The first is yeah moving say the tweeter physically back so its voice coil is above the woofers', or moving a horn backward or forward for the same effect. The second is where the baffle is sloped/angled back to essentially align the voice coils to the listening axis while keeping the front baffle flat without steps.
I feel skeptical these ideas can really work automatically and sufficiently, since they ignores the actual driver time response and the effects of the crossovers. But it's not an area I worked in a lot since by the time my career progressed to where I had tools to measure such I was in autosound where that was the least of my worries.
By the way here "time alignment" means the initial arrival sound from each drivers come at the same time. I do NOT mean "phase" at all, that's a different headache...
I feel skeptical these ideas can really work automatically and sufficiently, since they ignores the actual driver time response and the effects of the crossovers. But it's not an area I worked in a lot since by the time my career progressed to where I had tools to measure such I was in autosound where that was the least of my worries.
By the way here "time alignment" means the initial arrival sound from each drivers come at the same time. I do NOT mean "phase" at all, that's a different headache...
Short answer to the question in the title: no.
As you mentioned in your OP, the crossover filters often are the largest source of delay and contribute most to alignment problems.
Trying to make a linear phase loudspeaker is a silly goal IMO. The ear+brain displays certain limits to misalignment below which the effect is inaudible. Less so on synthetic signals via headphones and more so when a loudspeaker is reproducing music type signals in a real room.
As you mentioned in your OP, the crossover filters often are the largest source of delay and contribute most to alignment problems.
Trying to make a linear phase loudspeaker is a silly goal IMO. The ear+brain displays certain limits to misalignment below which the effect is inaudible. Less so on synthetic signals via headphones and more so when a loudspeaker is reproducing music type signals in a real room.
The voice coil position is irrelevant. What's relevant is the acoustic center of a driver. For a flat diaphragm this is trivial, for a cone or dome less so, notable with non-rigid cones/domes. As the important range is upper end of the intended passband the acoustic center of woofers tends to be closer to the cone-to-dustcap junction.
The idea of time-aligned centers is that you then can "blindly" rely on minimum-phase acoustic target functions without risk of delay/phase issues. In practise, it is not of much help, though (unless the difference is really huge, which then affects off-axis response even more than it is already). Set up your proper acoustical target functions (in magnitude, and more importantly, phase) and EQ you drivers until you get there...
The idea of time-aligned centers is that you then can "blindly" rely on minimum-phase acoustic target functions without risk of delay/phase issues. In practise, it is not of much help, though (unless the difference is really huge, which then affects off-axis response even more than it is already). Set up your proper acoustical target functions (in magnitude, and more importantly, phase) and EQ you drivers until you get there...
They are more the same thing than some realise. Delay is a rate of phase change. Phase doesn't mean that two sounds start at the same time at different parts of the cycle.
Well yes, delay causes phase wrap. What I meant was I was particularly regarding the arrival times, and not "what happens after that" ha ha. Jeez I'm flashing back to trying to cross over automotive coaxial with cheap capacitors only (due to cost and also physical space) in inexpensive speakers. There was inevitably a problem where the drivers would go in-out-in phase. It simply was not fixable without a significantly upgraded crossover.
These are 1965 problems. Buy a proper multi-channel plate amp with integrated digital crossover and adjustable delays and enjoy modern equipment. I configured a Hypex Fusion Fa503 for a friends Mageplanar 3.7i this week and it was just great. Measure each driver. Configure the crossover filters, notch out a resonance or two and adjust the delays to produce a nice flat response. Enjoy.
I did a direct ab comparison between a slope baffle and 16mm step baffle. Same speaker same everything. The step baffle sounded less edgy or smoother with no loss if dynamics. Never be afraid to give things a try regardless of what the theory supposedly tells us. Enjoy. Diy is about giving ideas a go.
Again, phase means they do start late. It doesn't mean they go in an out of phase if you combine it with a delay, phase and delay are related.
Here is an example of phase actually being delayed despite there being no added delay..
Last edited:
Is there a limit in the cut-off to fix a phase problem in the passband with an assymetric XO ?
With passive, the way to do is ? : try to phisically the Z delay at the targeted cut-off ; then try to correct the rest of passband phase missalignement by asymetric XO ?
If the magnitude is too much affected by the assymetric when you check with listening, then choose your trade off (voicing?) ?
With passive, the way to do is ? : try to phisically the Z delay at the targeted cut-off ; then try to correct the rest of passband phase missalignement by asymetric XO ?
If the magnitude is too much affected by the assymetric when you check with listening, then choose your trade off (voicing?) ?
When doing a thorough passive, you have the relative magnitudes to get to spec (sometimes you have a range available), and also the relative phase. After this you have global options for EQ. In the past I have done it using more extreme options like first order and fifth order, and have managed to retain proper control over the result.
You quited the first slope gang or you use it time to time yet ?
That's something I want to try with a 5" and 1" horned as I am not listening to loud levels those days. In mind, better transcient (myth?). While my actual is a 2 ways Bessel high pass on the mid and two serie second orders on the tweeter (a notch then the cut-off) and I have not the feeling of a lack of transcient...
Still have to find general litterature simple enough to understand the pors and cons of the op question (that pop up poften here at Diya) in order to choose a trade-off, mostly the poor and cons of the different slopes and transfer functions (Butt, bessel, LR, ...) for the off axis in order to have a good soundstage...
It is in the lines of Diya, but not easy to sum it whithout experiment it in real (I am not in the tons of saw dust and drivers purchase per year).
I stopped in the step of trying to make a good group delay enough in the bass and try to rule the whole magnitude as I can... 🙂
That's something I want to try with a 5" and 1" horned as I am not listening to loud levels those days. In mind, better transcient (myth?). While my actual is a 2 ways Bessel high pass on the mid and two serie second orders on the tweeter (a notch then the cut-off) and I have not the feeling of a lack of transcient...
Still have to find general litterature simple enough to understand the pors and cons of the op question (that pop up poften here at Diya) in order to choose a trade-off, mostly the poor and cons of the different slopes and transfer functions (Butt, bessel, LR, ...) for the off axis in order to have a good soundstage...
It is in the lines of Diya, but not easy to sum it whithout experiment it in real (I am not in the tons of saw dust and drivers purchase per year).
I stopped in the step of trying to make a good group delay enough in the bass and try to rule the whole magnitude as I can... 🙂
Last edited:
You might see whether the difference in distance is within the limits for group delay audibility. Then make sure you have a great enough slope on the one you aren't delaying, I mean sure you're looking for a difference between the orders but it's no good if your tweeter is fed too much bass or your woofer is allowed to breakup.
yeah,I see those times the fashion is to make a little lower mid and tweeter cut-off, often below 2k hz . While it is not always because the woof can not climb as high tha 3k hz with some 3" to 5" mid units... People seems to like the sound of the tweeter rulling early in frequencies...
Your advert about magnitude in relation about the two driver units is a good reminder...
Your advert about magnitude in relation about the two driver units is a good reminder...
No, it does not works automatically. It works only with symmetrical first order (acoustical) crossover and only with the correct baffle angle slope and center-to-center driver distance - to get acoustical origins form both drivers on the same vertical plane. In my opinion - too much trouble and too little benefits.
On the other hand, Time-Aligned coaxial 2-way loudspeaker - now we are talking...
We're off into semantics here. My use of "phase" was from another thread elsewhere, in a more colloquial or newbie/ignorant way, focused on are the drivers in/out of phase. Your screenshot is a good illustration and reminder that initial delay (from crossover and/or steps and slopes) can bring drivers "in or out of phase" and move crossover lobing up or down.
Hmmm now I'm wondering about the automotive crossover problems I referenced earlier...ugh how much "phase" is 2" like around maybe 4k-6k Hz...half a wavelength at 4 kHz. Not a coincidence!
Now if someone only made a surround decoding processor with multiple digital outs so you could do time alignment to multichannel. (Yeah I guess you can do that with a home theater PC but one look at a forum about that brought back too many horrors of past days messing with operating systems! 😵
Cool, glad you separated them.
I like to completely separate what is correct "time alignment" from what is correct "phase alignment".
Because imo they have nothing to do with each other.
To me, 'Time alignment' means a constant time or delay across all frequencies, and as such is not frequency dependent.
Whereas 'phase alignment' means a relative relationship between frequencies, and as such is completely frequency dependent.
The fact one is frequency dependent and the other is not, means they can't be treated as substitutes for each other and need to be addressed independently.
Ime, time alignment is properly accomplished without regard to xover phase rotation / group delay etc.
I've found time alignments are best when they do nothing more than achieve the alignment of acoustic centers to make for equal time-of-flights (ToF) to a single mic location.
This can of course be accomplished either through constant time delays, or geometric designs such as sloping baffles, or moving drivers forwards/backwards.
It is a mistake ime, to bring phase into the time alignment process...and a most common one.
Tis much better to simply time align raw drivers, then go to work on xovers and phase alignments.
Again all owing to the fact constant delays are not frequency dependant, and phase is.
Juggling them together simply gives suboptimal results ime.