mtg90â€™s multiple subwoofers + mains integration How To thread
I know there are a few of these threads floating around or posts buried in other threads but my hope for this is to become a useful reference being more detailed on certain aspects of it.
Note this thread does not deal with the act of finding the best locations for subwoofers in your room just dialing things inonce the subwoofers are in position.
Note this thread is a work in progress and will beupdated in the near future with more info/graphs.
I have used my room as an example (pictures and graphs) for the type of setup detailed here. This can be found here on post #2 .
Measurement microphone and software (Omnimic, miniDSP UMIK-1 + REWâ€¦)
DSP with enough channels for each subwoofer orgroup of subwoofers. This can be a MiniDSP, internal amplifier DSP (Behringer Inuke, Crown CDiâ€¦) or other external DSP.
AVR/processor with adjustable speaker distance settings (nearly all decent models from the past 10+ years have this feature).
Before we dive in I should warn you that this can be a tedious process, do not expect to start from scratch and speed though thisin 15 minuets. You may find yourself taking hundreds of measurements depending on how thorough you go with it. The results however can be well worth it, you can expect improved dynamics and definition and in some cases 6-10+dB more output capability over sections or all of the subwoofer range (that can be like doubling the number of subwoofers and amplifier power at no cost!).
Now to begin, Iâ€™ll break things down into steps for easier reading.
Step #1 Current setup baseline subs + mains.
You want to measure your current setup before making any changes so that you can use that to compare and make sure you are improving things. You only need to do this step if you want to compare your old settings to you new ones. If you have recently added new subwoofers and have yet to rerun your AVR/processorâ€™s room correction or you are just starting out with a new setup and have not yet run it for the first time it may make sense to wait until after the subwoofer integration steps as you will likely have to rerun that correction after the fact. If you donâ€™t mind running the room correction setup twice just to compare then by all means go right ahead.
If you are only concerned with the response at the main listening position that you only need to take measurements at that location. However it is best to try and optimize the response across all of the seating the best you can, therefore I will focus on that.
For most of these measurements it is best to send a two channel stereo signal into the AVR/processor, if you have your speakers set to large and do not run the subwoofer with stereo sources you may want to turn on the + setting for the subwoofer or similar option that sends the low frequencies to both speakers and subwoofer. The range we are concerned with is <200Hz so it may make sense to sweep below that frequency and adjust your softwareâ€™s graph axis to focus in on that range.
Depending on how thorough you want to go and how many seats you have you can measure at every seat. Or 2-3 positions per row (if you have multiple rows). At the very minimum take at least one measurement per row. It is useful to have one measurement centered between all the seating as well.
Once you have that baseline set of measurements saved you will want to start fresh.
Step #2 Baseline measurements for subwoofer integration.
First go into your processorâ€™s settings and change the subwoofer xover to the highest setting. This will allow a clear view over the subwoofers entire operating range without the rolloff from the subwoofer lowpass causing unneeded attenuation of the higher frequencies. Leave this at that setting until we go to combine the subwoofers with the main speakers. If using external amps turn off those for the main speakers or simply unplug them from the AVR.
Clear all settings in your DSP/s and adjust the amps so that the signal gain (power to each subwoofer) is matched. You want to measure the native response ofeach subwoofer individually at each location for which you choose to take measurements. You can usually get away with 1-2 measurements per row + the central location here but make sure you continue to use those same mic positions forthe other integration steps. Donâ€™t worry about getting the mic back exactly in the same spot an inch or three either way will not make too much of a difference at the bass frequencies. These measurements will be used as a reference when we start to add them together in order to make sure they combine constructively.
Now if any subwoofers are roughly at an equal distance to each seat for example multiple subwoofers at the front wall, these can be treated as a group since their outputs should already constructively combine. Thus they can be treated as asingle subwoofer and placed on a single DSP channel to be EQâ€™ed and delayed together.
Step #3 Combined response of two subwoofers.
Measure the combined responseof two of your subwoofers, if you have more then two subwoofer locations make sure one of those measured here is the subwoofer furthest from the LP and come back to this step after step #5 for each additional subwoofer location.
Step #4 Add delay
Ideally the response of the two subwoofers would constructively combine and you will see a net gain(roughly 6dB for two identical subwoofers) across the entire range without any dips caused by destructive interference. However this is not often the case and requires adjustment of delay to time align the individual subwoofers at the LP. Delay is likely required and should be added to any subwoofer that is closer then the furthest one from the LP.
To start physically measure the difference in length between the furthest subwoofer to the central measurement location and from the other subwoofer you have chosen to measure with it. Now add delay to the closer subwoofer equal to the difference indistance between the two. That will often get you close, now measure the combined response with the delay added. Adjust the delay +- a little (0.5ms or so) and re-measure to see if it can be improved upon. You want to try and get constructive summation of outputs at least up to your eventual crossover point for the mains(usually 80 Hz). Boundary interactions can cause issues here so you may have to settle for less then perfect summation, just choose something that looks best over the largest potion of the range.
One limitation of the 2x4 miniDSP here is its maximum delay, only 7.5ms. You can however squeeze a little more in if you use a low pass above the eventual crossover point. A higher order lowpass will result in more delay (or phase shift).
Step #5 Delay continued. (Skip if you have only one row of seating or care only about the response at the main LP.
If you have multiple rows you want good integration in each row, not only the central measurement location. Measure the combined and delay adjusted response from step #4 at the other rows of seating. You want to adjust the delay to achieve a decent result across multiple rows, this can be hard to achieve and is often impossible to get perfect. The amount of delay for ideal integration in one row will often leave a notch in another. Your goal is to find a balance in the delay settings that minimizes these notches.
Step #6 Additional Subwoofers (Skip if you have only two subwoofers)
Turn on another subwoofer and go back tostep #3 in order to integrate this subwoofer with the others. Treat the already integrated subwoofers as a group leaving them onbut continue to use the furthest sub as the physical reference point for adding delay. Once you have integrated all the subwoofers you can move on.
Step #7 Manual EQ of Subs (If using room correction with auto subwoofer EQ this step is optional)
Now that your subwoofers are time aligned and working constructively you can EQ them together and as one applying the same EQ to all which gives you equal power distribution, no sub will bottom or clip before another if using identical subwoofers. You could also EQ each flat individually if using non identical subs but that sometimes requires additional EQ work after they are combined and you will often reach the limits of one subwoofer before the others. However individual EQ can sometimes be beneficial if say you are getting too much output at the higher frequencies from nearfield or rear subwoofers causing localization issues when applying global EQ to all the subwoofers.
For this step if I usually measure 2-3 locations per row and average all of them applying EQ based off the averaged response.
Step #8 Integration with mains
Readjust your subwooferâ€™s xover back down to where you had it. Turn on the amps for your mains or plug them back in. Now is the time to rerun your processorâ€™s room correction if applicable.
After that is done measure the response ofthe mains with subs turned off. Then the subwoofers turned on with the mains turned off/unplugged. Finally both turned on for the combined result. Like the integration of multiple subwoofers the combined response should show constructive output through the crossover region without any dips that were not in the individual responses. If that isthe case you should be good to go.
However if they do not combine constructively you will need to play with the subwoofer delay in your processorâ€™s speaker distance settings. Unlike dialing in the subwoofers you can't often measure the physical distance offset between the subs and speakers here and expect the acoustic summation to be ideal at that distance, group delay in the low pass filter and subwoofer delays you set earlier effect the acoustic offset of the subwoofers. Your best bet is trial and error here, choose a distance, measure and adjust that distance + or - a little bit maybe 2ft and re-measure.
You will know you are moving in the right direction when the combined response has fewer peaks and dips around (below, at and above) the crossover. The same is true if you have the mains set as larger with the subwoofer in the plus (or similar) bass management setting. You may need to change it by a few feet to start to see a difference in the number of peaks and dips. When you get close the output from both should be constructive across the whole crossover range however if you see just one larger dip you may want to try flipping the phase 180 degrees on the subwoofers.
Like the subwoofer integration and if applicable you do want to check how the mains integrate with your subwoofers in your other rows (measuring subs only, speakers only and speakers+subs for each row). Adjust for the best balance across your multiple rows. Sometimes you can use the output from the main speakers to fix a dip in the subwoofer only response in one of your rows if itâ€™s near the crossover by adjusting the subwoofer distance setting to align the speaker output with that row.
You may or may not want to make some final EQ changes to the subwoofers based on the combined speaker/subwoofer response to flatten any output bump you might have created if they were not integrated correctly by the processor. Or manually dial in a house curve.
Put in a movie/music sit back and enjoy! :)
Sorry for the long read, I hope this was not too difficult to understand and will help anyone who reads it optimize their system to it's full potential.
Example of setup using my room.
The Room, 15.5' wide, 17' long open at back to rest of basement, 9' ceilings. I have four ported 15" subwoofers, two at the front wall below the screen, one even with the rear row of seating just to the right on the floor, and one just behind the rear row up on a shelf against the left wall. I should note that while the front and rear subwoofers are different they are both tuned to ~14 Hz.
Here are some pictures of the room so you can get an idea of where things are at:
Time for measurements.
I am going to skip step one because I just added the rear subs and am starting fresh. Now I know I say to bump the subwoofer highpass up to its highest setting but I forget to do that before I started taking measurements and did not want to go back and retake bunch so I left it at 80Hz. It is not necessary to do so but you don't have to deal with the early rolloff with it moved up.
Choosing a point in the center of the seating I start with some baseline measurements of each subwoofer individually:
Now being that the front two subwoofers are symmetrically placed they should already sum constructively:
While that is the case for the most part you can see an issue starting around 75hz, that is interaction with the TV on the left side of the front wall. It causes a sharp phase shift in the response of the left sub which cannot be easily taken out by conventional means. Stuff like that is just one of the issues that cannot be easily resolved (besides getting rid of the TV) and is best left alone. I tried fixing that but did more harm then good.
Now what happens when the rear subs are added to the front subs without any delay:
Uh Oh, that does not look good. It's constructive at parts and destructive in others some down more then 10dB. This is what happens when subs are added together without setting proper delays to align their outputs at the LP.
Now lets start adding delay. Since the front subs already work together for the most part I can treat them as one subwoofer, they are also the furthest away from the LP so they will be the reference distance. I picked the rear left sub as the second one that I will try and integrate with them. I measured the difference in distance between the LP and each sub and found the left rear sub was about 7.5' closer then the front subs so I will add 7ms of delay to the output driving that subwoofer. Here is what the responses look like without then with 7ms of delay added:
You can see how it went from constructive/destructive to all constructive with the delay added in. That is what this thread is all about right there!
Now for the same with the rear right sub treating the three that have already been integrated together as one. Again the RR sub was about 7.5' closer to the LP so it gets 7ms of delay:
You can see this time it went from mostly destructive to constructive!
Now that things are decently combined at the central measurement location I will measure how the front subs and rear subs combine at both the front and rear rows. This will tell me if I need to change the delay on the rear subs to shift the summation point forwards (less delay) or backwards (more delay) in order to better match the response at both rows. First the front row:
Things look good up until 70hz where some destructive interference begins.
Lets take a look at the back row:
The back looks better, good up to just about 90hz.
I could probably move the summation point forward a little to get things closer at each row but first lets compare the front and rear rows:
Now I took a bunch of measurements at different delay settings and settled at 5ms delay on the rear subs, here is how the front and rear rows look with that delay setting:
Now the front and rear rows are within +-2dB from each other, quite good is I say so myself.
Now global EQ can be applied to flatten things out and the bass will be very consistent between the rows which is what I am after. I won't cover the EQ phase here that should be a fairly straightforward process, REW even has the auto EQ function which is very helpful to get things started. I turned up the crossover for the EQ phase so I could get the response flat to about 120 Hz allowing which then provides a smooth response for the subs with the right slope when the crossover gets turned back down.
Before I dive into the Speaker/Subwoofer integration I do want to show a couple response graphs. This is the measured response at each seat before the subwoofer integration, in other words no delay added:
Now the same measurements after I have integrated the subwoofers using delay:
Speaker/sub integration will come soon...
Setup example continued, Mains & Subwoofer integration.
EQ took place between the last post and this one, that is fairly straightforward so I am going to skip over it and just cover the integration of subs with the mains.
I took a set of reference measurements with the mic centered between the two rows. This is with an 80hz crossover in place, you can see the mains don't roll off very quick, this is a combination of being a full range design and a bit of room gain at their location below 45hz.
The goal is to achieve constructive summation across that entire crossover overlap.
To find the correct subwoofer distance setting that gives you good integration between the mains and subs you need to do a little experimenting. For this I started with a high distance setting and worked my way down in distance until I found one that provides the best summation. So here is the first try with the sub distance set at 38':
How the two sum can tell you how close you are to a proper delay (distance) setting. You can see in the graph above there is destructive interference then constructive then destructive then constructive again. This tells you that the phase angle of the subs if much different then the mains. As you adjust the delay and it moves towards a more perfect overlap you will see the amount of flipping between constructive and destructive reduce until ideally you are left with a single constructive overlap.
Knowing I probably overshot the distance setting by quite a bit I will reduce it in fairly large steps. Here is 33':
Now looking at that you might first think "well that does not look too bad" And it doesn't between 60hz and 100hz however you can still see destructive interference both below and above the crossover which means that while the phase of the two overlaps around 80hz it is not a perfect overlay across the entire range.
I'll keep going, 28':
You can see here it there is only destructive then constructive summation.
Getting closer, it has flipped to starting to be constructive on the low end and destructive on the top end. Since delay effects the phase of higher frequencies quicker then the lower frequencies I know I am very close to the correct amount of delay.
I shall now move in smaller steps as not to pass over the correct distance setting. I have also flipped the phase of the subwoofers to find the best setting by looking for the largest suck out then they will be flipped back once that is found. Here is 22':
Getting closer, almost all one way now.
That's looking pretty good, all one side means good phase overlap across that whole range. The notch is not quite as deep as it would look if I did not have smoothing turned on here.
I am going to go just a little bit more at 19':
You can see while the notch did not get deeper at 80 Hz it did get wider at the deepest point which shows that the phase overlays just a bit better at that point.
Now I will flip the phase again (the 22-19' were actually normal phase and the 19' measurement below and the 23-38' measurements are reverse phase, I took those measurements after this one but forgot to flip the phase back to normal on those. Not that it matters which way you have the phase when taking measurements as long as you configure it correctly at the end.)
And that is what the goal is, constructive summation between the mains and subs across the entire range of crossover overlap.
An interesting points I would like to make, their are actually three distance settings where I saw good phase overlap between the mains and subs, the one above 19' is the middle setting which gave the very best phase overlay at the ends of the crossover range. I also saw respectable results at delays of 12' and 26', on those the subwoofer phase is normal. At 12' there is 180 degrees more phase rotation in the crossover then here at 19' and at 26' there is 180 degrees less. One could say that the 26' setting would be more appropriate inducing less overall phase rotation in the crossover. I have yet to try and do some comparisons between the 19' and 26' settings to see if I could notice a difference. Phase rotation/shift in a crossover is one of the more difficult things to pick out, ideally a speaker/system will have a flat phase response but that can be difficult to achieve without special hardware or a lot of design work.
Here is the measurements at 12' and 26':
Going back to my 19' measurement I wanted to check to see how it measured at the front and rear rows where people will be sitting. Here is the front row:
Now to check the back row:
Here you can see the top end of the crossover is not aligned as well.
I adjusted the delay to 18.5' which will shifts the integration bubble backwards about .5':
That looks a little bit better, it did not do much to the front row so I won't post that again. I felt pretty satisfied at this point, I did play around a bit more but there is not a whole lot that can be done to make it better at the rear row. Obviously the FR above ~80hz at front and rear rows don't quite match, the level is lower at the rear row because it is further from the mains where as the subs are spread out on the room so those levels don't really change. Not much I do about that.
Anyway that concludes the integration setup for now, I hope it helps some of you get more out of your systems.
Looking forward to going through this when I've got the measuring equipment and an evening to kill. I have a feeling it will really benefit my room and setup! Thanks for putting it together.
After spending months of fiddling I'm finally really happy with my setup. It will probably last a couple of weeks and then I will try your tutorial . Only thing you might want to add is some pics (graphs showing typical poor results and what it looks like fixed). Thanks for taking time to help the community [emoji457]
If running two subs off of one inuke, can you delay them separately? Does each channel have independent settings, delays, etc?
And should you theoretically gain 6db per identical sub you add? Or is it 6db gain when you "double" the number of subs? So if you have one sub then add another, you should have 6db gain, but to gain another 6db you would have to get two more subs for a total of four? So eight subs would theoretically be an 18db gain over one sub? Or would eight subs be a 42db gain over one sub? Or am I looking at it wrong and it's much more complicated?
I'd let him finish before asking cause it'll probably get covered, but yes you can delay subs individually on each channel of an iNuke and 6db for every doubling of subs AND doubling of power.
+3 dB for doubling subs (e.g. double the displacement)
+3 dB for doubling power
Reason why he says +6 dB when combined is he assumes when you double the amount of subs you also doubled the amount of power (e.g. dedicated amp channel for each sub).
Hey.. why is the first post all in black text?
Sorry, I copied it from MS word and lots of stuff got messed up. Among other things it put two extra spaces between each line and took out a space between every 6-10 words. :confused: It took me almost a half hour to fix everything.
Example graphs and stuff will be added to the second post when I get all that together.
Really looking forward to reading this.
Busy finishing up room treatments and plan on tackling this monster thereafter.
I spent days trying to get my 3 subs setup before I realized I really needed room treatments. About 100 REW measurements later...:o
One thing I remember was during my SUB-ONLY setup. I was trying to get not only delay correct but noticed that changing polarity on
any one sub changed things drastically. In the end I landed up having the polarity on the 3rd sub inverted as that gave me a vastly improved
response in 2 of 3 measurement points. I see you suggest only changing the polarity when integrating with the mains. Reasoning?
The one critical piece of advice I would give is - Make detailed CLEAR and ACCURATE notes in the REW measurements box.
I generally do the following
Speaker : eg Sub1, Fronts only, Front and Subs
Position : eg P1, P2, P3
SPL base from REW : 75db (bass / speaker)
Xover Setting : 80 / 250
DSP : None / DSP1 / DSP2 (I have a few settings that load into my minidsp)
Do you have any recommendations on the length and sweep qty one should perform? I normally run 2 sweeps of 256k between 5hz to 250hz.
How do you run a stereo signal to the AVR from REW? I can change my HDMI output to stereo but the ASIO settings in REW still show all of the channel options. Which channels do I pick to use the stereo output?
And does this consider the center channel? I've found that when I adjusted the subs to match with the L & R the center channel ends up integrating worse with the subs. But if I match the center with the subs the L & R get messed up. Is it necessary to match up the center channel? Seems like alot of movie content comes through the center; I would almost consider that more important.
Is the text color fixed?
I cleared the formatting then bolded the steps again. I don't see much change on my monitor.
Yes it is fixed. I use the dark theme and had to highlight your text to see it, but now it is fine.
The subwoofers should all be in phase and the delay adjusted to time align them so that they sum correctly. Now sometimes if you are using different subwoofers or more likely different types of subwoofers their natural phase response differs. For most they can still be brought in line by adjusting the delay. You want to look at how the bottom end of their responses sum, if you see gain at the the lowest frequencies their polarity is correct and you adjust delay to bring the rest of the range in line because delay will effect the higher frequencies first since their wavelength are much shorter. In most cases you should not have to change polarity with the mains, only if when adjusting distance setting you get a null (cancellation between the speakers/subs) over the entire crossover range.
If your center channel is aligned with the L/R there should not be too large a difference between how the center and subs combine and how the L/R and subs combine. That is assuming you use identical or matching speakers with the same crossover points for both.
Text is fixed, thanks!
Thanks for setting all that up matt! One suggestion to add: If people are struggling with getting their subs to blend well with the mains, what I typically do is flip the polarity 180 degrees out of phase for the subs (or speakers) and adjust the distance settings until I see the maximum suckout at the crossover freq. At this point the subs are 100% out of phase, at which point you would flip the polarity back and you will have yourself a perfectly blended XO region.
Example using my room/system for steps 2-6 added to second post!
Speaker/sub integration will be added soon, I am done for tonight.
Thanks for the guide. I am currently doing exactly this routine as I just rearranged my subpositions.
I'm placing subs on all wall midpoint but not exactly symmetrical due to furniture constraints so it's not easy to get right to be honest.
Kinda of a side note to your approach buddy.
Thx Matt for posting this, Mark Seaton posted his method in another thread, later I'll post that link here
Via Mikes brain/thumb interface, LLAP
A phase adjustment shifts the entire phase response back or forth, delay (which is what the speaker distance settings do) changes the slope of the phase response allowing for a more perfect overlay between the phase response of the subwoofers and that of the speakers.
|All times are GMT -7. The time now is 01:34 AM.|
Powered by vBulletin® Copyright ©2000 - 2019, Jelsoft Enterprises Ltd.
vBulletin Security provided by vBSecurity (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
User Alert System provided by Advanced User Tagging (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.