• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Speaker enclosure vibrations - a few measurements with accelerometer

Thomas_A

Major Contributor
Forum Donor
Joined
Jun 20, 2019
Messages
3,866
Likes
2,784
Location
Sweden
There are been quite a number of threads dealing with vibrations in speaker cabinets, both in relation to "speaker feet" and vibrations in enclosures themselves. I recently took out my ACH01 accelerometer to do some measurements, both speaker feet and enclosures. There was some initial measurements posted in another thread, and here I have repeated these with some modifications. For test one I wanted to test speaker feet, i.e. the so-called "spikes" against soft feet. However, I do not want to destroy my wooden floor or my speakers so I choose to put the speakers on the floor and use three metal nuts as a "tripod" as the coupling case. Ideally, coupling should be the whole area under the speaker bolted to the floor, but this is not practically done. The speaker is a two-way, 9 liter volume, ported speaker with 5-inch Perless driver. It has 7 sides, seen in the picture. The difference from yesterdays measurements were that I plugged the port and used a higher volume for the sweep. Also the use of metal nuts made a better stability than the metal bars used yesterday. Setup below, with the accelerometer fastened top left with double-adhesive tape. It is usually not recommended to use tape, but for lower frequencies this should be ok.


IMG_5218.jpg


The result from the measurement with hard steel nuts as feet (purple) or soft sylomer-based pads (cyan) below. (I think the 50 Hz peak is due to that my portable no longer can be run on battery only and thus s ground loop.) Note that the accelerometer outputs a signal proportional to acceleration (not SPL or displacement).
Floor soft vs hard feet acc.png


The next shows the second harmonic distortion of the above signals:
Floor soft vs hard feet dist.png

So conclusion is that hard coupling to the floor (oak mounted on sand in this case) causes more movement in the bass region of the speaker compared to soft feet. While the movement itself may not be a problem audibly, the use of so-called rigid coupling with spikes is probably not a good idea if you want the speaker to be fixed. What may or may not be audible is the transfer of vibrations to the support and other objects which I previously showed may lead to audible distortion. Se post:



So to the second part, I want to compare two speakers of about the same form and size, but where one is built with standard 19 mm MDF and the other with 9+9 mm MDF and a damping glue (constrained layer damping). The story is that I at heard some irritating distortion on my old speakers and also on my new built ones, both using 19 mm MDF and in principle identical encloure form but with different drivers. I rebuilt cabinets a couple of times but got the same annoying distortion around 500-550 Hz. So I decided to use constrained layer damping and after that I found no more distortion. The new cabinet also had a significantly different sound in "knock-test". Sow how do they measure? Below is the new and old speaker (I have lost the other enclosures, so this is a comparison with the original and my latest).

New:

IMG_5219.jpg


Old:
IMG_5220.jpg


I measured front, side and top of the speaker. The interesting part was the difference in distortion (below 2nd harmonic from the acc measurements) from the speakers:

Top:
Top baffle 2.png

Front:
Front baffle 2.png

Side:
Side baffle 2.png

Basically I could repeat the measurements from yesterday, shown in another thread. What is interesting are the lack of peaks the 500-700 Hz region in the constrained layer enclosure (but shifted to lower frequencies?). Now, I do not have any microphone measurements as yet on this since lack of time here, but I would almost be certain that these peaks are the reason for the audible distortion heard from the speakers.

So my conclusion is that 19 mm MDF is not always sufficient to get silent enclosures even in very small speakers as this. Bracing might help, but I think constrained layer damping is even better.

A link to the previous thread with measurements.

 
Last edited:
Very interesting, thank you!

As for constrained layer damping: I became somewhat interested in the concept when I was researching how to construct an isolation platform for my new turntable, several years ago. Ultimately the most effective isolation - both by subjective "feel" (hand feeling for vibration transfer) and objectively - measured by a vibration app - were Townshend spring footers under a butcher block, on which my turntable was placed.

But I had a bunch of materials left over from my experiments and I threw them together for the heck of it. I made sandwich of MDF (one side thicker than the other), some wall-damping material, a sheet of steel, and more wall damping material. I put it under my butcher block as well. I didn't end up measuring it with the vibration app, but my goodness that constrained layer damped piece was that one solid item! Rapping my knuckles on the MDF yielded no "wood" thonk sound whatsoever - it was like rapping rock!

ETA: Which reminds me: Sometimes I've wondered what the results would be of going in to a commercial speaker and re-enforcing the cabinet with such constrained layer damping. Clearly most speaker companies are constrained by costs, including anything that increased shipping weight etc, so they work with a best-bang-for-the-buck approach for cabinet design. But if you had speakers and could actually open them up, and layer them as I just described above, what type of plausible benefits might one hear? And what would be the liabilities? Might you change some carefully controlled Q-factor resonance in a speaker for the worse? Might you change internal cavity size in a way that affects the design negatively, etc?
 
Last edited:
Thank you very much for your effort! There is a long history in loudspeaker development regarding enclosure resonances and how to deal with them. And of course it is prone to some utter subjective bullshit with no relevant data to back it up.

I think the German manufacturer Audiodata was one of the first with a scientific approach back in the 80s of the last century. IIRC they had an agreement with the University of Aachen, to measure enclosure resonances with a laser in order to find out how to get rid of them. The result was the Audiodata Bijou, which was plastered with hawaphon damping mats at the inside of the cabinet.

The result was a remarkable "clean" sounding speaker (it had a lot of other flaws though) which I used to own for a few years.

Are you able to do some measurements on a speaker, which has already been tested here? Would be nice to see the effects of spikes and soft feet in comparison to the distortion measured by Amir...
 
Very interesting, thank you!

As for constrained layer damping: I became somewhat interested in the concept when I was researching how to construct an isolation platform for my new turntable, several years ago. Ultimately the most effective isolation - both by subjective "feel" (hand feeling for vibration transfer) and objectively - measured by a vibration app - were Townshend spring footers under a butcher block, on which my turntable was placed.

But I had a bunch of materials left over from my experiments and I threw them together for the heck of it. I made sandwich of MDF (one side thicker than the other), some wall-damping material, a sheet of steel, and more wall damping material. I put it under my butcher block as well. I didn't end up measuring it with the vibration app, but my goodness that constrained layer damped piece was that one solid item! Rapping my knuckles on the MDF yielded no "wood" thonk sound whatsoever - it was like rapping rock!

ETA: Which reminds me: Sometimes I've wondered what the results would be of going in to a commercial speaker and re-enforcing the cabinet with such constrained layer damping. Clearly most speaker companies are constrained by costs, including anything that increased shipping weight etc, so they work with a best-bang-for-the-buck approach for cabinet design. But if you had speakers and could actually open them up, and layer them as I just described above, what type of plausible benefits might one hear? And what would be the liabilities? Might you change some carefully controlled Q-factor resonance in a speaker for the worse? Might you change internal cavity size in a way that affects the design negatively, etc?
Regarding enclosures, it will always be that "it depends". I would guess that some of the distortions in the speakers measured by Amir indeed could be enclosure-related, especially if there are peaks around 400-700 Hz in smaller and simpler made MDF-based speakers. Bracing could be a simple way to reduce those peaks but I think constrained layer damping is quite expensive and is therefore seldom a choice. Modifying existing ones with constrained layer damping is not an option IMO, it will be complicated and will change too much regarding volume etc.
 
Last edited:
Thank you very much for your effort! There is a long history in loudspeaker development regarding enclosure resonances and how to deal with them. And of course it is prone to some utter subjective bullshit with no relevant data to back it up.

I think the German manufacturer Audiodata was one of the first with a scientific approach back in the 80s of the last century. IIRC they had an agreement with the University of Aachen, to measure enclosure resonances with a laser in order to find out how to get rid of them. The result was the Audiodata Bijou, which was plastered with hawaphon damping mats at the inside of the cabinet.

The result was a remarkable "clean" sounding speaker (it had a lot of other flaws though) which I used to own for a few years.

Are you able to do some measurements on a speaker, which has already been tested here? Would be nice to see the effects of spikes and soft feet in comparison to the distortion measured by Amir...
Thanks! I don't think I'll have time to look up and do more tests of other speakers. I just had some spare time now and this accelerometer has being lying around for years to just confirm the "distortion" I heard around 20 years ago. Perhaps in 10 years when I retire. :)

I think there are more members out there accelerometers. If not, this ACH01 accelerometer with its amplifier is not especially expensive.
 
When I was developing Silver 5L I used a tone arm and a MM cartridge to find out enclosure resonances. It worked pretty well.
 
with standard 19 mm MDF and the other with 9+9 mm MDF and a damping glue (constrained layer damping).
May I ask exactly what material you used as a damping glue?
I personally have used Green Glue for my current system: https://www.audiosciencereview.com/forum/index.php?threads/murphys-corner-line-array-project.25983/
And since I found that really successful I will also use that for this ongoing project: https://www.audiosciencereview.com/...ased-on-markaudio-chr-120.31219/#post-1255658

In the US you can get Decidamp but in Europe (where I am) it is kind of an unobtanium....
 
May I ask exactly what material you used as a damping glue?
I personally have used Green Glue for my current system: https://www.audiosciencereview.com/forum/index.php?threads/murphys-corner-line-array-project.25983/
And since I found that really successful I will also use that for this ongoing project: https://www.audiosciencereview.com/...ased-on-markaudio-chr-120.31219/#post-1255658

In the US you can get Decidamp but in Europe (where I am) it is kind of an unobtanium....
I used SWEDAC DG-A2:

 
Regarding enclosures, it will always be that "it depends". I would guess that some of the distortions in the speakers measured by Amir indeed could be enclosure-related, especially if there are peaks around 400-700 Hz in smaller and simpler made MDF-based speakers. Bracing could be a simple way to reduce those peaks but I think constrained layer damping is quite expensive and is therefore seldom a choice. Modifying existing ones with constrained layer damping is not an option IMO, it will be complicated and will change too much regarding volume etc.

Thanks.

Though perhaps addressing vibration in just some relevant portions could reap benefits? E.g. the front face of a speaker holding the drivers?

Anecdotally: I had the big floor-standing Thiel CS 3.7 speakers for quite a while and had to downsize to the slightly smaller 2.7 version. Same general design, same coax mid/high drivers (different woofer). I had them at the same time comparing over something like a year. The sound was very similar but
whenever I switched from the 3.7 to the 2.7 model, it felt like there was a slight obscuring of the sound, a bit like vocals were just a bit more obscured, and a general "airier" more unboxy, open and detailed sound on the bigger model. And with harder L/R panned sound, the imaging tended to float more free of the speaker on the 3.7s, where it gets a bit more "stuck in the speaker" on the 2.7s. Just a bit, but noticeable. I didn't know what to attribute this to, but then learned from someone who worked at Thiel that the 3.7s had used a (heavy) metal front baffle in which the drivers were seated, to help reduce vibration of the baffle. To get the cheaper price for the 2.7, that was abandoned and I believe it was just some form of wood.

Again...nothing but anecdote, but I wonder how much that change of materials figured in to the sense of un-obscured detail on the bigger model.

(And I've wondered...not that I'm going to do it...if re-enforcing the front baffle of the 2.7s would get closer to the bigger model's performance in that regard).
 
Very interesting!

While it’s fun to play with accelerometer it’s very difficult to translate their readings to actual dBSPL which makes it very difficult to judge whether the resonances matter in the great scheme of things.

Please correct me if I’m wrong.
 
Thanks.

Though perhaps addressing vibration in just some relevant portions could reap benefits? E.g. the front face of a speaker holding the drivers?

Anecdotally: I had the big floor-standing Thiel CS 3.7 speakers for quite a while and had to downsize to the slightly smaller 2.7 version. Same general design, same coax mid/high drivers (different woofer). I had them at the same time comparing over something like a year. The sound was very similar but
whenever I switched from the 3.7 to the 2.7 model, it felt like there was a slight obscuring of the sound, a bit like vocals were just a bit more obscured, and a general "airier" more unboxy, open and detailed sound on the bigger model. And with harder L/R panned sound, the imaging tended to float more free of the speaker on the 3.7s, where it gets a bit more "stuck in the speaker" on the 2.7s. Just a bit, but noticeable. I didn't know what to attribute this to, but then learned from someone who worked at Thiel that the 3.7s had used a (heavy) metal front baffle in which the drivers were seated, to help reduce vibration of the baffle. To get the cheaper price for the 2.7, that was abandoned and I believe it was just some form of wood.

Again...nothing but anecdote, but I wonder how much that change of materials figured in to the sense of un-obscured detail on the bigger model.

(And I've wondered...not that I'm going to do it...if re-enforcing the front baffle of the 2.7s would get closer to the bigger model's performance in that regard).
I am not sure if it works. There is an awful lot of stuff it can be, and I do not have any statistics of how often distortion occur due to the enclosure itself. It can be loose stuff inside the cabinet, port resonances, unusual driver resonances, speaker binding posts resonating etc. Once in a while those distortions in the "small speaker MDF range" pops up, and then I am always a bit suspicious of the enclosure. Female voices, soprano or tenors usually reveal these distortions.
 
Very interesting!

While it’s fun to play with accelerometer it’s very difficult to translate their readings to actual dBSPL which makes it very difficult to judge whether the resonances matter in the great scheme of things.

Please correct me if I’m wrong.
You are correct. It is not trivial to translate this to audible distortion; I mentioned I did not have the microphone set up now, but I may do it if I have time. As shown in the link in the post, distortion was audible when I used hard feet vs. soft feet, due to quite higher order distortion. Any distortion from the enclosure itself could also be measured e.g. if I test 520, 560 and 620 Hz signals to look for differences in distortion. My listening tests where done 20 years ago or so, and I did have recordings from signals that others also heard as distortion. So I have to redo some measurements and recordings from the old speaker. Again, I am quite confident that this is enclosure. Changing driver did not remove the distortion, neither did filling level of damping material, or rebuilding cabinets which I did twice.
 
What bracing and absorption material are you using inside your speaker?

I have found the most cost effective construction method is interlocked U-shaped braces, with 3" of mineral wool absorption on the back wall, and then a double layer baffle with constrained layer damping. That is how I build my subwoofers and speakers now. Doing the whole cabinet as CLD is better of course, but consumes more time and material.
 
What bracing and absorption material are you using inside your speaker?

I have found the most cost effective construction method is interlocked U-shaped braces, with 3" of mineral wool absorption on the back wall, and then a double layer baffle with constrained layer damping. That is how I build my subwoofers and speakers now. Doing the whole cabinet as CLD is better of course, but consumes more time and material.
These small 9 liter speakers do not have any internal braces I use mineral wool filling about 40% of volume.
 
I can just add that with respect to speaker movement in the bass region using hard or soft feet, accelerometer measurements has been done many years ago as well. Below is another thread regarding the matter.


I also know that there are several other discussions on the net regarding the audibility of using soft and hard feet as speaker feet, where measurements were made using frequency response measurements. This is quite meaningless IMO, since it is distortion and not SPL that is the issue. Sometimes there are measurable and audible distortion, sometimes not. It depends.
 
What strikes me the most is the amount of low end you loose when replacing spikes with soft feet. Below 100Hz that's consistently more than 10dB. This confirms soft feet are a spring mass damper system which needs to be tuned (appropriate damping coëfficiënt) to obtain an optimum result. That's never the case with generic spring feet being sold.
 
What strikes me the most is the amount of low end you loose when replacing spikes with soft feet. Below 100Hz that's consistently more than 10dB. This confirms soft feet are a spring mass damper system which needs to be tuned (appropriate damping coëfficiënt) to obtain an optimum result. That's never the case with generic spring feet being sold.
Note that measurements are not SPL or displacement of cabinet. The measured signal is proportional to acceleration. I don't have the skills to translate it to displacement. The soft feet makes the cabinet move less in the bass region. But it is true that the mass of speaker and compliance of the feet needs to be "tuned" so that the resonance is below the lowest frequency that the speaker can reproduce. 5 Hz is good. (And the different feet you see on the picture are optimised for different speaker mass.)
 
Last edited:
I can add a note regarding distortion measurements using microphone measurements, where I found one of my original threads from 2005 in a Swedish forum (time files... now I know how old my speakers are). I measured fixed tones with a microphone and found the following for an older-built enclosure, second and third distortion component.

120 Hz: -48.6 dB (0.37%); -53.2 dB (0.22%)
160 Hz: -51.8 (0.26%); -52.4 (0.24%)
200 Hz: -55.5 (0.17%); -57.9 (0.13%)
250 Hz: -54.3 (0.19%); -63.0 (0.07%)
315 Hz: -54.1 (0.20%); -56.0 (0.16%)
400 Hz: -51.6 (0.26%); -59.4 (0.11%)
500 Hz: -55.3 (0.17%); -38.6 (1.17%)!!!!
630 Hz: -51.7 (0.26%); -58.5 (0.12%)
800 Hz: -56.6 (0.15%); -47.6 (0.42%)
1000 Hz: -51.6 (0.26%); -53.7 (0.21%)

When I built the new constrained layer cabinet I got this:

100 Hz: 0.33%, 0.15%
125 Hz: 0.26%, 0.26%
160 Hz: 0.18%, 0.04%
200 Hz: 0.39%, 0.32%
250 Hz: 0.57%, 1.04%!!!
315 Hz: 0.28%, 0.07%
400 Hz: 0.27%, 0.14%
500 Hz: 0.21%, 0.09%
630 Hz: 0.03%, 0.47%
800 Hz: 0.13%, 0.14%
1000 Hz: 0.13%, 0.39%


So also the acoustic output from the speaker showed distortion components similar to the vibration of the cabinets shown, and shifted from 500 Hz to 250 Hz in the constrained layer cabinet.

Original thread in Swedish:
 
Back
Top Bottom