[JohnR]

I decided to work through the AMS to see if I understand what to do and if Nyal and Jeff have any comments. This is a pre-existing measurement, so there is one problem which is that the bass measurement is only individual channels, not both channels at once. The system is currently dismantled so I can build another prototype, so I can't redo them right now.

With that said, here we go!

A: Noise Control

I skipped this one for now.

B: Reflected Sounds

Here is the ETC of left and right channels. I can't really tell if this is considered smooth and decaying. It's decaying but it looks rather reluctant around 20-25 ms.



Doing filtered ETCs of the left and right channels as per Figure B.5 is more illuminating. Left channel:



Right channel:



The speakers have a fairly uniform off-axis response (at least, I thought they did), so it looks like the absorption that is in the room is taking effect only at higher frequencies. (?)

C: Low Frequency Decay Times

The AMS doesn't say whether this should be done with one channel or both. However I only have them individually. This is the spectrogram of the right channel. I have drawn the upper and lower limits for the 40 dB decay from Figure C.1 onto it. It looks pretty poor in the 40-70 Hz range, and to me it looks like it's starting to get rough again above 200 Hz. The thing at 240 Hz is a baffle resonance I think.



D: Midrange Decay Times

T20 and T30 for the left channel:



Right channel:



Looks OK if perhaps a little on the high side, but something wrong at 400-500 on the right?

E: Consistency of Midrange Frequency Response between Left and Right speakers

Left and right channels. Looks pretty well matched. Whether it's +/- 3dB over the range depends on whether you account for the "room curve" - the in-room response of the speaker has a slope down from 100 to 10 k. The AMS didn't mention anything about this.



F: Low Range Frequency Response

Again, I only have these individually, not together (Below 80 is mono though). 1/24 octave smoothing:



1/3 octave smoothing:



Looks OK except for the 20 Hz criteria. f3 so about 30 Hz, which sounds better than taking the f3 down to 20. That's because I can do 30 with dipoles, but going lower needs monopoles and the decay times look really bad then.

G: Room Size and Construction

Nothing I can do about this.

Conclusion

My take on this is that the first needed is absorption targeted at two ranges, 40-70 Hz, and 200-600 (ish) Hz. Nyal/Jeff?

[Jeffrey Hedback]

WOW JohnR...excellent job and thank you for embracing the guidelines and graphical presentations.

I'll start with comments and I'm sure Nyal won't be too far behind.

b: the overlayed L & R ETC shows very good consistency between speakers with the red having slightly more energy overall.  This could be assymetry within the room, location of test mic/LP or a slight indicator that the acoustical nature of one side of the room could be better matched to the other.  However, there is no red flag in this regard, again shows very good consistency.  There is in fact the 10dB decay.  This would be a bit easier to visualize with a range of 60 or 70dB on the amplitude axis.  The area you question from 20-25ms is likely the rear wall (maybe 8' behind you???) and this is very smooth with no stray singular spikes above the general energy level.

One octave band studies do indeed show too much 500Hz energy on the left speakers graph. Maybe the speaker is closer to a boundary?  The 4K is consistent between speakers and although is more reduced than 1K & 2K it is not by itself an issue.  What I would do is reference the midrange frequency response and any subjective impressions and if the room feels dry or there are issues in the frequency response this could indicate looking at absorptive factors in the room more closely (maybe in favor of diffusion or reflection).

C: the LF resonance...it looks posisbly to me that the measure may not have enough headroom above the room noisefloor or there are structural resonances (which could be it's own subtopic)...point being it seems that the decay down to -40dB is within targets...but the energy is on the verge of peaking it's ugly head.  I often think of room modes being like teenagers "they love to party with each other and cause problems...but on their own, they're not a bad thing".  So in summary, I would see what level your volume was at when the measurements where taken and look to 40dB above the noise floor of the room (or 75dBC to 85dBC).

D: interesting how the right speaker has more stronger decay at 400Hz in T20/30 and the octave filtered ETC has more at 500Hz in Left.  This would direct my attention to possible furniture/cabinetry near the right speaker and a possible wall that may be closer to the left speaker...either way, these are two areas where lo-mid absorption could indeed help.

E: fantastic.  Indicates the great likelihood that the imaging is fantastic.  If there where strange deviations here, we could cross-reference the ETC both full spectrum and octave filtered to see possible room issues that are conflicting with the speaker's designed response goal.  Again, WOW.

F: again, fantastic.  The paper does state to measure both speakers (it's in the asterisks John), but the consistency between L & F would indicate that the summed LF response would remain similar.  There certainly is a context to the speaker type you alluded to with yours being OB...I'll let Nyal respond there.

Conclusions:
- I would look toward identifying areas that could be causes of lo-mid excess
- I would double check the spectrogram and if the measurement was plenty above the noise floor of the room, then I would inspect the wall surfaces and other elements that could resonate (again it's own subtopic).

I look forward to your reply.  Great stuff and really fun to comment on.

[tubamark]

. . . However, other researchers have found that holes in the power response aren't necessarily detrimental, implying that the spectrum of the early reflections don't necessarily need to be the same (or else dips are less readily perceived than peaks, which is already known). This point has been the source of some controversy in large part because different listeners have different preferences.

So, yes, you don't NEED to absorb these early sidewall reflections. You should try it without and see what you prefer.

I won't argue with issues of preference.  However, I think Toole is too congenial, carefully downplaying the "widening of apparent source width (ASW)".  Ultimately, altered image width IS a form of distortion, albeit one that some prefer.
I can totally dig that.  However, I don't think a reproduction standard that specifically aims to minimize acoustic distortion should permit such a fundamental aspect of playback to be widely negotiable.  Bloated source width also impairs the quality of phantom / center imaging.

Again, not a problem for many listeners and recordings.  But If objective playback fidelity is the goal, it should be addressed.  Decisions to deviate from a standard can then at least be recognized for what they are: preferences.

Clearly, most instances of image bloat would be prevented by the reflection spectrum requirement.  I just hate to see anyone get the idea that parking a bipolar or omnidirectional speaker 2 feet from a wall is a good start to accurate playback.

-- Mark

[Nyal Mellor]

I won't argue with issues of preference.  However, I think Toole is too congenial, carefully downplaying the "widening of apparent source width (ASW)".  Ultimately, altered image width IS a form of distortion, albeit one that some prefer.
I can totally dig that.  However, I don't think a reproduction standard that specifically aims to minimize acoustic distortion should permit such a fundamental aspect of playback to be widely negotiable.  Bloated source width also impairs the quality of phantom / center imaging.

Again, not a problem for many listeners and recordings.  But If objective playback fidelity is the goal, it should be addressed.  Decisions to deviate from a standard can then at least be recognized for what they are: preferences.

Clearly, most instances of image bloat would be prevented by the reflection spectrum requirement.  I just hate to see anyone get the idea that parking a bipolar or omnidirectional speaker 2 feet from a wall is a good start to accurate playback.

-- Mark

MMM..interestingly when we were discussing with Dr Toole he classified (if my memory of our discussions serves me) apparent source width as defined in the listening experiments as not making images subjectively 'fuzzy' and 'vague' but meaning that the image was fleshed out in terms of its outline and body. He also noted that reflections with consistent spectrum give the ear a chance to 'listen in' further to the timbre of instruments and thus increase resolution. Apparent source width is, it seems, a bit of an unfortunate piece of terminology descended from concert hall acoustics with which we are stuck.

Now that's not to say that preference plays an important factor, but that should only come into play after you've assured yourself that the spectrum of reflected and direct energy in your space is in balance.

[JohnR]

I just hate to see anyone get the idea that parking a bipolar or omnidirectional speaker 2 feet from a wall is a good start to accurate playback.

From the info in Toole, 2 feet would be too close, because then a) the delays are too short, and b) the angles not sufficiently different.

[JohnR]

Conclusions:
- I would look toward identifying areas that could be causes of lo-mid excess
- I would double check the spectrogram and if the measurement was plenty above the noise floor of the room, then I would inspect the wall surfaces and other elements that could resonate (again it's own subtopic).

Thanks for your comments Jeff, much appreciated. I will look into both of those areas when I get far enough with the next prototype to take measurements and look seriously into the acoustics aspects again. I know what I need to do now - thank you for the paper 

[doug s.]

i have always been curious as to how i should be measuring my horn loaded speakers - where should i put the mic for nearfield measurement, and what should i plug in as the distance dimension?  do i place the mic at the face of the plane of the front of the horns, or in the throat of the horn, or?  i am using a deqx processor.  pic of driver; decware modded fe206e in oris horn:




(same question for the bass bin.)

thanks,

doug s.

[Nyal Mellor]

Hi John

In reference to the standard being down to 20Hz...both Jeff and I are proponents of a 'full range' listening experience, which means essentially flat to below 30Hz at the minimum and preferably 20Hz as defined in the standard. For those listening to a wide range of music including modern cuts with synthesized bass then you could go even lower than 20Hz, though this gets challenging! For those with preferences only for 'real' instruments maybe 30Hz is ok since as mentioned in the white paper really 30Hz is as low as the fundamentals of the lowest musical note from most instruments.

In reference to the house curve concept....well really any deviation from an essentially flat response is going to involve timbral distortion. With the fundamentals of most 'real' instruments topping out at about 4k we'd want the response at the listening position to be flat.

[Nyal Mellor]

i have always been curious as to how i should be measuring my horn loaded speakers - where should i put the mic for nearfield measurement, and what should i plug in as the distance dimension?  do i place the mic at the face of the plane of the front of the horns, or in the throat of the horn, or?  i am using a deqx processor.

Hi Doug, we didn't reference use of nearfield measurements anywhere in the acoustical measurement standards...if you are referring to how best to measure with the DEQX then best would be 6 feet if you can get a measurement without reflections that far away from the speaker (you'd need to move it outside onto a tall table or other structure to avoid floor bounce...it would be difficult to achieve 6 feet in room)

[doug s.]

Hi Doug, we didn't reference use of nearfield measurements anywhere in the acoustical measurement standards...if you are referring to how best to measure with the DEQX then best would be 6 feet if you can get a measurement without reflections that far away from the speaker (you'd need to move it outside onto a tall table or other structure to avoid floor bounce...it would be difficult to achieve 6 feet in room)

do i measure from the driver or from plane of the front of the horn, which is about 17" forward of the driver?  deqx asks for the measurement dimension.

thanks,

doug s.

[Nyal Mellor]

do i measure from the driver or from plane of the front of the horn, which is about 17" forward of the driver?  deqx asks for the measurement dimension.

thanks,

doug s.

PM sent, thought it might be better to do this off thread...

[brj]

Perhaps it deserves a separate thread, but I wouldn't resort to PM.  Knowing how to measure properly in different scenarios and to acquire different specific measurements is not an area I've seen well discussed.

[nwboater]

Perhaps it deserves a separate thread, but I wouldn't resort to PM.  Knowing how to measure properly in different scenarios and to acquire different specific measurements is not an area I've seen well discussed.

I agree. As the owner of Klipsch Klipschorns, La Scala and Heresy's this would be very helpful information.

Thanks very much.
Rod

[Nyal Mellor]

I agree. As the owner of Klipsch Klipschorns, La Scala and Heresy's this would be very helpful information.

Thanks very much.
Rod

We did it via PM because it was related to the DEQX specifically, not the acoustic measurement standards

[vettett15]

Figured I'd give it a shot:

Reflected Sounds:

Both L and R:



I can't figure out how you are showing multiple individual frequencies at the same time (i.e. 500, 1k, 2k, etc).  I know it is something stupid I am missing.

The spike after the direct sound is coming from the left channel, not sure what is causing it.

Low Frequency Decay:

Right:




Left:




Both seem a hot, need some low end absorption?

Mid-range Decay Times:

Right:




Left:



Really high, no?

Mid-range consistency:




I have a room/speaker placement issue causing a large dip in the 250-500 range. I need to work on fixing this.

Low-range consistency:

1/24




1/3



needs work...

[vettett15]

Nyal showed me how to overlay the filtered ETC measurements.
     You have to open the same measurement multiple times and then use the overlay feature.

Here are my overlays:

Right




Left





Are these as bad as they look?  Looks like heavy comb filtering is going on here. 

[Jeffrey Hedback]

Hello vettett15,

Glad Nyal has helped with the ETC overlay.  If you go to the Impulse tab of each file and smooth the filtered responses you'll see the relative spectral balance better.

Quick take:
- three biggest issues are (in no particular order): the hot reflection on right side*, the excessive decay time and the lack of energy in the lower mid/upper bass region (~200Hz-400Hz).
* you can use the time between first arrival and a hot reflection to track the trouble spot.  One millisecond=1.13ft (keep that in mind).  Now gather the distance between your speaker and the mic.  Next, gather the time (from ETC graph) from first arrival to the hot reflection.  It looks to be ~10ms.  If (for example) your mic to spkr distance was 9ft, the hot reflection would have a total path length from speaker to mic of ~17'-9".  You can take a tape measure and focus on the right side to track down a reflective object with that total path length.  You then reference the filtered ETC to decide the right tool to control the reflection

Overall, I suspect that your actual bass response sounds relatively balanced to the upper mids and highs.  I also suspect that imaging is poor and varies from content to content.  I finally, the lack of lower mid/upper bass energy would be a real buzz-kill...no punch.  I would suspect speaker location relative to room as primary cause of lower mid valley???

Not knowing anything about your room...it may only take a few broadband absorption panels and corner traps to bring the room down to ~.4s decay time with better spectral balance.

Might you have a room diagram and a few pics?

[vettett15]

Jeff,

    Thanks for the reply, the thought of improving my sound is exciting.  Couple comments:

  All of these measurements were with no EQ, when I add my sub and its eq along with eq for the mains, the data gets better.  I think it could be much better with room treatment.

   What hints at a hot reflection on the right side? I thought the spike on the left ETC would mean a reflection.

   I was wondering about those filtered responses being so crazy, but those are the 1/1 filtered responses you choose.  Is there further smoothing I can do to better see them?

   I will use the method you say to try and find the surface creating the reflections.  I think REW has a built in measure tool you can use between peaks.  I'm curious how you got 17'9".  Wouldn't the total length of the reflection causing the peak (between the speaker and mic) be 9' + (1.13*10) = ~20' 4"?  If that is the total length traveled, I could cut a piece of string that length and since I know the beginning and end (speaker and mic) I can tape the two ends at those respective locations and then bend the string to see what it hits.

I used the realtraps mode calculator the other night and it shows that a large amount of nodes building up between 250-500hz, so I'm guessing the room is playing havoc in this area.  I would love to take care of this somehow, I'm guessing absorption and speaker placement are the keys here?

I will upload some pictures later.

Thanks,
Pete

[vettett15]

Snapshot of the room:


I have curtains on the windows (not super heavy but decent).

Just to the right of the left speaker is the sub and then to the right of that is the entertainment stand.

[Nyal Mellor]

Hey good job on measuring the room!

The reflection on the right side is that energy shown in the right hand side ETC at around 45ms. It is close to the level of the direct sound. Really once you see a spike in the non-band limited ETC I would recommend looking next at the band filtered ETCs like you have done to determine reflection issues.

Glad you figured out the overlaying of the ETCs. Like Jeff said adding some smoothing to the ETC makes examining the trends for spectral balance easier to see. You can find it in the settings (the cog icon) in the ETC overlay screen.

One thing you can do is set the peak of all the impulse responses to 0. You do this by going to the Impulse tab and selecting Estimate IR delay from the settings menu (the cog icon). That will make it easier to find the time delay of the reflections. I might also suggest making the font of the labeling bigger so when you post the graphs it is easier to see. Find the option for this in the Preferences menu.

You should generally measure the L&R together first below 250Hz i.e. use a y-splitter to send the signal to both L&R inputs. Makes the first level of diagnosis a bit easier than posting a L&R separately.

Generally the low frequency decay spectogram shows evidence of some undamped room modes (sharp peaks and extended tails) and this extends quite low down in frequency (below 50HZ). Additional bass trapping definitely required, although there aren't many off the shelf products that work that low. RPG modex plate is one of them or else its custom design stuff.

The lack of energy from 200Hz-400HZ is going to have a very audible impact leading to some quite substantial timbral disortion in this region. 10dB is about half as loud, so notes in this area of the frequency spectrum have much less energy than they should. This could be speaker related or speaker boundary interference related or maybe both....

Certainly there is work to be done here and significant opportunity for sound quality improvement