The first part of this statement has no foundation in anything we know about speakers and the way we perceive sound. Therefore, if 2.5 kHz is the lowest your tweeter will play right, use that frequency, and if it is 3 kHz, use that. If it is higher than that, find a better tweeter.
Is there any scientific research showing that the ear is most sensitive to crosssover artefacts in the frequency range where it is most sensitive in term s of SPL ?
Regards
Charles
Regards
Charles
What do you mean precisely? Isn't Fletcher Munson is sufficient?
By logic and feeling alone I think it is understandable that we can recognize sound existence of the same SPL more easily if the frequency is not too high and not too low, and we cannot even hear it when the frequency is below or beyond certain frequencies.
So if we can hear most easily sound of 3kHz to 4kHz, then we can also recognize more easily if there is something wrong within that frequency. Regarding tonality, if we are not familiar with an instrument, wrong tonality can be acceptable, but everyone will more easily recognize voices tonality.
Many musics have SPL peaks [either vocal or instrument] at certain frequencies, I think 1kHz to 2kHz. Many woofer cones or tweeter domes have deficiency/breakup around this frequency...
So will we add crossover mistake to that?
I think it depends on the trade offs available. Our skill as the crossover designer is one of the most important trade off. If we aren't sure, I think full range option using e.g. ScanSpeak 10F which costs less than $100 is an option. No need tweeter, only one crossover at a point that is suitable with the chosen woofer.
If the above option is not acceptable in term of sound quality, then find better midrange. If too expensive or cannot find better midrange, take the risk or improve the skill 😀
What does Fletcher Munson teach us exactly ? IMHO it is showing averaged curves of hearing sensitivity vs frequency and SPL. The peak in the ear's response is mostly due to a resonance effect of the ear canal BTW.
But aren't the most relevant spectral parts below that ? I.e isn't the biggest part of musical (and other information) taking place below that ?
And are hearing sensitivity in terms of SPL and sensitivity to crossover artefacts really interrelated in such a simplistic way ?
Regards
Charles
https://www.perrymarshall.com/articles/ultimate2way/
"The oft-referenced Fletcher-Munson curves show that we hear best in the 1000 Hz – 5000 Hz band. Any anomalies in the crossover range become blazingly obvious. Moving the crossover out of that range reduces the severity of the problem."
Marco
Yes, most part of music is below that. But the point is...
When crossover mistake happens at LF such that there is a 3dB above reference, you may like the result. Now try 3dB above reference at 1k5! Even when it is flat [0dB from reference] the sound can hurt you [because it is a musical peak, and many drivers cannot cope with it].
Peak at HF hurts your ears, peak at LF hurts your stomach? 😀
Try piano keys from LF to HF at high SPL [measured at amp level]. Which one hurts your ears the most?
I think I understand your thinking. May be it is not that simple. Because wherever the mistake is, I still cannot accept it. Subwoofer integration is an example. I prefer no extra LF than poor integration.
But if we had to choose only to listen to 2 perfect octaves, which band will you choose for your speaker? 400-1k6? I think we want "this" band to be as perfect as possible.
What a crossover can do to natural sound is to change the relative phase. For certain sound that have wide overtones, we want them to be in the same phase. FR flatness can be achieved/maintained but phase coherence or delay is the price to pay for using crossover.
I'm one of those who cross mostly at critical band. I don't subscribe to this critical band dogma.
Again I remind over an instrument's own overtones, eg. violin has most differences in sound above 2kHz
http://www.akutek.info/Papers/AB_Violin_spectra_2003.pdf
http://articles.ircam.fr/textes/Vos03a/index.pdf
http://www.akutek.info/Papers/AB_Violin_spectra_2003.pdf
http://articles.ircam.fr/textes/Vos03a/index.pdf
I've been experimenting with some things lately, and if you make an acoustic lp filter for the rear output the piston diameter takes over the directivity a little before you reach 1wl dia.
The 12P80 seems a good match with the TPL150H as it appears to be very well behaved regarding cone break up all the way up to 2k.
The one thing I don't understand is why people pick the Nd version for home use since the Ferrite version shows lower distortion and is cheaper to boot.
Going by curves published by Beyma and BMS it appears to be normal that ferrite versions have less THD than their Nd brethren.
Makes no sense to me using Nds unless it is for a touring PA where weight equals money.
the fletcher curve is logical enough for me. do not put a crossover in the region where the ears is the most sensitive.
especially for DIY designer who may have trouble making the most transparent crossover.
Make a 2-way speaker with a tweeter that can be crossed under 2khz. (but then be sure that you dont require very high SPL)
Make a 2-way speaker with a mid bass that can cover up to 5-6-7 khz and cross it to a tweeter or super-tweeter.
Make a 3-way with a mid like a audax pr170 that can cover 400hz to 5-6khz...
ect.
if you really must put a xo in the 2-5khz region, you better be very good at XO design.
Last edited:
@ ScottG : thanks for the links, my tonight readings 🙂 !
@ Vaccuphile : ok, thank you for the precision, it's more what I learn from books and forums... but did you read the thread from Angeloitacare about a 12" in a sealed with a full range use : 50 hz to ...1600 hz !
and it seems to work pretty good ! (CharlesDarwin testimonial just above - thanks Charles (maybe they take the nd for the low inductance of the coil?).
@ Vaccuphile : ok, thank you for the precision, it's more what I learn from books and forums... but did you read the thread from Angeloitacare about a 12" in a sealed with a full range use : 50 hz to ...1600 hz !
and it seems to work pretty good ! (CharlesDarwin testimonial just above - thanks Charles (maybe they take the nd for the low inductance of the coil?).
Last edited:
Eldam, I have not seen that design. At such a high xover frequency, I suppose a narrow waveguide is employed. That kind of speaker usually creates a fairly small sweet spot.
Sometimes something that sounds logical just isn't true in reality and so it is in this case.
My personal opinion is that any crossover is generally going to be worse than a single driver reproducing that range(assuming the single driver was only driving the crossover range, not the whole range of both drivers). Even in a perfect crossover, the phase change, the change in directivity, and the position of the driver being different are unavoidable. I have heard Linkwitz's LX521's at his home and the only thing I could notice in the sound that I thought could possibly be improved was a first order crossover from an 8inch lower mid to a 4 inch upper mid at 1khz. The main thing I noticed was that it just seemed like the two halves of the midrange were coming from different places. Now I don't mean to talk negatively of the LX521's, by far they were the absolute best speakers I have ever heard, but this was the one thing I noticed, and Mr. Linkwitz chose to have that crossover for a reason. It gave the tradeoff of being able to push the lower crossover down to 120Hz and push the upper crossover to 7.5kHz, and therefore have the lower mid handled by an 8inch that was phenomenal from 120hz to 1kHz, while relieving the smaller mid of the excursion heavy lower mid and having it focus on the 1kHz to 7.5kHz range where it shines. Also it maintains constant directivity and acts like a point source.
I started a rant there to prove a point. While yes I believe the 1kHz crossover in linkwitz's speakers was noticeable and degraded the sound. But the advantages he gained from incorporating a crossover smack in the midrange were phenomenal and still made for the best speakers I have ever heard. I think the no-crossover-in-the-midrange rule is a simplification of a wide range of tradeoffs that are made when designing speakers, and yes, at all times crossovers in the midrange should be avoided IMO. But it is clear that having a crossover in the midrange frees up many other opportunities to improve the sound in other ways, like improving directivity and lowering the distortion of each driver by having drivers that handle each band more effectively.
Just as a sidenote, one thing I plan to try is a line array of 3-4 inch full ranges that can therefore effectively handle the bass down to around 100hz where it can cross to some woofers and up to 5-10khz where it can cross to a tweeter. In my opinion this is a good way to avoid the crossover in the midrange, while still maintaining a high sensitivity, good directivity at all frequencies and low distortion. Though some would say the cost of the extra drivers isn't worth the advantages, we again get back into the fact that every speaker has compromises, you just have to pick which ones you can live with, and that is a completely subjective opinion.
I started a rant there to prove a point. While yes I believe the 1kHz crossover in linkwitz's speakers was noticeable and degraded the sound. But the advantages he gained from incorporating a crossover smack in the midrange were phenomenal and still made for the best speakers I have ever heard. I think the no-crossover-in-the-midrange rule is a simplification of a wide range of tradeoffs that are made when designing speakers, and yes, at all times crossovers in the midrange should be avoided IMO. But it is clear that having a crossover in the midrange frees up many other opportunities to improve the sound in other ways, like improving directivity and lowering the distortion of each driver by having drivers that handle each band more effectively.
Just as a sidenote, one thing I plan to try is a line array of 3-4 inch full ranges that can therefore effectively handle the bass down to around 100hz where it can cross to some woofers and up to 5-10khz where it can cross to a tweeter. In my opinion this is a good way to avoid the crossover in the midrange, while still maintaining a high sensitivity, good directivity at all frequencies and low distortion. Though some would say the cost of the extra drivers isn't worth the advantages, we again get back into the fact that every speaker has compromises, you just have to pick which ones you can live with, and that is a completely subjective opinion.
why put a xo in the most critical band and where the ear is the most sensitive?
for diyers, it can help a lot to design your speaker with this in mind.
+1
Last edited:
But it's expensive as you said; it could be a good idea to make a speaker à la Wesayso but with high efficienty FR like the PRV Audio 5" neodynium serie... but the price of the wood cabinet and multiple drivers says to me to jump to a no-array solution !
maggiesnmacs gave some very good reasons why you would want to do so, plus a testimonial that a specific loudspeaker thus crossed over was one of the finest he had ever heard.
As a besides, did you know that distortion is usually lowest around xover points? Reason is that the drivers are each 6dB down @ xover f. Therefore, to minimize distortion in the ear's most sensitive region, that is exactly where you should xover.
It depends on your xover topology how far each driver is down.
L-R ones are 6dB down at the crossover point but a textbook Butterworth puts the drivers only 3dB down.
Consequently and presuming drivers with a linear response past their passband Butterworths produce a 3dB hump in the FR which should be particularly audible in the higher mid range.
May be the dogma at hand has its roots in a time before L-Rs became popular and Butterworth was the preferred alignment?
I haven't read all 10 pages of entries here, so forgive me if this has already been said, but here goes.
Setting aside the fact that the ear is most sensitive to amplitude from about 800HZ to 6kHZ, there's another issue.
Inter-aural crosstalk confuses the brains ability to locate stereo images below about 1kHZ. Room reflections can to some extent re-create a sense of space in that region of frequency (especially with open baffles), but only so good, it's fake, and it's always the same unless you change rooms. So the majority of real stereo imaging happens in the upper-midrange frequencies, arguably from about 1kHZ - 6kHZ, according to the fairly extensive research I've done.
So for best stereo effect imaging, you want perfect balance at all frequencies in this frequency region, 1kHZ - 6kHZ.
If you have a precise active or passive crossover at 2.5kHZ, for example, both sides perfectly matched (not likely but just suppose), there's still a problem, it's called the room. This is because, as phases change in the crossover frequency region, the angle of maximum amplitude projection will sweep (vertically in the case of drivers being mounted vertically) because of the way the 2 signals add with their differential phase shift. Reflections off the typical 9 foot ceiling will exacerbate this.
This may not matter in an anechoic chamber, but in a typical living room, the acoustics may affect one side differently than the other. Not a huge deal, but this can be significant depending on the room. The acoustics of the room cause a small amount of imbalance in the frequency region where both drivers are significantly putting out the same energy. If you have one pole crossovers, you might have imaging damage over 2 octaves. With a 4 pole active crossover, maybe only a half octave. 2 octaves can be most of this upper-mid frequency range that we don't want to damage. If X=2.5kHZ, imaging may be damaged from 1.2kHZ to 5kHZ with a simple one pole crossover. With an accurate active 4 pole, maybe only 2kHZ to 3kHZ. That's still not very good.
In the real world of passive crossovers, there will be more sloppiness. You won't even need a room, to have a differential beaming sweep issue. Are the impedance curves of drivers identical from driver to driver; batch to batch, and over time? I doubt it. But putting this more in perspective, most room acoustics will damage this balance regardless of what you do, even with no crossover points, so take it with a grain of salt.
But this is why I try to avoid putting a crossover between 800HZ and 7kHZ.
Setting aside the fact that the ear is most sensitive to amplitude from about 800HZ to 6kHZ, there's another issue.
Inter-aural crosstalk confuses the brains ability to locate stereo images below about 1kHZ. Room reflections can to some extent re-create a sense of space in that region of frequency (especially with open baffles), but only so good, it's fake, and it's always the same unless you change rooms. So the majority of real stereo imaging happens in the upper-midrange frequencies, arguably from about 1kHZ - 6kHZ, according to the fairly extensive research I've done.
So for best stereo effect imaging, you want perfect balance at all frequencies in this frequency region, 1kHZ - 6kHZ.
If you have a precise active or passive crossover at 2.5kHZ, for example, both sides perfectly matched (not likely but just suppose), there's still a problem, it's called the room. This is because, as phases change in the crossover frequency region, the angle of maximum amplitude projection will sweep (vertically in the case of drivers being mounted vertically) because of the way the 2 signals add with their differential phase shift. Reflections off the typical 9 foot ceiling will exacerbate this.
This may not matter in an anechoic chamber, but in a typical living room, the acoustics may affect one side differently than the other. Not a huge deal, but this can be significant depending on the room. The acoustics of the room cause a small amount of imbalance in the frequency region where both drivers are significantly putting out the same energy. If you have one pole crossovers, you might have imaging damage over 2 octaves. With a 4 pole active crossover, maybe only a half octave. 2 octaves can be most of this upper-mid frequency range that we don't want to damage. If X=2.5kHZ, imaging may be damaged from 1.2kHZ to 5kHZ with a simple one pole crossover. With an accurate active 4 pole, maybe only 2kHZ to 3kHZ. That's still not very good.
In the real world of passive crossovers, there will be more sloppiness. You won't even need a room, to have a differential beaming sweep issue. Are the impedance curves of drivers identical from driver to driver; batch to batch, and over time? I doubt it. But putting this more in perspective, most room acoustics will damage this balance regardless of what you do, even with no crossover points, so take it with a grain of salt.
But this is why I try to avoid putting a crossover between 800HZ and 7kHZ.
- Status
- Not open for further replies.