The distance from a speaker to a listener is usually about 3.5 meters - 12 feet, give or take.
How did the system get stuck with a 1 meter measuring distance when clearly there will be a response difference between measurements at 1 meter and measurements at 3.5 meters?
Seems pretty dumb to me. . . .
How did the system get stuck with a 1 meter measuring distance when clearly there will be a response difference between measurements at 1 meter and measurements at 3.5 meters?
Seems pretty dumb to me. . . .
Well Bill,
Will the mysteries of the world ever cease to amaze us?
I think they had to start somewhere. Maybe 1m made sence. What's your idea, Maybe the 3.5m as you suggest? You need quite a chamber to test that in.
Hard to call. Especially now with the car audio thing being so big. But you're right, I don't listen too much at 1m other than in autos.
Cheers,
Cal
Will the mysteries of the world ever cease to amaze us?
I think they had to start somewhere. Maybe 1m made sence. What's your idea, Maybe the 3.5m as you suggest? You need quite a chamber to test that in.
Hard to call. Especially now with the car audio thing being so big. But you're right, I don't listen too much at 1m other than in autos.
Cheers,
Cal
Yes, I do suggest 3.5 meters. But that doesn't mean squat. Too much momentum. Looks like were stuck with 1 meter.
But, it sure would make sense for those with their own measuring equipment to remeasure the drivers and their completed system at a more reasonable distance.
But, it sure would make sense for those with their own measuring equipment to remeasure the drivers and their completed system at a more reasonable distance.
OK Bill,
I think it's time to start a movement.
You got any friends in the media?
I'll cover Canada.
Can we do a 2m distance as well as the 3.5m? or how 'bout 1, 2, 4 and?
...On the trail of something good methinks
Cal
I think it's time to start a movement.
You got any friends in the media?
I'll cover Canada.
Can we do a 2m distance as well as the 3.5m? or how 'bout 1, 2, 4 and?
...On the trail of something good methinks
Cal
Bill Fitzpatrick said:
Speculation: Engineers might want to test the loudspeaker without getting too much room influence, but still include the baffle step. 1m fits that pretty well.
Well sensitiviity needs to be measured at 1m.
In anechoic chambers I believe if you measured further
back the chambers bass loss would be increased.
For most speakers the off-axis responses are not affected.
Large panel or multielement speakers are at a disavantage
meaured at 1m, the intelligent reviewer will change the
distance to 2m.
Measuring at real listening distances would cause
problems with the gating of modern digital techniques,
i.e. you can measure lower nearer to the speaker.
But note that a well measured RAR (room averaged response)
uses a range of typical listening positions averaged together.
🙂 sreten.
In anechoic chambers I believe if you measured further
back the chambers bass loss would be increased.
For most speakers the off-axis responses are not affected.
Large panel or multielement speakers are at a disavantage
meaured at 1m, the intelligent reviewer will change the
distance to 2m.
Measuring at real listening distances would cause
problems with the gating of modern digital techniques,
i.e. you can measure lower nearer to the speaker.
But note that a well measured RAR (room averaged response)
uses a range of typical listening positions averaged together.
🙂 sreten.
Measuring at one meter is used as a convenience for a number of reasons. First of all, since sensitivity is defined as sound pressure level at one metre for 2.83V input, then many people simply measure at one meter..!! One could of course measure in principle at any distance and correct for this, ie add 6dB for each doubling of distance.
However there are a number of difficulties with measuring either too close or too far.
The true farfield response means exactly that...farfield. Move in too close and you start to see nearfield effects, you change the pattern of diffraction response irregularities , you start to lose the baffle step loss and if you are making off axis measurement then unless you rotate around the true acoustic axis of the drivers rather than the convenient center of gravity of the cabinet you will get errors. You do however improve the signal to noise ratio of the measurement where noise can be true noise or also environmental contamination.
The farfield is actually defined in terms of wavelength and the size of the speaker. The bigger the speaker the further away you need to be to be in the farfield.
For the average bookshelf speaker, 1m is reasonablly OK, 2m is better. For big speakers, 2m is the minimum.
However, the further away you measure, the more you will then inccur problems from the influence of the measuring environment.
Anechoic chambers get progressively less acurate below 100Hz, and this is worsened by measuring further away.
Gated measurements also require sufficient time for the transient from the speaker to decay before the first room reflection, and this is made more dificult at larger measuring distances unless the room is very large.
Generally it is best to use a number of techniques simultaneously to accurately measure the low frequency response of the speaker.
Andrew
However there are a number of difficulties with measuring either too close or too far.
The true farfield response means exactly that...farfield. Move in too close and you start to see nearfield effects, you change the pattern of diffraction response irregularities , you start to lose the baffle step loss and if you are making off axis measurement then unless you rotate around the true acoustic axis of the drivers rather than the convenient center of gravity of the cabinet you will get errors. You do however improve the signal to noise ratio of the measurement where noise can be true noise or also environmental contamination.
The farfield is actually defined in terms of wavelength and the size of the speaker. The bigger the speaker the further away you need to be to be in the farfield.
For the average bookshelf speaker, 1m is reasonablly OK, 2m is better. For big speakers, 2m is the minimum.
However, the further away you measure, the more you will then inccur problems from the influence of the measuring environment.
Anechoic chambers get progressively less acurate below 100Hz, and this is worsened by measuring further away.
Gated measurements also require sufficient time for the transient from the speaker to decay before the first room reflection, and this is made more dificult at larger measuring distances unless the room is very large.
Generally it is best to use a number of techniques simultaneously to accurately measure the low frequency response of the speaker.
Andrew
To put some numbers on what people are saying.
In a typical room, a quasi-anechoic measurement length can be about 7ms before the first room reflection needs to be gated out. If the measuring distance is 1m, this means about 2.8ms is wasted in time of flight, leaving about 4.2ms for the meaningful measurement interval. This gives a lower frequency limit of about 200-400 Hz and this same interval resolution for data point spacing.
This is good enough for a simple 2 way. Can't tell much about the under 1k area, but ok for xover work.
Say you lengthen the measurement interval to 2m. Well, now ~5.6ms is wasted in time of flight and the meaningful interval is 1.4ms. Well, that won't do much good. A lower frequency limit of 1-2k and data points every 1-2k. Not very useful.
The only way around this is to go outdoors or a large room--not at all easy. But if you can go outdoors and get the loudspeaker unit 7-8 feet off the ground, you can get a measurement interval of ~15ms, so subtracting 2-3meters would still leave you with ~7-10ms-very usable.
In a typical room, a quasi-anechoic measurement length can be about 7ms before the first room reflection needs to be gated out. If the measuring distance is 1m, this means about 2.8ms is wasted in time of flight, leaving about 4.2ms for the meaningful measurement interval. This gives a lower frequency limit of about 200-400 Hz and this same interval resolution for data point spacing.
This is good enough for a simple 2 way. Can't tell much about the under 1k area, but ok for xover work.
Say you lengthen the measurement interval to 2m. Well, now ~5.6ms is wasted in time of flight and the meaningful interval is 1.4ms. Well, that won't do much good. A lower frequency limit of 1-2k and data points every 1-2k. Not very useful.
The only way around this is to go outdoors or a large room--not at all easy. But if you can go outdoors and get the loudspeaker unit 7-8 feet off the ground, you can get a measurement interval of ~15ms, so subtracting 2-3meters would still leave you with ~7-10ms-very usable.
One and 3.5 meter measurements will or will not differ depending upon what you are trying to measure. If you are looking at room interactions, then they will differ. If you are looking at the quasi-anechoic loudspeaker response and eliminate room interference, only a special few loudspeakers will vary. As stated by an earlier poster, loudspeakers with very large diameter drivers or drivers spaced far apart will vary.
While one meter fit an early theory of sound, now it is just a convenience, not a necessity. This distance is convenient because if no distance is specified, then we can assume one meter. If you want to measure at 10 cm, .5 m, 3m or 3.5m, it is okay. Just measure and when you publish your results, tell us your distance. That allows me to attempt to replicate your test if I so choose.
Looking at only quasi-anechoic loudspeaker response, I have measured at both one and 3 meters with drivers up to 10 inches in advertised diameter, two and three-way loudspeakers spread across two feet driver edge to driver edge, and planar speakers up to six feet in the longest axis. As long as the microphone distance was the only variable, measurement results varied by no more than .25 db. This is my limit of repeatability. Multiple tests with no changes in variables will show .25 db differences.
From my own empirical research I have found no compelling reason to measure from distances beyond one meter. And one meter is so convenient. I can compare and contrast measurement data from as long as 25 years ago and be comfortable that I am making valid comparisons. That is the only value in convention.
Great question (of course I love "why" questions), I only hope our responses have been helpful.
Mark
Lynn Olson advice
Hello
I read on the Ariel website that Lynn Olson consider best to measure at 2 meters. But that is for multi-speakers systems.
Anyway, i imagine that a 1 meter measurement for a single driver would correlate very well with a 2 meter measurement if there is no reverberation...
Also, I learned that for ANY good acoustical measurement, the microphone should NOT be fixed at one particular point but should be moved a little in an area of at least 10 cm diameter.
F
Hello
I read on the Ariel website that Lynn Olson consider best to measure at 2 meters. But that is for multi-speakers systems.
Anyway, i imagine that a 1 meter measurement for a single driver would correlate very well with a 2 meter measurement if there is no reverberation...
Also, I learned that for ANY good acoustical measurement, the microphone should NOT be fixed at one particular point but should be moved a little in an area of at least 10 cm diameter.
F
Lynn has not changed this part of his Web site in a long time. I am sure it is just word choice, but "crossover radiation patterns gelling" has never made much sense to me. Somehow it fails to match any of the physical processes of multi-way loudspeakers. Plus I have tried to replicate his results of measured differences at one and two meters for some early Ariels and could not. Maybe later Ariels are different. This is, however, only a minor concern. There is no problem with measuring at two meters. Again, if you measure at two, just let us know that you are measuring at two. In the end, the easy and inexpensive availability of acoustical measurement systems allows us to decide for ourselves what is best. We can be scientists and determine what works best.
gary f, Before I apply your suggestion to my testing program, would you kindly clarify this for me? Do you move the microphone to a different position when you repeat a measurement or are you moving it while you are measuring? Is this for any type of acoustic measurement or only some types. Is this a noise measurement, swept sine, mls, or impulse? What are the differences in measured results that make this a good acoustical measurement?
Thanks,
Mark
Well, I have to disagree on the point of measurement difference for multiple speaker units. Depending on the axis chosen, there can be quite a change in the relative ratio of acoustic center to measurement point. I will look for a prior post on this-in any case, just using simple geometry you can show significant phase errors introduced this way. Careful choosing of your axis can minimize this. Still, ideally you want to measure as far as you can and get good frequency resolution and a good lower limit.
You can do this and make a average
Yes, that's better
I do this for all acoustical measurements that i make in real world (with some physical boundaries around). It is simply to make sure that the microphone is not in a particular cancellation zone.
Put a microphone in 1 particular point in space, it will measure pressure in this particular point. But it makes no sense to measure a tiny particular point because we are, as listeners not always in the same position. If you measure a speaker for example, i would move the microphone around an imaginary sphere and make an average measurement for let's say 30 secs and that will give a better result (more conform to reality) that a microphone in a fixed position. If you try to measure lower frequencies, move the microphone over a larger sphere.
I guess the best way to see this is to make 2 mesurements with fixed microphones at slightly different positions and compare them. Maybe they will be similar but often, you can see some differences.
F
- Status
- Not open for further replies.