The No.1 Website for Pro Audio
 All  This Thread  Reviews  Gear Database  Gear for sale     Latest  Trending
Building an echo/reverb chamber
Old 6th July 2009
  #31
Gear Guru
Uncertainty

Hi Ethan, I think nonlinearity with SPL was probably mentioned as an interesting aside. I have no doubt that those posters know what they are talking about and I found the contributions a little intriguing. It is no surprise to me that things are a bit more complicated than they first appear. I mean we used to think the Speed of Light was constant, didn't we? Intuitively I doubt that the phenomenon is of enough magnitude to have an audible effect in the world of small studios, and even if it did, what could we do about it?
On the other hand I am very interested in the topic of LF measurements., in particular the variability we are experiencing. Several of us here experience that regularly. You have seen this yourself in ETF. Many overall results depend on accurate info in each band, e.g. rating of partitions, floors, etc. so I do think there room for improvement and discussion may help that along.
Personally, I find Decay spectra a lot more informative than Frequency Response graphs. I definitely want Decay information down to 40Hz.
FM uses the Sine Sweep also. It can perform what it calls synchronous averages. Each doubling of the number of sweeps adds 3dB to the S/N ratio. The method ignores once- off events such as cars passing or whatever, probably the same as REW. There is a MLS method (ETF uses it I believe) which some say is even more robust.
Even with all this, one rarely sees consistent EDT figures at LF. Now why is that, and can we improve it? For a start I suggest we stop use the term Decay time, rather than RT60 or even EDT. I like John's new notion of Topt.
B and K and others use Sine Sweeps also, but as I have said, when doing RT, they average several sweeps and each one is different. I have seen it. All these software packages are still extremely useful tools, particularly the Waterfalls which are the most reliable, but definitely room for improvement. John at REW in particular says he is working on new methods to look at LF. I look forward to that.

DD
Old 6th July 2009
  #32
Lives for gear
 
lord_bunny's Avatar
 

I've got a 1400' sq ft room that i was thinking of using as a chamber as well as my tracking room. I have EV Force One P.A. speakers and was wondering if there was ANY benifit using speakers of that pedigree (or lack thereof) for a chamber. The mic's I'd use, probably 414b-uls in omni.
Old 6th July 2009
  #33
Gear Addict
 

You want to capture as little as possible of the direct speaker signal anyway, unless you're going to re-record the entire track. Takes a bit of creativity regarding placement. We usualy point the speaker at the ceiling or the opposite wall.
Old 6th July 2009
  #34
Gear Guru
PA Speakers

I don't know those speakers in particular. However I do reckon that PA speakers would be ideal for this purpose. They are generally physically robust and capable of high SPL's. In context we don't need Monitor quality or even a full range signal.
DD
Old 6th July 2009
  #35
Motown legend
 
Bob Olhsson's Avatar
 

A 57 is by far the best chamber mike I've ever found. The off-axis response isn't muddy and the proximity effect ducks traffic noise like nothing else I've ever used while still preserving an amazing amount of low frequency reverberation. I aim it away from the speaker and aim the speaker into a corner.
Old 6th July 2009
  #36
Gear Addict
 

I suppose it's time to do a speaker/mike shootout for reverb chambers, including orientation. Any volunteers?
Old 6th July 2009
  #37
Gear Guru
 
Ethan Winer's Avatar
 

Lightbulb

Quote:
Originally Posted by DanDan View Post
Intuitively I doubt that the phenomenon is of enough magnitude to have an audible effect in the world of small studios, and even if it did, what could we do about it?
This is exactly the point. If air becomes nonlinear by a tiny amount at levels 20 dB higher than will blow out your ear drums, who cares?

Quote:
I am very interested in the topic of LF measurements., in particular the variability we are experiencing.
Yes, but the most sensible explanation is traffic rumble from outside. Or a distant plane, or your own footsteps, or similar.

Quote:
I find Decay spectra a lot more informative than Frequency Response graphs. I definitely want Decay information down to 40Hz.
I'm totally with you there. And I want to see to at least as low as the speaker's -10 dB point.

Quote:
The method ignores once- off events such as cars passing or whatever, probably the same as REW.
REW does that too, but I've never tried it. I should! From the REW Help:

Quote:
If Sweeps is more than 1 REW uses synchronous pre-averaging, capturing the selected number of sweeps per measurement and averaging the results to reduce the effects of noise and interference. The pre-averaging improves S/N by almost 3dB for each doubling of the number of sweeps. Averaging is particularly useful if the measurements are contaminated by interference tones, whether electrical or acoustic, as they typically will not add coherently in the averaging and hence will be suppressed by the process
Quote:
one rarely sees consistent EDT figures at LF. Now why is that, and can we improve it?
Beats me. But whatever the cause, it's not because rooms are nonlinear. heh

If John is still here, maybe he can chime in?

Quote:
when doing RT, they average several sweeps and each one is different. I have seen it.
But not with a moving microphone as in a reverb lab, right?

--Ethan
Old 6th July 2009
  #38
Gear Guru
Some Answers

Ethan, I have used the Synchronous Averaging. It does as it says on the tin. This, in conjunction with high test SPL's should help, but in reality I have not been able to improve the LF stuff. RT analysis is looking for nice Logarithmic decays. Modes screw this up big time. I have seen decays with several 'slopes'. This may be a contributory factor to the variations. The measurements I have referred to have indeed been done with rotating boom mics in certifed Rev rooms. Also in Semi Rev rooms. University of Liverpool.
Even in such august circumstances the sweeps vary and some are rejected.
I have also seen big differences when reading Impulse Responses taken with Dirac and read in FuzzMeasure, but that's another story.
DD
Old 6th July 2009
  #39
Gear Guru
 
Ethan Winer's Avatar
 

Lightbulb

Quote:
Originally Posted by DanDan View Post
The measurements I have referred to have indeed been done with rotating boom mics in certifed Rev rooms. Also in Semi Rev rooms. University of Liverpool. Even in such august circumstances the sweeps vary and some are rejected.
I was arguing the opposite. heh

When the microphone position keeps changing, of course you'll get different results every time. Versus tests like I did last weekend that gave the same results for each run.

--Ethan
Old 6th July 2009
  #40
Gear Guru
Opposite Opposite

Confused now. The moving mic is intended to spatially average. If done correctly, with multiple locations and enough samples, and averaging, the result should be the same every time. Statistics. And indeed it is if you go to all that trouble. Whether Boom mic or no, individual sweeps at the same location, with nothing else changing, give different results. I can see that you have some difficulty believing this Ethan, but surely you have noticed the problems with ETF and achieving Decay measurements?
DD
Old 6th July 2009
  #41
80425
Guest
Ethan is the man and knows his ****, you don't have to hop around here that much to know this.

Keep it going!
Old 7th July 2009
  #42
Gear Guru
Indeed

Clearly a discerning customer ;-)
Here's a picture of a very much used and loved echo chamber. A Bill Putnam 'design' from Cello Studios. Note the SM57's
DD
Building an echo/reverb chamber-putnamrooms.jpg
Old 7th July 2009
  #43
Lives for gear
 
nucelar's Avatar
 

Regarding linearity and non-linearity

IMHO, in acoustics measurements the observed non-linearities can only be attributed to factors outside the acoustic phenomenon itself.
Transducers are far from linear at high SPL, as well as the electronics and converters.
Another cause of observed non-linearity is the mechanical resonance of objects inside the room. We once measured a classroom and were getting very strange results. Turned out to be the chalkboard resonating strongly when a certain SPL was reached. Tables, chairs, and any other stucture will happily resonate non-linearlly at some point.

Exception to linearity is very high levels at ultrasonic frequencies. This effect is exploited by so-called directional speakers, which cause audible waves due to IM distortion in the air.
Old 7th July 2009
  #44
Gear Guru
 
Ethan Winer's Avatar
 

Lightbulb

Quote:
Originally Posted by DanDan View Post
The moving mic is intended to spatially average. If done correctly, with multiple locations and enough samples, and averaging, the result should be the same every time. Statistics.
For absorption measuring the tests I've seen are run 100 times and averaged. This gives high accuracy, but not necessarily high repeatability. This is why they have to run so many tests! (I'm kidding - the microphone is moving, so of course each run varies.)

Quote:
individual sweeps at the same location, with nothing else changing, give different results. I can see that you have some difficulty believing this Ethan, but surely you have noticed the problems with ETF and achieving Decay measurements?
I have gotten lousy data from ETF (and R+D) when measuring reverb, but that's because the software is buggy in that regard. Again, successive tests with REW do not have that problem. When I get a chance later I'll run five tests in a row in my large home studio, with the microphone way back in the middle of the room, and post the results here. Again, the issue is repeatability, not absolute accuracy for an RT60 measurement.

--Ethan
Old 7th July 2009
  #45
Gear Guru
 
Ethan Winer's Avatar
 

Lightbulb

Okay, I just ran five tests using REW, one after the other in rapid succession, and the results are below. As you can see, the differences are more or less "down in the noise." The only difference is very small, in the 2.5 KHz band, where tests #1 and #4 are a little higher than the others. It's raining lightly here so that might account for the difference, or maybe it's my cat Bear outside the door meowing to be fed, which he does constantly all day long. heh

This was done in my 34 by 18 foot one-room studio. The ceiling is 12 feet high at the peak halfway back in the room, and I put my DPA microphone up about 8 high, about 3/4 of the way back in the room. So the microphone was maybe 20 feet from the speakers in the front.

I have to assume that anyone here who has REW, or another program that uses a swept sine wave as the test signal, can duplicate this test and will get similar results.

--Ethan
Attached Thumbnails
Building an echo/reverb chamber-rt60x5.gif  
Old 7th July 2009
  #46
Gear Guru
No Bass

That test and REW only go down to 80Hz. Furthermore, that 80Hz is the lowest edge of a third octave band. That can hardly be said to fully, or even adequately describe the room in terms of it's usefulness for music. The largest amount of energy in music lies in the lower octaves.
So, clearly from 80 Hz up REW is quite repeatable. Below 80Hz is simply omitted. I reckon it is quite likely FM and ETF etc. are also quite repeatable above 80Hz. Below that they give variable and unreliable but still IMHO very useful results.
DD
Old 7th July 2009
  #47
Gear Guru
 
Ethan Winer's Avatar
 

Lightbulb

Then what's the disagreement?

This all started when I said rooms behave linearly, and what happens at low levels also happens the same at loud levels. When a few people objected I clarified and pointed out that high levels contain low levels, and any nonlinearity will manifest as distortion. We all know that's not the case at usual levels, and so far nobody has refuted what I said with a logical argument or graphs / data to the contrary.

So then the objections moved to my statement that reverb times are 100 percent repeatable as long as nothing changes in the test signal or the microphone and speaker placements. I have now proven this too is true. So now you want to move it yet again to only below 80 Hz, which conveniently happens to fall outside the range I can easily measure using REW. heh

Tell you what Dan (and others) - please give us a good explanation for why decay times below 80 Hz should vary from one run to the next, even though they don't vary above 80 Hz. (BTW, it looks like REW cuts off at 90 Hz, not 80.) I'm sure there's no plausible explanation. And if there is, I'll next measure the LF response and ringing in my studio five times in a row and post that as proof that subsequent runs are also the same on a waterfall plot. heh

Again, the only time you'll see variations from one run to the next at very low frequencies is when you use a rotating microphone stand. It's possible pink noise can vary too if the noise doesn't sustain long enough to self-average.

--Ethan
Old 7th July 2009
  #48
Gear Guru
None

Non linearity- the two posters who spoke of it have cried off. I don't blame them. I don't require proof from them. I found their contribution interesting. I doubt that it of significance in context here and there I would leave it.
Several of us have pointed out that LF Rev measurements do not come out the same each sweep. Fixed microphone. One position. That's an observation, it happens. I have used Bruel and Kjaer Building Acoustic Analysers, Dirac, FM etc. It happens. Interminable amounts of insistence that it cannot be will not make it so. It happens. Ethan since you have never seen this happen perhaps you could ask someone at a LAB to run a single sweep, fixed mic etc. Get them to derive Third Octave RTs from the sweep, then do it again. Or simply get them to record the two sweeps on a mic and send them to you. All should become clear then. This variability is well known and is one of the reasons that multiple sweeps are used. I don't know why it happens. Some hypothetical reasons have been put forward in earlier posts, certainly for pink bursts. I don't have a decent explanation for the variability using Sine Sweeps and MLS. I have suggested that software may have a difficulty establishing a 'slope'. John has gone into this in detail. Clearly it is not a simple matter, there are difficulties at establishing a 'slope' for LF. If each sweep resulted in exactly the same result, as Ethan insists, why are there confidence or quality indicators provided by B and K and REW and others on the results. These indicators are used to reject the results of some sweeps.
I don't consider it fair discourse to use the phrase 'shifting to below 90Hz'. I call omitting below 90Hz, well, an omission :-)
DD
Old 8th July 2009
  #49
Registered User
 

Can’t resist simply making an observation sure to step on everyone’s toes (after all, to do less wouldn’t be fair, would it? ;-)

First, the measurement of RT60s in a small acoustical space at frequencies where a statistically diffuse sound field does NOT exist, is a bit insane. Wrong tool, wrong environment! The very nature of the problem is operator error. RT60 measurements in acoustically large halls where they indeed do exist are not a problem regarding repeatability.

As far as small room measurements, a great deal of variability is to be expected! If for no other reason than: the methods being used to excite the room, the placement of, and the type of the mic(s) used, the fundamental nature of the measurement type itself, operator presence in the room during measurement and the interaction WITHIN a non-uniform energy distribution.

...Not to mention that if anyone is suggesting the averaging of measurements, this assumes the variability in such factors and fields (as well as the presence of other destructive stimuli - as a source of error or even mistakes...) – which sort of begs the question of the variability of small room sound fields which are by definition variable!!

But as far as the repeatability of measurements made with invariant stimulus, room conditions and measurement technique, and having the results vary in any way outside of a natural probability/error factor, something is amiss with the measurement apparatus, methodology, operation, or some combination of the various factors. And I would suggest that much of this would also point to the need for a more robust measurement methodology, platform and procedure.

As far as problems with diffusion and absorption measurements, wait for Ron Sauros' forthcoming paper on the matter. He blows this issue out of the water.

OK, you guys can go back to beating up on the proverbial horse and each other…;-)
Old 8th July 2009
  #50
Lives for gear
 
Berolzheimer's Avatar
 

I think the idea of using a PA speaker is a bad one, they tend to be pretty strident. Whatever characteristics the source has will be imparted to your reverb- I'd use the best, most linear or at least the most musical & pleasant sounding speaker you can. I agree about minimizing the direct path from the speaker to the mic, here the L shape of the chamber could be very advantageous- you can have the speaker around the corner from the the mics. I also agree that omnis are a good choice, as long as there is no line-of-sight between the speaker & the mics. I have one of those Behrengers too, it's a surprisingly good mic & I'd think it's a good choice for this application.
Old 8th July 2009
  #51
Lives for gear
 
andrebrito's Avatar
 

This has been my experience in large room acoustics: churches or concert halls.

I have tried different kinds of signal and different sound sources, pink, white, MLS, sweep... I NEVER obtain two identical measurements of RT or Lp or any other acoustical parameter particulalry at medium and low frequencies with the sound source and microphone position unchanged.

Changes in RT are more significant. Sometimes I even have to disregard some values since they don't make sense.
Old 8th July 2009
  #52
Gear Guru
Various

foxfyr, your point is well understood. I have earlier suggested using the term LF Decay and advocated such, to get away from the obvious misapplication of the RT60 term. I also suggest that useful Decay figures can be measured in less than perfectly diffuse spaces, by the use of judicious multiple measurement location. Not my idea, common practice.
To move forward I must say I like johnpm's Topt ideas very much.
Your confidence of non variance is simply at odds with my own and Andre's real life experience. I am afraid that divergence cannot be resolved by simple repetition.
May I pose a question here, which might shine a light?
Why does REW not present Decay figures below 90Hz?
Regarding the PA speaker, well obviously one would use a competent one. The suggestion was made in order to achieve high SPL, to get a decent S/N ratio. Obviously a bad anything is a bad idea. The Mackie I referred to is a fine transparent performer, with similarities in character to their excellent studio offering, which would also be a good choice IMHO
I reckon Ethan will agree on that one :-)
I would also guess an omni would be a good choice but I would certainly take big notice of the choice made by Bill Putnam and Bob O earlier, the SM57. Given the low costs perhaps several mics should be used, chosing the resulting varied tonal colours to suit the mix at hand?
DD

Last edited by DanDan; 8th July 2009 at 03:11 AM.. Reason: Typo
Old 8th July 2009
  #53
Lives for gear
 
andrebrito's Avatar
 

Just because measurements in a control room may not change that much with a fixed sound source and mic position does not mean the same will happen in a high reverberant church with a 900 m2 floor area...

Stating measurements are always linear not looking into what you are measuring and taking in account what room are you measuring is crazy. uncertainty is part of acoustical measurements, you can never excite the room in the same exact manner even using omni sources (been there done that!)
Old 8th July 2009
  #54
Gear Guru
 
Ethan Winer's Avatar
 

Lightbulb

Quote:
Originally Posted by andrebrito View Post
Just because measurements in a control room may not change that much with a fixed sound source and mic position does not mean the same will happen in a high reverberant church with a 900 m2 floor area.
I can believe the two types of spaces will respond differently. That makes sense, though I still do not understand the reason subsequent runs using a sweep tone would be different. Other than contamination by passing traffic etc. Pink noise is random so that could vary from one run to another. But why with a sweep? What exactly would cause different results?

--Ethan
Old 9th July 2009
  #55
Lives for gear
 
andrebrito's Avatar
 

Just the fact that a sound source can never excite a room the exact same manner.

This is much more perceptible in a high reverberant space and in large spaces
Old 9th July 2009
  #56
Gear Addict
 

Try exciting a small room at a modal frequency, put a mic in an antinode, and listen. You won't get a perfectly continuous signal. Why? Because we're dealing with reality here.
Old 9th July 2009
  #57
Gear Addict
 

BTW, we need a moderator here, to split up the thread. Perhaps a dedicated LF measurement thread, consolidated with the part in "Gap Thickness".
Old 9th July 2009
  #58
Gear Guru
Done

Probably better to the thread wander off into the distance. It has been keeping several balls in the air simultaneously. Confusing. Any wisdom gained, and I believe there is, will be used and stated much more clearly and in context throughout future threads.
DD
Old 9th July 2009
  #59
Lives for gear
 
Storyville's Avatar
Ok, couple of things to mention on this thread:

1. Room acoustics do not change unless the room changes. Now whether or not rooms react differently at higher levels of spls, I know when I track a guitar I "activate" the room and the cab to get the sound I want. but...

2. This could be more due to the microphones - which definitely react differently at higher spls.

The answer is you want a high enough to level to "activate the room", ie you can clearly hear the room generating vibrations. The only concern is not to distort your speakers. If you are using a 57, you pretty much don't have to worry about distorting the mic - but the spl coming into it will change it's tone a wee bit. 57s generally like it loud though.
Old 9th July 2009
  #60
Gear Guru
 
Ethan Winer's Avatar
 

Lightbulb

Quote:
Originally Posted by Storyville View Post
whether or not rooms react differently at higher levels of spls, I know when I track a guitar I "activate" the room and the cab to get the sound I want.
This is mainly psychoacoustics and Fletcher-Munson, plus making things loud enough to get the softer details above the noise floor and loud enough to hear. If you were to measure the room at soft and loud volumes you'd see the same response and decay times etc. So it's really our perception while listening in the room that changes, not the room itself.

--Ethan
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump