ZOOMING In REW (and a little Dirac)

[ttocs]

I wasn't careful enough and accidentally clicked on a checkbox I didn't mean to check. As a result, I spent most of Saturday chasing a inverted polarity problem thinking it was cabling or a component that failed.

This screenshot highlights the checkbox, and the popup dialogue box explains what it does. Either my mouse or keyboard must've been bumped and this checkbox got enabled.


I was in the middle of tuning some subs and running Dirac, and then checking results with REW. This plot shows what happens when the Right Speaker channel was inverted. In my process, I run REW in the following order: LFE, L, R, L+R, etc. It was when I ran L+R that the trace showed huge cancellation going on instead of a few dB of summing. The relatively straight faded traces are the Left and Right channels, The Purple trace is with the L&R setup as Large, and the Green trace is with them setup as Small so they are interacting with the subwoofer channel and the subwoofer channel is cancelling itself along with what the speaker channels are doing. Very messy!


This is what looks familiar to me when things are working ok and the polarity is correct. The higher level trace is the L+R.


I don't use L+R all the time, sometimes I just make quick REW runs, but it's one way to check summing/cancellation that one would not know about without this parameter.

[ttocs]

REW ALIGNMENT TOOL
This exercise uses a couple miniDSP 2x4HD for six subwoofers in two stacks of 3 each. The miniDSP is simply used to apply a delay to the audio signal and distribute the resulting signals to each subwoofer. I have not used any other capability of these miniDSP devices for this exercise. This exercise is to achieve the smoothest response curve, which does not necessarily mean the best impulse, but does have a nice smooth phase curve. In another exercise I'll go for the best impulse, which doesn't necessarily have the smoothest phase curve.

After having done similar routines which involve trying to align multiple subs together to form a group, this method has, so far, proven to be a good step forward in achieving a better result with less effort and time - at least to get this far with the result. There are other avenues of tasks that might achieve better results, but this is to show a basic, systematic approach, that I'm frankly a bit miffed I didn't come up with before last week because it's much more streamlined.

One must put some trust into the Alignment Tool, as I have learned to do. After countless times of the tool showing graphically what a particular amount of delay or gain might do when combining two individual measurements of two subwoofers, I'm still completely amazed at how accurate it is!

I use the Emotiva VSUB transmitter and receiver to send the signal from the Center Subwoofer Output of the XMC-2 to a Input on each miniDSP.
Outputs 1 through 3 are then connected to each of the 3 subwoofers in each stack of subs. I named them, from bottom up, L1, L2, and L3, for the Left Stack, and likewise for the Right Stack.
All subwoofers are setup the same, except that each subwoofer has a great feature built-in - Anthem ARC, so each sub has its own initial calibration which helps quite a bit.
Here's a photo of the system with the mess of wires, VSUBS, and miniDSP's on the floor. It's never neat and tidy when moving system components around and testing.


REW SETUP
Click on "Measure" in the upper left corner to bring up this window. In the "Name" use whatever you want each sweep to be named, but this should be changed for each sweep otherwise confusion will run rampant. I set "Range" for 20-300Hz. Just under where it says "Sweep" is the "Length" dropdown box for how long the sweep length should take, I've chosen 512k. Higher values make the sweep last longer. Below this is "Timing" and set it as shown with "Use acoustic timing reference" selected. "Output" is the speaker you want to measure. I setup the Fronts as Small and set the crossover to 200Hz, and turned off the amps for the Front Channels because I want to use this as an easy way to get a fuller range sweep going to the subwoofer channel to extend it in a flatter manner than if I chose LFE as the Speaker which has a 120Hz limit and rolls off the signal sooner at around 80Hz. The "Ref Output" is the speaker you want to use as a Timing Reference, and as shown, I use C, which is the Center Speaker Channel. It's convenient and logical as it's in the middle. I always test first using "Check Levels", and for newbies I suggest starting with very low processor volume levels and raise to your desired level. I usually aim for a result that gives around 70-80dB in the sweep plot. Depending on room or outside noise you may want this to be higher, but I wouldn't suggest much lower. And in the case of this exercise I knew I'd be adding lots of sum gain to the equation, as you'll see. 


STEP ONE
The first step is to measure each subwoofer alone. I first Mute each output of both miniDSP units, then, unmute and measure each subwoofer individually. It doesn't really matter what order you use, but I started from the bottom up on the left, then same for the right. Meticulous record keeping keeps my sanity in check. After measuring each subwoofer there is no more measuring needed till the very last step. After the first sweep, REW lands you with the first Tab selected for "SPL & Phase", so click on "All SPL" to get the screen to look like the image below. Blue is for Left subs and Red is for Right subs.


ALIGNMENT TOOL
On the right just above the graph area is the "Controls" gear. Click this and in the resulting popup click "Alignment Tool", then click the gear again to close "Controls" to get it out of the way. Move Alignment Tool to a good spot on your screen so you can use the controls and see how the Aligned trace forms. As you drag the sliders for "Delay" the Aligned trace will change radically. I try to make it form a response that has a sort of "natural" amount of jagged-ness to it, as well as the kind of response curve I'm hoping for. The only real way to know what to look for is to do an alignment, run Dirac, then measure and hear the results, then do it again and again.

I've done this in the past where I aligned each stack of subwoofers, then aligned both stacks, but this was pretty tedious to get decent results.

FIRST ALIGNED PAIRS
This week I started with the bottom pair of subs, L1 and R1, and aligned them. The L2 and R2 and aligned them. Then L3 and R3 and aligned them. So now I have three aligned pairs of subwoofers. White is the "Aligned Sum" of each pair of traces. I named each Aligned Sum: L1/R1, L2/R2, L3/R3. These represent "Group-1".


RECORD KEEPING
In each Aligned Sum is the amount of delay, if any, that needs to be entered accordingly into each channel of each miniDSP. miniDSP can only accept Positive Delay numbers in ms (milliseconds), so only move the Delay sliders to the right of Zero. What I did was to create a spreadsheet to gather the delay and be able to read all the numbers in a logical fashion. I also made a formula to add the delay numbers for me so there's no mistake. You also must pay attention to "if" any subwoofer needs to be Inverted, which is also done in the miniDSP. There are a few settings which only pertain to Martin Logan subwoofers.

In the spreadsheet there are several "Group" columns. These show the subwoofer groupings as they "grow" in number. First it's just a pair of subs, then another pair is added, then another pair. It's kinda like building a three story building using two building blocks for each level.


ALIGNING THE ALIGNED PAIRS
Next, I needed to align each Aligned Sum. So for Group-2 I chose to align L1/R1 and L2/R2 first, bottom subs with middle subs. Next I took that aligned sum Group and aligned it with the top sub Group, L1&2/R1&2 and L3/R3. In the image below it shows the aligned bottom subs and middle subs, then the bottom four subs and top subs which is the final alignment. After this I entered all the delays and measured the result, which matches almost PERFECTLY with the prediction! This is always amazing, but always true. REW hasn't failed yet. The measured result is in purple and can be seen intertwining with the predicted in White.
 

Total time involved from measuring each subwoofer to the final alignment took 11 minutes. Now that each sub has been measured you don't need to measure again unless something on the sub has changed. You can use these measurements and play with aligning in various progressions to get various results. This is an especially important benefit when more sub locations are involved and an increased amount of outcomes are possible.

The next step is to see if Dirac likes this result and can work with it, which Dirac usually does very well. I'll post the Dirac plot soon.

[ttocs]

Should there be more locations for subwoofers, my progression would be to do what I've done with the front subs first, then add the next subwoofer location to the group.

For example, I've been running three sub locations up to now. Two in front, one in back. So I would use the back subwoofer location to aid the front subs and smooth the response, hopefully.

[Germotiva]

Jan 8, 2023 1:18:51 GMT -5 ttocs said:

Should there be more locations for subwoofers, my progression would be to do what I've done with the front subs first, then add the next subwoofer location to the group.

For example, I've been running three sub locations up to now. Two in front, one in back. So I would use the back subwoofer location to aid the front subs and smooth the response, hopefully.

Hi ttocs,
Happy new year from Germany 🎉
Here is my new Multisub lineup. I would also like to try this with REw. Active sub absorbers are used in the right.
🙋‍♂️Paule

[ttocs]

Jan 8, 2023 17:17:44 GMT -5 Germotiva said:

Jan 8, 2023 1:18:51 GMT -5 ttocs said:

Should there be more locations for subwoofers, my progression would be to do what I've done with the front subs first, then add the next subwoofer location to the group.

For example, I've been running three sub locations up to now. Two in front, one in back. So I would use the back subwoofer location to aid the front subs and smooth the response, hopefully.

Hi ttocs,
Happy new year from Germany 🎉
Here is my new Multisub lineup. I would also like to try this with REw. Active sub absorbers are used in the right.
🙋‍♂️Paule
View Attachment

Happy New Year Paule!

Since REW can only align two measurements at a time, some thought about how to group the subwoofers is needed. In my example it's just two subwoofer locations but six subs. I found it to be more successful to align each "level" of both stacks, so, Level 1 left and right, Level 2 left and right, then Level 3 left and right. I then decided it would be best to start at the bottom and work up from there for combining the three aligned pairs. So it was an easy choice to align Level 1 and Level 2 and then to the aligned Levels 1+2 result with Level 3.

I see many subwoofers in your room layout. Divide and conquer by aligning logical pairs of subs. Front pair, Rear pair, Left pair, Right pair. Then Front and Rear is what I would align next. But, I'd try multiple ways of combining Left and Right into the final mix. Could be to align Left and Right together, then align the final two groups together - Front/Rear and Left/Right. Or, align Front/Rear and Left, then, Front/Rear/Left and Right.

There are many choices for which sub should be aligned with which. Heck, you could align the Front Left with the Rear Left, then the same with Front Right and Rear Right, and then add the left and right subs into the mix. It really isn't easy to know which will give the best result until it's actually tried.

I'll use your layout to show everyone just a few scenarios of sequence for the initial alignment to reduce the individual subwoofers into four initial groups. The #1 example is one that I believe would be the most problematic of the four. Aligning each pair of numbered subs together for the initial alignment, is probably the easy first step, but then, while there are still multiple possible sequences, I'd start with the most logical of Front + Back, Left + Right, then the last one which is Front/Back + Left/Right. Lots of possibilities to have fun with!


But the premise is still the same. Simply measure every subwoofer individually. After that, there is nothing else to measure until after coming up with alignment scenarios, loading the delays into the miniDSP(s), and then measuring to confirm. But make duplicates of the REW file so you can try many scenarios of aligning all the subs. It really doesn't take long to come up with a sequence. Try a bunch of scenarios, then compare by using Overlays in REW.

You can see that in my exercise with six subs, I ended up with six measurements, three initial alignments, then two more alignments, and one last measurement. But I tried other alignment sequences before coming up with the example I posted which happens to be the best one - so far. There is a lot of judgement involved in deciding which curve shape to accept before moving on to the next one because there are variables in play throughout the process. REW makes it fairly effortless to try many combinations and comparing immediately.

[Germotiva]

Thanks for the execution and tips.
Yes, there are many ways to get to the destination.
I also will make many saves in the dsp.
To find the best with the ear.
I get the umik-x 4(16) from a deal as an acoustic consultant in a disco.
I'm happy about it. I'll wait until I get it. I will report.

[Germotiva]

Jan 8, 2023 19:52:39 GMT -5 ttocs said:

Jan 8, 2023 17:17:44 GMT -5 Germotiva said:

Hi ttocs,
Happy new year from Germany 🎉
Here is my new Multisub lineup. I would also like to try this with REw. Active sub absorbers are used in the right.
🙋‍♂️Paule
View Attachment

Happy New Year 

Hi ttocs, The next attempt is the 2 subs in the niche as an active absorber to improve my WTF with REW.
The 30Hz must be attenuated. But before that, the front is still with rockwool 246 inches deep.
On the back wall is already 246 inches Rockwool,to. Not in the niche..


Do you  have a wtf of your room
🙋‍♂️Paule

[ttocs]

Jan 10, 2023 4:20:38 GMT -5 Germotiva said:

Jan 8, 2023 19:52:39 GMT -5 ttocs said:

Happy New Year 

Hi ttocs, The next attempt is the 2 subs in the niche as an active absorber to improve my WTF with REW.
The 30Hz must be attenuated. But before that, the front is still with rockwool 246 inches deep.
On the back wall is already 246 inches Rockwool,to. Not in the niche..
View Attachment

Do you  have a wtf of your room
🙋‍♂️Paule

I hope you mean to say 246mm instead of inches, because that would be a really quiet room with 20 feet of rockwool at each end of the room! 

My room is not treated much at all. To my right is a wall of glass windows, floor to roof, cathedral ceiling. But I would like to pad the back wall, which might happen this coming weekend. But, here's a waterfall from recently, but still after I began moving the subs around and trying different approaches with timing, which will be a work-in-progress for another couple weeks or so. Below 50dB there's really nothing I can do to squelch the stuff under 30Hz which is mostly caused by a highway about 1km away. And some of the ringing is from appliances I don't turn off when testing, I only turn off everything when I get serious about things.


I'm in the process of comparing some timing variations with the Alignment Tool. I'm trying to find answers to questions I haven't been able to discover on the www, so I'm left with my experiments. What I'm seeing so far is that when I align 6 subs, in only 2 locations, "how" the alignment sequence is performed, and the goal of the alignment, make a difference for how Frequency Response, Impulse, and Phase look. Aligning for best Impulse doesn't make for the best Freq. Response or Phase curve. But here's the kicker, Dirac behaves a little counterintuitively as well. Best Impulse method can look lousy after Dirac gets done with it. So some "fudging" seems to help, in other words, best impulse may not be best overall when using limited subwoofer locations because it's difficult to get good results at the crossover. But it's still early yet, so maybe next week I'll be able to post about it. I think Natural1 is on to something with adjusting the Distance, and it would be helpful if we could have an "Expert Mode" in our processors with more settings for such things.

[Germotiva]

Jan 10, 2023 11:33:16 GMT -5 ttocs said:

Jan 10, 2023 4:20:38 GMT -5 Germotiva said:

Hi ttocs, The next attempt is the 2 subs in the niche as an active absorber to improve my WTF with REW.
The 30Hz must be attenuated. But before that, the front is still with rockwool 246 inches deep.
On the back wall is already 246 inches Rockwool,to. Not in the niche..
View Attachment

Do you  have a wtf of your room
🙋‍♂️Paule

I hope you mean to say 246mm instead of inches, because that would be a really quiet room with 20 feet of rockwool at each end of the room! 

My room is not treated much at all. To my right is a wall of glass windows, floor to roof, cathedral ceiling. But I would like to pad the back wall, which might happen this coming weekend. But, here's a waterfall from recently, but still after I began moving the subs around and trying different approaches with timing, which will be a work-in-progress for another couple weeks or so. Below 50dB there's really nothing I can do to squelch the stuff under 30Hz which is mostly caused by a highway about 1km away. And some of the ringing is from appliances I don't turn off when testing, I only turn off everything when I get serious about things.
View Attachment

I'm in the process of comparing some timing variations with the Alignment Tool. I'm trying to find answers to questions I haven't been able to discover on the www, so I'm left with my experiments. What I'm seeing so far is that when I align 6 subs, in only 2 locations, "how" the alignment sequence is performed, and the goal of the alignment, make a difference for how Frequency Response, Impulse, and Phase look. Aligning for best Impulse doesn't make for the best Freq. Response or Phase curve. But here's the kicker, Dirac behaves a little counterintuitively as well. Best Impulse method can look lousy after Dirac gets done with it. So some "fudging" seems to help, in other words, best impulse may not be best overall when using limited subwoofer locations because it's difficult to get good results at the crossover. But it's still early yet, so maybe next week I'll be able to post about it. I think Natural1 is on to something with adjusting the Distance, and it would be helpful if we could have an "Expert Mode" in our processors with more settings for such things.

😂😂🙈It's really 625mm per back wall. This strength is gladly taken here in Germany on a back wall for an SBA system. If a DBA has not been installed. I'm more of an outsider with the Multisub. Your WTF looks good without treatment. Will currently discuss which level and settings can be used as a standard reference to better compare with each other. 100db level with -60dB drop at 300ms in scale.

[Natural1]

Jan 10, 2023 11:33:16 GMT -5 ttocs said:

Jan 10, 2023 4:20:38 GMT -5 Germotiva said:

Hi ttocs, The next attempt is the 2 subs in the niche as an active absorber to improve my WTF with REW.
The 30Hz must be attenuated. But before that, the front is still with rockwool 246 inches deep.
On the back wall is already 246 inches Rockwool,to. Not in the niche..
View Attachment

Do you  have a wtf of your room
🙋‍♂️Paule

I hope you mean to say 246mm instead of inches, because that would be a really quiet room with 20 feet of rockwool at each end of the room! 

My room is not treated much at all. To my right is a wall of glass windows, floor to roof, cathedral ceiling. But I would like to pad the back wall, which might happen this coming weekend. But, here's a waterfall from recently, but still after I began moving the subs around and trying different approaches with timing, which will be a work-in-progress for another couple weeks or so. Below 50dB there's really nothing I can do to squelch the stuff under 30Hz which is mostly caused by a highway about 1km away. And some of the ringing is from appliances I don't turn off when testing, I only turn off everything when I get serious about things.
View Attachment

I'm in the process of comparing some timing variations with the Alignment Tool. I'm trying to find answers to questions I haven't been able to discover on the www, so I'm left with my experiments. What I'm seeing so far is that when I align 6 subs, in only 2 locations, "how" the alignment sequence is performed, and the goal of the alignment, make a difference for how Frequency Response, Impulse, and Phase look. Aligning for best Impulse doesn't make for the best Freq. Response or Phase curve. But here's the kicker, Dirac behaves a little counterintuitively as well. Best Impulse method can look lousy after Dirac gets done with it. So some "fudging" seems to help, in other words, best impulse may not be best overall when using limited subwoofer locations because it's difficult to get good results at the crossover. But it's still early yet, so maybe next week I'll be able to post about it. I think Natural1 is on to something with adjusting the Distance, and it would be helpful if we could have an "Expert Mode" in our processors with more settings for such things.

Yep...If you are using Dirac and a minidsp for multi-sub optimization, the one thing you can't correct/tweak for is the integration with mains, unless of course Dirac somehow manages to set delays such that the subs lead the mains by too much time. Then you can adjust with a global positive delay in your minidsp.

I like what Dirac does in general, and I wanted to keep using it, but Dirac does 2 things consistently "wrong" in my particular setup...

1. Post-Dirac, the subs are always out of phase with the mains. I correct this with a physical XLR phase inverter on the Center Sub output. I could do this with minidsp, but my alignment between the subs is very good without it, and Dirac does a very good job of EQ-ing the 2 subs as one (Center sub mono). I'd rather not introduce another piece of DSP in the system if I can avoid it.

2. The integration with the mains is always slightly out of time, with the subs firing "too late" causing minor issues in the crossover region.

Obviously you can't fix #2 with a minidsp or similar because you can't apply a negative delay to the subs. The only way to do this is to delay the other speakers. You can't do this in the G3P menu, but you can do by carefully modifying the Dirac settings export.

From what I've read and watched, and from a bit of playing around myself, I agree that trying to integrate the subs based strictly on impulse response doesn't work very good. In my testing, just using the alignment tool for phase alignment (subs combined with LR Vector average at MLP) produces the best overall integration for me.

I used the LR Vector average at MLP because I want to prioritize stereo music at the MLP, and personally I'm not too concerned if the center or surrounds aren't crossed over "ideally" to the subs. As we all know, everything is a compromise anyway. The good news is that in my particular situation, using that LR Vector average for integration also produces very good results for the center and surrounds at the MLP. Maybe I got lucky, or maybe it's the fact that I also have all speakers tightly time aligned at the MLP? I'm not an expert, so I really don't know.

At the end of the day, right now I've got a good overall Dirac calibration that takes a wider area into account, combined with sub crossover integration that's L/R prioritized at the MLP. For me that's a good overall compromise and I've yet to find a real particular weakness in this approach, at least for my purposes.

[ttocs]

An Alignment Procedure. Just one of many ways to align, all with various results, good, bad, indifferent. This one happens to consistently get decent results.

The first step in any of these alignment procedures is to measure each subwoofer, one by one. Once these measurements have been made, and if the subs don't move, and if the MLP stays in the same spot, then there is no reason to do this part again. These measurements can be used again and again to try various alignment sequences for a variety of end results. In this case, I am using 3 locations for 6 subs, three stacks of 2 subs each. Two stacks in Front, one stack in back.

You will see in the graphics Group 1, Group 2, and Group 3. These are the groupings of two "Aligned Sum" plots as made by REW and/or Measurements of individual subs.
Group 1 is the Front Left Bottom sub and Front Right Bottom sub - L1+R1 for one alignment, and Front Left Top sub and Front Right Top sub - L2+R2 for the second alignment.
Group 2 is the Aligned Sum of each of the first two alignments PLUS+ the corresponding Back Bottom or Back Top sub, B1 or B2 respectively.
Group 3 is the Aligned Sum of the Bottom subs PLUS+ the Aligned Sum of the Top subs - LRB1+LRB2.

The sequence is: Left front bottom + Right front bottom = Aligned Sum. Then align this Aligned Sum with Back bottom. This will combine all three Bottom subs.
Repeat the above for the Top Subs. Now there are 2 Aligned Sums that will be combined for the final Aligned Sum.

For this exercise I used the exact delay numbers REW recommended for Group 1 and Group 2, but I "fudged" the alignment by using the slider for delay when aligning Bottom and Top Aligned Sums to get the response to be smoother and more to my liking. Dirac agreed with my result.

I made a spreadsheet to help gather the delays and add them for me. It also helps to graphically show the sequence and which numbers go where. I've got the number 1 subs below the number 2 subs so it makes it easier to comprehend, I hope. I also have screenshots of the REW tabs for each Aligned Sum which shows which item, A or B, is to get a delay or be inverted. All I had to do was to enter the delay numbers into the corresponding cell on the spreadsheet for each group in succession. After posting this I'll probably see easier ways to explain things and make edits as needed.

This first image is the floor plan showing where the subwoofer stacks are located and what they are named.


This is the spreadsheet showing all the aligned sums and the resulting delay number for each subwoofer. Keep in mind that for the final alignment, that one number needs to be entered for each of the three subs in the same level, in this case Level 2, or LRB2, the top subwoofers. The delay number for Groups 1 and 2 don't get entered into the miniDSP, but the final Group 3 numbers in the black column do get entered for each subwoofer in all the miniDSP units. So I input delays for 5 of the 6 subwoofers in the Output section of each miniDSP.


These next two plots are of the actual measured result, exactly as REW predicted. The first one is a before Dirac in Green, and after Dirac in Blue. The output of the XMC-2 didn't change, so this shows how Dirac arrived at its calibration and what is squashed or boosted. The second is a waterfall plot at a higher volume level. As noted in an earlier post my room is not fully treated.

(Please note: The image above has been corrected. The actual measured trace in this image correctly matches the 80Hz crossover setting used for the Aligned Sum as predicted in REW. The amps for speakers are turned off for testing subwoofers so they can more easily be used for this purpose, so, I normally use a 200Hz crossover setting so it will extend the subwoofer range farther before rolling off the frequency response. But when I was engaged in this exercise I mistakenly didn't change the crossover setting for SBR (Surround Back Right) which was set at 80Hz, so that's why the rolloff is what it is. Neither crossover setting invalidates anything in this exercise, but the previously posted comparison image was erroneous because the Aligned Sum is with 80Hz XO and the Dirac filtered actual measurement was at 200Hz, hence the correction. It was only a crossover setting mismatch.) 

[Natural1]

Very cool ttocs, and nice results!

I'm curious, did you ever try it summing/aligning the subs in a more positional arrangement, like... ((L1+L2) + (R1+R2)) + (B1+B2)

I only have 2 subs, but I wonder if it really makes a difference how you order the alignments. It seems like it might, but I really have no idea.

[ttocs]

Jan 11, 2023 16:02:33 GMT -5 Natural1 said:

Very cool ttocs, and nice results!

I'm curious, did you ever try it summing/aligning the subs in a more positional arrangement, like... ((L1+L2) + (R1+R2)) + (B1+B2)

I only have 2 subs, but I wonder if it really makes a difference how you order the alignments. It seems like it might, but I really have no idea.

It makes a big difference as to the order and combination of locations. I've tried 2-stacks and 3-stacks where I align each stack to itself - as in your groupings, then align the front two stacks together, then add another single sub or stack of subs in the back of the room, or in one scenario, just behind the sofa. The results vary quite a bit.

I have not yet tried L1+B1, and then add L1B1+R1, then do the same with Level-2, and then sum both levels together. But I want to try this and the reverse of it to see if I can tailor the left subs' weaknesses a bit better. There's also Inverting that can be added in the mix - which I try as a matter of course with every alignment step, as well as some PEQ within miniDSP. Also, one can run Dirac, get results that make it difficult to get good crossover blending, and then try inverting the phase to see if it works better, which I plan to do soon with this latest setup.

So there seems to be quite a few options during aligning, after aligning, and even after creating Dirac filters - as you know! Lots of ways to skin cats, if that's even a thing anymore.

Right now I'm just wanting to document some REW ideas to share what I know of the app and to gather more info from others about other features I have yet to know how to use. In the process I've been learning about things I haven't been able to read about elsewhere. A few years ago when I first began using REW I scoured the Triple-W trying to even just get started, and failed for a long time until I learned that I could just use my Mac connected directly to the processor via HDMI. Duh, who knew? I was using an audio interface and setting the controls manually for each REW session, which was tedious to say the least.

[Natural1]

Jan 11, 2023 16:22:33 GMT -5 ttocs said:

Jan 11, 2023 16:02:33 GMT -5 Natural1 said:

Very cool ttocs, and nice results!

I'm curious, did you ever try it summing/aligning the subs in a more positional arrangement, like... ((L1+L2) + (R1+R2)) + (B1+B2)

I only have 2 subs, but I wonder if it really makes a difference how you order the alignments. It seems like it might, but I really have no idea.

It makes a big difference as to the order and combination of locations. I've tried 2-stacks and 3-stacks where I align each stack to itself - as in your groupings, then align the front two stacks together, then add another single sub or stack of subs in the back of the room, or in one scenario, just behind the sofa. The results vary quite a bit.

I have not yet tried L1+B1, and then add L1B1+R1, then do the same with Level-2, and then sum both levels together. But I want to try this and the reverse of it to see if I can tailor the left subs' weaknesses a bit better. There's also Inverting that can be added in the mix - which I try as a matter of course with every alignment step, as well as some PEQ within miniDSP. Also, one can run Dirac, get results that make it difficult to get good crossover blending, and then try inverting the phase to see if it works better, which I plan to do soon with this latest setup.

So there seems to be quite a few options during aligning, after aligning, and even after creating Dirac filters - as you know! Lots of ways to skin cats, if that's even a thing anymore.

Right now I'm just wanting to document some REW ideas to share what I know of the app and to gather more info from others about other features I have yet to know how to use. In the process I've been learning about things I haven't been able to read about elsewhere. A few years ago when I first began using REW I scoured the Triple-W trying to even just get started, and failed for a long time until I learned that I could just use my Mac connected directly to the processor via HDMI. Duh, who knew? I was using an audio interface and setting the controls manually for each REW session, which was tedious to say the least.

Thanks for the info!

It's interesting... I think I've seen/read where folks say the order (probably) doesn't really matter. It certainly makes more sense that it would make a difference, like you've determined.

I'm guessing someone will eventually figure out a somewhat reliable formula for determining a good order for this kind of alignment. Maybe based on the relative phase differentials at MLP for each individual measurement, or something crazy like that. With 6 subs, if my brain is working right, I think you have like 360 different potential combinations (6 factoral / 2)?!?! Even if I'm wrong, it's still a lot!

I think I'm glad I only have 2 subs

[ttocs]

Jan 11, 2023 16:46:17 GMT -5 Natural1 said:

It's interesting... I think I've seen/read where folks say the order (probably) doesn't really matter. It certainly makes more sense that it would make a difference, like you've determined.

I'm guessing someone will eventually figure out a somewhat reliable formula for determining a good order for this kind of alignment.

The order in which multiple subs in more than one location and more than one elevation (stacked) are aligned does indeed matter. In my exercise above I chose to keep myself from having any input into the alignment calculation. I simply kept the A and B settings at zero until the final alignment, which is when I fudge the sliders a bit to smooth things as much as is possible without going too far with delays. So, with a "horizontal" alignment, combining the 3 subs at floor level, then combining the 3 subs stacked on top, then aligning top and bottom groups and fudging the slider for one of them a little has had pretty consistent results.

As to the magic formula (I added the word magic, because REW is already pretty magical!), (edit) using MSO (Multi Sub Optimizer) it's possible to toss all the measurements into a bin and have the app spit out a near perfect scenario. But keep in mind that there is some amount of finesse' involved with this stuff. I choose to make the traces have the fewest amount of JaGgEd edges, or fewer "troughs" (nulls), or not so many lumpy looking huge bumps, etc. I don't mind a few narrow width deep spike type nulls, as these seem to be either smoothed out by Dirac, or simply not noticeable psychoacoustically.

Jan 11, 2023 16:46:17 GMT -5 Natural1 said:

I'm curious, did you ever try it summing/aligning the subs in a more positional arrangement, like... ((L1+L2) + (R1+R2)) + (B1+B2)

Here is that result. I aligned the 3 individual stacks first. Then aligned L12+R12, then aligned LR12+B12. The results are below. The "vertically" aligned sum of the 3 stacks doesn't look as neat as the horizontally-then-vertically aligned groups from the previous post.

Green is from the previous post (LRB1+LRB2), and Red is newly aligned result (L12+R12+B12). It's lacking a little bit of overall output, not much, but that "trough" from 60-80Hz is something I'd avoid working with because it's difficult to work with most of the time.


This is with the respective Dirac filters applied. Both look good, but the new vertical aligned Red trace has 1-2dB less output below 70Hz and is not as smooth. This is nit picking, I know that, but there are differences. From 80-130Hz the Red trace is more "wobbly" (I hope everyone can understand this technical jargon), again, not quite as smooth. But I will say that either one would work great and nobody would be able to tell the difference. There's really only differences that can be picked on with measurements like this, so . . . .


So at the end of the day, I'd pick the one with a touch more output from 30-40Hz.

Now with all that said, aligning using this vertical method is quicker because there aren't any delays to be entered until the final alignment, at least when just working with 2 stacked subs. With 3 stacked subs there are delays that were entered for at least one sub in each stack when aligning (L1+L2, then L12+L3).

I'll post a waterfall comparison later. The one I just ran doesn't look near as good for this new alignment, but might be environmental, from outside noise. So waiting for the world to quiet down will be better for waterfall plots.

[Natural1]

Good info, thanks!

In this video from Jeff Mery there's some really interesting info about the spectrogram and group delays. My current calibration is very good in terms of frequency response, but not quite "there" in terms of group delay and decay time targets being discussed. I need to dig into that aspect of it sometime soon.

Maybe not to others, but this particular side of things is very new to me in terms of what this means to the listening experience and how you might get there. Again, it's really all about compromise.

www.youtube.com/watch?v=bEEOzhFk-kk

[ttocs]

Jan 11, 2023 21:29:03 GMT -5 Natural1 said:

Good info, thanks!

In this video from Jeff Mery there's some really interesting info about the spectrogram and group delays. My current calibration is very good in terms of frequency response, but not quite "there" in terms of group delay and decay time targets being discussed. I need to dig into that aspect of it sometime soon.

Maybe not to others, but this particular side of things is very new to me in terms of what this means to the listening experience and how you might get there. Again, it's really all about compromise.

www.youtube.com/watch?v=bEEOzhFk-kk

Good video by a very knowledgeable guy. Now I have a better idea of how to read GD and Spectrogram. I'll post the link on the first page along with one of Jeff's other videos.

I've never checked out MSO, but it's that magic app you were asking about before. I just watched another of Jeff's videos all about MSO. I'm just not sure I will be able to use it with my multiple miniDSP's going to 6 subs. I bought them over time, but if I had to do it over I'd get one of the models with more outputs. Oh well.

[Natural1]

Jan 11, 2023 23:33:29 GMT -5 ttocs said:

Jan 11, 2023 21:29:03 GMT -5 Natural1 said:

Good info, thanks!

In this video from Jeff Mery there's some really interesting info about the spectrogram and group delays. My current calibration is very good in terms of frequency response, but not quite "there" in terms of group delay and decay time targets being discussed. I need to dig into that aspect of it sometime soon.

Maybe not to others, but this particular side of things is very new to me in terms of what this means to the listening experience and how you might get there. Again, it's really all about compromise.

www.youtube.com/watch?v=bEEOzhFk-kk

Good video by a very knowledgeable guy. Now I have a better idea of how to read GD and Spectrogram. I'll post the link on the first page along with one of Jeff's other videos.

I've never checked out MSO, but it's that magic app you were asking about before. I just watched another of Jeff's videos all about MSO. I'm just not sure I will be able to use it with my multiple miniDSP's going to 6 subs. I bought them over time, but if I had to do it over I'd get one of the models with more outputs. Oh well.

Ahh I completely forgot about MSO in those terms, because I haven't used it either. I do remember watching one of his videos on using MSO a while back and it does a LOT of iterative calculations. Maybe more brute force but apparently it does a very good job. Very cool for those with more than 2 subs for sure.

[ttocs]

Jan 10, 2023 12:09:54 GMT -5 Germotiva said:

It's really 625mm per back wall. This strength is gladly taken here in Germany on a back wall for an SBA system. If a DBA has not been installed. I'm more of an outsider with the Multisub. Your WTF looks good without treatment. Will currently discuss which level and settings can be used as a standard reference to better compare with each other. 100db level with -60dB drop at 300ms in scale.

That is a very thick amount of rockwool !!   That's about 2 feet thick, so now I understand what you meant when you were using 245 inches, which was meant to be 24.5 inches.

Regarding REW Waterfall settings, I will use the default settings for the most part. I did this yesterday and think things show up well. Next one I post I will list the settings I change from default. Also, I use 45-100dB because my noise floor is just under 45dB a lot. Also, I'm learning about Spectrogram and what to look for, which is very informative.

[Germotiva]

Jan 12, 2023 0:26:18 GMT -5 ttocs said:

Jan 10, 2023 12:09:54 GMT -5 Germotiva said:

It's really 625mm per back wall. This strength is gladly taken here in Germany on a back wall for an SBA system. If a DBA has not been installed. I'm more of an outsider with the Multisub. Your WTF looks good without treatment. Will currently discuss which level and settings can be used as a standard reference to better compare with each other. 100db level with -60dB drop at 300ms in scale.

That is a very thick amount of rockwool !!   That's about 2 feet thick, so now I understand what you meant when you were using 245 inches, which was meant to be 24.5 inches.

Regarding REW Waterfall settings, I will use the default settings for the most part. I did this yesterday and think things show up well. Next one I post I will list the settings I change from default. Also, I use 45-100dB because my noise floor is just under 45dB a lot. Also, I'm learning about Spectrogram and what to look for, which is very informative.

Hi, yes is a lot of mass. But to tame the modal wave in space, there is no way around it. Unless you use air spaces. I look and calculate with the absorber calculator (Multi Layer). This reduces the mass but not much the depth. Why I use 2 sides of the damping. Firstly to tame the multisubs and secondly to tame the mains and surrounds additionally. I deal in the home theater with the additional location bass. I work with the bassmanagment in the dsp. I put close to the surrounds the subs. These work with LFE in the group. In addition, the individual sub gets the signal info of the locations. I mix both like you in the mini dsp. Only individually refined. This gives a great atmosphere in the location.
Yes I also dealt with the spectogram 2-3 years ago when I built Limp absorbers. From this you can see the WTF from above. I always looked at the holes / islands from above for the test. Since from the front these were not visible and were hidden. But as I see I still have to learn much much more to be able to understand some better, because this hides a lot of information.
Here are some old measurements. Unfortunately, I did not notice the scale on the right which is set incorrectly in the standard. Also here should be scaled-60db.