ProSoundWeb Community

Sound Reinforcement - Forums for Live Sound Professionals - Your Displayed Name Must Be Your Real Full Name To Post In The Live Sound Forums => Audio Measurement and Testing => Topic started by: Raimonds.Skuruls on February 12, 2014, 06:04:29 am

Title: delay estimation and measurement
Post by: Raimonds.Skuruls on February 12, 2014, 06:04:29 am
Everyone knows how delay estimation is challenging.
What about such kind of view on delays of 4 band loudspeaker system?
(http://R2008_pl04.png)
We can see that driver band is delayed about 1 ms after tweeter.
Midwoofer is delayed about 0.7 ms after tweeter.
If time span of graph is changed to 20 ms, we can see that LF band is delayed for about 8.3 ms.
(http://R2008_pl20.png)
And same graphs in isometrics.
(http://R2008_3D04.png)(http://R2008_3D20.png)
It is possible to create Delay Frequency Response from such graph.
(http://R2008_DFR.png)
You can find few more examples here
http://aplaudio.com/downloads/Introduction-to-TDA.pdf
Can this be helpful for your system tuning and commissioning everyday job?

BR,

Raimonds Skuruls
http://aplaudio.com

Title: Re: delay estimation and measurement
Post by: Tim McCulloch on February 12, 2014, 10:33:33 am
Looks like a "pretty picture" way of doing an impulse alignment.  Ivan? Chris? Timo? Langston?
Title: Re: delay estimation and measurement
Post by: Timo Beckman on February 12, 2014, 02:41:09 pm
I've seen this kind of a plot before and it's a nice picture but to me (underscore please) that's all it is.

I can do a lot more working with a phase display (or it's derivative group delay although the last time i looked at that is about a year ago after a "pissing contest" on a dutch forum).

Impuls alignment is something i do not use on system aligning a 2/3/4 or more way system. Has to do with High-pass and Low-pass filters causing group delay and thus changing "time" and the ir.
@ the topic starter you want to see something like the screen shots in this post at my blog.
http://timobeckmangeluid.wordpress.com/2013/01/12/again-playing-with-phase/
Here you find screenshots of the phase display and it's derivative the group delay screen which is what you're asking in my opinion?
 
Title: Re: delay estimation and measurement
Post by: Chris Tsanjoures on February 12, 2014, 03:31:16 pm
I would agree with Timo, the phase plot is the preferred view for time measurements. I can imagine the 3d spectrograph being useful for things like analyzing decay times, however rendering the 'slices' on the frequency scale rather than time is a red flag for me.
Title: Re: delay estimation and measurement
Post by: Ivan Beaver on February 12, 2014, 06:53:07 pm
I am not understanding this question

"It is possible to create Delay Frequency Response from such graph."

If the question is can you show the freq response from such a graph-I highly doubt it-mainly because there is not out of band information.

So I doubt it would be much good in trying to do a system alignment.

A simple phase repsonse is easier to read and show the interaction between different freq bands.

This seems to "stop" at the edges with no overlap
Title: Re: delay estimation and measurement
Post by: Arthur Skudra on February 12, 2014, 08:25:45 pm
Sveiki Raimonds!

Interesting software you have there!  I certainly like the 3D view, helps visualize certain characteristics of the loudspeaker.  Not sure if it's application in live sound reinforcement is useful for us in live concert sound, especially when room acoustics has considerable influence on what we see with our measurements, and in some ways makes it challenging to interpret the data displayed on the computer.

In measuring large scale sound reinforcement systems in very large rooms, most of us audio practioners use the impulse response for finding the general delay of a loudspeaker to get the dual channel FFT to be properly synchronized, then either use the phase plot or the group delay to figure out where in time certain frequencies arrive.

Though for some beginners it's hard to grasp the concept of phase at first, when it becomes clearer it is a very powerful tool for sound system alignment.

Arthur Skudra
Title: Re: delay estimation and measurement
Post by: Raimonds.Skuruls on February 13, 2014, 09:46:56 am
Yes, we can try to catch parallel phase responses by switching on and off some bands, but,
as you are not so close to a loudspeaker, as you are going down on frequency scale, as you have particular loudspeaker resonances, it is more and more difficult to realize that basic approach.
That’s because we need to get the reliable phase response. We can get it from impulse response by use of FFT. But impulse response (in real acoustic environment) is very noisy because of random reflections of environment. It means, the phase response is also heavy disturbed, noisy. Only serious smoothing helps little bit, but by the price of loosing valuable information. And if we are trying to calculate group delay from such phase response, we are going into deeper errors because of calculation of a derivative of the phase response.
This insoluble problem stirs me to action. Especially in car audio field, where problem of delay evaluation is very serious. (By the way, the car equipped with 6 channels of APL correction, performed the best sound in European final of EMMA car sound competition...)

I have processed lot of my previous medium and large scale (concert) loudspeaker system (including serious brands) measurements and did not find even one that was correctly tuned.
Title: Re: delay estimation and measurement
Post by: Timo Beckman on February 14, 2014, 04:36:58 am
So what would a correctly tuned set-up look like to you on a analyzer if i may ask?
Title: Re: delay estimation and measurement
Post by: Raimonds.Skuruls on February 14, 2014, 05:28:16 am
So what would a correctly tuned set-up look like to you on a analyzer if i may ask?
Thank you for a question!
If we have an ordinary IIR crossover, we must see just delays introduced by a crossover unit.
Picture  „crossover4way4order.png” is showing  delays of 4 way 4 order crossover.

If we have linear phase crossover we must see ideal transfer.
Picture „1.png”.

Real speaker must looks like this (2 way, 2 order crossover, no place to make delay mistake).
Picture „1_NS10.png” - NS10 monitor.

But we can improve it by minimum phase correction of Amplitude Frequency Response.
Picture „2_NS10_APL_mp_AFRcor.png”

And than, by introducing time correction – picture „3_NS10_APL_mp_AFRcor_TIMEcor.png”


Title: Re: delay estimation and measurement
Post by: Arthur Skudra on February 14, 2014, 09:56:14 am
Hi Raimonds,

Perhaps these links might be of interest to you!

http://www.fulcrum-acoustic.com/assets/pdf/whitepapers/spectrogram-loudspeaker-transient-response-2005.pdf
http://www.fulcrum-acoustic.com/assets/pdf/whitepapers/improving-loudspeaker-transient-response-with-dsp-2005.pdf
http://www.fulcrum-acoustic.com/assets/pdf/whitepapers/us8081766B2-creating-dsp-filters-improve-transient-response.pdf

Best regards,

Arthur

Title: Re: delay estimation and measurement
Post by: Raimonds.Skuruls on February 16, 2014, 07:05:38 am
Hi Raimonds,

Perhaps these links might be of interest to you!

http://www.fulcrum-acoustic.com/assets/pdf/whitepapers/spectrogram-loudspeaker-transient-response-2005.pdf
http://www.fulcrum-acoustic.com/assets/pdf/whitepapers/improving-loudspeaker-transient-response-with-dsp-2005.pdf
http://www.fulcrum-acoustic.com/assets/pdf/whitepapers/us8081766B2-creating-dsp-filters-improve-transient-response.pdf

Best regards,

Arthur
Sveiks, Artūr,

Thank you very much for links to those papers!
But seems like those works are still ideas (patent law view) for almost 10 years.
Evan patent is not describing the invention such way that it can be implemented without additional research work. It can not be prior art.
Pictures of those works have some similar properties to my work due to Heisenberg-Gabor limit.
Anyway, I have not seen any, ready to use in field, solution to show delays in easy readable way allowing to judge systems tuning quality in some seconds and than allowing tuning delays by direct measurement thus excluding errors due to derivate calculation when information about delay is acquired from heavy disturbed phase measurements.
BR,
Raimonds Skuruls
http://aplaudio.com

Title: Re: delay estimation and measurement
Post by: Timo Beckman on February 17, 2014, 07:03:50 am
Would be nice to see this kind of "perfect" looking measurements at different mic positions so onax and more important ofax 15/30/45/60dg etc mic positions.

If i do a measurement on what ever brand of speaker it can be nice to have a "perfect" response onax but if the ofax positions suck i have some more work to do (at least in my book) because the majority of people are at the ofax positions

If a loudspeaker has a perfect onax phase/magnitude response but a not so perfect ofax response there has to be a compromise between the all mic positions.
Title: Re: delay estimation and measurement
Post by: Martyn ferrit Rowe on February 17, 2014, 07:19:15 pm
Hi Raimonds,
You might find this interesting, Wavelet analysis that's used in Clio by Audiomatica.
http://www.audiomatica.com/wp/wp-content/uploads/2012/03/audaesny2007.pdf

Also there's a couple of MatLab packages for wavelet analysis.
cheers
ferrit
Title: Re: delay estimation and measurement
Post by: Raimonds.Skuruls on February 18, 2014, 06:40:40 am
Would be nice to see this kind of "perfect" looking measurements at different mic positions so onax and more important ofax 15/30/45/60dg etc mic positions.

If i do a measurement on what ever brand of speaker it can be nice to have a "perfect" response onax but if the ofax positions suck i have some more work to do (at least in my book) because the majority of people are at the ofax positions

If a loudspeaker has a perfect onax phase/magnitude response but a not so perfect ofax response there has to be a compromise between the all mic positions.
The „perfect" looking measurement” of delays is almost not dependant on a direction. Especially, for the coaxial design. If we have 3 ... 4 bands stocked vertically, we will have no dependence in horizontal plane and very small (negligible) in vertical plane. Hardest case is when different bands are fare from each other. We can make timing alignment just for one, most important direction in such case.


My friend John from Los Angeles sent me his first measurement of his studio monitors. I never can predict, that studio monitors may have such serious timing error – LF band is delayed for about 20 ms and crossover point is smear out. Please find picture attached.

The problem of controlling the speaker’s radiation in all directions is my previous work.
The key word in such problem is Sound (Acoustic) Power Frequency Response.
It is historically well forgotten thing. Tens of years ago the Reverberation Chambers (opposite to anechoic chamber) have been used to measure sound (acoustic) power emitted by speaker. And lot of works are pointing to importance of this and talking about this. Let’s stop talking and start doing. Let’s measure SPFR and create correction for it.  http://aplaudio.com/conc2/spfr/
Quote
Hi Raimonds,
You might find this interesting, Wavelet analysis that's used in Clio by Audiomatica.
http://www.audiomatica.com/wp/wp-content/uploads/2012/03/audaesny2007.pdf

Also there's a couple of MatLab packages for wavelet analysis.
cheers
ferrit
Thanks, Martyn, for valuable information. It will be addition to my collection of project names that are working on similar problems.
Title: Re: delay estimation and measurement
Post by: Raimonds.Skuruls on March 01, 2014, 08:17:27 am
At least two topics are about “align subs to mains” here.
Let’s see how to achieve this taking in to account everything – crossover filter responses, equalizer filter responses, particular loudspeaker resonances, physical placement of boxes.
http://forums.prosoundweb.com/index.php/topic,148587.msg1364464.html#msg1364464

BR,

Raimonds
http://aplaudio.com
Title: Re: delay estimation and measurement
Post by: Raimonds.Skuruls on April 01, 2014, 08:33:18 am
Two posts early I wrote about John’s problems with his monitors.
Actually, the monitors are Quested V3110  for about 4000$ each.
We did some work and get results.
Please find pictures attached.
1)   an initial AFR,
2)   an initial TDA delay graph,
3)   an AFR after minimum phase correction of AFR,
4)   a TDA delay graph after minimum phase correction of AFR,
5)   a TDA delay graph after delay correction.

And impressions of John:
 "Last night I listened to a project I recorded in 2005, I have listened to the project hundreds of times over the years. I heard details that I forgot were there like speed changes in the Lesile speaker and small inflections in the back up vocal. The reverb detail was un believable. It brought back memories of every detail of the sessions. The guitar tone reminded me of the guitar players amp and the mic I used. Over all I am very impressed with the correction, the detail is like headphones. I still need to work on the room some more and I would like a little more high end. More as I listen.
Thanks, John"

BR,

Raimonds
http://aplaudio.com
Title: Re: delay estimation and measurement
Post by: Raimonds.Skuruls on November 19, 2014, 07:09:58 am
TDA software gets new function – Non Linear Distortion analysis.
Now we can see frequency response of each particular harmonic.
Please find attached Non Linear Distortion analysis graph that is showing
almost 10% NLD at 2 kHz for some concert hall system.
(https://dl.dropboxusercontent.com/u/1893135/GORS_Rc.png)


BR,
Raimonds
http://aplaudio.com
Title: Re: delay estimation and measurement
Post by: Raimonds.Skuruls on November 26, 2015, 02:54:48 am
Hello Colleagues,

The discussion with Dr. Floyd Toole about his paper
"The Measurement and Calibration of Sound Reproducing Systems"
should be interesting to you.
The paper and start of discussion is available at AES forum:
https://secure.aes.org/forum/pubs/journal/?ID=524
Than the administrator of AES forum William McQuaide have decided:
"So we do not think that our readers will benefit from further postings" and denied next post.
All discussion is available here: http://aplaudio.com/downloads/Reading_Dr_Toole.pdf