Bob Katz on Loudspeaker Isolation

(Photo by Mary Kent)

Of late, Stereophile has written a lot about vibration-isolating footers under loudspeakers. I have been listening to the Magico M-Pods (under the Magico M2 loudspeaker, which won our Loudspeaker of the Year Award for 2020, and which I'll review, in follow-up fashion, in an upcoming issue of this magazine). Following an impressive demo at a 2019 audio show, I also tried IsoAcoustic footers under the Revel Ultima Salon2 loudspeakers. Michael Fremer tried them under his Wilson Alexxes. We both reported positive results.

The idea of isolating loudspeaker vibrations from floors is controversial. Many (perhaps most) designers believe that dynamic loudspeakers in particular—those with significant moving mass in their cones—should be rigidly connected to the floor as is typically done with spikes. A rigid connection of the speaker to the floor reduces the Newton-3 reactive motion (footnote 1) of the cabinet in response to the motion of the cones, heavy woofers in particular. Cabinet motion could be expected to smear the loudspeaker's sound.

Against that is another idea: With spikes, vibrations will be transferred from the loudspeaker to the floor, and the transfer of vibrations to a suspended wood floor will be quite different than, say, to a concrete floor, or wood on concrete, and so on. This transfer of energy could affect the sound of a loudspeaker both directly—by altering its output in ways that depend on the floor type—and indirectly, via the subsequent emission of low-level sound from energy transferred to the floor and attached walls. Loudspeaker performance, then, will vary from system to system, not just due to room acoustics—that's well-known—but also due to complexities of the loudspeaker's interactions with the room's surfaces.

In a recent conference paper delivered online to the Audio Engineering Society, mastering engineer and Stereophile contributor Bob Katz investigated in detail the coupling of a loudspeaker to a room with and without isolation, and the effect of vibration isolation on a loudspeaker's measured response.

The room in question is Katz's Studio B, which is modestly sized, with a volume of 34.3m3. The floor is well-damped and suspended. Some walls are plaster-and-lath while others are made from gypsum board. The loudspeakers are the Kii Audio Three. The isolators used are EVPs from; these are essentially low-pass filters with a claimed transition frequency of about 6Hz; above that frequency they are estimated to block about 90% of sound.

When isolation was not desired, a set of blocks were substituted that were identical in form but with absorbing material replaced by kiln-dried white oak, a very hard wood.

Three measurement microphones detected changes in the frequency response (with and without isolators): at the listening position, near the front wall on the right side, and near the back wall on the right side. An accelerometer was used to measure vibrations in various surfaces, on the base of the loudspeakers and at locations around the room.

First the left loudspeaker: At lower frequencies, as expected, the isolated loudspeaker moved more than the loudspeaker mounted on the wood blocks. (Decoupling it from the floor means freeing it up to move.) Measured at the loudspeaker's base, the difference was most pronounced around 200Hz, where vibrations in the vibration-isolated loudspeaker base were some 20dB higher than when the speaker was mounted on wood. A surprise: At 1kHz, vibration was about 8dB higher with the wood base than with the isolating base.

Throughout most of the audible spectrum, the floor moved more with the speaker on wood blocks than with it on the isolating device; the difference ranged from 5–8dB, rising to 12dB at 1kHz. However, at lower frequencies—below about 60Hz—floor vibrations increased when the isolators were used—another surprise, at least to me.

Results were qualitatively similar for the right loudspeaker, but in detail they were quite different, which reinforces the assumption that the coupling of a loudspeaker to a floor is a complicated business, depending on local acoustics and details of the floor's construction.

Wall vibration was also complex, the two loudspeakers yielding different results at different frequencies, again demonstrating (or seeming to) that coupling depends on structural details.

Acoustical measurements were taken using Room EQ Wizard, at the microphone positions mentioned previously. Measured at the listening position, there were differences with and without the isolators, mainly in the midrange, but they were modest and inconsistent between the L and R loudspeakers.

Katz believes that the front microphones—the ones closest to a loudspeaker and also to a corner—tell a more interesting story. Starting at about 600Hz and extending through the higher part of the audio band, the SPLs of the loudspeakers with the isolators were lower by an average of as much as about 1dB. (These measurements were reported with one-octave smoothing, to better show trends.) This behavior was common to both speakers. Over such a wide frequency range, the difference is likely to be audible. Between 300Hz and 600Hz, there's a smaller difference in the other direction, the isolators slightly increasing loudspeaker output in that range.

Response at the rear microphone was once again complex, the two loudspeakers giving divergent results. Katz made several other acoustical measurements, including total harmonic distortion; he found very similar results with the isolators and the wood blocks, except between about 45Hz and 100Hz, where distortion with the isolators was reduced by as much as 6dB. This suggests (to me) the existence of a floor resonance that is suppressed with the use of isolators; other interpretations are possible. Energy Time Curve analysis revealed many short, negative-going spikes in the wood-block measurements that are significantly reduced with isolators (although a smaller number of short spikes are increased in amplitude with the isolators). Measurements of frequency-dependent reverberation time (T30) were complex.

To me, the most interesting results were the waterfall plots (figs.22 & 23 in Katz's AES paper), which show a dramatic reduction in delayed energy for the left loudspeaker, with the isolators, compared with the wood blocks, from about 40Hz to above 200Hz, with one strip of delayed energy remaining—even reinforced—between about 95Hz and 110Hz. For the right loudspeaker, the results are ambiguous (figs.24 & 25). Note that the left loudspeaker is proximate to the gypsum-board wall, which could be expected to vibrate more, while the right loudspeaker's adjacent wall is plaster.

Katz also reports, without providing details, that "some critical listeners have observed" a "subtle increase in clarity . . . when a loudspeaker is isolated from its supporting surface." Katz concludes that "In the main, isolators placed under floor-standing loudspeakers over a wood floor produce a measurable and often strong improvement in performance." The exceptions, he writes, "seem to be related to the unique and non-symmetrical shell construction in this room." In his video presentation, Katz emphasized this effect on subjective sound quality, concluding that "isolation placed under loudspeakers has proved to be an effective tool, resulting in sound quality that is cleaner, tighter, quieter, and deeper." That's consistent with Michael Fremer's experiences (and to a lesser extent my own), but it is merely mentioned, not supported by evidence presented in the paper or the presentation.

I'm inclined to spin the results a little differently. The large differences observed from one loudspeaker to the other show, I'm thinking, how important the fine details of room construction are in affecting the interface between the loudspeaker and the room. In standard construction, such details vary greatly from structure to structure and spot to spot. How closely spaced are the floor joists? How thick is the plywood under a wood floor? How thick is the flooring itself? If the floor is nailed, how closely spaced are the nails? If it is glued, how much and what kind of glue was used? How uniform is all of this? The results also appear to show that coupling via the air is important, which is hardly surprising. I also find myself wondering how the measurements would differ with other commercially available isolators (footnote 2).

Footnote 1: For every action, there's an equal and opposite reaction.

Footnote 2: For an examination of how the interface between a bookshelf loudspeaker and the stand on which it sits affects the vibrational behavior of its cabinet, see my article here.—John Atkinson

jimtavegia's picture

As if we didn't have enough to worry about. Knowing what Bob has invested in his mastering suite, this had to be eye opening even to him.

I hope we get to read more from the Adventures of Bob Katz.

tonykaz's picture

You lads are building loudspeakers where the listener is inside the 2,000 cu.ft. enclosure.

Our little hobby has migrated to 1,000 Watt Amplification, 20db of dynamic range, deep bass capability, significantly better music source sound quality and staggering prices.

Each listening room is becoming a laboratory of experimental scientists.

I admire your devotion and determination for this listening environment refinement quest.

Headphones are my short-cut to the needed solution.

Tony in Venice

ps. I wonder if a person could build a gigantic clone of one of the better Loudspeaker transducer systems, climb inside and listen away? Maybe that's what Mr.Fremmer has created with his Basement Hermitage.

ps.2) Mr.Katz, you are hot stuff in that Red outfit. I've been hearing nice things about y'all

jimtavegia's picture

I might think that in 2020 and beyond we would not see so many 4 star or less ratings of "sound" for music releases. I cannot believe that we still need more plug-ins to make a 5 star recording, the material is another matter and why there is another rating.

Ortofan's picture

... was on the floor in Katz's studio?

How much difference is there between bare concrete or wood flooring, vinyl covering, tile, area rugs, wall-to-wall carpet with underlaid padding, etc?

If the woofer and midrange/tweeter units can be separated, should they be separated and should they be coupled to or decoupled from the floor in the same or different manner?

(Where did Katz find a suit that matches the finish of the Kii speakers?)

Gojira's picture

Vibration isolation is not an invention of the hi-fi industry.
They are mostly used in scientific applications such as electron microscopy. The results of a long series of research reveal that in our case, the loudspeakers transmit vibrations into the floor, where some of them move further and a considerable proportion work back into the loudspeakers. In addition to the diaphragm movements, there are also some from the vibrations, which leads to a reduction in accuracy and consequently to smearing of the reproduction. I've now isolated all the devices in my chain from vibrations, with a dramatic increase in sound quality of all aspects. Unfortunately, this fact has still not been sufficiently recognized by the hi-fi industry.

PeterG's picture

Good point about the importance of reducing vibration from the floor in microscopy. An import difference though is that in electron microscopy, there are virtually no vibrations coming from the instrument, and the instrument is designed to be insulated from the room. But with speakers we have huge vibrations from the instruments, and those instruments are designed to interact with the room.

Gojira's picture

Thanks for your thought. It is true that anti-vibration measures under the loudspeakers only solve part of the problem. I am convinced that a room and the elements housed in it, like our audio systems, can never be completely freed from undesired events such as vibrations. But you can at least reduce the effects so that the sound improves massively. I have dealt with this topic for a long time and have succeeded in converting sound into heat without much effort. Quite simply with the use of foam elements that are under the speakers and also, for example, under my turntable. I plan to equip the listening room with further elements of this kind. The sound quality has improved spectacularly.

mav52's picture

How is speaker cabinet motion impacted by a solid concrete floor. and what isolation devices are used as not everyone has wooden suspended floors.

CG's picture

Just recently we tried an experiment. This experiment was new to our listening experience, not new to the world.

The floors in our house are suspended like most in our area. The basic construction is an array of 2x10's that sit on the concrete building foundation. Halfway from the front to the back of the foundation there is also a steel beam running left to right, also supported by the foundation that helps support the floor above. On top of the 2x10 frame is a couple layers of plywood. On top of that is carpet. Some people have hardwood floors on top of the plywood, then put rugs or carpeting over the hardwood. But, not us.

Anyway, our loudspeakers previously were sitting on this carpet. Two really pointy cones in the front of each speaker poked through the carpet and probably the surface of the plywood. A spike did the same in the rear of each loudspeaker.

Based on a comment from the loudspeaker designer, we tried adding a granite support under each loudspeaker. This support is slightly larger in footprint than the speaker itself and consists of two standard 3 cm thick pieces glued together to become, ahh, 6 cm thick. These platforms weigh in at around 60 pounds each and were pretty inexpensive at the local countertop fabrication shop.

Two other details.

One was that we are using small metal "shoes" under each cone and spike so that the granite doesn't get damaged and so the speaker doesn't slide around on the hard granite, probably making terrible noise on the way.

The other is that although the speakers are sitting in the same X-Y positions in the room, they obviously moved 6 cm in the Z axis. So, the spike in the rear of each loudspeaker was adjusted so the tilt put the listener on the same vertical axis position as before. This was carefully measured using a laser measuring device, before and after. The left-right tilt was also checked.

Well, guess what? The designer was right. The sound tightened up all around and became clearer. Imagine that - the designer knows exactly what he is doing!

There could be lots of reasons for this. The woofers et al are obviously 6 cm higher, so they could be exciting the room differently. The speaker could be more rigid relative to the listener than before. The overall resonant structure could be changed. Or, d) all of the above. Or, e)...

I didn't bother to measure before and after, because frankly I didn't expect a lot of difference. But, it was a cheap experiment as audio stuff goes, so, why not? It's easy enough to reverse out if we wanted to. But, we don't.

The thing is, my empirical results seem to be at odds with what Mr. Katz found. In technical terms, WTF?

The only other observation I will offer is that we tried a couple different kinds of shoes under the cones and spikes. One was made from titanium, with a fairly hard thin rubber like backing backing beneath. The other, from the loudspeaker manufacturer, appears to be made from stainless steel and has a felt backing. The sizes are almost identical. Guess what: They sound different! Neither is bad, but the difference is there. I could imagine where some people would prefer one over the other, but we preferred the version from the loudspeaker manufacturer. Incidentally, the two are priced almost the same. No expectation bias based on "more expensive is better" here. Perhaps this throws some weight to the idea that the resonant structure change is the big thing, but it's hard to separate the variables accurately. So, I don't know. I'm not looking to publish any papers. Typing this comment was work enough.

The other truly tangible advantage of a platform like this is that the loudspeakers are much easier to vacuum around now.

Varejao17's picture

FWIW, Newton-1 (assuming you mean the First Law of Motion) is (more or less): objects at rest remain at rest, and objects in motion remain in motion at the same direction and speed, unless acted upon by an outside force. I think you meant Newton-3: Every action creates an equal and opposite reaction.

Jim Austin's picture

I do, honestly, have a PhD in physics. It's really embarrassing to make mistakes on stuff like that that most people learn in middle school. I'll just say that it has been a long, long time since I learned about such things, so I need to learn to be more careful. Thanks for the note, and apologies for the error.

Jim Austin, Editor

Varejao17's picture

...sorry that I came across as a pedant about it. :-)


bobkatz's picture

I'm very impressed by Jim Austin's thorough study of my AES article. My journey to a peer-reviewed paper began many months before, when I put some specially-constructed EVP isolators under my 300 pound Dynaudio Evidence M5P speakers in my mastering room, Studio A. Unlike the wood planks in the second floor Studio B, this room has a concrete slab on the ground, covered by a thick carpet with matting. There are holes in the carpet under the speakers so the isolators rest on the concrete floor. My mentor in this is Norman Varney of A/V, manufacturer of the isolators, who points out that sound travels much faster through concrete than through air and thus the ear receives a double image that softens the focus of sound when speakers are mounted solidly to the concrete. I experienced a significant quieting and tightening of sound with the isolators than with the wood blocks I had previously placed under the speakers. The area between the back of the speakers and the front wall has quieted enormously.

But when I reported these improvements in various Forums, and showed my comparative measurements with Room EQ Wizard as well--- a couple of strong objectivists called foul, saying that since my measurements with the wood blocks were taken a month before the measurements with the isolators, they must be ruled invalid.

So when I had the opportunity to try again in a much smaller room with a wood floor and ~100 pound speakers, I did. However, I believe the evidence is even stronger and more consistent in Studio A. The room is larger, with solid block walls on 3 walls stereophonically symmetrically, and the back wall is double reinforced styrofoam with two Gypsum layers. I have published comparative waterfall measurements which are far more symmetrical from left to right and show similar improvements as illustrated in the left hand speaker of Studio B. The congruence circa 300 Hz in the two measurements is a testament to my care in reproducing microphone position and loudspeaker height, even though the two measurements were taken a month apart. So, once again I have objective evidence of the improvements gained with isolators under loudspeakers. Anyone who wishes to get the REW mdat files for either Studio with and without the isolators is welcome to contact me through my website, Here's a link to a mini-article at my website with that report:


Jim Austin's picture

Bob, thank you for your kind words and your contribution to the conversation.

It's sometimes tempting to be disappointed by inconsistency in results--but inconsistency can be very interesting. Of course it would be good to have a firm understanding of what's happening, but I am most intrigued by the complexity--specifically, the difference in behavior between the two loudspeakers placed symmetrically in a (dimensionally) symmetric room. I could mention several measurements, but the waterfall plots, to me, tell the story most clearly (figs.22-25 in the AES paper). There is an obvious reduction in delayed energy in the left speaker between about 35 and 300Hz (figs 22 and 23). In the right speaker, it's difficult to discern any such improvement (figs 24 and 25).

One cannot say with any certainty what this difference is attributable to, but one likely possibility is the difference in the construction of the adjacent walls, as you point out in the article. Wouldn't it be interesting if you were able to correlate vibrations in the walls with the change in delayed energy in the left (but not right) speaker!

There are some well-known stories relating construction of performance spaces to structural factors--like the famous story of the stage at Carnegie Hall. If you don't know the story, check out the video below. The relevant discussion start after the 4-minute mark.

Jim Austin, Editor

Sandy Gross's picture

I think that this research and article are very special. I am certainly a subjective kind of guy, but one who has a tremendous respect for measurements. The whole subject of loudspeaker physical support has been a rather hotly debated subject since it began in the mid/late 70s when Ivor from Linn suggested taking a cassettes case and using it, lengthwise, to couple (using mortite) the top back of a loudspeaker to the wall behind it to brace and minimize cabinet movement. Now how many people remember that? And in the ensuing years we have had all kinds of devices (Tiptoes, spikes of all sorts, Stillpoints, Iso-Pods, Isoacoustics, and all kinds of new ones at all kinds of price points): some stiff, some soft, some with all kinds of unique and not so unique technologies. And, in terms of their use with loudspeakers (I won't go into use with other gear) we have listened, heard all sorts of things and opined and reached all sorts of conclusions. Now, with thIs article and its measurements, (although it certainly does not cover all possibilities), we can clearly see that there are dramatic measurable differences with different kinds of coupling to the floor (and different results with different kinds of floors). My Goodness!!! So we were not hearing or imagining subtle nothings. Now, we are still faced with deciding, subjectively, what we prefer. And, of course, as a loudspeaker designer, this dramatically complicates the process. Kudos to Stereophile to help to open up in a measured objective way this whole subject. No, it is not snake oil or black magic!!!

tonykaz's picture

But it sure reads like it.

Should we consider our listening room variables as tuneable? I think we should. It's definitely DIY in a world where the Super expensive gear room is the most needy.

Little suspension footers are not the quick easy fix for a problem that is still not well or properly stated.

We should be reading about how to diagnose SoundQuality issues in our listening rooms, shouldn't we.

Probably Bob Katz is the only one that ever attempted to describe tuning a room as he has done in his Florida Location.

We've all had these same Room issues for all our decades of listening but we didn't have Kilowatt Amplifiers driving low impedance, low efficiency loudspeakers with an increased possibility for feedback loops from sensitive record players and Digital players playing Audiophile grade high dynamic range recordings featuring unprecedented transparency.

Personally, I've set-up my listening rooms as semi-anechoic chambers, wall bounces loosing 9db, ambient SPL of 35db & remotely located signal sourcing.

All the above seems like Acoustical Engineering problem stating, trial & error, solution building with the final product being a well designed Music Review Studio.

Copy Bob Katz! or make a hobby out of Diy gizmo devices and hope to get a little lucky.

Headphones remain the super efficient solution and diagnostic tool to evaluate room performances.

Tony in Venice sorting thru >>>)))))'> oils