The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
Measurements Comparison Dynamics Plugins
Old 19th December 2010
  #1
Lives for gear
Measurements Comparison

Hello

After three weeks of room tuning, I'm at the last steps..
Just need some suggestion/opinion

Which of these two measurements is better?
What changes between the two is speaker distance to the wall and to the desk

Thank you!

p.s. oh well, the measurements are two, one not approximated, the other with 1/3 approximation
Attached Thumbnails
Measurements Comparison-1.jpg   Measurements Comparison-1b.jpg   Measurements Comparison-2.jpg   Measurements Comparison-2b.jpg  
Old 19th December 2010
  #2
SAC
Registered User
 

Hmmm.

First of all, smoothing is simply like squinting - it makes nothing clearer.

In deference to some who seem to like bit of smoothing, nothing is gained by minimizing the degree of comb filtering! its not like it goes away - but it will , to someone unaware, potentially fool one into thinking that it is less severe than it really is and thus result in contributing issues being ignored!

And your frequency response, aside for its examination below ~300 Hz and then examined in the form of a waterfall/CSD plot in order to examine modal behavior, only tells us that there is comb filtering.

The frequency response cannot directly answer your question of which reflections are contributing to any destructive interference by virtue of the nature of the frequency response. It simply shows us the result of all of whatever superposition (summing of direct and indirect signals) combined with whatever response the speakers themselves contribute.

In other words, the FR is like looking at a picture of the aftermath of a riot and asking who did what.

Thus, in order to identify the individual signals that cause the interference displayed in the FR, you need to generate ETC responses from the impulse response made with each speaker operating independently - and with the mic remaining in the EXACT same position for ALL measurements.

From the ETC response, you can then resolve each direct and indirect signal and associate them with their appropriate paths, points of boundary incidence, etc.


Oh, and one more thing concerns me. You say you are just about done with the treatment and you are just beginning to make measurements now??? Do I understand you correctly? I ask as that is backwards. And trying to evaluate the room now, we will be unable to easily tell what behavior is due to the room itself, and what might be contributed by the treatment as well - and yes, treatments , if misapplied, can indeed contribute - as for instance, broadband absorbers are not necessarily absorbent at all frequencies and at all incidence angles - where they become reflectors...

So, I hope I have not scared you...I am just very quickly trying to run through a few concerns you might want to investigate - and hopefully, in the process stimulated a few more questions...
Old 19th December 2010
  #3
Lives for gear
Quote:
Originally Posted by SAC View Post
Hmmm.

First of all, smoothing is simply like squinting - it makes nothing clearer.
I did the smooth because someone says it helps understanding the big picture and the not smooth graph to people like you an me who use no approximation everytime

Quote:
Thus, in order to identify the individual signals that cause the interference displayed in the FR, you need to generate ETC responses from the impulse response made with each speaker operating independently
ok, could you tell me how to do that with room eq?

Quote:
- and with the mic remaining in the EXACT same position for ALL measurements.
the mic is always in the same position

Quote:
Oh, and one more thing concerns me. You say you are just about done with the treatment and you are just beginning to make measurements now??? Do I understand you correctly? I ask as that is backwards. And trying to evaluate the room now, we will be unable to easily tell what behavior is due to the room itself, and what might be contributed by the treatment as well - and yes, treatments , if misapplied, can indeed contribute - as for instance, broadband absorbers are not necessarily absorbent at all frequencies and at all incidence angles - where they become reflectors...
every changes I do to the room I measure.. I'm going crazy... every added traps, every inch I move stuff... I measure
sometimes is hard to understand if a measurement is better than another.. that's why I posted these two

Quote:
So, I hope I have not scared you...I am just very quickly trying to run through a few concerns you might want to investigate - and hopefully, in the process stimulated a few more questions...
yeah you have scared me, are 2 years I'm doing measurements, traps, studying stuff.. trying figuring out how things are... and I thought that was a quite good result.. so, how it is? here are the waterfall.. all under 200ms .. so I haven't decay problems...
Attached Thumbnails
Measurements Comparison-1w.jpg   Measurements Comparison-2w.jpg  
Old 19th December 2010
  #4
Gear Guru
 
Ethan Winer's Avatar
 

Lightbulb

Quote:
Originally Posted by elan View Post
Which of these two measurements is better? What changes between the two is speaker distance to the wall and to the desk
When using measurements to find the best places for speakers and traps etc, all that matters is the response at 300 Hz and below. Problems above 300 Hz are easily fixed with relatively thin absorbers.

As for averaging, 300 Hz and lower should be shown at maximum resolution, and above 300 Hz should be averaged to 1/6 octave. Hey, that was easy!

--Ethan

________________
The Acoustic Treatment Experts
Old 19th December 2010
  #5
Lives for gear
Quote:
Originally Posted by Ethan Winer View Post
When using measurements to find the best places for speakers and traps etc, all that matters is the response at 300 Hz and below. Problems above 300 Hz are easily fixed with relatively thin absorbers.

As for averaging, 300 Hz and lower should be shown at maximum resolution, and above 300 Hz should be averaged to 1/6 octave. Hey, that was easy!

--Ethan
But isn't ok now I've posted the waterfall?

Anyway, what do you think about measurements? Are that bad?

Overall seems around +/- 5 db... shouldn't be that bad

I had a lot of comb filtering in that room and dips of 40db and more in some places..

What I done has been, starting from the 77hz dip, I took a sine pointed at particular frequencies that null in the listening position, and I tried to find the places where that frequency was louder, using ears and a spl meter

In many places, placing a trap over them, didn't change anything and often made things worse (nulls became deeper), in other places the null filled in.. so, steps by steps the nulls that were -40 db become much less.. as you can see

BTW I still have comb filtering and honestly it's hard to figure out from where it comes

Come on, give me some opinion :P is that bad?

(consider also the speaker are not toed in in one measurement and a little toed in in the other, because as Dan Dan suggested, is a way to tame the highs)
Old 19th December 2010
  #6
Gear Guru
 
Ethan Winer's Avatar
 

Lightbulb

Quote:
Originally Posted by elan View Post
what do you think about measurements? Are that bad?
No, not bad at all.

--Ethan

________________
The Acoustic Treatment Experts
Old 19th December 2010
  #7
Lives for gear
Quote:
Originally Posted by Ethan Winer View Post
No, not bad at all.

--Ethan

________________
The Acoustic Treatment Experts
Ok, that's a good start :P
Which of the two graphs is better? And why? I mean what is better in one compared to the other

BTw, then, what could I do to improve the comb filtering?
I have the ER points covered (ceiling, front, right) I avoided the left because placing even a 2" rockwool sheet, made things much worse in the mid, gave me more comb filtering.. I don't know why.. and even a 2" panel in front of the desk (for the floor/desk reflection)... the back wall has 2 of 4" panels one over the other till 4' height , but the speakers are lower so I think the ER point is covered even in the back

Do you think now comb filtering is the main problem or I have to fix more the lows, mids?

Unfortunately what scares me is comb filtering and the fact that the two speakers response is quite different
Old 19th December 2010
  #8
Lives for gear
Some pictures of my studio
Attached Thumbnails
Measurements Comparison-img_0120.jpg   Measurements Comparison-img_0121.jpg   Measurements Comparison-img_0122.jpg   Measurements Comparison-img_0123.jpg   Measurements Comparison-img_0124.jpg  

Measurements Comparison-img_0125.jpg   Measurements Comparison-img_0126.jpg   Measurements Comparison-img_0127.jpg   Measurements Comparison-img_0128.jpg   Measurements Comparison-img_0129.jpg  

Old 19th December 2010
  #9
Lives for gear
Other pics..
Attached Thumbnails
Measurements Comparison-img_0130.jpg   Measurements Comparison-img_0131.jpg   Measurements Comparison-img_0132.jpg   Measurements Comparison-img_0133.jpg  
Old 19th December 2010
  #10
SAC
Registered User
 

Again, the frequency response is not the tool you use to analyze and treat early reflections.

You use the ETC.

From the ETC you can identify and treat each individual reflection that may be causing a problem.

The various measurements, if they are not the same, will vary if there is a lack of LR symmetry.

You keep saying that you 'do not know why" to a variety of issues regarding comb filters. Then generate left and right ETCs and address the specific early reflections that exhibit excessive gain that appear within the region of a defined ISD

That would be an excellent starting point.

Again, just to repeat for emphasis,-the frequency response does not provide any insight into the causes of comb filtering - it just shows that there is. And if that is what you are using it for, you do NOT use smoothing, as you want to see the comb filtering details and the depths of the nulls - and i know of no one who tests that smooths to greater than 1/48 octave. but all of that is rather moot if you use the proper response to evaluate reflections in time - and that is the ETC.
Old 19th December 2010
  #11
Lives for gear
Ok, I'm searching about ETC, I just have never used it, so I don't understand how to read the graphs and how to proceed in the measurements

Anyway SAC, what to you think about the measurements in general?
The method I used has been, emptying most of the room (I couldn't entirely because the wall full of rockwool behind the speakers and the corner filled, couldn't be removed.. )

Then I found the right placement for the speakers and placed the traps, with a sine on 77hz, 194, 488 etc.. many more, then walked on the room to find the places where it is louder and so on.. and that's the result I got..

I mean, the results I got are not good to mix in it? The comb filtering I have is a big problem or "last tuning" of the room? I mean I should start from scratch or am I at a good point?
Old 19th December 2010
  #12
Lives for gear
ok here are the ETC

now I'm figuring out how to read it.. :P
Attached Thumbnails
Measurements Comparison-stand33desk37.jpg   Measurements Comparison-standtoedin.jpg  
Old 19th December 2010
  #13
SAC
Registered User
 

We go in circles.

I do not use the frequency response as a measure of the quality - well, with the exception that big anomalies indicate other problems we should have looked at either in the waterfall modal analysis, or at the ETCs showing behavior in the time domain.

I do not expect a flat frequency response. Thee will be a degree of closely spaced comb filtering in a good room (look at the FR plot of a QRD) as we will have reflections.

Again, looking at the ETC is the tool for analyzing this - the frequency response is essentially worthless.

Here, let me put it this way. For modal analysis below ~300Hz, use the frequency response and the waterfall/cumulative Spectral Decay. Above 300 Hz, use the L&R ETCs. Show me the ETCs and I can tell you exactly what is going on. And if the ETCs and the subsequent reflections are addressed appropriately, the frequency response will (magically) be fine.

So continuing to ask me if everything is OK/good/or bad by looking at a very compressed or smoothed frequency response where I cannot separate the speaker behavior and the room behavior..... my response will be the same - generate and give me the ETCs....heh
Old 19th December 2010
  #14
Lives for gear
I gave you the ETC

I just don't know how to read it.. I'm googling..
Old 19th December 2010
  #15
Lives for gear
maybe zoomed is better?
Attached Thumbnails
Measurements Comparison-stand33desk37hi.jpg   Measurements Comparison-standtoedinhi.jpg  
Old 19th December 2010
  #16
Lives for gear
 
amishsixstringe's Avatar
 

Show an ETC graph that is windowed from 0sec to about 40ms. That is all that is important really. 300ms is sound that has traveled over 300 feet.

Neil
Old 19th December 2010
  #17
Lives for gear
 
amishsixstringe's Avatar
 

Wow. Even seeing it that zoomed out, there's ton of reflections in the first 25ms. Zoom way in.

Neil
Old 19th December 2010
  #18
Lives for gear
Here we go
Attached Thumbnails
Measurements Comparison-stand33desk37hii.jpg   Measurements Comparison-standtoedinhii.jpg  
Old 19th December 2010
  #19
Lives for gear
what are those big white parts from 100ms to 250?
Old 19th December 2010
  #20
Lives for gear
 
amishsixstringe's Avatar
 

Ok, now make it so you see 0dB down to about 30dB. We don't really need to see a 120dB range.

We're getting close!

Neil
Old 19th December 2010
  #21
Lives for gear
Ok, thank you for your patience :D
Attached Thumbnails
Measurements Comparison-stand33desk37hiii.jpg   Measurements Comparison-standtoedinhiii.jpg  
Old 19th December 2010
  #22
Lives for gear
 
amishsixstringe's Avatar
 

...well now you're cutting off that first impulse. Does it go all the way up to 0dB? It is important to know how much lower the first reflection is from the original impulse. Can you post it one more time with that information there?

Neil
Old 19th December 2010
  #23
Lives for gear
Sorry :/
Attached Thumbnails
Measurements Comparison-stand33desk37hiiii.jpg   Measurements Comparison-standtoedinhiiii.jpg  
Old 19th December 2010
  #24
Lives for gear
 
amishsixstringe's Avatar
 

Ok, cool. I don't see super gnarly reflections. Your room actually looks quite dead. Effectively anechoic, even. You may have said this already, so I'm sorry if I'm having you repeat yourself, but are your earlier FR graphs shown with both speakers firing at the same time? I'm having a hard time seeing why you'd have so much comb filtering if that's only one speaker, given the ETC you have presented.

So, that is next. Is there one speaker or 2 being tested at the same time for the FR graph you've shown? For the ETC?

Also, just for a quick reference...what was changed in the room between the two measurements again?

Neil
Old 19th December 2010
  #25
Lives for gear
Two speakers... what changed is just the stand and desk position, of 2" more or less and the yellow one ave the speaker a little toed in while the first has the speaker parallel
Old 19th December 2010
  #26
Lives for gear
 
amishsixstringe's Avatar
 

Sorry, I see you moved the speakers. In the Yellow ETC you can see the very first Very early reflection (probably the desk) is just slightly lower than the pink one. It also looks like the reflections come slightly sooner. So the Yellow one is with the speakers closer to the listening position. Correct?

It's hard to tell, but it looks like there's a reflection at like 1ms. I hate to have you keep posting graphs, but can you zoom WAY in? Only show about 10ms total on the graph. At first glance I didn't see that really early reflection, but I think there is one there. It is likely a desk reflection, if so.

Neil
Old 19th December 2010
  #27
SAC
Registered User
 

OK.
I will say that I am not sure of the actual gain levels - so i will proceed as if the dBfs values actually correspond to the dB SPL levels........and for this I am also assuming the ETCs were correctly generated - as this is simply a rough guide of some basics of what to look for in the ETC that you will be able to do at your leisure...
Also, for he ETCs, you make them with only one speaker firing at a time, so you will have a left and a right ETC corresponding individually to the left and right speaker. Yes, and for the FR, you onl yuse one speaker! otherwise you ae just verifying that comb filtering is a result of superposition - as you have two sources - i from each speaker, plus whatever
boundary reflections are present.

The area in the red box indicates an estimated 10 ms ISD region (selected arbitrarily).

All of the spikes in this region indicate reflections exceeding a 20 dB down gain level that you want to 'remove' through absorption of redirection.

Also, as indicated in the oval, the near clutter will be due to the speaker driver alignment itself and/or cabinet/spkr mounting diffraction and reflections from the spkr shelf mount. You want to clean this up as these are deadly with such a short delay.

Assuming the plot is correct, you get a pretty good idea of what is causing the comb filtering in the Freq response.

Once this is addressed, we can discuss what you might want to as far as ISD termination (if you want to go that route) and with the later arriving soundfield.

But this is simply intended to be a very quick cursory look with a few pointers of what to look for.

Note, the largest spikes in that period are going to correspond to the desk, ceiling, and side walls... Without knowing the exact dimensions, and without seeing the room as I type this, by determining the ~ distance between the speaker and the mic by way of incidence with each of those boundaries, and then set them in order of shortest to farthest, you will have a pretty good guess of which surface corresponds to each of the largest early reflections. But you will have to verify this empirically as we don't have the offset time (rrr.....)

So, again, assuming the gain scales are equivalent to dB SPL (which I have questions about here) you have an idea of the reflection paths and the location of each region of incidence on the various boundary surfaces - as this is what would specifically need to be treated with absorption.

How's that for a start?

Here, I will post a very simple and quickly annotated plot.
Attached Thumbnails
Measurements Comparison-12-19-2010-etc.png  
Old 19th December 2010
  #28
Lives for gear
 
amishsixstringe's Avatar
 

If you test the frequency response with one speaker at a time you will see the comb filtering go away. You're just confirming HF interference by testing both at the same time. Like SAC said, this is pretty irrelevant. If you make the ETC right, the frequency response will be right.

So, zoom in on that first reflection and see if we can locate and destroy it.

How deep is your room? IE, what is the longest dimension?

Neil
Old 19th December 2010
  #29
Lives for gear
No problem, it's just the price of my ignorance about ETC.. that's the fist time for me.. and still I haven't understood how to read them
Attached Thumbnails
Measurements Comparison-stand33desk37hiiiii.jpg   Measurements Comparison-standtoedinhiiiii.jpg  
Old 19th December 2010
  #30
Lives for gear
 
amishsixstringe's Avatar
 

Cool. I was just mostly trying to get your graphs set up so somebody like SAC or Lupo could look at them and tell you what's really going on. I'll let him take over as long as he's committed. He's a wealth of knowledge in regards to ETC and ISD termination.

I was asking you the depth of your room to decide how we can fix the fact that you don't have any late arriving reflections at all. I'm sure SAC would like to see that changed, as would I.

Neil
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
Similar Threads
Thread
Thread Starter / Forum
Replies
soundgeezer / Studio building / acoustics
26
kas_lar / Studio building / acoustics
2
nixoblivion / Bass traps, acoustic panels, foam etc
13
Audiop / Gear Shoot-Outs / Sound File Comparisons / Audio Tests
11
iquinn / Mastering forum
9

Forum Jump
Forum Jump