Bruno has developed a unique measuring method that eliminates the need to use an anechoic chamber. In the middle of the assembly hall, far enough away from all walls, there is a vertically extendable pole on which the loudspeaker is mounted, and the measurement is then done at a height of about four meters with two microphones: one close to the loudspeaker, and one at a somewhat greater distance
We have been doing this for years and years. Then we went to ground plane. No appreciable difference.
Dual differential microphone setup. Since 2014. Not new! This is for a nearfield tweeter test.
And when we want highest resolution now, it is a groundplane measurement. It saves me setting up a 12 foot and 15 foot ladder. The sheet of plywood gives you a smooth surface no matter where you are situated.
8 x 16 inch baffle and a Mundorf AMT
Pew pew
An AMT I am working on currently.
25 measurements via REW each driver Vector averaged. Each measurement is done via pulling the mic back 100mm starting at 1 metre. The is smoothed to 1/12th octave. Umik microphone. I have much more expensive mics, but it is as accurate as the good ones as long as the SPL is not to high.
Mark
Agreed. It is rather disappointing to see how strongly the idea that Klippel is the only worthy method has taken internet comments, implying that nothing but Klippel obtained data to be of any value. I suppose that's the result of a technical/engineering discipline becoming sort of "pop scienced".I missed typing another agreed to thought. Klippel is great at making equipment. I respect that. What I do not respect is the blanket marketing that basicly says that everything else is not Klippel. What I do not like is that these Nearfield scanner measurements are basically the same as what we can do with REW. In almost every way. But that is not getting the attention that the marketing machine at Klippel is getting. Solid measurements should be recognized as such.
I see the value of a test room, but as DIY/custom is more interesting to me, I'm ok with the test room being the room that the speaker is being built for. 🙂Klippel is a best guess method of testing using derivations that are used to smooth over the inherent spurious response to the item under test. It goes from nearfield to pseudo farfield and make computations that are presented as anechoic. These by themselves are pretty for marketing. But as mentioned do not reflect what will happen in a real room. For that you need the room approximation graph. The question for this is what room? The standard IEC test room? I have been to this room. I know of one set up not that far from me, and the original setup at the NRC in Ottawa. It is a modest sized living room. A shoe box shape. NOt reflective of modern rooms either here in North America, or the construction of homes in Europe where the walls are much more reflective all the way from the base on upwards. Here with standard stud walls our rooms are basically invisible below 30 hertz. Europe is by and large block wall construction. Very different.
So, taking a few of these things into consideration I submit that a real test, in a room that incorporates the rooms reflections at distance is a much more valueable measurement for actual design than anything that is anechoic.
These things do seem to take longer when they're not the day job. 🙂No need to redevelop it! I'm still in the thread, just crazy busy at work, for the last 3 years.
I look forward to seeing it!I will try to post a picture of what I have so far, but probably not until next weekend.But the basic idea is that the mic can be positioned conveniently in spherical co-ordinates, manually at first, stepper motors added easily as a second step.
Sounds like some good hardware. That's the M215? Is it's factory calibration any good? I have an NTi Audio M2010 but I've been interested in getting an M215 because the "nunchuck" style of the M2010 has disadvantages sometimes (although possibly not in this application)... and I'd like to have another Type I mic that's not crazy expensive.I now have a precision measurement mic, a Bruel and Kjaer clone from MicW, and it calibrated pretty well.
Structure is nice thin wall, cold drawn aluminium tube from cross country race ski poles.
I need a mic interface probably a Topping E2x2.
A bit as if they are literally the ONLY engineers in the world with experience to pull this of.Agreed. It is rather disappointing to see how strongly the idea that Klippel is the only worthy method has taken internet comments, implying that nothing but Klippel obtained data to be of any value.
So if we use some basic math and statistics, people are basically saying that only 26 people of 8 billion is capable of understanding loudspeaker acoustics?
Because that is exactly what it comes down to.
I guess all other companies and universities are worthless.
There is no hope anymore for humankind, all we have left is Klippel.........
That momentum on that wooden board scares me though lol.
Well, there is not really a one solution fits all.Have you considered documenting your experience and recommendations for those methods? I know some of it is more well known, but it could be beneficial to have the information centralized with some professional insight.
But when designing a loudspeaker and doing already a lot of preliminary research, testing and investigation, there are many things one can already expect from the get-go.
Think about diffraction issues, port resonances, directivity miss match with certain drivers, internal cabinet resonances etc etc. Another good example here are distortion measurements.
In general, one doesn't need a full fancy "spinorama" to see and understand these issues.
For many speakers, this fully off-axis and directivity behavior is already fixed and can be understood from just a handful of angles.
Unless we go into pixel peep mode, and move us into all kinds of discussions.
Also near-field burst-decay (CSD in periods) plots and impedance graphs can already reveal a lot.
Rarely given by Klippel btw (and I found them very useful).
I personally always work from a top to bottom approach.
Meaning that beforehand you make a "map" of what problems to expect.
Quantitative assessments here can be already very helpful.
With enough experience, you can do the same for already made systems.
Granted that it won't give any qualitative numbers, so therefore not very useful for reviews.
But to put that simply, the results rolling out of Erin's measurements often don't surprise me.
Well, except how bad certain brands perform with even just the basics.
Basics that can be measured and found with the cheapest uncalibrated measurement microphones for 50 bucks.
Speaking of calibration.
Most issues that cheap mics have, is just the freq resp above 8-10kHz.
Everything below, down to 100Hz is fine.
After enough years of listening, someone can perfectly balance that by listening in combination with the "uncalibrated (off-axis) response".
Is that ideal and optimal? Obviously not.
But it's still possible to make good sounding speakers that way.
Yes it's the M215, the factory calibrations seems pretty decent, I compared it to a calibrated Earthworks and it all looked consistent. (thanks to @cowanaudio ! )Sounds like some good hardware. That's the M215? Is it's factory calibration any good? I have an NTi Audio M2010 but I've been interested in getting an M215 because the "nunchuck" style of the M2010 has disadvantages sometimes (although possibly not in this application)... and I'd like to have another Type I mic that's not crazy expensive.
It rolls off a little earlier than I would prefer, the calibration takes care of that but it doesn't look as nice.
So I would take your NTi M2010 in preference, you could always put it in a slim tube if you don't want a nunchuk.
But the M215 is pretty nice for the money, construction quality is impressive and less than $350 US is definitely not crazy expensive.
What do you use as an audio interface?
The E2x2 looks a decent match for the M215, not the absolute best but not far behind, for a fraction of the money.
Best wishes
David
Last edited:
What find interesting is that the mic close to the speaker is not on-axis nor inline but next to the midrange and at 90 degrees. Looks at first glance to me to measure primarely modes not speaker response given size of the hall ?Seems that Kii is using a similar approach, see picture:
It codes from an factory visit article in hifi.nl:
https://hifi.nl/artikel/33047/Fabrieksbezoek-Kii-Audio.html
Is in Dutch, but basically they measure in a fabrication hall with 2mics and special software, it removes the need for costly anechoic rooms etc.
Yes, given the shape of the mic capsule. Still curious as to why the mic close to the speaker is not in line. It is close to the midrange, yet the far mic is on about a meter and on-axis and measures the whole speaker. Assumption here it is measuring the whole speaker and not just the midrange.
Could also be that the close mic measures bass as well, and therefore is positioned at a decoder for point between mid and bass.
Like i said interesting, but then the kiiseven is interesting;-)
Could also be that the close mic measures bass as well, and therefore is positioned at a decoder for point between mid and bass.
Like i said interesting, but then the kiiseven is interesting;-)
Since Kii is an active cardioid system, which are a bit fiddly to get "perfect" , I can totally see why they wanted 180 degrees measurements for nearfield as well.
Also, we don't know the context of this picture.
So maybe they were just doing some specific or non related tests
Also, we don't know the context of this picture.
So maybe they were just doing some specific or non related tests
Solid like rock. 238mm wide x 38mm thick. Clamped across two stiff ladder rungs. So it is a little springy, but not much. This used to be my go to method. Then I discovered this thread and did a lot of comparative measurements. Everything anechoic messes up your calculations in that you are missing the inherent bounces that effect the midrange greatly in any realistic room from just the floor, then the ceiling. I would make best guesses, and then measure in room anyway. So, why go through all the fuss? Do the measurements insitu. You mentioned how Anechoic chambers are now not really needed. I agree except for low frequency measurements.That momentum on that wooden board scares me though lol.
Last edited:
Cardioid over a rather narrow bandwidth. This part gets skipped over by everyone in the audio press. It is literally less than 2 octaves. And only in the midbass where you have much of a cardioid response.Since Kii is an active cardioid system, which are a bit fiddly to get "perfect" , I can totally see why they wanted 180 degrees measurements for nearfield as well.
But isn't low frequency behavior very predictable?I agree except for low frequency measurements.
Which is I guess the extremely brief summary of what I mentioned before to @aslepekis
It's nice to be able to visualize things, but below 150Hz or so, measurements follow simulations almost perfectly.
Not only that, but their "far field" behavior will be very predictable as well.
It's mostly the frequencies higher up that start with us giving headaches and grumpy days. That's where all other issues start as wel.
There are a few exceptions, but these are very specific.
It's been awhile that I looked at their measurements, so I don't remember.Cardioid over a rather narrow bandwidth. This part gets skipped over by everyone in the audio press. It is literally less than 2 octaves. And only in the midbass where you have much of a cardioid response.
This is a nice example why a Klippel system is great, because we gave wonderful OBJECTIVE data from people like Erin.
Personally I think it's not 100% complete, but good enough to compare things.
But it's interesting why it's only such a narrow bandwidth.
An active cardioid up to 800-1000Hz or so is perfectly doable.
This technique was first revealed way back in the 40's in JASA by Gabriel Weinreich at the university of Michigan. All of the Klippel, et. al. techniques are based on this paper.Bruno has developed a unique measuring method that eliminates the need to use an anechoic chamber.
I completely concur. I used the TS parameters to model the LF sound field and merge it with the gated HF response. It worked very well. It's a little harder to do with ported systems since they are both monopoles and dipoles, but it's still doable.But isn't low frequency behavior very predictable?
Yes, for the purposes of simulating maximum SPL from specifications I agree. Simulations are very accurate. Especially from Hornresp. But, Many many woofers are very optimistically specified using the geometric distance of the voicecoil minus the top plate, or at a BL of 80% calculated and simulated! When I measure subwoofers I do this outside and I can tell you it is a rare woofer that gives you honest specifications. I recently read of a car audio driver that purported 3 inches of travel in each direction. When you design woofers, you need to pay attention to the mechanics of the basket. And I knew the basket that they attributed this massive excursion too. To get that mechanical motion was simply impossible. Secondarily to have a surround that has that much mechanical motion you loose so much SD that it becomes a game where you gain almost nothing due to the smaller and smaller Sd. Through many years of design I have come up with a rule of thumb to not try to do anything greater than 34mm one way. Most of the driver designs that I do start at 10mm one way. Examples abound of wishful engineering. Some I cannot mention. But for real world performance testing of a subwoofer you need the great outdoors. Your building will not like 130db at 14 or 16 hertz believe me!But isn't low frequency behavior very predictable?
Which is I guess the extremely brief summary of what I mentioned before to @aslepekis
It's nice to be able to visualize things, but below 150Hz or so, measurements follow simulations almost perfectly.
Not only that, but their "far field" behavior will be very predictable as well.
It's mostly the frequencies higher up that start with us giving headaches and grumpy days. That's where all other issues start as wel.
There are a few exceptions, but these are very specific.
Mark
A very long excursion 100mm driver I recently designed. Not quite 10mm real world mechanical throw.
This technique was first revealed way back in the 40's in JASA by Gabriel Weinreich at the university of Michigan. All of the Klippel, et. al. techniques are based on this paper.
It's been awhile that I looked at their measurements, so I don't remember.
This is a nice example why a Klippel system is great, because we gave wonderful OBJECTIVE data from people like Erin.
Personally I think it's not 100% complete, but good enough to compare things.
But it's interesting why it's only such a narrow bandwidth.
An active cardioid up to 800-1000Hz or so is perfectly doable.
Earl, I have to find this paper. We all stand on the shoulders of giants!
b_force, you know how an octave works. Yes up to 800 hertz is easy 2 octaves down from there is 200 hertz. You can effectively go a little lower on a small cabinet, but not too much lower. PA is a totally different animal. You can camp out in some of those subwoofer stacks at concerts. And the subwoofer stack can be near a half wavelength of the lower frequencies they are asked to reproduce.
Mark
yes, maybe you don't mind sharing it?This technique was first revealed way back in the 40's in JASA by Gabriel Weinreich at the university of Michigan. All of the Klippel, et. al. techniques are based on this paper.
- Home
- Design & Build
- Software Tools
- Klippel Near Field Scanner on a Shoestring