I need to eliminate 43 Hz noise

[vinc]

Hi,

I have what is probably a very unusual acoustics problem.

<b>The problem</b>

I'm a graduate student in physics at a large US university. I trying to do a certain exquisitely vibration-sensitive experiment. Unfortunately, what I am seeing is an effect consistent with my experiment being shaken, ever so slightly, at 43 Hz. (The intensity varies over time. The peak is only a couple Hz wide.) This corresponds exactly to the resonance frequency of one dimension of my room. So something is exciting that mode and messing up my data. I'm not sure what the noise source is, exactly. It's very soft--I can't make it out, anyway--but my experiment is more sensitive than my ears are. Anyway, it's likely that even if I could figure out the noise source, I couldn't do anything about it--there are pumps down the hall, trucks driving by outside, fans for circulation in the next room and so on and I can't shut down the entire area. The room is already pretty well isolated and I doubt I can improve it much. I've ruled out other sources of noise--groundbourne vibrations, electronics, etc--but going into that will make this post much too long. Also, playing a 43 Hz tone on a speaker outside the room will increase the problem. So I'm almost certain it's acoustic.

<b>The room and the instrument</b>

The room is 157" by 168" by 133" with 157" being the problem. The walls are made of sheet rock, then a small gap (not sure how much--though I could find out of people really need to know), then thick concrete (several inches). There's some foam at a few points on the wall to absorb high-frequency stuff.

The thing that's vibrating can be thought of as a small wire, with one end firmly bolted down, and the other hanging in space. This wire bolted down inside a vacuum chamber (so sound is being transmitted to the chamber, and *then* to the wire, not directly) which is bolted to a two-ton granite table resting on airsprings. There's a bunch of other stuff in the chamber, some of which might have a 43 Hz resonance, but I've looked and I can't figure it out.

<b>Potential solutions</b>

So what I want to do is soak up the 43 Hz noise. I bought two MondoTraps from RealTraps, and they had very little effect. I'm thinking I need more absorbing area, and also something which is more optimized for the relevant frequencies than the MondoTraps are. So now I'm looking at resonant traps.

Unfortunately, I have absolutely no practical experience with acoustics--my knowledge of the subject comes from a couple days of reading stuff on the Internet--so I was hoping for some commentary on what I can do. I'm considering the following three courses of action:

1. Buying a whole slew of traps. Perhaps the RPG <a href="http://www.rpginc.com/products/modex/in ... ">Modex</a>. However, these are $250 apiece for a four square foot panel. Covering one whole wall would be about $10,000 and I'd really prefer to find a cheaper solution. How much good would this do? How much coverage do I really need in order to see improvement? (I need at least 20 dB improvement at 43 Hz, preferably more...) What other options are there?

2. Building my own trap, probably a panel resonator. I've got access to a full wood/machine shop so I can build almost anything. However, I'm worried that designing a resonator at such low frequencies is going to be more complex than simply taking the panel resonator formula and plugging in the correct parameters to get a 40 Hz resonant frequency. In his <a href="http://www.ethanwiner.com/basstrap.html">article</a>, Ethan Winer tells us how to make 80-160 Hz bass traps. Can I naively (say) double his board thickness and quadruple his air gap and get a 30-50 Hz trap? Or is that pushing the formula beyond the limits of its applicability? It seems like too thick a board would lead to board stiffness becoming an important issue. Would some other form of resonator be better at such low frequencies? Any advice on this, or should I just try it and see? I'm drawn to this idea because I could get a lot more wall coverage than those Modex traps would give.

3. Buying a whole LOT of fiberglass and stacking it in the back of the room. I'm reluctant to do this, because it would take up so much space... one quarter wavelength would be half the room, so I probably can't do that, but I might be able to get 3/8 of a wavelength of thickness and that would lead to some cancellation. I guess if I did this I'd just rent a truck and go over to Home Depot and buy a whole lot of fiberglass insulation, right? Is there anything else I should be aware of?

In general, I don't care one bit about aesthetics, or how the room "sounds." The deader the room is, the better. As for budget, I'm willing to pay the $10,000 for the commercial panels but only as a last resort.

Thanks in advance for your help.

[Ethan Winer]

Vin,

> I bought two MondoTraps from RealTraps, and they had very little effect. I'm thinking I need more absorbing area <

Yes, I remember that, and I remember suggesting that you'd need a lot more than just two traps to make a meaningful improvement at 43 Hz.

> How much coverage do I really need in order to see improvement? <

You really need to cover as many corners as possible, especially those in the wall-wall and wall-ceiling junctions for the dimension that's resonating at 43 Hz. I imagine 10 MondoTraps would be a good start.

> Can I naively (say) double his board thickness and quadruple his air gap and get a 30-50 Hz trap? <

Yes, you can do that, but you'll still need a large number of them.

> It seems like too thick a board would lead to board stiffness becoming an important issue. <

Yes, if you use, say, 1/2 inch plywood you'll want to make the traps 4 by 8 feet rather than 2 by 8 feet as shown in my plans. But you'll still need to cover a large amount of the room's surface to make a real improvement.

> I'm willing to pay the $10,000 for the commercial panels but only as a last resort. <

Understood, though you probably wouldn't have to pay that much to get the job done with MondoTraps. If you have a truck and would like to drive down and borrow a large number of MondoTraps to see how many are needed to truly do the job, give me a call. I'm pretty sure you'd be able to fix this problem using MondoTraps for less than $10,000.

--Ethan

[vinc]

Hi Ethan,

I really appreciate the offer, and I might take you up on it later this week. I would like to spend a bit of time exploring cheaper solutions first, though.

[knightfly]

What's the physical size of the chamber itself, and what's the wall thickness/material, and how hard a vacuum are you reaching? Steve

[vinc]

The chamber is oddly shaped so it's hard to give a "size." I figured out the cross-section a while ago and it's about two square meters. It's made of stainless steel, and the thickness is 1-2 mm depending on position.

The vacuum level is 10^-7 torr.

[vinc]

It turns out that Poland Spring 5 gallon water bottles have a resonance frequency right on 43-44 Hz. Is there any reason I shouldn't just buy a couple dozen of these and use them as Helmholtz resonators?

[knightfly]

That might work, although you might also end up with MORE 43 hZ instead of less; it would depend on damping of the container somewhat. Also, placing a plug in the neck of 000 steel wool might broaden the response if you need to shift the Fo slightly.

How are you pulling 10^-7 on that large a vessel with only 2mm wall thickness, without the whole thing imploding? And, are you sure it's not your thin stainless that's the resonance culprit? Steve

[vinc]

It's not a sphere. It can be thought of as four cylinders of different dimensions, meeting up at different points. There's nothing with a radius of curvature greater than eight inches.

The chamber dimensions I gave included pumps and the dewar too, which are a significant fraction. If you tell me why you want to know I can be more specific.

I'm pretty sure it's not a mechanical resonance of the chamber. I've looked at the likely modes and they're nowhere near 43 Hz.

[knightfly]

OK, that makes more sense; I deal with vacuum chambers where I work, but they're for vacuum arc melting and a typical one is 30 feet long, 3 to 4 feet in diameter, and made of 1" thick stainless with ribs added for strength - 10^-7 is quite a bit harder vacuum than we pull, more like 10^-5 to 10^-6.

When you asked why I wanted to know, it just dawned on me that my confusion about you getting sound thru a vacuum was due to my forgetting you have one end of your wire hard attached to the chamber

So far, your idea of the water bottles is probably the easiest that could work; otherwise, maybe build a slat resonator tuned to that frequency. One thing that concerns me about that is that when you damp the 43 hZ enough, there may be other room modes that get "uncovered" also. For that reason, maybe the brute force idea of trying lots of Ethan's traps may work better - especially since he's willing to let you try before you buy... Steve

[len-morgan]

Just a thought:

Is there any chance of creating a "control" sensor that ONLY picks up the 43 Hz? You could then flip/adjust the phase of the control and mix the two together to cancel out the offending frequency.

A company I used to work for designed and built an attenuator calibrator that used this principal to measure attenuation down to 1,000th of a dB. (Not a typo). The measurement took quite a while because the unit would try and null out the two signals and had to integrate the noise out so the true measurement could be made.

Like I said, it's just a thought.

len

[vinc]

Thousandths of a dB? Wow.

I had some thoughts along these lines. Unfortunately this won't work for us. The relationship between incoming vibration and the induced noise is highly dependent on the geometry of the particular test run, the samples used, etc. Also it's highly nonlinear.

[len-morgan]

While you may have different sample geometries, doesn't the same rule still apply? I.e., wouldn't the control sample be subjected to the same FREQUENCY and AMPLITUDE of interference as what you are measuring? If this is the case, even though the two samples would have different absolute outputs (perhaps gain differences in the system), I would think you would have enough information to pick out the 43 Hz from the real signal since you "know" what to look for. It might take a little math to get level in the control to match the phase/frequency response of the sample you're trying to measure but once you figure it out, I would think it would a reusable formula.

Since we don't know exactly what it is you're doing or measuring, it's hard to give you an exact answer. Perhaps a picture?

len

[lovecow]

vinc,

Are you using a microphone or a vibration transducer of some sort? It appears as though you're using the latter, attached to some sort of wire? Which is then connected to a slab on springs? If I've got this right, have you calculated whether this arrangement resonates at 43 Hz? 'Cause if it does, changing its properties so they're not susceptible to external noises at 43 Hz could help.

IMHO, "trapping" is not going to help you a bunch. If your equipment is that sensitive to a room mode, no amount of "trapping" is going to eliminate the effect on your equipment.

Other thoughts:

What's the source of the noise in the room? How is 43 Hz being excited? Can you hear it? (If you cannot hear it, all the more reason "trapping" is not a likely fix. That low of a level of noise or vibration will probably not be affected by some fuzz in the corner.) Do you have machinery operating in the room? Nearby mechanical equipment? Knowing more about the exact source would help a bunch.

Any chance you can move the experiment elsewhere?

Also, if you aren't concerned with the experimental data in the 43 Hz range, you should be able to remove the noise electronically.