Phase-alignment based method of designing multi-way speakers

[turbodawg]

Originally Posted by speaker dave

You have to separate frequency response errors from all-pass errors. If you have an electronic crossover and create a system with perfect response at some exact delay between woofer and tweeter then advancing one unit or the other in small steps would likely be audible. This is not due to the phase or time error, but due to the large frequency response error that would be introduced.

If, on the other hand, you adjust an all-pass network (flat response but no frequency response error) then errors below 1ms are hard to hear on test signals and usually inaudible on music.

Perhaps you should read the entire thread. Better yet, the published papers on the subject.

David S.

Actually, no. I'm referencing a 3-way active system (2-way mains plus mono sub) with imperfect frequency response (minor PEQ), 1st order crossovers, and time correction values arrived at by distance measurement vs. the speed of sound. Adjustment of the time delay on any of the 6 channels is absolutely noticeable as a shift in the soundstage for the mains or loss of integration of the sub. No great noticeable change in frequency response.

Have you personally used time correction on an active system...?

[speaker dave]

You say 6 channels but you name 5.

Its not clear if you are shifting left and right elements together or just a single of the 6 channels. If you shift a single channel (say, left tweeter) then you are creating a balance shift per the precedence effect that will shift centered treble to the nearer (time wise) tweeter. This is a well know phenomenon of stereo hearing and has nothing to do with audibility of phase shift within a single source.

Either way a shift of an element of a 2 way system will clearly create a frequency response error as the usual nulling at crossover is tilting upwards or downwards with the varying time delay. You can't really claim that there isn't an audible frequency response error with this shift and that is what many documented tests have shown: fairly small response errors are audible (say 1dB over one Octave), while full millisecond delay errors without response changes are hard to hear.

And yes, I have used time correction on active systems.

David S.

[turbodawg]

Originally Posted by speaker dave

You say 6 channels but you name 5.

Its not clear if you are shifting left and right elements together or just a single of the 6 channels. If you shift a single channel (say, left tweeter) then you are creating a balance shift per the precedence effect that will shift centered treble to the nearer (time wise) tweeter. This is a well know phenomenon of stereo hearing and has nothing to do with audibility of phase shift within a single source.

Either way a shift of an element of a 2 way system will clearly create a frequency response error as the usual nulling at crossover is tilting upwards or downwards with the varying time delay. You can't really claim that there isn't an audible frequency response error with this shift and that is what many documented tests have shown: fairly small response errors are audible (say 1dB over one Octave), while full millisecond delay errors without response changes are hard to hear.

And yes, I have used time correction on active systems.

David S.

I am discussing a system with the right and left sub channels being summed to mono in the amp. So the processing is 6 channel, 3 way system, right/left, dsp crossovers and time correction on all channels. Only becomes mono at the sub amp and driver.

Once the system is time aligned properly, if you are to shift time alignment of a woofer or tweeter, you will get a quite clearly audible shift in the soundstage. The image moves, smears, and eventually just sounds wrong depending on how far you go. .1 ms changes on one channel are noticeable.

If you take both the sub channels and delay or advance them vs. mains, you will have bass that goes from perfectly integrated with the woofers, to either too soon or too late - which is also clearly audible.

Any minor frequency response changes that occur because of time shift I have never, ever noticed, and should only occur in the crossover region. However, I am not claiming there isn't. and you are presenting a straw man here.

For me, there is a major audible difference between time shifting and frequency response changes, and I am certainly not confusing the two.

In fact, I'm sitting next to not just one, but two parametric equalizers as I write this, one analog and one digital.

Here is your original quote:

Originally Posted by speaker dave

It is rubbish when tossed in as an absolute requirement without any proof.

As a diy forum there is always a lot of idealism involved here in discussing speakers and how they should perform. Since diy types don't need to make a living with their designs they can proclaim what the ideal criteria are and pursue them to the very ends.

If we want to advance the art, though, we need to really research what can and can not be heard. Even with highly audible criteria there are thresholds of audibility and we need to figure out what those are and not waste time and money pushing performance well beyond the thresholds of "good enough". (To do so generally means we are make do in other performance areas where we might not yet be to the threshold of inaudibilty.

Many studys have been done on phase audibility and there is a general consensus that minor phase errors are not audible. Of course extreme phase errors will be audible, but errors of less than a foot or a mSec in time are generally not heard except with some specialized test tones.

I realize that there is great appeal to having a speaker that will pass a square wave, conversly, to have one with poor square wave response and say "it doesn't matter" takes a lot of will power for some.

I always refer others to the listening tests of Toole. He found that a lot of speakers could sound quite fine if the designer concentrated on on-axis and near on-axis flatness and smoothness. Power response, phase response, and some distortion characteristics could be surprisingly non-ideal and not downgrade rank ordering in his tests.

Worth your reading his book.

David S.

Have you ever considered that these studies may not be applicable, and that a good percentage of speaker designers who believe this are completely wrong?

[ra7]

David,

Is the step response a good way to see proper time alignment? For example, in a 2-way horn+woofer, if the phase tracks properly at the crossover, would it result in a smooth hand off from the tweeter to the woofer, as is visible in the step of the tweeter blending nicely with that of the woofer?

A deep null at the crossover with the polarity of one driver reversed is also a good indication. Another way would be that the ripples near the crossover start reducing as the time alignment improves.

[fastbike1]

Do you have any idea who you are talking to (i.e. the background)? And so, have you considered that you may be completely wrong?

Originally Posted by turbodawg

. . . .

Have you ever considered that these studies may not be applicable, and that a good percentage of speaker designers who believe this are completely wrong?

[turbodawg]

Originally Posted by fastbike1

Do you have any idea who you are talking to (i.e. the background)? And so, have you considered that you may be completely wrong?

Nope, and don't really care either. Regardless of his background, he made a false statement.

[speaker dave]

Originally Posted by turbodawg

I am discussing a system with the right and left sub channels being summed to mono in the amp. So the processing is 6 channel, 3 way system, right/left, dsp crossovers and time correction on all channels. Only becomes mono at the sub amp and driver.

Understood. Be aware that any mono bass may be cancelled with time shifts and that would again be an audible frequency response error.

Once the system is time aligned properly, if you are to shift time alignment of a woofer or tweeter, you will get a quite clearly audible shift in the soundstage. The image moves, smears, and eventually just sounds wrong depending on how far you go. .1 ms changes on one channel are noticeable.

As I mentioned, that is a perfect description of a balance control using time shift rather than level shift. If you delay the left tweeter the treble image shifts right. Meridian has used this approach for a number of years. It is a well known stereo effect. This is not a proof of audibility of phase shift or time delay, in the context of our discussion we are talking about single channel effects rather than stereo.

If you take both the sub channels and delay or advance them vs. mains, you will have bass that goes from perfectly integrated with the woofers, to either too soon or too late - which is also clearly audible.

Any minor frequency response changes that occur because of time shift I have never, ever noticed, and should only occur in the crossover region. However, I am not claiming there isn't. and you are presenting a straw man here.

You can't actually claim this. You can say you hear a difference and you assume that it is delay related rather than response related, but since there will be a response error caused by all the scenarios you describe you can only assert that response isn't the cause. That is why any proper academic study would (has) separated phase from amplitude response by testing the audibility of all pass networks.

For me, there is a major audible difference between time shifting and frequency response changes, and I am certainly not confusing the two.

In fact, I'm sitting next to not just one, but two parametric equalizers as I write this, one analog and one digital.

Impressive!

Here is your original quote:

Have you ever considered that these studies may not be applicable, and that a good percentage of speaker designers who believe this are completely wrong?

No, not really. I stand by both quotes.

Regards,
David

[speaker dave]

Originally Posted by ra7

David,

Is the step response a good way to see proper time alignment? For example, in a 2-way horn+woofer, if the phase tracks properly at the crossover, would it result in a smooth hand off from the tweeter to the woofer, as is visible in the step of the tweeter blending nicely with that of the woofer?

I have dabbled with that once, taking an Advent system and reversing the phase of the tweeter, then angling the system backwards until a square pulse looked pretty decent. (I think this predated the Spica system that was time alligned that way)

It worked pretty well as I recall but it is generally hard to optimize a step response and get a flat response, etc., since in the time domain it is hard to infer much about response beyond the broad group delay aspect.

It is also possible to have a multiway with alligned impulses and poor phase blend at crossover. Remember that the peak energy of the impulse response tells more about group delay in the center of the passband. Little about the band edge.

David

[turbodawg]

Originally Posted by speaker dave

As I mentioned, that is a perfect description of a balance control using time shift rather than level shift. If you delay the left tweeter the treble image shifts right. Meridian has used this approach for a number of years. It is a well known stereo effect. This is not a proof of audibility of phase shift or time delay, in the context of our discussion we are talking about single channel effects rather than stereo.

You can't actually claim this. You can say you hear a difference and you assume that it is delay related rather than response related, but since there will be a response error caused by all the scenarios you describe you can only assert that response isn't the cause. That is why any proper academic study would (has) separated phase from amplitude response by testing the audibility of all pass networks.

Thanks for clearing up the stereo vs. mono point. I mainly observe soundstage shifts with the mains, as you say. Questions...

- Shift the time delay of the sub that is corner loaded, crossed at 80 hz / 1st order, from what is an audibly and distance calculated correct setting, by say, .5 ms ahead of the mains. The sub goes from being audibly invisible, to sounding out of sync, sped up. Or laggy, if you were to delay it the same amount. That's equivalent to a physical shift of about 6". Wavelengths at 80 hz are 14ft, so phase shift is minimal. I do understand that there is higher frequency distortion from the sub that may blend better with good time alignment. But why is a .5ms shift so audible here?

- Woofer to tweeter crossover is 3.5k, 1st order. If I shift on tweeter channel by .3ms, this results in a roughly 360 degree full phase shift at the crossover frequency. The positioning of the stereo image will be relatively unchanged, but it will sound "off", and not be as clear. Thoughts?

- Why does 1st order sound and image much much better than 4th order?

Also, I should also mention that is in an SUV, with a eclipse CD8053 head unit (dsp, preamp only), 3 vintage class a/b amps, custom door pods, morel/peerless front stage, 15" sub in the back in very well built 4 cube box. Sounds better than some $10K+ home stereo's I've heard. LOL.

[DBMandrake]

Originally Posted by ra7

Is the step response a good way to see proper time alignment? For example, in a 2-way horn+woofer, if the phase tracks properly at the crossover, would it result in a smooth hand off from the tweeter to the woofer, as is visible in the step of the tweeter blending nicely with that of the woofer?

Speaker dave touched on it but I think it bears expanding on as it's so often a point of confusion, and that is time alignment vs phase tracking, they are not one and the same, but two different things.

Time alignment refers to the inter driver time delay in the centre of their respective passbands, (away from the crossover frequencies) which is controlled by the acoustic centre offset, while phase tracking is always considered in the crossover overlap region where both drivers have significant output, and is dependent on several factors including the acoustic centre offsets, filter slopes, and the drivers own response curves.

It's possible as dave points out to have a design where the acoustic centres are aligned (at least on one axis) but phase tracking is poor or non existant. It's also possible to have a design where acoustic centres are not aligned but with the right crossovers phase can track well in the crossover overlap region. (many speakers fall in this category)

Finally you can also have a design where acoustic centres are aligned and phase tracks through the overlap region.

Trying to interpret the impulse response to check time alignment of drivers is not going to get you far because you can't visually separate the effects of different time delays at different frequencies...it's crude at best and inaccurate at worst.

The best way to measure physical time alignment of drivers IMO is excess group delay, where you can directly compare the time delay in the middle of each drivers passband. Looking at excess group delay instead of group delay eliminates the minimum phase parts of the group delay that come from amplitude response errors of each driver which otherwise tend to dwarf and obscure the physical delay.

Using this technique you can measure the relative acoustic centre offset of two drivers down to just a few millimetres with a single measurement.

A deep null at the crossover with the polarity of one driver reversed is also a good indication. Another way would be that the ripples near the crossover start reducing as the time alignment improves.

A deep null at the crossover frequency when you reverse one of the drivers tells you nothing about time alignment of the drivers, it only tests phase tracking.