Well, this ought to shake things up a bit:

Testing Loudspeaker Isolation Products

Enjoy!

--Ethan

Some points why your test is flawed.

A. The old table is already lossy, use a pro studio stand on a compromised floor.
B. You are measuring in the near field, the direct speaker response is not going to change much due to decoupling, so you are minimizing possible differences.
C. You could use an accellerometer on the speaker, plus one on the floor.
D. You are forgetting the differences in the IRs seems rather pronounced. Imo a lot more is happening in the time domain than a waterfall or FR plot will show.

I seem to remember a 20 year old test in hfn/rr, where the measurements showed significant differences with eg. spikes under the speakers. I do not have those anymore...

Thanks for your effort! This will be be an interesting read!

Cheers

Hi Ethan. Thanks for taking the time to test and publish. I'm not complaining, but do not know whether what you tested is relevant to questions I have. I don't know if my concerns are valid, but the theory of my concerns may be somewhat different. So here are some comments which may or may not be BS.

The resonant frequency of the stand may not matter to my curiosity questions-- For instance, if we whip out an old-fashioned set of piano tuners chromatic reference forks. Sit down at a nice sturdy hardwood dinner table, not especially resonant. Or the hardwood lid of a grand piano will do. Strike any tuning fork and hold it in air. None of the forks are very loud held up in the air. Strike a fork and set the stem on the hardwood surface, and it gets louder, easier to hear.

Any fork in the set gets louder when you set the stem on the thick hardwood. Apparently this amplification doesn't rely on the resonant frequency of the surface. Seems hardly likely that the surface would be resonant at all 12 chromatic intervals within an octave. I think touching the fork to the surface just multiplies the surface area which sound can radiate from. The mechanical vibration grabs more air so it sounds louder.

****

OK, it may be a dumb way to do it, but I have a fairly rugged, 3" thick shelf which puts the monitors at ear level, mounted above and to the rear of my keyboard stand. The shelf vertical supports are screwed and glued to the fairly heavy keyboard box. Both the keyboard box and the 2X6 shelf supports only touch a concrete floor. I am not worried that the speakers might vibrate thru the stand and shake the concrete floor. The shelf doesn't touch any walls.

However, this thick shelf stretches all the way from the left speaker to the right speaker, near 8 feet wide. A video monitor sits in the center of this shelf at eye level, between the two speakers. I have not tested, but am purt sure if I was to set a tuning fork stem on the shelf, that the shelf would be "resonant" enough to make the tuning fork louder, though the shelf isn't what I would consider especially resonant.

When playing keyboards or computer mixing, each speaker is about 39 inches from my head, so the center of the shelf would be perhaps 27 inches from my head.

Speed of sound in wood along the fiber looks to be more than 10 X the speed of sound in air, so if we just do a rough calculation neglecting the speed of sound in the wood-- If sound travels from the speaker thru the wood to the center of the shelf, the sound radiated from the center of the shelf would hit my head about 0.846 ms earlier than the sound traveling thru the air from the speakers.

So that multipath, if loud enough, ought to have full, half, and quarter wave relationships at 1182, 591, and 295 Hz.

But it ain't so simple, because sound would be radiating off the entire length of the shelf, causing a smooth mix of pre-echoes spanning from 0.846 ms all the way down to 0 ms.

Perhaps as you say, the time-smear of the shelf audio emanations would not be loud enough to affect the frequency response at the listening position. However, if the shelf audio emanations ARE loud enough to audibly cancel and reinforce, it could be a problem.

My frequency response is pretty good at listening position. Maybe better isolation would straighten out some of the midrange squiggles or maybe not. Maybe good enough isolation would smooth out the midrange squiggles from about +/- 2 dB, down to +/- 1.8 dB or whatever. Or maybe it would make better improvement. Or maybe no detectable improvement at all, similar to your test results. Dunno.

Am just saying, I am not complaining about your test method. I just don't know if it is relevant to my questions.

Thanks

‘Hi Fi News’ [July and August 2002 issues of Hi Fi News], Keith Howard used an accelerometer to measure the levels of vibration in various objects. When he tried measuring the vibrations produced in a loudspeaker stand by playing the speaker he found a result that surprised him. With the speaker unit sitting on cones the level of vibration of the stand was over one hundred times greater than if the cones were replaced with small rubbery feet. This indicates that cones are of doubtful use if the intention is to stop vibrations passing from the speaker to the stand, or to any other solid objects against which the speaker may sit.

Now surely, if a speaker sitting on cones induces vibrations over hundred times greater than decoupled using "small rubbery feet", this must make a subjective difference ? Let alone if the floor is not solid/well damped concrete.

It is a pity I threw away all those old HFN magazines. I remember those measurements, they were quite spectacular. I cannot find them online. I have been using my speakers decoupled ever since...

EDIT: not in the least, as JCJR remarks, because those vibrations travel A LOT faster through the room structure into your legs and buttocks. Of course your nearfield measurement mic won't pick this up, butt I can assure you, your behind will ...

LOL, thanks but these are easy enough to deflect.

The main claim made by these vendors is that speaker vibration travels to the surface the speaker rests on, thereby causing that surface to vibrate in sympathy and "cloud" the sound. My flimsy table is a perfect test because its surface is wobbly enough to vibrate in sympathy, and is ten times larger than the speaker driver's cone.

Look, we can test literally a hundred different scenarios: A rigid speaker stand on a rigid floor, a rigid stand on a flexible floor, a flexible stand on a rigid floor, a flexible stand on a flexible floor, loudspeakers on a meter bridge, loudspeakers on a metal rack cabinet, loudspeakers on an MDF rack cabinet, and so forth. If you believe a different set of surfaces will yield more difference when using isolation, I look forward to seeing your measurements. I mean this! If I missed something I want to know. But arguing theoreticals doesn't convince me of anything.

Most people using these products listen in the near field, so that's where the measuring microphone should be too.

Sure, but that gives no useful information. That's like my example of AC power products that show a few less millivolts on the power line, but ignore what happens at the output of the connected audio equipment. If an accelerometer shows less vibration, but actual audio measurements show little or no difference in the air, then who cares?

All the differences I see in the Impulse Responses are 10+ dB down. So much for "obvious" improvements. As with the Response and Waterfalls, there's much more difference between No Isolation three inches apart than between No Isolation and any of the isolation devices. So I'm confident that the differences are due entirely, or at least mostly, to position errors. Also, a waterfall shows the time domain and especially resonance.

Again, I urge you and anyone else to do their own tests and make the results public. More tests like this will only help us all get to the truth.

--Ethan

Sure, though the tines on a tuning fork move quite a lot compared to the cabinet of a decent near field monitor. I have a nice "pro quality" A-440 tuning fork here, which I just whacked, and the tines move at least 1/8 inch. I bet the tines on a larger C-262 tuning fork would move even farther. This is why I measured these devices! To see how much the sound in the air actually changes, as opposed to a bunch of vibration theories we read about in glossy advertisements.

Yes, exactly.

Yes, and the key words here are "if loud enough."

--Ethan

Ethan, I actually agree with your response (you asked for possible flaws, so...)

Except for
"All the differences I see in the Impulse Responses are 10+ dB down. So much for "obvious" improvements."

If I find the time next week, I will try some tests. But the last few years, I am on a 16 cm thick concrete floor, on sylomer strips. It is quite non-resonant.

And you did not respond to early arrival through the structure into the mixers' behind and feet. This alone can make a big subjective difference in sound quality.

I do understand the problem with different arrival times due to sound traveling faster through a structure than through the air. But in order to create substantial peaks and nulls (comb filtering) the sound sources need to have a similar volume level. If sound coming from a table vibrating in sympathy is 20+ dB softer, which I think is realistic, the response won't be affected very much. (At -20 dB the peak and null are both less than 1 dB.)

--Ethan

Yes, but FR is quite unimportant if you hear or feel something a fraction of a second before the direct sound, which will be very coloured anyway.

I wonder if soffit mounted speakers really measure different in the FR when properly uncoupled as opposed to coupled to the wall ?

It might have been interesting to run the tests on a nice sturdy, massy speaker stand too. That table doesn't look like it would couple very effectively with the floor.

I think some kind of vibration measurements would indeed be useful (although tricky with all that air moving things!). I always thought the idea of speaker decoupling was more about protecting the environment from the speaker rather than the other way round, whether its reducing transmission to other rooms or just stopping your console/desk buzzing and rattling. We want vibes in the studio, not vibrations!

I don't want to say this testing is flawed, but it is misleading, and I don't want to go into the details either. SPL (and the resulting displacement) matters quite a bit.

Thanks Ethan. Yes the tines move a lot and have nearly inaudible output in free air. I ignorantly suppose that this is because the tines are so small in diameter compared to the resonant frequency, that the air just "diffracts around" the edges of the vibrating tines, rather than propagating much displacement to neighboring air particles.

Also, maybe the two tines tend to be out of phase or in quadrature in order to make such an effective mechanical oscillator, dunno. Even the simplest things often turn out complex under the surface.

However the tuning fork stem hardly vibrates at all compared to the tine motion, but that tiny stem vibration can couple to a plank of hardwood quite nicely. Apparently the tines store the oscillatory energy, which slowly leaks out of the stem into an "amplifying surface".

I make no pretense to understand the physics. It was just an analogy that PERHAPS the "tiny" motion of a speaker cabinet could couple quite nicely with a piece of wood, making bigger spl in a wood stand than one might expect. I don't recall trying a tuning fork against concrete or dense stone. Maybe that would amplify the tuning fork, though it seems doubtful that it would amplify very much. Am interested in the wood scenario because wood is my current speaker platform.

Am currently using generic fairly dense closed-cell foam as speaker isolator. Which was added as an afterthought. The closed cell foam is most likely quite ineffective, though perhaps marginally better than the speakers directly coupled to the shelf.

As previously noted, in my situation I would be more concerned with coupling problems in the mids rather than the bass, and completely unconcerned with treble. The tweeter kicks in a 1.8 kHz, and the tweeter just ain't gonna couple into the speaker cabinet to any practical amount, so far as I could imagine.

It does, just as much as any driver, it's just that mass matters here.. or rather mass ratio.

Thanks Ethan

You are right. Practical testing trumps theoretical guessing, regarding "how loud does it have to be to matter". I'll probably do at least some cursory A/B measuements after building the new speaker cabs. Too lazy to tear down the current speakers to test different damping versus none at all.

I am not a dis-interested observer. I would PREFER a finding that "speaker isolation makes no significant improvement". There are enough real problems that it is not worth spending much time fixing what ain't broke.

Surprisingly small time-smeared amplitudes MIGHT make a measurable difference. The speakers + low level pre-echoes from the shelf, would make a real-world Finite Impulse Response filter.

The funny thang about FIR filters, is that incredibly tiny coefficients on the borders of the filter kernel can make easily measurable results in the filter transfer function.

For instance if we make a sinc FIR filter for interpolation, sample rate conversion. Even with a filter kernel a mere 16 points length, the edge coefficients have got very tiny. But it is easy to measure, and even audible to the ear, that the 16 point kernel performs better than an 8 point kernel.

People trying to do high-quality interpolation for hifi audio sample rate conversion, might use kernels hundreds of samples long, where most of the coefficients in that kernel are stupid-small numbers. But it makes an easily measurable difference. The 300 point kernel performs so much better than the 16 point kernel. Though even the edge coefficients of a 16 point kernel have got rather dang small.

Put another way, the coefficients near the middle are typically the largest in this kind of filter, but ALL the coefficients usually add up to 1.0. So given a 300 point kernel, with fairly large coffs in the middle, its easy to see how tiny the outside coffs get.

In the audio programming trade, there is the temptation to try to make it as good as can be measured, even if maybe the typical ear might not be able to hear it. There is always the open question of what some customers might be able to hear. If you can measure it, some guy might be able to hear it. So overkill seems better than underkill, so long as the overkill doesn't get too expensive to run on a computer or DSP.

Just saying, a pre-echo spanning from 0 to 0.85 ms, if expressed in a computer FIR filter at 44.1 kHz, would be a kernel about 37 samples long. In a computer FIR filter of that length, typically at least half of the coefficients will be rather tiny, and changing the tiny values would make easily measurable differences in the frequency response and time response.

I hear you jcjr and I don't think it's wise to ingore something 10dB down on the IR as Ethan mentioned above.

I believe this would be due to loading in half space. (Unless you're saying the volume changes even if you hold it up, say, a mm above the surface.)

Thanks Mike. That is an interesting idea, hadn't thought about it. Maybe sometime I'll drag out the long-neglected piano tuning toolbox and play with the forks.

I don't recall ever holding a tuning fork very close, parallel to a reflecting surface. Presumably it would get a little louder due to half space reflections off the surface. Am not a physicist. Proudly switched my major away from Physics at the beginning of my junior year. The math was hurting my brain.

Usually a tuner will strike a tuning fork then touch the stem to some convenient wood part of the piano. With the vibrating tines vertical an inch or two above the surface. Depending on however long the tuning fork stem happens to be.

I don't recall it getting loud enough to clearly hear the fork unless the stem actually touches the wood, but could be mistaken. The stem hardly vibrates. Otherwise, the mere act of holding the stem in the fingers would quickly damp the oscillation, negating the purpose of the tuning fork.

What's your impression of the different isolators... using your ears?

I didn't know either, so I found this video on YouTube. It shows the tines moving in opposition, closer to each other, then farther apart, rather than side to side in sync:

https://www.youtube.com/watch?v=VCER...outu.be&t=1m7s

I never even played music while testing these devices. My impression based on my ears is irrelevant for a few reasons:

1) I'm 67 years old so my ears aren't as sensitive as young ears, though I have very good learned hearing acuity and know well what to listen for.

2) No matter what I heard, or thought I heard, there's no reason anyone here should believe me. I don't believe most of the people who claim to hear stuff, so why should they believe me? Just look at all the people who swear they hear an improvement after buying a replacement AC power cord!

--Ethan

BTW Ethan I must thank you for including the pic of the cable elevators in your report, not something Ive encountered before. It made my day

Anyone know what the SI unit for quietude is ... ?

With all these kind of things I have always felt that if you can afford it and it makes you happy then go and buy it. I don't buy many

Thanks Ethan. Neat video. That fella at the end of the video, whacking the crap out of his tuning fork-- Hopefully he doesn't need to keep it in calibration for tuning purposes!

Found a couple of pages with some details--

Vibrational Modes of a Tuning Fork

https://en.wikipedia.org/wiki/Tuning_fork

It says that the side-to-side motion of the tines is converted into a (small) longitudinal motion in the stem. Interesting.

Was earlier guessing either out of phase or quadrature, guessing maybe it would work similar to electronic oscillators.

Did not see an explanation why, when one tine is struck, how does the other tine know that it is supposed to start moving in the opposite direction? Presumably transfer of energy between the tines. There are videos about coupled harmonic oscillators, coupled pendulums and such. Dunno if it would be that principle, or some other principle at work.

Thanks Ethan!

Andre

Thanks for this Ethan!

I tested MoPads, Recoils Stabilizers, Soborthane and Aperta, I own all of them

I always put the Tweeter at the same height and position.

The best to me are the Apertas, and then Soborthane... the rest far from them.

My experience with these things have been exactly as Ethan's test has concluded. A good damped stand in the right spot with minor / cheap isolation is fine.

Having said that, where i think these are really needed are those large and resonant desks people are using. You know, the ones with built-in shelves that have a place for speakers on top of them. Or any desk for that matter, or where the amount of surface area for vibration transmission is large. I'd like to see a test on those....

My test includes that. The table I used is about four times larger than the speaker, yet there was no change in ringing. Even if the table was a flimsy board ten feet long anchored at the ends only, I doubt the board would vibrate enough to make sound. A big part of my test showed that competent speaker cabinets don't vibrate or flex enough to make sound in the first place. But, as always, I welcome others to do tests like this.

--Ethan

Ethan,

If we keep with the tuning fork analogy, I would also guess that table area makes a big difference. The movement induced by the stem is tiny, but you can hear it if a large surface moves with this tiny amount. With a 3x5 desk, the area is about 10 times larger than your test table. Does this translate to 10 dB more? Anyway, I can't do that test so I'm not adding much here to the discussion.

But thinking about the fork made me realize something else. To get some decent transfer, you need to apply decent pressure from the stem to the surface. And the position on the surface also matters, sometimes a lot. You demonstrated quite well in the video that you need to hit the table in the center to get 298 and on the side to get 475Hz. The speaker might not be exciting these spots.

If I could redo the experiment, I would try to use spikes to ensure a very firm contact between the table and the speaker, and I would also try to position the spikes in order to excite the table modes. Hopefully, I would measure a little something. This would comfort me that the phenomenon is there, but that you have to try real hard, or be really unlucky to experience it in real life.

My personal opinion was that some felt or thin rubber might be required to eliminate potential rattling between the speaker and the surface. Your experiment seems to show that it's not even required. (or maybe not, have you looked for harmonics ?)