Home Theater Forum and Systems banner

21 - 40 of 48 Posts

·
Registered
Joined
·
23 Posts
Discussion Starter #21
Here's the mdat file from last night. I hope that helps, and I can also try to alter the graphs tonight when I'm home from work. The distance from the tweeters to the mic is 8' 8" as measured with the Bosch laser measuring device, so it's pretty accurate.

Thanks,

Andrew
 

Attachments

·
Banned
Joined
·
516 Posts
Below is the ETC windowed for normal viewing.



Something is not right - in addition to the pre-causal behavior (energy arriving prior to the direct energy!!!). And I do not think the pre-causal energy is due to mechanically coupled and transmitted energy. The floors are not 'sprung' wood are they? But more than that, I have reason to doubt ~7 direct and indirect sources within ~-5 dBfs spread over ~38ms.

Also, for our purposes, can you measure from the center of the speaker baffle (not the tweeter) to the mic capsule to the proper precision. (The precision of the measurement is determined by the measuring device. The precision is to the nearest tenth of the smallest unit division...meaning, for instance, if the marking of a measuring device is marked to tenths of a unit, then we can estimate to the nearest tenth of that scale - meaning we are precise to the nearest 100th. Or if the device is marked only to integer units, then we can estimate precisely to the nearest tenth of the unit. Stating the measurement to less or greater precision than that determined by the unit sale is a mistake. This is important as the total calculation determined by the use of he measurement is determined by the least precise measured unit. So we will benefit from the most 'correctly' precise we can be... )

I know many are of the opinion that the tweeter possesses some magic quality, but to clarify this issue - the tweeter is ofter a concern ONLY in terms that its dispersion is normally the most limited (narrow). Thus in order to be 'on axis', as the tweeter tends to beam, we think interns of the tweeter being the limiting agent.

But here that is not our concern. We are concerned with the energy that is emitted by the loudspeaker. And in this frame of reference, the tweeter emits much less energy than the other drivers. Here we are concerned with what is often referred to as the acoustic center of the unit - similar to the center of mass or a unit's acoustic 'moment'. In other words, the 'point' from which all of the energy appears to radiate from the entire loudspeaker. And as we have not measured the actual acoustic center, we can use the geometric center of the loudspeaker's front baffle as the reference point.

Thus, the more precise we can be with that measurement, the better off we will be using that measurement summed with the data from the ETC, where the limiting precision will be that of this distance measurement from speaker baffle to mic.
 

·
Registered
Joined
·
23 Posts
Discussion Starter #23
Thanks again for all detailed info. A lot of it is going over my head, but some it is starting to sink in. I will take a more accurate measurement when I get home from work. My speakers look like ones below, so should I measure from the mic to the point between the mid and bass drivers?

 

·
Banned
Joined
·
516 Posts
Yes, that would be a better point from which to measure.

Also, using the picture as a source, note how the cabinet extends a bit in front of the speaker to the left side (viewing from the front).

This small jump discontinuity will function as a source of diffraction and reflection - causing energy to emanate from that cabinet corner as if it is yet another small speaker, further acting to superpose with the direct signal and thus harming intelligibility, imaging and localization.

Assuming one does not have the option to remove the impediment, it would be preferable to position the speaker such that the cabinet face was flush with the speaker baffle front in order to (try t0) act as an extension of the speaker baffle to minimize this effect.

I mention this simply as a point of illustration, as I know the picture may not be of your particular installation.
It seems everything is out to get us, doesn't it?:D
 

·
Registered
Joined
·
23 Posts
Discussion Starter #25
Something is not right - in addition to the pre-causal behavior (energy arriving prior to the direct energy!!!). And I do not think the pre-causal energy is due to mechanically coupled and transmitted energy. The floors are not 'sprung' wood are they?
Sorry, but I just noticed the date on the mdat I posted was Aug 23rd. I meant to upload the file from the 25th, but was rushing before I went to work. I'll redo the ETC's tonight and post the graphs based on your recommendations.
 

·
Registered
Joined
·
23 Posts
Discussion Starter #26 (Edited)
Here's a couple new graphs with the distance from the mic to speaker baffle marked on them. I've also attached the mdat. I tried to measure as accurately as I could with the Bosch DLR130 laser measuring device, which is accurate to 1/16" or 1.5mm. However, it was hard to line the edge of the device up perfectly with the mic, and I might have moved it slightly when measuring.

I read what you said earlier about the pre-causal behaviour. Are those the peaks on the graph before 0s? My apartment building is made of concrete, with parquet wood floors, and I live on a busy street. So there are often loud traffic noises outside. I thought that might have been the cause for some of the peaks, but I ran the tests a couple times and tried to wait for the traffic to die down, and the graphs look similar. So I don't think that's it.

I also hope I'm doing the measurements correctly. I have the ECM8000 mic pointed straight up and tilted forward about 20 degrees. The top of the mic is set to ear level when I'm in my listening chair, about 36". The measurement settings are "256k log sweep from 20 Hz to 300 Hz at -12.0dB taking 5.9s". I'm using the Tascam US-144mkII as the USB preamp.
 

Attachments

·
Banned
Joined
·
516 Posts
Why are you band limiting the sweep from 20 Hz to 300 Hz? Technically, we are not even driving the room with energy that will have substantial specular behavior.

Use full range sweeps! And leave off all smoothing for any responses. And process/convolve only unsmoothed responses.

Do not try to play with the measurement by thinking in the frequency domain.

As far as the graph, there should be more definition to the energy arrivals. And it is highly unlikely that you will have reflections persisting at 25ms (~283 feet of travel including bounded incidence) that are barely 2dBfs down from the direct signal (or ~-6.5 dBfs at ~44ms (after 49.7 feet of travel including bounded incidence!).

And with a severe lack of L/R symmetry, you should not have such a plethora of high intensity early arriving reflections except off the floor (especially without substantial energy being delivered that behaves specularly at those frequencies)

Plus something is amiss with the pre-causal signal arrival of ~-2dBfs at ~-2ms relative to the direct signal - unless the signal and drivers are THAT much (~2.26 feet) out of alignment....

Try a full range sweep. We'll figure this out!:D
 

·
HTS Senior Moderator
Joined
·
5,288 Posts
20-300 is easier to identify (with smoothing off), where any frequency response problems are more specifically rather than a full range sweep where it's a guess at best.

Bryan
 

·
Banned
Joined
·
516 Posts
If one wants to examine the modal range from 20 Hz to 300 Hz, use the proper stimulus and response.

Truncation of sweeps has unintended consequences, and folks would do well to stop thinking solely in the frequency domain.

For insight into modal behavior from 0-300 Hz, generate the waterfall response driven with a full range signal and then window the response appropriately.

In order to generate the ETC response, drive the system with a fullrange stimulus.

There is a fundamental conflict in trying to use a gated signal from 20 Hz to 300 Hz to generate an ETC that provides insight in to specular energy behavior while restricting the stimulus energy to frequencies that behave (primarily) modally and expecting to see the full range (where our region of concern is actually from about 300 Hz to 10,000 Hz) specular energy behavior.

Thus, for an examination of the specular energy region, drive the space with energy spanning the full range of interest - at least 20-20,000Hz for instance (0-22Khz is even better). ( As an aside, I would normally also suggest using frequency weighted sweep were this feature available in REW. This makes good use of the pink noise behavior and improves the signal-to-noise ratio for the high frequencies simultaneously with more energy being applied to the low frequencies and less to the high frequencies in order to avoid the potential to blow up tweeters, where the difference between bass and treble in the linear plot amounts to only 18dB between 20Hz and 20kHz, in the octave band plot to only 10dB, and the mostly high-level mid-band range is even lowered a bit (5dB).)

I wouldn't recommend using a Porsche GT3 to tow a 26 foot long boat or house trailer either. :D
 

·
HTS Senior Moderator
Joined
·
5,288 Posts
Understood completely. That said, using a tool that's well supported here which does do sweeps is very easy and relatively accurate for assessing the basics of what's happening in the room. Yes - you can certainly run a full range sweep and then look at just the 20-300 area.
 

·
Registered
Joined
·
23 Posts
Discussion Starter #31 (Edited)
Doing a full range sweep seems to have made a big difference and I think I'm getting a proper looking ETC's now. I think had also read that it was easier to limit the sweeps to 300Hz, so I guess that's why I was doing it.

Here's some new graphs. Please let me know what you think.

I was also wondering if disconnecting the speakers is the only recommended way to make these measurements. Is there any reason why I can't disconnect the input for right or left, or just use the balance control? Disconnecting the speakers is the least convenient way for me, so I would prefer one of those other methods as I continue my testing.
 

Attachments

·
HTS Senior Moderator
Joined
·
5,288 Posts
Using a balance control or disconnecting a channel from the source can certainly also work.

Bryan
 

·
Registered
Joined
·
23 Posts
Discussion Starter #33 (Edited)
I did some more testing last night, and I'm starting to understand how to use the Impulse measurements to identify the various reflection points. By taking some measurements and experimenting with a piece of 3" upholstery foam and heavy blankets to treat the right speaker reflection points on the floor, back of the door and ceiling. I could easily see how the Impulse response changed when each reflection point was treated.

Below is the right speaker Impulse with no treatments. The three spikes after the initial spike at 0 correspond to the floor, back of door and ceiling. The next picture shows the Impulse with foam on the ceiling and blankets on the back of the door, and you can see a drop in the second and third spike.

However, I'm still not sure where to start with the proper acoustic panels and/or bass traps. If we forget about aesthetic concerns for now, what would you guys suggest? Would a Super Chunk in the right corner and some panels at the reflection points be a good place to start?

Thanks again for your patience and assistance while I work through this.
 

Attachments

·
HTS Senior Moderator
Joined
·
5,288 Posts
Yes. That would be a good place to start in addition to right next to the speaker. Impulse response can be very useful in terms of finding specific reflections but will not show you boundary interactions.

Once you individually measure the left and right channels, you'll see pretty quickly the difference due to the boundary proximity.

Bryan
 

·
Registered
Joined
·
23 Posts
Discussion Starter #35
What do you think if I treated the corner like the picture below? It's basically a Studiotips Corner Absorber plus a 3" panel with a 1" air gap on the side wall beside the speaker. Rather than wrapping each panel in fabric, I was thinking about a screen that would sit in front of the insulation. It might look neater and require less fabric.
 

Attachments

·
HTS Senior Moderator
Joined
·
5,288 Posts
You could certainly do that and it will help a good deal with the difference in boundary gain between the 2 speakers and also likely address at least some of the near wall reflections on that side. Not going to fix all the issues but it's a good starting point.

Bryan
 

·
Banned
Joined
·
516 Posts
A couple of comments...

As your primary concern is modal behavior, you would do well to address the modes first.

Regarding the proposed traps, you might consider a Superchunk style corner trap using the 'cheap pink fluffy stuff" and utilizing several pieces of plastic orchard bird netting to create several partitions that will act to support the material such that it will not compress as much over time.

Also note that a 3 inch panel mounted adjacent to a wall will not act as a bass trap. in fact, a 3 " panel with a 3" gap, utilizing Rockwool, will only exhibit an ~.3 absorption coefficient at 100 Hz! In fact, a 4 inch panel with a 4" gap is about the 'smallest' you want to use for a broadband (mid and higher) absorptive panel.

To be effective below 100 Hz, you need to go with at least a 6" panel with a 6" gap ([email protected] Hz and [email protected] Hz).

AFMG Reflex simulations curves are available for the above.


As far as the ETC...

I am not sure how much you want to go into this, as it can quickly become a extended discussion of the theoretical and conceptual underpinnings that are quite different than that with which most are familiar. We can certainly do it, but its not worth it if no one is interested and if we just get comments from folks saying that their 'brain hurts'.

Stated quickly, the ETC can indeed identify each reflection, provide information about the nature of the reflected energy, and provide specific detail not only about its precise path, but also identify the precise location of boundary incidence!

In other words, it can tell you precisely where to place treatment material appropriate to your desired response model. And it can verify its effectiveness.

Below is a very quick perusal of one the ETC responses annotated generally.
If you are serious in following up and determining the paths and specific points of boundary incidence for each anomalous reflection, let me know - and post the mdat files.

Also, I will provide a synopsis of the manner in which the reflections can be identified and also resolved into their paths and walked back to their boundary incidence locations. (As soon as I figure out the interface sufficiently to attach a PDF...:D)



 

·
Registered
Joined
·
23 Posts
Discussion Starter #39
Regarding the proposed traps, you might consider a Superchunk style corner trap using the 'cheap pink fluffy stuff" and utilizing several pieces of plastic orchard bird netting to create several partitions that will act to support the material such that it will not compress as much over time.

Also note that a 3 inch panel mounted adjacent to a wall will not act as a bass trap. in fact, a 3 " panel with a 3" gap, utilizing Rockwool, will only exhibit an ~.3 absorption coefficient at 100 Hz! In fact, a 4 inch panel with a 4" gap is about the 'smallest' you want to use for a broadband (mid and higher) absorptive panel.

To be effective below 100 Hz, you need to go with at least a 6" panel with a 6" gap ([email protected] Hz and [email protected] Hz).
So there's seems to be some consensus that a Superchunk in the right corner is a good start, but the effectiveness of any further treatment of the wall beside the right speaker is still questionable.

If I was going to use a 4 to 6" thick panel of OC703 with a 4 to 6" gap, could I skip the Superchunk and just fill that 4'x8' side wall with the panel from the back wall to the door? Something like the picture below might suit my aesthetic concerns better, and be easier to build, than separate corner and side panels, especially if then need to be so thick.

Thanks again for the incredibly detailed response. I'll take some more time to read and digest the ETC info.
 

Attachments

·
Banned
Joined
·
516 Posts
A few suggestions that are not only more effective, but much less expensive.

For the corner bass trap, stop messing with expensive OC703 panels and the additional complexity of complex multi-element face frames and simply make a Superchunk corner trap that straddles the entire space that you are imagining being treated with a corner trap and adjacent panels.





The additional space that is filled by the resultant hypotenuse is negligible in terms of room use, but substantial in terms of bass absorption.

And fill the entire corner region with cheap fluffed 'pink fluffy stuff' available at any home improvement store. Just provide a few horizontal layers of plastic orchard bird netting (cheap at any nursery supply or Wal-Mart in season) dividing the Superchunk into sever 2-3' compartments. You can fill these with fluffed 'pink fluffy stuff' and the netting will minimize the insulation settling and compressing due to gravity as the mass will be less in each compartment and it will not all press on the lower portions.

This will outperform the Superchink made with the more expensive lower performing semi rigid OC703.

{Part of the reason the corner placement will outperform the planar application is that some of the energy reflects ;laterally off the wall surface and is directed 'sideways' through the absorbent material, effectively increasing the reactive area whereby the energy interacts with the absorptive material. In other words, porous traps are velocity based devices. Velocity goes to zero near a boundary, so by moving the absorbent material further out from the wall, you lower the effective frequency that corresponds to the maximum velocity quarter wave spacing, and then you also further increase the amount of interaction by having some if the energy be redirected off the wall as it travels 'sideways' over a longer path through the absorbent material, thus increasing the energy dissipation.}

And its MUCH simpler to construct as well! A case where the easier cheaper alternative outperforms the more expensive and complex variation.
 
21 - 40 of 48 Posts
Top