Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Frank Koenig]

A great article on sub alignment IMO...
http://www.excelsior-audio.com/Publications/AES_NAMM_2018_Subwoofer_Alignment_with_Full-Range_Systems,_rev02.pdf

Well it's an interesting approach. At the end of the day it sounds like all he's doing is finding a delay value that is likely to keep the phase difference within 75 deg (2 ms @ 100 Hz) across the venue, which is an admirable goal. His suggestion, however, of using lots of overlap sounds like a recipe for mud when using non-collocated subs and tops. I prefer a little bit of a hole, if anything. Furthermore, while initial arrival may get you through a field alignment with a given set of speakers and crossovers, it's pretty useless for designing a crossover from scratch. For that I prefer the frequency domain with a time-frequency-energy (tone-burst) verification. It's a big subject. --Frank

[Mark Wilkinson]

Well it's an interesting approach. At the end of the day it sounds like all he's doing is finding a delay value that is likely to keep the phase difference within 75 deg (2 ms @ 100 Hz) across the venue, which is an admirable goal. His suggestion, however, of using lots of overlap sounds like a recipe for mud when using non-collocated subs and tops. I prefer a little bit of a hole, if anything. Furthermore, while initial arrival may get you through a field alignment with a given set of speakers and crossovers, it's pretty useless for designing a crossover from scratch. For that I prefer the frequency domain with a time-frequency-energy (tone-burst) verification. It's a big subject. --Frank

Hi Frank, let me set aside aside his technique for determining an optimized delay for non-collated subs and mains.....still need to think through all that.

But, I am 100% convinced perfect phase alignment, and an accompanying perfect impulse response, comes from initial waveform alignment, plain and simple.

That dawning has cleared up so much mud in my mind it's hard to describe...
Waiting for the mental train wreck when it unravels, Lol

[Frank Koenig]

But, I am 100% convinced perfect phase alignment, and an accompanying perfect impulse response, comes from initial waveform alignment, plain and simple.

Mark, no question that the impulse response begins when the first energy arrives. Anything else would violate causality. The peak occurs sometime later depending, generally, on the upper cutoff frequency of the system. In theory, time- and frequency-domain representations are equivalent provided they both have infinite extent. I'm with Ivan though in believing that looking for overlapping phase curves is the easiest and most reliable way to achieve time alignment in the vicinity of the crossover. In any case, it's always a compromise given the crazy response of most speakers and the constraints of the filters used. And sometimes it's not clear, to me at least, which of several alternatives that achieve reasonable time alignment is the best. It may depend on the situation or what sonic quality is most desired.

As an example of the messiness of it all, here is the phase plot and the tone burst response at crossover frequency for a crossover I recently worked on. Those tone bursts are pretty representative of musical signals. Look how the apparent frequency of the carrier varies although this is a completely linear representation. (The acoustic responses on which these are based are an average of multiple pseudo-anechoic measurements made over a range angles.) --Frank

[Mac Kerr]

It depends on which direction you stir first as to how long it takes to get to the Same Old Place (obscure Fire Sign Theater reference).

You must mean the old Same place. It's out back.

Mac

[Tim Tyler]

You must mean the old Same place. It's out back.

Mac

...here's the key...

[Dave Garoutte]

Regnad Kcin.

[Luke Geis]

Thought of from a different perspective, the group delay of the subs is behind in time as much as the mains are ahead in time. The filter network advances the mains ahead in time by half the group delay and retards the lows by half the group delay. Since we can only measure from initial energy, the total group delay adds up to the sum of all the filters total time difference. I.E. A 2 pole filter advances the hi's by 90* and retards the lows by 90* for a total group delay of 180* acoustically speaking.

Assuming the physical location of the acoustic centers of the mains the subs were exactly aligned at 1 superposition, the subs would lag in time behind the mains by some amount which is frequency and filter type dependent. A 90* phase shift @ 80hz. would be roughly 3ms. and 180* of phase shift would be roughly 6ms at sea level. Electrically the time is different than the acoustic outcome. This is because the physical superposition of each driver is not at the exact same point and the superposition of the listener can only be 100% in the middle of the two points on only 1 of the 3 possible planes at any given point in time. If the listener's point of incident between the two drivers moves, then so does the phase and delay time.

All this means is that you can only ever get the phase relation 100% correct at only 1 superposition. Anywhere else will be wrong by some amount. It is prescribed that anything within 1/4 of a wavelength will effectively couple which then means that you need to be phase accurate to within 90*  in order to effectively work well enough. 90* of phase at 80hz is roughly 3ms in relationship to time, and most well-trained engineers can determine at least that much difference in the field ( the very best are sensitive to 1ms. or less ). BUTTTTTTTTTT we are blind to phase shift when the sources of sound emanate from two distinct locations ( within reason of course ). When two different drivers are producing similar frequencies we can easily hear the phase difference between them. This is because the phase shift and amplitude of detection exist over a relatively large region. Once we get beyond about a 3-6db difference in amplitude between two different sources we can't really hear the sound of one driver combined with the other anymore. Majority rules in this case and since we hear one driver doing one thing and the other something else, the brain simply correlates that as being the way it is. It is the small region where both drivers meet at relatively the same SPL that we can sort of hear some funny stuff, but with a crossover, that region should be relatively small enough that even just being slightly off-axis should throw credible information out the window. Or simply put, the phase shift from our point of reference was already far enough out that any change made is too slight to make a definitive distinction. The summation region of a low and hi pass filter should have a fairly small sonic region in which they fully couple and outside of that region should be distinct from each other acoustically speaking. This is why having a " small hole " in the crossover region electrically and acoustically can actually help a lot.

Since we can't really detect the phase shift between two distinct devices, having the small hole around the crossover point helps keep it that way; rather undetectable. Sonically speaking a small dip in frequency response is not a deal breaker especially when utilizing a system where one half of the PA is truly independent of the other ( such as subs on an aux ). A 90* phase shift even around 100hz is not too hard to determine and correct even using guesstimation or basic math. 100hz is = to roughly 10' and 90* of phase shift at 10' is roughly 2.5ms. If your mains are essentially in line with the subs and you are using a 24db 4 pole filter ( 360* of phase shift ) a polarity inversion of either the subs or the mains will typically do the trick. 2.5ms is roughly 2.5' physically speaking, so this is the margin of error in physical location difference between the drivers you can get away with. With most subs that have a pole mount on them, the speaker is placed close enough to the physical centerline of the subs that a basic polarity shift ( assuming a 24db filter as is most common these days ) will bring the mains and subs close enough inline to be acceptable. Again since we can't really hear the phase shift even though the subs are lagging by 1 full cycle, we don't really notice it.

[Kenny Phillips]

Thank you so much for all the enlightening answers.  I now have some homework and testing to do to help me wrap my head around the concepts you guys have brought up.  Understanding why something is true is very helpful to me. 

Kenny

[Mark Wilkinson]

I'm with Ivan though in believing that looking for overlapping phase curves is the easiest and most reliable way to achieve time alignment in the vicinity of the crossover. In any case, it's always a compromise given the crazy response of most speakers and the constraints of the filters used. And sometimes it's not clear, to me at least, which of several alternatives that achieve reasonable time alignment is the best. It may depend on the situation or what sonic quality is most desired.

Yes, I too believe overlapping phase curves is the easiest and most reliable way to align.  I imagine anyone who has spent much time with Smaart or any other dual FFT comes to this conclusion also.
I guess the point of my linking to Charlie Hughes' paper was to help provide an understanding of what overlapping phase curves really mean, particularly in terms of what overlapping phase curves mean as far as an impulse response. I hadn't even read the rest of his paper about optimizing versus placement and distance.

I believe the very confusing subject of phase and time alignment gets greatly simplified, and much easier to see intuitively, when it is recognized that overlapping phase curves are the means for achieving the closest synchronization possible of initial rise times for all frequencies, for the given system under alignment.

I think the mere graphical image of an impulse response leads folks to falsely intuit that overlapping phase curves align the peak energy of all frequencies.  Or at least that's the false path I had to learn wasn't correct, and I see evidence in many threads that others are hung there too.
So just trying to help a little here...

It's a crying shame we can't learn about phase, time, crossovers, etc under linear phase first.....and then move to the much more complicated world of traditional IIR.  It would make learning so much easier IMHO.

[Tim McCulloch]

Thought of from a different perspective, the group delay of the subs is behind in time as much as the mains are ahead in time. The filter network advances the mains ahead in time by half the group delay and retards the lows by half the group delay.

Advances time?  Really?  Is this a Behringer product, the "Super Speeder-Upper Pro?

Where where you when I needed lotto numbers yesterday?

[ProSoundWeb Community]