The importance of time alignment for subwoofers

[edd9000]

Speakers are not linear phase, so they are not "time aligned" (actually, Danley unity horns can be)

With passive (or IIR active or digital) crossovers you aim to at least have the crossover region where the drivers over lap in phase, with short wavelengths placement of the drivers can be critical for even response, especially in the vertical axis.

With subwoofers the group delay can be rather large, so physical time alignment will only be over one range, and it depends on the mains bass tuning etc too.

But there is also the fact, with 20 feet wave lengths, does a foot of delay matter. With such long wave lengths and a modal room, does delay/alignment matter. No

Much better to just position for the best room behaviour.

 

[Fitzcaraldo215]

Speakers have group delay.

Yes, I understand that. But, I doubt that a single driver sub has group delay that varies much over its intended 2 octave frequency range. Also, it is entirely possible via measurements that any delay differences in a sub/satellite setup can be adjusted to be minimal.

 

[edd9000]

Yes, I understand that. But, I doubt that a single driver sub has group delay that varies much over its intended 2 octave frequency range. Also, it is entirely possible via measurements that any delay differences in a sub/satellite setup can be adjusted to be minimal.

It can be up to 100ms 20-100hz, not insignificant. Sub/sat with a crossover and delays set on a receiver is fine. But I thought we were talking stereo with no delays and no extra crossover on the mains.

Edit: To clarify my stance, I don't think you can "time align" and it doesn't matter even if you could. Compared to room behaviour, distortion, and all those other things its nothing to worry about.

 

[DonH56]

Hmmm... I sort of skimmed this but not sure I got it all, and at the risk of incurring the wrath of Thomas am not quite sure what to make of this. For myself, time and phase go hand-in-hand since simple calculations get you one from the other. Regarding alignment, my thoughts match everything I have read and experienced myself over the decades:

1. Alignment is most critical at the crossover frequency so that the sound waves from the mains and subs strike the listener's ears in phase to avoid a peak or large null. I consider this phase alignment though it is readily accomplished by playing with the distance/delay time adjustment in your system. Getting sub(s) and mains in phase at the listening position at the crossover frequency is the easiest thing to do but only the start...
2. Flat frequency response (FR) requires that time or phase alignment, natch, so I am not sure how you would separate the two. One caveat is that you could delay a full wavelength, or multiple wavelengths, at the target frequency and obtain good FR for steady-state inputs but presumably corrupt the impulse response. The solution is to use REW or whatever and check both frequency response and impulse response. One of the issues with most FR tools is that they only report magnitude and not phase (REW will do both).
3. At 20 Hz a wavelength is about 56 feet, dropping to 14 feet at 80 Hz (typical crossover to sub) and 9 feet at 120 Hz. The critical distance is about 1/4 wave so divide all those by four and you can see why phase/delay matters much more at higher frequencies than down around 10~20 Hz. Room modes, reflections, SBIR, etc. all play into this, natch, by determining the frequencies at which your room will boost and cancel frequencies at the listening position (and from which speakers).
4. Crossovers are not brick walls. Set to 80 Hz, the sub is probably still putting out appreciable sound at 100 Hz and above. The higher the crossover, the easier it is to localize the sub, and the more likely it is to smear the image if the alignment is not perfect. That can be very difficult without a powerful room correction program or the equivalent since things change rapidly around the crossover frequency and most likely the sub and mains are changing with the same slope and such. That means frequency-dependent filters to correct the alignment over frequency so sub(s) and mains continue to work together at the MLP (main listening position) well above and below the crossover frequency.
5. There have been papers discussing if absolute polarity matters. That is another aspect of this while alignment issue and one that I am not sure is completely resolved. For me I not sure it matters as the polarity of the source is usually unknown and/or questionable. For a kick drum if your head is next to the front the pressure waves are well-defined, but out in the audience? What about a kettle drum? Etc. I tend to think getting the magnitude right is more important and thus focus more on frequency response and sometimes impulse response.

Still not sure this is relevant to Thomas' question but onwards - Don

 

[dallasjustice]

Thomas,
I will assume we are talking about frequencies below 80hz.

Time domain is something that is frequently discussed when integrating subwoofers. The subwoofer's proximity to R/L speaker has nothing to do with the system's time alignment.

There is no subwoofer which can time align itself. That's impossible. If you really are interested in time alignment, you must use a multi-way DSP controlled system. To do this, you'll need:
1. microphone and DAC/ADC
2. REW
3. acourate or audiolense (DIRAC isn't very useful at integrating subwoofers)
4. a convolver

When it's done right, the result is astounding. IMO, it's impossible to integrate subs well without at least a cross-over of some sort. I know your subs may have a crossover knob on them. That's not what I'm talking about. You will need the R/L speakers to be crossed over to the subs. There are many ways to do it. IMO, digital FIR crossovers are superior to anything else.

There are also many acceptable subwoofer configurations. There's no "best" or "right" way to set them up, because all rooms are built differently. However, Welti has shown that mono subs produce a smoother frequency response vs. stereo subs. IME, this is correct.

 

[dallasjustice]

Btw, I am going to do a new subwoofer tear down and setup article soon. I'm in the middle of setting up a new subwoofer array. I am just waiting on my new Audiolense license to get to work. I hope I can put something up here soon.

 

[fas42]

To be completely non helpful, I have yet to hear a system with a subwoofer where anything was gained by having these beasts - if the main speakers were able to go down to 50 or 60Hz. Music which specifically relies on extremely low frequency throbbing as part of the effect, the 'hit', would be a different matter - but more conventional compositions, sound making? I can't see it ...

The chap up the road using DEQX had deep bass aligned to as good as DSP would give one - didn't do an ounce of good as far getting the grandeur of "big" pipe organ playing, though.

 

[mitchco]

@Thomas savage - I have read Bob McCarthy's book on this subject and agree. However, modern software DSP is very powerful and can overcome the limitations that Bob has indicated, especially in the home environment, instead of large physical spaces with many drivers. as he is discussing.

For example, here is a view of the step response of the individual drivers in my 3 way system, including sub. The first positive peak, is the tweeter (i.e. compression driver arriving first). Then shortly after, is the positive going step of the mid bass speaker, followed later by the negative going subwoofer. None are time aligned. The mid bass speaker is about .71 ms behind the tweeter and the sub is a about 3 ms behind the tweeter:

Here is the same setup, but time aligned at the listening position. Note the step response is almost a textbook result of what it should look like. In this particular measurement, the top end is rolled off a bit too much as I am in a bare room at the moment (walls just got painted), but one can see the direct sound is arriving at my ears, all at the same time, a.k.a time aligned:

The chart below is showing the frequency and phase response at the listening position from 200 Hz on down using 1/6 octave smoothing:

The chart below shows group delay, but a small amount based on the actual roll off of the speaker. Note no non-minimum phase issues at the listening position:

I wrote a detailed article on Audiolense that goes into more detail than I can here: https://www.computeraudiophile.com/...nd-room-correction-software-walkthrough-r682/
Note that non-minimum phase low frequency reflections can be removed without affecting frequency response. This is not at just the one mic location, but in my case, I measured across my seating area and the time alignment, flat fr, phase and GD still holds true. Those details are in my ebook in my sig.

As mentioned, another article will be published in the New Year going through the deets of time aligning a sub with mains. From a listening perspective, to me, it is the difference between mushy sounding bass (non-time aligned) to bottom end that has transient impact at ones ears. It is both measurable and audible - and a real treat to hear!

 

[Thomas savage]

Hmmm... I sort of skimmed this but not sure I got it all, and at the risk of incurring the wrath of Thomas am not quite sure what to make of this. For myself, time and phase go hand-in-hand since simple calculations get you one from the other. Regarding alignment, my thoughts match everything I have read and experienced myself over the decades:

1. Alignment is most critical at the crossover frequency so that the sound waves from the mains and subs strike the listener's ears in phase to avoid a peak or large null. I consider this phase alignment though it is readily accomplished by playing with the distance/delay time adjustment in your system. Getting sub(s) and mains in phase at the listening position at the crossover frequency is the easiest thing to do but only the start...
2. Flat frequency response (FR) requires that time or phase alignment, natch, so I am not sure how you would separate the two. One caveat is that you could delay a full wavelength, or multiple wavelengths, at the target frequency and obtain good FR for steady-state inputs but presumably corrupt the impulse response. The solution is to use REW or whatever and check both frequency response and impulse response. One of the issues with most FR tools is that they only report magnitude and not phase (REW will do both).
3. At 20 Hz a wavelength is about 56 feet, dropping to 14 feet at 80 Hz (typical crossover to sub) and 9 feet at 120 Hz. The critical distance is about 1/4 wave so divide all those by four and you can see why phase/delay matters much more at higher frequencies than down around 10~20 Hz. Room modes, reflections, SBIR, etc. all play into this, natch, by determining the frequencies at which your room will boost and cancel frequencies at the listening position (and from which speakers).
4. Crossovers are not brick walls. Set to 80 Hz, the sub is probably still putting out appreciable sound at 100 Hz and above. The higher the crossover, the easier it is to localize the sub, and the more likely it is to smear the image if the alignment is not perfect. That can be very difficult without a powerful room correction program or the equivalent since things change rapidly around the crossover frequency and most likely the sub and mains are changing with the same slope and such. That means frequency-dependent filters to correct the alignment over frequency so sub(s) and mains continue to work together at the MLP (main listening position) well above and below the crossover frequency.
5. There have been papers discussing if absolute polarity matters. That is another aspect of this while alignment issue and one that I am not sure is completely resolved. For me I not sure it matters as the polarity of the source is usually unknown and/or questionable. For a kick drum if your head is next to the front the pressure waves are well-defined, but out in the audience? What about a kettle drum? Etc. I tend to think getting the magnitude right is more important and thus focus more on frequency response and sometimes impulse response.

Still not sure this is relevant to Thomas' question but onwards - Don

That’s perfect don, I just wanted to nail down what was happening and have a discussion on the issues surround it.. I know the ultimate solution is multiple subs in a configuration like @dallasjustice but in my relatively small space and L shaped room that’s probably not a option for me.

But the thread was about the issue not about me. I guess there’s a difference well I knew there was a difference between the approach to sub setup of the average guy/the odd person I might know who put it in the corner and play with the knobs on the back of the sub, maybe they look at frequency response maybe not and the way more technical minded folk like yourselves approach it.

I just wanted to shine a light on the importance of more than just frequency response, there’s more to consider and time/phase integration within the rest of the system seemed to be largely ignored and have you all ( if intrested) dig into and discuss it as I’m sure there’s intresting things of technical merit here to chew over.

The thread started ok, then got turned into “ Thomas needs help with subs” rather than “Thomas stumbled into comb filtering and phase issues while playing with subs let’s talk about those issues , explain what they are “ .

I should of just put up the thread heading and left a blank space lol

 

[Thomas savage]

@Thomas savage - I have read Bob McCarthy's book on this subject and agree. However, modern software DSP is very powerful and can overcome the limitations that Bob has indicated, especially in the home environment, instead of large physical spaces with many drivers. as he is discussing.

For example, here is a view of the step response of the individual drivers in my 3 way system, including sub. The first positive peak, is the tweeter (i.e. compression driver arriving first). Then shortly after, is the positive going step of the mid bass speaker, followed later by the negative going subwoofer. None are time aligned. The mid bass speaker is about .71 ms behind the tweeter and the sub is a about 3 ms behind the tweeter:


View attachment 9815


Here is the same setup, but time aligned at the listening position. Note the step response is almost a textbook result of what it should look like. In this particular measurement, the top end is rolled off a bit too much as I am in a bare room at the moment (walls just got painted), but one can see the direct sound is arriving at my ears, all at the same time, a.k.a time aligned:

View attachment 9816

The chart below is showing the frequency and phase response at the listening position from 200 Hz on down using 1/6 octave smoothing:


View attachment 9817

The chart below shows group delay, but a small amount based on the actual roll off of the speaker. Note no non-minimum phase issues at the listening position:

View attachment 9818

I wrote a detailed article on Audiolense that goes into more detail than I can here: https://www.computeraudiophile.com/...nd-room-correction-software-walkthrough-r682/
Note that non-minimum phase low frequency reflections can be removed without affecting frequency response. This is not at just the one mic location, but in my case, I measured across my seating area and the time alignment, flat fr, phase and GD still holds true. Those details are in my ebook in my sig.

As mentioned, another article will be published in the New Year going through the deets of time aligning a sub with mains. From a listening perspective, to me, it is the difference between mushy sounding bass (non-time aligned) to bottom end that has transient impact at ones ears. It is both measurable and audible - and a real treat to hear!

Thanks Mitch this is fantastic and I really appreciate your time putting this up.

 

[Thomas savage]

Thomas,
I will assume we are talking about frequencies below 80hz.

Time domain is something that is frequently discussed when integrating subwoofers. The subwoofer's proximity to R/L speaker has nothing to do with the system's time alignment.

There is no subwoofer which can time align itself. That's impossible. If you really are interested in time alignment, you must use a multi-way DSP controlled system. To do this, you'll need:
1. microphone and DAC/ADC
2. REW
3. acourate or audiolense (DIRAC isn't very useful at integrating subwoofers)
4. a convolver

When it's done right, the result is astounding. IMO, it's impossible to integrate subs well without at least a cross-over of some sort. I know your subs may have a crossover knob on them. That's not what I'm talking about. You will need the R/L speakers to be crossed over to the subs. There are many ways to do it. IMO, digital FIR crossovers are superior to anything else.

There are also many acceptable subwoofer configurations. There's no "best" or "right" way to set them up, because all rooms are built differently. However, Welti has shown that mono subs produce a smoother frequency response vs. stereo subs. IME, this is correct.

Thanks Michael, Iv followed your various setups as much as you’ve detailed them publicly anyhow and I know you’ve spent considerable time and effort learning about all the issues involved plus the various software/hardware setups needed to gain the best results as you see it.

I’d be really intrested to read the tear down of your latest setup and discover the various practical issues and solutions you’ve found along the way and that would be a great asset for the forum.

For me personally I need to relate theory ( words on a page) with a ‘working’ physical reality I can expirence first hand in order for anything to stick up in my brain. I’m a bit odd that way, best way I can describe it is if you imagen you’ve never tied shoelaces before, no amount of reading about it would be enough to learn.. well I’m like that but with everything ( not that weird as that’s how our brain and senses learn to form the reality we expirence in the early stages of our life’s ..., apparently I can’t remember that far back)

If I spent a hour or two with a setup like yours and could physically relate all the theory and graphs with what’s in front of me then I’m fine, Iv got a world in my head to relate information to and in fact I can explore and understand a great deal, basically it all becomes relatively simple even though it’s not.

Now there’s plenty of normal people out there, folk that aren’t strange like me that would enjoy a discussion on all the things highlighted in this thread. Maybe some causal audio enthusiasts, people that have bought a sub maybe some room correction (like I did) but discover there’s more to it than meets the eye and certainly more to it than sub manufacturers let on. Plus all you guys that like to dig a little deeper and mull over various POV’s in a far more advanced way. My intention was to marry the two together, use my clumsy discovery of a issue to kick off a wider debate, a altruistic ambition maybe.. a little too ambitious likelier still

 

[Thomas savage]

This is probably a better place to continue the discussion.., ( just a suggestion, my endless explaining has rather diluted this thread)

https://audiosciencereview.com/foru...-they-add-dimension-to-music.2165/#post-58585

Thanks for all your contributions, your all generous with your time and I appreciate it.

 

[watchnerd]

5 big actives!

Subs are for wimps.
S.

I only listen to 2 channel content.

 

[Fitzcaraldo215]

There may be other issues to be sorted out, and perfect timing may be elusive. But, I still believe in setting up subs to be acoustically equidistant with main/satellite channels from the central sweet spot, even if just as a starting point for further tweaking.

For those who wish to use equal distances to all speakers including the sub, here is another consideration. Many, if not most, subs today use DSP on their input circuitry to control a number of their operating settings. That introduces considerable delay in their output relative to the main/satellite channels. And, the upshot is that physical tape measure distance to the sub understates the true acoustic distance, often considerably.

Most sub makers also do not specify exactly how much delay is involved in their subs. So, mic measurement of timing/distance is the only sure way to achieve this. Again, many Mch AVR/prepros automatically do this via mic calibration, though many expensive ones do not. In my case, Mch Dirac Live sets this acoustically measured distance automatically as part of the room EQ calibration.

That may be sufficient for many, but not exacting enough for others. @dallasjustice has suggested going further via REW measurements to better align phase/timing in the xover region relative to main channels. Mark Seaton has also suggested someting similar via mic measurements, varying the sub distance parameter in Mch setups to achieve minimal measured phase misalignment in the xover region.

As always, careful actual measurements are far more useful and a big time saver vs. just trying to do it by ear alone, as so many audiophiles attempt to do.

 

[DonH56]

Note the delay at any given frequency depends upon the crossover design and setting; it is more than a physical distance. That is likely one reason it is so hard to specify.

 

[RayDunzl]

Alignment Thought Experiment:

Here is a Kick Drum whack...

In Audacity, I "crossed" the original signal, 80Hz/24dB/octave, creating high and low pass signals - as might go to mains and sub.

Simple Filtering apparently adds a delay to the low pass (subwoofer) signal.

Recombining the low and high pass signals, as would occur in the air between the speakers and the ear, doesn't reproduce the original wave.

Full Range, 80Hz High Pass, 80Hz Low Pass, re-combined High and Low Pass, Original again for comparison.

 

[sergeauckland]

I only listen to 2 channel content.

OK then, Two big 'speakers. You can have even bigger ones of there are only two.

S

 

[Fitzcaraldo215]

Note the delay at any given frequency depends upon the crossover design and setting; it is more than a physical distance. That is likely one reason it is so hard to specify.

Don, agreed. Acoustic time/distance measurements are typically just an average over several octaves. Better time/phase integration might be achieved in the crossover region by careful adjustment and tweaking.

 

[mitchco]

Agree with Don and Ray. I have tried many approaches using REW, rePhase, HOLMImpulse and other software, including Acourate. Attached is a special procedure for Acourate... I have combed forums like REW, diyAudio, AVS, pro audio forums and much of Bob McCarthy's work. Using Rythmik's calculator, etc. All of which produce varying results, none of which produce an exact time alignment as seen in Post 48.

Check out the measured delays in Audiolense that correspond to the non-time aligned step response in post 48:

The delays for the bass (i.e. subs) are 2.75 ms

and 3.38 ms, even though I measured the distance from the wall behind the subs to be exact down to a 1/16th of an inch. Note that the sub voice coils are more or less in line with the midrange (i.e. bass cabs) of the JBL 4722's in the horizontal plane. Given that sound travels roughly a foot per millisecond, and with a 3 millisecond delay, one would think the subs were "physically" ~3 feet behind the voice coil of the bass speaker (midrange in the config above). But they are physically side by side...

Audiolense is the first and only software product I have used that time aligns subs correctly and is repeatable, with no manual intervention. More amazing is how they possibly can be aligned when I have them crossed over so low at 40 Hz!

 

[svart-hvitt]

Let me explain the way of Genelec, which is the only subwoofer I've tested in my setup (with 8351a). The Genelec way is fully standardized and software based. I will explain the digital AES-EBU setup, which is the preferred way (analog setups are possible but not of great interest if you use Genelec monitors).

Cabling is a non-issue. You can cable anyway you want using AES-EBU cables (XLR connectors). Then you start up the Genelec software. You tell the software to fix all the connected speakers; then after a short period of audio test signals, the frequency curve is cut where needed and all speakers are time corrected.

What's the result when mixing 8351a with just one smaller 10 inch 7360a sub? Well, people - myself included - were hard pressed to hear any difference between 8351a speakers on a stand alone basis compared with the setup including the 7360a 10 inch sub. The difference was quite subtle. Which it should be when everything works as intended, right? Maybe the sub offloaded the 8351a speakers somewhat resulting in a cleaner sound (less distortion due to the sub taking some of the pressure off the main speakers)? And the bass tones became a bit less physical when using the sub; the bass of the 8351a is pretty physical, maybe due to its coaxial single point design?

(I had the same experience when I listened to an engineer's 8351a with the bigger 7370a sub).

Genelec sets the sub crossover at 85 Hertz by default. Probably with good reason to avoid subs that are easy to locate? However, you're free to set the crossover at any point under 100 Hertz (Genelec subs top at 100 Hertz).

When I read this thread, it becomes clear why the paternalistic Genelec way has its appeal. The Genelec way can be copied by cutting the process into smaller "modules" or parts by very competent people like Mitchco (see his comment above), but for the 99,99 (!?) percent that don't have the right software and enough time to play around, it's a mystery that so few producers copy the Genelec way.

To Thomas: Your "problem" - or headache? - would be solved if you had Genelecs. Because that's sort of an irrelevant reply because you want to keep your existing speakers, I think the best option is to get a grip on the Audiolense software. I heard earlier this year a setup with eight 24 inch subwoofers plus a range of big mid bass woofers that was time aligned using Audiolense. And it worked magically well!