2 channel, left and right speakers only graphs

[jaspal kallar]

Hello

I've a got 2 channel system, no sub,

I'm going to be adding room treatments (some corner triangles first) but wanted to do some measurements first to find what peaks (and valleys effect my room). I'm awaiting some y-adapters so in the mean time I did measurements on the left and right speakers separately form my listening position. Later I'll do the left and right together. I'm still finding my way around REW. so let me know if I've missed some information. These are my first graphs with 1/3 smoothing, target setting with full range, measured at 75db:

The right speaker:


The left speaker:


Any comments?

My room has alot of echo. What part of REW can I use to check/measure echo (reverberation) times?

Under 'Filter Tasks' I clicked on "Find Peaks" and had 3 peaks for both left and right speakers. I'm wandering if these are the peaks one obtains when using a SPL meter and test tone cd (what I call the manual method ;-) ? I ask since the person helping me with choosing room acoustics wanted this information after I've installed the acoustic triangles. He recommended a SPL meter and test tone cd to obtain the peaks along with applying manually some SPL meter corrections .

Btw can these Find Peak values be exported to file ?

Cheers

jaspal

[brucek]

What part of REW can I use to check/measure echo (reverberation) times?

Waterfall plots from 0-200Hz and RT-60 when above 200Hz..........

I'm wandering if these are the peaks one obtains when using a SPL meter

These are peaks that REW passes as modal resonances above the target you have set.
Room treatment won't have a lot of effect at modal frequencies (unless they were extremely large). Resonances are usually first dealt with by careful speaker placement and subsequent measurement to get the smoothest response. Then for pesky cases, EQ is applied.

can these Find Peak values be exported

You can export Filter settings, Measurements, Impulse Response and RT60 data.

brucek

[jaspal kallar]

Thanks for your answers Bruce. I'll look into the waterfall and RT-60.

I was wondering if you or anybody else could say if the graphs (for right and left speaker) are typical for an untreated, sparsely furnished room like mine which has approx. dimensions of 14' x 16' x 8'?


Cheers,
jaspal.

[Wayne A. Pflughaupt]

Can't really tell anything about that with those graphs, since they are limited to the lower end. Full-range readings will show really ragged response in the upper frequencies in an overly live room, due to comb filtering. This graph is a fairly severe example from a room that needs more treatment:

The better the room is dampened, the "smoother" the line will be, like so:

That's about the best you can tell about upper frequency dampening with a standard response graph. As far as the lower frequencies go, a standard response graph will tell you absolutely nothing. As brucek noted, waterfall and RT-60 graphs are better for gauging the effectiveness of room treatments, especially "before" and "after" comparisions.

Regards,
Wayne

[jaspal kallar]

Hi Wayne, thanks for your post.

Below is the right hand speaker with no smoothing and a wider range. It looks similar to your green graph:



I'm looking at the Waterfall. I have to generate it. Are the following default values okay before I generate:

The time range (ms) = 300
Window (ms) = 300, 3.3 Khz

x-axis (on) = -16
y-axis (on) = 100
z-axis( (on) = 150

Below is the RT 60 for the right hand speaker:



I'm not sure how to interpret the RT 60 results above and in particular what the vertical (y) axis denotes?

Cheers.

jaspal

[brucek]

Below is the right hand speaker with no smoothing and a wider range

Always start your graphs horizontal axis at 15Hz. There is generally no usuable information below that unless you have a very large subwoofer.

I'm not sure how to interpret the RT 60 results above and in particular what the vertical (y) axis denotes?

Note that RT60 isn't meaningful below ~200Hz.

The graph shows the decay across an octave (or 1/3 octave) band of frequencies as a single line denoting time. The vertical axis is from 0 to 1 second in time.

For domestic sized rooms a sensible figure would be a nice flat line around 0.3 seconds. HT setups are considered better on the dead side (lower RT60) and a music setup a little more lively (higher RT60). So, your RT60 is not too bad.

I'm looking at the Waterfall. I have to generate it. Are the following default values okay before I generate:

The time range (ms) = 300
Window (ms) = 300, 3.3 Khz
x-axis (on) = -16
y-axis (on) = 100
z-axis( (on) = 150

I would use 600ms time window.
x-axis (on) = 1
y-axis (on) = 100
z-axis( (on) = 150
Change to scale of vertical = 45dB-105dB and horizontal = 15Hz-200Hz.
Change mode to LOG from LIN (important)


brucek

[jaspal kallar]

Here's the right speaker Waterfall graoh:





Once again I'm not sure how it looks. Has anybody any comments on this waterfall graph?

What is one looking for by looking at the waterfall graphs?

Cheers

jaspal

[brucek]

What is one looking for by looking at the waterfall graphs?

If you move the slice slider back from 30 to 0, you will be looking at your standard frequency response graph. If you move the slider to 2, you will be looking at the signal that is still present 20 milliseconds later (600ms/30 slices = 20msec per slice). Keep moving the slider out to 30 and you will be at 600msec.

Notice your 40Hz-60Hz area. You still have a fairly large signal still present at 600msec. That's over a half second later after the original stimulus, you're still hearing the signal at 40-60Hz. You can understand how that won't sound so good. This ability to see what happens in the time domain in addition to the standard frequency response amplitude, allows you to position the speakers to try and eliminate that ringing out at certain frequencies.

brucek

[Wayne A. Pflughaupt]

jaspal kallar wrote:

What is one looking for by looking at the waterfall graphs?

Basically, it's showing you how long it takes for a linear broadband bass signal to fade away. I prefer a lower noise floor for these graphs, say 30-35 dB, because most rooms are quieter than 45 dB. Even with the 45 dB floor, your signal has pretty much fully attenuated before 500 ms, which isn't bad for an untreated room (i.e., no bass traps).

I wouldn't concern myself with the lingering signal at 50 Hz. It probably isn't from your home theater system, but from some other noise source inside the house - A/C system, refrigerator, etc. If you live in close proximity to other residences, like in an apartment or townhouse, you could also be picking up some noise from the neighboring residences. For instance, when everything's quiet in the house I can hear my next door neighbor's air conditioner running. That kind of stuff will show up in a waterfall.

It might be beneficial to extend the graph time to 1000 ms and see what the 50 Hz signal is doing. If it's a legitimate audio signal from your system, it should show signs of diminishing. If it appears to steady-state, then it is from some other source. You can see in this graph what constant noise generated from an extraneous source looks like - note the 125 Hz signal:

In any event, make sure every potential source of noise in the house is turned off when taking a reading for a waterfall graph.

Waterfalls are also useful post-equalizing to see how well you've addressed a room mode.

Regarding RT60, a little Googling tells me there is some debate in pro audio circles as to its usefulness in small rooms (recording studios and control rooms are typically small, for instance, so we have something in common with those professional settings). I've pasted some quotes I culled from various sites below, along with hyperlinks. You can take a look at them and draw your own conclusions.

Regards,
Wayne



LSI’s technical answer man fields a multipart question from a reader
If I want to measure an acoustic parameter such as RT60 or frequency response in a small room like a home theatre or a recording studio, can I use a (SIA) SMAART Live acoustic-based (computer) measurement system? Danny Murniadi

A small room really doesn’t have an RT60 (reverberation time). Reverberation time requires an even slope of level decay, which is a smoothly sloping, dense field of reflections that has an even and measurable drop in level with respect to time. This smooth decay slope is what determines how many seconds it takes the sound to drop by 60 dB (RT60).

Without a dense field of decaying reflections, RT60 cannot be determined. A small room only has relatively few early reflections before the sound dies altogether. You may be able to get some measurement systems to produce a number for RT60, but this really is meaningless until you get to larger rooms with a mathematically statistical (sufficiently dense) decay field.

See figures Small Room 1, 2, & 3 (below). Note that a variation of only 0.24 seconds in RT60 indicates more than a 50 percent difference in room absorption, comparing the shortest calculation to the longest ( 0.47 and 0.71 seconds). All that was done differently in the three small room RT60 calculations was to change the data points that determine the calculations.

___________________________________________


When a loud short sound is made in a room, the sound is reflected and it takes time for the sound to completely dissipate. RT60 is a measurement term that means "the time it takes for the initial sound level to decay by 60 decibels". The RT60 is one of the easiest to determine, and most useful, measurement techniques that we can employ to "tune a room" for recording or playing back audio.

A racquetball court has an RT60 of many seconds, which is basically useless for anything but, well, racquetball. A symphony hall may have an RT60 of 1-3 seconds, which is appropriate for this type of performance. But it is generally recommended that a small room for recording or playback should have an RT60 of less than 3/4 second.

An important attribute of the RT60 is that it is frequency-dependent. This means that it may take a different amount of time for a low frequency to dissipate than it takes a high frequency to dissipate. In addition to the fact that the total RT60 should be maintained below 3/4 second, a critical point is that the RT60 should be equal, or as close as possible, across the audio spectrum. A room with an RT60 of 1 second for low notes, but only 1/3 second at high notes, will be perceived as boomy, or muddy. The same room with a 1 second RT60 for high notes might sound very reverberant, but it is less likely to be described as boomy, even though the amount of bass reverberation has not changed.

___________________________________________


Raza,
> i am am currently calculating the best RT60 times possible for a control room which is 7m x 4.5m x 3.5m ... i have found that for the lower frequencies RT can be a little higher <

RT60 is not as useful as you might think for small rooms. At mid and high frequencies simple echoes dominate the top 20 dB or so, and anything below that is not so important. And at low frequencies RT60 is meaningless in a small room. What really matters there is "modal ringing," which refers to the decay time at the individual mode frequencies.

This does not mean you don't need to treat a small room! Small rooms need acoustic treatment, and especially bass trapping, even more than large rooms. But curtains are not an appropriate material because they absorb only higher frequencies. What's really needed in all small rooms is broadband absorption that's effective to as low a frequency as possible.

--Ethan

___________________________________________


4. RT-60 is generally not even a consideration in our control rooms. For large performance rooms and recording rooms, we will sometimes calculate RT-60 but rooms less than 5000 CSF it's pretty much a waste of time (*IMHO).

___________________________________________


Lastly, let’s take a look at our listening rooms. Listening rooms have short reverberation times relative to our church, but long reverberation times relative to the control room mentioned above. Reverberation times for listening rooms vary depending on listener preferences including listening levels, and whether or not it is designed for multi-channel or 2 channel use. This is also an area where we can scientifically measure the reverberation times, but it’s somewhat of an art to get the reverberation times correct for a particular listener—not unlike the differences in studios—what are the goals and what do we want to achieve. Some people feel that the listening room should not interact with the speakers, and it should perform much like the control room. Having listened in rooms designed like this, I can tell you I completely disagree. It takes away the kinetic energy of the music and having no interaction with the room has a very unnatural sound. In general we would like to achieve something between 0.34 and 0.39 seconds for an RT-60 from about 200 Hz on up. RT-60 measurements in small room acoustics below 200 Hz have little meaning and are flawed by the energy buildup of room modes. To evaluate bass response below 200 Hz, we have to use other methods.

[DrWho]

brucek wrote:

Waterfall plots from 0-200Hz and RT-60 when above 200Hz.........

RT-60 doesn't really exist in a home environment - it's a concept based on having sufficient randomized reflections, which isn't the case until your room is huge (like the size of a gym).

A much better measure would be the ETC, which will show the relative decay, but more importantly the specular reflections - which are what you're hearing as the echo's.

RT-60's are used when trying to shorten the reverberant decay, which are often used in conjuction with ETC's so that absorption can knock out the specular reflections while also reducing the length of the reverb. In the home, the reverb is never close to being long enough...which means there is no point to looking at an RT-60 because you're not trying to shorten the reverb.

The typical solution in a home environment is to use diffusion...which, when properly implemented, increases the number of random reflections (lengthening the reverb) while cutting apart the specular reflections (getting rid of the echo's).

Specular reflections that happen within the Haas window introduce a comb-filtering effect which reduces intelligibility/clarity while creating tonal imbalance from the crazy frequency response aberrations. The Haas window is the shortest length of time between two sounds in order for them to be perceived as separate sounds...it's typically on the order of 20-50ms. So if you snapped your left finger about 20ms after your right finger, it would be perceived as a single (and likely louder with different timbre) snap. When you get to about 60ms between finger snaps, you would start hearing it as two different snaps (and the timbre difference and apparent loudness return to normal).

All that to say, the most powerful part of the ETC is being able to identify specular reflections that occur within the Haas Window....getting rid of these is absolutely vital for maximum clarity.

[jaspal kallar]

Thanks to Bruce, Wayne and Mike for your responses.

I can attach the ETC (Energy Time Curve) graphs say but I wanted to check what X and Y axis values I should use. Are the default vales okay?

Left: -1.0
Right: 2.0
Top: 0
Bottom: -120


I've added some some acoustic Triangles (Corners), one in each corner and I did some measurements. I'll will therefore attach the before and after waterfall and ETC graphs.

Once again being new to all this I don't really know how to interpret the ETC graphs?

Thanks,

jaspal.

[DrWho]

An ETC is showing the relative magnitude and time-arrival of every reflection in your room.

So for example, at time=0, you have the largest spike because that is the direct sound of the speaker (which has the least amount of attenuation since it is traveling the shortest distance). If your speaker is 3 feet from the side wall, then you will see another spike in the ETC at a little over 6ms (depending on how far away you're sitting). Sound travels roughly 1 ft every 1 ms (so 6ms corresponds to 6ft). If your listening position is 10ft from the rear wall, then you'll see another spike at ~20ms.

You could sit down and try to calculate all of the reflection paths in your room, but the ETC will show you exactly what is going on and remove all the guesswork (well most of the guesswork).

Next time I get my laptop fired up, I'll try to post some ETC's in a room I just measured a couple weekends ago. That room had some fairly good acoustics, which makes the measurement a lot easier to interpret and describe. A bad room will be so chaotic crazy that it barely makes any sense at all.

As far as the scales for posting plots...I have always felt that you should choose the scale based on the information you're trying to present. So you're probably going to want to pick a scale that shows the initial energy decaying down into the noise floor. As far as the vertical-axis, I would use a dB scale and normalize the data. All that matters is the reflection intensity relative to the initial signal.

[thewire]

So then why can't we look at the REW window for a meter rather than when I move my mouse ever so slightly it is more like 10 meters?

[brucek]

So then why can't we look at the REW window for a meter

You can. Just zoom in the X-Axis. It will zoom to a time resolution far greater than you would ever require.

brucek

[thewire]

For some reason each time I try to set the axis in the windows settings it always goes to fully unzoomed. When I zoom in with my mouse I can't see the top of each to measure between. Is there a shortcut key to zoom only X-Axis?

[brucek]

When I zoom in with my mouse I can't see the top of each to measure between.

You've confused me for sure...

ETC is no different than any other graph with regard to zooming.

Just click the plus or minus zoom feature and then use the slider to position where you want to look (as shown in the pic below).........




brucek

[thewire]

Yes that works thank you. I was trying to use the Graph Limits tab and had some difficulties before.

[jaspal kallar]

Hello again

I've added acoustic treatments. First I added some Echo Buster Corners and a about one week later I added some Cathedral Sound Panels (for 0-200 Hz). As a reminder I have a 2 channel system and no subs.

I attach below the some waterfall graphs at the different stages. I'm still looking at the ETC graphs. Of course I've used my ears to hear how the sound has changed also but I wanted to see if anybody can see any (great) differences between the graphs below?


First no treatments:




Next, 4 (one in each corner) Echo Buster Corners:



Lastly, 4 (one in each corner) Echo Buster Corners and 4 (one in each corner) Cathedral Sound Panels:





Thanks

jaspal

[brucek]

To look for changes in waterfall plots, use its overlay function to compare two at a time.

brucek

[Wayne A. Pflughaupt]

Hi Jaspal,

There is no appreciable difference in these waterfalls. That's not necesarily a cause for disappointment, however.

The Echo Busters are designed to prevent upper-frequency reflections. As such, any improvement they get you will not show up on a 15-200 Hz waterfall. For that, a full-range reading would be better. The treatments should show less comb filtering and/or better RT60 times.

As for the Cathedrals, it often takes a lot of large bass traps to show a difference in a waterfall. So apparently you don't have enough of them at this point to register an improvement.

Regards,
Wayne

[jaspal kallar]

Hi Jaspal,

There is no appreciable difference in these waterfalls. That's not necesarily a cause for disappointment, however.

Hi Wayne

You confirm what I see in the three waterfalls - hardly a difference. Bruce I've did the overlay as you suggested and again hardly any difference.

The Echo Busters are designed to prevent upper-frequency reflections. As such, any improvement they get you will not show up on a 15-200 Hz waterfall. For that, a full-range reading would be better.

Wayne by For that, a full-range reading would be better" do you mean I should see what is seen under the "Measure" tab? If so is it useful to click the "Separate traces" box when looking ?

The treatments should show less comb filtering and/or better RT60 times.

I can attach the RT60 times although the discussion so far in this thread suggests it's not that useful in small rooms.

Thanks,

jaspal

[Wayne A. Pflughaupt]

I can attach the RT60 times although the discussion so far in this thread suggests it's not that useful in small rooms.

Yeah, I realized that after I dropped that post...

Wayne by For that, a full-range reading would be better" do you mean I should see what is seen under the "Measure" tab? If so is it useful to click the "Separate traces" box when looking ?

You'll have to change the parameters and take a new reading.

Under the Target Settings tab to the left of the screen, change the setting from "Subwoofer" to "Full Range.

Under the Graph Limits icon, top right of screen, change the "right" axis limit to 20,000.

Under the Settings icon, top left of screen, click the "Soundcard" tab, then toggle the "Levels" box from "Use subwoofer" to "Use main speaker."

I think that's it. brucek will correct me if I left out something. (I haven't done much full-range measuring).

Generally the Radio Shack SPL meter (if you're using that) is unsuitable for full-range readings, but you're ot shooting for pinpoint accuracy, just trying to do is see if your treatments have improved comb filtering. So, if you can do "with" and "without" treatment readings, that would be the most beneficial for you.

Regards,
Wayne