Hello to everyone, this is my first thread (second after the Introduction), I'm new to speakers design so i'm still adapting to new knowledges.
Well, the title says it all, I've looked over the web and this forum and I can't find something about this topic, at less not something that someone at my level could understand.
Most crossover designs i see are done with the graphic responses provided by the factory/vendor, if i'm not wrong these test are done in a something like a "infinite baffle" or free air... but as low as you go in frequencies, subwoofers, bass woofer or low mids the higher impact will have the enclosure design in the speaker response.
So... how? when? I've seen some softwares that allows to simulate the speakers response on a given enclosure, you simulate that response and then proced to design the corssover with a new dbspl? is there any other method? closed boxes are usually much simpler and predictable, the procedure will be same if it is ported, closed, reflex? Can you share what is your approach on this matter?
Well, the title says it all, I've looked over the web and this forum and I can't find something about this topic, at less not something that someone at my level could understand.
Most crossover designs i see are done with the graphic responses provided by the factory/vendor, if i'm not wrong these test are done in a something like a "infinite baffle" or free air... but as low as you go in frequencies, subwoofers, bass woofer or low mids the higher impact will have the enclosure design in the speaker response.
So... how? when? I've seen some softwares that allows to simulate the speakers response on a given enclosure, you simulate that response and then proced to design the corssover with a new dbspl? is there any other method? closed boxes are usually much simpler and predictable, the procedure will be same if it is ported, closed, reflex? Can you share what is your approach on this matter?
Always build the enclosure first. Measure in your enclosure, not manufacturer data or subwoofer enclosure software. Now you may begin your crossover design. This article from a real expert explains it very well:
Crossovers
Crossovers
Enclosure simulation software can only predict the low frequency response where the driver acts as a near-perfect piston. This is where you put the electro-mechanical (T/S) parameters of a driver (as measured by the manufacturer and published in the datasheet) into a CAD program that predicts the low frequency response in a given enclosure.
At higher frequencies where the cone 'breaks up' into different resonant modes, there is no end user software which can accurately model the response. Not all manufacturers follow the same measurement procedures to obtain a frequency response to put in the datasheet, so you should take them with a grain of salt. It is better to actually measure the driver you have received yourself if you have the capability. You can measure T/S parameters too with a DATS, or REW and a soundcard, as those vary with different manufacturing batches as well.
Definitely measure in the actual enclosure for high frequencies (>1kHz), where edge diffraction effects will be significant.
Measuring low frequencies is a whole separate kettle of fish, especially if you don't have access to a very large room or open outdoor space.
If you have the space to avoid reflections (several meters open space all around the speaker), making a quasi-anechoic measurement of a speaker is pretty easy. A ground-plane measurement (mic level with the ground) can be performed for the low frequencies while an on-axis measurement (at listening height) can be performed for the higher frequencies. Mic should be placed at the intended listening distance from the speakers. Matching the levels is easy - just make sure the distance from the drivers to the mic is the same for both measurements. It should be obvious where you should cross your measurements - the on-axis measurement will develop dips at low frequency due to reflections off the ground, while the ground plane measurement will become in-accurate at high frequencies due to being placed off-axis to the drivers. Both measurement need to be time-gated to avoid reflections as much as possible without overly-smoothing the response.
If you only have access to a small indoor space, reflections from nearby walls/objects become too problematic to achieve an accurate low frequency response with the mic placed away from the speaker. To avoid reflections, you'll have to do a near-field measurement (mic very close to speaker driver) to capture the low frequencies without significant room effects. Then you need to apply a baffle diffraction sim to the near-field measurement since baffle diffraction does not occur in the near-field. For a ported/reflex enclosure, you'll have to measure both the woofer and the port and sum the measurements.
High frequencies can still be captured far-field as gating will remove reflections from the measurement with the mic ~1m and walls/objects more than 1m away from the speaker and mic. Expect to cross the measurements somewhere between a few hundred Hz to 1kHz.
Matching levels of near-field and far-field measurements is somewhat of an artform if using this method as it depends on the size of the drivers and the mic distances. Some sims and estimations may be required. If using this method for the first time, expect that you may get the amount of baffle step compensation wrong and the speaker may be too light or heavy on bass, requiring a crossover design tweak.
At higher frequencies where the cone 'breaks up' into different resonant modes, there is no end user software which can accurately model the response. Not all manufacturers follow the same measurement procedures to obtain a frequency response to put in the datasheet, so you should take them with a grain of salt. It is better to actually measure the driver you have received yourself if you have the capability. You can measure T/S parameters too with a DATS, or REW and a soundcard, as those vary with different manufacturing batches as well.
Definitely measure in the actual enclosure for high frequencies (>1kHz), where edge diffraction effects will be significant.
Measuring low frequencies is a whole separate kettle of fish, especially if you don't have access to a very large room or open outdoor space.
If you have the space to avoid reflections (several meters open space all around the speaker), making a quasi-anechoic measurement of a speaker is pretty easy. A ground-plane measurement (mic level with the ground) can be performed for the low frequencies while an on-axis measurement (at listening height) can be performed for the higher frequencies. Mic should be placed at the intended listening distance from the speakers. Matching the levels is easy - just make sure the distance from the drivers to the mic is the same for both measurements. It should be obvious where you should cross your measurements - the on-axis measurement will develop dips at low frequency due to reflections off the ground, while the ground plane measurement will become in-accurate at high frequencies due to being placed off-axis to the drivers. Both measurement need to be time-gated to avoid reflections as much as possible without overly-smoothing the response.
If you only have access to a small indoor space, reflections from nearby walls/objects become too problematic to achieve an accurate low frequency response with the mic placed away from the speaker. To avoid reflections, you'll have to do a near-field measurement (mic very close to speaker driver) to capture the low frequencies without significant room effects. Then you need to apply a baffle diffraction sim to the near-field measurement since baffle diffraction does not occur in the near-field. For a ported/reflex enclosure, you'll have to measure both the woofer and the port and sum the measurements.
High frequencies can still be captured far-field as gating will remove reflections from the measurement with the mic ~1m and walls/objects more than 1m away from the speaker and mic. Expect to cross the measurements somewhere between a few hundred Hz to 1kHz.
Matching levels of near-field and far-field measurements is somewhat of an artform if using this method as it depends on the size of the drivers and the mic distances. Some sims and estimations may be required. If using this method for the first time, expect that you may get the amount of baffle step compensation wrong and the speaker may be too light or heavy on bass, requiring a crossover design tweak.
Last edited:
Tango,
yes published FR graphs are almost always infinite baffle response. When you put a driver in a baffle the FR will be modified by the baffle step and diffraction effects. The kind of enclosure itself will modify the bass only, but it is pretty straightforward to modify the infinite baffle FR to add the enclosure effect. This sometimes doesn't event need to be done: if you are developing a 2-way crossed say at 2KHz, what happens below 200Hz is not important.
So before your crossover design you really only need to simulate what happens to the published FR on the chosen baffle, i.e. baffle step and diffraction. This can be done with specific tools, or using a crossover simulator that has also the baffle simulator. I think that boxsim and vituixcad are able to.
For an overview of the whole process look for the pdf at "How to Design Loudspeakers without Performing Measurements": software
Ralf
I agree that doing your own measurement is always better, but proper simulations can be very close to reality, and help in deciding beforehand what to build with what drivers.
yes published FR graphs are almost always infinite baffle response. When you put a driver in a baffle the FR will be modified by the baffle step and diffraction effects. The kind of enclosure itself will modify the bass only, but it is pretty straightforward to modify the infinite baffle FR to add the enclosure effect. This sometimes doesn't event need to be done: if you are developing a 2-way crossed say at 2KHz, what happens below 200Hz is not important.
So before your crossover design you really only need to simulate what happens to the published FR on the chosen baffle, i.e. baffle step and diffraction. This can be done with specific tools, or using a crossover simulator that has also the baffle simulator. I think that boxsim and vituixcad are able to.
For an overview of the whole process look for the pdf at "How to Design Loudspeakers without Performing Measurements": software
Ralf
I agree that doing your own measurement is always better, but proper simulations can be very close to reality, and help in deciding beforehand what to build with what drivers.
Thank to all of you for your responses; to anyone who is at my newbie stage it's highly recomended to read the link shared by cyberstudio, while it's seems pretty basic it's realy good straight info from experienced people (thank you for that) It's seems that all of you consent that for any god result there is no replace to take propper measurements with the speaker on it's respective baffle
Thanks TTM for your explanation, most of us try to get into this world and maybe the last thing we take into acount or read it's about how to do propper measurements, your reply it's for sure a good place to start reading and learning more on the subject that it seems to be a fundamental step in order to achive good results on our projects.
Giralfino, i'll be following your advices, (i haven't read the PDF, already bookmark it so I can read it tonight) i've been using Xsim and Vituix and it's mostly the area bellow 500 hz where i was initially concern but actually is the low ends what seems to be more initially easy to predict, i was going with closed baffle design because even that i really like deep bass on music, kick drums, deep well defined bass lines i litteratly hate when the low end bass "extends" and gets loose, that low ends that you "feel" all the time even you can't hear it... but i also find awfull when you can't get that "punch" in your stomach when you hear a good kick drum. Being low frequencies less directional and easier to "spread", it's seems that a theorical approach in this area at first stages it can be more accurate than with the mids/highs where the frequencies are more directional and diffraction starts to be really important, and after all it's where music comes "alive". Thank very much for your time guys, you gave me a better idea on the path that i should follow.
Thanks TTM for your explanation, most of us try to get into this world and maybe the last thing we take into acount or read it's about how to do propper measurements, your reply it's for sure a good place to start reading and learning more on the subject that it seems to be a fundamental step in order to achive good results on our projects.
Giralfino, i'll be following your advices, (i haven't read the PDF, already bookmark it so I can read it tonight) i've been using Xsim and Vituix and it's mostly the area bellow 500 hz where i was initially concern but actually is the low ends what seems to be more initially easy to predict, i was going with closed baffle design because even that i really like deep bass on music, kick drums, deep well defined bass lines i litteratly hate when the low end bass "extends" and gets loose, that low ends that you "feel" all the time even you can't hear it... but i also find awfull when you can't get that "punch" in your stomach when you hear a good kick drum. Being low frequencies less directional and easier to "spread", it's seems that a theorical approach in this area at first stages it can be more accurate than with the mids/highs where the frequencies are more directional and diffraction starts to be really important, and after all it's where music comes "alive". Thank very much for your time guys, you gave me a better idea on the path that i should follow.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.