FFT and Windowing - Page 3 - diyAudio
Go Back   Home > Forums > Loudspeakers > Multi-Way

Multi-Way Conventional loudspeakers with crossovers

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 2nd September 2013, 12:49 AM   #21
gedlee is offline gedlee  United States
diyAudio Member
 
Join Date: Dec 2004
Location: Novi, Michigan
I fear from reading this that it is not well understood that the window being used is highly dependent on what is being done. For example when one has an impulse response and needs to "window" out the reflections then all one needs to do is taper the window towards the end, a "typical" window like cosine or Hanning, etc. across the entire data set is not ideal in this case. One should not tamper with data within the window that you know is reflection free. Only taper near the end of the window. For example, I will use a 6 ms. window with only the last 1 ms. tapered as a cosine function. There are times when one wants to vary the taper length slightly to suppress a DC residual that can creep into the data when a window is used.

Now the case of steady state pure tone measurements like THD and or IMD are completely different and one needs to use one of the more typical windows with all the associated tradeoffs.

Then one can sync the data set to the sample rate and window length which will allow for the use of no window at all for the highest possible resolution. With this later technique one can also average across data sets thus reducing the noise and further enhancing the measurement resolution. With this technique you can measure THD below the noise floor of say an amplifier. (This probably would not work for an acoustic measurement like a loudspeaker because of delay issues.)
  Reply With Quote
Old 2nd September 2013, 11:10 AM   #22
diyAudio Member
 
scott wurcer's Avatar
 
Join Date: Jan 2004
Location: cambridge ma
Quote:
Originally Posted by gedlee View Post
With this technique you can measure THD below the noise floor of say an amplifier. (This probably would not work for an acoustic measurement like a loudspeaker because of delay issues.)
Earl, are you familiar with Angelo Farina and his work. This paper is worth recommending, http://pcfarina.eng.unipr.it/Public/...226-AES122.pdf
__________________
"Greetings from The Humungus! The Lord Humungus! The Warrior of the Wasteland! The Ayatollah of Rock and Rolla!" aka the Wizard of Wrestling.
  Reply With Quote
Old 2nd September 2013, 12:56 PM   #23
JMB is offline JMB  United States
diyAudio Member
 
Join Date: Apr 2004
Location: Texas
Hi Earl,

I think that most of the software in use for measuring loudspeakers does just what you suggest (at least from the shape of the window that appears when you select a method) and it generally boils down to which of these methods are best for the tails. I could be wrong in my interpretation of this but if one was to look at a Soundeasy FR when you apply a time window, that is what it looks like to me. ARTA looks more rectangular but it may be just the way it is depicted, I am not certain.

I was feeling more confident after reading the NI docs and explanations by feyz and SY but now you have me re-evaluating my understanding (perhaps that was your purpose?).

Would you be so kind as to explain the advantages of different windows from your perspective? In particular, if one time windows an impulse response to window out reflections, why then window for CSD or Tone Burst; seems to me it might be because they continue outside of the time window that was set. In addition, the tail of the Impulse response is windowed to eliminate the reflections, for FR, it seems that you would want a narrow window with low side lobes to eliminate cross contamination from neighboring frequency bins. In fact, I would appreciate your weigh in on how to weight the importance of a more narrow main lobe vs. lower side lobes in FR measurements (say, 7 term blackman harris with the extremely low side lobes vs. Hann with a more rapid decent (more narrow main lobe) but side lobes that are not as deep for FR, THD, IMD, Tone Burst Testing, etc, with loudspeaker measurements.

My take from the information that I have obtained is that for FR testing & CSD & Tone Burst, the lowest side lobes with the narrowest main lobe is beneficial and you would have to balance which is more important in comparing something like a Hann vs BH (your input here would be appreciated as to how to balance this) while for THD and IMD, since relative amplitudes are important something with a wider main lobe might be of benefit?
  Reply With Quote
Old 2nd September 2013, 01:11 PM   #24
JMB is offline JMB  United States
diyAudio Member
 
Join Date: Apr 2004
Location: Texas
Just to clarify my understanding further, a flat top or Hanning seems like it would be advantageous for CSD or Tone Burst since they continue outside the window and a Hanning brings down the side lobes well while a 7 term Blackman Harris or a Kaiser Bessel might be good for FR due to a narrow main lobe?
Attached Images
File Type: jpg FFT Window Comparisons from ni.jpg (74.2 KB, 61 views)
File Type: jpg FFT Window Comparisons from ni06.jpg (41.0 KB, 61 views)
File Type: jpg FFT Window Comparisons from ni08.jpg (19.1 KB, 61 views)
  Reply With Quote
Old 2nd September 2013, 01:26 PM   #25
JMB is offline JMB  United States
diyAudio Member
 
Join Date: Apr 2004
Location: Texas
I am also adding one more picture with a few other comparisons for anyone else following this thread with interest for reference.
Attached Images
File Type: jpg FFT Window Comparisons from wikipedia.jpg (74.6 KB, 45 views)
  Reply With Quote
Old 2nd September 2013, 01:55 PM   #26
gedlee is offline gedlee  United States
diyAudio Member
 
Join Date: Dec 2004
Location: Novi, Michigan
Scott

I use Prof. Farina's technique as implemented in HolmImpulse.

JMB

As to the shape of the window, your plots exhibit the tradeoffs between width of the main lobe and suppression of the side lobes. I would tend to prefer the narrower main lobe to suppression of the side lobes.
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
FFT, now even faster benb Everything Else 9 19th January 2012 05:21 AM
FFT Analyzer Radioman62 Parts 0 1st April 2009 08:13 PM
Waterfall windowing problems elvischan917 Multi-Way 3 7th October 2008 09:54 AM
what is an FFT graph? keantoken Everything Else 23 10th September 2006 12:13 AM
FFT Analysis RobPhill33 Everything Else 22 19th March 2003 07:01 PM


New To Site? Need Help?

All times are GMT. The time now is 06:11 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2