If you could put the x-over at any frequency without concern for physical constraints, what would be the least damaging x-over frequency for a 2-way or 3-way speaker?
For example, many speakers cross the tweeter over around 2kHz (just an example- may not be correct) because it works best for the drivers even though the ears are most sensitive to anomalies at that range.
Another example, what would be the least damaging when crossing over a bass driver - 50Hz, 200Hz, or 400Hz?
For example, many speakers cross the tweeter over around 2kHz (just an example- may not be correct) because it works best for the drivers even though the ears are most sensitive to anomalies at that range.
Another example, what would be the least damaging when crossing over a bass driver - 50Hz, 200Hz, or 400Hz?
Here's a page with frequency ranges for a lot of instruments https://www.zytrax.com/tech/audio/audio.html
Just my opinion here, but I like to get the crossover out of the vocal band. On the low end I like crossovers around 300hz and on the high end I like them at 5,000. The 300hz crossover gets the bass out of the vocals and lowers distortion and allows you to put more power into the midrange frequencies. 200-300hz is great crossover range for a WAW.
The high crossover of 5,000hz reduces distortion in the tweeter which now does what a tweeter is supposed to do, tweet. 1” Tweeters aren’t midranges. Also, this high crossover requires a midrange that doesn’t beam at 3-5,000hz. They are hard to find so I use so called ‘full ranges’. Now having said that, I’m not the best crossover designer guy, I have limited success making the very popular 6 1/2” woofer to a 1” dome tweeter crossed at 2,500hz type of loudspeaker. I might try again with a waveguide on the tweeter.
I have a not so economical econowave system that is great through the crossover at 1,500hz. So, I’m not sure that frequency matters as much as time/phase, and even power response.
I’m interested to hear what others think.
The high crossover of 5,000hz reduces distortion in the tweeter which now does what a tweeter is supposed to do, tweet. 1” Tweeters aren’t midranges. Also, this high crossover requires a midrange that doesn’t beam at 3-5,000hz. They are hard to find so I use so called ‘full ranges’. Now having said that, I’m not the best crossover designer guy, I have limited success making the very popular 6 1/2” woofer to a 1” dome tweeter crossed at 2,500hz type of loudspeaker. I might try again with a waveguide on the tweeter.
I have a not so economical econowave system that is great through the crossover at 1,500hz. So, I’m not sure that frequency matters as much as time/phase, and even power response.
I’m interested to hear what others think.
For a 2-way:
Small Fullrange + midbass where the midbass‘s low-pass occurs at the lower-end of the pressure-loss region from baffle-step-loss. This really depends on the width of the baffle.
For a 3-way:
Similar idea as the 2-way but with a B&O filter, and with the tweeter low enough and close enough to be within a 1/4 wavelength with the mid..
Small Fullrange + midbass where the midbass‘s low-pass occurs at the lower-end of the pressure-loss region from baffle-step-loss. This really depends on the width of the baffle.
For a 3-way:
Similar idea as the 2-way but with a B&O filter, and with the tweeter low enough and close enough to be within a 1/4 wavelength with the mid..
Gone are the days where it was advised to try and crossover where the ear is least sensitive.
This was primarily said decades ago when good simulation software was not available and measurement systems expensive.
If you were going to try and design a speaker then it was advised to avoid certain areas because if you messed things up there was less to go wrong.
I suppose that still holds true but there are so many compromises involved when you do this that is not worth it.
This was primarily said decades ago when good simulation software was not available and measurement systems expensive.
If you were going to try and design a speaker then it was advised to avoid certain areas because if you messed things up there was less to go wrong.
I suppose that still holds true but there are so many compromises involved when you do this that is not worth it.
Depends on the technology you're using.
If you're making a passive crossover, I'd aim for the 400Hz-800Hz octave. My current speakers crossover at 1kHz, because that's as low as the HF driver will go. The wavelengths are long enough that CtC distancing isn't a huge problem, but component sizes are still pretty sensible. NB - it's perfectly possible to make a decent passive crossover at some other frequency, but 400Hz-800Hz is probably where it's easiest.
If you have some good DSP (FIR with lots of taps etc), then you can be much more flexible. Pretty much anything goes, so long as it makes sense for the drivers.
Chris
If you're making a passive crossover, I'd aim for the 400Hz-800Hz octave. My current speakers crossover at 1kHz, because that's as low as the HF driver will go. The wavelengths are long enough that CtC distancing isn't a huge problem, but component sizes are still pretty sensible. NB - it's perfectly possible to make a decent passive crossover at some other frequency, but 400Hz-800Hz is probably where it's easiest.
If you have some good DSP (FIR with lots of taps etc), then you can be much more flexible. Pretty much anything goes, so long as it makes sense for the drivers.
Chris
Hi,
coffee break: it's interesting question. I read your post that you want to know if a crossover in itself has a sound, and if it has is it audible and is there some bandwidth where it's least audible?
I made simple listening test just for this using web audio api few years back
https://audioexperiment.com/pages/crossover-filters-on-headphones/It's just crossover filter, which combiness back in electrical domain, so doesn't represent a real loudspeaker system as such, but gives perspective if a crossover itself is audible.
Not sure about you guys, but to me there is very little or no audible difference, so if any two crossover filters sound different on real loudspeaker, it's mostly due to how the drivers work and what the resulting acoustic radiation (pattern) is and most importantly how's your room acoustics and how you have positioned the system. How the various sound sources combine acoustically and interact with the environment. Basically sound is what and how the system radiates, while the crossover itself doesn't seem to be an issue of any kind.
Imagine you'd buy best drivers available and implement ideal crossover so that the drivers itself sound as good as they can, work on their linear region. Now only variable is the pattern and how it interacts with the environment, right? For any given speaker ideal crossover is just matter of time and effort, and good judgement to understand what can work and then ability to experiment. Free tools like VituixCAD and allowing oneself to use DSP make crossover a no problem. DSP allows any group delay removed as well. Buying best drivers suitable for the system design is not a concern, they are not toooo expensive, save up for few years.
So, the actual problem is only how well you know your own auditory system, what is it that you want to hear from the playback system. Then, you'd need to know how the room affects things so that you can now understand what kind of a system would work in your room, a system which would provide your auditory system what you want to perceive. Now, when you eventually arrive with a good design it's trivial to make "best" crossover for the thing you've built. In general, I don't see crossover be any kind of a problem, it's just a hurdle, only acoustic design is which you must do before any crossover. If you don't want to spend time on this stuff, or consider it important, then you could buy any drivers and just make optimal crossover for those, for what you have there, optimal for measurements you got from the speakers.
Not sure if I'm clear enough here, crossover of course matters and is part of acoustic design, but things don't start from crossover, that's about the last thing that needs to be done for a loudspeaker. Your underline though is correct, what is audible and what is not, and why, and that's the main nugget. I just don't think crossover is the problem, mainly the listening skill is and ability to reason over it is, then overall system design based on the reasoning, and crossover then falls in place over time. Power of DIY, make something tailor fit to you in your context.
Hope it helps
Last edited:
IME, you are working with different constraints, depending on your budget. By this, I mean what parts you can afford: drivers, primarily, but also crossover components.
At the entry level, you want to select well-behaved drivers, having smooth response and good value for money. Fortunately, there is a lot of shared community information, both anecdotal (user recommendations) and objective (published user measurements). At this level you may not be able to deliver a design that has good power response, but you can definitely work to restrict a driver within its linear range. So, here, your crossover frequency is governed by controlling driver distortion and frequency response along chosen design axis.
At the next level, where you might have a better budget, your choices become quite wide. I would recommend using community "staple" drivers: those that have been proven over time. You will have access to (for a 2-way case) tweeters that have sufficient low-end extension to match the lower driver before beaming sets in. So, then, you can add good power response to your list of achievements. In this case, your crossover frequency was determined by the additional requirement for good power response.
If we stick to conventional drivers, then the next step is multi-way, which will then allow you to achieve better dynamic range and power handling (there are other ways to achieve this, using waveguides, horns, etc.). In this case, you can practically designate each driver to do duty according to the band in which it performs best, as opposed to maximising useable bandwidth, as you would in the lower level categories. But, again, the criteria for choosing the crossover frequency would be different.
At all the described levels you can build a speaker that sounds good and avoids sounding objectionable within the ear's most sensitive range. But, of course, as you progress, the speaker's dynamic limits will set in later; the cheaper speaker being most constrained and the better speakers potentially appearing unconstrained.
IMO, all this only makes sense when you walk the walk. Try it and see. You will never appreciate a next-level performance without discovering the limitations of the prior levels.
At the entry level, you want to select well-behaved drivers, having smooth response and good value for money. Fortunately, there is a lot of shared community information, both anecdotal (user recommendations) and objective (published user measurements). At this level you may not be able to deliver a design that has good power response, but you can definitely work to restrict a driver within its linear range. So, here, your crossover frequency is governed by controlling driver distortion and frequency response along chosen design axis.
At the next level, where you might have a better budget, your choices become quite wide. I would recommend using community "staple" drivers: those that have been proven over time. You will have access to (for a 2-way case) tweeters that have sufficient low-end extension to match the lower driver before beaming sets in. So, then, you can add good power response to your list of achievements. In this case, your crossover frequency was determined by the additional requirement for good power response.
If we stick to conventional drivers, then the next step is multi-way, which will then allow you to achieve better dynamic range and power handling (there are other ways to achieve this, using waveguides, horns, etc.). In this case, you can practically designate each driver to do duty according to the band in which it performs best, as opposed to maximising useable bandwidth, as you would in the lower level categories. But, again, the criteria for choosing the crossover frequency would be different.
At all the described levels you can build a speaker that sounds good and avoids sounding objectionable within the ear's most sensitive range. But, of course, as you progress, the speaker's dynamic limits will set in later; the cheaper speaker being most constrained and the better speakers potentially appearing unconstrained.
IMO, all this only makes sense when you walk the walk. Try it and see. You will never appreciate a next-level performance without discovering the limitations of the prior levels.
The great sound engineers of Cerwin Vega agreed with @Arthur Jackson when they decided on 300hz for the bass on their S1 model(3-way)
To me that tells me that one should look for bass and midd drivers that behave well 1/4?/1/2? Or a whole octave? (I don't know) past this frequency.
Its something I want to explore in some future builds.
I so far have very limited experience apart from modifications.
Cheers!
Edit:
P.S. I imagine better crossover technology. -Both passive and active, has given us more slack since late 70ies? ,when Cerwin Vega S1 came out.
Cheers again! Glob, Glob, Glob.
To me that tells me that one should look for bass and midd drivers that behave well 1/4?/1/2? Or a whole octave? (I don't know) past this frequency.
Its something I want to explore in some future builds.
I so far have very limited experience apart from modifications.
Cheers!
Edit:
P.S. I imagine better crossover technology. -Both passive and active, has given us more slack since late 70ies? ,when Cerwin Vega S1 came out.
Cheers again! Glob, Glob, Glob.
Crossover frequency set at equal to or less than the quarter wave centre-to-centre of the 2 drivers being crossed over.
dave
In any speaker system, you want to avoid harmonic distortion above the 40dB (1%) threshold (0.5% ideally) from 100Hz up and a smooth power response / even DI, aka:
This will help make the speaker sound even in room, avoid tonal balance changes / "bumps and dips" as a listener moves off axis and provide a wider sweet spot when listening.
The driver selection and crossover point then comes down to achieving these goals (peak SPL desires aside)
This will help make the speaker sound even in room, avoid tonal balance changes / "bumps and dips" as a listener moves off axis and provide a wider sweet spot when listening.
The driver selection and crossover point then comes down to achieving these goals (peak SPL desires aside)
For me that answer is at least 3 different answer´s, depending on your listeninglevel "max SPL", and roomsize. (to simplify what I mean)
1) If you okay with max 85-90 db listeninglevel and also a speaker that "go/starts" from 35-40 hz, this DIY have many many 1000 solutions, because most drivers is capable of this task "in a good way" (2 way zone, prefer conetype paper etc, xover at about 1600-1800 hz , maby lower if you use cd and the driver has to cover a large frequency range, which limits things)
2) If you insteed need 105-110 db max listeninglevel and still a speaker that "go/starts" from 30-35 hz, then the supply of drivers that can handle this task well becomes much smaller, and now you have to spending time finding suitable drivers & simulating these carefully as well as also calculating box volume etc (3 way zone prefer conetype paper,mms etc, xover often at around 300, 2500 hz, and the driver has to cover a smaller frequency range, which limits things less)
3) A DIY speaker that can handle +115 dB and play from 20-25 hz and do all that in a very good way, now you have 10 times more challenge than 105-110 dB diy-speaker, and 100 times more challenge then 85-90 dB diy speaker. Now you really have to look for the possibly few drivers that would handle this in a very good way, & calculate & simulate exactly EVERYTHING that is included. ( 3 way zone or more, and have to use what conetype and mms etc that exist on the market, xover at and the drivers has to cover a smaller frequency range, which limits things "less", but this is a real hard task)
In addition to this, there is the problem of beaming, which is sometimes a really annoying "sound phenomenon", & sometimes not even perceived by the listener.
It's about how high up in frequency a driver can be used before it can start sounding like a megaphone, and what affects Beaming is the size of the cone surface in combination with the angle of the speakers as well as the room & listening position in the room.
So this can limit you a lot when choosing the size of drivers for the various examples above.
Regards John
1) If you okay with max 85-90 db listeninglevel and also a speaker that "go/starts" from 35-40 hz, this DIY have many many 1000 solutions, because most drivers is capable of this task "in a good way" (2 way zone, prefer conetype paper etc, xover at about 1600-1800 hz , maby lower if you use cd and the driver has to cover a large frequency range, which limits things)
2) If you insteed need 105-110 db max listeninglevel and still a speaker that "go/starts" from 30-35 hz, then the supply of drivers that can handle this task well becomes much smaller, and now you have to spending time finding suitable drivers & simulating these carefully as well as also calculating box volume etc (3 way zone prefer conetype paper,mms etc, xover often at around 300, 2500 hz, and the driver has to cover a smaller frequency range, which limits things less)
3) A DIY speaker that can handle +115 dB and play from 20-25 hz and do all that in a very good way, now you have 10 times more challenge than 105-110 dB diy-speaker, and 100 times more challenge then 85-90 dB diy speaker. Now you really have to look for the possibly few drivers that would handle this in a very good way, & calculate & simulate exactly EVERYTHING that is included. ( 3 way zone or more, and have to use what conetype and mms etc that exist on the market, xover at and the drivers has to cover a smaller frequency range, which limits things "less", but this is a real hard task)
In addition to this, there is the problem of beaming, which is sometimes a really annoying "sound phenomenon", & sometimes not even perceived by the listener.
It's about how high up in frequency a driver can be used before it can start sounding like a megaphone, and what affects Beaming is the size of the cone surface in combination with the angle of the speakers as well as the room & listening position in the room.
So this can limit you a lot when choosing the size of drivers for the various examples above.
Regards John
Sorry all for the delay. Some posting issue they cleared it up now.
I'm a layman. Hope this makes more sense.
Yes, something like this. Putting driver limitations aside, trying to squeeze the crossover into the least active part of the frequency spectrum. Like for bass 200Hz is above bass but below vocals. So trying to maintain coherence for the bass / midrange / treble by letting each driver in those areas run without obstruction.
I'm a layman. Hope this makes more sense.
This sounds like it might be right, but I’ve no idea what it means! Please explain 😁