KeithL
Administrator 
Posts: 10,275
Member is Online
|
Post by KeithL on Aug 28, 2018 10:12:44 GMT -5
As a very general concept, frequencies that appear on two or more lists are likely to cause problems, because your room is likely to experience standing waves at those frequencies in multiple directions at once.
It's basically telling you the various frequencies at which standing waves will occur in that room. You get standing waves at frequencies where the wavelength is a multiple of the dimension. So, for a room that's eight feet high, ten feet wide, and fifteen feet long, you get standing waves at multiples of those three wavelengths. You will note from this that ALL rooms have room modes - frequencies where you get standing waves. The trick is to spread them out or control them. So, for example, in a room with three non-multiple dimensions, you will have a list of frequencies where room modes will occur for each dimension. However, in a room where the dimensions are even multiples, the entries on each list will coincide,
Having a room with room modes at a bunch of different frequencies isn't so bad....
If you have a room that's six feet, by ten feet, by sixteen feet, you will get a list of room mode frequencies for each dimension, and they'll all be different. (So you may end up with a bunch of little dips and peaks but they won't add up at any one particular frequency.)
But, if that room was six feet, by twelve feet, by eighteen feet, the room mode frequencies for six feet would occur on ALL THREE LISTS.... and so would add together or aggravate each other.
(And, in that case, you would be likely to see serious peaks or dips at frequencies that show up on two or more of those lists.)
The way you "use" this information is to try your best to AVOID situations where the three dimensions are even multiples of each other. For example, when you design a speaker cabinet, you should avoid choosing dimensions that are even multiples of each other (for example, use 3 x 4 x 5 or 30 x 40 x 50, and NOT 4 x 8 x 12 or 8 x 8 x 16).
And the so-called "magic ratio" is simply a ratio of length to width to height where the three dimensions are as far from being multiples of each other as possible. Note that the object of the game is to avoid having "high energy room modes" that coincide.... so the ones with lower energy matter less. (Alternately, you can avoid rooms with parallel walls.... as many speaker designs in fact do.... but that can introduce other issues  ). www.mcsquared.com/wavelength.htmAnother potentially useful calculator. This one allows a frequency to wavelength conversion. This kind of information is useful in designing listening areas. Speaker designers use it in baffle design and enclosure proportioning. www.mcsquared.com/modecalc.htmThe ROOM MODE CALCULATOR is useful for deign and execution of listening rooms. Checking in advance for Standing Waves allows for the construction of rooms with fewer problems and less need of Room Treatments (not to the exclusion of such treatments, however) Entering square floor plans shows why square rooms and rooms with dimensions that are multiples of one another should be avoided. Even a room with an 8 foot ceiling which is 16x24 will probably give problems. So I entered my room dimensions into The ROOM MODE CALCULATOR and was rewarded with this gibberish: Room Dimensions Room length 22.5 Feet Room width 14.5 Feet Room height 6.5 Feet Axial Room Modes Axial room modes 25.11111111111111 Hz 38.96551724137931 Hz 86.92307692307693 Hz 50.22222222222222 Hz 77.93103448275862 Hz 173.84615384615387 Hz 75.33333333333333 Hz 116.89655172413794 Hz 260.7692307692308 Hz 100.44444444444444 Hz 155.86206896551724 Hz 347.69230769230774 Hz 125.55555555555554 Hz 194.82758620689657 Hz 434.61538461538464 Hz 150.66666666666666 Hz 233.79310344827587 Hz 521.5384615384615 Hz 175.77777777777777 Hz 272.7586206896552 Hz 608.4615384615385 Hz 200.88888888888889 Hz 311.7241379310345 Hz 695.3846153846155 Hz 226 Hz 350.6896551724138 Hz 782.3076923076923 Hz Tangential Room Modes Tangential room modes have 1/2 of the energy of axial modes (-3dB). Tangential Room modes 46.35600754080096 Hz 90.47756187591328 Hz 95.25719309145833 Hz 92.71201508160192 Hz 180.95512375182656 Hz 190.51438618291667 Hz 139.0680226224029 Hz 271.43268562773983 Hz 285.77157927437503 Hz 185.42403016320384 Hz 361.9102475036531 Hz 381.02877236583333 Hz 231.78003770400485 Hz 452.3878093795664 Hz 476.28596545729164 Hz 278.1360452448058 Hz 542.8653712554797 Hz 571.5431585487501 Hz 324.4920527856068 Hz 633.342933131393 Hz 666.8003516402083 Hz 370.8480603264077 Hz 723.8204950073062 Hz 762.0575447316667 Hz 417.2040678672087 Hz 814.2980568832194 Hz 857.3147378231249 Hz Oblique Room Modes Oblique room modes have 1/4 of the energy of axial modes (-6dB). Oblique room modes 98.5114243978735 Hz 197.022848795747 Hz 295.5342731936205 Hz 394.045697591494 Hz 492.55712198936743 Hz 591.068546387241 Hz 689.5799707851145 Hz 788.091395182988 Hz 886.6028195808614 Hz Sooo... as a less informed layman, how am I supposed to interpret this?😕🙉 |
|
|
|
Post by leonski on Aug 28, 2018 11:45:58 GMT -5
Your're going to have to do a LITTLE reading to find out what this means and how the information might be applied. Generall, I'd tend to ignorore any frequency over maybe 250hz to 300hz artsites.ucsc.edu/ems/music/tech_background/TE-02/modes/Modes.htmlLink to description of Room Modes and their importance. Than, know what the 'modes' are. They generally refer to the number of 'bounces' needed to coincide with that frequency. Axiel will be 2 walls. Always Parallel. Either long way, short way, or uP / DOWN. These IIRC will be at the highest amplitude. If you look at your readout, you'll see RELATED frequencies. 25hz turns into 50 and the cog @39hz cogs @78hz. but You're looking for multiple frequencies across ALL modes that are the same. These will be where sound 'piles' up and creates problems. Amd as I hope you saw, the modes DECREASE in energy as you move UP from a simple 2-wall bound. 3 wall is Tangential while Oblique is 4 wall. And they get FEWER in number. The room is soaking up energy. I'll do some further reading myself. Of course if you can identify a problem, the NEXT step is 'what to do'? Smooth bass is a PIA and sometimes not easy to get. This depends somewhat on how FREE you are to modify the space and your proposed budget. DIY bass traps are a viable option. I'm a little leery of using a MIC and REW for measuring since the bass wavelengths are so very long that moving the mic just a foot either way alters the LF result. Quite the learning curve for that tool. |
|
|
|
Post by craigl59 on Aug 28, 2018 12:33:39 GMT -5
leonski:
As regards REW I use the mic at the listening position and have no problems with accurate bass correction. Yes, there is a substantial learning curve with REW and have been learning for some 5 years now. Once you feel comfortable with the engine and the design, it is so effective that I would not use anything else -- and have Dirac available.
For me, the key to using REW in an easy fashion is having a ASIO converter available along with a quality measurement mic. Since I am fortunate to have studio stuff around, can use a RME UCX converter and an Earthworks QTC-40 mic. These are so accurate they make REW relatively simple to record. Once the data is there then you have to learn to correct the wave or use the Auto function in REW; the former is more effective.
Have become addicted to the REW convolution file (*.wav) that corrects both impulse and frequency response. This is created as a standalone file then can be imported into JRiver and other media centers.
Looked over the calculator engine being discussed here and always find that these are of limited help because I never have a truly rectangular room. Right now have quality systems in two living rooms and the one that is more rectangular is better acoustically and easier to tune.
|
|
|
|
Post by Jean Genie on Aug 28, 2018 12:46:34 GMT -5
Thanks both Keith & Leonski, for the good clear explanation.
I feel confident now I can, with a bit of research, go back to the readout and glean something of value.
Much, much appreciated! 🙊🙉🙈
|
|
|
|
Post by leonski on Aug 28, 2018 13:33:33 GMT -5
leonski: As regards REW I use the mic at the listening position and have no problems with accurate bass correction. Yes, there is a substantial learning curve with REW and have been learning for some 5 years now. Once you feel comfortable with the engine and the design, it is so effective that I would not use anything else -- and have Dirac available. For me, the key to using REW in an easy fashion is having a ASIO converter available along with a quality measurement mic. Since I am fortunate to have studio stuff around, can use a RME UCX converter and an Earthworks QTC-40 mic. These are so accurate they make REW relatively simple to record. Once the data is there then you have to learn to correct the wave or use the Auto function in REW; the former is more effective. Have become addicted to the REW convolution file (*.wav) that corrects both impulse and frequency response. This is created as a standalone file then can be imported into JRiver and other media centers. Looked over the calculator engine being discussed here and always find that these are of limited help because I never have a truly rectangular room. Right now have quality systems in two living rooms and the one that is more rectangular is better acoustically and easier to tune. You have OBVIOUSLY spent a lot of time learning and DOING. Can you answer a question for me? How much difference to your bass measurement does moving the mic 2 feet in any direction? even UP? My suspicion is that when you design such a correction as you refer, that you are getting GOOD to TERRIFIC (based on user / error / learning) and that somebody sitting 4 or 5 feet away might NOT have the same experience. Putting the mic where your HEAD is going to be would appear to be the smart move. Can you take several measurements and 'average' the curves to produce a wider sweet spot? And YES, for sure. These calculators are of NO use to me. My main (only, actually) listening space is very asymmetric and 8 walls along with a peaked ceiling which is ALSO off center. This is a 2-edged sword. First? Nearly impossible to predict. Second? Except for ONE wall, it is easy to listen around and discover the Bass is fairly uniform. Once-upon-a-time, I had substantial problems with bass in my DEN which adjoins the listening room. It was like being in a 55Gallon drum! I'm thinking 'HelmHoltz Resonator'. But moving the sub to another location substantially fixed that at the same time restoring Musicality to the bass. I presented the calculator as a means for people to start to Visualize what's going on in their rooms. Once you can 'SEE' the sound bouncing around, the rest is EASY. |
|
|
|
Post by donh50 on Aug 28, 2018 13:39:50 GMT -5
Nulls due to room modes tend to be fairly sharp and so a foot or two can make a difference even in the bass region IME. At least that was true in my room, which has unfortunate dimensions after taking out room for another bedroom and hallway to it in the basement. Cut the media room volume by a third and trashed my nicely primed dimensions. I implemented my room mode calculator in Matlab and Mathcad but a spreadsheet works. Here is a nice one from Harman: www.harman.com/room-mode-calculatorHTH - Don |
|
|
|
Post by Boomzilla on Aug 28, 2018 13:46:33 GMT -5
And I'd contend that despite the technical differences, the difference between efficiency and sensitivity is totally academic. Why? Because the manufacturer's standardized SPL at one watt at one meter measurement provides a "level playing field" for speaker comparisons. Note that this standardized measurement already takes into account:
variations in speaker impedance
variations in the presence or absence of acoustic couplers such as horns
variations in frequency response
Therefore, when the consumer is comparing a pair of speakers with 90 dB and 93 dB relative outputs, it can fairly be said that the 93 dB pair will play equivalently loudly with less amplifier power, period.
And THAT'S all that the consumer need know.
|
|
|
|
Post by leonski on Aug 28, 2018 13:47:49 GMT -5
As a very general concept, frequencies that appear on two or more lists are likely to cause problems, because your room is likely to experience standing waves at those frequencies in multiple directions at once.
It's basically telling you the various frequencies at which standing waves will occur in that room. You get standing waves at frequencies where the wavelength is a multiple of the dimension. So, for a room that's eight feet high, ten feet wide, and fifteen feet long, you get standing waves at multiples of those three wavelengths. You will note from this that ALL rooms have room modes - frequencies where you get standing waves. The trick is to spread them out or control them. So, for example, in a room with three non-multiple dimensions, you will have a list of frequencies where room modes will occur for each dimension. However, in a room where the dimensions are even multiples, the entries on each list will coincide,
Having a room with room modes at a bunch of different frequencies isn't so bad....
If you have a room that's six feet, by ten feet, by sixteen feet, you will get a list of room mode frequencies for each dimension, and they'll all be different. (So you may end up with a bunch of little dips and peaks but they won't add up at any one particular frequency.)
But, if that room was six feet, by twelve feet, by eighteen feet, the room mode frequencies for six feet would occur on ALL THREE LISTS.... and so would add together or aggravate each other.
(And, in that case, you would be likely to see serious peaks or dips at frequencies that show up on two or more of those lists.)
The way you "use" this information is to try your best to AVOID situations where the three dimensions are even multiples of each other. For example, when you design a speaker cabinet, you should avoid choosing dimensions that are even multiples of each other (for example, use 3 x 4 x 5 or 30 x 40 x 50, and NOT 4 x 8 x 12 or 8 x 8 x 16).
And the so-called "magic ratio" is simply a ratio of length to width to height where the three dimensions are as far from being multiples of each other as possible. Note that the object of the game is to avoid having "high energy room modes" that coincide.... so the ones with lower energy matter less. (Alternately, you can avoid rooms with parallel walls.... as many speaker designs in fact do.... but that can introduce other issues ). So I entered my room dimensions into The ROOM MODE CALCULATOR and was rewarded with this gibberish: Room Dimensions Room length 22.5 Feet Room width 14.5 Feet Room height 6.5 Feet Axial Room Modes Axial room modes 25.11111111111111 Hz 38.96551724137931 Hz 86.92307692307693 Hz 50.22222222222222 Hz 77.93103448275862 Hz 173.84615384615387 Hz 75.33333333333333 Hz 116.89655172413794 Hz 260.7692307692308 Hz 100.44444444444444 Hz 155.86206896551724 Hz 347.69230769230774 Hz 125.55555555555554 Hz 194.82758620689657 Hz 434.61538461538464 Hz 150.66666666666666 Hz 233.79310344827587 Hz 521.5384615384615 Hz 175.77777777777777 Hz 272.7586206896552 Hz 608.4615384615385 Hz 200.88888888888889 Hz 311.7241379310345 Hz 695.3846153846155 Hz 226 Hz 350.6896551724138 Hz 782.3076923076923 Hz Tangential Room Modes Tangential room modes have 1/2 of the energy of axial modes (-3dB). Tangential Room modes 46.35600754080096 Hz 90.47756187591328 Hz 95.25719309145833 Hz 92.71201508160192 Hz 180.95512375182656 Hz 190.51438618291667 Hz 139.0680226224029 Hz 271.43268562773983 Hz 285.77157927437503 Hz 185.42403016320384 Hz 361.9102475036531 Hz 381.02877236583333 Hz 231.78003770400485 Hz 452.3878093795664 Hz 476.28596545729164 Hz 278.1360452448058 Hz 542.8653712554797 Hz 571.5431585487501 Hz 324.4920527856068 Hz 633.342933131393 Hz 666.8003516402083 Hz 370.8480603264077 Hz 723.8204950073062 Hz 762.0575447316667 Hz 417.2040678672087 Hz 814.2980568832194 Hz 857.3147378231249 Hz Oblique Room Modes Oblique room modes have 1/4 of the energy of axial modes (-6dB). Oblique room modes 98.5114243978735 Hz 197.022848795747 Hz 295.5342731936205 Hz 394.045697591494 Hz 492.55712198936743 Hz 591.068546387241 Hz 689.5799707851145 Hz 788.091395182988 Hz 886.6028195808614 Hz Sooo... as a less informed layman, how am I supposed to interpret this?😕🙉 I think the idea of 'Non-Multiple Dimensions' may be explained. In many rooms you have an 8 foot ceiling which is 'standard' height for such. But a room with a MULTIPLE of that number in width or length IS going to be a problem. The frequency which is the SAME for several dimensions or combination will tend to BUILD UP in a given spot or might 'suck out' at a certain location. You sit in a 'null' location and that bass line from a tune you know simply either disappears or is altered enough that you start looking in back of your speakers for a Wiring Problem. Or suspect you finally DID blow up the crossover! Nope. If you go to the CARDAS website, you will see they advocate the use of the number PHI. this is a very OLD number, known in some fashion to both the Ancient Greeks and Egyptians of 40 centuries ago. The value is one of those weird numbers. As it turns out, a room constructed using that as a 'ratio' (golden section) will have an advantage when it comes to standing waves and such. Some dimensions like 8 feet high X 13 feet wide X 21 long, produces a reasonable result. And can be 'scaled' to larger or smaller sizes. Doing the math? You'll see the aprox ratio of lenghts is 1.6 give or take. Now? If you only had access to a designer who would work with you in the design and construction of a 'proper' room, you'd be 'golden'. And probably Broke. www.cardas.com/room_setup_main.php |
|
|
|
Post by craigl59 on Aug 28, 2018 15:13:39 GMT -5
leonski: As regards REW I use the mic at the listening position and have no problems with accurate bass correction. Yes, there is a substantial learning curve with REW and have been learning for some 5 years now. Once you feel comfortable with the engine and the design, it is so effective that I would not use anything else -- and have Dirac available. For me, the key to using REW in an easy fashion is having a ASIO converter available along with a quality measurement mic. Since I am fortunate to have studio stuff around, can use a RME UCX converter and an Earthworks QTC-40 mic. These are so accurate they make REW relatively simple to record. Once the data is there then you have to learn to correct the wave or use the Auto function in REW; the former is more effective. Have become addicted to the REW convolution file (*.wav) that corrects both impulse and frequency response. This is created as a standalone file then can be imported into JRiver and other media centers. Looked over the calculator engine being discussed here and always find that these are of limited help because I never have a truly rectangular room. Right now have quality systems in two living rooms and the one that is more rectangular is better acoustically and easier to tune. You have OBVIOUSLY spent a lot of time learning and DOING. Can you answer a question for me? How much difference to your bass measurement does moving the mic 2 feet in any direction? even UP? My suspicion is that when you design such a correction as you refer, that you are getting GOOD to TERRIFIC (based on user / error / learning) and that somebody sitting 4 or 5 feet away might NOT have the same experience. Putting the mic where your HEAD is going to be would appear to be the smart move. Can you take several measurements and 'average' the curves to produce a wider sweet spot? And YES, for sure. These calculators are of NO use to me. My main (only, actually) listening space is very asymmetric and 8 walls along with a peaked ceiling which is ALSO off center. This is a 2-edged sword. First? Nearly impossible to predict. Second? Except for ONE wall, it is easy to listen around and discover the Bass is fairly uniform. Once-upon-a-time, I had substantial problems with bass in my DEN which adjoins the listening room. It was like being in a 55Gallon drum! I'm thinking 'HelmHoltz Resonator'. But moving the sub to another location substantially fixed that at the same time restoring Musicality to the bass. I presented the calculator as a means for people to start to Visualize what's going on in their rooms. Once you can 'SEE' the sound bouncing around, the rest is EASY. |
|
|
|
Post by craigl59 on Aug 28, 2018 15:48:53 GMT -5
leonski:
The current system I am measuring with REW is in a poor acoustic environment and the right speaker, in particular, has serious bass dip problems because of a closeby hallway.
So have experimented a lot with different mic placements and they do show differences but not, oddly, that much in the bass correction area. The overall pattern there stays pretty much the same contour.
Mostly have used the mic in the listener's head position as you suggest. Unlike Dirac (and other engines) there is no need for multiple averaging for my single-listening chair -- although that could be done by hand in REW. Multiple frequency graphs can be viewed there at the same time and then you would average the graphs and apply correction.
The Lyngdorf system has you measure all across the room and this has some advantage for that system.
Here is what I have noticed that surprised me a bit.
We dealt, as you remember, with a lot of issues concerning the power requirements for the XPA-1s -- and I appreciated very much your help, btw.
As I made any change with cables and circuits, I would redo the REW measurement, check the response, and create a new convolution file.
Any change, REPEAT, any change altered the wave structure in the room. Mostly, the overall pattern was similar BUT there were differences appearing -- especially in the midrange and treble. Sometimes these differences could be significant and the largest changes were shown when one amp was on a much longer power cord.
Then, would experiment with trying swapping in old convolution files after making a change. And the results, again surprised me. After a relatively small change in the new frequency map, the old file could create a poor sound -- shown especially in timbral accuracy and high-frequency shrillness.
In REW you can define a contour and have that app provide EQ correction for a measured waveform. This can be used as further refinement in the room correction process.
BTW Can someone explain to me how you add text to a post after quoting? The quote appears but there is no place to add text. Do you click on one of the buttons in the menu? Thanks.
|
|
|
|
Post by leonski on Aug 28, 2018 20:29:14 GMT -5
Thanks for the Insigts from someone who is 'in the trenches' and has experimented with REW and related issues.
My first thought is that s SINGLE measurement sweep, while neat, is no substitute for several identical sweeps followed by some form of 1/3 octave averaging. Tiny swoops and dips might not be as audible as Visible!
If you compare a file with a 'peak' at certain wavelengths, to a file with 'dips' at the same point, you could get what you notice. Moving around the room might cause some of these problems.
Measurement REPEATABILITY is not perfect, either, and you might be up against +-2db or more of variance under otherwise identical conditions. I don't know what these kinds of systems are 'capable' of until I either
parse somebodies numbers or some manufacturers data.
The 'trick; would be to REPEAT an earlier experiment. Right down to what cords were plugged in WHERE. And of course mic / furniture / YOU positions all must be the same. I'd be real surprised if you got a perfect duplicate.
And almost as surprised if you didn't get some differences which were 'out of character' with the original measure. I would personally REPEAT each measure under each set of conditions 3x. It is than a question of applying a
1/3 octave averaging to each file THAN averaing the 3, or averagine the 3 files THAN apply some form of 1/3 octave averaging. I'm out of my experience zone here.
Very high frequency wavelengths are VERY small and interact with room and room objects in unpredictable ways. SAMENESS from sweep to sweep followed by a form of averaging would SEEM to be a way to avoid some of
those pitfalls. A 1khz 'wave' is about 1 foot long. Just over, actually, and that should give you some idea of the potential. Higher frequencies are yet shorter in wavelength and interact with smaller and smaller objects.
An 11khz wave might be about 1/10 of a foot long. Or about 1 1/4" or so. You'll see effects like these in speaker measurements where as the frequency RISES the dispersion FALLS.
As kind of a sidenote, this also an issue of speaker design. You might have a pair of good drivers capable of say.......35hz to 20khz. But to get the 'low' you might need a 10" driver. Which starts getting 'beamy' above a certain point while
at crossover the tweeter may have pretty WIDE dispersion. Both play fairly loudly thru the passband but where you SIT or speaker positioning will make a BIG change in what you hear.......or MEASURE. What a can of worms.
A note about 'averaging'. 2 quick ways, but not the ONLY way to 'average' 3 measures:
#1 High value PLUS Low value sum divided by TWO.
#2 Add all 3 values and divide by THREE.
#3 Do a real statistic where you get AVERAGE and STANDARD DEVIATION. This requires more than 3 measures, but will also be the gateway to how much systemic and measurement error is inherent in
your measuring system. And is a REAL PIA to do. At work, I NEVER wrote a control chart with any less than 20 measures under unchanging conditions.
|
|
).
