[NegativeEntropy]

Hello,

In http://www.hometheatershack.com/forums/rew-forum/51192-recommended-hardware-rew-w-win-7-64-bit-family-room-setup.html I asked about hardware. I ended up with the Tascam US122MKII and a Dayton mic, calibrated by Cross Spectrum labs. I also picked up an inexpensive mic stand and clip.

The setup has worked well (insofar as it has behaved as expected).

The room in question is setup as described in this early thread here, with the sub in the corner location shown. Please note the asymmetry. Major changes are a no go (e.g. moving the main speakers to another wall); small changes (exact speaker, seating locations) are a possibility.

Mains are Ascend Sierra Towers, sub is Rythmik FV12.

I experimented with sub position by placing it in the main listening position, then running sweeps with the microphone at viable listening positions. The best one ended up being the corner, with the speaker facing toward the far side of the room.

The main speakers are ~3' from the wall behind them. The room itself, including next to the main speakers, is asymmetrical as shown in the linked thread.

I chose the crossover frequency by running sweeps and varying the crossover from 40-80 Hz. 70 Hz gave the smoothest frequency response graph. I did the same with sub phase after running audyssey, 180 degrees ended up being the setting that produced the smoothest graph.

I tried to evaluate my results based on this paper.

Everything looked pretty good except for the very low frequency decay (<35Hz) and potentially for a "peaks smooth in pattern density" in the ETC. Unfortunately my ignorance is vast here, even after reading here and elsewhere for some time.

Frequency response of both individual left, right and sub is good post Audyssey, as well as the combined result which measures +/- 4dB from 17Hz to 24kHz with 1/3 octave smoothing when I trim the sub for a flat response (my standard setup includes a bit of a house curve, at least as much as can be accomplished w/o an outboard EQ).

Inserted below are the ETC results for the left main, the first showing 40 ms after peak, the second 500 ms. This is measurement sweep #1 in the attached REW file.

Next is the same for the right main. These are sweep #2 in the attached REW file.

Next is the frequency response of the sub + both mains, smoothed to 1/3 octave. This is measurement #3 in the REW file. This shows a ~10 dB house curve.

Last is the low frequency waterfall for the same sub + both mains measurement. It exhibits significant ringing at 20 and 26 Hz.

I welcome any suggestions on both interpretation as well as reasonable measures to improve things. I think it sounds good today, but, for example, if an outboard EQ or wife approved room treatments would help appreciably, they are a potential option.

Thanks!

 

[SAC]

A few comments about process.

First, this is an iterative process, not an all in one do everything at the same time operation.


First, the crossover issues is to resolved and corrected. It is not some step in the room correction process.
And a NECESSARY part of the crossover configuration is the signal alignment of the sources.

After that is done, then focus on the modal behavior.
And please loose the FR smoothing. We are not trying to sell this or to make things 'look' nice. Instead we WANT to know the detailed issues that are obscured by smoothing! (Sorry, not angry with you, but seeing this predilection with smoothing remain an almost universal fixation with folks who claim that they want to identify and resolve problems in the process of obscuring them grows old...)

First map the space. The mapping of the region can be done via waterfalls, or if you are not comfortable with this, perhaps more easily by the playing stepped tones (in REW) in order identify modes and to then identify the location of regional peaks and nulls. Assuming L/R symmetry, explore the region about the desired listening position, both forward and backward from the current position, in order to determine the location least subject primarily to modal nulls, and then to a lesser degree to peaks. A problem avoided by relocation is much easier than attempting to utilize brute force.

Then you will want to employ 'pink fluffy' Superchunk style porous corner traps featuring 6mil or greater faced fronts. And then after bass traps are installed and the space remeasured, one may then choose to employ limited EQ to mitigate the PEAKS (only, and then only up to 3dB) for modal peaks below 80 Hz.

After all of this is done, THEN individual ETC responses are derived for each source.
You will want to resolve each of the high gain sparse indirect energy arrivals into their vector paths and point of incidence.

A couple quick observations about the ETC above....aspects that can provide a quick indication of potential issues.
First, is the direct arrival REALLY at ~35 ms, or 39.6 FEET from the source speaker?!? I suspect not.
Thus, depending upon the method you intend to use to resolve the vector paths, you might want to consider using the hardware loopback correction for system propagation delay so that the arrival times displayed are accurate.

Also, you have quite a few significant sparse (isolated) high gain 'reflections' relative to the surrounding space, as well as a at least 4 Very significant late arriving indirect returns.

But becoming overly concerned with specifics aside from the basic configuration of the ETC process at this point is premature. Now just focus on configuring the platform in anticipation of the measurements to be performed with the mic remaining in the same precise position (a position 'marked' so that it can be Precisely re-established at any point in time later).

Also, when you do post results, please post the REW .mdata files.

 

[Wayne A. Pflughaupt]

Hey Ryan,

I notice that your frequency response graph shows a really low level. Did you do the SPL calibration routine? If you did and this is your actual measurement level, then it is low enough that ambient room noise might be encroaching on the results.

Of course, you don’t necessarily need to do the SPL calibration; it only means the vertical graph dB markings aren’t relevant in your case. As long as REW didn’t indicate low level or clipping when you took the measurement you’re fine.

As you have probably figured out, 1/3- or 1/6-octave smoothing for the upper frequencies eliminates the comb filtering most untreated rooms display lets you see the underlying trend in response. It looks like you have a fairly large plateau between ~1500-2500 Hz. It’s broad enough and “tall” enough (~4 dB) to probably be audible. You might be able to address this with your receiver’s built-in equalization.

A separate unsmoothed graph breaking out the bass frequencies (below 200 Hz) would have permitted better evaluation of what the sub is doing, but it looks like you’ve adequately smoothed response. However, the plateau between 30-60 Hz might be making the bass sound a bit bloated. House curves often sound best with a sloping curve that plateaus at about 23-30 Hz. It will most likely require an outboard EQ to accomplish this.

Re the low-freq ringing the waterfall graph displays, there isn’t much you can reasonably do for ringing below ~50 Hz or so. It doesn’t appear than you have any significant room modes or nulls.

Regards,
Wayne

 

[NegativeEntropy]

SAC,

Thanks for responding.

A few comments about process.

First, this is an iterative process, not an all in one do everything at the same time operation.


First, the crossover issues is to resolved and corrected. It is not some step in the room correction process.
And a NECESSARY part of the crossover configuration is the signal alignment of the sources.

Crossover setting:
Please describe the issues apparent with the crossover. My apologies that it is not clear to me that there are issues by looking at the measurement results.

My mains are capable to 40 Hz, so that is not a limiter. I took sub+mains sweeps with the crossover at 40-80 Hz in 10 Hz increments. A crossover setting of 70 Hz produced the visually smoothest response (with no or at most 1/24th smoothing).

What is the proper method for determining the crossover frequency? I will gladly read a reference source if you could point me to one.

Phase alignment of sub to mains: I have read of a few ways to do this. One is using the group delay plots, but these are usually too noisy to properly judge at the low frequencies a crossover is at. My experience was similar.

I adjusted the sub phase until the sub+mains SPL was loudest when playing a sine wave at the crossover frequency (70 Hz). Comparing frequency sweeps at 0, 90 and 180 degrees confirmed 180 gave the visually smoothest results.

Again, I would gladly read a reference source for the proper way to do this if my approach is not appropriate (e.g. I'm not trying to ask you to write paragraphs in response - I recognize responding in detail is time consuming).

After that is done, then focus on the modal behavior.
And please loose the FR smoothing. We are not trying to sell this or to make things 'look' nice. Instead we WANT to know the detailed issues that are obscured by smoothing! (Sorry, not angry with you, but seeing this predilection with smoothing remain an almost universal fixation with folks who claim that they want to identify and resolve problems in the process of obscuring them grows old...)

First map the space. The mapping of the region can be done via waterfalls, or if you are not comfortable with this, perhaps more easily by the playing stepped tones (in REW) in order identify modes and to then identify the location of regional peaks and nulls. Assuming L/R symmetry, explore the region about the desired listening position, both forward and backward from the current position, in order to determine the location least subject primarily to modal nulls, and then to a lesser degree to peaks. A problem avoided by relocation is much easier than attempting to utilize brute force.

Then you will want to employ 'pink fluffy' Superchunk style porous corner traps featuring 6mil or greater faced fronts. And then after bass traps are installed and the space remeasured, one may then choose to employ limited EQ to mitigate the PEAKS (only, and then only up to 3dB) for modal peaks below 80 Hz.

Heh, I understand your frustration with smoothing. I only used some because otherwise the High frequency response is unusable; 1/3 seems to be standard for above ~250 Hz and below that none or 1/24th? I guess I should have broken it into 2 different plots.

I did this when selecting a location for the subwoofer but limited myself to the frequency response graphs. Will this not show peaks/nulls?

I have not experimented with relocating the mains. I will do so. I also have port plugs coming that will grant more placement flexibility for the mains (they have a rear port).

Unfortunately, bass traps are likely a no go. This is a multipurpose space, not a dedicated theater and the WAF is not there.

After all of this is done, THEN individual ETC responses are derived for each source.
You will want to resolve each of the high gain sparse indirect energy arrivals into their vector paths and point of incidence.

A couple quick observations about the ETC above....aspects that can provide a quick indication of potential issues.
First, is the direct arrival REALLY at ~35 ms, or 39.6 FEET from the source speaker?!? I suspect not.
Thus, depending upon the method you intend to use to resolve the vector paths, you might want to consider using the hardware loopback correction for system propagation delay so that the arrival times displayed are accurate.

Also, you have quite a few significant sparse (isolated) high gain 'reflections' relative to the surrounding space, as well as a at least 4 Very significant late arriving indirect returns.

But becoming overly concerned with specifics aside from the basic configuration of the ETC process at this point is premature. Now just focus on configuring the platform in anticipation of the measurements to be performed with the mic remaining in the same precise position (a position 'marked' so that it can be Precisely re-established at any point in time later).

I was using a loopback cable. The receiver has 14' distance for the mains in it (which matches the distance measurements), but even with that (14' receiver delay + 14' actual propagation delay), as you indicated, it does not add up to 35 ms. I should have noticed this and investigated. Perhaps I was using the loopback incorrectly. I'll have to play with this more when the time comes.

Regarding mic position, yes, I will have to break out the plumb bob and painters tape as you advised in another thread.

Also, when you do post results, please post the REW .mdata files.

It was attached (hence my references to it in the text).

 

[NegativeEntropy]

Hey Wayne, thanks for responding.

I notice that your frequency response graph shows a really low level. Did you do the SPL calibration routine? If you did and this is your actual measurement level, then it is low enough that ambient room noise might be encroaching on the results.

Of course, you don’t necessarily need to do the SPL calibration; it only means the vertical graph dB markings aren’t relevant in your case. As long as REW didn’t indicate low level or clipping when you took the measurement you’re fine.

SPL was calibrated, I was just taking these sweeps while the wife and kids were sleeping, so I moderated the levels a bit. If memory serves, I had REW generating the sweep at -15dB and it resulted in a remaining headroom of between 10 and 15 dB (still in the green). Background noise level was 25dBA.

So, given that, do you think the levels were too low? REW did not indicate a problem.

As you have probably figured out, 1/3- or 1/6-octave smoothing for the upper frequencies eliminates the comb filtering most untreated rooms display lets you see the underlying trend in response. It looks like you have a fairly large plateau between ~1500-2500 Hz. It’s broad enough and “tall” enough (~4 dB) to probably be audible. You might be able to address this with your receiver’s built-in equalization.

Unfortunately this is already with Audyssey, so I'd need to add an outboard EQ to tweak this further. The Onkyo 709 has an EQ in it, but it is bypassed with Audyssey is engaged. If I disabled Audyssey I could make use of it. It would be fun to see what I could do with it, but I would hope Audyssey would be better
I am pondering an outboard EQ for the mains, but if this is the only area I would use it in, it would be difficult to justify adding another box.

A separate unsmoothed graph breaking out the bass frequencies (below 200 Hz) would have permitted better evaluation of what the sub is doing, but it looks like you’ve adequately smoothed response. However, the plateau between 30-60 Hz might be making the bass sound a bit bloated. House curves often sound best with a sloping curve that plateaus at about 23-30 Hz. It will most likely require an outboard EQ to accomplish this.

Re the low-freq ringing the waterfall graph displays, there isn’t much you can reasonably do for ringing below ~50 Hz or so. It doesn’t appear than you have any significant room modes or nulls.

Regards,
Wayne

Thanks Wayne, a BFD is within my discretionary spending range, so I'll have to ponder it. I'll play more with a flatter response (trimming the sub to -4 in the receiver does this) and make sure I prefer the house curve approach, then go from there.

--Ryan