Quantcast
Large peaks in ETC - Gearspace.com
The No.1 Website for Pro Audio
Large peaks in ETC
Old 22nd January 2013
  #1
Gear Maniac
 
soundthinker's Avatar
 
🎧 15 years
Large peaks in ETC

I've noticed a series of peaks in the ETC of my living room when viewing with a large (5s) graphing window. Are these meaningful and if so, what do they indicate? I've done some searching online and in my acoustics books and haven't found an answer.

This room is connected to the dinning room via an open arched wall and a smaller den via a standard doorway. My best guess is that the peaks are somehow related to these adjacent spaces. But user error is always a possibility.

I've attached a link to the .zip'd impulse as well as the ETC graph.

Thanks.

PS: I see the same peaks when viewing the ETC in REW.

http://www.filedropper.com/livingroomaiff

Old 22nd January 2013
  #2
Gear Maniac
 
soundthinker's Avatar
 
🎧 15 years
hmmmm.... The peaks go away when looking a non-minimum phase adjusted measurement.

I still have no idea why they are there, but suspect the answer is more involved than I thought.

-Jim
Old 22nd January 2013
  #3
Gear Guru
 
1 Review written
🎧 15 years
ODD

The vertical scale you are using is too extreme.
60 dB or so should be plenty.
Room noise is typically 40dBish and your sweep should be around 90-95.
At that low level it is possible that you are looking at a digital artifact.
If your capture A/D was set to 16 Bit, and no dither, you would see something like that looping below 96dB or even higher if you didn't record at absolutely Full scale.
Here's a different view.
Large peaks in ETC-screen-shot-2013-01-22-01.41.48.jpg

DD
Old 22nd January 2013
  #4
Lives for gear
 
🎧 10 years
if the measurement is an accurate reflection (no pun) of what is happening within your bounded acoustical space, i would first investigate a flutter echo between two parallel surfaces of which the energy makes multiple passes across the mic (receiver position)
Old 22nd January 2013
  #5
Gear Guru
 
1 Review written
🎧 15 years
Time

A flutter echo starting at 700mS, gradually decreasing to 100mS?
Unless the OP is Marcel Marceau, I don't think so.
DD
Old 22nd January 2013
  #6
Lives for gear
 
Jolida's Avatar
 
🎧 5 years
Quote:
Originally Posted by DanDan ➑️
A flutter echo starting at 700mS, gradually decreasing to 100mS?
Unless the OP is Marcel Marceau, I don't think so.
DD
Old 22nd January 2013 | Show parent
  #7
Lives for gear
 
🎧 10 years
Quote:
Originally Posted by DanDan ➑️
A flutter echo starting at 700mS, gradually decreasing to 100mS?
Unless the OP is Marcel Marceau, I don't think so.
DD
why do you think i stated "if the measurement is an accurate reflection of what is happening within the bounded acoustical space".

post #3 shows indirect energy passing over the mic at common intervals that could be the result of two parallel surfaces.
Old 22nd January 2013
  #8
Gear Maniac
 
soundthinker's Avatar
 
🎧 15 years
Thanks for the replies. No mimes here.

I understand that for most practical treatment purposes I should be looking at a smaller window in both time an amplitude. I was mucking around with window settings, saw those peaks, and had absolutely no idea what they could be caused by.

DanDan, are you saying my sine sweep should be at a higher spl or does it look like my measurement technique is correct and I should just view those later peaks as artifacts?

I seem to recall a post that addresses dither in this context. I'll look that up, thanks.

localhost127, this is an asymmetrically arranged, untreated room. You are entirely correct about the many parallel surfaces.

Any idea why the peaks go away when not viewing a minimum phase adjusted measurement?

Thanks.

-Jim
Old 22nd January 2013
  #9
Gear Guru
 
1 Review written
🎧 15 years
Level

I am not sure what is going on soundthinker. I cannot recreate your graph.
The information on mine looks fine for purpose. I guess you normalised the measurement. In future just be sure to check your settings including in AudioMidi Setup. 24 Bit is good.

I do not see a regular repeating echo here.
Large peaks in ETC-livingroom_zps6a424385.jpg

EDIT, I just did succeed to replicate your graph. Very odd. I recommend you contact Chris Liscio at FM, he is very helpful.
I was about to suggest a feedback loop but again the repeats get shorter as time passes.

DD
Old 22nd January 2013 | Show parent
  #10
Lives for gear
 
jim1961's Avatar
 
🎧 5 years
Quote:
Originally Posted by DanDan ➑️
I am not sure what is going on soundthinker. I cannot recreate your graph.
The information on mine looks fine for purpose. I guess you normalised the measurement. In future just be sure to check your settings including in AudioMidi Setup. 24 Bit is good.

I do not see a regular repeating echo here.
Attachment 326640

EDIT, I just did succeed to replicate your graph. Very odd. I recommend you contact Chris Liscio at FM, he is very helpful.
I was about to suggest a feedback loop but again the repeats get shorter as time passes.

DD
I have one suggestion. Divide the ETC into one octave bands. Then see if the character of the repeats is in one area of the frequency range or encompassing all of it. It might give a hint to its cause.
Old 22nd January 2013
  #11
Lives for gear
 
Jens Eklund's Avatar
 
🎧 10 years
Looks like distortion components to me.
Old 22nd January 2013
  #12
Lives for gear
 
jim1961's Avatar
 
🎧 5 years
Large peaks in ETC-01-22-spectrogram-rew-gearlutz.gif

Here is what the spectrogram looks like. Sorta like an echo.
Old 22nd January 2013 | Show parent
  #13
Gear Maniac
 
soundthinker's Avatar
 
🎧 15 years
Quote:
Originally Posted by jim1961 ➑️
I have one suggestion. Divide the ETC into one octave bands. Then see if the character of the repeats is in one area of the frequency range or encompassing all of it. It might give a hint to its cause.
I've done that using the decomposition function in FM. The peaks seem to become more pronounced the higher the octave.
Old 22nd January 2013
  #14
Lives for gear
 
Jens Eklund's Avatar
 
🎧 10 years
Distortion components:

Large peaks in ETC-distortion-components.gif

When using sine sweeps for measuring harmonic distortion, the distortion components are usually found before the center peak of the impulse response.
Old 22nd January 2013 | Show parent
  #15
Lives for gear
 
jim1961's Avatar
 
🎧 5 years
Quote:
Originally Posted by Jens Eklund ➑️
Distortion components:

Attachment 326677

When using sine sweeps for measuring harmonic distortion, the distortion components are usually found before the center peak of the impulse response.
So mic or pre-amp overload?
Old 22nd January 2013
  #16
Lives for gear
 
akebrake's Avatar
 
🎧 5 years
New measurement-other room

Soundthinker,

Why don't take a new measurement in another room e.g. mic 3 ft from the loudspeaker and see if the same late reflections appear? As a reference so to say...

Make sure to avoid leakage (skip the loop-back) between the measuring signal and the reference. A shot in the dark...

I have around 300 ms latency in my macbook with FM. I am not a computer guy so I don't know if that's an operator error or part of the system.

Cheers

Ake
Old 22nd January 2013
  #17
Gear Guru
 
1 Review written
🎧 15 years
Mystery

I am inclined towards something electronic/digital. Definitely lose the Loopback and try again.
If these were discrete slapback from all the varied walls 10-20 feet away, they would not have the symmetrical ramp up and down to and from the peak.
Chris at FM, has probably seen this before.
DD
Old 22nd January 2013 | Show parent
  #18
Gear Maniac
 
soundthinker's Avatar
 
🎧 15 years
Quote:
Originally Posted by DanDan ➑️
Definitely lose the Loopback and try again.
DD
My signal path was Saffire Pro40 S/PDIF optically to receiver -> speaker. The only input was the measurement mic into the Pro40. No loop-back if I understand the term correctly.

I have contacted Chris and included a link to this thread.

-Jim
Old 23rd January 2013 | Show parent
  #19
Gear Nut
 
🎧 10 years
Quote:
Originally Posted by soundthinker ➑️
hmmmm.... The peaks go away when looking a non-minimum phase adjusted measurement.

I still have no idea why they are there, but suspect the answer is more involved than I thought.

-Jim
Don't combine a minimum phase record with ETC. You are triggering some DSP weirdness by combining the two operations.

Think of it as 'rounding error', but worse.

In general, you should avoid the minimum phase copy unless you understand the consequences. If you want to normalize the delay, please make sure that you are using automatic device correction (which you appear to be doing, so there's absolutely no need to use minimum phase copies.)

Chris
Old 23rd January 2013 | Show parent
  #20
Gear Maniac
 
soundthinker's Avatar
 
🎧 15 years
Quote:
Originally Posted by liscio ➑️
If you want to normalize the delay, please make sure that you are using automatic device correction (which you appear to be doing, so there's absolutely no need to use minimum phase copies.)

Chris
Thanks for responding so quickly.

Regarding the automatic device correction and my own FuzzMeasure education: I was feeding the sine sweep to my receiver digitally and was not looping back. The response I posted was a minimum phase copy of a non-auto-corrected measurement. Would an Impulse Response export of the original measurement differ from the impulse posted above?

My understanding is that every display is derived from the impulse response which is calculated after the sweep and doesn't change.

I'm talking over my head right now so feel free to just point me to another resource instead of spending your time trying to explain it to me.

Thanks.

-Jim
Old 23rd January 2013 | Show parent
  #21
Gear Nut
 
🎧 10 years
Quote:
Originally Posted by soundthinker ➑️
Thanks for responding so quickly.
Regarding the automatic device correction and my own FuzzMeasure education: I was feeding the sine sweep to my receiver digitally and was not looping back.
OK, so you're not really "eligible" for the loopback.

Quote:
The response I posted was a minimum phase copy of a non-auto-corrected measurement. Would an Impulse Response export of the original measurement differ from the impulse posted above?
It should differ considerably, yes.

Quote:
My understanding is that every display is derived from the impulse response which is calculated after the sweep and doesn't change.
Yesβ€”graph displays are derived from the impulse response. However, once you make a "copy" of an impulse with some kind of operation, you're modifying the impulse in some way. That's why I was careful to offer the minimum phase operation it as a "copy" routine, because the operations would destroy the original.

Quote:
I'm talking over my head right now so feel free to just point me to another resource instead of spending your time trying to explain it to me.
No worries. It can get over my head at times as well. The more I engage customers to try and explain this stuff, the more I firm up my own understanding.

Cheers,

Chris
Old 23rd January 2013
  #22
Gear Maniac
 
soundthinker's Avatar
 
🎧 15 years
Thank you Chris. Top notch service.

For anyone still interested the impulse and ETC from the original measurement are below:

http://www.filedropper.com/originalmeasurementimpulse



And with a more useful window:



-Jim

Last edited by soundthinker; 23rd January 2013 at 06:41 PM.. Reason: Added second picture
πŸ“ Reply

Similar Threads

Thread / Thread Starter Replies / Views Last Post
replies: 50 views: 45758
Avatar for anilnc
anilnc 23rd June 2011
replies: 2372 views: 403443
Avatar for TJ99
TJ99 2 days ago
Post Reply

Welcome to the Gearspace 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…

Forum Jump
Forum Jump