Archive for March, 2010

Project 1

March 16, 2010

This is my webpage for Project 1 of Sam McGuire’s Graduate Audio Seminar Class.  For my project, I wanted to do something that would require me to beef up on my knowledge of acoustics, specifically acoustic testing using Smaart Live RTA software.  I wanted to do this in anticipation of my upcoming internship at Walters-Storyk Design Group, because I feel that having a solid baseline of acoustics knowledge and experience will help me immensely.

With that in mind, I decided to do some acoustic testing of Studio D in The Core.  I chose Studio D per Sam’s recommendation, as it is widely known as a poor sounding studio, in need of serious acoustic treatment.  I will talk about my testing process, my results, and finally some recommendations for improving the sound in Studio D.

First, I needed to setup my acoustic testing equipment to properly analyze the room.  Below is a block diagram copied from Smaart Live’s instruction manual, showing how to properly set Smaart Live up for doing real time frequency analysis.

For the measurement microphone, I used an Earthworks MC30, for its flat frequency response and omnidirectional pickup pattern.  The microphone preamp was a M-Audio MobilePre, which I could use to interface with my computer, a laptop running Windows 7.  For a reference signal, I used pink noise generated from the internal pink noise generator in Smaart Live.

I also needed to split the signal back into my computer, because to compute a transfer function (the difference between the reference signal and the measured signal), you need both the measured pink noise in the room, as well as a baseline measurement of pink noise to compare it to.  I accomplished this by using a microphone splitter to route the pink noise out from Smaart both to the Mackie console, and also back into my MobilePre, using a line-level input.  This allowed me to simultaneously measure the pink noise coming out through the speakers through the microphone, as well as measuring the pink noise directly.  In this picture, you can see the entire setup:

from left to right: the microphone, the mic splitter, the MobilePre and my laptop.

After getting everything setup, it was time to get busy.  I started the pink noise generator in Smaart Live, and observed the real-time frequency analyzer on my laptop.  Here you can see the real-time analyzer, each bar represents 1/12th of an octave.

The green graph represents the measured pink noise, coming out the speakers and into the measurement microphone.  The blue represents the pink noise being routed directly out of the signal generator and back into the laptop for comparison.  As you can see, the two graphs differ significantly.  As expected, the pink noise measured directly back into the preamp is relatively flat, as pink noise has equal energy in each octave.

However, the pink noise measured through the microphone is quite uneven.  These graphs tell us a little bit, but to get the full story, we need a single graph that shows us the difference between the two measurements.  This is what is known as a transfer function, a graph that gives a visual representation of the difference between the measurements.  For the rest of this report, I will be referring to these.

Here you can see what I am going to refer to as the “baseline” transfer function.  This was made with the microphone in the listening position (marked by the ‘x’ on the floor), with the microphone at ear level.  All acoustic treatments that are usually in the room were in their normal spots, which include a membrane absorber on the back and side walls and two bass traps in the rear corners.  For the rest of my graphs, I will be changing aspects of the room, including the mic position, the location and presence of the bass traps, and introducing baffling, and comparing the results to this baseline graph.

Baseline

As you can immediately see, this graph is quite far from being flat, which would indicate a well-tuned room.  There are major peaks around 80 Hz, 110 Hz and 300 Hz.  There are large troughs around 1 kHz and 2.5 kHz.

The first thing I wanted to try was to setup a baffle across the small “extrusion” where the doorway is.  I wanted to see if there would be a large difference if I made the room more rectangular.  Here is the result, superimposed on the baseline graph (the baseline graph will always be the same green color in all of my pictures).  Below the graph is a picture of the baffled doorway.

Baffled Doorway

The graphs are relatively similar, leading me to believe that the doorway was not affecting the sound too much, and for all intents and purposes, the room was acoustically rectangular.  One thing to notice though, is how the dip at 2.5 kHz is greatly reduced, leading me to believe that those frequencies may be getting trapped in that doorway.

Next, I wanted to test the room without the bass traps in the corners, to see if they were having a noticeable effect on the room response.

Baffle Doorway No Bass Traps

The bass traps do not appear to have much of an effect.  I don’t think they are large enough or constructed out of the correct materials to appropriately dampen the problem frequencies.

Next I wanted to explore whether or not more high frequency absorption may be needed on the rear wall behind the listening position.  Additionally, it seems to me that the existing absorber is placed about a foot too high, if you trace a line from the monitors to the rear wall, the sound hits the wall about 6 inches below the absorber.  I brought in two extra baffles and placed them under the absorber as shown in the picture below

I hoped that this would at least control some of the early reflections and reduce comb filtering at the listening position.  Additionally, I tried moving the microphone about six inches forward to further reduce the effect of reflections off the back wall.  This is the resultant graph:

Mic Six Inches in Front of Listening Position

Interestingly, this not only got rid of the huge dip around 2.5 kHz, but also seemed to level out the high frequency somewhat.  Although there was also a dip introduced around 2 kHz, which leads me to believe that there are many unwanted interactions of high frequency waves as they bounce off untreated surfaces in the room.

I also wanted to try moving the mic back from the listening position to see what effect that might have.

Mic One Foot Back From Listening Position

This also seemed to level out the high frequencies somewhat, severely lessening a couple of the major nulls in the baseline graph.  This reinforces my belief that there is some serious comb filtering occurring at the listening position that needs to be addressed with high frequency absorption.

Just for kicks, I experimented with placing the mic in front of the left and right speakers, to see if that might lessen the effects of the comb filtering resulting from reflections off the back wall.

Mic in Front of Left Speaker

Mic in Front of Right Speaker

Clearly, there are still issues even here, but on the whole, the high frequency response is somewhat flatter.

The final graph I would like to post is a composite of all the different acoustic tests I did, the ones I’ve talked about so far, plus a few other tests I did.  I think this gives a great, overall picture of the room.

I think this graph illuminates a couple things about Studio D.  First of all, there are certainly two very large room modes, one around 110 Hz and one around 300 Hz.  Most experts say that room modes don’t affect frequencies above 250 Hz or so, but the consistency of that frequency spike around 300 Hz really couldn’t be the result of anything but room modes.  Since these measurements were taken at different locations around the room, with varying amounts of treatment, it is hard to speculate that those consistent frequency spikes are the result of anything but room modes.

As you can also see, the high frequency response (above 1 kHz) is very uneven, fluctuating wildly between tests.  This  is almost certainly due to a lack of high frequency absorption in the room.  Additionally, there are a number of highly reflective surfaces in the room, such as the two large panels of glass, the console and the furniture that reflect sound in many directions.  Comb filtering is creating many large peaks and valleys throughout the high frequencies.

Summary

Probably most of my recommendations are going to seem pretty obvious based on what I’ve already noted, but I will summarize them anyway.  First of all, Studio D is in dire need of effective bass trapping, to reduce or eliminate those huge room modes around 110 and 300 Hz.  The current “bass traps” do almost nothing as you can see from this graph I made that has the room configured the exact same way, but with the bass traps in the corners in one graph and no bass traps at all in the other graph.

Bass Traps vs No Bass Traps

Obviously, the current bass traps are woefully inadequate.  I would suggest at the minimum, replacing those existing bass traps with ones that are constructed out of more effective trapping material, such as fiberglass.  Additionally, they should be much larger, stretching from the floor to the ceiling, and triangular shaped as to fill in the entire corner.  This room might even need more than that, including possibly other bass traps along the floor of the rear wall, in the doorway, and even a “cloud” hanging above the listening position.  At the very least, I think that these additions would flatten out the low frequency response enough to make the studio acceptable for mixing.

The other major problem is lack of high frequency absorption.  I am not sure how effective the existing hanging absorbers are, as I couldn’t remove them to test the room with them absent.  However, I feel like they are probably not helping that much.  I think the entire rear wall should be covered with either high frequency absorptive materials, or QRD diffusers that would scatter the sound coming from the speakers instead of reflecting it directly back at the listening position.  I think that would go a long way towards evening out the high frequency response across the spectrum, but additionally a case could be made for adding high frequency absorption to other surfaces in the room, including the side walls and ceiling.

So, if I were on a budget and needed to prioritize what to do in Studio D, I would apply treatments in this order:

1) Large bass traps in the rear corners

2) High frequency absorption/diffusion on rear wall

3) Added bass traps hung from ceiling, or along floor

4) High frequency absorption on other surfaces


Follow

Get every new post delivered to your Inbox.