emrr said:referenced to theoretical Brownian motion at -23dB.
Why?JohnRoberts said:I've been inside a couple anechoic chambers (Bozak, and Peavey).
Even my brief visits were uncomfortable.
JR
emrr said:reads the headline.
Nov issue of Pro Sound News page 54 references an anechoic chamber built for Microsoft as the new world record holder at -20.6dB, referenced to theoretical Brownian motion at -23dB.
I can only speculate, but we are used to hearing sound reflections from the space around us that give us a sense of whether we are in a closet or a large auditorium, even with our eyes closed. In an anechoic chamber we only hear direct sounds, or a very low noise floor.Phrazemaster said:Why?JohnRoberts said:I've been inside a couple anechoic chambers (Bozak, and Peavey).
Even my brief visits were uncomfortable.
JR
spica said:To expand on what Rochey stated in response to the question of "dB what?", it is dB(A) referenced to 20 uPa (micro Pascals) sound pressure level (aka SPL as Rochey stated).
I work for B&K and helped come up with the methodology for testing that chamber. It is very challenging to measure something that quiet.
Two of my colleagues are in the video: https://www.youtube.com/watch?v=Cyqc_4ixV80
Not sure who they are trying to impress or if there is a real need.spica said:I wasn't involved in the discussions with Microsoft regarding their exact reasons for wanting the quietest testing room in the world but I would imagine it is for two reasons:
1) To simply have the quietest room in the world. When you're a company as large as Microsoft (or, really, a company of any size), it is a powerful statement to say you have the best available environment in the world for the testing you are performing. If you want to be a serious/top player in your field, having the best equipment/environment to prove/test it is a good start.
That is a rule of thumb regarding relative contribution for incoherent sources (square root of sum of the squares, -10dB results in roughly 10% error). You can parse out measurements without that much noise floor but lose resolution/accuracy.2) It depends on what they want to measure. It is important to remember that you ALWAYS measure at least two things when making a measurement - even an electrical measurement. There is some level of background noise ALWAYS - whether that is ambient noise in a room, electrical noise in test equipment, etc. Sometimes it doesn't matter - ie, the thing you are trying to test (and information you are trying to extract from the test) are completely out of the noise floor of everything else and unaffected. For really low level acoustic measurements, like emitted electrical noise of a screen, power supply, component, etc, the background noise has to be very low (at least -10 dB from the level you are trying to measure - for a good measurement). Since many of those components (the quiet ones at least) may already be at or below 0dB (re:20 uPa) (quieter than we can hear in very quiet environments) the test environment must be at least -10 dB below that to make a very good measurement.
Unless they are designing measurement microphones I can't guess what they're up to, and can't imagine them pursuing that. In one of my dad's old acoustics text books there was a reference to a unit of absorption described as an open window of unit area. (like I said a very old book).CJ makes a good point that the room (and any room) will not have a consistent noise floor. There is no such thing as "sound proof" (if you hear someone use that term, slap them) - we can only attenuate what is going on outside of the room. For Microsoft, that attenuation is so good that the room gets to -20 dB(A) ambient background noise... when it is quiet around that room. During a work day with very loud things going on, perhaps the room is -15 dB(A) or ? By making the room VERY quiet (ie the attenuation is VERY good), you can be assured that at any time of the day/year/whatever that your measurements are still valid because who cares what is going on outside - the attenuation into the room is just that good. You shouldn't assume that, of course, but having a really good room is a very good starting point.
It is perhaps ironic that Microsoft known as a computer software company is building a modern physical anechoic chamber. Computer technology has all but obsoleted the need for "anechoic" measurements these days (using time windowing).You might argue what is the point of measuring a component that measures -10 dB(A)... and you might have a point if that is the only thing in the system. However, when you have 100 of those components they would total +10 dB(A) (assuming they are incoherent to each other) and you might start to have something that is audible depending on its frequency range ( and any other noise present). If there are tonal components to the sound, you may have an actual sound quality "problem" on your hands.
The technique for measuring the room isn't really that complicated. The real problem with the measurement is that the room is below the noise floor of our quietest microphone. Therefore, a single microphone will not do. If you measure with two microphones, they will both have a noise floor that is incoherent (ie not related to each other). If the microphones are both measuring VERY close to the same point in space, they will be measuring the same signal from the room and it will be in phase between the two microphones. If you look at the coherent power between the two, you can very effectively remove the electrical noise because that is incoherent.
JohnRoberts said:I am very curious about what they are up to, I am not aware of them working with exotic low noise semiconductors, or transducer hardware. A quiet box is different (simpler) than an anechoic chamber.
JR
Enter your email address to join: