dB, in addition to reduced Gain settings that were needed in the miniDSP. MSO tries to raise up whatever it's allowed to see, which in the case of the LFE channel or any channel with a crossover setting applied includes the range of frequencies that are being rolled off. So you want a flat response to work with that is within the range of the subwoofers involved. Anything outside that range of what the subwoofer can really cope with will reduce the final output dramatically.

So just make it simple and swap cables with a speaker channel, or two, or three, whatever floats your boat.

I'm still finding that I get the best results, those more to my liking, when I diligently use Impulse information to align the subs to each other, then run a combined sweep in REW, then use REW EQ to make an EQ filter. This has more consistent results and also the highest output level I could ever ask for prior to running Dirac. This method also doesn't mess up the Impulse response of the final filter as much as MSO does, and this is where I notice the most loss of dynamics with MSO filters.

[marcl]

Jan 30, 2023 15:30:35 GMT -5 ttocs said:

One really important thing to note about MSO is the importance of how the system is setup and measurements taken by REW.

When getting serious about tweaking my subwoofers, I swap XLR cables around so I can use a Speaker Channel setup as Large so the signal sent to any and all subwoofers is not getting rolled off like what happens if Channel 4 (the LFE channel) in REW is used, or, if one wanted to use a processor channel setup as Small with a 200Hz crossover setting which also starts to rolloff way too soon to be useful when trying to work with EQ apps like MSO.

I ran some MSO runs yesterday and decided to try to use the LFE channel just to see what would happen, and it was a disaster. Setting up MSO requires some guessing as to how to tell MSO to only go so far with the EQ by way of using the Center Frequency setting. Then you're left with some spillover at a certain point, not to mention how drastically MSO cuts the levels both in the PEQ and just the regular Gain cuts. I ended up with a filter that buried everything MSO could find way down at -20dB, in addition to reduced Gain settings that were needed in the miniDSP. MSO tries to raise up whatever it's allowed to see, which in the case of the LFE channel or any channel with a crossover setting applied includes the range of frequencies that are being rolled off. So you want a flat response to work with that is within the range of the subwoofers involved. Anything outside that range of what the subwoofer can really cope with will reduce the final output dramatically.

So just make it simple and swap cables with a speaker channel, or two, or three, whatever floats your boat.

I'm still finding that I get the best results, those more to my liking, when I diligently use Impulse information to align the subs to each other, then run a combined sweep in REW, then use REW EQ to make an EQ filter. This has more consistent results and also the highest output level I could ever ask for prior to running Dirac. This method also doesn't mess up the Impulse response of the final filter as much as MSO does, and this is where I notice the most loss of dynamics with MSO filters.

What is the advantage of doing some EQ in the miniDSP based on MSO, and then letting Dirac do it also?  I understand the benefit of MSO to get the alignment, but seems like the miniDSP filters will just add some phase shift that Dirac will have to deal with.  

[ttocs]

Jan 30, 2023 15:48:31 GMT -5 marcl said:

What is the advantage of doing some EQ in the miniDSP based on MSO, and then letting Dirac do it also?  I understand the benefit of MSO to get the alignment, but seems like the miniDSP filters will just add some phase shift that Dirac will have to deal with.  

Phase shift might explain some of the lackluster performance of otherwise great "looking" response curves. I don't know how to answer your question. I just started using MSO a few weeks ago just to see what it's all about, and using some settings can get some good results, but none so far have gotten close to what REW can do via REW EQ + Dirac.

Dirac can eek out some good stuff on top of MSO filters, but it always seems to be just a bit tooooooo much loss of SPL and dynamics, and it's the loss of dynamics that frustrates me the most with MSO.

So this week I'm just concentrating on REW + Dirac only. Last night I re-tweaked the alignment of the subs - again. Each time I go through this I discover something I hadn't done before, or how to do something a bit better.

For the last bunch of weeks I've been using the Phase Control on each subwoofer to get the Impulse peaks exactly lined up. Prior to this I was either using the Alignment Tool to adjust the sliders for the best looking response and throwing caution to the wind, and got some great results by the way, or, I was using the delay that REW calculated for each sweep as the number to work with (which I could not get consistent results with). But using just IMPULSE is resulting in way more output level to work with prior to any EQ. Then, I use EQ to make a filter which doesn't have near as much PEQ stuff in it as MSO has but Dirac seems to work with it pretty well, and, it sounds way better than what MSO + Dirac can come up with. Anyone who might not be interested in anything audio can hear the difference, especially with something that can make you jump out of your seat like that first hit of the drum in SRV's Tin Pan Alley!

[donh50]

MSO allows you to integrate the subs together to optimize the overall (net) subwoofer response, and then Dirac Live (or whatever) has just one channel to work with instead of however many independent subs you have. There are all sorts of debates about the best way to integrate a group of subs, whether as a group or individually and adding the results, but "most" articles I have read say getting all the subs working together to resolve room issues, then treating them as a single channel to the processor provides better results. Less load on the processor, and the hard work of actually aligning the subs so they work together to produce the optimal response at the MLP is handled by MSO (or some other tool like SFM or an Antimode). This is a big deal when you have two or more subs, probably not so much with a single sub. Older Audyssey program had far fewer LF filters, some other programs (Pioneer's MCCAC comes to mind) only had 2-3, and few processors have more than two sub channels. For that matter, a few years ago some of the AVRs with two sub outputs just placed them in parallel, and even the better ones would often just delay them independently with the same correction filters applied to both. Dirac Live, at least the non-DLBC version, treats the subs like any other speaker. I know I had to do some tweaking to get my subs aligned using Dirac Live and my XMC-1. And of course there are many AVRs that only have one sub output.

Once MSO, or something like it, has all the subs doing their thing (setting amplitude and phase/delay for each), the resulting single channel that the AVR deals with is more easily (and properly, IME) integrated with the rest of the speakers, crossover, and such compared to having individual subwoofer channels.

But, as always, YMMV - Don

[KeithL]

I can think of a few "affirmative arguments" for the idea of using multiple correction mechanisms.
Each of those solutions has a limit on the number, magnitude, and type of the correction it can do.
None of them can create or process an unlimited number of filters or create filters of every known type... 
So, arguably, like having a huge number of different files in his or her toolbox, a master craftsman has a better chance of having exactly the one they need. 
Also, while we know in general what each is doing, we do not in fact know all the details.
For example, with Dirac Live, we really don't get the details of how many biquads are being used, and the exact characteristics of each.
And the same is technically true foe each of those other solutions as well.
And I'd rather not think about the potential number of permutations possible when "stacking" and "overlapping" different types of filters.

However there are also an equal number of rather obvious negative possibilities... although those fall into two main categories.
The first is that you are applying multiple levels of processing, and potentially multiple conversions, to your audio signal.
In some cases, for example if you use Dirac Live in one of our processors, and then an external MiniDSP, you may actually have multiple A/D and D/A conversions.
But, even if you somehow manage to avoid that, there are various "conversions" and "transformations" going on inside each DSP itself.
(If you believe that processing the audio in a single DSP has any deleterious effects then, by using multiple levels, you are "stacking up" those negative effects too.)

The second category of negative effect, which also counts as a positive effect, is that of adding complexity and reducing "process transparency".
The simplest way to explain this is by saying that "by the time you're done you really don't know exactly what you did".
(When I make chili I use a certain brand of stewed tomatoes, and a certain brand of barbecue sauce, and some ketchup, and a bunch of spices.
Even though I know what spices I put in, since I don't know the details of what's in the tomatoes, or the chili powder, I really don't know the whole recipe.)
So, when all is said and done, for example, you could have an error in the MiniDSP's corrections that, by chance, just happens to cancel out an error in Dirac Live's corrections.
And, while this may work out well, it makes predicting exactly what will happen when you change something extremely difficult (or just plain unlikely).
It's also worth noting that, when Dirac Live shows you "the corrected response" on-screen it's showing you a digital model of what it expects that response to be.
And, likewise, when you take room measurements, in a real room, with a microphone, they aren't going to be totally accurate or repeatable either.

In short, there are a LOT of variables with room correction and measurement to begin with, and if you combine multiple solutions you're adding a lot more to the pile...
And, of course, there is the little matter of more stuff to buy, more wires to connect, and more controls to fiddle with...
(Which I guess you might count as a positive or a negative.)

My guess is that, unless you do this for a living, eventually you'll get bored and actually sit down and listen to some music... 
(And do try your best to keep your eye on that point of diminishing returns.)

Jan 30, 2023 15:48:31 GMT -5 marcl said:

Jan 30, 2023 15:30:35 GMT -5 ttocs said:

One really important thing to note about MSO is the importance of how the system is setup and measurements taken by REW.

When getting serious about tweaking my subwoofers, I swap XLR cables around so I can use a Speaker Channel setup as Large so the signal sent to any and all subwoofers is not getting rolled off like what happens if Channel 4 (the LFE channel) in REW is used, or, if one wanted to use a processor channel setup as Small with a 200Hz crossover setting which also starts to rolloff way too soon to be useful when trying to work with EQ apps like MSO.

I ran some MSO runs yesterday and decided to try to use the LFE channel just to see what would happen, and it was a disaster. Setting up MSO requires some guessing as to how to tell MSO to only go so far with the EQ by way of using the Center Frequency setting. Then you're left with some spillover at a certain point, not to mention how drastically MSO cuts the levels both in the PEQ and just the regular Gain cuts. I ended up with a filter that buried everything MSO could find way down at -20dB, in addition to reduced Gain settings that were needed in the miniDSP. MSO tries to raise up whatever it's allowed to see, which in the case of the LFE channel or any channel with a crossover setting applied includes the range of frequencies that are being rolled off. So you want a flat response to work with that is within the range of the subwoofers involved. Anything outside that range of what the subwoofer can really cope with will reduce the final output dramatically.

So just make it simple and swap cables with a speaker channel, or two, or three, whatever floats your boat.

I'm still finding that I get the best results, those more to my liking, when I diligently use Impulse information to align the subs to each other, then run a combined sweep in REW, then use REW EQ to make an EQ filter. This has more consistent results and also the highest output level I could ever ask for prior to running Dirac. This method also doesn't mess up the Impulse response of the final filter as much as MSO does, and this is where I notice the most loss of dynamics with MSO filters.

What is the advantage of doing some EQ in the miniDSP based on MSO, and then letting Dirac do it also?  I understand the benefit of MSO to get the alignment, but seems like the miniDSP filters will just add some phase shift that Dirac will have to deal with.  

[405x5]

One of the first things I ever did regarding Subwoofer integration was to the dispense with the redundancy……(decades ago)

[KeithL]

That is a perfect "high-level summary" of how it all works in theory.

However you always need to bear in mind that many of the blocks in your block diagram are just convenient approximations.
The reason all those debates still exist is because there is no single solution that actually solves everything.
Even with a single sub, and a single listening position, you're forced to choose between the flattest frequency response and the best impulse response.
Add more subs, and you have that same choice for each, then a choice for how to optimize the way they interact.
For example, unless your subs are stacked in one pile, they cannot be in phase, at all frequencies, relative to your listening position. 
And, again, more choices and compromises between the size of the listening area, and how accurately you want to correct things.
(So MSO really can't "make all of your subs act exactly like a single sub".... only "sort of"...)

However, there is also a "meta-problem" in your suggestion.
Because all of this is so complex, and there is no single complete solution, MSO is going to include a variety of "compromises and choices".
And, similarly, Dirac Live is ALSO going to incorporate a whole bunch of assumptions, compromises, and "artistic choices" into the optimizations it uses.
And, to put it bluntly, their "optimizations", "assumptions", and "artistic choices" may not work well together.

As many people have noted... in the real world room correction is "part art and part science"...
And, that being the case, you run the risk of "hiring three different artists to decorate your living room"...
Their artistic choices may or may not work well together...

In this case, while the "goal" of MSO may be "to get all of your subs to work together as a single optimized sub"....
In practice, when MSO finishes "doing its thing", what you end up with is a LOT more complex than a single sub at a single location....
And, because of that, you really cannot count on Dirac Live being able to handle it as effectively as if it really was a single sub.... 
(So, in the end, the combination may work great, or it may produce... somewhat unexpected results.)

My point is merely that, by using multiple complex solutions, like MSO and Dirac Live, you really are adding yet another level of complexity to the situation.
MSO's treatment of your subs as "a single sub" is still just an approximation... which may serve in some contexts...
But, at another level, you do still have a bunch of separate subs, at different locations, and a bunch of separate speakers, all interacting with the room and each other...
I would agree that it makes things easier... but I would also suggest that this is more true with something like the simple room correction in an AVR than with something complex like Dirac Live.

Jan 31, 2023 10:14:19 GMT -5 donh50 said:

MSO allows you to integrate the subs together to optimize the overall (net) subwoofer response, and then Dirac Live (or whatever) has just one channel to work with instead of however many independent subs you have. There are all sorts of debates about the best way to integrate a group of subs, whether as a group or individually and adding the results, but "most" articles I have read say getting all the subs working together to resolve room issues, then treating them as a single channel to the processor provides better results. Less load on the processor, and the hard work of actually aligning the subs so they work together to produce the optimal response at the MLP is handled by MSO (or some other tool like SFM or an Antimode). This is a big deal when you have two or more subs, probably not so much with a single sub. Older Audyssey program had far fewer LF filters, some other programs (Pioneer's MCCAC comes to mind) only had 2-3, and few processors have more than two sub channels. For that matter, a few years ago some of the AVRs with two sub outputs just placed them in parallel, and even the better ones would often just delay them independently with the same correction filters applied to both. Dirac Live, at least the non-DLBC version, treats the subs like any other speaker. I know I had to do some tweaking to get my subs aligned using Dirac Live and my XMC-1. And of course there are many AVRs that only have one sub output.

Once MSO, or something like it, has all the subs doing their thing (setting amplitude and phase/delay for each), the resulting single channel that the AVR deals with is more easily (and properly, IME) integrated with the rest of the speakers, crossover, and such compared to having individual subwoofer channels.

But, as always, YMMV - Don

[KeithL]

From a technical perspective, I'm also inclined to prefer a single sub, in terms of simplest setup and interactions with the room.
Although I've never tried it, if I was going to have multiple subs, I would stack them in a single floor-to-ceiling line array.
(A full line source actually makes the math and the required corrections simpler... at least in theory.)

Jan 31, 2023 10:36:52 GMT -5 405x5 said:

One of the first things I ever did regarding Subwoofer integration was to the dispense with the redundancy……(decades ago)

[ttocs]

Jan 31, 2023 10:30:40 GMT -5 KeithL said:

It's also worth noting that, when Dirac Live shows you "the corrected response" on-screen it's showing you a digital model of what it expects that response to be.
And, likewise, when you take room measurements, in a real room, with a microphone, they aren't going to be totally accurate or repeatable either.

The Dirac predicted is usually pretty optimistic, but quite often looks pretty close to measured, with some caveats. But every single time I've ever used REW's Alignment Tool, the actual measured response matches almost exactly to what the Alignment Tool predicts. So much so, that I can absolutely trust what it predicts.

MSO's final response curves are pretty close - mostly, but can have more problematic nulls when measured, they can "grow" badly. So one needs to get a bunch o MSOf filters loaded and measured to get a sense of the pitfalls MSO can get into so as to avoid wasting time with them just by looking at the final response curve when the Optimization has ended. Whatever looks like a slight problem can grow into one that's unrecoverable via Dirac correction. In other words, Dirac can hate what it has to work with.

So the two greatest issues with MSO for me are nulls that can be worse than predicted, and, the serious amount of reduced final output level if MSO is not reigned in a lot - which can reduce its capability - which can then reduce its attraction to being used in the first place.

When megash0n was having trouble with the output of his subs, I now see the problem was that MSO simply reduced the output a lot to get great looking flat response curves and hiding much of that reduction in the PEQ filters. Remove the filters, and the SPL goes way up. I can live with gain being reduced by around -6dB, but not -12dB or even -26dB (overall vs no filter)!  Give me peaks/nulls with higher output, at least, the peaks/nulls I'm used to seeing in my room.

[marcl]

Jan 31, 2023 10:30:40 GMT -5 KeithL said:

I can think of a few "affirmative arguments" for the idea of using multiple correction mechanisms.
Each of those solutions has a limit on the number, magnitude, and type of the correction it can do.
None of them can create or process an unlimited number of filters or create filters of every known type... 
So, arguably, like having a huge number of different files in his or her toolbox, a master craftsman has a better chance of having exactly the one they need. 
Also, while we know in general what each is doing, we do not in fact know all the details.
For example, with Dirac Live, we really don't get the details of how many biquads are being used, and the exact characteristics of each.
And the same is technically true foe each of those other solutions as well.
And I'd rather not think about the potential number of permutations possible when "stacking" and "overlapping" different types of filters.

However there are also an equal number of rather obvious negative possibilities... although those fall into two main categories.
The first is that you are applying multiple levels of processing, and potentially multiple conversions, to your audio signal.
In some cases, for example if you use Dirac Live in one of our processors, and then an external MiniDSP, you may actually have multiple A/D and D/A conversions.
But, even if you somehow manage to avoid that, there are various "conversions" and "transformations" going on inside each DSP itself.
(If you believe that processing the audio in a single DSP has any deleterious effects then, by using multiple levels, you are "stacking up" those negative effects too.)

The second category of negative effect, which also counts as a positive effect, is that of adding complexity and reducing "process transparency".
The simplest way to explain this is by saying that "by the time you're done you really don't know exactly what you did".
(When I make chili I use a certain brand of stewed tomatoes, and a certain brand of barbecue sauce, and some ketchup, and a bunch of spices.
Even though I know what spices I put in, since I don't know the details of what's in the tomatoes, or the chili powder, I really don't know the whole recipe.)
So, when all is said and done, for example, you could have an error in the MiniDSP's corrections that, by chance, just happens to cancel out an error in Dirac Live's corrections.
And, while this may work out well, it makes predicting exactly what will happen when you change something extremely difficult (or just plain unlikely).
It's also worth noting that, when Dirac Live shows you "the corrected response" on-screen it's showing you a digital model of what it expects that response to be.
And, likewise, when you take room measurements, in a real room, with a microphone, they aren't going to be totally accurate or repeatable either.

In short, there are a LOT of variables with room correction and measurement to begin with, and if you combine multiple solutions you're adding a lot more to the pile...
And, of course, there is the little matter of more stuff to buy, more wires to connect, and more controls to fiddle with...
(Which I guess you might count as a positive or a negative.)

My guess is that, unless you do this for a living, eventually you'll get bored and actually sit down and listen to some music... 
(And do try your best to keep your eye on that point of diminishing returns.)

Jan 30, 2023 15:48:31 GMT -5 marcl said:

What is the advantage of doing some EQ in the miniDSP based on MSO, and then letting Dirac do it also?  I understand the benefit of MSO to get the alignment, but seems like the miniDSP filters will just add some phase shift that Dirac will have to deal with.  

Some things we do know ... sort of ...

Although the miniDSP HD can do FIR filters, my guess is most people don't use them and instead use the IIR filters with biquads generated by REW.  Or they use a miniDSP 2x4 that only has IIR filters.  IIR filters always move phase and amplitude, while FIR filters can adjust amplitude without affecting phase.  It's a bit of a generalization since Dirac's filter technology is their "secret sauce", but I have heard them say in various interviews that they use IIR filters in the lower frequencies and FIR elsewhere ... and some proprietary mix that may have other filter algorithms.  I think it's safe to say that even though the number of biquads and FIR taps available is limited, Dirac has far more available than REW.  As for A/D/A conversions ... once there's a miniDSP in the picture that conversion happens once whether you use the PEQ or not.

So I agree that if you do both there can be some doing by one and undoing by the other.  I've tried EQ'ing each sub in the miniDSP and then doing Dirac overall, EQ'ing the subs overall in miniDSP and then doing Dirac overall again, and doing no PEQ in miniDSP and using Dirac overall.  I can't say I found any advantage to the first two approaches so I settled on the simple latter approach.  Time alignment in miniDSP is the real bang for the buck for sure ... it's the thing that Dirac can't optimize with multiple subs without DLBC.  Also, I never count on any EQ to lift nulls below 200Hz.  No matter what Dirac says theoretically about some degree of boost, they really don't and can't do that much with modal resonances.  So I use every method including speaker placement and especially time alignment to get sub levels as high as possible .... then I never raise the bridge, I only lower the river.  And for me this means using a Dirac sub target that's about -3db overall, then boost the sub level in the miniDSP to compensate.

[marcl]

Worthy of a reminder ... Welti did science ...


Welti_Multisub.pdf (1 MB)

multsubs.pdf (814.85 KB)

He came up with the simplest answer for an acoustically small room, which was two or four subs.  This as a basis before going further with phase alignment or EQ.

He didn't do stacked subs.  I've looked for technical info on using stacked subs in acoustically small rooms and haven't found anything other than speculation or sales pitches ... certainly nothing with the rigor of Welti's methods.  Stacks are discussed for use in large sound reinforcement scenarios, and there is always a bit about the lobes, nodes and the space that it takes for them to spread and come together.  One sound tech told me the technique puts a null at the stage so the musicians don't explode ... then they get their bass near-field.  That doesn't work for a small room.  I would like to see the science around whether stacks can work better than four corner subs in an acoustically small room, and if so, how does all that beaming stuff work to combat modal resonances (which are generally not a factor in large venues), and seat to seat consistency ... and eliminating cancellations.

[ttocs]

Jan 31, 2023 11:03:35 GMT -5 KeithL said:

However, there is also a "meta-problem" in your suggestion.
Because all of this is so complex, and there is no single complete solution, MSO is going to include a variety of "compromises and choices".
And, similarly, Dirac Live is ALSO going to incorporate a whole bunch of assumptions, compromises, and "artistic choices" into the optimizations it uses.
And, to put it bluntly, their "optimizations", "assumptions", and "artistic choices" may not work well together.

As many people have noted... in the real world room correction is "part art and part science"...
And, that being the case, you run the risk of "hiring three different artists to decorate your living room"...
Their artistic choices may or may not work well together...

In this case, while the "goal" of MSO may be "to get all of your subs to work together as a single optimized sub"....
In practice, when MSO finishes "doing its thing", what you end up with is a LOT more complex than a single sub at a single location....
And, because of that, you really cannot count on Dirac Live being able to handle it as effectively as if it really was a single sub.... 
(So, in the end, the combination may work great, or it may produce... somewhat unexpected results.)

Yep.

And I'm finding that MSO, while convenient in one sense - being able to do its supercomputer thing and try a gazillion combinations in mere minutes - is very complex and needs exacting execution by the user to measure, import, export, import and change settings, without a single error otherwise the result will be mass confusion.

I still say MSO is a great app and has its place. And I think that place is squarely aimed at HT and the broadening of the sweet spot for multiple seats so each is less affected by peaks and nulls. So long as the user has enough subwoofer output to overcome the expected reduction of overall SPL, then it's all hunky dory. If I expect output to be reduced by say, -12dB, then I just need more subs to compensate.

[ttocs]

Jan 31, 2023 12:01:45 GMT -5 marcl said:

Worthy of a reminder ... Welti did science ...
He came up with the simplest answer for an acoustically small room, which was two or four subs.  This as a basis before going further with phase alignment or EQ.

He didn't do stacked subs.  I've looked for technical info on using stacked subs in acoustically small rooms and haven't found anything other than speculation or sales pitches ... certainly nothing with the rigor of Welti's methods.  Stacks are discussed for use in large sound reinforcement scenarios, and there is always a bit about the lobes, nodes and the space that it takes for them to spread and come together.  One sound tech told me the technique puts a null at the stage so the musicians don't explode ... then they get their bass near-field.  That doesn't work for a small room.  I would like to see the science around whether stacks can work better than four corner subs in an acoustically small room, and if so, how does all that beaming stuff work to combat modal resonances (which are generally not a factor in large venues), and seat to seat consistency ... and eliminating cancellations.

I was at the Riviera Theater a week ago and was backstage for the event (my brother represents one of the bands and I was helping out). The stage wasn't overwhelmed with bass due to how the subwoofers are arranged, but when I walked around a bit I found the bass collected to the sides of the stage and the audience was treated with really tight bass along with great sound overall from a well setup sound system. I don't know if they employed any cardioid type arrangements, I only know from listening to a couple of the sound engineers that the subs were under the front of the stage horizontally which helped to get it dispersed up to the balcony, and that the sound guy for one of the bands was impressed with the setup. This experience made me think more about one setup I have yet to try, which is to arrange four subs horizontally under the tv panel in a tight configuration.

While I've experimented with stacks taller than just 2-high, the advantage to stacking is mostly for increased output. I don't think it compares favorably to 4 subs in corners or mid-wall setups. If you have a room that can allow 4 subs to be placed symmetrically, I think that's the best thing going. And if you need more output, then stack more in each location. I haven't found stacking to be detrimental in any way.

[donh50]

Random thoughts:

The advantage of multiple subs is in placing them to excite room modes from differing locations and thus provide the ability to "drive" or cancel things like nulls caused by strong fundamental room modes ("nodes"). A stacked array in a corner will provide more output but excites basically "all" room modes so does not help in correcting them at the listening position. You may be able to place the array in a different spot to excite fewer modes, but fundamentally a closed rectangular room has a node (null) at the center so don't sit there or add subs strategically placed to counter the null (Todd Welti's papers, among others, are a good reference for sub placement).

I do not know REW's limits; I suspect the filter limit (number and length) is due more to HW limitations (e.g. miniDSP) than SW. There are other programs that allow you to create an "infinite" number and length of filters, up to the memory limitations of your PC, but I have not tried any of them (I have Mitch's book and keep thinking about trying it, but Life and Work keeps getting in the way).

IIR filters are often easier to implement for low frequencies and do not have the potential for pre-ringing that FIR filters have. A 10 Hz FIR filter may need a lot of taps...

I set my four subs as two pair, front and back, strategically placed per Welti's guidance. My XMC-1 set the delays reasonably well, and Dirac Live handled their response reasonably well, but I had to adjust their phase manually (Rythmik subs, with an analog continuous phase control knob) to dial them in. My current processor has four independent sub channels and I just let its Optimizer do it's thing. Sounds and measures OK, though someday I may try MSO or SFM.

I think everyone realizes there is no perfect solution. Just gotta' find a compromise that's best for me (you).

[KeithL]

Subs act rather differently in very large rooms... and especially in huge venues with very high ceilings.
And, as that sound engineer mentioned, sometimes the goals aren't at all the same either.
(You wouldn't be too happy with creating a deep null at your listening position... and then using an IEM to hear the bass notes.)
Very large rooms also have room modes that can be multiples of wavelengths (so you've got even and odd multiples to think about).
And, in very large rooms, the locations of subs can sometimes be audible...
(That's why "theater Atmos" uses multiple sub channels even though Dolby only ever specified a single sub channel for home systems.)

Also, the way we visualize "what's going on with speakers" is different in VERY large rooms.
We tend to think of a sort of dividing line between "higher frequencies that bounce around like light" and "low frequencies that act more like pressure".
However, in a very large venue, even low bass is somewhat directional, and "pressurizing that whole huge room" just isn't a thing.
But, unlike in a small room, you need to worry about things like actual echoes... 
(Once the delay exceeds some point, reflections can still cause cancellations, but they can also be audible as echoes.... )
(And you can end up with really nasty situations where a bass note that persists for more than a faction of a second builds up into a sort of unintelligible roar...)

And, of course, in many situations there, you have no option to add absorbers or diffusers, but that role is handled by your audience...
(The catch there is that it will vary depending on how many seats are filled... and you don't get to hear or test it until the audience actually arrives.)
(That wouldn't be a problem in "a permanent installation"... but it sure is in situations where "the band brings their own gear"... )

However, the biggest difference is that, in a small room, we assume that all of the speakers will interact, and plan to optimize this.
Whereas, in a very large venue, where speakers are going to be more than thirty of forty feet apart, you generally try to avoid interaction.
Let's say you have a concert hall that is three hundred feet square.
In order to maintain good levels you're probably going to need multiple rows of speakers.
However, if the people in back can hear both the front speakers, and the middle speakers, what they'll hear will be a jumbled mess.
The solution is to arrange things so that the people in the front part of the room hear the front row of speakers...
But the people in the back half of the room ONLY hear the middle row of speakers...
Now everybody gets to hear a nice clean presentation emanating from one set of speakers...
You do this by facing the middle row back so they dominate the back half of the room... and angling the front row down so they only reach as far back as the middle.
(And, since the middle row of speakers will be pretty loud to the folks sitting right behind them, they will drown out any sound that reaches there from the front row of speakers.) 
Forget the details... but my point is that setting up speakers for a very large venue is nothing like setting up a system that will sound good in a normal sized room.  

I personally think it would be interesting to try a full floor-to-ceiling stack of subs... essentially making a "subwoofer line array".
This would essentially "reduce some of the math to a two-dimensional problem".
You can't have floor-to-ceiling nulls and nodes when your speaker actually stretches completely from floor to ceiling.
Technically that's an oversimplification of how a line array works, and a bit optimistic, but I think the idea is sound.... 
I would be thinking of something like a 16" triangular column, from floor to ceiling, with a vertical row of 8" subs in it.
But, to be fair, I don't know of anyone who's tried it. 

I have seen "whole wall subwoofer arrays" described, but I've never heard one, and that seems rather impractical to me.

I have also seen the idea of "a full dipole subwoofer array".
Literally line the front and back walls of a room both with subs... then drive the subs in the two walls out of phase.
You end up with no "pressure buildup" but with the output of the subs acting as "a wavefront that passes you".
This pretty well meets the description of what you would experience sitting next to a massive dipole subwoofer.
(And, no, I've never heard one, nor spoken to anyone in person who has.)

Jan 31, 2023 13:15:13 GMT -5 ttocs said:

Jan 31, 2023 12:01:45 GMT -5 marcl said:

Worthy of a reminder ... Welti did science ...
He came up with the simplest answer for an acoustically small room, which was two or four subs.  This as a basis before going further with phase alignment or EQ.

He didn't do stacked subs.  I've looked for technical info on using stacked subs in acoustically small rooms and haven't found anything other than speculation or sales pitches ... certainly nothing with the rigor of Welti's methods.  Stacks are discussed for use in large sound reinforcement scenarios, and there is always a bit about the lobes, nodes and the space that it takes for them to spread and come together.  One sound tech told me the technique puts a null at the stage so the musicians don't explode ... then they get their bass near-field.  That doesn't work for a small room.  I would like to see the science around whether stacks can work better than four corner subs in an acoustically small room, and if so, how does all that beaming stuff work to combat modal resonances (which are generally not a factor in large venues), and seat to seat consistency ... and eliminating cancellations.

I was at the Riviera Theater a week ago and was backstage for the event (my brother represents one of the bands and I was helping out). The stage wasn't overwhelmed with bass due to how the subwoofers are arranged, but when I walked around a bit I found the bass collected to the sides of the stage and the audience was treated with really tight bass along with great sound overall from a well setup sound system. I don't know if they employed any cardioid type arrangements, I only know from listening to a couple of the sound engineers that the subs were under the front of the stage horizontally which helped to get it dispersed up to the balcony, and that the sound guy for one of the bands was impressed with the setup. This experience made me think more about one setup I have yet to try, which is to arrange four subs horizontally under the tv panel in a tight configuration.

While I've experimented with stacks taller than just 2-high, the advantage to stacking is mostly for increased output. I don't think it compares favorably to 4 subs in corners or mid-wall setups. If you have a room that can allow 4 subs to be placed symmetrically, I think that's the best thing going. And if you need more output, then stack more in each location. I haven't found stacking to be detrimental in any way.

[thompson12]

So I finally got a chance to play with MSO again I ran a few different optimizations with target levels from 70db up to 100db and it optimized them all pretty flat from 10db to 200db put them all in mini DSP all of them resulted in no bass. So I then tried REW's sub optimizers better but still not as much bass as with mini dsp settings flat with no adjustments at all. I'll try the old way again and sum the subs one at a time and see how that works out.


Mitch

[ttocs]

Feb 4, 2023 15:11:39 GMT -5 thompson12 said:

So I finally got a chance to play with MSO again I ran a few different optimizations with target levels from 70db up to 100db and it optimized them all pretty flat from 10db to 200db put them all in mini DSP all of them resulted in no bass. So I then tried REW's sub optimizers better but still not as much bass as with mini dsp settings flat with no adjustments at all. I'll try the old way again and sum the subs one at a time and see how that works out.

Mitch

When you look at the squiggly diagram of the PEQ filter in miniDSP, you can see how much negative gain has been applied, which can easily be -20dB - just in the filter. Then, when you take into account the amount of negative gain we are told to input into each miniDSP channel, the total negative gain can be -32dB. So yeah, this can result in no bass.

For now, I've played with MSO for a few weeks and can see it has a place for helping some systems, but not mine.

I'm pretty happy with EQ in REW. I'm going to be posting some things in the REW thread, maybe tomorrow.

[thompson12]

Yah. I liked how easy it seamed to be to use and looking at the graphs in MSO it looked like it was going to be real good but it didn't work out that way and to get the gains and delays for mini DSP was easy as well.

I'll be looking forward to your REW post.


Mitch