Damping/Decoupling material experiment

[Thephantompsychic]

Bill,
Firstly I note that the upper crossover frequency doesn't really matter to me. I could cross it at roughly 2k with an 18db/Octave crossover and genuinely believe the tweeter would do quite well. In this case, the mids crossing over at 2k would do the job. The sb26adc can be crossed really low, so anywhere from 1700+ should do the trick (depending on crossover slope obviously). I keep referring to thermal power ratings and that's because I like playing it safe. I don't want to blow a driver, but I am aware it varies on a LOT of things like temperature and frequency band and voltage.
As far as separate subs go, you've mentioned a few times that the subwoofer being on the speaker itself isn't ideal in most cases which I find very intriguing. Obviously it varies from room to room, but what would be the rule of thumb for placing subwoofers? (Assuming I split it to be a 2 way + sub)

Andy,
thank you for the very interesting read. I find it intriguing that the resonance (whatever the part that improved with decoupling is) is located at roughly 400hz on this experiment. Certainly this is an indication that a midrange would be better to de-couple than a subwoofer but I should probably do some more independent research to confirm this. The article I had read is at this link:
https://audioxpress.com/article/spea...ced-vibrations
If I have misrepresented something from this article, please tell me as I didn't mean to.

As for the gaskets/well nuts and etc: I will have to look into that. They seem useful.

[billfitzmaurice]

You wouldn't use only one sub, as you can't correct for room modes and boundary reflection sourced nulls with one sub. You'd place one sub using the 'sub crawl' method, either by ear or with measured results at the listening position, then place the second where it gives the best results filling in the response nulls at the listening position.

[augerpro]

Not saying you are wrong, but why is that? Here are #6-32 neoprene well nuts 0.5 inch and 1.0 inch. This weekend I'm going to use some to attach an RS100-8 full-range to a small speaker's baffle. BUT, I'm not doing any testing with NOT having the well nuts, so I won't know if it makes a difference or not.

See attached. They could probably hold a driver in just because they will compress and press against the wood, but they are really meant to expand behind the mounting surface.

[a4eaudio]

Got it! Thanks for sharing that image.

[andy19191]

I find it intriguing that the resonance (whatever the part that improved with decoupling is) is located at roughly 400hz on this experiment.

It will almost certainly be the driver bouncing/rocking on the baffle. This mode is almost always the most troublesome because it is low frequency, efficiently driven and radiates sound straight at the listener.

Note that when the driver is (successfully!) isolated from the baffle the magnet movement is a bit below -40 dB below that of the cone. Now a bit below -40 dB distortion in the midrange can be expected to be borderline audible depending on kind. Good but possibly not a full solution. Rigidly attaching the mass of a midrange enclosure and then isolating that from the woofer cabinet will push this down another 10-20 dB making this particular source of unwanted sound comfortably inaudible. (But would need to check other sources like the frame resonance shown in the example).

Certainly this is an indication that a midrange would be better to de-couple than a subwoofer but I should probably do some more independent research to confirm this.

Whether isolating a midrange driver is better depends on context. You wouldn't want to do it with a separate isolated midrange+tweeter cabinet. You may want to do it when all drivers are attached to the same enclosure.

It isn't physically reasonable to isolate a sub for reasons mentioned earlier and I am not sure how one could go about it given the softness required for whatever would take the place of the grommets used for a midrange. The resonant frequency of the driver on the "grommets" needs to be well below the passband of the driver for isolation to work (e.g. something like a sprung turntable). If this resonant frequency is within the passband it will amplify the cabinet motion rather than reduce it.

The article I had read is at this link:
https://audioxpress.com/article/spea...ced-vibrations

Thanks and it was interesting albeit not in the same way as the Andrew Jones article. The experiment involving a sounding board (?) with driven and undriven woofers seems rather distantly related to drivers exciting a cabinet. Not sure one can draw many reliable conclusions from it. (And good luck pushing the resonances in a large plywood woofer cabinet containing heavy drivers well above 1000 Hz!).

[Paul Carmody]

I think this is one of the better "thought experiments" we've seen on here in a while. Unfortunately we can argue hypotheticals until we're blue in the face. It really does call for some experimentation. I see that augerpro has done some experimentation in the past. To which I would like to ask "How do you measure the effectiveness of this isolation?" and "What exactly are you measuring for?" And then of course, "Is it audible in a blind test?"

[billfitzmaurice]

I tested the benefits of driver isolation by building identical cabs with and without, comparing their measured responses. They were not significantly different. IMO too much concern is made over the effect of cabinet vibration being transferred to the driver frame. If the cabinet isn't vibrating that concern isn't warranted, especially when one considers that said vibration would have to affect the driver cone, which is already isolated from the frame by the suspension.

[andy19191]

I think this is one of the better "thought experiments" we've seen on here in a while. Unfortunately we can argue hypotheticals until we're blue in the face. It really does call for some experimentation. I see that augerpro has done some experimentation in the past. To which I would like to ask "How do you measure the effectiveness of this isolation?" and "What exactly are you measuring for?" And then of course, "Is it audible in a blind test?"

There is little hypothetical about the measurements in the Andrew Jones document. Some of the quantities we want to know are being directly measured with a real driver in a real cabinet. Would need more quantities to get the full picture and the unwanted sound at the listening position. This is less the case with audioexpress article which doesn't use a real speaker or directly measure what we want to know hence requiring a degree of hand waving/hypotheticals.

The problem with performing a fully useful experiment is that it requires measuring the normal velocity component over the whole of the outside of a speaker cabinet. This is tedious to do by hand but feasible with a laser vibrometer (common in every acoustics lab but not DIYers garages) which can be moved from point to point fairly quickly. Likely to take most of a day for one set of measurements for one configuration. The only person I have come across that had a go at sticking and unsticking strain gauges never produced any results. It also requires a bit of software to calculate the acoustic field which can be a hurdle for some.

So people tend to perform less useful experiments involving drivers in rigs rather than drivers in speaker cabinets and measuring quantities that might be related in some approximate way to what we would like to know. They might provide some qualitative insight but this can also be obtained from 3D simulations along with a great deal more. Simulations are likely to be a lot more useful at understanding what is going on with various resonances and assessing options to modify them. Unlike laser vibrometers, software to perform simulations is freely available to DIYers. Not in a full commercial form and one or two desirable features are missing but it is around.

[Paul Carmody]

What is the Andrew Jones document?

[andy19191]

It is linked in post #12

[mcargill]

ThePhantomPsychic, You threw us a bit when you called it a subwoofer. It seems you are doing a very traditional 3-way design. Typically, it would follow the traditional "decade" design, dividing the sound spectrum with crossover points at 200Hz and 2000Hz. The woofer's band would be 20Hz-200Hz, which is pretty much what you say you are shooting for. Some good and famous speakers have been designed that way. There are acoustic advantages to go to a 4-way, or even a 5-way, so that you have a true subwoofer, but choosing a traditional 3-way is not obviously a mistake.

To address your specific question: If someone did want to decouple a driver from the cabinet, would Green Glue be a good idea? Absolutely not. Because Green Glue stays sticky forever, it should not be used anywhere exposed, where it might be touched. Under the flange of a driver is exposed around the edge, and under screw heads and washers. It would ooze out as the screws are tightened. It would be a mess. And if you ever wanted to take the driver out, it would be ten times a mess. Green Glue works as a layer hidden, confined, enclosed, entombed, secreted between two other layers of material. If you really want a glue, ordinary caulk would work, because it does not remain sticky forever.

Since it is the screws that are holding the driver on and holding it up, the sealing material does not have to be sticky. It could be foam, like weatherstripping.

It is easier to use a decoupled mounting in a design where there is less concern with sealing the chamber behind the driver air tight, such as in an open-baffle design, where there is no rear chamber at all. Then the issue of how well the foam or rubber grommets seal matters less, and decoupling from the expansive baffle can be made a priority.

But I agree with everyone else that decoupling is for midranges, not for woofers. The reactive force of the cone motion needs the driver's flange to be held solidly to a massive cabinet, not held in a squishy, bouncy gasket that allows the flange to move.

[mcargill]

My observations about Andrew Jones' experiment in "Loudspeaker Driver Decoupling":

The most telling result is the first one he reports--It makes no measurable difference to the motion of the cone whether the driver is mounted in a cabinet or hanging from bungee cords. This indicates that all our talk about the importance of mounting to a massive enclosure might be mistaken. It could turn out that even if a subwoofer exhibits reactive forces strong enough to walk the box across the floor, that little bit of motion might not have any audible or measurable effect on the sound we hear at those wavelengths. As long, that is, as the only sound we hear comes from the cone, and not from the walls of the cabinet or the box rattling against the floor.

The biggest difference Jones measured was in the motion of the magnet. Even then, it was in the time domain that the difference was most evident. In the frequency domain the magnet of the coupled driver was quieter than the decoupled driver below 250Hz. He measured displacement, not SPL produced. The SPL of that magnet motion would depend on its surface area. And if the magnet is enclosed in a box, it is not likely to make any audible sound outside the box. The only situation where that measurement might matter is if someone mounts the driver with the magnet to the outside of the box, as in a push-pull or W-baffle woofer, and then there are other factors affecting the sound, so the magnet motion Jones measured may be the least of one's troubles. If the sound that is being emanated by the backside of the driver inside the cabinet matters, then there are other things about that sound to worry about besides the magnet's vibrations.

Jones reasons that a moving frame would have to affect the cone's motion because the frame is the "ground reference" for the cone motion. But the front edge of the cone is decoupled from the frame by the surround and the rear of the cone is decoupled by the spider. The place where the magnet does "couple" to the cone is in the magnetic gap, where the electromotive force is applied. Measuring the movement of the magnet is measuring the movement of the magnetic field around the voice coil. The question then becomes whether we can reasonably expect the motion of the cone to be affected by the motion of the magnetic field, when the motion is on the scale Jones measures.

The driver Jones used was a 5-inch midrange. The results might be different for a subwoofer. The most relevant result of Jones' test was the movement of the cabinet walls, which would radiate sound into the room. In those tests, the only significant difference between coupled and decoupled drivers came in the band 100Hz-500Hz, which is upper bass and lower midrange, not subwoofer.

The cabinet he built used 1/2-inch thick material. The chances are good that a thicker cabinet, or a little attention to damping the walls, would have much more significant effect than decoupling the driver from the baffle.

An "initial" listening test tells us nothing if it is not blind and controlled (A-B-X) and several listeners, and specifies how much "preference" and in what ways.

[Steve Lee]

That ^ was a really nice synopsis, mcargill - Thanks.

[andy19191]

My observations about Andrew Jones' experiment in "Loudspeaker Driver Decoupling":

Interesting analysis which I started to comment on but then stopped. Correcting the interpretation of measurements without the relevant physics being recognised/understood is unlikely to be accepted. It is this that seems more important to get across but it is far from straightforward how to achieve it. Thoughts?

Nonetheless your post did prompt me to look more closely at the measurements and the results at higher frequencies are not as expected. There is a comment about the pressed frame resonating (which is expected with pressed frames and why better quality frames are stiffer) which would then affect the motion of the cone via an increased motion of the magnet. However, isolation is expected to be more effective at higher frequencies than low with little force being exerted on the baffle. The measurement of the motion at a point on the side and back panels shows the driver is exerting essentially the same force on the baffle with and without the grommets. I tried to find the material of the grommet (some do stiffen at higher frequencies) but without success although without more details of the mounting there would still be relevant information missing. My suspicion is that the isolation isn't as well implemented as it might be and further work may be useful. I have added it to my examples list for simulation.