Manually time-aligning subwoofer(s) to mains - how to

[ernestcarl]

@ernestcarl the file below has measurements with the new delays, I actually ended up limiting Dirac's correction range up to only 2KHz but it shouldn't make much of a difference regarding bass.

I don't think the response changed much other than the extra headroom (alignment wise). which is great ofcourse.

https://drive.google.com/file/d/1qbz9ko-puvpimiG3Baxb5aqcZ0WHnyCY/view?usp=sharing

Looks better and smother actually with the subs aligned. I don't know if this is feasible in your case, but bass traps might improve even further that range between 200-500Hz where we see a bit of some uneven & extended GD and decay.

 

[abdo123]

Looks better and smother actually with the subs aligned. I don't know if this is feasible in your case, but bass traps might improve even further that range between 200-500Hz where we see a bit of some uneven & extended GD and decay.

That's pretty much the next step, I'm bombarded by SBIR and LBIR because both the coach and the speakers are located at two opposite edges of the room. and maybe a third subwoofer, I'm baffled by how much multisub smooths out the frequency response.

 

[bbizzle]

This is such a good thread! I'm learning so much.

I'm trying to follow along and I'm not sure what I can do about what I'm seeing. I think I need to delay my sub about 10 ms to align it, but I'm using a nanoDIGI which only allows up to 9 ms of delay. I'm using a umik-1 with nearfield genelec 8030cs about 3 ft away. Sub is an SVS 2000 pro sealed sub and it's right in between my monitors. I know there is some delay in the sub from the dsp, but I didn't think it would be beyond my ability to delay it with my nanoDIGI.

Could someone please take a quick look at my files and let me know if I need to come up with a new approach? It would mean the world to me .

To summarize my setup:
-Nearfield Genelec 8030cs (L and R) with a SVS SB2000 Sub.
-All connected using a NanoDIGI with 2 SCHIIT DACS to power the monitors and the Sub.
-I followed the requirements given in the initial post (no crossovers or PEQ, bandwidth limited sweep 20-250 hz).
-Scans were windowed with 6 cycles.
-I also smoothed (1/24 db) for clarity.

Impulse:

Phase:

Step:

 

[Daverz]

This is such a good thread! I'm learning so much.

I'm trying to follow along and I'm not sure what I can do about what I'm seeing. I think I need to delay my sub about 10 ms to align it, but I'm using a nanoDIGI which only allows up to 9 ms of delay. I'm using a umik-1 with nearfield genelec 8030cs about 3 ft away. Sub is an SVS 2000 pro sealed sub and it's right in between my monitors. I know there is some delay in the sub from the dsp, but I didn't think it would be beyond my ability to delay it with my nanoDIGI.

Could someone please take a quick look at my files and let me know if I need to come up with a new approach? It would mean the world to me .

To summarize my setup:
-Nearfield Genelec 8030cs (L and R) with a SVS SB2000 Sub.
-All connected using a NanoDIGI with 2 SCHIIT DACS to power the monitors and the Sub.
-I followed the requirements given in the initial post (no crossovers or PEQ, bandwidth limited sweep 20-250 hz).
-Scans were windowed with 6 cycles.
-I also smoothed (1/24 db) for clarity.

Impulse:
View attachment 137796

I think you mean the delay for the mains?

Are you sure you are keeping the sub from playing when the mains are and vice versa for the band-limited sweep?

I used a sweep file with the sweep on the left channel only and the timing signal on the right channel. Then the routing I use on my nanoDIGI is

mains only

main right (to play timing signal) and sub only

Though I suppose one could do the same thing by just turning off one DAC and/or unplugging a channel on one of the DACs.

With my mains 2.3m from the listening position, I ended up with the SVS SB1000 Pro 3m away and a delay of 8.6 ms for the mains. In other words, I can only put the sub a couple feet further away than the mains before I run out of delay in the nanoDIGI.

 

[bbizzle]

I think you mean the delay for the mains?

>>Ah, yes you are correct. I I need to "push back" my mains.

Are you sure you are keeping the sub from playing when the mains are and vice versa for the band-limited sweep?

I used a sweep file with the sweep on the left channel only and the timing signal on the right channel. Then the routing I use on my nanoDIGI is

mains only

View attachment 137814

main right (to play timing signal) and sub only

View attachment 137815

Though I suppose one could do the same thing by just turning off one DAC and/or unplugging a channel on one of the DACs.

With my mains 2.3m from the listening position, I ended up with the SVS SB1000 Pro 3m away and a delay of 8.6 ms for the mains. In other words, I can only put the sub a couple feet further away than the mains before I run out of delay in the nanoDIGI.

Yep, this is exactly how I did it as well. What I'm so confused about is why the phase is so different for the sub compared to my monitors. They are almost the same distance away from my monitors but the phase response looks so different.

SVS has mentioned that the dsp in their subs adds about 5.5-6 ms of latency, so I expect that there would be some difference, but the phase response doesn't seem to improve when I delay by this amount - the changes in the step/phase plots only change by a very small amount.

 

[bbizzle]

After adding the max amount of delay to my mains, the step looks closer to aligned (i think). Is the goal to just make the initial peaks aligned?

 

[Daverz]

After adding the max amount of delay to my mains, the step looks closer to aligned (i think). Is the goal to just make the initial peaks aligned?

I used the impulse response to align and checked it with the step response. One of my DACs is inverting, and I caught it because I could not align both until I undid that.

 

[fluid]

The best alignment will depend on where you want to set the crossover and you have a bit of boundary interference that is going to make getting a really good response difficult with a straight crossover.

Somewhere around 12ms seems to work by reading the peaks of the waveshapes which align quite nicely with 100Hz filtering
Attached is a step response to show you what it looks like and the alignment tool which shows pretty good phase tracking.

It also seems like both your speakers have the highest peak in a negative direction which may or may not be correct.

 

[bbizzle]

The best alignment will depend on where you want to set the crossover and you have a bit of boundary interference that is going to make getting a really good response difficult with a straight crossover.

Somewhere around 12ms seems to work by reading the peaks of the waveshapes which align quite nicely with 100Hz filtering
Attached is a step response to show you what it looks like and the alignment tool which shows pretty good phase tracking.

It also seems like both your speakers have the highest peak in a negative direction which may or may not be correct.

Thanks so much for your insight! I'm still very new to this and trying my best to learn. How did you identify the boundary interference from the figures? Just loss of low frequency information? Or is there some feature in the plots that is a giveaway?

Also, I'm using a nanoDigi which only allows 9 ms of delay compensation. Is there anything else I could do other than physically move my speakers to more closely phase align my speakers or improve my speaker integration with my room in general? Thanks again! Really appreciate your time.

 

[fluid]

Thanks so much for your insight! I'm still very new to this and trying my best to learn. How did you identify the boundary interference from the figures? Just loss of low frequency information? Or is there some feature in the plots that is a giveaway?

You're welcome. Boundary interference is usually easy enough to spot when looking at an SPL response with a Frequency dependent window or the spectrogram view. In your sub measurement there is a sharp dip at 82Hz with an associated phase wiggle. In the spectrogram view you can see that the peak energy arrives at 53ms or so. There is a null early in time which is eventually overcome from sound arriving from the rest of the room.

Also, I'm using a nanoDigi which only allows 9 ms of delay compensation. Is there anything else I could do other than physically move my speakers to more closely phase align my speakers or improve my speaker integration with my room in general? Thanks again! Really appreciate your time.

The extra delay is most likely coming from the SVS sub as they do have quite a large propagation delay. I have never used one or looked at one to know how they can be wired but it might be possible to daisy chain through the sub so that the nanodigi gets a delayed input which would then only need 6ms or so to align it. The other option is to consider changing polarity of the sub as that can have a delay effect as well.

 

[GalZohar]

Hi,

What I tried to do was to change the delay on my subwoofer in the receiver until I received what appeared the most flat response in REW. Ended up with pretty much the same delay being optimal for both L, R, and L+R measurements (using 80Hz crossover and Audyssey on with Dynamic EQ off). Ended up increasing the distance from 3.8m to 5.8m. The improvement was audible even for a beginner like myself. Had the same results both with speakers 0.5m from wall and 1m from wall (subwoofer remaining in the corner behind and to the right of the right speaker) - Audyssey measured the same subwoofer distance and needed the exact same manual correction of +2 meters to the subwoofer.

After reading all these complicated ways to measure, I'm wondering if I was doing something wrong? Could the optimal delay be different and I can make things sound even better? Or did I just take the slow way to do things (having to measure over and over with different delays to find the optimal one), but the result should still be optimal?

I'm using SVS pb-2000 with svs ultra bookshelves on stands, with a denon X2700h.

 

[ernestcarl]

Hi,

What I tried to do was to change the delay on my subwoofer in the receiver until I received what appeared the most flat response in REW. Ended up with pretty much the same delay being optimal for both L, R, and L+R measurements (using 80Hz crossover and Audyssey on with Dynamic EQ off). Ended up increasing the distance from 3.8m to 5.8m. The improvement was audible even for a beginner like myself. Had the same results both with speakers 0.5m from wall and 1m from wall (subwoofer remaining in the corner behind and to the right of the right speaker) - Audyssey measured the same subwoofer distance and needed the exact same manual correction of +2 meters to the subwoofer.

After reading all these complicated ways to measure, I'm wondering if I was doing something wrong? Could the optimal delay be different and I can make things sound even better? Or did I just take the slow way to do things (having to measure over and over with different delays to find the optimal one), but the result should still be optimal?

I'm using SVS pb-2000 with svs ultra bookshelves on stands, with a denon X2700h.

It's only complicated in the sense that we are scrutinizing the additional component of "time" as a form of a trace or graph for each individual speaker & sub rather than just focusing on the overall look of the summed magnitude trace. The time aspect can be observed and adjusted virtually by looking at an overlay of the impulse graphs and/or phase traces to predict what the summed outcome would be even before measuring the speakers and sub in unison. But it's not necessarily going give one a better result from your more manual (adjusting the time by small increments) /automatic (Audyssey) way of doing things.

 

[GalZohar]

It's only complicated in the sense that we are scrutinizing the additional component of "time" as a form of a trace or graph for each individual speaker & sub rather than just focusing on the overall look of the summed magnitude trace. The time aspect can be observed and adjusted virtually by looking at an overlay of the impulse graphs and/or phase traces to predict what the summed outcome would be even before measuring the speakers and sub in unison. But it's not necessarily going give one a better result from your more manual (adjusting the time by small increments) /automatic (Audyssey) way of doing things.

Thanks!

Physically speaking, the only thing I might be doing poorly is that I could delay the subwoofer by an entire 360 degrees at 80Hz more than I should. Although I suppose if I try other delays (which give the same phase at 80Hz) I could see a difference in the other frequencies.

Supposedly if the 2 signals arrive at the same time I would expect the best summation... But then I just don't know how to explain Audyssey running my subwoofer 2m too short... Maybe the timing of Audyssey was correct but there is still a phase difference between the speakers and subwoofer that requires fixing by mis-aligning the delays on purpose? I suppose this is something I should be able to somehow see with REW measurements?

 

[ernestcarl]

Supposedly if the 2 signals arrive at the same time I would expect the best summation...

Not necessarily... if focusing only at 80Hz. The phases may still not be fully aligned throughout the entire relevant crossover region/zone of interaction -- a manual change in xo filter type for the speakers and/or sub -- or use of additional FIR phase manipulation may prove beneficial. I'm a very visual type of person so I would rather see the actual traces in something like REW, Smaart, or similar graphical software rather than second-guessing whether I or something like Audyssey (supposed equivalent/estimated distance) already has it optimized as best as possible:

How to use delay to match slopes in your crossover alignment

I suppose this is something I should be able to somehow see with REW measurements?

Yes, you should be able to do/see this with your AVR and REW. Just don't ask me "how" cause I've never owned an AVR in my life!

 

[GalZohar]

I meant not only at 80Hz. Supposedly if the sources are the same phase and arrive at the same time then they should arrive at the same phase. But maybe they just don't leave the speaker and the subwoofer at the same phase? But then we actually mis-align their timings in order to align their phases and therefore time aligning is pointless when we just want to avoid phase cancellations which I'm assuming is more audible than a tiny time delay?

Don't all frequencies arrive at the same time (due to speed of sound)? I'm guessing dispersion in air is negligible?

It's still not 100% clear to me regarding what we actually want and of course also how to achieve that. All I know how to do right now is play with values until the frequency response in REW appears the flattest.

If there are more sophisticated ways to modify phase that's cool and probably for the higher-budgeted systems, my AVR (x2700h) can only do some automatic filters of Audyssey and has automatic level and delays (distances), where the latter can be manually modified. Can't even modify the crossover filters, just the crossover frequency.

 

[ernestcarl]

I meant not only at 80Hz. Supposedly if the sources are the same phase and arrive at the same time then they should arrive at the same phase. But maybe they just don't leave the speaker and the subwoofer at the same phase? But then we actually mis-align their timings in order to align their phases and therefore time aligning is pointless when we just want to avoid phase cancellations which I'm assuming is more audible than a tiny time delay?

Don't all frequencies arrive at the same time (due to speed of sound)? I'm guessing dispersion in air is negligible?

It's still not 100% clear to me regarding what we actually want and of course also how to achieve that. All I know how to do right now is play with values until the frequency response in REW appears the flattest.

If there are more sophisticated ways to modify phase that's cool and probably for the higher-budgeted systems, my AVR (x2700h) can only do some automatic filters of Audyssey and has automatic level and delays (distances), where the latter can be manually modified. Can't even modify the crossover filters, just the crossover frequency.

Why not just check the time domain stuff in REW? e.g phase curves, group delay etc.

The phase curves/slopes of different speakers/subs are not always the same/fully compatible -- unless if they were purpose built for each other. Which means that even if you use the same xo slopes, you're not always going to get the perfect expected summation even if you get the set time delays dialed in as best as possible. Additionally, the room and positioning can alter the phase response. I don't know about your receiver specifically, but, yes, some do already have automatic FIR phase manipulation capabilities.

 

[GalZohar]

Why not just check the time domain stuff in REW? e.g phase curves, group delay etc.

The phase curves/slopes of different speakers/subs are not always the same/fully compatible -- unless if they were purpose built for each other. Which means that even if you use the same xo slopes, you're not always going to get the perfect expected summation even if you get the set time delays dialed in as best as possible. Additionally, the room and positioning can alter the phase response. I don't know about your receiver specifically, but, yes, some do already have automatic FIR phase manipulation capabilities.

Can I get anything useful from the combined measurement or do I have to measure the subwoofer and speaker separately (that would take some more work).
Also, Audyssey uses FIR filters but I think it's only on a per speaker basis, and it doesn't do anything special for speaker-subwoofer integration except for setting the delays and levels. So even if I had more information, what could I do with it?

 

[ernestcarl]

Can I get anything useful from the combined measurement or do I have to measure the subwoofer and speaker separately (that would take some more work).
Also, Audyssey uses FIR filters but I think it's only on a per speaker basis, and it doesn't do anything special for speaker-subwoofer integration except for setting the delays and levels. So even if I had more information, what could I do with it?

Sure you can… don’t need separate measurements to check the phase, group delay, or even a spectrogram of the summed response. It should at least give you some idea how delayed the sub/bass frequencies are in relation to the main speakers.

I suppose if one of your sources comes from a PC’s HDMI out, you could do some “extra” analysis or DSP convolution processing beforehand via something like Dirac or Audiolense or maybe even rePhase if desired… not that I think it’s something one absolutely “has to do” really.

 

[thorvat]

After adding the max amount of delay to my mains, the step looks closer to aligned (i think). Is the goal to just make the initial peaks aligned?

Goal is to match the phase between sub(s) and mains so they are as close as possible at and around the crossover frequency.

As you have done good job with measurement this can be done quite easilly using "Align phase at cursor" feature available within Alignement tool which can be activated under "All SPL" tab.

As shown here, choose your sub in upper dropbox and your left main in lower dropbox. Set your crossover frequency by clicking it on the graph (in this example I chose 100Hz) and click on "Align phase at cursor" button. REW will calculate necessary delay for chosen channel so the phase at the chosen frequency is matched. Pressing "Aligned copy" button will generate measurement with aligned phase. Repeat the same procedure for right channel.

Once you do that phase graph will look like this (upper 2 curves are your mains, blue curve is sub and lower 2 curves are aligned mains):

Impulse:

And step:

Setting mains delay to 9ms should still result in pretty decent phase alignement. You can check final result by measuring FR when all 3 speakers are playing together pink noise using moving microphone method. After doing that you can adjust amplitude response of your sub to align it with mains as this is as important as aligning phase. Finally, you can activate the crossover and run sweep on left and right channel to check if sub was properly integrated with mains. In ideal case you should see smooth phase and GD curve around the crossover region for both channels.

If that goes well you can apply room EQ via PEQs as you would without the sub as sub should now act as if it was part of your mains.
If you have the ability to implement FIRs as well you can use them to correct phase of your XOs, but be sure to include XO between sub and mains in that correction in addition to the XOs of your main speaker drivers.
Good luck!

 

[SitronNO]

While I am trying to align the sub to my fronts, I followed the steps from the first post: Measured my speakers+sub, one by one, with XO turned off, and did a 30-150Hz sweep. Now I am trying to figure out if it's all correct, by looking at the graphs, and to me it seems that all speakers are aligned properly, but please do check!

But by looking at the post right above from @thorvat , I tried using the "Align phase at cursor" in the Alignment tool. And when comparing Sub to FL vs Sub to FR, or comparing FL to FR, the alignment tool sees a 5-6 meter diff? That my FL is 5-6 meter closer to MLP that FR? Using a tapemeasure, I know they are exactly 215cm from the MLP.

Have I done something wrong, or are there something wrong with my understanding?
MDAT: XO Off - 30-150 - 20211104.mdat