RealTraps - Optimizing Acoustic Treatment using the ETF software.

Exclusive for AcoustiSoft.com

..USING ETF EFFECTIVELY..

More RealTraps Articles

 

 

"I separate room acoustics into two distinct frequency ranges, with the dividing line around 300 Hz."

By Ethan Winer

I'll admit up front that I'm a huge fan of the ETF software. As a manufacturer of acoustic treatment, I need to assess different room treatment choices and placements, and measure the performance of our products in real-world applications as accurately as possible. I eat, sleep, and breath acoustic treatment so you don't have to, and ETF delivers for us in spades. I'll start by describing the ETF features that are most important to me, and then explain how we use ETF at RealTraps and how it helps us to create high performance acoustic treatment products.

Added 11/23/2007: For Mac users I recommend FuzzMeasure.

I separate room acoustics into two distinct frequency ranges, with the dividing line somewhere around 300 Hz. At low frequencies, the main problems are a badly skewed frequency response - numerous peaks and deep nulls - and modal ringing. Modal ringing is a type of reverb that causes some bass notes to sustain for longer than is on the original recording. Where normal reverb affects all frequencies more or less evenly, modal ringing occurs only at frequencies related to the room's dimensions. At higher frequencies, the main acoustic problems are single echoes that harm imaging and clarity, and repetitive echoes - called flutter echoes - that impart a specific pitch onto the music. If you clap your hands in a room that has large bare parallel surfaces, you can hear the pitched "boing" sound that is characteristic of flutter echo. The goal of acoustic treatment, therefore, is to eliminate all of these problems, or at least reduce them as much as possible. Top

 

 

"Third octave analysis and correction has been abandoned by most professional acousticians simply because it doesn't work very well."

 

Click to see a larger version
Figure 1 - This is the kind of low frequency response you can expect in a room without any bass traps. Click the image for a larger version.
A BRIEF HISTORY OF ROOM ANALYSIS

In the old days the most common way to measure a room's response was to play pink noise through the loudspeakers and measure the frequency response at the listening position in third-octave bands using a high quality calibrated microphone and sound level meter. If the measured response was found to deviate from flat, a third-octave equalizer would often be added to the playback chain to compensate for the disparity. This practice was common in recording studio control rooms from the 1970s through the early 1990s, but it has since been abandoned by most professional acousticians simply because it doesn't work very well.

There are two basic problems with third-octave frequency response measurements and EQ correction, especially in smaller rooms that you'll find in most homes. One problem is that the low frequency response can vary wildly, with numerous peaks and deep nulls all within a range narrower than one-third of an octave. The graph in Figure 1 at left shows the response I measured in the RealTraps test lab when empty. This room is about 16 by 11-1/2 by 8 feet, and you can see the horribly skewed low frequency response that's typical of all untreated rooms. Look at the range centered around 164 Hz, where there's a peak, a null, another peak, and another null, all very close together. It is impossible to see this response with third-octave analysis, and likewise impossible to correct using a third-octave equalizer because the only available frequencies are 125, 160, and 200 Hz.

The other big limitation is that frequency response is only half the story with room acoustics. The other half is timed-based problems - echo, reverb, and ringing. For example, if a room has a reverb time that's longer in the range from 500 to 1,000 Hz than at other frequencies, that range will sound louder simply because frequencies in that range sustain longer. More energy is present in that range even though the measured level is no higher. So if you add an equalizer to the monitoring path, you'll reduce the level below what it should be, and the excess reverb that clouds the music will still be present. Top

 

 

"ETF measures to a resolution of 0.7 Hz, which is mandatory for assessing low frequency response."

 

Click to see a larger version
Figure 2 - The graph above is derived from the same data shown in Figure 1, but this one also shows modal ringing and its decay time. Click the image for a larger version.
ENTER ETF

ETF solves these limitations because it not only measures raw frequency response - and to a much finer resolution than one third octave - but also displays time-based phenomena. Frequency response is measured to a resolution of 0.7 Hz, which is mandatory for assessing low frequency response. ETF also shows modal ringing, individual reflections that harm imaging, ringing caused by flutter echo, as well as third-octave reverb time across the entire audio range. The graph in Figure 2 is for the same empty lab room as above, but this one is shown as a low frequency "waterfall" plot that reveals the modal ringing. In this type of graph, the "mountain" peaks come forward over time.

Now you can clearly see that each peak in the response is also accompanied by ringing that sustains for longer than half a second at low frequencies. This ringing is a huge problem because it causes some bass notes to sustain and overlap subsequent notes. This results in a muddy and boomy sounding low end - an effect known as "one note bass" - because no matter which bass note is being played, the room's ringing sounds at the same frequency (or frequencies). Using a test tone CD or less capable analysis program with a sound level meter measures only the frequency response, and completely hides the ringing. Clapping your hands creates mostly midrange frequencies, so that won't reveal this problem either.

Another great feature of ETF is its ability to show individual reflections over time. Because you can see the time delay between the original and reflected sounds, you can more easily determine the source of the reflections. For example, each millisecond of delay represents about one foot of distance. Another feature of ETF is being able to see comb filtering, a different but equally important artifact caused by reflections. Examples of using ETF to identify early reflections and comb filtering are shown on the ETF web site, so I won't belabor those here. Top

"We place enough absorbers in our test room to make a substantial change in decay time at the frequencies of interest - the design that reduces the decay time the most wins."

 

 

 

 

 

 

 

 

 

 

 

 

"My personal philosophy is to add as much bass trapping to a room as you can possibly manage."

USING ETF TO OPTIMIZE ACOUSTIC TREATMENT

At RealTraps we use ETF to determine the efficiency of various bass trap and other absorber panel designs. We simply place enough absorbers in our test room to make a substantial change in decay time at the frequencies of interest and compare one design with another. The design that reduces the decay time the most wins. This is not the same as a formal test performed in a real acoustics lab that yields certified absorption data, but it's quite useful for comparing the efficiency of one absorber versus another.

You can use ETF in a similar way to compare the improvement from different trap placements in your own room. The effect of even small changes in placement - too small to necessarily be audible - can be easily seen. Besides flattening the low frequency response and reducing modal ringing, bass traps also increase the bandwidth of modal peaks. This is an important feature because broad peaks are not as damaging musically as peaks that boost a very narrow range of frequencies. Narrow peaks exaggerate the effect of making all bass notes sound like they have the same pitch.

Besides showing exactly what's happening in a room at low frequencies, ETF shows reverb time over the entire audible range in third-octave bands. Reverb time is also called RT60 because it's defined as the length of time it takes for the reverberant sound to decay by 60 dB. If you know the recommended RT60 times for different room uses, ETF makes it easy to tell when you have the right amount of absorption. For example, in a room primarily used for two-channel music, the ideal RT60 is somewhere under 0.5 seconds over as wide a frequency range as possible. However, in home theaters it's common to have even less reverb because movies usually have the desired amount of ambience already added to the sound track by the mix engineers. So for a home theater, an RT60 of 0.3 or even 0.2 seconds is a common target. Top

Note that the measured - and actual - reverb time rises at very low frequencies. Much of this is due to modal ringing explained above. At higher frequencies furnishings in a room, such as plush carpet and soft upholstered chairs and couches, contribute to reducing the overall reverb time. This makes it difficult to predict the reverb time in advance, making accurate measuring using ETF all the more important. Also note that much of the "reverb" in small rooms is really a smaller number of individual reflections. To be considered true reverberation the reflections need to build over time and fuse into a coherent "haaah" sound and, likewise, decay over time. In rooms the size you'll find in most homes, most of the decay is either individual reflections off a few nearby surfaces or repetitive flutter echoes that sustain between two parallel surfaces, such as two walls, or between the floor and ceiling.

To measure reverb time accurately with ETF, you need to determine where the decay slope is most linear and tell ETF to consider only that range. Here are the steps:

1. Run a new test or load a previous test.
2. Select the Energy Time Curve display.
3. Increase the X/Y Axis Limits so the Horizontal Limit displays 5-200 milliseconds.
4. Set the Gating dialog to also display the Schroeder plot.
5. Identify the longest contiguous linear region on the Schroeder decay.
6. Switch to the RT60 display.
7. Set the Time Limits to use the linear Start and Finish times noted in Step 5.

Now you can see the reverb time in each frequency band, though as explained earlier, below about 200 Hz the decay times displayed represent modal ringing at specific frequencies more than true reverb. Top

Click to see a larger version
Figure 3 - This graph shows how low a reverb time we were able to achieve using bass traps in the RealTraps showroom. Note the uniform reverb time versus frequency. (The sudden rise at 20 Hz is an artifact.) Click the image for a larger version.

 

Click to see a larger version
Figure 4 - This graph shows the ringing and low frequency response in my living room with 38 traps. Click the image for a larger version.

 

 

 

 

MondoTraps rule!
This MondoTrap from the author's company RealTraps is effective down to about 40 Hz. In a large room you can use EQ to reduce modal peaks at even lower frequencies if needed. Click the image for a larger version.
TREATING THE ROOM

Now that we've examined how to measure the room at both low and mid/high frequencies, the next step is to treat the room and measure the effectiveness of that treatment. My personal philosophy is to add as much bass trapping to a room as you can possibly manage. At mid and high frequencies you can definitely have too much absorption, which makes a room dead sounding and unnatural. But I don't think it's possible to have too much absorption at low frequencies.

I have 38 RealTraps bass traps in my fairly large home theater (the RealTraps showroom), and as each round of traps was added the low end continued to become flatter and tighter sounding. One big feature of RealTraps products is that they absorb a huge amount at low frequencies, with intentionally less absorption at higher frequencies. So even with that many bass traps, my home theater is not too dead sounding, having an RT60 of about 300 milliseconds over most of the audio range. Figures 3 and 4 show the short reverb time, wider modal peaks, and reduced ringing at all but the lowest frequencies that I was able to achieve in this room. You can clearly see the enormous improvement in flatness, peak bandwidth, and ringing compared to the untreated room shown in Figure 2.

Often, concerns over the appearance of bass traps and acoustic panels dictate how much treatment can be added. Although I'm not a fan of active room correction devices, such as equalizers and DSP processors, these can be useful when it's not possible to install a sufficient number of bass traps. In particular, equalizers can reduce the severity of modal peaks at the very lowest frequencies. However, they can reduce only the level of these peaks, not the longer decay times associated with modal ringing. Likewise, EQ and DSP devices cannot replace absorption for reducing the effects of early reflections, comb filtering, and excess ambience. Top

Note that EQ is most appropriate when used only for the lowest frequencies, that is, the primary mode frequency for each of the three room dimensions. This is also the range where most bass traps have less efficiency (though the bass traps I manufacture can help at frequencies down to 40 Hz and sometimes even lower). And even if EQ won't reduce ringing at the lowest frequencies, there's little musical content there anyway. If a cannon shot or rocket blast from a DVD movie lingers for an extra half second, that's not nearly as damaging as ringing that affects bass notes from a musical instrument.

Finally, I'll explain an interesting problem that happens when relying on low frequency response only for assessing the effect of placing bass traps. Sometimes moving a trap to a better location, or even adding another bass trap, can appear to make the response worse even though this is not really the case. The low frequency response at any given location in a room is the sum of the direct sound from the loudspeakers plus a large number of competing reflections from all of the room's surfaces. Some combinations create peaks, and some create nulls, but sometimes a reflection that would have made a null combines with one that would have made a peak. So together the result is somewhere between the two. If a trap reduces a reflection that had negated a null caused by a different reflection from somewhere else, the null will then appear. The null-causing reflection was always present but was partially cancelled by another reflection that has now been trapped. This is another reason that ETF's waterfall plots are the best way to assess the improvement by adding or positioning bass traps. With this type of graph, adding traps will always show the ringing being reduced, even if the raw response happens to be worse where you measure. Top


Ethan Winer heads RealTraps in New Milford, CT, where he designs and manufactures high-performance acoustic treatment.

Entire contents of this site Copyright 2004- by RealTraps, LLC. All rights reserved.
No portion of this site may be reproduced in any form without the express permission of the copyright holder.