OK. Both papers are available for download and I will read them over the weekend. It will be interesting to try this method against a true anechoic measurement. It will take me a while to do it. But I will report back. What mic did you use? And are you using a calibration file? That in itself may be the difference between your measurements and Amir's. I have 13 measurement mics and they are not always in perfect agreement. My 6mm mics are the most revealing. But for low end high SPL I use the 1 inch GenRad mics.
Oh, I am not just talking about using the ground plane measurements by themselves, but combining them with gated measurements as well as nearfield measurements. So stitching it all together.
I don't have all that nice outdoor luxury, just a small apartment or at the office/workshop/outdoor area
That may be asking more out of my available tools than they can provide.
I probably didn't do it the way you were suggesting, but this is the result I got:
Nope, I meant short. Bigger IR windows are great. Best I can get is a 3-ish millisecond IR window when I measure in my house (green line), compare that to a 500 millisecond IR window (red). My point was that as window length decreases, so does lower frequency detail.
It'll be a fun read!
I look forward to see what your results are.I used an uncalibrated ECM8000. I do have a calibrated ECM8000 too, but I decided not to use my "good" one on the bases that the uncalibrated one was already out and on the mic stand. I was thinking the same thing with regard to the difference between my measurement and Amir's, I have performed measurements with a more careful setup and the calibrated mic on the B2030P and the results were much closer to what's shown at ASR. There might be some unit to unit variation there too.
No. It is a discreet sone wave. eg 1000 hertz. A brick wall filter just below and just above it is used to filter out everything but 1000 hertz. It is ver hard on the drivers if it's loud and it is very immune to external noise. It is not fast. Back when I started measuring I did this with a function generator but did not know how to set up a twin T filter. But that was back when I had hair. And being taught via the library of the National Research Council in Ottawa I had some knowledge and made a ton of mistakes. Still do actually. It's what keeps me trying to learn more!
Oh yes, but that's a fundamental and intrinsic part of a time window
In this case I don't see any major differences, especially not ones that could not be see in any of those other measurements.
Btw, with beam forming mics we don't even need many mics, just one mic that we can move up and down.
Or maybe this was already proposed in the paper? Apologies, didn't have the time yet to read the whole thing atm
What was the idea of using MLS instead of a frequency sweep?
The S/N ratio is miles better from a frequency sweep.
I wish more people would see the world this way!!
But isn't a frequency sweep (like the one used in ARTA) basically doing the same thing?
Just auto correlating a frequency could also work, since at a given point in time we are only interested in the amplitude of the frequency that is just being measured at.
I cannot say for certain that a sweep has the ability to reject noise in the same manner as a individual sine wave measures with filters. You have to use many discreet sine waves to plot points over the frequency bandwidth of interest. The highest resolution I do is a 512 point measurement from 20 hertz to 20 kilohertz. But that takes about 40 minutes.
Yeah I definitely can see that, but with a swepped sine wave, a S/N ratio of at least 40dB is most of the time possible.
Since outside noise doesn't correlate with the actual source, one can do an average with swepped sine waves, reducing the noise with a 10*log(amount of measurements).
Which is another 9 or 12 dB extra, a little bit more if we measure for a minute or so.
When measuring around 1W an average speaker does around 80-90dB @ 1 meter, with averaging this will give you effective (relative) output of 92-102dB (seen from a signal to noise ratio point of view).
That is already quite substantial.
I guess even some kind of noise canceling principle can be used, since the signal is so heavily correlated.
Haha, well I am just brainstorming at this point, so definitely feel free to brainstorm along!
Yep! That's what they did in the paper and what I did too. Saves on expense and ensures phase and amplitude match between measurements. Also the measurement points were done radiating from the speaker so the sum results would be a specific axis.
Another test with a lower elevation (about 43" rather than 84") and a greater total distance for the measurement points per Gedlee's suggestion. Using the calibrated microphone this time too! Although it might need to be recalibrated...
A different B2030P from my collection was used this time also. 18 points were taken at 6" intervals from 36" away from the speaker out to 144". And as before, no smoothing or windowing was applied. The black line is today's measurement; blue is yesterday's; red is ASR's.
A different B2030P from my collection was used this time also. 18 points were taken at 6" intervals from 36" away from the speaker out to 144". And as before, no smoothing or windowing was applied. The black line is today's measurement; blue is yesterday's; red is ASR's.
It is good that you are trying different ideas. But using one loudspeaker is a wise idea as it will be the only constant to allow comparisons. Your loudspeaker not measuring precisely the same as Amir's is in no way a problem. Even in the factory not all brand new bouncing baby loudspeakers measure the same. You need serious quality control to accomplish this and Behringer is not know for this level of QC. I have worked on factory assembly lines as the measurement checker. And been in other factories that claimed to be doing QC is is farcical. Error windows large enough to fly a 747 through. But they passed QC! Tight QC starts with every driver being measured, then the system and then the final assembly. And keeping a tight 1 to 1.5 plus or minus standard. What I see in the differences between your measurements and Amir's is a reasonable deviation between samples. Possibly a lack of lower frequency resolution. But this may be a function of the math behind the Klippel measurements and curve fitting (Perhaps smoothing) versus actual measured points. Every method of measurement uses some smoothing. It is in the nature of the math that areas are defined as beneficial or not and displayed or rejected.
Don't forget that there is also quite a deviation in the microphones themselves, as well as in positioning.
The very vast majority checks in companies are superficial. Often even based on wrong references.
I can still remember when I was doing an intern at at acoustics company many many many years ago.
For vibration measurements they use some kind of mini steel coupling table/platform to mount accelerometers on.
So my very first question was what the upper frequency limit was of this thing, since this has a lowpass kind of behavior.
They were staring at me like I was asking something weird, lol
If I remember currently the upper limit was around 150-200Hz, after digging deep into the standards.
Almost every company I know uses exactly the same construction/platform totally wrong for the same reasons.
I have seen this with so many companies. I also know of certain flow meters that are calibrated on the wrong (rather incomplete) assumptions. Quite some bigger companies are using those.
But hack, to some extend measurements are sometimes even superficial.
I mean that from a consumer and sales point of view, not from a designer point of view.
The very vast majority doesn't bother to read them, the next portion of people don't understand them or how to read them and than there is a group who say they understand it but they actually don't. On top of that it's is very rarely apples vs apples. Even from speaker driver manufactures themselves it's horrible. Extremely rough response with so many interference and reflections that it's often impossible to recognize a reflection from something like a surround dip etc etc.
Or they put so much smoothing on it that it's also completely useless.
In general, in my opinion the standard Klippel NFS spinoramas never shows the whole picture.
Simply because they are not meant for that to begin with. For example, if you want to get a good picture of like issues inside the cabinet etc, it's much better to plot a near-field Burst Decay Envelop together with frequency response, the impedance as well as the distortion. Spinoramas are only good for getting an overall picture as well as a good sense of the directivity.
Agreed. And that actually hit me after I got everything put away, per my usual method of remembering everything for a project after I think I'm done.
That coincides with something that I've been noticing lately, that it has been getting to be more fashionable for consumer home audio and theater enthusiasts to be pro science and engineering... but not everyone thinks like an engineer or a scientist, and I believe that a person must think like an engineer or a scientist for that data to be truly useful to them.
True, it's another piece of the engineering puzzle to making a loudspeaker. A very important piece in my opinion, but a piece.
My old professor for control theory would stand up and applaud you for saying that. haha, no joke.
Yes, don't participate without insight, because then it will become a statement without a vision.
(little hard to translate, but I think it translates pretty well).
The "new" form of discussing/debating seems to be either just out of the blue responding with "can you prove it" or just linking to some kind of (3rd party) article without insight what the article exactly is about.
What I personally miss the most, is just debating with an open mind and curiosity.
You can still have a (strong) opinion, but always keep that little but of curiosity open.