My way to determine the optimum crossover frequency was based on physiology. In the medium-treble zone, for non coincident drivers, there is failure of human perception in the direction of sounds around 1800 Hz. Years ago, I discovered that speakers having a crossover around 1900 Hz give a very precise stereo image. So I am currently using 2000 Hz, the tweeter being a ScanSpeak resonating et 500 Hz.
For the bass, I found that, using a sine generator, there quite a change of the nature of sound under 140 Hz (which seems to belong to bass zone) and over 160 Hz (which seems to belong to the medium zone). After having set the crossover at 140 Hz for long, I changed to 160 Hz. The medium driver is a ScanSpeak 17 cm Revelator resonating at 50 Hz in a quite big box and the bass driver is a Peerless 31 cm XLS where an accident in the frequency response at around 450 Hz should be sufficiently attenuated.
Crossover sloped are 18 dB/o and the hi-pass behaviour of the drivers is taken in account.
I mainly choose my drivers according to the crossover frequencies which I think to be the best for the human perception. It's quite the contrary of the usual way, choosing the drivers first, and then looking for the optimal crossover frequencies.
For the bass, I found that, using a sine generator, there quite a change of the nature of sound under 140 Hz (which seems to belong to bass zone) and over 160 Hz (which seems to belong to the medium zone). After having set the crossover at 140 Hz for long, I changed to 160 Hz. The medium driver is a ScanSpeak 17 cm Revelator resonating at 50 Hz in a quite big box and the bass driver is a Peerless 31 cm XLS where an accident in the frequency response at around 450 Hz should be sufficiently attenuated.
Crossover sloped are 18 dB/o and the hi-pass behaviour of the drivers is taken in account.
I mainly choose my drivers according to the crossover frequencies which I think to be the best for the human perception. It's quite the contrary of the usual way, choosing the drivers first, and then looking for the optimal crossover frequencies.
I would say the best way to go about it is to determine what type of directivity you want in your speaker. IMO the only options to consider right now are:
1) True omni
2) True dipole (more like Note 2 RS not Orion)
3) CD e.g. Synergy horn or Autotech OS waveguide
Then from there you know which types of drivers you need to use based on the requirements of the directivity, and in what ranges.
1) True omni
2) True dipole (more like Note 2 RS not Orion)
3) CD e.g. Synergy horn or Autotech OS waveguide
Then from there you know which types of drivers you need to use based on the requirements of the directivity, and in what ranges.
Your options assume an environment where such speakers can be used. Quite the modern thought, but not everyone has that luxury. Not really part of a crossover point selection guideline though.
forr,
Selecting the drivers that meet your design goals is one of the few really smart things said in 24 posts. THEN design the crossover that is optimum for those drivers.
I don't quite get your correlation with physiology and crossovers. Have you done these same tests with many different drivers, or for these drivers that is where your perception tells you they are happy? For the drivers you mention, these seem very reasonable values, but if you were to try the listening tests with say, Vifa XT-25, you may find a very different result in they really do not like being used that low.
My most "precise" imaging speakers cross over at 2600. (Dayton/Seas). The original tweeters were Vifa's at a 1650 crossover and they were horrid. No imaging, and just plain ugly sounding. Miss-use of a less than optimum tweeter trying to avoid woofer breakup issues. My bad. All better now.
Selecting the drivers that meet your design goals is one of the few really smart things said in 24 posts. THEN design the crossover that is optimum for those drivers.
I don't quite get your correlation with physiology and crossovers. Have you done these same tests with many different drivers, or for these drivers that is where your perception tells you they are happy? For the drivers you mention, these seem very reasonable values, but if you were to try the listening tests with say, Vifa XT-25, you may find a very different result in they really do not like being used that low.
My most "precise" imaging speakers cross over at 2600. (Dayton/Seas). The original tweeters were Vifa's at a 1650 crossover and they were horrid. No imaging, and just plain ugly sounding. Miss-use of a less than optimum tweeter trying to avoid woofer breakup issues. My bad. All better now.
Indirectly it is, since the crossover point(s) should be the lowest + highest point that still approximates the directivity pattern goal of the speaker.
I respect your focus on something that is all too frequently not addressed, probably because it is hard. You are absolutely correct that poor decisions in the crossover can make it impossible to meet a design goal. But I suggest it is equally or more important to temper those goals with keeping away from breakup regions, resonance problems, energy decay, harmonic distortion, power compression and the rest. If you can't find happy ground with all the parameters, you will not be happy with the result. Pick a new design or different drivers. Sometimes, you just plain can't do what you want to.
Of course, optimizing for directivity assumes that the drivers can operate in the crossover region without any major issues.
You seemed to know the issues(?), but then your question is strange. So should we assume that the designer is 100% capable and the drivers are 100% perfect? Then may be the right answer for the question is that you can cross over at any frequency that you want.
But wait, then you will find out that each driver have their "sweet spot". But what if the driver is so 100% perfect so that every frequency is sweet spot?
BTW, as a rule of thumb, my style is to focus on the sweet spot, and try very hard to solve all problems associated with it.
Life is difficult only if you don't make too much assumptions. In general, with 2-way, common issue is that woofer cannot go too high and tweeter cannot go too low. Then you choose your compromises. And there you have your style.
First assumption is that everyone know that it is good to simulate two drivers before purchasing them, so to ensure better match. But we cannot know fully 100% without hearing them, so that is where I based on my decision to choose the optimum Fx.
Drivers can match in any frequency points or bands. If you limit your frequency to say 2K4 (or any frequency for reasons you are finding out, which I don't know the answer for) you may have difficulty to find a match drivers. Just buy the flattest drivers then (which are expensive), if you think that matching is an easy thing.
Then you must try to cross as low as possible, with the consequence of having to buy expensive tweeter. The higher you cross, the harder it is to do it right (in general), but the benefit is you can get good sound with cheap tweeter.
Do every options in computer. Purchase the driver, do a quick examinations of the drivers real capability. Do every options again by prototyping. Choose your poisons/compromises.
A nice case example is an XT25. You cannot easily see how bad this tweeter handles low frequency from simulation. But once you know its character, then it is your call how do you want to cross this tweeter. I have never paired an XT25 with big woofer (such as 8"). The tweeter has limitation below 3K, but the woofer will usually have limitation above 2K. So pick the best of the bad conditions.
I think, there are too many issues that choosing frequency is not the right thing to do as the first step in building a speaker. But of course it can be part of anyone design style to avoid issues from the first place. More likely if we have less confidence in recognizing and solving the issues.
Really? Check the sensitivity curves, you will be surprised.. It's not because some people have done some proeminent research in some fields that everything they say is true. The 100-500 area is VERY important.. The people who say the opposite have an agenda. Have you seen a waveguide managing 200hz? It's possible, but it would be huuuuuge. Not practical, difficult to sell..
BTW, where is the research saying that the highs need beaming? I believe the large Archimedes project held in Denmark 20 years ago came to the exact opposite conclusion..
Last edited:
Very definitive statement in a very undefined area, care to elaborate a bit?
It would exclude the use of Tannoy DualConcentrics and ATC 3" domes for ever, two drivers I personally am very fond of.
Indeed it is...very important. That's not what I said. We are less sensitive to diffraction and to the imaging being skewed by reflections at these frequencies. Reflections are still a problem, and not constraining these frequencies in a narrow pattern still leaves us to clean up the colourations.
It has been shown that diffraction and early reflections cause issues especially at higher frequencies. These define the choice to constrain the highs.
Those points have been the professional choice since the 50's, give or take a few Hz either way.
Not at all. I would say your prescription errs heavily on the high side, but it all depends on driver size, driver behaviour, filter order, desired dispersion, just to name a few.
Yep, similar to XT25, the Revelator has that unreasonable low crossover capability. I think even if the size of the dome is big, physically it is not possible for any dome to produce wide frequency band well. After XT25 and Revelator, I don't believe in wideband tweeter.
ATC has wide dome but for limited band use only (midrange). That's probably the best midrange.
I doubt that somehow.
Tannoy DualConcentrics have been a popular professional choice since the '50s.
Being 2ways the crossover points depend on the driver size and are 1000Hz for 15", 1200Hz for 12" and 1400Hz for 10".
ATCs are a very popular professional choice since the '70s. Crossover points are almost invariably 380Hz and 3800Hz.
Klein&Hummel/Neumann also use ATC 3" domes for mids with 600Hz & 2k points and Quested as well as PMC also use 3" domes (Quested use Volt VM752, PMC their own design) with almost identical crossover points to ATC.
RE Geithain use their own drivers AFAIK, crossover points depend on model and are either 550 and 2200Hz, 550 & 2.8k, 250Hz & 2.6kHz, 650 & 2.8k, 700 & 3.2k etc.
It all depends on the actual drivers used.
The ATC has been mentioned in these pages before. It's not widely available here and a bit on the steep side. Sure would like to play with a pair. I have one problem with midranges. Just too big to fit on a 17 inch high baffle with a 6 1/2 inch woofer and dome tweeter. You are forced to a bigger box. I am looking at smaller tweeters.
The reason why I like to take the crossover into consideration before anything else is because it will aid in choosing the right drivers to fit the design goal.
Personally I'm open for all sorts of ideas.
I could just as well opt for active digital crossover with a fullrange driver + woofer.
Add some FIR filters and maybe some ideas from Linkwitz to extend the bass.
Still I enjoy understanding the math and the mechanics behind it all.
No point in using a lot of high tech stuff to put a band aid on a poor design if I can do it right from the beginning.
Personally I'm open for all sorts of ideas.
I could just as well opt for active digital crossover with a fullrange driver + woofer.
Add some FIR filters and maybe some ideas from Linkwitz to extend the bass.
Still I enjoy understanding the math and the mechanics behind it all.
No point in using a lot of high tech stuff to put a band aid on a poor design if I can do it right from the beginning.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.