How much can the results of two identical scans vary?

[hilde45]

I hope you will indulge this question -- which may have been answered before.

I am measuring my room with a USB Umik and REW 5.20 on a Mac. I am doing my best to measure everything and keep everything stable as I change a single variable. Sometimes, when I take two scans at the same position, I get more than insignificant differences in the measurements. I'm trying to figure out if there's something I am doing which is introducing the variations.

QUESTION: Assume for the sake of argument that I do two scans without changing anything at all. To what degree can the two measurements differ? And how?

As you can tell, I'm trying to refine my experimental procedures to eliminate as much variation in my scans as I can. If you tell me what kinds of variations are natural between two scans where *nothing* has been altered, I will have a baseline. (I'm half assuming you'll just say "nothing" but acoustics seems a science which includes stochastic phenomena; if that's true, I need some practical advice about how much chance I can expect to see varying results in controlled, duplicate experiments.)

P.S. Perhaps there are even phenomena outside the listening space I should be aware of? A car driving by on the street? People walking above the room? Heating blowing air? I try to be aware of all the background noise and do the scans when it's quiet. I am in the basement room of a bungalow. I am not in the downtown but I'm not in the country, either.

 

[EarlK]

You are best to post an REW native .mdat file of the measurement examples that you're most concerned about ( best to limit the quantity to 3 so as to make the file small enough to post/attach ).

Others can then offer a better informed opinion.

If you create each measurement at around 75db then the differences between measurements should be minimized.

Use the "notes" window for each measurement to add details about the measurements ( time of measurement, hand-held mic or mic stand employed, distance to target, etc. )

 

[hilde45]

Wow! Ok that's very generous of folks to help me that way. I'll see what I can generate. Many thanks.

 

[hilde45]

Hi all, I'm uploading a scan here. This is the result of a lot of work -- hundreds of scans, many re-positionings of listening position (LP), speakers, even room furniture. I have tried, also, subs and a few OC 703 panels at different places. I'm on the verge of sharing this scan with some treatment companies (GIK, RealTraps, and maybe others). I cannot eliminate the nulls you see here and I do not know what else to try. I have read that I may need to try to trouble shoot some peaks to eliminate the nulls. Is that correct, and how would I experiment or deduce which peaks? I'm probably in over my head, but I want to know everything I can about my room's hard physical limitations before I buy a lot of treatments.

Speakers are stand mounts with a 35-30k range; Salk SS 6M. I have a REL 328 sub. Not in this scan.
Here is a diagram of my room and an .mdat file (small) Any help appreciated!

 

[EarlK]

@hilde45

Here's a document from a company of acousticians that has some guidelines for ( acoustics and measurements and how they contribute to ) good 2-chnl listening .

Some Suggestions ( to facilitate a professionals perusal ):
(i) Take three measurements ( each case executed without moving the test mic ) . Left Speaker individually, Right Speaker individually, Both Speakers together
(ii) In the "Notes" section for each measurement indicate distance from mic to speaker and approximate angle.
- Also include location of test mic ( include pertinent measurements of distances to walls . floor and Roof ).

I looked at your file and here are a couple of comments;

(a) Your RT60 times indicates that you have already treated the room sufficiently to get those times down to a sufficiently ( dry ) point.

(b) Unfortunately, other analysis shows ( Spectrogram and "Excess Group Delay" ) that you still have reflective surfaces creating quite distinct high energy reflections that will detract from good aural imaging ( IME ).

- I've high-lighted one such source of HF reflection seen in the spectrogram above ( 2ft 1.74" ).
- I would speculate that it's a significant contributor for the messy Excess Group Delay seen below ( note the correlation between the two graphs anomalies seen between 2K-10K )

(c) These two observations taken together suggest that you've simply treated the wrong surfaces ( IMHO ) rather than having not enough surface treatment ( as a percentage of all room surfaces ).



Take your files to a Pro and see what they say ( "Pro" meaning someone who will "monetize" their guidance and stand behind their opinions ) .

Treating the destructive ( reflective ) surfaces will certainly help linearize ( flatten ) the response of your speakers ( above the Schroeder frequency ) > typically existing somewhere between 150hz to 300hz ).

 

[hilde45]

Thank you so much, Earl, for your comments. And I will try doing the measurements as you suggest.
I've not done much professional treatment other than moving things around and adding some rugs, but I take your point. I need to look again at the source of reflections in the room, though with low ceilings, the answer may be (literally) right above my head.

I'm about to start contacting some pros. Gotta loosen the ol' wallet.