Well I imagine like me, many of you don't have an anechoic chamber sitting around. I also imagine that, like me, its not always convenient or practical to take outdoor measurements. Even if you can, we are limited by our equipment and rigs to get accurate measurements. In my case, I have Speaker workshop and ATB PC Pro. I predominantly use the ATB as its more accurate. I use a Dayton Woofer Tester 2 for impedance and electrical phase measurements as I find it more convenient and consistent than the ATB, but the ATB works for that as well. I use speaker workshop for manipulation, combining, splicing, etc of all the files, as I prefer the way it handles things.
For Microphones, I currently use the mic that comes with the ATB. It looks like a piece of crap computer mic, but according to ATB, they gutted a computer mic and attached a very accurate mic transducer with a flat response from 20hz to 20khz. Combined with the mic correction file they supply, I find I can get reasonably good measurements from 20hz to 20khz, but thats about it. I do also have a Behringer measurement mic, but have not had success getting better measurements, so I just use the other as its more convenient.
For my measurements in room, ATB prefers an averaging technique to the chirp method others use. I don't know if its any better, I can't speak to that, but I don't think the averaging method is very elegant or easy to achieve consistent results. In order to achieve this averaging, they suggest moving the mic around a 60 degree arc in front of the speaker. In order to do this, I have to manually move the mic while the test tone is running, and then manually stop the tone. If any of you have a suggestion on how to make this more precise I would appreciate that.
As for measuring at different angles, how do you all ensure the precision of these changes. I've been using a triangle marked with degrees placed at the center of the speaker, with a string attached to its center point. I then move it so that it aligns with the appropriate degrees (The string is roughly 1 meter long), and place the mic at the appropriate point. This works horizontally, but without attaching it to the baffle, it doesn't work vertically.
Then you have all those little things that seem to effect results. You have room reflections of course, inaccuracies because of distance (I use a close miking of the bass drivers below 300hz), phase of the moon, outside noises, wind direction, etc etc. I mean, I can not, for the life of me, get what I would call a good acoustic phase response measurement of the speakers. ATB claims that it automatically syncs itself with the speaker for phase, but makes it very clear that, according to them, all measurement systems are problematic for measuring acoustic phase accurately.
How you do position your mics? Do they need to point directly at the driver, or is that actually somewhat unimportant. I assumed they did, but I dont know. How you do hold the mic, I see some crazy looking mic stands in professional measurement rooms, things I've never even seen for sale anywhere before, but is a normal mic stand ok. What about the mic, someone told me to place a circle of acoustic foam or felt roughly 3" in diameter on the mic to improve high frequency measurements. Doing so does actually smooth things out, but is it correct to do this?
When then working with the files for crossover design, how much smoothing do you apply. I've heard people say none, but that seems impractical. Applying none gives you a porcupine graph that is almost impossible to decipher when designing the crossover. I find that 1/6th or 1/12th is ok, but is that reasonable?
Thanks in advance for any help you all can offer.
For Microphones, I currently use the mic that comes with the ATB. It looks like a piece of crap computer mic, but according to ATB, they gutted a computer mic and attached a very accurate mic transducer with a flat response from 20hz to 20khz. Combined with the mic correction file they supply, I find I can get reasonably good measurements from 20hz to 20khz, but thats about it. I do also have a Behringer measurement mic, but have not had success getting better measurements, so I just use the other as its more convenient.
For my measurements in room, ATB prefers an averaging technique to the chirp method others use. I don't know if its any better, I can't speak to that, but I don't think the averaging method is very elegant or easy to achieve consistent results. In order to achieve this averaging, they suggest moving the mic around a 60 degree arc in front of the speaker. In order to do this, I have to manually move the mic while the test tone is running, and then manually stop the tone. If any of you have a suggestion on how to make this more precise I would appreciate that.
As for measuring at different angles, how do you all ensure the precision of these changes. I've been using a triangle marked with degrees placed at the center of the speaker, with a string attached to its center point. I then move it so that it aligns with the appropriate degrees (The string is roughly 1 meter long), and place the mic at the appropriate point. This works horizontally, but without attaching it to the baffle, it doesn't work vertically.
Then you have all those little things that seem to effect results. You have room reflections of course, inaccuracies because of distance (I use a close miking of the bass drivers below 300hz), phase of the moon, outside noises, wind direction, etc etc. I mean, I can not, for the life of me, get what I would call a good acoustic phase response measurement of the speakers. ATB claims that it automatically syncs itself with the speaker for phase, but makes it very clear that, according to them, all measurement systems are problematic for measuring acoustic phase accurately.
How you do position your mics? Do they need to point directly at the driver, or is that actually somewhat unimportant. I assumed they did, but I dont know. How you do hold the mic, I see some crazy looking mic stands in professional measurement rooms, things I've never even seen for sale anywhere before, but is a normal mic stand ok. What about the mic, someone told me to place a circle of acoustic foam or felt roughly 3" in diameter on the mic to improve high frequency measurements. Doing so does actually smooth things out, but is it correct to do this?
When then working with the files for crossover design, how much smoothing do you apply. I've heard people say none, but that seems impractical. Applying none gives you a porcupine graph that is almost impossible to decipher when designing the crossover. I find that 1/6th or 1/12th is ok, but is that reasonable?
Thanks in advance for any help you all can offer.
Measurements can be tricky as the ability of the duplex function of a sound card is challenged, room interaction can require gating technique and so on. Instead of turning the mic around you have the alternative to turn the speaker but usually in professional measurements they use to have a turning "board" that does this automatically.
I´m going to read this book before I get "lost" in the world of measurements. Don´t forget to try to find some correspondence between your listening impressions and the measurements!
http://www.amazon.com/Testing-Loudspeakers-Joe-DAppolito/dp/1882580176
I´m going to read this book before I get "lost" in the world of measurements. Don´t forget to try to find some correspondence between your listening impressions and the measurements!
http://www.amazon.com/Testing-Loudspeakers-Joe-DAppolito/dp/1882580176
I pivot the speaker and leave the mic fixed. This ensures you maintain the same distance to the mic. I mark 15,30,45 and 60 degree angles on a board on which I place the speaker, then twist and remeasure. I measure individual off axis curves and simulate these as different drivers in SW so I can have both onaxis and all off axis curves available to me in the one model file.
For vertical response - just turn the speaker on its side and re-pivot on the board.
As long as the mic doesn't move then I don't think it has to be precise. I simply use a tape measure to get 1m to the mic (or more if I can gate high enough to get resolution in SW for xo modeling) and just use my eye to square up the mic. I just use a standard mic stand I bought cheap.
I can't speak for ATB but NEVER smooth your measurements when designing in SW. IF you do - you stuff up actual measured phase making XO summation inaccurate. If you want to smooth for XO modeling in SW - then you have to HAVE to HBT the data, re-import and set acoustic offsets on all your non-mic axis drivers (using a negative offset in the position field for each driver in the XO).
I only smooth (probably about 1/6th octave) on a copy of the final modeled / measured response to help me target the problem areas - or just to see how it looks.
Don't forget an in-room response at listening position (ungated) to sanity check your designs (ie. what your ears are going to hear, as opposed to a semi-anechoic designed speaker)
Cheers,
David.
I do always take in room response measurements, and those are relatively easy to measure. When you take those though, do you have both speakers going, or just one? I find the measurements aren't the same, which makes sense I guess. I keep wondering if designing a speaker to measure flatter in room with both going is important or not, or if its even possible.
Anyway, I'm mostly concerned with getting good measurements of the individual drivers and final systems in a quasi-anechoic nature for crossover design.
I use speaker workshop for my crossover modeling, atb doesn't have this ability. The ATB suite is only a measurement suite for speakers and electronics. I haven't used it this way, but I guess it can be used to measure distortion and things of electronics.
You know I've read about people designing completely with software and tracings, no actual measurements. I can't help but think that is a really bad idea, given my experience. I used to do that for some cheap basic projects, and I've since gone back and measured them. I'm finding that if I had just taken some actual measurements all along, I would have much more accurate models, and potentially better designs. In some cases the projects were designed to be cheap and simple, and anything better would have required more parts, but still, I'm intending to make some changes to the crossovers now that I know how they measure.
Anyway, I'm mostly concerned with getting good measurements of the individual drivers and final systems in a quasi-anechoic nature for crossover design.
I use speaker workshop for my crossover modeling, atb doesn't have this ability. The ATB suite is only a measurement suite for speakers and electronics. I haven't used it this way, but I guess it can be used to measure distortion and things of electronics.
You know I've read about people designing completely with software and tracings, no actual measurements. I can't help but think that is a really bad idea, given my experience. I used to do that for some cheap basic projects, and I've since gone back and measured them. I'm finding that if I had just taken some actual measurements all along, I would have much more accurate models, and potentially better designs. In some cases the projects were designed to be cheap and simple, and anything better would have required more parts, but still, I'm intending to make some changes to the crossovers now that I know how they measure.
Bruel & Kjaer has an excellent primer on microphones. You have to register to download but they should be required reading for anyone who wants to measure sound.
http://www.bksv.com/Library/Primers.aspx
http://www.bksv.com/Library/Primers.aspx
I think there are not a main road to speaker design and learning about passive crossover construction takes time and must involve awareness about psycho acoustics. Neil Patel of Avalon Audio replied in an interview to the question "Do you use computers in your design work" in (about) this way; by God, beware of these soul-less things, in the initial design they are banned but used later on by my staff.
Does a loudspeaker with the straightest on axis frequency response sound better than all others with more or less peaks and dips to theirs? Some say (small) dips are less "dangerous"; they are often passed unnoticed by the listener but peaks are generally easier to hear at least if we exclude the bass , some say that speakers with a declining response in the upper register are often preferred by listeners in a controlled experiment. I don´t think there are so many "easy answers"; the whole thing is quite complex.
The polar response of a loudspeaker with regard to different frequencies (directivity ) seems crucial , but this is also due to if room acoustics is good or bad. More to read on the subject here for instance:
http://www.gedlee.com/downloads/Cum laude.pdf
One hint is to really get to know the drivers you are about to use; their merits and drawbacks. This rises another question: does it make sense to try to compensate for their flaws when designing the crossover? Or is a simple approach better?
I can recommend downloading this demo program (link) perhaps not so much because you need yet another program but more to the fact that you will receive a series of articles from the construction of some loudspeakers, where the need to do measurements and listen are stressed. I have just received part 9 and all tough the articles have been published in the Hi-Fi World magazine before I like to have the whole series downloaded to my computer.
http://wduk.worldomain.net/acatalog/Software.html
Does a loudspeaker with the straightest on axis frequency response sound better than all others with more or less peaks and dips to theirs? Some say (small) dips are less "dangerous"; they are often passed unnoticed by the listener but peaks are generally easier to hear at least if we exclude the bass , some say that speakers with a declining response in the upper register are often preferred by listeners in a controlled experiment. I don´t think there are so many "easy answers"; the whole thing is quite complex.
The polar response of a loudspeaker with regard to different frequencies (directivity ) seems crucial , but this is also due to if room acoustics is good or bad. More to read on the subject here for instance:
http://www.gedlee.com/downloads/Cum laude.pdf
One hint is to really get to know the drivers you are about to use; their merits and drawbacks. This rises another question: does it make sense to try to compensate for their flaws when designing the crossover? Or is a simple approach better?
I can recommend downloading this demo program (link) perhaps not so much because you need yet another program but more to the fact that you will receive a series of articles from the construction of some loudspeakers, where the need to do measurements and listen are stressed. I have just received part 9 and all tough the articles have been published in the Hi-Fi World magazine before I like to have the whole series downloaded to my computer.
http://wduk.worldomain.net/acatalog/Software.html
- Status
- Not open for further replies.