Simplified REW Setup and Use (USB Mic & HDMI Connection) Including Measurement Techniques and How To Interpret Graphs - Page 520 - AVS Forum | Home Theater Discussions And Reviews
Forum Jump: 
 2276Likes
Reply
 
Thread Tools
post #15571 of 23561 Old 08-06-2015, 06:27 PM
AVS Forum Special Member
 
CherylJosie's Avatar
 
Join Date: May 2008
Posts: 1,375
Mentioned: 18 Post(s)
Tagged: 1 Thread(s)
Quoted: 732 Post(s)
Liked: 297
Quote:
Originally Posted by Alan P View Post
Same here, invalid zip file.
https://www.dropbox.com/s/w7h439mppn...%20eq.zip?dl=0
CherylJosie is offline  
Sponsored Links
Advertisement
 
post #15572 of 23561 Old 08-06-2015, 07:22 PM
AVS Forum Special Member
 
CherylJosie's Avatar
 
Join Date: May 2008
Posts: 1,375
Mentioned: 18 Post(s)
Tagged: 1 Thread(s)
Quoted: 732 Post(s)
Liked: 297
Quote:
Originally Posted by AustinJerry View Post
Treatments are designed to control bass ringing in the modal region, below 300 Hz (bass traps), or to control reflections in the spectral region, above 300 Hz (broadband traps),

To assess effectiveness of bass traps in controlling bass ringing, measure left+right+subs (or center+subs), and look at either the waterfall or the spectrogram.

To assess effectiveness of broadband traps in controlling spectral reflections, measure individual speakers and examine the ETC graph.
OK so the 'front audyssey reference' measurement that was actually taken with audyssey off (sorry about the mislabel) is one of the measurements to use. Actually the 'front' plot is less noisy but I forgot to prime the receiver with some noise first so the sound mode switching cut off the initial couple of hertz worth of sweep.

The rest of the traces are all there. I guess I can look them over but the ETC still looks bad to me at first glance especially the center channel and front left/right channels.
Attached Thumbnails
Click image for larger version

Name:	spectrogram fronts plus subs.jpg
Views:	86
Size:	66.8 KB
ID:	873602   Click image for larger version

Name:	etc center.jpg
Views:	80
Size:	108.3 KB
ID:	873666   Click image for larger version

Name:	etc front left.jpg
Views:	64
Size:	104.3 KB
ID:	873674   Click image for larger version

Name:	etc front right.jpg
Views:	57
Size:	99.2 KB
ID:	873682   Click image for larger version

Name:	etc high left.jpg
Views:	48
Size:	104.3 KB
ID:	873690  

Click image for larger version

Name:	etc high left 2.jpg
Views:	47
Size:	102.7 KB
ID:	873698   Click image for larger version

Name:	etc rear left.jpg
Views:	47
Size:	103.0 KB
ID:	873706   Click image for larger version

Name:	etc rear right.jpg
Views:	48
Size:	96.6 KB
ID:	873714   Click image for larger version

Name:	etc side left.jpg
Views:	47
Size:	95.7 KB
ID:	873722   Click image for larger version

Name:	etc side right.jpg
Views:	53
Size:	98.5 KB
ID:	873730  

Click image for larger version

Name:	etc wide left.jpg
Views:	48
Size:	101.3 KB
ID:	873738   Click image for larger version

Name:	etc wide right.jpg
Views:	51
Size:	96.1 KB
ID:	873746  
CherylJosie is offline  
post #15573 of 23561 Old 08-06-2015, 07:34 PM
** Man of Leisure **
 
AustinJerry's Avatar
 
Join Date: Jan 2003
Location: Austin, Texas
Posts: 16,244
Mentioned: 206 Post(s)
Tagged: 1 Thread(s)
Quoted: 7570 Post(s)
Liked: 4228
Quote:
Originally Posted by CherylJosie View Post
OK so the 'front audyssey reference' measurement that was actually taken with audyssey off (sorry about the mislabel) is one of the measurements to use. Actually the 'front' plot is less noisy but I forgot to prime the receiver with some noise first so the sound mode switching cut off the initial couple of hertz worth of sweep.

The rest of the traces are all there. I guess I can look them over but the ETC still looks bad to me at first glance especially the center channel and front left/right channels.
A quick way to analyze the etc is to place the cursor at the -20dB horizontal spot. Then look at any spikes that are above the line, especially in the first 20ms on the horizontal scale. And yes, the center is showing some pretty severe reflections. Reducing reflections improves many aspects of the audio quality, including imaging, sound stage, dialog intelligibility, etc. It's not easy work, but the rewards are significant.
ahblaza and PlasmaPZ80U like this.
AustinJerry is online now  
 
post #15574 of 23561 Old 08-06-2015, 08:03 PM
AVS Forum Special Member
 
CherylJosie's Avatar
 
Join Date: May 2008
Posts: 1,375
Mentioned: 18 Post(s)
Tagged: 1 Thread(s)
Quoted: 732 Post(s)
Liked: 297
Quote:
Originally Posted by AustinJerry View Post
A quick way to analyze the etc is to place the cursor at the -20dB horizontal spot. Then look at any spikes that are above the line, especially in the first 20ms on the horizontal scale. And yes, the center is showing some pretty severe reflections. Reducing reflections improves many aspects of the audio quality, including imaging, sound stage, dialog intelligibility, etc. It's not easy work, but the rewards are significant.
How do I find the distance? All the traces are in time. REW guide says to drag the cursor but that is not working in the Linux build on my Ubuntu 14.04 system.
CherylJosie is offline  
post #15575 of 23561 Old 08-06-2015, 08:53 PM
** Man of Leisure **
 
AustinJerry's Avatar
 
Join Date: Jan 2003
Location: Austin, Texas
Posts: 16,244
Mentioned: 206 Post(s)
Tagged: 1 Thread(s)
Quoted: 7570 Post(s)
Liked: 4228
Quote:
Originally Posted by CherylJosie View Post
How do I find the distance? All the traces are in time. REW guide says to drag the cursor but that is not working in the Linux build on my Ubuntu 14.04 system.
It is reasonably simple. Here is a typical ETC graph:



Place the vertical cursor on a peak you want to analyze, as shown in the screen shot. Read the elapsed time on the horizontal scale (13.39 mille-seconds in the example). The speed of sound is 1,126 ft/sec. Do the math to convert 13.39 ms into distance in feet. I created a simple worksheet that makes the calculations:



The calculation yields a distance of 15.07 ft, which matches the value in REW when you use the CTRL-Right Click method of showing the distance (which is unavailable to you in Linux).

I have placed the simple Excel worksheet in a Zip file that can be downloaded here. Of course, being a Linux user, Excel may not be in your toolbox.

BTW, I wouldn't spend a lot of time generating/analyzing graphs other than Right, Left and Center.
ahblaza and PlasmaPZ80U like this.
AustinJerry is online now  
post #15576 of 23561 Old 08-07-2015, 06:59 AM
AVS Forum Special Member
 
jim19611961's Avatar
 
Join Date: Mar 2011
Location: Texas
Posts: 1,827
Mentioned: 13 Post(s)
Tagged: 0 Thread(s)
Quoted: 449 Post(s)
Liked: 153
Quote:
Originally Posted by CherylJosie View Post
Click image for larger version

Name:	josie GD.jpg
Views:	84
Size:	73.1 KB
ID:	874154

Click image for larger version

Name:	josie phase.jpg
Views:	83
Size:	63.7 KB
ID:	874162

I would focus on getting your phase response flatter and lowering the Group Delay. What both of these graphs are saying is that the time arrivals vs frequency are uneven. The grey plots are yours and the green is close to where you should be.

Most of the rest doesn't look bad

My Room
My Music
Rega - Apollo, Rega - DAC, Goldpoint Passive, Classe CA-100, Crown XLS 1002, miniDSP
Jenzen Next
jim19611961 is offline  
post #15577 of 23561 Old 08-07-2015, 07:30 AM
** Man of Leisure **
 
AustinJerry's Avatar
 
Join Date: Jan 2003
Location: Austin, Texas
Posts: 16,244
Mentioned: 206 Post(s)
Tagged: 1 Thread(s)
Quoted: 7570 Post(s)
Liked: 4228
Quote:
Originally Posted by CherylJosie View Post
How do I find the distance? All the traces are in time. REW guide says to drag the cursor but that is not working in the Linux build on my Ubuntu 14.04 system.
I downloaded your MDAT file and agree with Jim's comments. A couple of additional comments on REW itself:

To identify your measurements, I recommend using the box where the date is displayed by default. Then the graph description appears in the legend, making it much easier to select the measurement you want to look at.

Your way:



My recommendation:



Which results in the measurement description appearing in the legend:




Second recommendation (and there are differing opinions on this), I would measure at a slightly higher output level so that you are sure to have ~40dB above the noise floor. JMO.
ahblaza likes this.
AustinJerry is online now  
post #15578 of 23561 Old 08-07-2015, 07:49 AM
AVS Forum Special Member
 
3ll3d00d's Avatar
 
Join Date: Sep 2007
Location: London, UK
Posts: 2,898
Mentioned: 100 Post(s)
Tagged: 0 Thread(s)
Quoted: 1653 Post(s)
Liked: 632
Quote:
Originally Posted by jim19611961 View Post
Attachment 874154

Attachment 874162

I would focus on getting your phase response flatter and lowering the Group Delay. What both of these graphs are saying is that the time arrivals vs frequency are uneven. The grey plots are yours and the green is close to where you should be.
which measurement is that? what configuration did you apply to produce that phase response? it looks quite odd
3ll3d00d is online now  
post #15579 of 23561 Old 08-07-2015, 08:09 AM
AVS Forum Special Member
 
jim19611961's Avatar
 
Join Date: Mar 2011
Location: Texas
Posts: 1,827
Mentioned: 13 Post(s)
Tagged: 0 Thread(s)
Quoted: 449 Post(s)
Liked: 153
Quote:
Originally Posted by 3ll3d00d View Post
which measurement is that? what configuration did you apply to produce that phase response? it looks quite odd
Left Front.

No special phase configuration. Just the overlay phase with the limiters (+1800 to -1800 degrees) to near maximum. > 4000 degrees swing from the highs to lows is not unheard of but certainly not good.. I have seen this several times. Mostly, I just dont think many look at phase, so it jumps out when it is shown.

The step response is another indicator of timing. Decided not to open that can of worms just yet.

What I have noticed is that rooms where reflections are not properly controlled, the reflections themselves compromise the phase response. That is, a 10ms 1k strong reflection will cause the phase response at that frequency to jump way up. Generally, a flat phase response is only realized when all early reflections (not what a full range ETC shows) are below -20db (-30db even better) and the speaker itself has a well designed XO.

I think most would be shocked at what their phase curve looks like

My Room
My Music
Rega - Apollo, Rega - DAC, Goldpoint Passive, Classe CA-100, Crown XLS 1002, miniDSP
Jenzen Next

Last edited by jim19611961; 08-07-2015 at 08:26 AM.
jim19611961 is offline  
post #15580 of 23561 Old 08-07-2015, 08:11 AM
AVS Forum Special Member
 
LastButNotLeast's Avatar
 
Join Date: Feb 2007
Location: 08077
Posts: 8,331
Mentioned: 35 Post(s)
Tagged: 1 Thread(s)
Quoted: 1793 Post(s)
Liked: 1404
So how does one "fix" it?

Did you really need to quote that entire post in your reply?
Downloadable FREE demo discs: Demonstration Blu-Ray Discs (Independently Authored)
Welcome to AVS - Get out while you still can!
Don't guess, measure: Getting Started With REW: A Step-by-Step Guide
LastButNotLeast is offline  
post #15581 of 23561 Old 08-07-2015, 08:20 AM
AVS Forum Special Member
 
jim19611961's Avatar
 
Join Date: Mar 2011
Location: Texas
Posts: 1,827
Mentioned: 13 Post(s)
Tagged: 0 Thread(s)
Quoted: 449 Post(s)
Liked: 153
Quote:
Originally Posted by LastButNotLeast View Post
So how does one "fix" it?
Assuming your speaker XO is well designed, properly tame early reflections. Late reflections will skew the phase response also, but you can window the IR so their influence doesn't affect the phase plot.

Remember, phase is indicative of timing. So strong delayed reflections will affect the perception and measurement of time arrival (phase, step and GD).

My Room
My Music
Rega - Apollo, Rega - DAC, Goldpoint Passive, Classe CA-100, Crown XLS 1002, miniDSP
Jenzen Next

Last edited by jim19611961; 08-07-2015 at 08:27 AM.
jim19611961 is offline  
post #15582 of 23561 Old 08-07-2015, 08:36 AM
AVS Forum Special Member
 
3ll3d00d's Avatar
 
Join Date: Sep 2007
Location: London, UK
Posts: 2,898
Mentioned: 100 Post(s)
Tagged: 0 Thread(s)
Quoted: 1653 Post(s)
Liked: 632
Quote:
Originally Posted by jim19611961 View Post
Left Front.

No special phase configuration. Just the overlay phase with the limiters (+1800 to -1800 degrees) to near maximum. > 4000 degrees swing from the highs to lows is not unheard of but certainly not good.. I have seen this several times. Mostly, I just dont think many look at phase, so it jumps out when it is shown.

The step response is another indicator of timing. Decided not to open that can of worms just yet.

What I have noticed is that rooms where reflections are not properly controlled, the reflections themselves compromise the phase response. That is, a 10ms 1k strong reflection will cause the phase response at that frequency to jump way up. Generally, a flat phase response is only realized when all early reflections (not what a full range ETC shows) are below -20db (-30db even better) and the speaker itself has a well designed XO.

I think most would be shocked at what their phase curve looks like
An ungated view on the phase response from an in room measurement is useless, simply the wrong tool for the job. All it shows is a load of wraps that are really reflections.
3ll3d00d is online now  
post #15583 of 23561 Old 08-07-2015, 08:51 AM
AVS Forum Special Member
 
jim19611961's Avatar
 
Join Date: Mar 2011
Location: Texas
Posts: 1,827
Mentioned: 13 Post(s)
Tagged: 0 Thread(s)
Quoted: 449 Post(s)
Liked: 153
Quote:
Originally Posted by 3ll3d00d View Post
An ungated view on the phase response from an in room measurement is useless, simply the wrong tool for the job.
I agree that different views show different things. Not useless. Just one way to see the data.


Quote:
Originally Posted by 3ll3d00d View Post
All it shows is a load of wraps that are really reflections.
Mine doesn't look that way (ungated room response)

Click image for larger version

Name:	my phase.jpg
Views:	72
Size:	65.2 KB
ID:	874362

My Room
My Music
Rega - Apollo, Rega - DAC, Goldpoint Passive, Classe CA-100, Crown XLS 1002, miniDSP
Jenzen Next

Last edited by jim19611961; 08-07-2015 at 08:58 AM.
jim19611961 is offline  
post #15584 of 23561 Old 08-07-2015, 09:49 AM
AVS Forum Special Member
 
3ll3d00d's Avatar
 
Join Date: Sep 2007
Location: London, UK
Posts: 2,898
Mentioned: 100 Post(s)
Tagged: 0 Thread(s)
Quoted: 1653 Post(s)
Liked: 632
Quote:
Originally Posted by jim19611961 View Post
I agree that different views show different things. Not useless. Just one way to see the data.




Mine doesn't look that way (ungated room response)

Attachment 874362
It is a useless view for analysing reflections, ungated (or rather, using the default rew configuration of a 500ms right window) it if useless for analysing the phase response of the speaker and any excess phase induced by the room or other factors.

Yours looks as it does because you have a v well controlled room. You can just show that more clearly via your etc.
3ll3d00d is online now  
post #15585 of 23561 Old 08-07-2015, 10:03 AM
AVS Forum Special Member
 
jim19611961's Avatar
 
Join Date: Mar 2011
Location: Texas
Posts: 1,827
Mentioned: 13 Post(s)
Tagged: 0 Thread(s)
Quoted: 449 Post(s)
Liked: 153
Quote:
Originally Posted by 3ll3d00d View Post
It is a useless view for analysing reflections, ungated (or rather, using the default rew configuration of a 500ms right window) it if useless for analysing the phase response of the speaker and any excess phase induced by the room or other factors.

Yours looks as it does because you have a v well controlled room. You can just show that more clearly via your etc.
I would say that a very well controlled room is the goal for many here. The ungated phase response exposes rooms that are not well controlled and that is useful. Yes, ETC will expose this also when utilized properly. The default full range ETC is useless as well in exposing midrange reflections.

No one view tells everything accurately.

My Room
My Music
Rega - Apollo, Rega - DAC, Goldpoint Passive, Classe CA-100, Crown XLS 1002, miniDSP
Jenzen Next
jim19611961 is offline  
post #15586 of 23561 Old 08-07-2015, 10:06 AM
AVS Forum Special Member
 
3ll3d00d's Avatar
 
Join Date: Sep 2007
Location: London, UK
Posts: 2,898
Mentioned: 100 Post(s)
Tagged: 0 Thread(s)
Quoted: 1653 Post(s)
Liked: 632
Well all I can say is I can't think of a less useful view on room reflections then the phase response. Your mileage may of course vary
3ll3d00d is online now  
post #15587 of 23561 Old 08-07-2015, 10:31 AM
AVS Forum Special Member
 
jim19611961's Avatar
 
Join Date: Mar 2011
Location: Texas
Posts: 1,827
Mentioned: 13 Post(s)
Tagged: 0 Thread(s)
Quoted: 449 Post(s)
Liked: 153
Quote:
Originally Posted by 3ll3d00d View Post
Well all I can say is I can't think of a less useful view on room reflections then the phase response. Your mileage may of course vary
I wasnt specifically trying to use the phase response to expose room reflections. It just so happens that reflections are among the main causes for a skewed ungated phase response.

Spectrograms, Group Delay, Phase Response and ETC all show the effects of reflections, but in different ways. Which is the more useful depends on what specifically your trying to address. But one of the things that I really like about measurement data is that all the various views tell an overlapping story that expose room issues.

My Room
My Music
Rega - Apollo, Rega - DAC, Goldpoint Passive, Classe CA-100, Crown XLS 1002, miniDSP
Jenzen Next

Last edited by jim19611961; 08-07-2015 at 10:45 AM.
jim19611961 is offline  
post #15588 of 23561 Old 08-07-2015, 12:30 PM
AVS Forum Special Member
 
CherylJosie's Avatar
 
Join Date: May 2008
Posts: 1,375
Mentioned: 18 Post(s)
Tagged: 1 Thread(s)
Quoted: 732 Post(s)
Liked: 297
Quote:
Originally Posted by AustinJerry View Post
Of course, being a Linux user, Excel may not be in your toolbox.
Thanks. Works just fine in LibreOffice Calc.
CherylJosie is offline  
post #15589 of 23561 Old 08-07-2015, 12:40 PM
AVS Forum Special Member
 
CherylJosie's Avatar
 
Join Date: May 2008
Posts: 1,375
Mentioned: 18 Post(s)
Tagged: 1 Thread(s)
Quoted: 732 Post(s)
Liked: 297
Quote:
Originally Posted by AustinJerry View Post
To identify your measurements, I recommend using the box where the date is displayed by default.
Thanks, nice tip.

Quote:
Second recommendation (and there are differing opinions on this), I would measure at a slightly higher output level so that you are sure to have ~40dB above the noise floor. JMO.
Reserving higher amplitude measurements for once I understand things better and fix as many obvious problems as I can. Neighbors are gonna throw a fit if I keep torturing them with the whoops.
CherylJosie is offline  
post #15590 of 23561 Old 08-07-2015, 01:53 PM
AVS Forum Special Member
 
3ll3d00d's Avatar
 
Join Date: Sep 2007
Location: London, UK
Posts: 2,898
Mentioned: 100 Post(s)
Tagged: 0 Thread(s)
Quoted: 1653 Post(s)
Liked: 632
Quote:
Originally Posted by jim19611961 View Post
I wasnt specifically trying to use the phase response to expose room reflections. It just so happens that reflections are among the main causes for a skewed ungated phase response.
I know you didn't mean to show specifically but there is no other information content in that graph. Nothing else is going to produce 3600 degrees of phase rotation except for this sort of v wide window on an in room impulse.

If people want to learn how to read this view and learn how to window a response then great. This might then tell them something about timing.
3ll3d00d is online now  
post #15591 of 23561 Old 08-07-2015, 02:09 PM
AVS Forum Special Member
 
CherylJosie's Avatar
 
Join Date: May 2008
Posts: 1,375
Mentioned: 18 Post(s)
Tagged: 1 Thread(s)
Quoted: 732 Post(s)
Liked: 297
Quote:
Originally Posted by 3ll3d00d View Post
which measurement is that? what configuration did you apply to produce that phase response? it looks quite odd
Every single phase response of my REW measurements looks like wormy snake thingy. I was meaning to ask why since my understanding of phase of a speaker does not include phase that wraps indefinitely toward increasing lag with monotonically changing frequency. For a 3-way system the phase should only go through at most 2 inversions due to crossover and then the rest is just rolloff at the extremes of extension. So maybe 720 degrees total phase over the full range?

My front left measurement is showing over 1000 degrees in the first 400Hz.

I wonder if the calibration (or lack thereof in Linux) is part of the problem. Accurate phase response depends on accurate delay calibration. Does REW calibrate for phase when it calibrates via SPL meter? I have no SPL meter. The only calibration I have done is:

  • load microphone 90 degree narrow band cal file
    (the USB microphone is not auto-detected in Linux)
    (the microphone cal data can be plotted)
    (the sensitivity factor cannot be plotted and does not eliminate need for SPL cal when taking measurements)
  • (first attempt at REW calibration - I used this method first because I have no SPL meter)
    start TX-NR929 auto setup and select the subwoofer cal 75dB signal
    (I adjusted both subs to 75dB independently so the receiver 'sees' 80dB with both on)
    start the REW SPL meter and push 'calibrate'
    select external cal signal
    received level wanders around +/-1.5dB at least
    set SPL meter calibration to 80dB
    watch the REW SPL meter and re-cal if it is not visually 'averaging' 80dB
    ignore the time delay through REW?
  • (bridge between subwoofer and speaker method)
    first, cal with the prior TX-NR929 auto setup subwoofer cal 75dB signal
    front in 'direct' sound mode
    set receiver output to -20dB
    start the REW SPL meter and push 'calibrate'
    select speaker cal signal
    REW reports approximately 80dB from the fronts
    received level is very stable and repeatable
    verifies that using the mains in direct mode to calibrate might be more repeatable
    test signal originates from REW and concievably has some sort of delay reference?
  • (current calibration method)
    front in 'direct' sound mode
    set receiver output to -20dB
    start the REW SPL meter and push 'calibrate'
    select speaker cal signal
    set SPL calibration to 80dB
    pray that the mag and phase through REW is now calibrated?
I saw nothing in the REW guide that indicates how to get accurate delay calibration. Loopback cannot do it because the microphone and air is out of the loop. Either REW susses it out magically from the received sweep data, or it does something else I am unable to imagine.

How will it know what frequency to use as the timing reference? Does it just assume something in the 50Hz range? How does it know it did not pick a really strange point in the sweep to base its delay on? If the timing reference is off, all the measured phase information will also be off.

Did I miss something? How does REW calibrate itself for delay/phase?

Quote:
Originally Posted by jim19611961 View Post
Left Front.

No special phase configuration. Just the overlay phase with the limiters (+1800 to -1800 degrees) to near maximum. > 4000 degrees swing from the highs to lows is not unheard of but certainly not good.. I have seen this several times. Mostly, I just dont think many look at phase, so it jumps out when it is shown.
I noticed it and shrugged. I just assumed that REW was not calibrating itself for phase and the phase info was nearly meaningless.

Quote:
The step response is another indicator of timing. Decided not to open that can of worms just yet.
Probably wise. One step at a time here or your helping hand may drag me face down.

Quote:
What I have noticed is that rooms where reflections are not properly controlled, the reflections themselves compromise the phase response. That is, a 10ms 1k strong reflection will cause the phase response at that frequency to jump way up. Generally, a flat phase response is only realized when all early reflections (not what a full range ETC shows) are below -20db (-30db even better) and the speaker itself has a well designed XO.

I think most would be shocked at what their phase curve looks like
Not me. I am well aware of phase issues (but not necessarily all that conversant -- it has been a while since I did any analog). I am not so familiar with REW, speakers, and acoustics though. What I do know makes me question why the phase response I measured has so many wraps. It is not bouncing back and forth, it just keeps accumulating and to me that looks like a timing calibration issue with REW where an uncalibrated delay error looks like perpetually increasing phase as frequency increases. I can think of no physical mechanism that would cause a speaker/room to accumulate so much phase.

Quote:
Originally Posted by jim19611961 View Post
Assuming your speaker XO is well designed
HA! These speakers were originally designed by Cary Christie (cofounder of Infinity) and sold with some help from Tom DeVesto of Cambridge Audio as a budget-minded pseudo 'house brand' at Tweeter Etc. and related chains. The drivers, cabinets, and crossovers all fit my own financial model -- cheap but not quite junk, pushed to the limits of their performance for the best 'bang for buck'.

Then I bought them at nearly 90% discount used, cut/drilled new outriggers from metal bar, put black marker on all the gashes and scrapes, and pieced the grills back together with superglue/paper bandages wrapped around shattered plastic.

The grill for the center speaker is flimsy and buzzes. The plastic basket of the front left lower midrange also buzzes (it is slightly warped or sticking out and not flush to the baffle on the bottom, and there are only 4 mounting screws). I have seen no impact from that buzzing in the measurements so far but I am sure it is there.

Some day I will address that but first I need to repair the crossover in the right rear tower -- that is the tower that was knocked over and had MTM motors sheared off. It now has the drivers from my original shortened TSC center speaker in it and I got a second center speaker of the original Sapphire (not rebranded 'The Speaker Company") design. It appears that the woofers are reproducing too much midrange or something so I suspect that a capacitor sheared off or a PCB trace broke from the impact when the drivers got damaged. The cal level comes out 3dB lower than the rest of the towers so the receiver is really backing down the output and the frequency response looks a little bit off too.

I also have to glue the grill for the high left back together. I pretty well shattered that one myself. It was in perfect condition when it arrived.

So when considering these plots please take into account that the speakers themselves are nothing to brag about. They do the job for me but the need for lots of EQ is probably a good indicator of the speaker performance as much as it is an indicator of the room that is also nothing to write home about nor are the treatments.

Goal here is just to maximize what I have using as much tech as I own while adding as little incremental cost as possible. Future room treatments will all be more DIY and probably also using found materials recycled into absorbers or resonators or diffusers, or at least inexpensive materials that can do the job adequately if not optimally and with as little labor input and tooling as possible. My nickname for this system is the 'Gutter Budget 11.1 Home Theater'.

All that said, I still do not understand the phase plots. Something seems way off.

Quote:
properly tame early reflections. Late reflections will skew the phase response also, but you can window the IR so their influence doesn't affect the phase plot.

Remember, phase is indicative of timing. So strong delayed reflections will affect the perception and measurement of time arrival (phase, step and GD).
Yes but none of these effects would explain phase that continually wraps. It could explain large local deviations but while the phase will push backward below resonance and forward above resonance, it will not wrap around and keep going from reflections. There should be wide ranges of frequencies that have relatively stable phase in the middle of each driver's operating range but none of my plots show that.

Quote:
Originally Posted by 3ll3d00d View Post
An ungated view on the phase response from an in room measurement is useless, simply the wrong tool for the job. All it shows is a load of wraps that are really reflections.
Quote:
Originally Posted by jim19611961 View Post
I agree that different views show different things. Not useless. Just one way to see the data.

Mine doesn't look that way (ungated room response)

Attachment 874362
So how exactly are you configured with REW? What microphone? Did you loopback to cal your sound card/receiver? Anything special to calibrate for phase? Analog or digital output from your PC?

I ended up using SPDIF and changing to Full Mono with most of the channels disconnected/disabled/full range (to keep the subwoofer output generated from a single pair with 6dB cut applied to the sub output in the receiver when measuring a single speaker) because with the USB mic there is no way to analog loopback for cal anyway. It just made things more consistent and less error-prone for me than using analog output of unknown fidelity though the EMU 1212m should be pretty good but Linux driver does not give me good control over its advanced functions like sample rate, word length, and level (diff at +4dB versus single at -10dB).

Quote:
Originally Posted by 3ll3d00d View Post
It is a useless view for analysing reflections, ungated (or rather, using the default rew configuration of a 500ms right window) it if useless for analysing the phase response of the speaker and any excess phase induced by the room or other factors.

Yours looks as it does because you have a v well controlled room. You can just show that more clearly via your etc.
Quote:
Originally Posted by jim19611961 View Post
I would say that a very well controlled room is the goal for many here. The ungated phase response exposes rooms that are not well controlled and that is useful. Yes, ETC will expose this also when utilized properly. The default full range ETC is useless as well in exposing midrange reflections.

No one view tells everything accurately.
Quote:
Originally Posted by 3ll3d00d View Post
Well all I can say is I can't think of a less useful view on room reflections then the phase response. Your mileage may of course vary
Quote:
Originally Posted by jim19611961 View Post
I wasnt specifically trying to use the phase response to expose room reflections. It just so happens that reflections are among the main causes for a skewed ungated phase response.

Spectrograms, Group Delay, Phase Response and ETC all show the effects of reflections, but in different ways. Which is the more useful depends on what specifically your trying to address. But one of the things that I really like about measurement data is that all the various views tell an overlapping story that expose room issues.
This sort of info is exactly what is lacking in my background. Acoustics is foreign territory to me.

These are all nice and maybe valid too but back to my main question: how does REW calibrate itself for phase? I found nothing in any of the documentation that indicates how it works.

The phase response in my plots seems totally out to lunch, with the extreme phase accumulation with frequency not explainable in my paradigm by anything other than uncalibrated delay in the measurement tool.

Maybe this is another limitation of the Linux version.

Opinions?
CherylJosie is offline  
post #15592 of 23561 Old 08-07-2015, 02:22 PM
AVS Forum Special Member
 
jim19611961's Avatar
 
Join Date: Mar 2011
Location: Texas
Posts: 1,827
Mentioned: 13 Post(s)
Tagged: 0 Thread(s)
Quoted: 449 Post(s)
Liked: 153
Quote:
Originally Posted by 3ll3d00d View Post
I know you didn't mean to show specifically but there is no other information content in that graph. Nothing else is going to produce 3600 degrees of phase rotation except for this sort of v wide window on an in room impulse.

If people want to learn how to read this view and learn how to window a response then great. This might then tell them something about timing.
The novice REW user is probably going to use defaults to look at everything initially. So for those viewing, now you know why your phase graph looks "out to lunch" as CJ put it

Your welcome to provide the instructional here on phase graphs.

My Room
My Music
Rega - Apollo, Rega - DAC, Goldpoint Passive, Classe CA-100, Crown XLS 1002, miniDSP
Jenzen Next

Last edited by jim19611961; 08-07-2015 at 02:51 PM.
jim19611961 is offline  
post #15593 of 23561 Old 08-07-2015, 02:45 PM
AVS Forum Special Member
 
jim19611961's Avatar
 
Join Date: Mar 2011
Location: Texas
Posts: 1,827
Mentioned: 13 Post(s)
Tagged: 0 Thread(s)
Quoted: 449 Post(s)
Liked: 153
Quote:
Originally Posted by CherylJosie View Post
All that said, I still do not understand the phase plots. Something seems way off.
Why my phase plot looks different from yours has already been explained. I have eliminated ALL relevant early reflections and you and most folks here have not. Its not a shortcoming of your setup (probably).

So in reality, your phase plot is not way off. Using the default phase settings, your simply picking up all the delayed energy (reflections) and REW is translating that into delayed timing. If you took your setup outside where no reflections exist, then you would have a much flatter looking curve.

You could do as my opponent (just kidding) suggested and restrict the IR window. But that has a drawback as well in that as you truncate the timing, you loose definition, especially in the bass region. Then the bass region phase info becomes invalid (if you squeeze the IR window down too much).


Quote:
Originally Posted by CherylJosie View Post
1) What microphone? 2) Did you loopback to cal your sound card/receiver? 3) Anything special to calibrate for phase? 4) Analog or digital output from your PC?
1) Umik USB mic. 2) No loopback is possible via USB. 3) No. 4) Digital

My Room
My Music
Rega - Apollo, Rega - DAC, Goldpoint Passive, Classe CA-100, Crown XLS 1002, miniDSP
Jenzen Next
jim19611961 is offline  
post #15594 of 23561 Old 08-07-2015, 02:52 PM
AVS Forum Special Member
 
3ll3d00d's Avatar
 
Join Date: Sep 2007
Location: London, UK
Posts: 2,898
Mentioned: 100 Post(s)
Tagged: 0 Thread(s)
Quoted: 1653 Post(s)
Liked: 632
Quote:
Originally Posted by CherylJosie View Post
Every single phase response of my REW measurements looks like wormy snake thingy. I was meaning to ask why since my understanding of phase of a speaker does not include phase that wraps indefinitely toward increasing lag with monotonically changing frequency. For a 3-way system the phase should only go through at most 2 inversions due to crossover and then the rest is just rolloff at the extremes of extension. So maybe 720 degrees total phase over the full range?

My front left measurement is showing over 1000 degrees in the first 400Hz.
Basically it is the noise (sound bouncing around the room and recorded by the mic) in the tail of the impulse that causes this. It is not real.

Quote:
Originally Posted by CherylJosie View Post
I saw nothing in the REW guide that indicates how to get accurate delay calibration. Loopback cannot do it because the microphone and air is out of the loop. Either REW susses it out magically from the received sweep data, or it does something else I am unable to imagine.

How will it know what frequency to use as the timing reference? Does it just assume something in the 50Hz range? How does it know it did not pick a really strange point in the sweep to base its delay on? If the timing reference is off, all the measured phase information will also be off.
It looks for the impulse peak and treats this as the timing ref point, you control the default behaviour via http://www.roomeqwizard.com/help/hel.../analysis.html

If you want an absolute time measurement then you need a loopback and configure REW to use it. This sends the same impulse via the selected channels and REW then compares e recorded impulse peak against the reference impulse peak to determine the delay. You can use a software loopback here BTW, it just needs the original impulse to come back in on the nominated timing input channel. The only real rule is that the impulses must have similar frequency content otherwise the shape of the impulse is v different so peak alignment will fail to give accurate data. Note this means you can use it to measure software delays as well.

REW also uses the peak to estimate the delay and can remove this in order to remove the delay induced excess phase. You may find that manually moving the impulse to bring T=0 to the initial rise gives more accurate results. There is some degree of interpretation required here to get the "right" answer.

Quote:
Originally Posted by CherylJosie View Post
Maybe this is another limitation of the Linux version.

Opinions?
I have used both Linux and Windows versions for 10 years now. I haven't noticed any limitations in the Linux version, do you have something specific that doesn't work?
3ll3d00d is online now  
post #15595 of 23561 Old 08-07-2015, 02:59 PM
AVS Forum Special Member
 
3ll3d00d's Avatar
 
Join Date: Sep 2007
Location: London, UK
Posts: 2,898
Mentioned: 100 Post(s)
Tagged: 0 Thread(s)
Quoted: 1653 Post(s)
Liked: 632
Quote:
Originally Posted by jim19611961 View Post

You could do as my opponent (just kidding) suggested and restrict the IR window. But that has a drawback as well in that as you truncate the timing, you loose definition, especially in the bass region. Then the bass region phase info becomes invalid (if you squeeze the IR window down too much).
FWIW there is a beta out there with an option for frequency dependent windowing. This gets rew closer to how we hear and so is a nice addition (removes one reason to use Holm anyway).
3ll3d00d is online now  
post #15596 of 23561 Old 08-07-2015, 03:05 PM
AVS Forum Special Member
 
jim19611961's Avatar
 
Join Date: Mar 2011
Location: Texas
Posts: 1,827
Mentioned: 13 Post(s)
Tagged: 0 Thread(s)
Quoted: 449 Post(s)
Liked: 153
Quote:
Originally Posted by 3ll3d00d View Post
Basically it is the noise (sound bouncing around the room and recorded by the mic) in the tail of the impulse that causes this. It is not real.
True as it relates to the phase of the direct response. But false as it relates psycho-acoustically to our brains integration period (up to 8-20ms depending on who you read).

My Room
My Music
Rega - Apollo, Rega - DAC, Goldpoint Passive, Classe CA-100, Crown XLS 1002, miniDSP
Jenzen Next
jim19611961 is offline  
post #15597 of 23561 Old 08-07-2015, 03:09 PM
AVS Forum Special Member
 
3ll3d00d's Avatar
 
Join Date: Sep 2007
Location: London, UK
Posts: 2,898
Mentioned: 100 Post(s)
Tagged: 0 Thread(s)
Quoted: 1653 Post(s)
Liked: 632
Quote:
Originally Posted by jim19611961 View Post
True as it relates to the phase of the direct response. But false as it relates psycho-acoustically to our brains integration period (up to 8-20ms depending on who you read).
Or even less than that at high frequencies (depending on who you read ) hence the utility of fdw
3ll3d00d is online now  
post #15598 of 23561 Old 08-07-2015, 03:16 PM
AVS Forum Special Member
 
jim19611961's Avatar
 
Join Date: Mar 2011
Location: Texas
Posts: 1,827
Mentioned: 13 Post(s)
Tagged: 0 Thread(s)
Quoted: 449 Post(s)
Liked: 153
Quote:
Originally Posted by 3ll3d00d View Post
FWIW there is a beta out there with an option for frequency dependent windowing. This gets rew closer to how we hear and so is a nice addition (removes one reason to use Holm anyway).
Are you referring to the Psycho-Acoustic/ERB smoothing features?

My Room
My Music
Rega - Apollo, Rega - DAC, Goldpoint Passive, Classe CA-100, Crown XLS 1002, miniDSP
Jenzen Next
jim19611961 is offline  
post #15599 of 23561 Old 08-07-2015, 03:22 PM
AVS Forum Special Member
 
3ll3d00d's Avatar
 
Join Date: Sep 2007
Location: London, UK
Posts: 2,898
Mentioned: 100 Post(s)
Tagged: 0 Thread(s)
Quoted: 1653 Post(s)
Liked: 632
Quote:
Originally Posted by jim19611961 View Post
Are you referring to the Psycho-Acoustic/ERB smoothing features?
I don't recall what name he gives it - http://www.hometheatershack.com/foru...8/topics/99673

It is an implementation of fdw though, possibly with some additional smoothing to produce a better resulting interpretation of the fats.
3ll3d00d is online now  
post #15600 of 23561 Old 08-07-2015, 04:22 PM
AVS Forum Special Member
 
CherylJosie's Avatar
 
Join Date: May 2008
Posts: 1,375
Mentioned: 18 Post(s)
Tagged: 1 Thread(s)
Quoted: 732 Post(s)
Liked: 297
Quote:
Originally Posted by jim19611961 View Post
1) Umik USB mic. 2) No loopback is possible via USB. 3) No. 4) Digital
Well that pretty much covers it then. Same measurement setup as me.

How do you calibrate SPL? REW speaker or subwoofer signal, externl signal, or auto detect the sensitivity factor from mic cal file?
CherylJosie is offline  
Sponsored Links
Advertisement
 
Reply Audio Theory, Setup, and Chat

Tags
Dayton , Dayton Audio , Room Equilizer Wizard Rew

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

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