Home Theater Forum and Systems banner

21 - 40 of 42 Posts

·
Registered
Joined
·
239 Posts
Timing reference set up is correct. Sub + Left timing is consistent with Just Left Speaker. But Sub alone timing is not. Two more variables to eliminate: First is the Decimate IR setting in Preferences, Analysis tab. Decimation changes sample rate used internally by REW in handling data, based on bandwidth of measurement sweep. So for sweep to 200Hz, REW is doing sample rate conversion from 44100 samples per second to 2756 samples per second. 44100/16 = 2756.25, so REW isn't doing an integer value sample rate conversion. I have not studied how well REW converter works, and eliminating it as a variable is easier.

Yes more details... It will cause files to be slightly larger without the decimation.

The 2nd variable to eliminate is to use same full range sweep for all measurements, this will ensure timing characteristics across all measurements, simplifying analysis.

And, all that really matters is how system sounds to listener(s). Chances are if system doesn't sound right, or clearly isn't performing relative to other listening experiences, then an underlying measurable difference with remedy exists. Typically the smoother a sound systems response, the greater the agreement is that the system sounds good.

I'll look into Anthem correction.

Andrew
 

·
Registered
Joined
·
114 Posts
Discussion Starter #23 (Edited)
Timing reference set up is correct. Sub + Left timing is consistent with Just Left Speaker. But Sub alone timing is not. Two more variables to eliminate: First is the Decimate IR setting in Preferences, Analysis tab. Decimation changes sample rate used internally by REW in handling data, based on bandwidth of measurement sweep. So for sweep to 200Hz, REW is doing sample rate conversion from 44100 samples per second to 2756 samples per second. 44100/16 = 2756.25, so REW isn't doing an integer value sample rate conversion. I have not studied how well REW converter works, and eliminating it as a variable is easier.

Yes more details... It will cause files to be slightly larger without the decimation.

The 2nd variable to eliminate is to use same full range sweep for all measurements, this will ensure timing characteristics across all measurements, simplifying analysis.

And, all that really matters is how system sounds to listener(s). Chances are if system doesn't sound right, or clearly isn't performing relative to other listening experiences, then an underlying measurable difference with remedy exists. Typically the smoother a sound systems response, the greater the agreement is that the system sounds good.

I'll look into Anthem correction.

Andrew
Thanks Andrew. I am attaching preferenes->Analysys. The decimate checkbox is already checked. Next time I will do full range sweep of 0 to 20khz for subwoofer alone , left speaker alone and sub+ left speaker. I will continue researching to understand.
Appreciate taking time to analyzing .mdat files.

Edit: sorry. I guess you want without decimation next time. I will do following test.
decimiation unchecked. Loop back connected right to right and following sweep range
1. sub 0 to 20k
2 . left 0 to 20k
3. sub+left 0 to 20k
 

Attachments

·
Registered
Joined
·
239 Posts
I see now post #22. Levels appear maybe a bit low, in Controls set Plot Responses Normalized. It will make Sub measurements bloated in display, but easy to see.

And yes, with timing reference in place, impulse responses of Left Speaker and Sub + Left start at 47.3ms and Sub impulse response starts at 43.6ms. This is additional time it takes signal to travel from soundcard, through AVR, out speakers, through air as sound, to microphone, and back to soundcard. Speed of sound is roughly 1.13ft/ms, so if microphone is 18ft from speaker, flight time is about 16ms, which means it takes signal 31ms to get through AVR. Much of this is probably related to speaker distances chosen by Anthem correction, operator input, or default settings.

Without experimentation it is difficult to rate how well Anthem correction optimizes system in terms of equalization and speaker distance settings. A baseline is formed by switching all equalization out, setting all distances to zero, setting Sub to full range and zero phase (assuming it has this feature), turning AVR crossover off, and measuring full range response of Sub alone, and Left alone or Right alone. Based off of Left or Right speakers natural low frequency response, and natural response of Sub, it is then possible to select crossover frequency, and if AVR is flexible the type of crossover filter to use.

It is easy to see why AVR manufactures seek automated tuning function. The actual step by step process becomes quite involved....

And to post #23: Yes, decimation unchecked.

1. sub 0 to 20k
2 . left 0 to 20k
3. sub+left 0 to 20k

and again with AVR corrections, equalization, crossover off as described above would be great. Might need to post as two posts due to size of mdat files. I imagine Anthem allows storing settings as presets, which is very useful.

Andrew
 

·
Registered
Joined
·
114 Posts
Discussion Starter #25 (Edited)
Thanks. Andrew. I am still digesting to understand the concepts. I will do more testing as you requested. The impulse graph I complained is beyond the time range. Since I specified -5msec to 50msec, I was not able to see.

I did change time range to -5 5o 150msec and increased level to 100% FS, now I got proper response as attached here. I guess with loop back, I have to change time range.

"Plot response is normalized" was already checked in controls of impulse tab. I see peak at 64.9msec and second peak at 67.9msec. I guess it took 64.9msec for sound to take path for example TASCAM->Receiver->Speaker->air->microphone->sound card
 

Attachments

·
Registered
Joined
·
114 Posts
Discussion Starter #26
I took measurements. I am going to post in two replies. I need to update these problems that I did face cause any incorrect measurement.

1. I have to increase/decrease volume from receiver during measurement. For example, just sub vs sub & left connected etc makes REW complain input levels are low or head room is in red after disconncect/connect combinations of speakers/sub.

2. I did REW SPL calibration one time. after that I did not touch either line out or line in. However, I changed volume of receiver couple of times. Not sure if REW SPL will affect measurement.

3. I don't know if the level is 75db. I just try to make sure input levels ok. No red head room.

Some times head room shows red, but it does not pop up "do you want to cancel". Don't know if this kind of measurement is allowed. I changed receiver volume to make sure head room is not red.
 

·
Registered
Joined
·
114 Posts
Discussion Starter #27
First measurment with decimation unchecked and 20 to 20k

I did this measurement with the following changes.

1. Anthem off
2. phase on sub changed from 90 to 0 (sub is located 2 feet from LP close to back and side wall, facing other side wall)
3. Cross over on sub -bypass (Not changed)
4. distance set to 0 for sub, left, right
5. calibration levels set to 0 for sub, left, right
6. left and right set to full

Attaching two parts
 

Attachments

·
Registered
Joined
·
114 Posts
Discussion Starter #28
I did measure with decimiation unchecked and 20 to 20k. Some how I lost SW and Left +SW. I only have Left measurement. I don't know "remove all" deleted. I thought it only removes from the screen. I am attaching here.
Basically, this is just before anthem off, distances set to 0, phase of SW 0 etc.
 

Attachments

·
Registered
Joined
·
239 Posts
Having established timing reference, it becomes necessary to remove excess delay/phase, bringing peaks back to near zero, in part because REW sets left and right windowing functions relative to t = 0. Procedure is to look at measurement set and select measurement with peak closest to t =0. Viewing the Impulses tab with Controls panel open click Estimate IR Delay. Note the values. Clicking the Shift IR button will apply calculated shift to current measurement. The rest of the measurements need shifting by the same amount. Switch to each remaining measurement and in the “t=0 (ms):” box enter the previously calculated delay and click “apply” once. This shifts all measurements by the same amount, and REW windowing function will work as intended.

Having applied shifts to the “Anthem off” measurements, and the “Left Anthem on” based on “Right only” for calculating shift, it is seen that Anthem Room Correction is creating 17ms additional propagation delay, and adding artifact. Artifact may simply be some form of channel cross talk, or processing artifact of Anthem’s DSP engine. The artifacts under some types of program material may be audible as a pre echo, but hopefully not. This can be problematic when using digital room correction (DRC) techniques.

Visually, comparison of Anthem correction v off suggests that Anthem correction may be effective for equalizing the system. It’s all about how it sounds….. Based on limited measurements, a dip in response exists in 200Hz region. Pinning down absolute source would require yet more measurements. More important perhaps is apparent channel imbalance of Left and Right speakers, both in general amplitude, and in frequency response, assuming that levels were not changed during measurement set. Impulse responses indicate microphone was several inches closer to right speaker. This could also impact 200Hz response dip to some degree, but is more likely do to asymmetry in placement of Left and Right speakers in distance from front wall, and in distances from left wall and right wall; or if left and right front corners of room have large feature differences such as furniture, proximity to doorways or closets. These impacts are greatest upon front speaker sound stage width, depth, detail, and coherence, with music being most affected. Small tweaks in speaker location can have big effects, as can fairly small movements in head position of listeners.

So, now I have some more questions and suggestions.

How far in front of back wall is microphone? What are Left and Right front speaker locations? More important, what are your listening impressions of system, especially with Anthem Correction?

How many measurements, and at what microphone locations, were done for making Anthem Correction?

I tried to reach anthemav.com website without luck this weekend with no luck. I did follow some reviews, which not surprisingly, mostly thumbs up for ARC.

Following THX Ultra II thread, jtalden raised question of user modification of settings when Audyssey is active being limited, or causing Audyssey to disengage when user modified.

This is legitimate concern with Antherm product as well. Given that heavy part of your outlay in system is for Anthem badge, and ARC feature, it make sense to work with these first to find settings that work well for music and movies in your new system.

I recommend completing your room treatment program and furnishing of your theater first.

Then without sub and speakers set full range I would work on tweaking speaker location and listening position for music, aiming for good sound stage depth, imaging detail, and stability of image to small head movements sideways.

Next I would do series of Anthem room correction measurements with Anthem suggested number of measurements and microphone locations for music, and storing the preset.

Following this a series of Anthem room correction measurements with Anthem suggested number of measurements and microphone locations for movies, and store the preset.

My expectation is that ARC allows 5+ measurements for calculating settings. For music, listening is focused on single seat or small area, and series of fairly close microphone placements are likely used. For movies, listening is typically for multiple seats, or a larger area, and a series of more widely spaced microphone placements are likely used.

ARC selection of crossover frequencies in your current set up appears to be 60Hz high pass for fronts, 100Hz high pass for surrounds, and 120Hz low pass for sub. These settings strike me as good movie settings for your speaker array. Initial REW measurements with ARC on support this view. Sub range overlap with fronts looks like it yields bass boost suited to movies.

Given ARC off measurements of you fronts shows your 100’s to be fairly capable in your theater room. For music I would keep 60Hz high pass for fronts, and find way to convince ARC to use 60Hz or 80Hz low pass for sub while getting ARC equalization to work.

In all cases I would set up ARC to perform its corrections from as low a frequency as possible to 5kHz, which I am led to understand is the high frequency max for ARC correction.

In testing the ARC music and movies settings I would suggest a fair amount of listening to wide range of music, and of course movies with LFE, action, music and scenes with multiple people talking. For analytical testing of ARC settings on your system using REW, I recommend full range sweeps, 10Hz to 20kHz length 256k. When assessing low frequency behavior, be it peaks, dips, or apparently really flat, I suggest at least three measurements with a change in microphone placement of 18”-30” that includes change along all three axes, before jumping to conclusions.

If unhappy with ARC, or just curious, I would rerun ARC setup using same microphone placements first, to check repeatability, and then different microphone placements (and or number of measurements) to see what happens.

Once well versed by the above process, if you are still curious or unhappy with results, then go for complete manual alignment and equalization with aid of REW.

I hope you get great enjoyment out of your new theater.

Best regards,

Andrew
 

·
Registered
Joined
·
114 Posts
Discussion Starter #30
Andrew,

Thanks for detailed reply. I am still digesting. I will answer some of the questions now. Both front speakers are around 17 to 18 feet from LP. Not sure if I did not unset distance of Right speaker to 0ft like left one. During measurement, I increased/decreased receiver volume particularly before/after sub is tested alone and with combination. I did got knack on avoiding either headroom is not enough to avoid clipping or input level(I guess sound card) not enough.

Is there any suggestion to avoid changing receiver volume with out getting complaints from software either headroom or input level?

I did not notice any difference with Anthem on/off. It could be my inability to not find difference. In AVS forum everybody is so happy to use their anthem correction. I need to find clip and what they are seeing difference. I took 5 measurements for ARC.

I started putting 4 bass traps today (4 corners from floor to ceiling) and adding two more on the front/black wall on the floor from left side wall to right side wall. Unfortunately, I have front speakers close to corner walls. With new bass traps I am moving little front but very close to side wall. I have lot of bass traps behind it, can add more as well. Don't know if it is problem. Since I did not know all these things at the time of screen selection (120 inch wide, 142 inch diagnol), there is not enough space.

First I am trying to understand IR and still digesting. I will post reply to confirm my understand on IR in next reply. Thanks again for all your help.
 

·
Registered
Joined
·
239 Posts
Can you show me floor plan with seating and speaker layout? For 21 x 15 floor and 18ft distance from front speaker to LP, puts front speakers really close to both side and front wall, and LP really close to back wall. Angles to fronts, and surrounds from LP are far from standards for 5.1 sound mixing, and far from equilateral triangle for stereo. I'm guessing that this is not a projection system?

Andrew
 

·
Registered
Joined
·
114 Posts
Discussion Starter #32 (Edited)
It is projector system. The room is 21 feet long. However at the screen side it is 13 feet 9 inch and width changes suddenly at around 7 feet from front wall to 15 feet 6 inches. Seating is close to back wall (2 or 3 feet). I am attaching photos of home theater. All my test that I did so far does not include any bass traps. One more is thing speakers were not so close to side wall. They were placed diagnose to current bass traps. Now they are moved close to side wall. My friend thinks that is how tweeter is exactly pointing to center seating.

I like this idea, since moving to as side as possible will reduce distraction while watching on the screen.If needed I can place more absorbers.

Thanks for your help Andrew. You can see slide show or see each photo with caption.
https://plus.google.com/u/0/photos/107811778043676490320/albums/5771003652663651985
 

·
Registered
Joined
·
114 Posts
Discussion Starter #33 (Edited)
Andrew,
I am understanding post #29. I did follow the instructions you provided which are listed here.
1. Viewing first impulse on left speaker measurement, clicked control->Estimate delay and wrote down the value (47.915) and did click Shift IR
2. Next for all measurements, clicked control-> set value for t=0 off set (msec) =47.915 and click apply. I did all the steps including second left speaker measure that was choose when anthem correction was there. I see following with Anthem, but all others have peak at t=0.

In the following picture, the peak is happening after 17msec extra time with Anthem. This is what you are referring right? You mentioned following.

Having applied shifts to the “Anthem off” measurements, and the “Left Anthem on” based on “Right only” for calculating shift, it is seen that Anthem Room Correction is creating 17ms additional propagation delay, and adding artifact. Artifact may simply be some form of channel cross talk, or processing artifact of Anthem’s DSP engine. The artifacts under some types of program material may be audible as a pre echo, but hopefully not. This can be problematic when using digital room correction (DRC) techniques.

What is artifact you are referring to? Is it signal before the peak? I some times saw some signal before the peak. Not sure what is this? How did you infer microphone is close of serveral inches from right speaker? Also, how do you see comparison of with anthem vs without showing equalizing the system?
 

Attachments

·
Registered
Joined
·
239 Posts
jtalden:

All DSP is applied by convolution operation, but yes ARC appears to utilize IR with inversion process in generating EQ, with multiple measurement averaging. Without great care in windowing and envelope smoothing of raw inverse, readily apparent, at least visually, pre causal artifacts occur, some are well masked with continuous time convolution with program material. Others are unmasked by certain types of material.

chinni123: post#33, Yes that is result of process I describe. If you display a left/right pair of as Left alone with Right alone or pair with subs, and zoom in close you will see peeks are separated by about 700u (that's 0.7 milliseconds) which works out to about 9.5 inches. Left peak is trailing, so real center line between speakers was about 4.75 inches left of microphone.

Anthem on v Anthem off: My response is based on observations from looking at info from all the mdats thus far, and lots of experience. Yes, some definitive measurements would clarify if my opinion is justified. Anthemav.com was available tonight, so I've read through MRX manual, and ARC usage. I'll cook up a measurement request after some more thought.

Andrew
 

·
Registered
Joined
·
114 Posts
Discussion Starter #36
chinni123: post#33, Yes that is result of process I describe. If you display a left/right pair of as Left alone with Right alone or pair with subs, and zoom in close you will see peeks are separated by about 700u (that's 0.7 milliseconds) which works out to about 9.5 inches. Left peak is trailing, so real center line between speakers was about 4.75 inches left of microphone.
Andrew
Based on limited measurements, a dip in response exists in 200Hz region. Pinning down absolute source would require yet more measurements. More important perhaps is apparent channel imbalance of Left and Right speakers, both in general amplitude, and in frequency response, assuming that levels were not changed during measurement set. Impulse responses indicate microphone was several inches closer to right speaker. This could also impact 200Hz response dip to some degree, but is more likely do to asymmetry in placement of Left and Right speakers in distance from front wall, and in distances from left wall and right wall; or if left and right front corners of room have large feature differences such as furniture, proximity to doorways or closets. These impacts are greatest upon front speaker sound stage width, depth, detail, and coherence, with music being most affected. Small tweaks in speaker location can have big effects, as can fairly small movements in head position of listeners.
Andrew,

I am trying to digest understanding on the above statements for couple of hours. Appreciate pointers to try it myself. Speakers are not perfectly placed to align exactly to microphone. Also, I might have increased receiver volume since what works for speaker may not have worked for subwoofer or combination of them. For example, I got complaint on head room or input level is too low. I need to find way to test them without changing levels next time.

To understand 700 micro second gap between Left and right speaker alone graphs, how do I verify in REW for my education? Since you mentioned time, I assume ETC is the one to compare. Do I need to use overlay?

I generated following graph. I don't know if it is right because I don't see difference of 700 micro seconds. Thanks for your time.
Chinni
 

Attachments

·
Registered
Joined
·
114 Posts
Discussion Starter #37
I want to mention that the microphone is in horizontal position facing front wall all the time. For Anthem correction, I was always pointing up. Don't know if there is any problem placing microphone in horizontal position.
 

·
Registered
Joined
·
239 Posts
Getting peak performance from great stereo recordings demands much from playback system. Linkwitzlab.com is extensive resource covering this topic. Excellent speakers set up in reasonable listening space, that typically doesn't require extensive treatments, will produce sound image where both the speakers, and the room itself disappear. The basic trick is getting the spatial cues contained in the recording into the mind, and adding minimal distractions such as reflections differing in spectral content compared to direct sound. This is where highly symmetrical room and speaker placement come into play.

Anyway, when looking at overlay of left and right speaker impulse responses after offsetting back to t=0 as you did for post #33, zoom in real close and you see right speaker with peak at t=0 and left with peak at 0.749m, which is 749 microseconds (749u):

right 0 left 749u.jpg

Plot results are not normalized, and it is easy to see that peaks are two very different levels. Same conclusion if viewing ETC.

Microphone primarily has different high frequency response when on axis (0 degrees) and pointing straight up (90 degrees). Calibrations for both positions are possible. I don't know about Anthem's microphone. I understand it is USB microphone. When plugged in does it show up in windows as sound device? Can you see it in REW preferences?

Andrew
 

·
Registered
Joined
·
114 Posts
Discussion Starter #39
Getting peak performance from great stereo recordings demands much from playback system. Linkwitzlab.com is extensive resource covering this topic. Excellent speakers set up in reasonable listening space, that typically doesn't require extensive treatments, will produce sound image where both the speakers, and the room itself disappear. The basic trick is getting the spatial cues contained in the recording into the mind, and adding minimal distractions such as reflections differing in spectral content compared to direct sound. This is where highly symmetrical room and speaker placement come into play.

Anyway, when looking at overlay of left and right speaker impulse responses after offsetting back to t=0 as you did for post #33, zoom in real close and you see right speaker with peak at t=0 and left with peak at 0.749m, which is 749 microseconds (749u):

View attachment 37448

Plot results are not normalized, and it is easy to see that peaks are two very different levels. Same conclusion if viewing ETC.

Microphone primarily has different high frequency response when on axis (0 degrees) and pointing straight up (90 degrees). Calibrations for both positions are possible. I don't know about Anthem's microphone. I understand it is USB microphone. When plugged in does it show up in windows as sound device? Can you see it in REW preferences?

Andrew
Andrew, I did use 0 degree calibration file for ECM 8000 microphone. I did not use Anthem microphone since calibration is their propriety.

Finally, after spending lot of time, I am able to verify the time delay around 750 micro. I am attaching graph. when I uncheck normalized, the graph is collapsing to tiny. I am having hard time to scroll left and right to go to the graph starting point.

I see legend in the picture you posted to have 3.0%fs for left and .1%fs for right and %fs to end at 4 on upper side. How you determine the values and apply them. Curious to know. zoom in/out and scrolling to search for 0 time is taking for ever. Thanks again for your time.
 

Attachments

·
Registered
Joined
·
16 Posts
Hi,

Can anybody help me interpret this IR?
I m reading in the manual but cannot say anything about this one, other measurements I do get to understand.

Its a measurement of both speakers and room has treatment
 

Attachments

21 - 40 of 42 Posts
Top