I would like to pose a question: "How are people measuring frequency response on drivers that have no documentation?" I highlight the question because there doesn't seem to be information about a system or quantified Equipment that one can take and produce frequency graphs. As Anachoic Chambers are out of the question, the following are what I have Same, heard or read about in terms of getting measurements
I read that people use a closed box for frequency measurements, I do not seem to get a feel for how big the boxes should be in relation to speaker size and type and whether to flush mount or surface mount To this closed box.
Large flat baffles, where the speaker is placed within the baffle, seem to have their own quirks which do not get described very often. I remember an article written by Zaph. (No longer writing?), Explaining some of the requirements that had to be added to the baffle sheet, To be able to get results, But this to me seemed incomplete.
I note a device that possibly changes the landscape Of driver frequency measurement called the tetrahedral test chamber by Geoff Hill. It seems like a great idea, put it in a corner, possibly Assemble and disassemble so it doesn't take up space, but there are probably requirements there that may be beyond the DIYer To execute, in order to be able to get good results.
So, how does one measure the frequency response on drivers?
I read that people use a closed box for frequency measurements, I do not seem to get a feel for how big the boxes should be in relation to speaker size and type and whether to flush mount or surface mount To this closed box.
Large flat baffles, where the speaker is placed within the baffle, seem to have their own quirks which do not get described very often. I remember an article written by Zaph. (No longer writing?), Explaining some of the requirements that had to be added to the baffle sheet, To be able to get results, But this to me seemed incomplete.
I note a device that possibly changes the landscape Of driver frequency measurement called the tetrahedral test chamber by Geoff Hill. It seems like a great idea, put it in a corner, possibly Assemble and disassemble so it doesn't take up space, but there are probably requirements there that may be beyond the DIYer To execute, in order to be able to get good results.
So, how does one measure the frequency response on drivers?
All I can offer is my opinion, based only upon experience. Even though I am an engineer I take a practical approach.
Sorry, I know my previous "bucket" answer to your introduction did not seem like a good "technical" approach, but if you don't have t/s parameters all you can really do is manually experiment with a sealed enclosure volume until you find one in which the speaker starts to sound good. Since you are building bookshelf speakers, it likely will be a sealed (smaller) enclosure, so that is the proper starting point. You can grossly estimate where to begin (I suggest on the large size then decrease) but it's trial and error at best. The frequency response of an open-back driver really depends upon having the proper air suspension, which is why volume is the first thing you calculate using the t/s parameters. I can't imagine frequency measurements without this (or with open baffle) would be meaningful. No parameters, you have to do this manually, either burning through lots of plywood, or, well, a bucket. Whichever easy or difficult method you use to do it, this rough adjustment you find by ear. Once relatively close, you can trot out the microphone and see how small adjustments change the frequency response curves to tune it further to the desired outcome. Flush mount vs surface mount is a second order consideration after you find a volume for the woofer you are happy with. Mids and tweets I would do on an open baffle, and probably does not matter as most are closed back. I think most folks design/find a volume for the woofer, mount the tweeter (and/or mid) in the cabinet, then start measurements to design the crossover. Might as well characterize and tune within the expected context of the design.
Sorry, I know my previous "bucket" answer to your introduction did not seem like a good "technical" approach, but if you don't have t/s parameters all you can really do is manually experiment with a sealed enclosure volume until you find one in which the speaker starts to sound good. Since you are building bookshelf speakers, it likely will be a sealed (smaller) enclosure, so that is the proper starting point. You can grossly estimate where to begin (I suggest on the large size then decrease) but it's trial and error at best. The frequency response of an open-back driver really depends upon having the proper air suspension, which is why volume is the first thing you calculate using the t/s parameters. I can't imagine frequency measurements without this (or with open baffle) would be meaningful. No parameters, you have to do this manually, either burning through lots of plywood, or, well, a bucket. Whichever easy or difficult method you use to do it, this rough adjustment you find by ear. Once relatively close, you can trot out the microphone and see how small adjustments change the frequency response curves to tune it further to the desired outcome. Flush mount vs surface mount is a second order consideration after you find a volume for the woofer you are happy with. Mids and tweets I would do on an open baffle, and probably does not matter as most are closed back. I think most folks design/find a volume for the woofer, mount the tweeter (and/or mid) in the cabinet, then start measurements to design the crossover. Might as well characterize and tune within the expected context of the design.
I guess what I'm hoping for in 2024, is to find a "cookbook" method that is reproducible. Collect the required tools together and complete the task. At present, even though I may have tools, I do not have a clear picture of execution. For example, if I want t/s parameters, I use the dats v3 and I can obtain some figures. I also learn at the same time. Surely, there has to be some way to streamline the process of measuring the frequency of drivers. I kind of like this tetrahedron concept by Geoff Hill, because of its streamlined, linear approach. I'm probably a little frustrated and still learning. There is almost a "dark art" associated with getting these measurements.
I'm sure hoping someone more skilled in the methodology of working without t/s parameters can chime in, as I would like to know too.
My novice understanding is that some of these parameters are referring to physical properties of the driver- how stiff the spider/surround are, how far the cone can travel, how strong the motor is, resonant frequency and Q of the moving structure, etc, in addition to the electrical properties (impedance vs frequency, inductance, resistance, etc). I know you can look up great explanations for what all of these parameters mean, but how to measure them all correctly is beyond me. Certainly you can do a sweep to measure the impedance of the driver, but what volume of air cushion the cone needs as a proper suspension is calculated based upon the physical properties. Experimenting with different volumes to achieve the best result may be the only way to establish a good starting point.
My novice understanding is that some of these parameters are referring to physical properties of the driver- how stiff the spider/surround are, how far the cone can travel, how strong the motor is, resonant frequency and Q of the moving structure, etc, in addition to the electrical properties (impedance vs frequency, inductance, resistance, etc). I know you can look up great explanations for what all of these parameters mean, but how to measure them all correctly is beyond me. Certainly you can do a sweep to measure the impedance of the driver, but what volume of air cushion the cone needs as a proper suspension is calculated based upon the physical properties. Experimenting with different volumes to achieve the best result may be the only way to establish a good starting point.
Hi,
see https://kimmosaunisto.net/ VituixCAD measurment manual. VituixCAD is todays best loudspeaker design suite, revolving around all measurements and all relevant information that is needed to successfully measure ( and simulate ) a speaker is in the manuals. Since it might be quite a boat load of information here is a simple explanation hopefully help you to get started:
In general, when you measure a speaker, reflections from surrounding environment would make additional sounds being measured and your data would not be only speaker response but also what ever the environment happened to be and affect sound. Thus, one should strive for anechoic measurement data somehow. Anechoic data would contain only the speaker frequency response. Since most of us do not have access to a proper anechoic measurement facility, best we can do is take various types of measurements to make somewhat anechoic response from the speaker.
In a domestic room there is about 3-5ms time before reflections arrive after direct sound if your DUT is about middle of a room and microphone 1m away. If the data is windowed to contain only sound that arrives within the 3-5ms time, it'll show anechoic response. But, since lowfrequency wave, say at 100Hz takes 10ms for one cycle, not all of it fits into this short 3-5ms measurement window. For this reason these "far field" measurements at home have inaccurate low frequency data, since the data lacks. Also, if speaker is big, not all edge diffraction / and box effects arrive within the window. One could measure lows separately, for example outside to make first reflections come significantly later, or put mic at very close proximity so that direct sound is much louder than any reflections. Lows can be measured from very close, as wavelength is much bigger than the speaker size the speaker has literally no effect on the sound so nearfield measurement is about same as far field measurement would be. It has some effect, which VituixCAD has tools to add as well. For higher frequencies the mic just needs to be far enough. 1m is the standard distance, but try if you can extend 3x longest dimension of baffle to get more to "far field", to include most baffle effects. After sound leaves a transducer the box is at the immediate acoustic environment the box itself affecting frequency response to various directions because sound reflects and diffracts.
Now with relatively good high frequency far field response and near field low frequency response one would stitch data to arrive approximation of anechoic response for the whole bandwidth. In addition, usually these are not just one measurement on-axis of the speaker, but a "spinorama" for each individual trasnducer! Multiple measurements from multiple angles to be able to calculate total system power and be able to see polar maps and so on, to get more complete view on acoustic radiation of the speaker in general. See VituixCAD measurement manual.
TS parameters data is relevant when you want to build some suitable cabinet for bass, and impedance measurement functions as basis to calculate the TS parameters and affects how the system works electrically which is important when you make passive crossover network. Impedance and TS params can be measured with the DATS, or just with reqular computer audio interface with two channels and few resistors. See VituixCAD measurement manual. Impedance plot could show you also some information how well the box works for example. TS params aren't that relevant for mid or high frequency transducers, as they aren't operated at the main resonance, which the TS params mostly describe. Usually one would mesaure impedance and TS parameters from the drivers at hand, even though the datasheet would give these. There is some manufacturing tolerances, possibility spot broken driver, and so on. This is why one would measure also distortion at some point, to get hunch whats the good operation band and max SPL for any of the transducers.
As you see most of the data can be obtained only after buying drivers and building cabinets so this could be a rinse and repeat process. Thus if you are looking for very good sound prepare to build multiple boxes and buy and sell drivers, to find performance what you need for the budget and aesthetics you might want. To speed dial this process go out and listen, make sure you have heard the sound you want and know why it happened to be able to come up with a speaker that could recreate experience something like that within your environment. Without a goal the rinse and repeat process could continue for ever, you'd be lost 😉 Thus, taking the path and being philosphical with it, experiment a lot, makes sure you have some progress on the fun path 🙂 perhaps you get to an end as well.
Have fun!🙂
see https://kimmosaunisto.net/ VituixCAD measurment manual. VituixCAD is todays best loudspeaker design suite, revolving around all measurements and all relevant information that is needed to successfully measure ( and simulate ) a speaker is in the manuals. Since it might be quite a boat load of information here is a simple explanation hopefully help you to get started:
In general, when you measure a speaker, reflections from surrounding environment would make additional sounds being measured and your data would not be only speaker response but also what ever the environment happened to be and affect sound. Thus, one should strive for anechoic measurement data somehow. Anechoic data would contain only the speaker frequency response. Since most of us do not have access to a proper anechoic measurement facility, best we can do is take various types of measurements to make somewhat anechoic response from the speaker.
In a domestic room there is about 3-5ms time before reflections arrive after direct sound if your DUT is about middle of a room and microphone 1m away. If the data is windowed to contain only sound that arrives within the 3-5ms time, it'll show anechoic response. But, since lowfrequency wave, say at 100Hz takes 10ms for one cycle, not all of it fits into this short 3-5ms measurement window. For this reason these "far field" measurements at home have inaccurate low frequency data, since the data lacks. Also, if speaker is big, not all edge diffraction / and box effects arrive within the window. One could measure lows separately, for example outside to make first reflections come significantly later, or put mic at very close proximity so that direct sound is much louder than any reflections. Lows can be measured from very close, as wavelength is much bigger than the speaker size the speaker has literally no effect on the sound so nearfield measurement is about same as far field measurement would be. It has some effect, which VituixCAD has tools to add as well. For higher frequencies the mic just needs to be far enough. 1m is the standard distance, but try if you can extend 3x longest dimension of baffle to get more to "far field", to include most baffle effects. After sound leaves a transducer the box is at the immediate acoustic environment the box itself affecting frequency response to various directions because sound reflects and diffracts.
Now with relatively good high frequency far field response and near field low frequency response one would stitch data to arrive approximation of anechoic response for the whole bandwidth. In addition, usually these are not just one measurement on-axis of the speaker, but a "spinorama" for each individual trasnducer! Multiple measurements from multiple angles to be able to calculate total system power and be able to see polar maps and so on, to get more complete view on acoustic radiation of the speaker in general. See VituixCAD measurement manual.
TS parameters data is relevant when you want to build some suitable cabinet for bass, and impedance measurement functions as basis to calculate the TS parameters and affects how the system works electrically which is important when you make passive crossover network. Impedance and TS params can be measured with the DATS, or just with reqular computer audio interface with two channels and few resistors. See VituixCAD measurement manual. Impedance plot could show you also some information how well the box works for example. TS params aren't that relevant for mid or high frequency transducers, as they aren't operated at the main resonance, which the TS params mostly describe. Usually one would mesaure impedance and TS parameters from the drivers at hand, even though the datasheet would give these. There is some manufacturing tolerances, possibility spot broken driver, and so on. This is why one would measure also distortion at some point, to get hunch whats the good operation band and max SPL for any of the transducers.
As you see most of the data can be obtained only after buying drivers and building cabinets so this could be a rinse and repeat process. Thus if you are looking for very good sound prepare to build multiple boxes and buy and sell drivers, to find performance what you need for the budget and aesthetics you might want. To speed dial this process go out and listen, make sure you have heard the sound you want and know why it happened to be able to come up with a speaker that could recreate experience something like that within your environment. Without a goal the rinse and repeat process could continue for ever, you'd be lost 😉 Thus, taking the path and being philosphical with it, experiment a lot, makes sure you have some progress on the fun path 🙂 perhaps you get to an end as well.
Have fun!🙂
Last edited:
That's what I'm saying. Your starting point is not the special environment to reduce second order effects of reflections and scattering, etc, but to determine an initial volume to place the speaker in for that test- a test enclosure that allows it to perform at least in the same rough ballpark as it will in the final enclosure, then walk it in. I don't like building multiple plywood enclosures, or even one larger enclosure that you slowly fill with scrap plywood until you get the right volume, or a more complicated enclosure with a sealable moveable wall. A bucket or tote sounds stupid, but at least it lets you quickly and cheaply walk to an initial rough ballpark volume that you can work with. I've done it several times and it does work. So far I stop at a particular volume (assuming the test setup is working well) because it sounds "good enough" for what I am trying to achieve. In the future, I will use the above system to measure behavior and fine tune further.As you see most of the data can be obtained only after buying drivers and building cabinets so this could be a rinse and repeat process. Thus if you are looking for very good sound prepare to build multiple boxes and buy and sell drivers, to find performance
I have a philosophy. Either experiment and experiment, to get The answer, or listen to other peoples experience and work from there. So from what I'm understanding, the size of the driver may be a beginning point. From your experience, if I have a 4 inch, 6 inch, 8 inch driver, what containers should I use to begin the speaker making journey? Would this be a good starting point for any speaker making journey?
Download WinISD (free) and select some popular drivers (Dayton RS series, for example) and play around with those. Compare a specific driver, how it plays in a sealed enclosure vs bass reflex. Compare frequency responses of different tunings, effected by changing chamber volume and port tuning. There is too much free information available these days to go in blind. You can also compare what others have done in their published DIY projects.
VituixCad is a bit overwhelming for first timers/starters. Furthermore, it is a simulation suite, not a measurement tool. ARTA and REW (there is more Freeware!) are your friends.
Experimenting without measurements is what kept all of us diy-ers pretty much in the dark untill the early 90ties, when Bill Waslo gave the diy community IMP. IMP and later IMP/M was the first affordable and circuit- and PCB-wise relatively easy diy loudspeaker measuring tool. In combination with Calsod 3.0 by Wittold Waldman, an extremely powerful measurement and x/o filter optimization system could be had for less than some 400 $. Both Bill and Wittold are members here.
To get started in 2024 this may help:
http://audio.claub.net/tutorials/FR measurement using ARTA.pdf
Experimenting without measurements is what kept all of us diy-ers pretty much in the dark untill the early 90ties, when Bill Waslo gave the diy community IMP. IMP and later IMP/M was the first affordable and circuit- and PCB-wise relatively easy diy loudspeaker measuring tool. In combination with Calsod 3.0 by Wittold Waldman, an extremely powerful measurement and x/o filter optimization system could be had for less than some 400 $. Both Bill and Wittold are members here.
To get started in 2024 this may help:
http://audio.claub.net/tutorials/FR measurement using ARTA.pdf
https://hificompass.com/en/speakers/measurements
click "Measurement setup description" Large woofers must have even wider baffle
IEC 60268-5
https://audioxpress.com/article/spe...ng-iec-60268-5-part-1-what-comes-next-and-why
https://audioxpress.com/article/speaker-measurements-replacing-iec-60268-5-part-2
https://ampslab.com/RECOMMMENDED READ/LOUDSPEAKER MEASUREMENTS/AUDIO PRECISION.pdf
https://www.klippel.de/fileadmin/klippel/Bilder/Know-How/Literature/Papers/TUTORIAL_W.KlippeL-IEC Standard Project 60268-5 A and B AES 2015.pdf
click "Measurement setup description" Large woofers must have even wider baffle
IEC 60268-5
https://audioxpress.com/article/spe...ng-iec-60268-5-part-1-what-comes-next-and-why
https://audioxpress.com/article/speaker-measurements-replacing-iec-60268-5-part-2
https://ampslab.com/RECOMMMENDED READ/LOUDSPEAKER MEASUREMENTS/AUDIO PRECISION.pdf
https://www.klippel.de/fileadmin/klippel/Bilder/Know-How/Literature/Papers/TUTORIAL_W.KlippeL-IEC Standard Project 60268-5 A and B AES 2015.pdf
Last edited:
Kpan- Did you strike out at the wayback machine Dynaudio links you were pointed to in a different thread? I looked at this link here and it seemed to have the t/s parameters for a bunch of these drivers (several of those links did) Even if the Dynaudio number is not printed on the back of the woofer itself I would think you could compare the construction and appearance of yours and find a decent match to these. I bet the data on that page will point you to a relatively close initial volume. It sucks entering t/s parameters into WinISD (it's finicky) but there is a good YouTube video that walks you through that process.
I also like Parts Express better than experimenting with the pre-baked offerings already coded into WinISD (That you cannot see pictures of, just model numbers listed)- You can look through pictures of the selection of drivers that Parts Express carries, and for each you can scroll down and find a spec sheet and a quick "recommended volume" for both sealed and vented enclosures by the PE staff. If nothing else this will at least point to an initial ballpark. Use a test enclosure a little on the large size and reduce with uncompressible matter. Sometimes I just look at the rough size of speakers that use drivers in the range I have and start that way, keeping in mind that most speakers are manufactured in undersized cabinets, where size is more important to them than selling you an optimal system.
I also like Parts Express better than experimenting with the pre-baked offerings already coded into WinISD (That you cannot see pictures of, just model numbers listed)- You can look through pictures of the selection of drivers that Parts Express carries, and for each you can scroll down and find a spec sheet and a quick "recommended volume" for both sealed and vented enclosures by the PE staff. If nothing else this will at least point to an initial ballpark. Use a test enclosure a little on the large size and reduce with uncompressible matter. Sometimes I just look at the rough size of speakers that use drivers in the range I have and start that way, keeping in mind that most speakers are manufactured in undersized cabinets, where size is more important to them than selling you an optimal system.
To all , thanks again. I think I think the acquaintance will go down the line. of following the totem one speaker. which utilizes the dynaudio 17w75. At least there is a volume size that seems to be used for most of these type of drivers. Make The box a little bigger and then fill and check the sound to work out a volume.
I feel I may have skewed the topic on a different tangent. " Is there an approach, that when one gets all their equipment together, they can confidently get good frequency readings. What sort of baffle set up, anything special that needs to be considered on the back of the baffle? As I may have said before, I am trying to start to cut thru the mystery of this topic.
You miss the point of many hints given here: you must start actually doing the measurements in order for you to get an appreciation of what is effective and what factors compromise your measurements. The information you seek is not purely conceptual. At a minimum you should have a microphone and REW before things start to click. Every speaker builder has their own, slightly different, best practice - often dictated by their personal circumstances. My recommendation: start making some measurements and present your results here for further critique and guidance.
Being bed ridden at the moment, I am trying to plan an approach that will possibly save some time when I am up and about. At this point I have no starting point. I have got a Behringer ecm800 at the moment with a focusrite interface to connect to windows 10 computer. I have a Dats3. The garage space has obstructions so not the best for measurements. What else do I need?
- Home
- Design & Build
- Equipment & Tools
- Streamlining speaker driver frequency measurements