[musiccritic5088]

Hi, my name is James and I am new to this forum so please excuse me if I do something I’m not supposed to.

I wanted to create this thread in order to help me and anyone else trying to learn about the fine details of speakers design and I’ll start this thread off with my first set of questions (these are the types of q&a I want in this thread).

My first questions regard to crossover design (somewhat).

     In a crossover, I am aware that all components add specific delay amounts to the audio signal, with some components having little to no delay and with others having a big delay to the audio signal. My questions are as follows.

Question 1
Are there specific crossover components that I can use to solely adjust the time delay of each driver in a multi-way speaker? Components that won’t effect anything except time delay? The issue im trying to resolve is PHASE ALIGNMENT of the drivers. My concern is that once my speaker drivers are locked in place on the baffle, once I build my crossover, some of the drivers will be out of phase with the others to a certain degree. My first instinct was to create individual moveable cabinets for each driver so that even after I design my crossover I can physically time align my speakers. I’m told that the issue with this is the possibility of reflections and diffraction due to the uneven baffle surface which will cause coloration to the audio signal. With this in mind, my next instinct is to have one baffle for all the drivers, and mount each driver a certain depth into the baffle.

Question 2 (also a continuation of question 1)
The issue I’m worried about in doing this is, for the mid and high frequency drivers, how will I mount the drivers deeper into the baffle in a way that I won’t cause coloration to the drivers due to the throat I will be making that the driver will be sitting behind.. My first instinct to solve this issue is to round the corners of the cutouts I make to mount the drivers deeper into the baffle. My concern in doing this is that I will create a boost/Horn effect to the driver at a specific frequency and up due to an effect similar to baffle step loss.

Question 3
Is there a way I can mount my drivers deeper into my baffle without coloration to the sound? Is it a matter of me being unable to mount the driver past a certain depth into the baffle so that I can avoid having a throat deep enough to effect the frequency range of a specific driver? If this is the case then it’d be nearly impossible to have no coloration to the tweeter as this is the driver I will have to mount the deepest if I want to time align each driver at the horizontal axis.

Question 4
With all this in mind, is there a shape that I can make the cutouts so that the cutouts won’t add coloration to the signal?

Question 5
Is the idea behind a tractrix horn to amplify the signal without coloration of the sound?

[FullRangeMan]

Welcome  again
Starting Block is really for people to introduce themselves, as you already open a thread in the Starting Block Circle I will move this topic to the Lab Circle for you now.

[FullRangeMan]

Question 5
Is the idea behind a tractrix horn to amplify the signal without coloration of the sound?
By coloration do you mean say the duck voice?
If so its caused by the plastic horn material, you would use a real wood horn.

[musiccritic5088]

Question 5
Is the idea behind a tractrix horn to amplify the signal without coloration of the sound?
By coloration do you mean say the duck voice?
If so its caused by the plastic horn material, you would use a real wood horn.

I’m talking about the color caused by internal diffractions and reflections in the Horn. When I say coloration I am referring to a boost in a specific frequency range. For an example, when cupping you hands around your mouth you’ll notice that your voice doesn’t increase in volume evenly across the whole frequency spectrum. You’ll notice that you end up making your voice sound “midrangy”. This is the effect I’m trying to avoid.

[richidoo]

Hi, my name is James and I am new to this forum so please excuse me if I do something I’m not supposed to.

Hi James, welcome to AudioCircle!

     In a crossover, I am aware that all components add specific delay amounts to the audio signal, with some components having little to no delay and with others having a big delay to the audio signal. My questions are as follows.

Question 1
Are there specific crossover components that I can use to solely adjust the time delay of each driver in a multi-way speaker? Components that won’t effect anything except time delay? The issue im trying to resolve is PHASE ALIGNMENT of the drivers. My concern is that once my speaker drivers are locked in place on the baffle, once I build my crossover, some of the drivers will be out of phase with the others to a certain degree. My first instinct was to create individual moveable cabinets for each driver so that even after I design my crossover I can physically time align my speakers. I’m told that the issue with this is the possibility of reflections and diffraction due to the uneven baffle surface which will cause coloration to the audio signal. With this in mind, my next instinct is to have one baffle for all the drivers, and mount each driver a certain depth into the baffle.

Time alignment of the wavefront coming from different drivers is affected by their relative distance to the listeners ears, and the electrical phase shift imparted by the crossover components. Each pole added to the filter increases the phase rotation by 90 degrees. You can determine the displacement by multiplying the phase rotation angle by the wavelength.

Individually positioned separate cabinets per driver can allow time alignment, but complex and adds unnecessary edge diffraction and associated FR aberrations. Some high end brands do this for marketing purposes because it does allow very fine tuning of the front lobe.

You can embed the drivers into the box at different distance, but this can create a hard baffle edge around the driver which will cause diffraction. A waveguide is a small horn that can be used to smoothly deliver sound waves to the baffle surface while minimizing diffraction. Some waveguides also give some dispersion control.

Question 2 (also a continuation of question 1)
The issue I’m worried about in doing this is, for the mid and high frequency drivers, how will I mount the drivers deeper into the baffle in a way that I won’t cause coloration to the drivers due to the throat I will be making that the driver will be sitting behind.. My first instinct to solve this issue is to round the corners of the cutouts I make to mount the drivers deeper into the baffle. My concern in doing this is that I will create a boost/Horn effect to the driver at a specific frequency and up due to an effect similar to baffle step loss.

Yes, you are describing a waveguide. What you want is a waveguide design that gives you the depth you want for driver time alignment, but adds minimal diffraction to avoid coloring the sound. You can explore the various horn profiles to see how horn shape affects SQ.

Question 3
Is there a way I can mount my drivers deeper into my baffle without coloration to the sound? Is it a matter of me being unable to mount the driver past a certain depth into the baffle so that I can avoid having a throat deep enough to effect the frequency range of a specific driver? If this is the case then it’d be nearly impossible to have no coloration to the tweeter as this is the driver I will have to mount the deepest if I want to time align each driver at the horizontal axis.

Other options for time alignment without recessing drivers deeper into the baffle are angled baffle and stepped baffle. Angled baffle works well because usually the listener prefers toeing the speakers out a little anyway, so listening on tweeter axis is not desired. Might as well point it upward to allow time alignment with simple build and good looks. Works well for normal dynamic tweeters, more complicated with ribbons with limited vertical dispersion.

Question 4
With all this in mind, is there a shape that I can make the cutouts so that the cutouts won’t add coloration to the signal?

A straight walled waveguide is popular in some high end designs, like YG Acoustics, or the new Channel Island Audio speaker. There is no perceptible color from the designs, but a very smooth handoff between the tweeter dome and the waveguide is critical. Gaps and protrusions into the horn surface disrupt wave motion and create HOMs (high order modes)  HOM is more noticeable in larger horns that play midrange frequencies. It is less noticeable in the high frequencies which is why you see some speaker with straight wall waveguides. But other brands like Revel make the effort to use curvilinear shaped waveguides instead of straight horns to more smoothly deliver the wave to the baffle curface with minimal disturbance.

Question 5
Is the idea behind a tractrix horn to amplify the signal without coloration of the sound?

Yes, the first horn was a straight horn, with segmented sides. It amplifies the sound pressure by increasing efficiency of the diaphragm by increasing acoustic impedance and limiting dispersion. But the tonal distortion of a straight horn is very noticable due to internal reflections. Tractrix was one of the first attempts to lessen the reflections to make a more realistic tone. It still has abrupt exit which causes diffraction. The most natural sounding horn shape is the Le Cleach horn. But it is very large in width which create other design problems.

[richidoo]

Because physical appearance is such a big part of selling speakers, and a plain, vertical baffle without angles and horns is the preferred fashion, designers must be able to time-align the drivers using the electrical crossover only, which drivers mounted flush on a vertical baffle. They do this by using asymetrical crossover filters. Most common example of this is a 2nd order filter on the woofer and a third order filter on the tweeter. Using an extra electrical pole (aka order) on the tweeter delays the launch of the tweeter wave by 90 degrees, because each pole adds 90 degrees of delay. 90 degrees of a 360 degree wave is 1/4 wavelength. A wavelength calculator converts your crossover design frequency to wavelength, then you divide that by 4 to determine the distance of delay per pole at the xo freq. At typical tweeter crossover frequency of 2400 Hz with 2nd order filter on woofer, and 3rd order filter on the tweeter (asymmetrical filters) makes the tweeter act as if it was recessed relative to the woofer by 1.4".

The problem with asymmetric crossover filters is their rolloff slopes are not the same steepness, and steepness relates directly to phase rotation. So the filters do not remain in phase with each other throughout the entire crossover band, they're exactly in phase only at the nominal xo freq. But this is one of many design compromises made in speaker design and its very common. Symmetric filters is preferred because they will remain in phase throughout the crossover band.

In the real world, bandaids are often required to reach the design goals and it's rare that a perfectly harmless speaker design is bestowed upon the world. One useful bandaid is the use of notch filters in a xo design to give small tweaks to the slope and phase to fine tune the time and phase alignment through the entire xo band. Filter order is primary effect, while notch ("EQ") filters have secondary effect for fine tuning FR and phase.

[musiccritic5088]

Because physical appearance is such a big part of selling speakers, and a plain, vertical baffle without angles and horns is the preferred fashion, designers must be able to time-align the drivers using the electrical crossover only, which drivers mounted flush on a vertical baffle. They do this by using asymetrical crossover filters. Most common example of this is a 2nd order filter on the woofer and a third order filter on the tweeter. Using an extra electrical pole (aka order) on the tweeter delays the launch of the tweeter wave by 90 degrees, because each pole adds 90 degrees of delay. 90 degrees of a 360 degree wave is 1/4 wavelength. A wavelength calculator converts your crossover design frequency to wavelength, then you divide that by 4 to determine the distance of delay per pole at the xo freq. At typical tweeter crossover frequency of 2400 Hz with 2nd order filter on woofer, and 3rd order filter on the tweeter (asymmetrical filters) makes the tweeter act as if it was recessed relative to the woofer by 1.4".

The problem with asymmetric crossover filters is their rolloff slopes are not the same steepness, and steepness relates directly to phase rotation. So the filters do not remain in phase with each other throughout the entire crossover band, they're exactly in phase only at the nominal xo freq. But this is one of many design compromises made in speaker design and its very common. Symmetric filters is preferred because they will remain in phase throughout the crossover band.

In the real world, bandaids are often required to reach the design goals and it's rare that a perfectly harmless speaker design is bestowed upon the world. One useful bandaid is the use of notch filters in a xo design to give small tweaks to the slope and phase to fine tune the time and phase alignment through the entire xo band. Filter order is primary effect, while notch ("EQ") filters have secondary effect for fine tuning FR and phase.

Thank rhichidoo! You’ve covered every base of my questions regarding to this specific aspect of crossover design, but this still leads me to more questions. My first question is unrelated to this topic but has to do with how this forum works. If I reply to your message as I am doing right now, how easy is it for you to notice I replied? Do you get a notification when I reply or do you have to manually check this thread to see whether of not I replied? Second Question is, let’s say you have a 5th order XO (I assume this is how you achieve 360° phase shift). Does this really delay the driver in time, or does it just change the phase of the driver back to normal? For instance if a have a 2-way design with my tweeter on a 5th order and my woofer on a 1st order, does this mean that if I play a frequency sweep through the speaker, that they will be playing the sweep at different times regardless of listening position? Is adding poles to the crossover really the same as physically moving the drivers backwards when looking specifically at time arrivals of the drivers? Also would you say that physically moving the drivers while keeping a symmetrical XO would be a superior solution vs having them rest ok the same vertical axis and adjust their phase relationship through the use if an asymmetric XO?

Also as a main question I want answered, Can you adjust the actual physical time delay of the drivers, using the crossover? Can I actually delay the whole signal with the crossover and not just the phase shift?

[FullRangeMan]

If I reply to your message as I am doing right now, how easy is it for you to notice I replied?
See this feature:
New replies to your posts
https://www.audiocircle.com/index.php?action=unreadreplies

Do you get a notification when I reply or do you have to manually check this thread to see whether of not I replied?
One can get email notification if it was configured.

[richidoo]

Thank rhichidoo! You’ve covered every base of my questions regarding to this specific aspect of crossover design, but this still leads me to more questions. My first question is unrelated to this topic but has to do with how this forum works. If I reply to your message as I am doing right now, how easy is it for you to notice I replied? Do you get a notification when I reply or do you have to manually check this thread to see whether of not I replied?

It's all manual, but the "Notify" button at the bottom of every page in a thread subscribes you to that thread to receive an email when a new post is made since the last time you read it. The button toggles to "unnotify" so you can shut it off.

Second Question is, let’s say you have a 5th order XO (I assume this is how you achieve 360° phase shift).

4th order filter makes 360 phase shift. 1st order is 90. 5th order is rarely (never) used. Above 4th order usually only the even orders are used on speaker filters.

Does this really delay the driver in time, or does it just change the phase of the driver back to normal?

Yes, it delays the driver in time. Phase can only be delayed, never advanced. If you need something advanced you do that by delaying everything else in the system.

For instance if a have a 2-way design with my tweeter on a 5th order and my woofer on a 1st order, does this mean that if I play a frequency sweep through the speaker, that they will be playing the sweep at different times regardless of listening position?

Yes, but they would only be playing portion of the sweep that passes their filter. None of it would ever be in phase, because...
Adjacent drivers sharing a crossover band are usually symmetrical in order, or differ by only one order. Most often it is 2nd and 3rd for the flat baffle trick. Keep in mind, the delay is in degrees, so the delay of the filter varies with wavelength/freq. Acoustically symmetrical high pass filter and low pass filter will remain in phase through the crossover band, even though their phases are constantly changing. High pass filter's phase decreases while the low pass filter's phase increases rotation.

Is adding poles to the crossover really the same as physically moving the drivers backwards when looking specifically at time arrivals of the drivers?

Yes.

Also would you say that physically moving the drivers while keeping a symmetrical XO would be a superior solution vs having them rest ok the same vertical axis and adjust their phase relationship through the use if an asymmetric XO?

Yes, a symmetric XO is superior to asymmetric XO because the drivers stay in phase through the crossover band, while asymmetric crossovers are only in phase at the nominal crossover freq. and the phase error increases as you move above and below the Fxo. It's acceptable distortion in low cost speakers, but for hifi you want symmetric acoustic response.

By acoustic response I mean the net acoustic rolloff that's put into the air, which is the sum of electrical filter's function plus the natural rolloff of the unfiltered driver. A tweeters natural low end rolloff and a woofers high end rolloff contribute their own phase rotation and this can be used to reduce the number of poles in their filter. The crossover design affects the total response and all the physics that are involved with both drivers.

Also as a main question I want answered, Can you adjust the actual physical time delay of the drivers, using the crossover?

Yes, but compensating for physically time-misaligned drivers requires asymmetric xo. A symmetric xo would shift phase of both drivers equally, so no relative delay occurs. They remain "phase coherent" singing together in a symmetrical xo.

Can I actually delay the whole signal with the crossover and not just the phase shift?

Delay is phase shift and phase shift is delay in the analog world. Delay is caused by shifting the phase by passing signal through a filter.

Technically, you could put a steep passive high pass filter at 5Hz and the whole speaker response would be delayed equally at all freqs in the pass band of the subsonic filter, but it would sound bad and waste a lot of power and money. A digital crossover box would do a better job at delaying a whole speaker.

[FullRangeMan]

Does this really delay the driver in time, or does it just change the phase of the driver back to normal?
If you mean say delay the signal a few seconds or even more you will need a place to store the signal as an HD and a another amp to release the signal to the driver when desired, what seems a advanced military weapon.

[musiccritic5088]

Does this really delay the driver in time, or does it just change the phase of the driver back to normal?
If you mean say delay the signal a few seconds or even more you will need a place to store the signal as an HD and a another amp to release the signal to the driver when desired, what seems a advanced military weapon.

Lol, thanks!

[musiccritic5088]

It's all manual, but the "Notify" button at the bottom of every page in a thread subscribes you to that thread to receive an email when a new post is made since the last time you read it. The button toggles to
Yes, it delays the driver in time. Phase can only be delayed, never advanced. If you need something advanced you do that by delaying everything...

Thanks for answering all of my questions! I really appreciate you took lots of time to give me thorough answers. As of right now you’ve answered all the questions I can remember I had (maybe I should write them down somewhere next time). Again thank you so much! This is all stuff that will really help me in design speakers!

[musiccritic5088]

Because physical appearance is such a big part of selling speakers, and a plain, vertical baffle without angles and horns is the preferred fashion, designers must be able to time-align the drivers using the electrical crossover only, which drivers mounted flush on a vertical baffle. They do this by using asymetrical crossover filters. Most common example of this is a 2nd order filter on the woofer and a third order filter on the tweeter. Using an extra electrical pole (aka order) on the tweeter delays the launch of the tweeter wave by 90 degrees, because each pole adds 90 degrees of delay. 90 degrees of a 360 degree wave is 1/4 wavelength. A wavelength calculator converts your crossover design frequency to wavelength, then you divide that by 4 to determine the distance of delay per pole at the xo freq. At typical tweeter crossover frequency of 2400 Hz with 2nd order filter on woofer, and 3rd order filter on the tweeter (asymmetrical filters) makes the tweeter act as if it was recessed relative to the woofer by 1.4".

The problem with asymmetric crossover filters is their rolloff slopes are not the same steepness, and steepness relates directly to phase rotation. So the filters do not remain in phase with each other throughout the entire crossover band, they're exactly in phase only at the nominal xo freq. But this is one of many design compromises made in speaker design and its very common. Symmetric filters is preferred because they will remain in phase throughout the crossover band.

In the real world, bandaids are often required to reach the design goals and it's rare that a perfectly harmless speaker design is bestowed upon the world. One useful bandaid is the use of notch filters in a xo design to give small tweaks to the slope and phase to fine tune the time and phase alignment through the entire xo band. Filter order is primary effect, while notch ("EQ") filters have secondary effect for fine tuning FR and phase.

Lately I’ve been thinking about phase rotation and I’ve been wondering about the fact that each frequency correlates to a specific phase rotation distance. For instance let’s use a 2700hz signal with a wavelength of about 5 inches (I chose this frequency to make things simpler). If my current understanding of phase shift serves me right, if you move 2.5 Inches toward or away from the speaker, you change the phase angle by 180°.
This 2.5” difference only works at 2700hz and if you did this at any other frequency, you wouldn’t change the phase angle by 180°. So this brings me to my first question.

Question 1
When designing a crossover, each pole adds 90° of phase rotation. Does this mean exactly 90° at only the crossover frequency? If this is true, then a 2nd order high-pass crossover at 2700hz would be equivalent to a 2.5” distance change away from you?

Question 2
You said the the roll off slope directly relates to phase shift rotation. Does this mean that in a high pass filter, the phase continuously changes rotation while moving down in frequency? Does the phase rotation remain the same above the high pass xo frequency? Also if the phase rotation is continuously changing above or below the crossover point, doesn’t this mean that in a two way speaker, the tweeter will be out of phase at certain frequencies with the woofer below the crossover frequency and vise versa for the woofer?

[richidoo]

When designing a crossover, each pole adds 90° of phase rotation. Does this mean exactly 90° at only the crossover frequency?

No. It's rotated exactly 90 only in the pass band, or the frequency range outside of the filters influence. In the filters influence, the phase rotates greater than 90 with increasing filter effect. The more attenuation, the more phase rotation.

The nominal frequency of a filter is called the knee or F3 because it happens at the -3dB rolloff point. But few crossovers (two filters crossing) happen at -3dB, most modern crossovers happen -6dB below the input signal. Read "Loudspeaker design cookbook" to get filled in about crossover design.

If this is true, then a 2nd order high-pass crossover at 2700hz would be equivalent to a 2.5” distance change away from you?

In theory, above the crossover band, where frequency is flat again, the phase of the electrical output of the filter, relative to the input signal would be +90 degrees. I don't know exactly what frequency the filter becomes flat and 90 would begin. I think you trig and need calculus to figure that out but it's irrelevant to the XO design process. At the crossover freq where attenuation is -6dB, the phase rotation would be more than 90 degrees, and increases with attenuation.

You said the the roll off slope directly relates to phase shift rotation. Does this mean that in a high pass filter, the phase continuously changes rotation while moving down in frequency?

Yes! You got this.

Does the phase rotation remain the same above the high pass xo frequency?

You are thinking correctly, in that phase will flatten above the filter band, but technically the crossover frequency is at the -6dB point  and therefore still within the filter band so the phase at crossover is still greater than final phase. Phase only flattens (stops rotating) when amplitude rises back to 0dB = input signal amplitude.

Also if the phase rotation is continuously changing above or below the crossover point, doesn’t this mean that in a two way speaker, the tweeter will be out of phase at certain frequencies with the woofer below the crossover frequency and vise versa for the woofer?

Good question, you're getting it! Actually, symmetrical order filters rotate phase at same rate and same direction so they do stay in phase through the crossover band, like magic!

[musiccritic5088]

No. It's rotated exactly 90 only in the pass band, or the frequency range outside of the filters influence. In the filters influence, the phase rotates greater than 90 with increasing filter effect. The more attenuation, the more phase rotation.

In this message when I say “at the crossover point” I mean to say “at the roll off point”

If you rotate the phase 90°, don’t you also have to specify at which frequency the phase rotation will be 90°? From my understanding, you can’t just rotate the phase by 90° at every frequency. Let’s say you rotate the phase by 90° at 2000hz specifically. Doesn’t this mean that you rotated the phase by 180° at 4000hz? Because if you shift the phase physically (by moving the speaker forwards or backwards) I can almost guarantee that a 90° shift at 2000hz is equivalent to a 180° shift at 4000hz. The only way you could shift phase by 90° at 2000hz, while also shifting phase by 90° at 4000hz (and all other frequencies above 2000hz) at the same time, is if the phase shift amount, continuously and proportionately changes while going up in frequency. This would be the only way to shift phase by a certain degree at all frequencies at the same time. For instance you could have 180° of phase rotation (or completely cancellation) throughout the whole frequency range. By simply moving a driver, you can only reverse phase by 180° at one frequency at a time, based on the distance you move one driver away from another. So all this basically leads me to the question. When you have a 2 pole crossover (180° of phase rotation), are all frequencies of the band pass rotated by exactly 180° (again for this to be true, the phase rotation would have to be continuously changing to stay exactly 180° out of phase at each frequency. I know this sounds counter intuitive, but I have no other way of describing what I’m trying to say, but basically, use the idea of physically moving a driver to change phase rotation. In this idea, you can only make it out of phase with another driver at one frequency at a time. This means the phase is rotated by 180° at one frequency at a time and not at all frequencies at once.), or is only the set xo frequency rotated by 180° and everything in the band pass is continuously changing in phase rotation. I know this message is hard to read but I did my best.

[musiccritic5088]

Good question, you're getting it! Actually, symmetrical order filters rotate phase at same rate and same direction so they do stay in phase through the crossover band, like magic!

I was planning on not having a high pass filter for my midrange driver in a 3-way design while having a low pass filter on the woofer. Based on what you just said, this would be a bad idea because the woofer would be out of phase from the midrange at specific frequencies. Correct? Also when I’ve been saying “at the crossover point” I should have been saying “at the roll off point”.

[musiccritic5088]

New question! When building a 3-way speaker where each driver has an impedance of 8 ohms, how can I get the whole speaker to be 8 ohm without sacrificing sensitivity? I know you can use resistors to up the impedance, but won’t this have an effect on efficiency? Also, more speakers in parallel=more efficiency, and more speakers in series = less efficiency, right?

[FullRangeMan]

how can I get the whole speaker to be 8 ohm without sacrificing sensitivity? I know you can use resistors to up the impedance, but won’t this have an effect on efficiency? Also, more speakers in parallel=more efficiency, and more speakers in series = less efficiency, right?

Any xover part will decrease driver efficiency and loss low level music signal as heat, tô avoid it you would use a FR driver direct from the amp.

[musiccritic5088]

Any xover part will decrease driver efficiency and loss low level music signal as heat, tô avoid it you would use a FR driver direct from the amp.

I’m guessing, based on your name, you’d like this solution? Lol! No but realistically is there a way I can change impedance with little effect to efficiency in a 3-way speaker where each driver is 8 ohms?

[FullRangeMan]

There is no free Lunch in áudio,  music is a very demanding and sensitivity art when recorded, but there are some speaker builders that will say xovers dont harm the music.

Usually they use a small cap in the tweeter.