Introduction to designing crossovers without measurement

And yes I have done it all by ear with no measuring tools or software, just a friend's Proac Studio 100's as reference.

Yes, comparing side by side with a proven design is a good practice. I still do it until today.

Without comparing, actually I can detect what is wrong or missing, only by ears. But direct comparison makes it easier and I can also compare the "quality" of the new design relative to the proven speaker.

To me speaker design is an art. The hardest part is to "understand" why one option should be better than the other, when you understand that you cannot avoid compromises in engineering decisions.

We have to understand measurement and simulations, but those will only be meaningful if we know its correlation with sound quality perceived by ears. Drawing this relationship is an "art". Of course, first we must understand the physics (measurements, simulations), then we must have good ears, and the worst thing is that it is not enough. Okay, we can hear differences and we know what physical properties are in effect, but we cannot gather all the good things in one design, we have to choose.

I can see that most speakers have "signatures". It may come from cone material or even deliberately designed to give an "impression" to the listeners. (Well, may be some just have it by accident). Usually there are prices to pay for. Certain speakers are said to be "neutral". But I think only God knows if we refer to the same thing when we mention "neutral" :D

I think we cannot find the more advanced engineering wisdom that is deliberately written in bold points on the internet. May be because only few who knows it, or just don't want to share.

Many times people ask "If you think that the current measurement techniques are flawed, why can't you propose one that work?". I think if the questioners ask nicely, may be he can get the answer, may be.
 
diyAudio Moderator
Joined 2008
Paid Member
I am STILL getting a little sibilance on the tweeter which I guess is due to a rise at the tweeter crossover point and I just cant be bothered to mess around with notch filters or re designing the x-over to a 3rd order which I know would help.
There are a few things that are very important in a speaker but can be very difficult to achieve. One is the correct weighting of the human voice. Another is the correct reproduction of sibilants. Of these, some aspects can be sorted through the crossover and some can not. Some are related to baffling and the room.

The sound that comes directly to the listening position from the driver should have a response that for arguments sake let's say should be flat, or in any case smooth. Most of the sound that a speaker produces does not travel directly toward us though. Some of this sound may affect imaging and much of it has the potential to colour the sound.

Speaker baffling can be responsible for diverting some of this energy via diffraction, with a particular response and timing which may be audible. A baffle also controls the entry of this energy into the room in a way that can be beneficial, or not.

I am of the opinion that fixing sibilants begins with weighting voices. This begins at a couple of hundred Hertz or lower. This is basically about making a vocalist sound the correct size, with the goal of being able to hear their breathing and movements as these seem to be related to this. A crossover here is responsible for equalising the sound, naturally. Managing room reflections here can be critical. Even moving a speaker about one foot can potentially make the difference between just right, and awful to the point of being hard to listen to, when it causes a critical problem in the low hundreds of Hertz.

As you know, dome tweeters don't respond well to being overloaded. Their response also needs to be smooth. The smoother the better, particularly where we are the most sensitive (in the low to mid kHz region). Apart from low variations (peaks and dips) in the response, it should blend with the midrange and again, the direct sound is only part of the story here with regards to whether the midrange driver will match up. Controlling diffraction and managing reflections from a tweeter can also contribute to a smooth response.
 
diyAudio Moderator
Joined 2008
Paid Member
Certain speakers are said to be "neutral". But I think only God knows if we refer to the same thing when we mention "neutral" :D
I'd agree. Fortunately there comes a point where you stop asking yourself that question. At this point you can hear the performers through the recording studio, the recording, your room and your speakers, and for me this was the point all along yet I'm still not sure what neutral means.

I thought I'd found it once, about 20 years ago when I'd managed to find the 'clinical' sound. I guess the point would be that if I liked it enough I may not have built any more speakers.

Fortunately also, there is enough information on this site (on the net, to be fair) to build an almost perfect speaker. I say almost here, noting that the experts can't decide what perfect means in this case. There is much written on measurement techniques and I believe they will serve their purpose to this end.
 
I'd agree. Fortunately there comes a point where you stop asking yourself that question. At this point you can hear the performers through the recording studio, the recording, your room and your speakers, and for me this was the point all along yet I'm still not sure what neutral means.

I thought I'd found it once, about 20 years ago when I'd managed to find the 'clinical' sound. I guess the point would be that if I liked it enough I may not have built any more speakers.

Fortunately also, there is enough information on this site (on the net, to be fair) to build an almost perfect speaker. I say almost here, noting that the experts can't decide what perfect means in this case. There is much written on measurement techniques and I believe they will serve their purpose to this end.

Yeah, I'll tell you what neutral means to me, Boring and uninspiring! Accurate it may be (technically) but this type of sound does not inspire me to want to listen to the next CD in my chosen pile for the evening. If I find the music I'm listening to does nothing for me I would turn the system off and watch TV instead which I'd rather not do. I've heard many reputable (and expensive) designs from big manufacturers and have been very unimpressed with some of them. When the sound is RIGHT it brings a smile to your face and a feast of audio delight to your ears, this is what music SHOULD do, and CAN be achieved with relatively budget priced equipment if chosen or designed properly.
I have found over the years of just messing around with speaker designs that even removing or adding a small piece of internal wadding, it's position in the cabinate and type of wadding can have a massive effect on the overall sound produced, getting this wrong can cause all sorts of headaches for the uninitiated builder as they cannot understand why the $1000's worth of expensive drivers doesn't give them the results they expect.
Yes of course measuring software and the math etc IS important but I believe the final stage of a loudspeaker design should be done by thorough listening tests with a large variation in source material, volume levels, speaker positioning room acoustics etc etc and by the ear of the primary listener.
What might sound great to one person may sound horrible for a different listener due to their auditory anatomy and a thousand other variables. To me its how the music makes me feel emotionally and if the electronics and acoustics are not right, the end result is a big waste of time and money which equals a depressing disappointment.
Iv'e always craved the "live sound" or "performers in the front room" sound which is why I am so critical of my own work, maybe a little too fussy sometimes but that's just me, and what gives me the most pleasure is: I didnt have to spend $200'000 on Wadia CD players, Krell amps and Wilson speakers to get the end result.:D
 
about 20 years ago when I'd managed to find the 'clinical' sound. I guess the point would be that if I liked it enough I may not have built any more speakers.

Isn't most systems clinical? Long ago I was familiar with simple systems. Tube or class-A solid state amps driving fullrangers. Lowther, Fostex, Jamo, Davis Acoustic and now a modified local driver selected by ears (that I use specially for testing new amp designs).

It is this enjoyable experience, sitting all day listening to music, and cannot part long enough with the system, just like listening to certain live music, that makes it harder for me to accept modern systems.

I parted with tube amps because of my perceived low speed of bass slam with rock music. I parted with fullrangers because I don't like the breakups at high SPL, and I want extra details from multi-way speakers.

There are many known compromises, but enjoyment is really a double edge sword for me. It usually doesn't want to go hand in hand with the most transparent, the most detailed, etc. For men, it is clearly different between having a beautiful wife and having one that makes us happy (tho not that beautiful).

Fortunately there comes a point where you stop asking yourself that question. At this point you can hear the performers through the recording studio, the recording, your room and your speakers, and for me this was the point all along yet I'm still not sure what neutral means.

I have a design that in term of details, I have never heard from other systems. Each instruments have body and as if comes from different sources. But it failed the ultimate test. Listening to it while sleeping didn't give me "beautiful dreams". I got "bad dreams" instead :(

Neutral to me is more positive. It doesn't try to impress you with too much of anything.

Fortunately also, there is enough information on this site (on the net, to be fair) to build an almost perfect speaker. I say almost here, noting that the experts can't decide what perfect means in this case. There is much written on measurement techniques and I believe they will serve their purpose to this end.

What a fullranger can easily provide (as compared to multi-way with complex crossover) is the perfect low level reverberation. Many complex multi-way speakers failed to recreate this low-level info. Funny thing is that a resonating baffle can help recreate this reverberation. So...

Those who are familiar with non-resonating enclosures will look down on resonating enclosures. Switching from non-resonating enclosure to resonating one, we will clearly hear that "box sound". But another funny thing is, just give us some time with the resonating box and we will get along well with it and forget the strange sound :D

Harbeth, Spendor, the LS3/5A variations, Audio Note (The Orang Utan), they all sound "wrong" but have higher possibility of enjoyment experience than those speakers with overdamped enclosures.

Try this: change your box with concrete and tell me if you like it better.

Overdamped enclosure only works perfectly with expensive drivers, especially those who can use very simple crossover. Second order is maximum. Sonus Faber speakers can get away with 6db/octave with good measurement results.
 
Isn't most systems clinical? Long ago I was familiar with simple systems. Tube or class-A solid state amps driving fullrangers. Lowther, Fostex, Jamo, Davis Acoustic and now a modified local driver selected by ears (that I use specially for testing new amp designs).

It is this enjoyable experience, sitting all day listening to music, and cannot part long enough with the system, just like listening to certain live music, that makes it harder for me to accept modern systems.

I parted with tube amps because of my perceived low speed of bass slam with rock music. I parted with fullrangers because I don't like the breakups at high SPL, and I want extra details from multi-way speakers.

There are many known compromises, but enjoyment is really a double edge sword for me. It usually doesn't want to go hand in hand with the most transparent, the most detailed, etc. For men, it is clearly different between having a beautiful wife and having one that makes us happy (tho not that beautiful).



I have a design that in term of details, I have never heard from other systems. Each instruments have body and as if comes from different sources. But it failed the ultimate test. Listening to it while sleeping didn't give me "beautiful dreams". I got "bad dreams" instead :(

Neutral to me is more positive. It doesn't try to impress you with too much of anything.



What a fullranger can easily provide (as compared to multi-way with complex crossover) is the perfect low level reverberation. Many complex multi-way speakers failed to recreate this low-level info. Funny thing is that a resonating baffle can help recreate this reverberation. So...

Those who are familiar with non-resonating enclosures will look down on resonating enclosures. Switching from non-resonating enclosure to resonating one, we will clearly hear that "box sound". But another funny thing is, just give us some time with the resonating box and we will get along well with it and forget the strange sound :D

Harbeth, Spendor, the LS3/5A variations, Audio Note (The Orang Utan), they all sound "wrong" but have higher possibility of enjoyment experience than those speakers with overdamped enclosures.

Try this: change your box with concrete and tell me if you like it better.

Overdamped enclosure only works perfectly with expensive drivers, especially those who can use very simple crossover. Second order is maximum. speakers can get away with 6db/octave with good measurement results.

Damn! I think we must be of the same stable.

One of the things I actually left out in my last reply WAS that I also believe a loudspeaker cabinet should not be totally damped (or made of stone) but act as an instrument itself with its natural coloration's (in the right place of course) Like a 200 year old Stradivarius it's the tone in the wood (and resonance) that gives it the rich sound. I didn't mention this because I expected a thousand annoyed forum members jumping down my throat and disagreeing with me, but you know what? I don't really care, that is my opinion (after lots of experimentation) and am sticking to it. At least now I know that I'm not alone in my thinking.

I believe a loudspeaker should have its own character whether it adds or takes away from the music (Gasps of disapproval) there is nothing wrong with that as long as it gives pleasure to the listener, and some of the best I've ever heard have some kind of character.
I was lucky enough to hear a Sonus Faber cremona auditor a few years ago and was very impressed with it's tonal accuracy with classical music, but obviously it is a different story with rock music. I was also very impressed with some Neat Petite speakers at a British HiFi show, but again no good for headbangers. Maybe this is the biggest holy grail of all -trying to design a speaker system (or complete audio system) that does everything well and I'm not sure this is really possible, I think the only thing that came close for me were some oldish very large KEF reference speakers developed for professional recording studios.
Obviously yourself and AllenB have an extensive experience of design and you've had the pleasure of listening to many very good systems for which I am totally jealous -in a nice way of course. I was 13 years old when I found I was obsessed with HiFi and my dream was to get into audio design, unfortunately I didn't get the opportunity and now at the age of 51 I have missed the boat for which I really regret, so I am now left to tinker at home, Its just a shame I cannot share my results with others that matter.:sad:
 
Last edited:
I also believe a loudspeaker cabinet should not be totally damped (or made of stone) but act as an instrument itself with its natural coloration's (in the right place of course) Like a 200 year old Stradivarius

The effect of the baffle is actually well known. LS3/5A for example, deliberately used the baffle to “tune” the sound. What is probably not well known is how it affects our enjoyment, and how enjoyment aspect should be prioritize in the context of design compromise. Imho, not many people care about enjoyment aspect as much as I do.

Sonus Faber has more design with over damped enclosures. But yes, the Stradivari is classic. Imo, a good speaker must have no sound from box resonance. It is again about compromises. For expensive design such as Sonus Fabers’, I don’t expect a resonating box, but a resonating box is much much better than fatiguing speakers.

I believe a loudspeaker should have its own character whether it adds or takes away from the music (Gasps of disapproval) there is nothing wrong with that as long as it gives pleasure to the listener, and some of the best I've ever heard have some kind of character.

I believe it is about design compromise. “Pleasure” is a rare thing in audio world. “Impressive” is the common one. Impress the listeners so they will buy your speakers.

Yes, some of the best speakers have some kind of character. Honestly, at that cost-no-object level, I don’t expect a speaker to have “character”. Ask those who have good ears, they will know that different cone materials have different character. This is because the design is not perfect. If the design is perfect, there should be no difference between “magnesium” and paper except that you will get more detail with magnesium.

Maybe this is the biggest holy grail of all -trying to design a speaker system (or complete audio system) that does everything well and I'm not sure this is really possible

Funny thing is, I think that it is absolutely possible, to design a speaker that is RIGHT. It doesn’t have to do everything well, it just have to do things “right”. This is very philosophical but is deep.

What I notice as the most outstanding “character” of my current speaker is the VOCAL. I have lived with Lowther and tube amps but I don’t think they (Lowther+tube) are at the same level, even with my speaker driven by a poor TDA2030 chip amp. It is not the SWEETNESS (which usually is a sign of distortion) but the NATURALNESS. The proof is in the listening of singers who have the angel voices. Great singers sound great, amateurs sound amateur.
 
If anyone of you guys have ever been in a recording studio, and I mean a recording studio and not some friends basement with a 8 track machine and speakers on a table close to the wall, so you must have heard the sound of real studio monitors which were mated to the room they play with. Those rooms are designed to have a short reverberation time over the whole audible frquencie range. That is the enviroment where the sound engineer monitors his work, where he decides about different microphones used to record as natural as üossible or as artificial as he wants it. That's where the final mix is made by deciding about the colour of the various instruments and putting room information in to the context wether they were natural from recording in larger rooms or artificial digital reverbs.
Those speaker systems don't have "resonating baffles" or any such things derivated from a instrument like a old violine. They are build to reproduce as precisely as possible.
That involves a enclosure that has carefully designed inner proportions to not avoid but spread standing waves across the spectrum. Then timeallignement between woofer and tweeter to stay in the here often used terms is used to induce less damage by the necessary crossover. Of course the used chassis are of high quality and also quality tested, means selected to be well within thenspecs.
 
Last edited:
Not to forghet the non compromise placement of such speakers to be able to really elaborate a stereo mix, spezially with mono recordings and multitrack stuff that is mostly done with one microphone per instrument.
After this all was mixed and mastered, why would we go and add early reflexions to this records by designing speakers with sharp edges?
Why should we add colouring by having a vibrating enclosure?
What happens to the sound when our DIY crossover messes with the phase response of my speaker and gives me crazy coverage patterns?
It is certanly not a enhancement to the original record but a distoted picture of it.
So you guys better take care of trying designing a as good as possible flawless reproduction loudspeaker and placing them as good as you can in your listening enviroment.
Then you will really here what the soundengineer recorded, what the musicians played and how good the room sounded.
That is pleasure and you can listen to it for long times before you fatigue.
No studio by the way uses Lowthers or Fostex fullrange drivers. The studio systems are designed by people who can make a good speaker and have ways around the problems of enclosures or crossovers.
 
The thing is that some records just don't sound good.
The better your speakers are, the more you will hear it.
So this can become a bit of a problem.

Does this happen to ALL "better" speakers?

My current speaker is very transparent. Actually it is part of the design objective. I wanted it to be the best at any price at its size category. In terms of its transparency, I think I have only heard one speaker having the same signature, it was a Martin Logan. But I don't think I want to keep this design :RIP:

Or may be I want to blend specific features to get a good cross breed :D

But I still don't understand what is happening scientifically. I cannot buy that it is just like an amplifier with ultra low THD, where the output will be just similar like the input. Imo, speaker design is not as simple as that :(
 
It is simple if you want to talk science.
Flat frequencie response on axys. Yes.
Flat energy response, means as even as possible off axys response.
Then very important: no phase jumps at transitions between drivers.
All this is allready not that easy to achieve.
If you have that, it should allready sound not too bad. But you have just started.
Doppler distortion is very audible because it actually changes the pitch of the record. Comes from one driver reproducing a too large part of the spectrum.
Same is also a problem if you want the driver to follow the waveforms of the recorded programm. If the driver is not reproducing vaweforms correctly, it will loose important texture and detail especially overtones and change harmonic content to the bad. So it is going to "not sound good" because it will introduce small amounts of disharmonic overtones. That can spoil a good record.
Sort out all those things and you will enjoy listening for hours.
 
Doppler distortion is very audible because it actually changes the pitch of the record.

How do you describe this doppler distortion sound-wise? I mean, can you explain what it is sound like? I know about pitch, but I think there is more to it?

Sort out all those things and you will enjoy listening for hours.

My experience cannot confirm that unfortunately.

There are many high-end speakers out there. And there has been nobody who can sort these things out? :confused: :D
 
Hi end is about making money.
It is not about sorting out every problem.
Hi end has to look good. Sorry to say that. A speaker designed a 100% following
The rules will not look spectacular.
Doppler effect in music will let high midrange frequencies ho up and down in pitch
quite fast and so will the harmonics. The harmonics will then not anymore be harmonic to all the music. Hard to explane. Just when it is not happening you will notice time between the notes. A certain precision and air between everything.
 
Doppler effect in music will let high midrange frequencies ho up and down in pitch quite fast and so will the harmonics. The harmonics will then not anymore be harmonic to all the music.

I don't understand about the harmonics go up and down. I agree that there is a possibility that a "phase distortion" plays the most critical role in high end speakers' problems. But as I can see it, speaker design (or issues) is not so simple. Science so far cannot prove the importance of doppler distortion in speaker. There are too many parameters that can affect sound quality.

Hard to explane. Just when it is not happening you will notice time between the notes. A certain precision and air between everything.

I think I have noticed/heard this many times with high end speakers and my own (it is not rare at all). The questions are, am I experiencing the same thing? Is it the sound of no doppler distortion?

Many times ago I wrote in a thread about my observation of cone movement (I can tell the speaker design perfection from the cone movement).

Because the cone must produce or move in different frequencies, you will notice the small vibration to produce HF and big vibration to produce LF. The best sound was associated with a well controlled cone movement. It should (almost) always vibrate, no abrupt stops or such things, and changes are smooth.

To do the above we need almost everything:
(1) The cone must be stiff and light so it can vibrate along with LF vibration
(2) There should be a perfect compliance or suspension, from the driver itself (Qms/spider/surround/etc)
(3) There should be a perfect compliance or suspension, from the air behind the cone (Vas/enclosure design/enclosure type)
(4) The crossover effect is usually very noticeable (visually with the cone movement and sound-wise) in the use of LCR trap, especially in the mid band.

I think this explains why sealed enclosure tends to be always enjoyable, not because of the transient (which is also true) but because there is no issue with this "certain" thing that I associate with phase, may be the doppler distortion you are talking about.
 
Hey Jay, you can not see frequencies.
You can visually distinguish about 25 movings per second.
You will not "see" midrange frequencies move.
Doppler is clearly changing pitch up and down by the movement of the bass speaker and is interfering with the midrange and is therefore changing the colour of the sound in a very important part of the spectrum.
 
diyAudio Moderator
Joined 2008
Paid Member
What happens to the sound when our DIY crossover messes with the phase response of my speaker and gives me crazy coverage patterns?
I don't see how messing with the phase response can affect coverage patterns (unless you're referring to vertical lobing?). This is a function of the relative levels between drivers and the resultant acoustic characteristic.
The thing is that some records just don't sound good.
The better your speakers are, the more you will hear it.
This is an often reported phenomena. I'm not sure I agree with it.

Speaker issues are typically more of a problem than recording issues. I'll cite: The Animals, House of the Rising Sun as an example of a recording that includes gross overload. This overload is far greater than many of us would like to accept in a speaker and yet, is capable of sounding reasonable.

The problem may be acoustic. When a speaker cabinet is malfunctioning it may give itself away. Any sound energy which is allowing us to locate and 'see' the speaker is wayward energy, let alone up to no good.

This may also be an issue where a crossover (or equalisation in general) has been used incorrectly to fix a problem that cannot be fixed that way. The result is made tolerable, nothing more. A recording with response issues that may highlight this range of frequencies in question could re-introduce the problem.

Take the example of listening to an ordinary television in a room. No matter how it sounds, it is clear that you are sitting there listening to a television. If you recorded this, re-creating the illusion of sitting there through a set of speakers is difficult. With most systems you will not be convinced. If you use such a speaker/setup that can, then the recording will sound much better despite the fact that it may be a recording of a simple speaker from within a room, reproducing distorted program material. It may actually sound like the original situation (and that is what it is...good or bad. It will at least sound correct). This implies that the speaker isn't changing anything significant from the recording..

Faulty speakers may also make the acoustic mistake of sending a significantly different characteristic of the sound to different parts of the room, betraying the intended illusion.

no phase jumps at transitions between drivers.
Are you referring to linear distortions or time domain distortions?

Perhaps you're referring to Group Delay?

GD is usually measured in milliseconds and refers to the rate of change in phase, just as when a speaker is measured from an increasing distance, the phase response will wrap.

Group delay has been demonstrated to have a specific threshhold of audibility.
 
Last edited:
diyAudio Moderator
Joined 2008
Paid Member
But I still don't understand what is happening scientifically. I cannot buy that it is just like an amplifier with ultra low THD, where the output will be just similar like the input. Imo, speaker design is not as simple as that :(
The first thing to accept might be that speaker function (including some speaker issues) happens in three dimensions.

I know about pitch, but I think there is more to it?
When two tones are reproduced together in the presence of non-linear dstortion, a sum and difference signal are created between the tones, as well as between their harmonic distortion components etc. This sum and difference are not harmonically related to the intended signal and don't fit.

I agree that there is a possibility that a "phase distortion" plays the most critical role in high end speakers' problems.
Phase distortion is not necessarily a problem even when it is present. Of course there are limits.

(2) There should be a perfect compliance or suspension, from the driver itself (Qms/spider/surround/etc)
This may make sense with stiff cones crossed at a low frequency. Otherwise the outer cone is just hanging there, with nothing to 'hold its hand' if something goes wrong.
 
I don't see how messing with the phase response can affect coverage patterns (unless you're referring to vertical lobing?).

I am talking about lobe distortion. Yes. Drivers can also be arranged horizontally. Then your lobing takes place in the horizontal plane. If x-over frequencie is in the right relation to center-center distance between the drivers and the phase relation is good, nothing bad happens.
 
AllenB; said:
Are you referring to linear distortions or time domain distortions?

Perhaps you're referring to Group Delay?

GD is usually measured in milliseconds and refers to the rate of change in phase, just as when a speaker is measured from an increasing distance, the phase response will wrap.

Group delay has been demonstrated to have a specific threshhold of audibility.

Same as my last posting. Group delay happens if you deploy steep x-over slopes,
which I believe should be avoided in the lower frequencie regions.
But phase wraps and I don't think we hear this as a problem.