Specifying required acoustic environment 3

[KENAT]

I've got a feeling I and my colleagues are being dumb but please bear with me and help me out.

We manufacture precision metrology equipment that is very vibration sensitive - including to acoustic noise especially below around 2 kHz.

In our customer facilities requirement we specify the required acoustic environment for us to maintain our measurement performance.

At present this is essentially just saying something like "acoustic noise should not exceed X dBC".

However, a problem we're seeing is that although they may be below XX dBC overall, if there is concentration of power into small frequency bands then we see degradation of our performance - especially if those bands match the natural frequency of our structure or major component there of.

We'd like to come up with a better spec something like "acoustic noise should not exceed X dBC total with no more than Y% of power in any Z Hz frequency band".

Does anyone have any guidance they can share, or examples of spectra specified like this?

Thanks,

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[GregLocock]

Great question, BTDT and had to burn the T shirt. There is no harm in specifying a spectrum of noise that must not be exceeded.

Now there are several issues, first of all the frequency basis. Typically you'd use 1/3 octaves as a frequency but this may hide specific resonances. So you could use narrowband, but you need to specify how exactly this is measured.

Next up is how the acoustic environment is actually measured. Some ding-dong with a RadioShack SLM won't understand or care, but you need to consider spatial averaging and temporal. I'd cut to the chase and specify 5-6 microphone locations around your equipment, and specify a peak SPL in each frequency band while the equipment is in use, using settings X Y Z on a given acquisition unit.

Consultants may try and bamboozle you with weightings, (you suggest C weighting), that really does depend on what your instruments are sensitive to. For example A weighting is much loved by many industries, not least because it suppresses so many low frequency problems.

As ever, more details would help, but I think you are asking the right sort of questions.

As an aside, you could even include a mic and accelerometer in your instrument that alerts the user to a poor acoustic/vibration environment. Patent pending G Locock 2016.



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[IRstuff]

I would take a page from military vibration or acoustic noise specifications, which often have the entire power spectral density, which would seem to be what you need to go after, particularly in the ranges where your structure or equipment are especially sensitive. Like:

So, in your case, you'd notch out the frequencies that are resonant.

TTFN
I can do absolutely anything. I'm an expert!
homework forum: //

http://www.engineering.com/AskForum/aff/32.aspx

faq731-376 forum1529

 

[KENAT]

Thanks Greg - I'll be happy to try and supply more details if you can suggest what info might help.

Probably not what you're looking for but the systems I'm worried about are always on vibration isolation and in an acoustic isolation enclosure. That's how we do most of the testing but we will crack open the enclosure to try and better identify what frequency our actual systems themselves are vulnerable to.

We will on occasion work with the customer to do/get site surveys of noise & vibration done but it's not as routine as you might expect given the sensitivity of our equipment.

IRstuff, that's the kind of thing I/we was originally thinking but trying to get clean repeatable data of what frequency we're vulnerable to has been a struggle. It can vary quite a bit even between nominally similar systems - perhaps because of the limited control we have over the environment we test in. Best case we set up in a nominally quiet lab and then blast white noise at our instrument at increasing levels.

Looking again at spectrum we captured on a couple of very similar systems (once you get past the sub 15 Hz stuff) the noise is dominated by peaks of similar amplitude at 60 Hz, 180 Hz & 300 Hz with continued peaks but of ~ 1/2 the amplitude gradually diminishing at 420, 540 ... upto about 1.5 kHz where they become indistinguishable from general signal fluctuation. Initial suspicion was actually electrical noise from mains frequency (& harmonics) in our electronics but we've more or less ruled that out.

I also fear anything too complicated we come up with will be vetoed by marketing due to 'specmanship' games with our competitors.

I'm really feeling like a dummy and it gets worse with every page I read on the web, especially as several of my classes at university were in the

http://www.southampton.ac.uk/engineering/research/centres/isvr.page:-(

Though my physics PhD boss isn't doing much better and non of the other many Doctors & Masters have jumped in with much useful.

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[IRstuff]

The go-to approach for things like that is to either envelope the range of frequencies or to insert the notch with a narrative description of what is not allowed, e.g, "There shall be no a single frequency which exceeds 15 dB above the background ambient"

TTFN
I can do absolutely anything. I'm an expert!
homework forum: //

http://www.engineering.com/AskForum/aff/32.aspx

faq731-376 forum1529

 

[KENAT]

Thanks IRstuff, your last sentence seems like something we might be able to work with.

So if we have data showing we can operate our system in its enclosure up to say 90 dBC white noise before we see unacceptable 'average measurement performance' is it really as simple as setting a spec like:

"Acoustic noise should not exceed 80 dBC with no single frequency which exceeds 10 dB above the background ambient."

Seems a little too simple and I'm missing something. We'd be leaving some performance on the table but I'd rather that than upset customers.

The other problem we have is that even if our 'average measurement performance' meets our spec, strong dominant frequencies can still affect measurement quality and I'm not sure above entirely resolves that.

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[Strong]

Consider supplying a noise-controlled vibration-isolated enclosure that could be used in most lab environments meeting specified NC or NR curve rating and with a floor vibration level/spectrum rating. Otherwise you are in a no-win situation with no control of standing waves and transient vibrations and noise at your customer's facility. Your enclosure could be an extra cost option that your customer should want, if your metrology instruments are good.

Walt

 

[KENAT]

Strong, see above - we do offer acoustic enclosure.

Getting an acoustic enclosure that works adequately in even a nominally quiet fab or similar lab is challenging - our systems are more sensitive to acoustic noise than many sound meters.

Our newer hoods preserve performance up to quite high white noise levels. Trouble comes when we install in customer facility that has an environment which meets our X dBC spec over all but has strong peaks at narrow frequency bands which overwhelm the hood dampening abilities & show up in our measurements as anomalies or worst case excite resonant frequencies in our system so our overall 'average measurement performance' fails to meet spec.

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[Strong]

"our systems are more sensitive to acoustic noise than many sound meters." If this applies to a Type I precision sound level meter then you do have a problem. I have only been doing noise control engineering since 1968, so I am very surprised that anyone would use dB-C overall level for any interior environment specification, but perhaps I missed something along the way. I guess that you need a better "box".

Walt

 

[GregLocock]

dBC could be handy as it doesn't completely hammer the low frequency stuff. I've never used it in anger myself. However, it doesn't actually matter once you switch to a spectrum based requirement.

So, I'd set up 5 mics round the cabinet in a reverberant room, and do a sine sweep while monitoring the performance of the device. It shouldn't take long to map the allowable limit for single frequencies, but you'd need to think about combinations of frequencies as well.





Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[GregLocock]

Did you see this, I haven't read it yet

http://www.colingordon.com/upload/document/69d0c2725ba5ace76c1b8c6d81319879.pdf

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[IRstuff]

I think that still can be handled. The broadband specification is a PSD, which means that it assumes that the frequencies are random and non-stationary. So you can still do the PSD curve, but with a similar caveat, "There shall be no frequency that is more than -XX dB below the ambient for longer than yy seconds."

This can be verified by averaging a series of spectrum analyzer scans. The random frequencies will average down, but any fixed frequency will stay put.

Or, something like, "When unit audio noise is averaged over xx seconds, there shall be no frequencies that exceed the average more than yy dB."

TTFN
I can do absolutely anything. I'm an expert!
homework forum: //

http://www.engineering.com/AskForum/aff/32.aspx

faq731-376 forum1529

 

[KENAT]

Greg - that article looks interesting I'll make sure and finish reading it as soon as I can.

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[KENAT]

Well I just re-read that document more thoroughly and as I thought it basically says the same thing we've determined - that the environmental noise spec needs to be better specified than we're currently doing given our product's sensitivity.

On the bright side it's mention of scanning electron microscopes definitely puts it in our nanotechnology ball park. Shame non of my colleagues picked this up back when it was written when we might have had a bit more resource to throw at the issue.

Thanks again Greg & IRstuff.

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[Strong]

Google: precision metrology equipment that is very vibration sensitive

Perhaps this is what you need:

http://mikes.fi/research-the-mikes-building

Walt

 

[KENAT]

Thanks Strong,

I did search quite a bit before posting. Searching the terms you gives mostly results related to the kind of mechanical 'floor vibration' that various types of air table or similar fairly effectively isolate. Plus this thread comes up on the first page!

If I swap acoustic instead of vibration in the Google search then I get this thread as my top result, and one of our acoustic enclosure vendors is also in the top 3.

I'm not looking to build a nanotechnology facility, I'm trying to better specify the environment our instruments need in order to meet a stated performance level - pretty much the issue the paper Greg found also points out.

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[MikeHalloran]

Maybe you could add Active Noise Cancellation to the enclosure.

... or map the instrument's susceptibility, and stiffen or add damping to the parts that are the biggest contributors.

... or add microphones or accelerometers to the instrument, and basically add ANC to the core instrument itself.

Nothing cheap comes to mind, except maybe bolting the whole mess to a heavy concrete slab supported by partially inflated inner tubes.



Mike Halloran
Pembroke Pines, FL, USA

 

[KENAT]

Thanks Mike, primary consideration at the moment isn't making design changes to fix things but to better specify the acoustic environment we need to meet our performance specs.

That said.

Some optimization of the instruments structure etc. has been done as part of the original design and my various 'stretches' of our base platform. Damping doesn't work too effectively on primary structure from our experience, instead we try to keep resonant frequencies high (i.e. relatively light* & stiff structure). *Light is a relative term here, main structural element is a 60lb casting.

The idea of active noise cancellation has come up in one form or another and been played with a little, also tuned mass dampers etc.

We do make occasional use of active vibration control platforms in lieu of more typical passive air tables or other technology.

Industrial customers and some labs will create large concrete pedestals for our systems but that address mechanical vibration not noise.

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​