Active crossovers - a search for the more desirable solution

Ionmw

Member
2015-10-15 11:19 pm
*** Disclaimer : "FIR is off-topic, not in the scope of the thread intent" ***
Thank you all for due consideration.
-----------


Hi all,

I am seriously wondering which type of active crossover is most desirable:

1. the square wave fidelity type (total phase zero throughout) but with 90° dephase between speakers at XO frequency. Typical example the 1-order RC with 6dB/octave.

2. the on-phase at (and around) XO freq. between the speakers, but with a total linear offset throughout the spectrum? Typical examples the 2+order RC with 12dB/octave or more.

As reference, Mr. Pass has mentioned here
Phase coherent crossover networks | Pass Labs
the two configurations, but he did not mentioned which case is more desirable, assuming perfect speakers.

I mean, I must agree from experience that total summed phase of output signals being a straight and flat line is a fine and dandy thing for any EE guy to see his test triangular or rectangular signals transmitted perfectly, but... sadly this happens only if we will add those HI and LO signals electrically! In real life, and the real-world purpose of XO in acoustics is so that we want send these two Hi and LO signals at separate speakers(!) and then we are then suppose to add them acoustically. It is this little script detail of 90° dephase between Hi and LO at XO freq. that concerns me. I should know how to solve this phase alignment or else the flow-media physics whispers to me that this might actually be a big problem. But I do not know how to solve the 90° dephase. It worth our attention and some questions? I know that Mr. Pass invited us to do that in his article.

So, for 6db Systems the acoustical Problem is our physiological timeline trigger value (for clarity. Suppose you have a 6dB slope XO at 125Hz, like for example between our Sub and Woofer. Then the 90° dephase between these signals would mean that our Sub will kick with 20ms before our Woofer will start to move. Reading that any signals that are not spaced at least 4ms in time domain are audible as separate signals, then such a negative blurring effect will be clear until 625Hz and should remain an issue until 6250Hz! At least anything separated by more than 4ms is known by some guitar guys among us as "Reverb" in their nice Fender amp. I wonder how would other affirm that their 6dB XO-system does provide them clarity in this audio range.

Additionally, I also do not imagine how one would do time alignment for 20ms. With 6m offset this is helpless (except large venues).

So, I would personally incline to believe for now that Case 2 is the more desirable as it helps clarity of the fundamental harmonics. Considering arbitrarily a "worst case" when total dephase of square wave between first 5 harmonics could be around 90° with weight at around or below 40dB (volts), this still means that a triangular or square signal would be totally recognizable.

But, I am not sure... since all these numbers are academic in nature. I need your feedback.

Would you please share what your real-world experience tells?

Or, anybody, can tell at least where can I find a thread concerning this topic?
Thanks to all,
ion
 
Last edited:
Neither

Any crossover (analog active, analog passive, or DSP) which does NOT take into account the "natural" frequency (and phase) bandpass responses of the raw (unfiltered) drivers to which it is connected is sub-optimal (read: often woefully inadequate), irrespective of its theoretical "textbook" characteristics when ideally connected to "perfect", DC-to-MHz flat drivers.
 

Ionmw

Member
2015-10-15 11:19 pm
Hi Marco,

If I understand you correctly, does it mean that if we find the suitable real Hi and Lo drivers to handle the oddity of phase quadrature (otherwise so much loved and magically useful in EE!) then we should build only bottom flat total phase loudspeakers. Yes, why not? Maybe these results are all around me.

But... this also means we should rely on speakers imperfections. This is somehow opposite to our intuition that tells to stay away from problems, i.e. to operate speakers in their "innocent" and linear range. This is the first contradiction between what we tell and what we do.

Actually many, if not most of the books and articles tell us again and again that speakers should not be used with XO near any resonance especially near the main one.


Another aspect is that HI speaker will be used near resonance, while the LO speaker will be used in its linear zone. Fine. Engineering speaking, this means that because the task is to compensate ("coherency") by a local 90° phase shift then this means we MUST actually hit XO very close the resonance frequency of HI speaker. This is not actually what the manufacturer advise, or any other reading source.

I do not mind hunting rare articles, but in this field I found yet nothing helpful or described in a helpful way. Generally they tell to listen with our ears and judge success based on them. Maybe... like the artists that we all are. Then for the rest 99.99% not DIY, probably the LS engineers that design XO for mass products do not need any measurements or design flow... but they must only understand the ears and preferences of this not small 99.99% of population better? Fine. This was my questions actually, I wanted to know if they consider more important this or that and which scale to use from terrible to mediocre to good and excellent.

I am looking yet to find an independent critic for technology side, former XO designer for mass produced LS, that takes 4-5 good and extreme good quality examples and dissect, analyze them and tell us why the design is so as it is found. I found None, for a span of many decades. Curious, this is yet to be declared by me as the single technological field (audio) where such thing is happening, that we actually do not find any suitable example among the tons of books and papers.

We are here curious about this "how-to" in real world. Do you have your own preferred flow and experience with some words on good and bad cases (results) to share to us?


Many thanks and regards,

ion
 
Last edited:

AllenB

Moderator
Paid Member
2008-10-18 11:31 am
But... this also means we should rely on speakers imperfections....
that speakers should not be used with XO near any resonance especially near the main one.
Sometimes these issues should be fixed, sometimes left alone, and sometimes they should be crossed out. I don't like to ignore them, I'd rather make a decision and do what needs to be done.
 

Ionmw

Member
2015-10-15 11:19 pm
Hi Allen

If you know such product containing an active XO most probably bound to a specific LS product that has bottom flat acoustic phase response and no quadrature problem(!), would you share its name, for the sake of giving us mortals some hope? Almost any example that costs below 20K is good.
Assuming you have really good speakers to work with (and which will be operated in their regions with no resonances), how one can solve the quadrature at XO frequency? Is there any solution for this problem, that comes also with phase bottom flat response too?!
Can you comment if you have a preferred design flow and design philosophy (general approach markers and clear don't-s) concerning the topic?
Do you have any particularly useful preference for a book or article that teaches us the real life of XO+speakers with details and full measurements for some examples?


If it is a DIY, not a commercial product, would you share its design details and measurements too?
Thanks and greetings,
 
Last edited:

AllenB

Moderator
Paid Member
2008-10-18 11:31 am
If it is active or passive, there is no difference. The acoustic issues are the same and I do the same thing for both.

Can you give an example of an issue that you deal with, for example..
most of the books and articles tell us again and again that speakers should not be used with XO near any resonance especially near the main one.
It is OK to use a speaker near its resonance if you get the response/phase sorted out, if the power handling is acceptable, if the dispersion is not a problem, if the harmonic distortion is not extremely high (some is acceptable).
 

Ionmw

Member
2015-10-15 11:19 pm
To be more specific, I come from the world where I deal with resonances daily, from breakfast to sunset.

In audio (and in my work) I dislike the problems they come with. Sensitivity of parameters, group delay for giving some examples, among many other more. So I tend to avoid the main resonances.
My DIY design concept for active LS at home should use some good speakers but not near resonance.
I stumble on either fixing the quadrature and obtain sound good at bass and mids or obtain the lovely square at the output and obtain the "all natural" sound from 20Hz to 20KHz (BW around 50-70KHz).

I only need hope (or better see an example) that fixing quadrature + bottom flat phase acoustic is possible. If this hope exists, I would strive to find and understand how. In this moment I do not know hot to solve both problems.
Else, somebody with experience should feel empathy for me and others like me looking for the same answer, and commit to write a NO statement (with his explanations why not, of course), and hence put me out of current misery looking for unobtainum.
 
Last edited:
I admit I don't know all the details, and indeed can't fully follow the arguments here :mischiev: but a comment:


If only active crossovers are being considered, I thought that a single amp directly driving a single driver had few to no problems due to inductance, capacitance, etc. compared to when the paleolithic :) passive x-over is used?
 

Ionmw

Member
2015-10-15 11:19 pm
Hi Allen,


Hm, interesting note you made... or question.

I assume you know already if two versions of the same harmonic signal have 90° phase shift, they are called "in quadrature". Does it matter if he uses the word or not? Maybe he wanted to reach to larger audience and considered that word "quadrature" is too high level for some audience.

In link from Mr. Pass, the Figure 2 plots the phase for Hi and Lo. There can be seen for 1st-order the quadrature of Hi and Lo. Phase plots for higher orders are not completely accurate, nor some of his statements, I mean that there are very slight plotting errors there but this is not the topic: main trend is correct as plotted and his analysis, as much extent as he wished to share was also correct. My questions do not have answers in his thread, but he offered us all an open invitation to give our thoughts in this problem. Maybe this is why I asked them here...?

De-phase was a slang from my work and I wondered if I should use "phase align" too, for clarity. But in context I thought it will confusion with time align (many make this confusion). It seemed fine for me. Sorry, I did not realized is slang. Please read "make a 90° phase shift" instead.
Also, the 90° phase shift for either Lo OR Hi, or alternatively the +/-45° for both Lo AND Hi should be full-spectrum, not just local to the XO range or even worse local to only a region in the XO.

regards,
ion
 

Ionmw

Member
2015-10-15 11:19 pm
Speakers have impedance variations even without passive components. Those paleolithic drivers ;)
Thanks for your answer, I wanted to answer him the same:) I agree also to your additional notes, but this aspect is not of my concern, definitely not the question of this topic.
Assume amplifier has sufficient damping factor.

For this topic: assume the paleolytic drivers have only their main resonance and a subsequent resonance-free region (not too large) where the driver is intended to be used with active N-way crossover (DIY is not limited to 2-3-4 drivers as commercial products might be).

Microresonances are not considered.

What other drivers you use, or suggest for us better alternatives? Interesting to know...
 
Last edited:
...My DIY design concept for active LS at home should use some good speakers but not near resonance....

Absolutely true. Therefore, you must agree with me that all woofers everywhere are a real dumb idea since their pass band includes their resonance.

BTW, even more deceptive than "hearing with your eyes" (by eyeballing FR and other plots) is "hearing with your theory" which is what you seem to adhere to. The notions that are swell in say, mechanical design, may be without significance to human hearing, and vice versa.

This forum has devoted vast and continuing effort to argue about things like phase that may in fact influence hearing music zilch, at least within reasonable limits.

B.
 

Ionmw

Member
2015-10-15 11:19 pm
If only active crossovers are being considered, I thought that a single amp directly driving a single driver had few to no problems due to inductance, capacitance, etc. compared to when the paleolithic :) passive x-over is used?


...indeed this is the correct working model/assumption here in this topic.

But, I might emphasize that it "should have", not "had".
Any experiences to share? Please tell what arguments needs more clarification, or maybe just a rephrasing to be faster to understand?
Why exactly is the Hi and Lo quadrature a problem sonically but not electrically? This is what I should clarify?


Kudos,
ion
 
Last edited:

Ionmw

Member
2015-10-15 11:19 pm
BTW, even more deceptive than "hearing with your eyes" (by eyeballing FR and other plots) is "hearing with your theory" which is what you seem to adhere to. The notions that are swell in say, mechanical design, may be without significance to human hearing, and vice versa
B.
Thanks,

this note will take me some longer time to understand.


Else, I tend to agree with you on woofer point, but not sure. Surely there should be a "if" instead of "since". Else, I know commercial examples of good work on woofer side. But is the Mid's region that troubles me.
 

AllenB

Moderator
Paid Member
2008-10-18 11:31 am
1. the square wave fidelity type (total phase zero throughout) but with 90° dephase between speakers at XO frequency. Typical example the 1-order RC with 6dB/octave.
Now I am clear. The phase looks like it is 90 degrees apart, but away from the crossover only one driver is playing so there is no phase separation. At the crossover, the total acoustic phase is an average of the two.

This changes off-axis response. Power and frequency response are flat on-axis, but not off-axis.

If you do first order, you will have a challenge with cone breakup and tweeter low frequencies.

Square wave is nice but maybe not audible. There is another way to check. Make a normal good crossover, then use an FIR filter to make phase flat.
 
Bla bla bla and speakers are a compromise and there is no perfect speaker :p

sidenote
active N-way crossover (DIY is not limited to 2-3-4 drivers as commercial products might be).
We should define why 3 way is the best - and not only because 3 is the magic number (mmmhhh...maybe implies 2 as the crossover points + 2 as the out-of- passband points+ One is what we hear but the ears aretwo...and so on!)
 

Ionmw

Member
2015-10-15 11:19 pm
To keep on-topic we should stay with 2way.

More useful for dissection, consider your answers if the XO is at 500Hz, between Bass and Mid drivers.

Suppose the drivers are in close proximity to each other.
Suppose too that the drivers are of about the same area.
Both drivers operate away from their main resonances.



@Allen: while the point is indeed using a 2way example, I want to exclude the typical 60Hz-20KHz 2-way speaker case. Main question and problems are described here for low frequency active XO between drivers having roughly the same surface.
 

AllenB

Moderator
Paid Member
2008-10-18 11:31 am
All these transient perfect crossovers you describe give good square wave performance on-axis (FWIW, but is it audible?). The differences are out of band levels and off-axis.

Out of band.. cutting cone breakup, relieving a tweeter of bass, these are good things.

Off-axis.. you are trying to remove this variable by specifying drivers that are close together. Same size is not the issue, I'm talking about vertical issues related to crossover lobing.

All these things have been done passively before.