Setting up 4 subs and phase alignment...

[Sonnie Parker]

While this thread is really about sub placement and phase alignment between subs, I hope that it can further elaborate on other setup options for subwoofers, and phase alignment between the subs and mains. Typically I have two subs up front and let Dirac do all the work, with excellent results, but now I've added two more subs. My goal is to learn more about phase as it relates to setting up subs and mains, while also doing all I can do to optimize my subwoofers and get the most out of my investment for my room, while also getting the smoothest transition from sub to mains.

I won't mention everything in my system as you can click on My System below my avatar and see the equipment. We are working with four SVS SB16-Ultras in a room that is 19.5ft wide by 23.5ft deep. The main front speakers are MartinLogan ESL Classic 9 powered by Emotiva XPA-1 Monoblocks... with the NAD M17 processing the system with Dirac Live.

I have found a home for two of the SB16's on each side of the back row of seating... and prefer to keep them there, but the fronts I am moving around to experiment where I can get the best response. I have thus far tried front corners, out from the front corners to the front edge of the front riser, various locations down the side wall to about the half way point, then out from the side wall in several locations, and currently next to the front speakers angled towards the listener and parallel with the mains. All of these positions ultimately sound very good after processing, so it's not that they are bad, but the question is can they be better in some ways. I'm not sure that my perception is effected just by seeing a cleaner response. I have not tried inside of the front speakers yet, but that seems like a good next place to check and see how they perform there.

Thus far I have kept both front subs equidistant from the main listening position... and the same with the rear subs. I keep them symmetrically placed in the front and then in the rear... so it should be the same as having one sub up front and one sub in the rear. However I am not against placing them asymmetrically in the front, although that will create the necessity of time aligning 3 sub locations.

The mains are pulled on out into the room for the best sound stage, imaging and depth acuity. Several of us here, Wayne, Dennis, and Leonard and a few others from time to time... have all experimented with countless speakers in this room... evaluations, professional reviews, and my personal speakers of many different flavors. They have all performed best pull on out into the room. However, this is always at the sacrifice of the low end of the mains (save a few models), where the response is less than desirable below 120Hz.... so subs are a must.

I have tried using the acoustical timing and basically take a bunch of measurements of the front and several for the rear... until I get two measurements with the same reference point. Then I adjust phase on the rears, which are closer... then do more measurements until I get two matching the same reference point again... probably an unfruitful waste of time, as I don't think it's working properly. I have never really understood phase.. just seems too complicated for me. I am looking for options on how best to accomplish this task correctly.

So... I have called on jtalden for assistance in this thread. He is very knowledgeable with phase and time alignment of speakers... and has helped several others with excellent results. He has agreed to evaluate my setup and help me figure it all out, while at the same time agreeing to allow this to be a guide on what and how he does it. Hopefully it will be help for those of us who don't quite understand it as well as we would like.

@jtalden ... you can tell me what you need from me and I'll work on getting it for you. However you would like to preface what you will be doing is fine by me... I sincerely appreciate your assistance.

Here are pics of the rear subs that I want to keep as they are... and the front subs next to the mains.

 

[jtalden]

Hi Sonnie,
General thoughts/comments

Setting aside room treatment and automated setup capabilities in most AV processors, a manual multi SW setup can be approached in 4 steps:

• Select individual SW positions/delays and the LP position to achieve a reasonably smooth SPL at the LP. [The group of SWs is then considered as a single speaker.]
• Select a favorable frequency for the SWs to main speaker crossover point based on favorable modal relationship between the SW group and the mains. [Often we instead just set the XO to the recommended 80 Hz, or randomly pick another frequency we think will work better.]
• Select favorable delay timing for the SWs to mains crossover.
• EQ

The initial question here is related to step 1. With identical SWs a close phase relationship of the direct sound of the 4 subs is achieved at the LP by just setting delay timing for each of the SWs to account for any difference in the distances to the LP. With 2 fronts equidistant and 2 backs equidistant and using the phase control to adjust the 2 rear SW, the direct sound phase is possibly already reasonably well aligned. The steady-state phase relationship at the LP may be very different, however.

The only delay timing I see in this equipment list is the SW distance setting in the NAD M17. A quick look into the owner’s manual suggest there is only one SW distance adjustment. There is 1 XLR sub output jack and 2 RCA SW output jacks, but I assume all 3 jacks are controlled by that one SW distance setting. That being the case, we currently cannot adjust the relative timing alignment between the 2 front Vs 2 back SWs hence your use of the phase control on the rear SWs. Using REW for analysis however we can compare the SPL and phase that exist now with those that can be achieved at various other SW locations and by adding in a MiniDSP, DCX2496, or other box to allow for independent delays for all 4 SWs. Another box may, or may not, provide a significant improvement. An analysis in REW can help sort this out.

I should emphasize that in the bass range it’s the steady-state SPL smoothness at the LP that is the primary indicator of sound quality. The phase relationship of direct sound is not the major factor. Large direct sound timing differences can be problematic, but sometimes smaller ones may be beneficial. It is probably wise to stay reasonably close to the direct sound timing so I start there first and look as minor adjustments to see if that is helpful.

[The direct sound is a more dominate factor in sound quality at the higher frequencies and close phase tracking handoff between the drivers in the main speaker is the preferred target.]

Following are my thoughts on how to proceed if we want a comprehensive study of different SW locations with the addition of a box to provide independent SW timing:

Setup

• For the front SW locations that are being considered, provide a location number and a description of the location. [or possibly mark the locations on a photo?] Confirm which numbers correspond to the current front 2 positions now in use.
• The SW distance setting in the NAD M17 can be left unchanged (it doesn’t impact this analysis)
• * Disable Dirac in the NAD M17
• * Set the XO for the SWs to the highest frequency in the NAD M17
• * Set any LFE filter in the NAD M17 to the highest frequency
• * Set any lowpass filter in the SWs off or to its highest frequency setting
• Use REW acoustic timing on all measurements
• All sweep measurements can just be left to your standard full range (10-20k Hz?)

* Optional settings - these would allow using this same data in Step 2 above – ‘Select a favorable frequency for the SWs to main speaker crossover point’

Measure

• A SW in each position. [It need not be the same SW just pick the easiest to move.]

[This step creates a measurement for each of the back SWs (with the phase control at the current setting) and a measurement for each of the possible locations in front.]

• Measure each rear SW, but this time with the phase control set back to 0°.

Analyze

• The current combination of the 4 SWs in their current positions and with the current phase settings will be used as the reference condition.
• REW will be used to analyze all or most of 4 SWs combinations with favorable delay timing applied.
• I will select 1 or 2 favorable cases to show comparison charts and offer my analysis.

Is this the scope you were think of?

If you are instead more interested in confirming what the current setup looks like regarding direct sound phase and LP steady-state phase that can be done by just measuring each of the 4 SWs with all the current XO and filter settings. You may already have those measurements. We could start there and decide later whether you want to do something more comprehensive. If there are other locations that you would really like to use for the front SWs and you don’t mind another box then that would be a strong motivating factor in the a comprehensive analysis.

 

[Sonnie Parker]

I think we are on the same track of thinking here... I'll get these measurements done hopefully on Saturday at some point during the day... being Bama has the day off, although I still like to watch some football.

 

[AustinJerry]

To facilitate time alignment for four subs, I recommend adding a MiniDSP 2x4, which allows a unique delay setting on four output channels. At ~$100, it is a very cost-effective solution. Plus, it allows you to use REW’s EQ tool to create PEQ filters to address any remaining response issues with the combined sub signal. I also use Dirac Live, and if you are interested in my approach, I have attached a link to my setup guide.

Dropbox

www.dropbox.com

 

[JStewart]

Select a favorable frequency for the SWs to main speaker crossover point based on favorable modal relationship between the SW group and the mains. [Often we instead just set the XO to the recommended 80 Hz, or randomly pick another frequency we think will work better.]

@Sonnie , does your NAD allow distance changes when using Dirac filters? If not @AustinJerry s minidsp suggestion may prove necessary.

 

[Sonnie Parker]

Hi Jerry...

Yep... read your guide... great information for sure. Unfortunately I don't have a miniDSP at this time, although as cheap as they are, I may just go ahead and pick one up. However, i was trying to accomplish this without having another source in the signal path. I also have a DEQ2496 that would separate timing from front to rear with two inputs, although the DCX would be better if my front and rear subs were not equidistant.

About to get into some measurements now.

 

[Sonnie Parker]

@Sonnie , does your NAD allow distance changes when using Dirac filters? If not @AustinJerry s minidsp suggestion may prove necessary.

It does not, which is why we are using the variable phase adjustment on each SB16.

 

[Sonnie Parker]

I have a miniDSP DDRC-88A w/BM ... so I could insert it into the chain and solve the timing issue pretty easily. It is quite larger than the 2x4 ... but it would work.

However... I want to learn more about phase and how it interacts with these subs, so we will continue with this method and then later I might stick the DDRC in there to see how it shapes things up.

 

[JStewart]

It does not, which is why we are using the variable phase adjustment on each SB16.

Sorry to not be clear, the comment was more toward the sub-group to mains alignment that Dirac might change after you set it manually as outlined by @jtalden . I guess OTOH Dirac may choose the same relationship between the sub-group and mains. For sure with a minidsp that relationship could be adjusted post-Dirac if desired.

Edit: Oops. Got sidetracked for a moment. Didn't see your last post.

 

[AustinJerry]

I have a miniDSP DDRC-88A w/BM ... so I could insert it into the chain and solve the timing issue pretty easily. It is quite larger than the 2x4 ... but it would work.

However... I want to learn more about phase and how it interacts with these subs, so we will continue with this method and then later I might stick the DDRC in there to see how it shapes things up.

While the 88A would certainly give you the ability to set individual delays, it seems like overkill. The 88A is a popular device which I am sure you would find no difficulty in selling, thereby funding the purchase of a 2x4 with money to spare. Just a thought...

 

[Sonnie Parker]

Sorry to not be clear, the comment was more toward the sub-group to mains alignment that Dirac might change after you set it manually as outlined by @jtalden . I guess OTOH Dirac may choose the same relationship between the sub-group and mains. For sure with a minidsp that relationship could be adjusted post-Dirac if desired.

Edit: Oops. Got sidetracked for a moment. Didn't see your last post.

Yeah... the DDRC I have could cure that problem. Once DL is engaged... it turns off level and distance adjustments... and if any were made pre DL... it resets them to 0. The only thing that can be changed is the speaker configuration/crossover setting.

It might be interesting to see the difference and learn what DL is doing pre and post with the phase to mains alignment.

I have mentioned to both NAD and DL that they should look at adding BM to the M17... similar to what a few others are doing with their higher end units... then we would have level, distance/delay, and PEQ for each channel for post DL adjustment. it's really not a complicated or costly implementation... and I'd pay $100 for the software if that is what it took.

 

[Sonnie Parker]

While the 88A would certainly give you the ability to set individual delays, it seems like overkill. The 88A is a popular device which I am sure you would find no difficulty in selling, thereby funding the purchase of a 2x4 with money to spare. Just a thought...

It was a loaner from miniDSP, so I would not sell it. I could probably send it back and have them send me a 2x4 as a loaner... but as cheap as it is, I could also just buy one and it would be easy to conceal behind my equipment rack.

I really only need two channels as long as I keep the front equidistant and the rears equidistant.... being my room is so symmetrical in every way.

 

[Sonnie Parker]

@jtalden

Here is the room and the sub locations along with the main listening position and distance measurements. REW measurements forthcoming.

Purple speakers are the MartinLogan main speakers.

 

[Sonnie Parker]

For clarification... I set the front speakers to Large in the M17, since this effectively turns off the crossover and allows the subs to run full range. My preferred crossover setting in the M17 for the mains to subs is 80Hz.

I turned off the PEQ filter I had set in the subs... and turned off the crossover setting for the front subs that I was using. I also had the front subs with an additional 60Hz crossover setting which smoothed out the response a little and took away a small dip I had at 74Hz. Those crossovers on the front subs are turned off for these measurements.

The first round of measurements on the rear subs... SUB 1 and SUB 2 are with the 30 degree phase adjustment and labeled SUB 3 LEFT REAR 30* and SUB 4 RIGHT REAR 30*.... and the second round of measurements are labeled 0* for zero phase adjustment.

If this file is too large, let me know and I'll upload the measurements separately.

 

[jtalden]

Thanks, I will take a look at these measurements.

 

[jtalden]

The analysis indicates the current 4 SW setup is very favorable. The front pair has very similar SPL timing and phase. The same is true of the back pair. This is not surprising given the very symmetrical setup of the room. Front and Rear SWs also complement each other well in that some of the deeper SPL nulls in one pair are mitigated by the other pair. This current 4 SW setup (rear phase set to 30°) results in direct sound timing identical for all 4 SWs.

Charts

Below is the impact of the 30° phase setting on the 2 rear SWs. I found that the 30° phase setting results in a 1.66ms timing delay.

Below is the impact of the 0° vs 30° setting on phase rotation.

[I have found the phase control on my DCX2496 to also to be simple timing delay. That seemed redundant to delay controls in the DCX so I never used it. In this case the SW provides an additional timing delay to the ‘distance’ setting in the AV processor. I have always been curious as to what the frequency reference is used to call this a ‘30° delay’. The 1.66 ms delay measured here would suggest that it is referenced to a 60 Hz frequency. Maybe mains frequency is the standard reference for these SW phase controls?]

Back to the analysis:

Below is the relative phase of the current 4 SW setup with a 4 cycle FDW applied. This approximates the phase of the direct sound at the mic. A 4 cycle FDW still allows some indication of early room effect like the irregularity 60-80 Hz range. The lower the setting the smoother the chart, but a setting of 1 are even 2 cycles starts to distort the rotation at the frequency extremes. This is excellent direct sound timing for the reference setup and thus the 0° phase setting would not be as favorable.

Below is the impulse timing of this same reference setup. It supports the conclusion that the 4 SWs are timed very closely. The initial impulse rise is at the same time for all 4 SWs.

Now let’s see what is going on at the LP where we want the steady-state SPL to be smooth.

Below is the phase with the FDW removed providing the steady-state condition. The phase tracking still looks great from 10-55 Hz. The room effects are disrupting the tracking in the 55-105 Hz range. Green is Front pair, Purple is Rear Pair

Below is the SPL chart. It’s a little hard to read. Shown are; front pair (green), rear pair (purple), and the resulting 4 SW group (red). The phase irregularities only resulted in relatively minor SPL issues. The final SPL response is favorable for EQ.

Below is another SPL chart hinting at the difference in the SPL of the early direct sound arrival at the LP vs the steady-state LP situation. Both traces are the same data. The smoother green trace utilizes a 4 cycle FDW. The red trace is the same steady-state response shown above.

Below is the steady-state phase result. There is much less room impact than what is normally seen in typical listening rooms. This results from a well-designed setup. It requires favorable; room size, room dimensions, symmetry of the layout, acoustic treatments, and well selected LP and SW locations.

Considering the very smooth SPL result with Dirac engaged there is little reason to expect that finding other SW locations that works better than this for the initial SPL will result any significant sound improvement. Even if the small SPL dips your saw after Dirac could be addressed, I suspect that no significant improvement in perceived sound quality will be achieved.

I will answer any questions about this work. We can also proceed to do any additional work you would like. We could confirm the XO handoff timing for instance; it’s entirely your call.

For my benefit I would like to see an mdat containing left channel and right channel measurements using all your reference settings. I have seen lots of Dirac SPL charts, but have not yet had the opportunity to review any measurements.

 

[Sonnie Parker]

I certainly appreciate the analysis... I was eager to find out if it was as good as it was likely to get. I have no complaints with the sound at this point, but knowing that it's pretty much on the money makes it sound better huh? lol... at least that's my perception and I'm sticking to it.

How do we go about confirming the XO handoff timing? We can do that if you'd like.

I thought maybe I would try the subs on the inside of the mains to see if I would feel anymore impact from the subs, being they would be a little closer to the LP and also literally firing directly at and in front of the LP.

On my last session I changed the xover on the front subs from 80Hz to 60Hz after running Dirac Live to get rid of that small issue at 74Hz, so I want to set the xover on the front subs to 60Hz before DL and see what it looks like.

I would be happy to send you some files on the left and right channel measurements.

What would you like me to do first for your further analysis. I am thinking you will want me to do more measurements where the subs are currently before anything else.

 

[jtalden]

Let's first address your thoughts - move the SWs, set the XO to 60 Hz, or both.

[Wouldn't changing the XO on the front SWs change all SWs? Maybe you changed the FL, FR to 60 Hz and left the CC and surrounds to 80 Hz? Please clarify.]

If you have a setup combination you would like to try, I would tend to just evaluate the setup with all your intended settings. We still will be able to see the net change of the new SW group to the reference SW group we just completed we just won't have the detail of the contribution of SW in the group.

My initial thought is:

• Set all the new settings in the SWs and AVP (XO, SW Distance, SW phase, SW locations, etc., that you are interested in trying)
• Run Dirac Setup
• Use full range sweeps
• Use acoustic timing

Measure with Dirac off:

• SW group (redirected bass output from the FL or FR, not LFE)
• FL main (by itself)
• FR main (by itself)

Measure with Dirac on:

• SW group (redirected bass output from the FL or FR, not LFE)
• FL (by itself)
• FR (by itself)

We can calculate the FL+SWs and FR+SWs in REW or you can measure them also if you like.

This plan may not be appropriate if; we need to first investigate to determine any of the settings, or if Dirac setup constraints don't make this workable. We will then have to adjust the plan accordingly.

If this can be done we can:

• Compare the SWs group to the reference settings.
• Deterine if the XO distance setting is favorable
• Compare the result with Dirac to the current reference settings (This assumes you already have FL and FR channel measurements of the reference setup.)

Does this help?

 

[AustinJerry]

Just curious, why does Sonnie's impulse look different from my impulse? Sonnie's impulse has a jagged line, while mine is a smooth line. Is this a REW measurement setting? Which is correct?

 

[Sonnie Parker]

I noticed it too Jerry... my previous measurements were smooth as well. I think the only thing different was the frequency sweep on my previous measurements were only 0-400Hz, while these are 0-22khz. Not sure if that would make a difference, but easy enough to check.

@jtalden ... I had the crossover in the M17 set to 80Hz, which applies to the mains and the subwoofers... then I further added the 60Hz on the front subs.

I will work on the other items you mentioned later this evening... as I will be out of pocket for a while. The plan you have outlined looks good and should be interesting. Thanks!

 

[AustinJerry]

I noticed it too Jerry... my previous measurements were smooth as well. I think the only thing different was the frequency sweep on my previous measurements were only 0-400Hz, while these are 0-22khz. Not sure if that would make a difference, but easy enough to check.

No, that's not it. I checked several of my 15-20KHz measurements, and the impulse is smooth. I'm sure it is a setting somewhere, and regardless, it doesn't affect the validity of the analysis.

 

[jtalden]

Jerry,
My guess - both correct.
My SW measurements are jagged also. Measuring just the SW, the noise floor is down only about 25 dB from a 75-80 dB measurement. A clean SW impulse measurements may result if the sweep is not full range. A 10-500 Hz measurement range may smooth it out a lot. Possibly, testing at a much higher level may also help smooth the result.
I would have to experiment to verify these thoughts.

 

[Sonnie Parker]

I had this happen once before, and was able to make it go away... I will see if I can figure it out if I get a chance to do so a bit later.

 

[Sonnie Parker]

It is the sweep limits that cause it... if I change it from 400Hz to 22kHz... I get the very unsmooth sweep line.

 

[Sonnie Parker]

@jtalden ... I moved the subs to inside of the front mains and reset the phase... resulted in 21° phase setting on the rear subs, as the front subs are now a little closer.

I ran several measurements full range with and without timing and also a few only up to 400Hz to smooth out the impulse response lines.

.mdat files attached

I noticed the timing appears to be off from the subs to the mains, but not sure if the reading is caused my crossovers. I know when I set the xover in the front subs (although they are not set in these measurements), that it greatly effects the timing impulse. if I set the front subs to 60Hz and then compare timing, I have to set the rear subs at 57° to get the timing aligned from front to rear.