-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Subwoofer Time Alignment
- Thread starter Ron Texas
- Start date
Apologies to revive and old thread, but what is the consensus on figuring out time alignment between speakers and subs? If I look at the Wavelet in REW, I notice that my speakers crossed at 80Hz (and subs shut off) start to deviate from time 0 at around 140Hz. Around 80Hz they have 15 ms of delay. So should my goal would be that individual subs' distances give me 15ms of delay (same as speakers) at the cross over frequency? Or should I be aiming for 0 ms delay as my speakers are at say 500Hz?
Also, what is the maximum delay in bass frequencies that is not audible, basically it would not matter if a little late?
I find that Audyssey is not able to correctly determine the Distances of individual subs, but perhaps does not emphasize much on that because it does not matter? In which case we should be tweaking for the smoothest frequency response.
Also, what is the maximum delay in bass frequencies that is not audible, basically it would not matter if a little late?
I find that Audyssey is not able to correctly determine the Distances of individual subs, but perhaps does not emphasize much on that because it does not matter? In which case we should be tweaking for the smoothest frequency response.
What makes you say that? How many subs do you have?
Depending on which audyssey you have, it can do 1 or 2, My AV7703 does 2 and when I've compared it's results against the actual measured distances it is very close, close enough that I'm comfortable with the wizards result. When you run the wizards setup it first measures the distance and level of each sub individually, then corrects the overall FR at the listening chair.
KSTR
Major Contributor
For a "standard" subwoofer integrations, which means the sub and mains should have a resulting 4th-order LinkwitzRileay acoustic crossover at ~80Hz, a possible system check would consist of several stages:
Stage 1 : Establish correct individual frequency response for sub and mains measured in (quasi)-anechoic conditions to see if they follow the LR4 requirement (you have to look on magnitude and phase for this, and 360° offsets as well are to be avoided). The most important part is getting the phase right at and way beyond the crossover point, that is neglect magnitude perfection and rather focus on the phase match. Full or partial cancellation (phases being more than 120° apart at and around the XO point) is something we really want to avoid here.
Reset and polarity and phase controls on the sub for this. Sadly, it is not guaranteed that this starting goal can be fully met when an arbitrary sub is thrown in to assist arbitrary mains, not all sub designs are that flexible to adjust and ideally the sub must know exact FR of the mains to introduce the proper electrical XO. A way to do this is the mains having a switch that adjusts acoustic bass roll-off to 2nd order Butterworth at 80Hz/85Hz, along the lines of some THX specs. The sub then applies another electrical Bu2 on top of this so the LR so that the effective LR4 is created. For the sub's own LR4 slope, things are a bit more tricky. The phase of its own bass rolloff may leak into the XO range (and when adjustable to other freqs: the lower the hi-cut the larger the additional phase).
Stage 2 : Compensate for any time-of-flight differences for sub vs. mains, ideally by lining up subs and mains along the same listening distance at first (so, zero compensation). BTW, LR4 was chosen for the advantage of being the XO type being that is least affected by by time-of-flight delta. A mismatch of 30cm (1ft) is completely irrelevant. Most subs have a simple "phase control" (besides a polarity switch) that works by introducing additional phase at the XO region with an adjustable allpass filter. While this can and should be used to match phases it also introduces significant additional phase distortion (and corresponding group delay), so use with a bit of care. Polarity switch setting (if any) is done at stage 1 (for example, some 3-way mains may have "inverted" bass channels).
Stage 3 : Confirm FR at the listening position for sub and sat is close to identical at the XO region, again magnitude and phase, L and R seperately. You have to look at the pure delta, really. The FR might look rugged as hell from reflections and room modes but to 1st order the shape of mag and phase plots for each subsystem should be identical (and L and R seperately). This also applies to stage 1 as quasi-anechoic measurement are close to imposible at home. So you might start with measuring stage 1 right at the final listening position.
Stage 4: Now you have a system with a sub that is aligned almost as well as if it were integrated in the mains but you might still suffer from room modes. Ideally one might actually start with this step first, finding locations that are acceptable for the mains for correct imaging etc while giving a balanced bass FR, then do the same with the subs moving them around (also in height) while trying to keep the aproximate listening angle and a well-matched distance. What is left may be addressed in a final step by applying full "room correction" at the listening position that also will establish resulting linear phase XO.
Bottom line of this: When integrating subwoofer(s), notably arbitrary subs with arbitrary mains, you will happen to find yourself in the position of a speaker designer, no way around this.
The only simple way to resolve this without the tedious process outlined above is trusting (pro) manufactures that offer dedicated subs for their main speakers, and following their install guidelines.
-- All of the above strictly IME and IMHO ;-)
Stage 1 : Establish correct individual frequency response for sub and mains measured in (quasi)-anechoic conditions to see if they follow the LR4 requirement (you have to look on magnitude and phase for this, and 360° offsets as well are to be avoided). The most important part is getting the phase right at and way beyond the crossover point, that is neglect magnitude perfection and rather focus on the phase match. Full or partial cancellation (phases being more than 120° apart at and around the XO point) is something we really want to avoid here.
Reset and polarity and phase controls on the sub for this. Sadly, it is not guaranteed that this starting goal can be fully met when an arbitrary sub is thrown in to assist arbitrary mains, not all sub designs are that flexible to adjust and ideally the sub must know exact FR of the mains to introduce the proper electrical XO. A way to do this is the mains having a switch that adjusts acoustic bass roll-off to 2nd order Butterworth at 80Hz/85Hz, along the lines of some THX specs. The sub then applies another electrical Bu2 on top of this so the LR so that the effective LR4 is created. For the sub's own LR4 slope, things are a bit more tricky. The phase of its own bass rolloff may leak into the XO range (and when adjustable to other freqs: the lower the hi-cut the larger the additional phase).
Stage 2 : Compensate for any time-of-flight differences for sub vs. mains, ideally by lining up subs and mains along the same listening distance at first (so, zero compensation). BTW, LR4 was chosen for the advantage of being the XO type being that is least affected by by time-of-flight delta. A mismatch of 30cm (1ft) is completely irrelevant. Most subs have a simple "phase control" (besides a polarity switch) that works by introducing additional phase at the XO region with an adjustable allpass filter. While this can and should be used to match phases it also introduces significant additional phase distortion (and corresponding group delay), so use with a bit of care. Polarity switch setting (if any) is done at stage 1 (for example, some 3-way mains may have "inverted" bass channels).
Stage 3 : Confirm FR at the listening position for sub and sat is close to identical at the XO region, again magnitude and phase, L and R seperately. You have to look at the pure delta, really. The FR might look rugged as hell from reflections and room modes but to 1st order the shape of mag and phase plots for each subsystem should be identical (and L and R seperately). This also applies to stage 1 as quasi-anechoic measurement are close to imposible at home. So you might start with measuring stage 1 right at the final listening position.
Stage 4: Now you have a system with a sub that is aligned almost as well as if it were integrated in the mains but you might still suffer from room modes. Ideally one might actually start with this step first, finding locations that are acceptable for the mains for correct imaging etc while giving a balanced bass FR, then do the same with the subs moving them around (also in height) while trying to keep the aproximate listening angle and a well-matched distance. What is left may be addressed in a final step by applying full "room correction" at the listening position that also will establish resulting linear phase XO.
Bottom line of this: When integrating subwoofer(s), notably arbitrary subs with arbitrary mains, you will happen to find yourself in the position of a speaker designer, no way around this.
The only simple way to resolve this without the tedious process outlined above is trusting (pro) manufactures that offer dedicated subs for their main speakers, and following their install guidelines.
-- All of the above strictly IME and IMHO ;-)
I have XT32 and 2 pre-outs... and 2 subs. I say I do not trust the auto sub distances discovered because it differs between the AVR runs vs the MultEQ app runs. Additionally, if I use the Rhythmik guide for @80Hz I arrive at different distances than Audyssey. Lastly, the REW wavelets show delay in the sub frequencies and are not at 0 ms where my speakers are at 500Hz
The first image attached is the result after MultEQ app set distances of 20+ feet. The second image is after AVR set distances about 10 feet. Totally different, and the Wavelet shows that difference too.
Attachments
Last edited:
Thank you for the detailed explanation! I am re-reading your steps a few times as it is a bit too complex for me at this stage. I was hoping that someone would advise to do steps 1, 2, 3 in REW, using graphs A, B, C or the Alignment Tool by doing xyz. Perhaps something like guide: https://www.minidsp.com/applications/auto-eq-with-rew/measuring-time-delay but with just REW and what I can input in the AVR as distances. I do not have a miniDSP
I am pleased this fabulous thread has been revived. I am even more confused after having read it 
. I give a lot of eight to the works of Geddes and it seems that down low, things are steady state. It takes a while for us to even register a low bass signal , last I read, it was 50 ms for 50 Hz, by then the wave would have traveled 7 meters/23 feet. Let's put it in perspective, the wave would have likely hit a wall in all but the most palatial listening rooms... before we register it, thus the insistence of Geddes to obtain the most uniform response before EQ with quai-random placement of multiple subs, then EQ-uing since we are then in a quasi-steady state region. I question then, the effectiveness or audibility of 6 or 8 ms delay in the low bass region. Not everything that shows up on a graph is audible. With respect to KSTR and apologizing my laziness, what is an Acoustic LR4 crossover?
It is my opinion that Audyssey does a commendable job on the issues of distance and timing, at least in my system and my room
I really would like other heavyweights to join the debate and intervene. This is where we need their august inputs. The subject is replete with misconception and misunderstandings. Also ... I understand the value of Science in our daily lives but I would like the experts to distillate it to make things easy on us.
. I give a lot of eight to the works of Geddes and it seems that down low, things are steady state. It takes a while for us to even register a low bass signal , last I read, it was 50 ms for 50 Hz, by then the wave would have traveled 7 meters/23 feet. Let's put it in perspective, the wave would have likely hit a wall in all but the most palatial listening rooms... before we register it, thus the insistence of Geddes to obtain the most uniform response before EQ with quai-random placement of multiple subs, then EQ-uing since we are then in a quasi-steady state region. I question then, the effectiveness or audibility of 6 or 8 ms delay in the low bass region. Not everything that shows up on a graph is audible. With respect to KSTR and apologizing my laziness, what is an Acoustic LR4 crossover?It is my opinion that Audyssey does a commendable job on the issues of distance and timing, at least in my system and my room
I really would like other heavyweights to join the debate and intervene. This is where we need their august inputs. The subject is replete with misconception and misunderstandings. Also ... I understand the value of Science in our daily lives but I would like the experts to distillate it to make things easy on us.
Last edited:
dasdoing
Major Contributor
you can test what you can hear here: https://www.audiocheck.net/blindtests_timing_2w.php?time=10
btw: I would simply use excess group delay and/or step response
btw: I would simply use excess group delay and/or step response
I just did the test. 8 times out of 10 I was able to identify the 10ms delay. Here is my result:
Correct — Current score: 8/10 (80%) — Confidence : 94.53%
Could you please elaborate what you mean by "simply use excess group delay and/or step response." More specifically, what should we do?
dasdoing
Major Contributor
next step would be trying 5ms:
https://www.audiocheck.net/blindtests_timing_2w.php?time=5
well, step response at LP would be fine tuning; forget it atm
look at this excess group delay example:
in the pefect world it should be all 0. ignore the big spikes. the subwoofer is 25 to 30ms late here. play around with the delay until you got the average near 0. this actualy means that not the subwoofer needs to be delayed, but the mains*
*this is probably the reason this method is not realy popular, you end up with a little latency. now this can be compensated in most movie players
There is a know bug in Editor App that requires connecting the closest sub to sub 1 or it will give incorrect results?. Try reversing your connections. Otherwise I'm aware of many users that have checked ruler measured distance against the Audyssey and get very acceptable results.
YMMV?
Interesting... OK, will have to try that then.
This basically means we have to live with the sub delay otherwise if we try to time align, we need to add delay to speakers which is going to have some sync issues during playback - a no, no. In other words, we better give up on sub and main alignment for HT and focus on smoothest FR response?
dasdoing
Major Contributor
This basically means we have to live with the sub delay otherwise if we try to time align, we need to add delay to speakers which is going to have some sync issues during playback - a no, no. In other words, we better give up on sub and main alignment for HT and focus on smoothest FR response?
I am actualy no HT guy. For music it was an easy choice for me favoring aligned bass over less latency.
I am also not sure if all subwoofers are late, it seams all that have any type of DSP are. traditional passive subwoofers probably don't have that problem
ernestcarl
Major Contributor
Unfortunately, while only using a single sub, I'm one of those who has found themselves using rather unusual xo settings.
I'm not sure what the concensus is, if any, has been established as universal and 'best practice' over the course of this thread. KSTR mentioned a lot of good points for a start.
My own journey with doing things the manual way was pretty interesting... and frustrating.
Here's some measurements from my latest endeavor:
Some rudimentary adjustment options at the back of my sub:
I have to crawl down under a desk and blindly turn knobs just by feel, hence the different knob for volume control.
Having something like a miniDSP or software to make xo, delay, and FR, and volume adjustments is a godsend, over fidling over with physical knobs in each speaker each and every time.
The way the driver is positioned, phase looks good enough 'as is' with zero delay/phase change for the sub and no polarity inversion.
You can actually use REW's acoustic timing reference option to get good approximate delay needed for the mains.
Before doing anything with the sub(s)...
I first set my microphone exactly at midpoint between my left and right channels (stereo setup, in this case) at the MLP. You take several quick measurements and adjust the mic (or speakers) accordingly, until you have it situated pefectly at the exact midpoint between L&R speakers in the the MLP. Or adjust delay per channel via DSP...
Mine's already perfectly aligned:
You can also then check what suggestion REW thinks as a good delay for your mains using the same option:
*I think it was the left channel instead of right that was muted, so the image on the right is wrong for the settings used in REW!
One speaker channel is used as your acoustic timing reference, and the other channel for your sub only (mute opposite channel).
Of course, in the real world, things aren't always so easy and given to you on golden platter.
As it turns out, with many a trial and error runs, 13ms of delay for the mains gave me the best result -- least amount of cancellation and most boost around xo -- not only for the main listening position, but also in all seats of my couch area.
REW's 'Overlays' module is extremely helpful in making comparisons between individual measurements in one window:
Some people have less exacting standards when it comes to what's acceptable GD for subs...
That spike in delay around 130Hz or so can be seen in the wavelet as well:
Just the speaker:
Just the sub:
That odd gap certainly was not from the sub/caused by it...
Overlay of two plots vs all drivers summed:
So let's eliminate the room out of the equation using nearfield measurements.
Zoomed in using Linear (% peak) amplitude instead:
Makes a whole lot more sense than the 20-25ms I was seeing (and probably hearing?) at the MLP.
I'm not sure what the concensus is, if any, has been established as universal and 'best practice' over the course of this thread. KSTR mentioned a lot of good points for a start.
My own journey with doing things the manual way was pretty interesting... and frustrating.
Here's some measurements from my latest endeavor:
Some rudimentary adjustment options at the back of my sub:
I have to crawl down under a desk and blindly turn knobs just by feel, hence the different knob for volume control.
Having something like a miniDSP or software to make xo, delay, and FR, and volume adjustments is a godsend, over fidling over with physical knobs in each speaker each and every time.
The way the driver is positioned, phase looks good enough 'as is' with zero delay/phase change for the sub and no polarity inversion.
You can actually use REW's acoustic timing reference option to get good approximate delay needed for the mains.
Before doing anything with the sub(s)...
I first set my microphone exactly at midpoint between my left and right channels (stereo setup, in this case) at the MLP. You take several quick measurements and adjust the mic (or speakers) accordingly, until you have it situated pefectly at the exact midpoint between L&R speakers in the the MLP. Or adjust delay per channel via DSP...
Mine's already perfectly aligned:
You can also then check what suggestion REW thinks as a good delay for your mains using the same option:
*I think it was the left channel instead of right that was muted, so the image on the right is wrong for the settings used in REW!
One speaker channel is used as your acoustic timing reference, and the other channel for your sub only (mute opposite channel).
Of course, in the real world, things aren't always so easy and given to you on golden platter.
As it turns out, with many a trial and error runs, 13ms of delay for the mains gave me the best result -- least amount of cancellation and most boost around xo -- not only for the main listening position, but also in all seats of my couch area.
REW's 'Overlays' module is extremely helpful in making comparisons between individual measurements in one window:
Some people have less exacting standards when it comes to what's acceptable GD for subs...
That spike in delay around 130Hz or so can be seen in the wavelet as well:
Just the speaker:
Just the sub:
That odd gap certainly was not from the sub/caused by it...
Overlay of two plots vs all drivers summed:
So let's eliminate the room out of the equation using nearfield measurements.
Zoomed in using Linear (% peak) amplitude instead:
Makes a whole lot more sense than the 20-25ms I was seeing (and probably hearing?) at the MLP.
Last edited:
This time alignment thread got me thinking and playing around with the speakers' delay. I have a question specific to REW though:
After a measurement, what does the 'Estimate IR delay' option do exactly? I see a value generated like ~3ms, is that something I can 'apply' in equalizer apo, to delay say my left channel?
After a measurement, what does the 'Estimate IR delay' option do exactly? I see a value generated like ~3ms, is that something I can 'apply' in equalizer apo, to delay say my left channel?
ernestcarl
Major Contributor
Yes. One of the impulses is arriving late. You can move yourself (I guess for a near-perfect stereo image), the drivers, or the mic a bit to get it zeroed in exactly. It’s better to have it aligned down to zero when you’re summing multiple channels.
dasdoing
Major Contributor
So what happens if you add another 12ms delay to the mains to let that excess group delay kiss the 0 line at 40Hz?
I did a measurement of the subwoofer alone (mic at my listening position), before and after EQ. After eq, when I click on estimate IR delay, I got ~-17ms. Before eq, I got around -0.04ms. The subwoofer is right below my desk, ~15cm away from me.
What does the -17ms value mean? How do I use this value?
This is my rew measurement (The measurements are done at 2 different Windows volume, because it's too loud unequalized): https://app.box.com/s/6p5mmci1xgnjtf3cox7sratszhqq6mi9
These are my subwoofer eq filters:
Subwoofer before IR delay shift
After Estimation and IR delay shift. So what's the significance of this?
What does the -17ms value mean? How do I use this value?
This is my rew measurement (The measurements are done at 2 different Windows volume, because it's too loud unequalized): https://app.box.com/s/6p5mmci1xgnjtf3cox7sratszhqq6mi9
These are my subwoofer eq filters:
Subwoofer before IR delay shift
After Estimation and IR delay shift. So what's the significance of this?
Last edited:
ernestcarl
Major Contributor
27ms works better around that time delay area -- but even that ain't perfect. Why? A small bit of 'suck out' develops around the crossover region.
Phase looks better, but that ain't everything...
My old delay setting is still consistently better. I've just about tried everything in between...
You can sort of see why in the wavelets
That sharp bend is caused by adding even more delay and this forms a discontinuity in the peak energy (or rather it looks stretched out) around that 60Hz region.
Also it doesn't sound good. The 'boom' or bass thumps in between between 30-60Hz is arriving too early in relation to a lot of the stuff above now -- and it can sound smeared. I'd much rather have it a little late.
Similar threads
