Hi,
Well, given that the underlying principles are really "tres primitive", maybe spending some time on the topic may help.
Box vibration is a big problem IME. JUst take a generic MDF box, a stethoscope, play some wideband orchestral music and listen to drone of the box panels.
Now sit down and listen again, same piece of music.
You can suddenly hear the "MDF Drone", which previously you could not.
Note, the acoustics have not changed, your perception has. What was previously an unidentifiable artefact, the contributed to "boxiness" and the like.
If you have sufficient contol of your senses, you can now assign the "box talk" noise to a filter and ignore it, knowing what it is, if not it will bug the heck out of you.
Diffraction is another issue. Play pink noise and move your head around the speaker and you will easily hear both the tonality and relative levels of the various secondary diffracted sound sources. You can then also hear them more easy in the room afterwards.
Hmm, I am not sure what I am supposed to do with this statement.
It sounds like religion to me.
I know that too small baffles for the tweeter give rise to frequency response ripples (comb filtering) that are both measurable and audible. Of course, some people upon hearing a severely comb filtered signal will perceive this as "very spacious sound", others just as "bad sound".
I have only made few series filters, most of the time they simply cannot be made to work right or are too sensitive to component tolerances. As said, they are one option.
Yes?
Sorry, but regarding the drivers "lookback impedance" (think about why this would be important) given a first order filter, you cannot.
I started out with the drivers, in a different box with a parallel crossover. It never got past "ok", even with quite extensive compensation it was not really that good. I normally always start with "parallel" as default...
Absolutely. Some driver combinations lend themselves best for 1st order series crossovers, others are better off using 10th order parallel, There are no universal panaceas, no solutions that are always right under all conditions.
The Drivers are quite smooth, so there was no need for notches etc. If these are needed series designs become a lot more complex and difficult, perhaps enough so to leave no advantage over parallel ones.
I do have this funny old rule about drivers I use. Any driver (for Mid/Bass/Mid) that, when listened to in open air, boxless and with music does not offer reasonable performance I simply do not use, as it will be too hard work to make it listenable. It usually only takes a few seconds to reject most drivers outright.
Well, if you understand how all these "foreshortened transmission line" Boxes (especially the PMC ones - these where the trigger of my investigation) work, you will also know how to design the reflex box.
You need a certain minimum amount of smoothness in simple series crossovers, difficult drivers with heavy resonance peaks/suckouts close to the
operating range are indeed unusable.
While the impact of compensation circuitry is largest in series crossovers, if designed following the "intelligent design" principles it is possible to mostly omit ny compensation for driver impedance peak/rise problems.
However, this is a bit of a balancing act, as you need to trade off a number of parameters simultaneously while having only a single variable influencing all of them. On the other hand, I found the payoffs worthwhile, in this particular case.
Ciao T
Well, given that the underlying principles are really "tres primitive", maybe spending some time on the topic may help.
Box vibration is a big problem IME. JUst take a generic MDF box, a stethoscope, play some wideband orchestral music and listen to drone of the box panels.
Now sit down and listen again, same piece of music.
You can suddenly hear the "MDF Drone", which previously you could not.
Note, the acoustics have not changed, your perception has. What was previously an unidentifiable artefact, the contributed to "boxiness" and the like.
If you have sufficient contol of your senses, you can now assign the "box talk" noise to a filter and ignore it, knowing what it is, if not it will bug the heck out of you.
Diffraction is another issue. Play pink noise and move your head around the speaker and you will easily hear both the tonality and relative levels of the various secondary diffracted sound sources. You can then also hear them more easy in the room afterwards.
Hmm, I am not sure what I am supposed to do with this statement.
It sounds like religion to me.
I know that too small baffles for the tweeter give rise to frequency response ripples (comb filtering) that are both measurable and audible. Of course, some people upon hearing a severely comb filtered signal will perceive this as "very spacious sound", others just as "bad sound".
I have only made few series filters, most of the time they simply cannot be made to work right or are too sensitive to component tolerances. As said, they are one option.
Yes?
Sorry, but regarding the drivers "lookback impedance" (think about why this would be important) given a first order filter, you cannot.
I started out with the drivers, in a different box with a parallel crossover. It never got past "ok", even with quite extensive compensation it was not really that good. I normally always start with "parallel" as default...
Absolutely. Some driver combinations lend themselves best for 1st order series crossovers, others are better off using 10th order parallel, There are no universal panaceas, no solutions that are always right under all conditions.
The Drivers are quite smooth, so there was no need for notches etc. If these are needed series designs become a lot more complex and difficult, perhaps enough so to leave no advantage over parallel ones.
I do have this funny old rule about drivers I use. Any driver (for Mid/Bass/Mid) that, when listened to in open air, boxless and with music does not offer reasonable performance I simply do not use, as it will be too hard work to make it listenable. It usually only takes a few seconds to reject most drivers outright.
Well, if you understand how all these "foreshortened transmission line" Boxes (especially the PMC ones - these where the trigger of my investigation) work, you will also know how to design the reflex box.
You need a certain minimum amount of smoothness in simple series crossovers, difficult drivers with heavy resonance peaks/suckouts close to the
operating range are indeed unusable.
While the impact of compensation circuitry is largest in series crossovers, if designed following the "intelligent design" principles it is possible to mostly omit ny compensation for driver impedance peak/rise problems.
However, this is a bit of a balancing act, as you need to trade off a number of parameters simultaneously while having only a single variable influencing all of them. On the other hand, I found the payoffs worthwhile, in this particular case.
Ciao T
Why another thread?? This is the same topic as your first thread. 😕
I think a woofer with its size requirements to get 20Hz, etc has fundamental flaws playing that high at all. I have never even seen a good measured woofer that can do this. The closest so far was the AV15X and the TD18H both from AESpeakers.
, I believe the XO being below 500Hz is a lot better then 800Hz which is in the critical voicing range. The Ideal 300Hz to 3000Hz should be untouched by an XO and handled by one source for complete phase coherence throughout that critical range.
Last edited:
You need to build subwoofers and stop trying to build a 2-way 20Hz - 20Khz. Even professional studios do use subwoofers.
btw, there is no room gain in the 30Hz to 60Hz. Most rooms do have gain below 20Hz unless you live in a room the size of a car.
I see you have a 3rd thread on this topic "Manger + IB15" and you are considering subwoofers.
FWIW, you should have posted your complete requirements from the start in one thread and many of the experts here would have given you the best solutions. The discussions are jumping around in multiple threads when its all just one discussion.
Last edited:
Hi,
I would not be so draconian. For example, Tannoy crosses over most of their decent coaxials at around 1200 - 1400Hz (acoustic) and the results are very good. If you consider the human hearing, it is probably a better choice than (say) 500Hz.
Ciao T
I would not be so draconian. For example, Tannoy crosses over most of their decent coaxials at around 1200 - 1400Hz (acoustic) and the results are very good. If you consider the human hearing, it is probably a better choice than (say) 500Hz.
Ciao T
Zero issues with phase, no CTC issues. There is a valid reason to find a single point source to run the 300Hz to 3000Hz critical voicing range. That does not sound darconian to me, it sounds like a solid foundation for speaker design.
Heck, there are thousands of designs with XOs in all over the map from 500Hz to 2KHz that sound amazing but that does not replace the fact that in theory no XO from 300Hz to 3KHz and single point source are would give optimal design performance. Wether that is an audbile thing depends on the designer and the preference.
I would not call those draconian issues. I would call it choice.
I do know one thing for sure you posting Tannoy sound amazing because of the XO being at 1200Hz is 100% anecdotal and meaningless to the discussion. The XO is a necessary evil in its design because the HF driver can only go so low. The reason it probably still sounds incredible is less about the XO point and more about CTC not being an issue.
All speaker designs have specific accepted compromises. You are the expert builder so you know its simply a choice of compromise.
Hi,
Sure there is valid reasoning, from a purely technical viewpoint, which excludes the human auditory system.
By the same logic BTW, one might also declare that a single point source from 30Hz to 30KHz is a good prescription.
If we take the human auditory system into account I would suggest that both 300/3KHz (which I incidentally used in the 3-way I commented on in this thread) and 1.3KHz for 2-Way are very valid choices, that have a better chance than other options to have low audibility.
I find normally 3KHz still too low, I'd ideally like 300/7,500 or 1,300Hz...
Then again, I have in another recent 3-Way design used 60Hz and 600Hz crossover points and the results are very good.
Then, with respect, your knowledge could use improvement. You may wish to study the human hearings sensitivity to phaseshift/groupdelay/amplitude variation, it is quite illuminating, at least I found it to be (e.g Blauert et al. and others).
The bigger Tannoy coaxials do sound amazing (go and listen) and the fact that they successfully integrate (at least subjectively speaking) a metal diaphragm compression driver and a paper cone driver with a crossover smack in the middle of the "critical range" has a lot to do with the precise crossover frequency they used.
Later Tannoy drivers that deviated much from this crossover point do not sound equally well integrated or good, despite having reduced the apparent differences in drivers (Non-compression HF driver) designs.
Ciao T
Sure there is valid reasoning, from a purely technical viewpoint, which excludes the human auditory system.
By the same logic BTW, one might also declare that a single point source from 30Hz to 30KHz is a good prescription.
If we take the human auditory system into account I would suggest that both 300/3KHz (which I incidentally used in the 3-way I commented on in this thread) and 1.3KHz for 2-Way are very valid choices, that have a better chance than other options to have low audibility.
I find normally 3KHz still too low, I'd ideally like 300/7,500 or 1,300Hz...
Then again, I have in another recent 3-Way design used 60Hz and 600Hz crossover points and the results are very good.
Then, with respect, your knowledge could use improvement. You may wish to study the human hearings sensitivity to phaseshift/groupdelay/amplitude variation, it is quite illuminating, at least I found it to be (e.g Blauert et al. and others).
The bigger Tannoy coaxials do sound amazing (go and listen) and the fact that they successfully integrate (at least subjectively speaking) a metal diaphragm compression driver and a paper cone driver with a crossover smack in the middle of the "critical range" has a lot to do with the precise crossover frequency they used.
Later Tannoy drivers that deviated much from this crossover point do not sound equally well integrated or good, despite having reduced the apparent differences in drivers (Non-compression HF driver) designs.
Ciao T
I have spent quite some time. I think the material I choose have been too advanced to be useful.
Now you see why my objective had always been speaker disappearance.
If the box vibrate (standing wave etc), you will hear the box, it will not disappear.
If the tweeter signal is "reflected" by whatever happens on the wide baffle, the tweeter will not disappear. (Sorry, minimal but "controlled" shape has been the best)
If the tweeter has intrinsic high distortion, it will be very difficult (if possible) to make it disappear.
But the most important thing is... if you put any driver in a wrong filter, it will give you distortion that is audible but UNMEASURABLE. It will not disappear, and your measurement trick you into thinking you have a good speaker, but a speaker that you don't want to live with.
So once you can make the speaker disappear, no need to care about tonality, dispersion, group delay, fatigue, directivity, etc. Speaker disappearance is a sign that you have achieved the good things.
Exception for the simple first order. Actually in the past I always started with first order series as default if I heard the driver is smooth. But never 100% satisfied with the drivers that I have used.
Not funny at all 😉
This has been listed on the first row of my wish list 😀
Right!
If you can do it then go for it but again the laws of physics gets in the way. We have to make compromises to find the best solution.
Then can be valid and both can sound great ( I have 3 ways with an XO at 2KHz that I love and my waveguides at 1200Hz rock the HT room) but again one follows a solid theory about phase coherence and the simple fact that no XO is better then any XO when the single driver is perfect for that range. Its just a choice, you can choose to pick your priorities. As you posted some of your designs do meet that requirement. My next build is getting the horn down to 500Hz so its going to be a 3-way with the Coax horn doing 500Hz to 20KHz and a bass bin below (horn or PR).
Well, thanks for telling me I need improvement, that is a choice too 😉
Im more interested in just speaker design but I have played with all of the above using the DCX, it was fun. I also know they do matter to some people more then others. My point isn't even about if they are audible, its simply about measurement and design theory and in that context having phase coherence is a great idea. We all know there are hundreds of design choices and compromises. We simply have to list our priorities, etc and start designing.
BOSE is still loved by 100,000+ individuals so obviously there is a broad range of acceptability.
I said the Tannoy's sound great above. I even have a pair of low cost Tannoy V8s. You do not need to convince me.
Last edited:
The idea is to control the directivity of the speaker. If a narrow baffle suits your needs then you use one.
A waveguide on the other hand is typically designed to keep the sound mostly forward, not to wrap around the baffle. The best would be a very large waveguide. Second best would be a normal waveguide rounded onto a very large flat baffle. Third best would be a normal box. Most of the sound will not touch the box but some will... and it will cause diffraction.
I'm a big fan of basic 1st order crossovers, but I don't know how to implement a first order network into a 3 way speaker.... would you care to publish your crossover for your 3 way, or is that information propitiatory?
Doug, to be honest, I have many more than these three threads on approximately this topic, on this forum, over the last year. I apologize for the botheration this has caused you. But each thread is drawing a different kind of contribution, and I seem to be learning from all of them. As for "complete requirements"... this is not an ongoing project. I have drivers for two other (very different) projects which I will have to complete before touching this one.
I am considering subwoofers, but in that thread I have stated that the idea would be to crossover between 400 Hz & 1 KHz. That thread was created separately to draw the involvement of people who have experience with either of those two drivers... none forthcoming so far... maybe you're right and I should have put it in this.
I see many people here suggesting that what I'm aiming to do is impossible. I know that you can never have everything you want in a single box, its a balance of compromises. A certain author who has written multiple books on studio monitoring design once told me "loudspeaker design is a highly imperfect science". He also happens to be the designer of the Reflexion Arts 234 and 239X... two way speakers that can do 20-20. Some will insist that the performance of a 15" in the mid range will never be satisfactory, but the words of people who have heard this speaker in a properly designed room have convinced me that it is an alternative I cannot ignore. Now the 2235H is no longer available (in a thread dedicated to the purpose, I found that the TD15H is the closest match) and the TAD CD/axissymmetric horn combo will always be much too expensive for me. I started out looking to build a speaker designed with this approach. I tested it with the big box econowave and now I am thinking about the next step forward. I am open to other approaches as well...
Thanks to everyone, again... new to this and learning all I can. Really fascinating... the little line array drivers, the Beyma TPL, B&G RD-75... Multiple ways of skinning the cat!
Hi,
I have no drawing for the schematic even, that I could scan.
Basically it is very simple. A resistor in parallel with a resistor equalises the tweeter (can be omitted if tweeter and midrange have same sensitivity). Another resistor in parallel to this combo adjusts the impedance back to nominal for the speaker (6 Ohm in my case).
Then there is an inductor in parallel to the tweeter.
Midrange is in series with the tweeter, in my case in phase. Midrange is in parallel with a capacitor.
That is basically a classic series X-Over, like this:
To make it 3-Way you simply make another such crossover, Tweeter and Midrange with their crossover take the "tweeter" position and the woofer is the woofer.
This way you can also calculate the initial crossover in multiple stages with the Bagcby Crossoverdesigner for Excel...
Ciao T
I have no drawing for the schematic even, that I could scan.
Basically it is very simple. A resistor in parallel with a resistor equalises the tweeter (can be omitted if tweeter and midrange have same sensitivity). Another resistor in parallel to this combo adjusts the impedance back to nominal for the speaker (6 Ohm in my case).
Then there is an inductor in parallel to the tweeter.
Midrange is in series with the tweeter, in my case in phase. Midrange is in parallel with a capacitor.
That is basically a classic series X-Over, like this:
An externally hosted image should be here but it was not working when we last tested it.
(from humblehomemadehifi.com)
To make it 3-Way you simply make another such crossover, Tweeter and Midrange with their crossover take the "tweeter" position and the woofer is the woofer.
This way you can also calculate the initial crossover in multiple stages with the Bagcby Crossoverdesigner for Excel...
Ciao T
ThorstenL, thank you so much for taking the time to post your crossover information. My first order equipped two ways sound great to my ears, but I could never figure out how to effectively integrate a subwoofer into my system using a first order network (the inductor would be real big and likely saturate). I'll study your info and see what I can come up with.
I don't think Tannoy had a choice to go with it anywhere else, but sure it turned out right.
- Status
- Not open for further replies.