While the method I used wasn't perfect, it can put the "problem" of panel noise in perspective. The output from the panels will typically be so low that on their own they may be barely audible, but also completely swamped by the output from the speaker itself. The issue I have with my test setup is that I don't have an accurate measure of the difference between the output of my accelerometer and my measurement microphone. Putting the accelerometer on the woofer cone and then on the panel gives a very rough approximation of the difference between how much each of these are vibrating, and the amount I can use to offset the panel measurements. But even the conservative estimate of 40db pushes the panel output down below audibility. Obviously the panel output becomes more of a problem if the speaker has larger, thin unbraced panels that are undamped. So this isn't me saying that speaker construction can be slack. It's me saying that the single minded obsession with making a completely inert speaker cabinet is a waste of time and resources. Especially when you consider that the walls near the speaker will vibrate just as much (or more) as the speaker cabinet itself. Now I should point out that when the accelerometer is attached to a panel, it's measuring how much the panel vibrates, and how much the entire speaker is vibrating. Unless the speaker is rigidly held so that it cannot move, there will always be some contribution from the cabinet vibrating as a whole.
Interesting use of available tools and techniques to correlate and quantify the enclosure panel vibration conundrum. In spite of the 40-60dB offset (depending on the level of margin applied to accuracy), the open baffle enthusiasts often cite elimination of enclosure panel resonances as a key advantage. Listening to music at 85dB, I'm not sure if 25dB of resonance is something that would tangibly interfere with the pure driver output in a way that it would be "clearly obvious" if the enclosure were then removed. Not to mention, the directivity of the panels might also be a factor. How much output from a -50dB side panel is actually making it's way to the listener's ears? Very interesting work. You certainly get the neurons firing on all 8.
Hi John! I've been thinking about this issue now. Recently, I rounded the edges of the baffle and it made a huge difference. So much so that I thought I'd tackle box resonance. Thinking on this I think that people are confused about these box resonances. In your video, you show that the box panels don't make sound, and draw a comparison to the wall to show that resonances are happening everywhere. I agree with your findings, that the box panels are not adding sound. But that's not the problem. The problem is that box vibrations color the sound because they vibrate the drivers themselves, as they are in the act of making sound. For bookshelf speakers the answer is simple, foam under the speaker absorbs resonances, also known as decoupling. I use it, and the results are great! However, for large speakers like my floor standers that weigh 60 pounds, it's not really an option, and all bass notes are not fully absorbed by foam. What I'm thinking of trying is coupling the baffle to the back of the speaker box using dowels, well fitted, and screwed in from both sides. The vibrations from the baffle will be roughly halved, transmitted directly to the back panel. Those vibrations will be more or less out of phase with the front panel, causing destruction of the vibrations meeting on the side panels. Coupling these panels over nearly a foot should do more to stiffen the baffle than anything I can think of, basically creating a stressed skin panel whose rigidity is a function of the distance between panels. The way we brace boxes does not seem in harmony with trying to deal with these resonances by cancellation. In the current bracing model, vibrations go directly to the sidewall first basically just spreading out vibrations to all the panels, but with no objective of cancellation. The only thing I can't account for in this mental exercise is the sound waves coming from the back of the woofer. When it hits the back wall, will it be cancelled out by the vibrations transmitted to the wall, or will it be additive? Will it be in phase with them, or out of phase? Will the dowels be able to transmit the vibrations immediately enough to cause cancellation? For years I've read that the box is the worse shape for these resonances, but if these equidistant walls are coupled, there might be a way to make box speakers great again.
I am o think an audiophile. We are insane, the hobby is insane, we, all of us will never be 100% happy with our gear. All of this and cables , cable lifters, the twice clock. It is all nonsense , treat your room and hope no one of importance notices this nonsense. Go listen to some music! Live a little!
Use the exciter as active noise cancellation on the panel. Flip the polarity of the accelerometer recording and play it back into the panel. Would be interesting to see the result even if you couldn't hear the difference.
Man, I am really digging your channel! Working on a pair of partially open baffle speakers right now and your past few videos have made me consider a lot of things.
yeah I think that relative measurement cone vs cabinet is a valid one and about what I'd expect. Although the cabinet surface area is much larger than the speaker but also most is facing away from the listener. Maybe it would be a different story once you hit the cabinets resonance frequency.
In a 400 level engineering course i was assigned a project to measure the reduction in vibration hammering on a chisel bare & with various chisel safety top-caps. Setting up the accelerometers and processing the raw output to be meaningful for analysis turned out to be such an involved task that it was eventually broken out to a separate project for another student to tackle while i worked specifically on the mechanics of the apparatus.
Interesting subject John. I think music that is bass heavy and have transients from drum solos would be the best samples for cabinet resonance measurements. The key point to remember is that since everything has a resonant frequency the challenge in cabinet design is to keep that resonant frequency below the lower range of the speaker. Otherwise, the whole cabinet will vibrate uncontrollably whenever that frequency is hit. Adding mass lowers the resonant frequency. Of course, this is not a problem in open baffle designs.
I've done a little testing with various materials(cement with dampening, ceramic, composites...), best one I came up with was with constrained layers..., Haven't really implemented it yes as right now I'm just building smaller bookshelf speakers witch don't suffer as much. Plus I've also never thought this was something that is degrading the sound... It's the whole diminishing returns thing.
One way to measure the actual sound produced by a side panel is to do the recordings in an anaechoic chamber with the front panel isolated from the sides?
Nice video but I wonder if the accelerometer should have been taped to the side panel in the center (assuming its a cheap cabinet without bracing), maybe it would have the largest amount of vibration there rather than say 1/3 of the way down. Just a thought.
I'm positive that you could also use the exciter like a microphone. It will need some sort of a pre amp to make it work properly but it is viable I think. It might be a more accurate measurement device for vibrations in this fashion. Very cool content!!
You did this with an Elac. Those things are built far better than average, as far as store bought speakers go. -67 db is probably a significantly better reading then you'd get from a lot of other speakers. See if you can borrow some Klipsch RP-600's or Wharfedales. The GR-Research channel has done measurements of those, and they're not as well built as the Elacs. Also, maybe you can test the application of the GR-Reserarch No-Res product?
John this was excellent and answered the concern fairly well. I'm especially impressed by all the work you did to try and put it into perspective with the decibel levels between resonance vs driver output. 40 Db difference is huge and wouldn't be very audible. Heck my forced air heating and cooling runs about 20 Db lower and I can't hear that when playing music. I'm into speaker building as well and may try the exciter when deciding where to put bracing for the greatest effect. Honestly, I think that most of the benefit of a well braced cabinet is more for a placebo effect when one feels better about the sound because the cabinet passed the knock test. That being said, I wouldn't be surprised if a completely unbraced, untreated cabinet made with thin material may exhibit some audibility when playing music continuously vs a sweep, but only at substantial SPL. Recently in a current project I'm working on, I could swear there was an audible increase in Resonance during guitar playback... It was a much larger cabinet with substantially more potential for resonance due to panel size vs a well built, thick bookshelf speaker.
I'd consider resonance and bracing tuning to be a more technical and complex level of speaker building. Does it make a difference? absolutely. The goal of building a speaker yourself is essentially making it sound how you want, at a price you can afford. Resonance is a variable that changes sound, almost always in a detrimental way. The biggest piece of information missing from this experiment (aside from only using one unit) is failing to account for off-axis sound. The biggest issue with resonance is that the box itself is now emanating sound in all directions, not just the drivers. This results in peaks or dips that interact in unpredictable ways in a room, and the frequency response and cannot be EQ'd out. On-axis measurements may still look very good, but when you do various measurements off axis, you'll see resonances have the same peak or dip in all measurements. The most direct impact of this is poor soundstage/ imaging, as a given instrument may seem to 'move around' randomly as it gets amplified and quieted by the box itself. For the record I'm not just parroting theory written by others, I've incorporated adding bracing and there's definitely an impact.
quite interesting... excellent little experiment. i was a bit surprised @6:35 by the high fqs coming from the speaker cabinet. some of those vibrations are out of phase w/ the driver, right? also the separate direction... i'm imagining the wall is stealing energy thru it's vibrations, like a membrane absorber, and what we hear is what gets reflected. like when a sound wave applies pressure on the wall the wall deform away under that pressure...
This is a really interesting test which certainly gives some interesting results worth exploring. With small bookshelf models like the Elac, panel resonance isn't often a major issue. Where it becomes more of an issue is with large, unbraced cabinets like many of the larger Klipsch "Heritage" models, especially the Forte, Cornwall and La Scala, along with many older, boxy-style speakers. The bigger those panels are, the lower their resonant frequencies and the more freely those panels will resonate, making them much more noticeable, especially during playback. It would be interesting to se you do a similar test with a much larger cabinet, then repeat the test after adding bracing throughout the cabinet.
@@IBuildIt in some sense, sure. but it's also not receiving the same pressures as what would be seen within the box, just the parts of the front waves that wrap around the speaker or comes out from a rear port, if one is present.
If I can come up with a more accurate way to set the offset between the mic and the accelerometer, I'll try it again with something bigger. I doubt there will be much of a difference, unless it's something very poorly made.
Worst case enclosure vibrations are when a musical section sound notes are approximately matched...mostly these will be sustained for some time enough to become discernable because basically that's what music really is..those musical notes may repeat/ sustain and is not like a frequency sweep... that's done only once , probably for a few hundreds of millisecs and then gone.... the panel mass has a latency/ delay that may not respond as quickly to vibrate... so it's best to test with actual sound clippings / track in the +/-3dB frequency range that causes the worst enclosure performance...@ different SPL levels and correct for that accordingly....
I hear these resonances on certain songs, like a bloated bass or drone at certain frequencies, if the note or passage is sustained or recurring. Would you say this is cabinet resonance I am hearing? sometimes I find certain song unlistenable due to this. I have the speakers positioned well out from the front or side walls with plenty of soft furnishings so i don't think its the room. Have you any experience with No-Rez? I am considering trying this but want to know for sure whether it is indeed cabinet resonance I am hearing. speakers in question are wharfedale Linton Heritage.
Thanks for more great stuff! Could you disconnect any tweeters and mid-ranges and put a tube running from the woofer into a nearby room and heavily insulate that tube. Then the sound of the room picked up by a microphone would be predominantly that transmitted by the cabinet and all its surfaces at that? Thanks again!
Interesting analysis and presentation. Fascinating to see the vibration in the walls. The only problem is when you brace a speaker of take care to engineer out box vibration, the perceived sound improves. In open baffle, I think the biggest contribution is the front baffle being stiff and massive. Also a rigid frame in the drivers. With this, you get the full dynamic punch of the cone vs a mixture of the cone moving and the baffle recoiling.
Hearing perception is influenced by what we know and what we see. If you know the speaker is solidly made and you think that's an important feature, you will perceive the sound to be better.
Agreed. I epoxied 1/4" steel semi rings behind the midwoofers in my Polk 7B's and lets just say that the improvement in bass punchiness was not subtle. I also feel less vibration through my fingertips when I touch the baffle. Could be better still and I'm thinking about a couple of 3/4" oak dowel braces going from the front baffle to the rear enclosure. Lots of little incremental improvements are additive and I think each one of them could fail a blinded A/B test but no way would the sum of them be an insignificant improvement. So, do we convince ourselves to not even do the small incremental improvements? Not me.
@@captainwho1 Hey! I'm on this channel looking to find exactly what your describing. I'm thinking that dowels from the baffle to the back would not only reinforce the baffle, but also vibrate the back panel opposite to the front. These vibrations could also cancel out the sound waves from the front panel because the back panel is vibrating oppositely in a push pull configuration. The side panels wouldn't vibrate because the destructive vibrations from the front and back would meet on the sides. Is this what your thinking too? Could you elaborate on what you did with the polks? In a constrained layer thread some guys were saying that opposing woofers fix this problem, so I'm thinking that coupling panels would do a similar thing.
@@headphone307 By coupling the baffle with the rear, you increase the effective rigidity of the baffle and also raise it's resonant frequency and lower it's amplitude of vibrations. Another method is to use an adjustable tensioning rod that couples the woofer to the enclosure rear. Jean Mauer uses this, combined with a sand filled rear wall. B&W Nautilus development may have employed woofer coupling to rear as well.
Very nice video and tests! One thing occurred to me though and I don't know if it means anything. A microphone is taking a sample of the sound energy that has made it into the air and is indicative of what our ears will pick up. The accelerometer is taking a sample of the vibration that has made it into a hard surface. The size of that surface will influence how much sound volume is transmitted into the air. The side of the speaker vibrates much less than the driver, but is also much larger. The wall in the room may vibrate less than the side of the speaker, but the wall has far more surface area than the side of the speaker. (And if the wall doesn't, you have a VERY understanding partner!) Anyway, what you might want to try next is to take your exciter and stick it to the side of the speaker. Measure the vibration with the accelerometer. Use just the exciter to reproduce the same vibration level as the speaker playing normally. Then see how much volume the microphone picks up from whatever distance you measured the driver. That said, I don't know if the exciter itself will make noise and invalidate such a test. I suppose if you really wanted to, you could pull one of the drivers and put the exciter on the inside surface of the cabinet.
I did another test yesterday that's similar to what you suggest and I'll put that video out tomorrow. In it I used a tone from my function generator to use as a reference for both measurements. The tone is sent to the panel exciter and is picked up by both the mic and the accelerometer.
@@IBuildIt I guess brilliant minds think alike :) My only suggestion then is to run two separate tests. Run the tone through the exciter and measure the output with the mic like you said, but also run the tone through the speaker at a sensible listening volume to get a baseline measurement for the accelerometer.
@@zackw4941 I'm surprised that there weren't more suggestions (even ones that aren't good) on how to improve the test. That's usually the first thing you get whenever you do something, a bunch of "I would have done it this way" type comments :)
Thank you John for this very interesting investigation. For the sake of bringing it down to a practical level it would be nice if you did a few sweeps on the Elacs but change their position or change what their surroundings. Many years ago I bought a small Denon system that my wife and I really liked when we listened to it in the show room. Brought it home and put the speakers up on the mantle in front of a mirror ( you already know where this is going ). They sounded horrible and I just thought that it was the demo room that made them sound so good. We rarely used them and listened to an old "boom box" on the floor. Years later I put them on the floor beside the couch to get them out of the way and decided to turn the system on. Out came a really nice sound. It was my first lesson in how important it is that you set up your speakers in a good location. In my case the position and environment had much more impact than the quality of the speakers.
Dear John, As I understand it, panel resonance is not so much about panels radiating sound themselves but rather about vibrating panels modulating the amplitude of the drivers and maybe even compromising the workings of the crossover network. The amplitude of the tweeter can be extremely tiny -think of 10E-9 m-, so even a tiny movement of the cabinet may distort that ... . When you put an accelerometer of the woofer cone, you are measuring the combined vibration of woofer and cabinet ... . For my very heavy Thiel CS6 loudspeakers I use IsoAcoustics feet and Omicron Stabilizers on top. Both clearly bettered the sound. Anyone can hear that. Even stranger: in my small home cinema set, I use a B&W subwoofer with opposing drivers. The cabinet doesn't vibrate. Placing the sub on Aurios feet made the total sound of that system al lot better ... how can that be? Or am I not allowed to hear that? Just "making it up?" These phenomena need to be explained rather than "debunked" by measurements that may not be that relevant after all ...".
What about putting exciters on cabinet walls, and feed them with 180 phase shifted signal? This would reduce cabinet vibrations, and make the leakage smaller, effectively acoustically "stiffening" the walls. Same idea can be applied for any parasitic resonant surface in studio. This technology is already used to reduce street and traffic noise inside apartments.
the thing is - accelerometer measures acceleration not amplitude. so data from accelerometer is not a frequency response. more so - the stiffer measured system is the larger response from accelerometer will be. if you want to measure amplitude of speaker panel vibration it must be more like magnet glued to a panel and a wire coil next to it (fixed independantly).
Good work!, But you can replace the woofer with a midrange (back closed) pointing in to the box and measure only at resonance (near 1 kHz) at different power levels to deduce panel "sensitivity" in SPL...fully calibrated
interesting test. Now im wondering what this test would be like if you tested a Harbeth speaker, on some models they tune cabinet resonance to help create warmth.
Every structure has a resonant frequency, and bracing affects what the resonant frequency is. I think an interesting project would be to start with an unbraced cabinet, and take measurements. Then add bracing and re-measure. The goal would be to move the peak resonance outside the range the speaker is meant to play. And to keep the resonance as low as possible. Maximum theoretical dynamic range of human hearing is ~90db, but 60db is plenty good enough when you're listening at 70-80db. (Minimum noise floor is going to be 20-30db in a very quiet environment.)
Nice video! When you have the accelerometer attached to the panel are you measuring the vibration at one point and if so, how do you generalize it to the vibration of the entire panel. Thanks!
Push-Push design will remove a significant amount of reaction energy from entering the box. The box can be underbuilt without concern for panel resonances.
Resonance doesn’t need debunking because it’s all about context. A large wide baffle floor stander, Inadequately braced, will absolutely colour the sound. An accelerometer proves the amplitude of resonance at each tested frequency. At high frequencies, the whole panel area acts like an out of phase drive unit so even 0.1mm amplitude will colour the sound. That’s why speaker engineers try and get panel resonance down to below 45dB under programme by design. It can indeed be seen in some waterfall decay plots for larger speakers. Little stand mounts will exhibit tiny amounts of resonance, probably not audible. Not so with large speakers, fact.
Name one medium to higher end speaker manufacturer that makes speakers with large unbraced panels. Time's up, I'll answer: none. Maybe Harbeth, but they are using the cabinet as if it's a musical instrument (their claim). What I'm "debunking" (word used to get people to click on the video, welcome to TH-cam) is the ridiculous idea that panel resonance is a major problem that needs to be addressed. It isn't. If the wall beside the speaker is resonating as much or more than the speaker panel itself, it's absolutely a non-issue for reasonably well made speakers. Even if your speaker panels are perfectly damped, the walls (and the room itself) will resonate, swamping anything you think you might hear from the speaker cabinet. That's basic physics. Speaker manufacturers that are bragging about using special materials to reduce resonances are doing it as a marketing point, since the idea that panel resonance is a major problem has been ingrained in the buyers.
Could you try the test again where you put the accelerometer in the geometric center of the side panel? Also, I think it would be interesting to put it on the baffle since that is probably where the most detrimental panel vibrations occur. Speakers with much larger unbraced enclosures of which there are no doubt a lot on the market, should show vibrations that are worse. Thanks!
There's a brace in the centre of the panel. And it wouldn't make that much of a difference, anyway. If the panel vibrations are the conservative -40db, the speaker output is 16 times louder. You are measuring tadpoles in the belly of a whale.
@@IBuildIt Hmm you are of the opinion that the efforts that high end manufacturers like Magico go to is just fooling people who have too much money burning holes in their pockets? I think those Elacs are very small panels and apparently well braced which is a good thing no doubt.
It's not opinion, it's fact. Audiophiles are some of the most gullible people on the planet and will buy into anything that sounds "science based" quite readily. Even people who seem to be very rational in their other affairs slip off into Wonderland where audio tweakery is involved. And to be completely fair to the manufacturers, you all make it easy for them to take full advantage. That inability to properly put issues related to audio quality into perspective drives a lot of sales.
@@IBuildIt If there is ever a class action lawsuit against these high end manufacturers where they need an expert witness maybe you are their man. I love all the clever things you have constructed in your shop and even if we don't agree totally on audio tweaking necessarily, I give you full respect for craftiness! Thanks for delving into the technicals of measurements instead of just giving subjective superlatives. I'm in the camp that the only thing I want vibrating is my voice coil and diaphragm but it is necessary to put things in perspective always. I also believe that having vibrational energy resonating in any components in the speaker can feed back into and distort the movements of the voice coil. Can you measure that? I think you can but it has to be done right and I'm sure it is not easy.
I don’t understand why you applied a 40-60dB offset on your measurements that were both made with the pink noise recorded into Audacity. With the same measurement device at the same gain (I assume from what you said in the video) wouldn’t the relative levels of each recording be on par with each other? Interesting video!
What would be interesting is to see how cabinet resonance causes cabinet internal presure fluctuation, causing variations in driver impedance and therefore sound signature.
this is interesting, and since you measuresed one speaker sidewall, would not all the cabinet walls vibrate and add to the output? then the inside also reflect back thru the woofer again or port, all this adds up.
The test was trying to isolate the panel vibration only and doesn't include and sound that just goes right through the panels or out the port or through the cone when the speaker is playing normally. More panels vibrating would mean more output, but still at a very, very low level.
Without sounding like I am invalidating your experiment altogether, I will say some of it is invalid. Hear me out please. double sided tape will have a very large effect on your measurements, much larger than you think. Your method of measuring the vibration of the cone and box is the correct method given your gear. This allows you to directly compare the *amplitude* of the box vs the driver. Attempting to normalize an air measurement against a box measurement becomes progressively less accurate. My estimate is that your confidence interval has ballooned to something on the order of +/- 10% (20% total). Unless you have a solid background in stats, you will not be able to overcome this deficiency in your toolset. I think your conclusion is based on an incomplete premise as well. Your box vs cone measurements showed between -60dB and -75dB with many large distinct peaks over the area that human hearing is most acute. You will certainly be able to hear this, and since the sound will arrive out of phase, and potentially mixed with wall reflections, this could have a pretty large impact on the sound.
What sort of results would get if you ran the same source audio but inverted phase through the panel exciter on the wall? Wondering if this result in cleaner audio as a result or if the harmonics etc would just compound.
What about if you left the accelerometer in free air and ran the sweep. Would that tell you the reference level and then see how much the wall vibrates in comparison?
4. There is this japanese company that makes bookshelf speakers from thin wood, idea is that they feel more roomy when the panels vibrate. I have also some thin walled speakers it kind of works when you give the speakers room etc . I will write the company name if i can remember it later.
Plato's cave for Audio.... A blind man hearing someone using headphones will barely hear no sound, yet the man using headphones will hear a symphony... As far as i know the main problems with enclosure resonances are internal, not external... Standing waves and enclosure vibrations bounce back and excite exposed driver's cone affecting the sound quality
If if your measurements are out by a factor of 2x it still shows a massive difference, so your point still holds. "Not everyone is good at keeping things in perspective"... I feel attacked... 😆🤣
Would be interesting to hear more of your opinion about those panel exciter speakers they rave about on those videos. I had my doubts when I first saw them!
I am 62 seconds into this video. I would expect more problems from the resonance of the cabinet causing the moving of the driver physically in space (and thus distorting the signal directly) than by acting as another diaphragm that causes additive/subtractive pressure waves in air.
What's hard to differentiate is the amount of sound that's coming from the speaker operating normally, and the amount of sound that's coming from the vibrating panels of the box. Or the amount of sound that's coming from the vibrating wall.
Tech Ingredients are far better at analysis than you are and having made his speakers can testify to the fact they are far far better than any other speaker I could afford to buy.
While the method I used wasn't perfect, it can put the "problem" of panel noise in perspective. The output from the panels will typically be so low that on their own they may be barely audible, but also completely swamped by the output from the speaker itself.
The issue I have with my test setup is that I don't have an accurate measure of the difference between the output of my accelerometer and my measurement microphone. Putting the accelerometer on the woofer cone and then on the panel gives a very rough approximation of the difference between how much each of these are vibrating, and the amount I can use to offset the panel measurements. But even the conservative estimate of 40db pushes the panel output down below audibility.
Obviously the panel output becomes more of a problem if the speaker has larger, thin unbraced panels that are undamped. So this isn't me saying that speaker construction can be slack. It's me saying that the single minded obsession with making a completely inert speaker cabinet is a waste of time and resources. Especially when you consider that the walls near the speaker will vibrate just as much (or more) as the speaker cabinet itself.
Now I should point out that when the accelerometer is attached to a panel, it's measuring how much the panel vibrates, and how much the entire speaker is vibrating. Unless the speaker is rigidly held so that it cannot move, there will always be some contribution from the cabinet vibrating as a whole.
Interesting use of available tools and techniques to correlate and quantify the enclosure panel vibration conundrum. In spite of the 40-60dB offset (depending on the level of margin applied to accuracy), the open baffle enthusiasts often cite elimination of enclosure panel resonances as a key advantage. Listening to music at 85dB, I'm not sure if 25dB of resonance is something that would tangibly interfere with the pure driver output in a way that it would be "clearly obvious" if the enclosure were then removed. Not to mention, the directivity of the panels might also be a factor. How much output from a -50dB side panel is actually making it's way to the listener's ears? Very interesting work. You certainly get the neurons firing on all 8.
Hi John! I've been thinking about this issue now. Recently, I rounded the edges of the baffle and it made a huge difference. So much so that I thought I'd tackle box resonance. Thinking on this I think that people are confused about these box resonances. In your video, you show that the box panels don't make sound, and draw a comparison to the wall to show that resonances are happening everywhere. I agree with your findings, that the box panels are not adding sound. But that's not the problem.
The problem is that box vibrations color the sound because they vibrate the drivers themselves, as they are in the act of making sound. For bookshelf speakers the answer is simple, foam under the speaker absorbs resonances, also known as decoupling. I use it, and the results are great! However, for large speakers like my floor standers that weigh 60 pounds, it's not really an option, and all bass notes are not fully absorbed by foam. What I'm thinking of trying is coupling the baffle to the back of the speaker box using dowels, well fitted, and screwed in from both sides. The vibrations from the baffle will be roughly halved, transmitted directly to the back panel. Those vibrations will be more or less out of phase with the front panel, causing destruction of the vibrations meeting on the side panels. Coupling these panels over nearly a foot should do more to stiffen the baffle than anything I can think of, basically creating a stressed skin panel whose rigidity is a function of the distance between panels.
The way we brace boxes does not seem in harmony with trying to deal with these resonances by cancellation. In the current bracing model, vibrations go directly to the sidewall first basically just spreading out vibrations to all the panels, but with no objective of cancellation. The only thing I can't account for in this mental exercise is the sound waves coming from the back of the woofer. When it hits the back wall, will it be cancelled out by the vibrations transmitted to the wall, or will it be additive? Will it be in phase with them, or out of phase? Will the dowels be able to transmit the vibrations immediately enough to cause cancellation?
For years I've read that the box is the worse shape for these resonances, but if these equidistant walls are coupled, there might be a way to make box speakers great again.
I am o think an audiophile. We are insane, the hobby is insane, we, all of us will never be 100% happy with our gear. All of this and cables , cable lifters, the twice clock. It is all nonsense , treat your room and hope no one of importance notices this nonsense. Go listen to some music! Live a little!
Use the exciter as active noise cancellation on the panel. Flip the polarity of the accelerometer recording and play it back into the panel. Would be interesting to see the result even if you couldn't hear the difference.
yes, this.
im not sure the wallhas the rigidity to do that compared to a 12 inch mdf board or a driver's cone? does it?
Man, I am really digging your channel! Working on a pair of partially open baffle speakers right now and your past few videos have made me consider a lot of things.
yeah I think that relative measurement cone vs cabinet is a valid one and about what I'd expect. Although the cabinet surface area is much larger than the speaker but also most is facing away from the listener.
Maybe it would be a different story once you hit the cabinets resonance frequency.
In a 400 level engineering course i was assigned a project to measure the reduction in vibration hammering on a chisel bare & with various chisel safety top-caps. Setting up the accelerometers and processing the raw output to be meaningful for analysis turned out to be such an involved task that it was eventually broken out to a separate project for another student to tackle while i worked specifically on the mechanics of the apparatus.
Fascinating test. Seriously good topic for a deeper dive.
Interesting subject John. I think music that is bass heavy and have transients from drum solos would be the best samples for cabinet resonance measurements.
The key point to remember is that since everything has a resonant frequency the challenge in cabinet design is to keep that resonant frequency below the lower range of the speaker. Otherwise, the whole cabinet will vibrate uncontrollably whenever that frequency is hit.
Adding mass lowers the resonant frequency. Of course, this is not a problem in open baffle designs.
I've done a little testing with various materials(cement with dampening, ceramic, composites...), best one I came up with was with constrained layers..., Haven't really implemented it yes as right now I'm just building smaller bookshelf speakers witch don't suffer as much. Plus I've also never thought this was something that is degrading the sound... It's the whole diminishing returns thing.
One way to measure the actual sound produced by a side panel is to do the recordings in an anaechoic chamber with the front panel isolated from the sides?
Nice video but I wonder if the accelerometer should have been taped to the side panel in the center (assuming its a cheap cabinet without bracing), maybe it would have the largest amount of vibration there rather than say 1/3 of the way down. Just a thought.
I'm positive that you could also use the exciter like a microphone. It will need some sort of a pre amp to make it work properly but it is viable I think. It might be a more accurate measurement device for vibrations in this fashion. Very cool content!!
You did this with an Elac. Those things are built far better than average, as far as store bought speakers go. -67 db is probably a significantly better reading then you'd get from a lot of other speakers. See if you can borrow some Klipsch RP-600's or Wharfedales. The GR-Research channel has done measurements of those, and they're not as well built as the Elacs. Also, maybe you can test the application of the GR-Reserarch No-Res product?
John this was excellent and answered the concern fairly well. I'm especially impressed by all the work you did to try and put it into perspective with the decibel levels between resonance vs driver output. 40 Db difference is huge and wouldn't be very audible. Heck my forced air heating and cooling runs about 20 Db lower and I can't hear that when playing music. I'm into speaker building as well and may try the exciter when deciding where to put bracing for the greatest effect. Honestly, I think that most of the benefit of a well braced cabinet is more for a placebo effect when one feels better about the sound because the cabinet passed the knock test. That being said, I wouldn't be surprised if a completely unbraced, untreated cabinet made with thin material may exhibit some audibility when playing music continuously vs a sweep, but only at substantial SPL. Recently in a current project I'm working on, I could swear there was an audible increase in Resonance during guitar playback... It was a much larger cabinet with substantially more potential for resonance due to panel size vs a well built, thick bookshelf speaker.
I'd consider resonance and bracing tuning to be a more technical and complex level of speaker building. Does it make a difference? absolutely. The goal of building a speaker yourself is essentially making it sound how you want, at a price you can afford. Resonance is a variable that changes sound, almost always in a detrimental way. The biggest piece of information missing from this experiment (aside from only using one unit) is failing to account for off-axis sound. The biggest issue with resonance is that the box itself is now emanating sound in all directions, not just the drivers. This results in peaks or dips that interact in unpredictable ways in a room, and the frequency response and cannot be EQ'd out.
On-axis measurements may still look very good, but when you do various measurements off axis, you'll see resonances have the same peak or dip in all measurements. The most direct impact of this is poor soundstage/ imaging, as a given instrument may seem to 'move around' randomly as it gets amplified and quieted by the box itself. For the record I'm not just parroting theory written by others, I've incorporated adding bracing and there's definitely an impact.
quite interesting... excellent little experiment. i was a bit surprised @6:35 by the high fqs coming from the speaker cabinet. some of those vibrations are out of phase w/ the driver, right? also the separate direction...
i'm imagining the wall is stealing energy thru it's vibrations, like a membrane absorber, and what we hear is what gets reflected.
like when a sound wave applies pressure on the wall the wall deform away under that pressure...
This is a really interesting test which certainly gives some interesting results worth exploring.
With small bookshelf models like the Elac, panel resonance isn't often a major issue. Where it becomes more of an issue is with large, unbraced cabinets like many of the larger Klipsch "Heritage" models, especially the Forte, Cornwall and La Scala, along with many older, boxy-style speakers. The bigger those panels are, the lower their resonant frequencies and the more freely those panels will resonate, making them much more noticeable, especially during playback.
It would be interesting to se you do a similar test with a much larger cabinet, then repeat the test after adding bracing throughout the cabinet.
Isn't the wall behind the speaker a big enough unbraced panel?
@@IBuildIt in some sense, sure. but it's also not receiving the same pressures as what would be seen within the box, just the parts of the front waves that wrap around the speaker or comes out from a rear port, if one is present.
If I can come up with a more accurate way to set the offset between the mic and the accelerometer, I'll try it again with something bigger. I doubt there will be much of a difference, unless it's something very poorly made.
Worst case enclosure vibrations are when a musical section sound notes are approximately matched...mostly these will be sustained for some time enough to become discernable because basically that's what music really is..those musical notes may repeat/ sustain and is not like a frequency sweep... that's done only once , probably for a few hundreds of millisecs and then gone....
the panel mass has a latency/ delay that may not respond as quickly to vibrate...
so it's best to test with actual sound clippings / track in the +/-3dB frequency range that causes the worst enclosure performance...@ different SPL levels and correct for that accordingly....
I hear these resonances on certain songs, like a bloated bass or drone at certain frequencies, if the note or passage is sustained or recurring. Would you say this is cabinet resonance I am hearing? sometimes I find certain song unlistenable due to this. I have the speakers positioned well out from the front or side walls with plenty of soft furnishings so i don't think its the room. Have you any experience with No-Rez? I am considering trying this but want to know for sure whether it is indeed cabinet resonance I am hearing. speakers in question are wharfedale Linton Heritage.
Thanks for more great stuff! Could you disconnect any tweeters and mid-ranges and put a tube running from the woofer into a nearby room and heavily insulate that tube. Then the sound of the room picked up by a microphone would be predominantly that transmitted by the cabinet and all its surfaces at that? Thanks again!
Interesting analysis and presentation. Fascinating to see the vibration in the walls. The only problem is when you brace a speaker of take care to engineer out box vibration, the perceived sound improves. In open baffle, I think the biggest contribution is the front baffle being stiff and massive. Also a rigid frame in the drivers. With this, you get the full dynamic punch of the cone vs a mixture of the cone moving and the baffle recoiling.
Hearing perception is influenced by what we know and what we see. If you know the speaker is solidly made and you think that's an important feature, you will perceive the sound to be better.
Agreed. I epoxied 1/4" steel semi rings behind the midwoofers in my Polk 7B's and lets just say that the improvement in bass punchiness was not subtle. I also feel less vibration through my fingertips when I touch the baffle. Could be better still and I'm thinking about a couple of 3/4" oak dowel braces going from the front baffle to the rear enclosure. Lots of little incremental improvements are additive and I think each one of them could fail a blinded A/B test but no way would the sum of them be an insignificant improvement. So, do we convince ourselves to not even do the small incremental improvements? Not me.
How many bacteria does it take to fill a shot glass?
@@captainwho1 Hey! I'm on this channel looking to find exactly what your describing. I'm thinking that dowels from the baffle to the back would not only reinforce the baffle, but also vibrate the back panel opposite to the front. These vibrations could also cancel out the sound waves from the front panel because the back panel is vibrating oppositely in a push pull configuration. The side panels wouldn't vibrate because the destructive vibrations from the front and back would meet on the sides. Is this what your thinking too? Could you elaborate on what you did with the polks? In a constrained layer thread some guys were saying that opposing woofers fix this problem, so I'm thinking that coupling panels would do a similar thing.
@@headphone307 By coupling the baffle with the rear, you increase the effective rigidity of the baffle and also raise it's resonant frequency and lower it's amplitude of vibrations. Another method is to use an adjustable tensioning rod that couples the woofer to the enclosure rear. Jean Mauer uses this, combined with a sand filled rear wall. B&W Nautilus development may have employed woofer coupling to rear as well.
Very nice video and tests! One thing occurred to me though and I don't know if it means anything. A microphone is taking a sample of the sound energy that has made it into the air and is indicative of what our ears will pick up. The accelerometer is taking a sample of the vibration that has made it into a hard surface. The size of that surface will influence how much sound volume is transmitted into the air. The side of the speaker vibrates much less than the driver, but is also much larger. The wall in the room may vibrate less than the side of the speaker, but the wall has far more surface area than the side of the speaker. (And if the wall doesn't, you have a VERY understanding partner!)
Anyway, what you might want to try next is to take your exciter and stick it to the side of the speaker. Measure the vibration with the accelerometer. Use just the exciter to reproduce the same vibration level as the speaker playing normally. Then see how much volume the microphone picks up from whatever distance you measured the driver. That said, I don't know if the exciter itself will make noise and invalidate such a test. I suppose if you really wanted to, you could pull one of the drivers and put the exciter on the inside surface of the cabinet.
I did another test yesterday that's similar to what you suggest and I'll put that video out tomorrow. In it I used a tone from my function generator to use as a reference for both measurements. The tone is sent to the panel exciter and is picked up by both the mic and the accelerometer.
@@IBuildIt I guess brilliant minds think alike :) My only suggestion then is to run two separate tests. Run the tone through the exciter and measure the output with the mic like you said, but also run the tone through the speaker at a sensible listening volume to get a baseline measurement for the accelerometer.
@@zackw4941 I'm surprised that there weren't more suggestions (even ones that aren't good) on how to improve the test. That's usually the first thing you get whenever you do something, a bunch of "I would have done it this way" type comments :)
@@IBuildIt That makes sense. I read through to make sure I wasn't repeating anyone else. It is a challenging problem you took on and very well done!
Thank you John for this very interesting investigation. For the sake of bringing it down to a practical level it would be nice if you did a few sweeps on the Elacs but change their position or change what their surroundings. Many years ago I bought a small Denon system that my wife and I really liked when we listened to it in the show room. Brought it home and put the speakers up on the mantle in front of a mirror ( you already know where this is going ). They sounded horrible and I just thought that it was the demo room that made them sound so good. We rarely used them and listened to an old "boom box" on the floor. Years later I put them on the floor beside the couch to get them out of the way and decided to turn the system on. Out came a really nice sound. It was my first lesson in how important it is that you set up your speakers in a good location. In my case the position and environment had much more impact than the quality of the speakers.
Dear John,
As I understand it, panel resonance is not so much about panels radiating sound themselves but rather about vibrating panels modulating the amplitude of the drivers and maybe even compromising the workings of the crossover network. The amplitude of the tweeter can be extremely tiny -think of 10E-9 m-, so even a tiny movement of the cabinet may distort that ... .
When you put an accelerometer of the woofer cone, you are measuring the combined vibration of woofer and cabinet ... .
For my very heavy Thiel CS6 loudspeakers I use IsoAcoustics feet and Omicron Stabilizers on top. Both clearly bettered the sound. Anyone can hear that.
Even stranger: in my small home cinema set, I use a B&W subwoofer with opposing drivers. The cabinet doesn't vibrate. Placing the sub on Aurios feet made the total sound of that system al lot better ... how can that be? Or am I not allowed to hear that? Just "making it up?"
These phenomena need to be explained rather than "debunked" by measurements that may not be that relevant after all ...".
What about putting exciters on cabinet walls, and feed them with 180 phase shifted signal? This would reduce cabinet vibrations, and make the leakage smaller, effectively acoustically "stiffening" the walls. Same idea can be applied for any parasitic resonant surface in studio. This technology is already used to reduce street and traffic noise inside apartments.
the thing is - accelerometer measures acceleration not amplitude. so data from accelerometer is not a frequency response.
more so - the stiffer measured system is the larger response from accelerometer will be.
if you want to measure amplitude of speaker panel vibration it must be more like magnet glued to a panel and a wire coil next to it (fixed independantly).
Good work!, But you can replace the woofer with a midrange (back closed) pointing in to the box and measure only at resonance (near 1 kHz) at different power levels to deduce panel "sensitivity" in SPL...fully calibrated
Speaker panel might be 60 dB down but you need to compare the relative area of the entire speaker enclosure versus the speaker cone
Another panel vibrating at the same amplitude will add 3db.
have you made the foam DML speakers and listen to them? would be cool to see.
interesting test. Now im wondering what this test would be like if you tested a Harbeth speaker, on some models they tune cabinet resonance to help create warmth.
Every structure has a resonant frequency, and bracing affects what the resonant frequency is. I think an interesting project would be to start with an unbraced cabinet, and take measurements. Then add bracing and re-measure. The goal would be to move the peak resonance outside the range the speaker is meant to play. And to keep the resonance as low as possible. Maximum theoretical dynamic range of human hearing is ~90db, but 60db is plenty good enough when you're listening at 70-80db. (Minimum noise floor is going to be 20-30db in a very quiet environment.)
Nice video! When you have the accelerometer attached to the panel are you measuring the vibration at one point and if so, how do you generalize it to the vibration of the entire panel. Thanks!
At 2:06, did I hear a short but audible buzz/rattle from the speaker?
Push-Push design will remove a significant amount of reaction energy from entering the box. The box can be underbuilt without concern for panel resonances.
Resonance doesn’t need debunking because it’s all about context. A large wide baffle floor stander, Inadequately braced, will absolutely colour the sound. An accelerometer proves the amplitude of resonance at each tested frequency. At high frequencies, the whole panel area acts like an out of phase drive unit so even 0.1mm amplitude will colour the sound. That’s why speaker engineers try and get panel resonance down to below 45dB under programme by design. It can indeed be seen in some waterfall decay plots for larger speakers. Little stand mounts will exhibit tiny amounts of resonance, probably not audible. Not so with large speakers, fact.
Name one medium to higher end speaker manufacturer that makes speakers with large unbraced panels.
Time's up, I'll answer: none. Maybe Harbeth, but they are using the cabinet as if it's a musical instrument (their claim).
What I'm "debunking" (word used to get people to click on the video, welcome to TH-cam) is the ridiculous idea that panel resonance is a major problem that needs to be addressed. It isn't.
If the wall beside the speaker is resonating as much or more than the speaker panel itself, it's absolutely a non-issue for reasonably well made speakers. Even if your speaker panels are perfectly damped, the walls (and the room itself) will resonate, swamping anything you think you might hear from the speaker cabinet. That's basic physics.
Speaker manufacturers that are bragging about using special materials to reduce resonances are doing it as a marketing point, since the idea that panel resonance is a major problem has been ingrained in the buyers.
Could you try the test again where you put the accelerometer in the geometric center of the side panel? Also, I think it would be interesting to put it on the baffle since that is probably where the most detrimental panel vibrations occur. Speakers with much larger unbraced enclosures of which there are no doubt a lot on the market, should show vibrations that are worse. Thanks!
There's a brace in the centre of the panel. And it wouldn't make that much of a difference, anyway. If the panel vibrations are the conservative -40db, the speaker output is 16 times louder. You are measuring tadpoles in the belly of a whale.
@@IBuildIt Hmm you are of the opinion that the efforts that high end manufacturers like Magico go to is just fooling people who have too much money burning holes in their pockets? I think those Elacs are very small panels and apparently well braced which is a good thing no doubt.
It's not opinion, it's fact. Audiophiles are some of the most gullible people on the planet and will buy into anything that sounds "science based" quite readily. Even people who seem to be very rational in their other affairs slip off into Wonderland where audio tweakery is involved.
And to be completely fair to the manufacturers, you all make it easy for them to take full advantage. That inability to properly put issues related to audio quality into perspective drives a lot of sales.
@@IBuildIt If there is ever a class action lawsuit against these high end manufacturers where they need an expert witness maybe you are their man. I love all the clever things you have constructed in your shop and even if we don't agree totally on audio tweaking necessarily, I give you full respect for craftiness! Thanks for delving into the technicals of measurements instead of just giving subjective superlatives. I'm in the camp that the only thing I want vibrating is my voice coil and diaphragm but it is necessary to put things in perspective always. I also believe that having vibrational energy resonating in any components in the speaker can feed back into and distort the movements of the voice coil. Can you measure that? I think you can but it has to be done right and I'm sure it is not easy.
I don’t understand why you applied a 40-60dB offset on your measurements that were both made with the pink noise recorded into Audacity. With the same measurement device at the same gain (I assume from what you said in the video) wouldn’t the relative levels of each recording be on par with each other? Interesting video!
I didn't. The Audacity recording was to figure out the offset.
Okay. That makes way more sense. Thanks
Mathias Wandel would be proud of you!
What would be interesting is to see how cabinet resonance causes cabinet internal presure fluctuation, causing variations in driver impedance and therefore sound signature.
Thx! Exactly what I was "theoreting" around with in my brain the last days✊🕴👍
Excellent! This is a no-nonsense, bullshit-free zone, that many audiophiles would be well advised to visit.
this is interesting, and since you measuresed one speaker sidewall, would not all the cabinet walls vibrate and add to the output? then the inside also reflect back thru the woofer again or port, all this adds up.
The test was trying to isolate the panel vibration only and doesn't include and sound that just goes right through the panels or out the port or through the cone when the speaker is playing normally. More panels vibrating would mean more output, but still at a very, very low level.
всегда смотрю вас с интересом
Without sounding like I am invalidating your experiment altogether, I will say some of it is invalid.
Hear me out please.
double sided tape will have a very large effect on your measurements, much larger than you think.
Your method of measuring the vibration of the cone and box is the correct method given your gear. This allows you to directly compare the *amplitude* of the box vs the driver. Attempting to normalize an air measurement against a box measurement becomes progressively less accurate. My estimate is that your confidence interval has ballooned to something on the order of +/- 10% (20% total).
Unless you have a solid background in stats, you will not be able to overcome this deficiency in your toolset.
I think your conclusion is based on an incomplete premise as well. Your box vs cone measurements showed between -60dB and -75dB with many large distinct peaks over the area that human hearing is most acute. You will certainly be able to hear this, and since the sound will arrive out of phase, and potentially mixed with wall reflections, this could have a pretty large impact on the sound.
What sort of results would get if you ran the same source audio but inverted phase through the panel exciter on the wall? Wondering if this result in cleaner audio as a result or if the harmonics etc would just compound.
What about if you left the accelerometer in free air and ran the sweep. Would that tell you the reference level and then see how much the wall vibrates in comparison?
4. There is this japanese company that makes bookshelf speakers from thin wood, idea is that they feel more roomy when the panels vibrate. I have also some thin walled speakers it kind of works when you give the speakers room etc . I will write the company name if i can remember it later.
I think you're referring to Audio Note UK which is historically related to Audio Note Japan.
Kiso Acoustic. Greetings from Denmark
I'll concur. Unless there is something afoul in your construction, the cabinet contributes very little to the perceived audible spectrum.
Got my back, brother! :)
@@IBuildIt and your sides! :D
The audibility of the resonance depends on the frequency of the resonance
Partly, but if it's 16 times less loud than the output of the speaker it doesn't matter what frequency it is - you aren't going to hear it.
Strongly agree
Plato's cave for Audio.... A blind man hearing someone using headphones will barely hear no sound, yet the man using headphones will hear a symphony... As far as i know the main problems with enclosure resonances are internal, not external... Standing waves and enclosure vibrations bounce back and excite exposed driver's cone affecting the sound quality
Wilson Audio, "Shut it down... shut it all down."
If if your measurements are out by a factor of 2x it still shows a massive difference, so your point still holds.
"Not everyone is good at keeping things in perspective"... I feel attacked... 😆🤣
Would be interesting to hear more of your opinion about those panel exciter speakers they rave about on those videos. I had my doubts when I first saw them!
I am 62 seconds into this video. I would expect more problems from the resonance of the cabinet causing the moving of the driver physically in space (and thus distorting the signal directly) than by acting as another diaphragm that causes additive/subtractive pressure waves in air.
Sound versus vibration is difficult to differentiate as sound is vibration!
What's hard to differentiate is the amount of sound that's coming from the speaker operating normally, and the amount of sound that's coming from the vibrating panels of the box. Or the amount of sound that's coming from the vibrating wall.
some known electronic attenuators would be a useful tool to bring into these measurements.
Audacity ❤️
oh i definitely get audible panel vibration.
from my furniture
You can use a laser to glance off the side and measure that.
Do you sleep with your writting device 🤔 🤣👍🏻
Myth buster ;-)
I say; Master Heisz
This guy is awesome
Tech Ingredients are far better at analysis than you are and having made his speakers can testify to the fact they are far far better than any other speaker I could afford to buy.
Just 😆
I don't agree with your results.
Good morning, professor, thanks for your teachings. I follow your advice and in a group of themakersmob and TH-cam... an excellent weekend
walls only vibrate if you make them from paper like they do in the freedom country
GR research Gary ? Is not going to like some of your material lol
Thanks for doing these mythbusting videos. There is too much useless folk wisdom that the DIY audio community could get rid of.