Part 1 covered the design, construction, and initial installation of existing equipment in the home theater.
Part 2 covered “The Great Upgrade” with the selection of new speakers, a 2nd subwoofer, a new preamp processor, amplifiers and source components, remote control, a 4K projector, a new screen, a new cabinet, and Dolby Atmos.
In Part 3 we will cover acoustic treatments used in the room, measurements of room acoustics using Room Eq Wizard software, subwoofer placement, room modes, and setting up a house curve or a customized frequency response using Anthem Room Correction.
As mentioned above, getting the acoustics right is critical for the full enjoyment of the home theater experience. It starts with room dimensions and ensuring that no two dimensions have a common divisor. Without going into too much detail, certain bass frequencies “fit well” into a certain dimension and will be reinforced compared to other frequencies. These are called room modes. Ideally, these room modes would be well spread out rather than being bunched up and having the same bass frequencies excessively reinforced by more than one equal dimension or multiple thereof. There are lots of web-based room simulation sites that will predict the room modes for a given set of room dimensions, or one can consult graphs like the one on the right that shows several of the more ideal room ratios. My targeted room dimensions were essentially near the point labeled 3 on this graph, but when we changed the shape of the front window in the room from an arch to a pentagon, it raised the ceiling height from 9’ to 10’8” and moved my point on this graph down a bit below the point labeled 2, but still close to the favorable area.
In my case, the ceiling dimension may not be critical as it is not a hard, reflective surface but is made from porous expanded polypropylene plastic beads. The panels came from the company, Acoustical Surfaces, and are said to offer sound blocking, sound absorption, and sound diffusion all in one unit. The porosity of the material and all the exposed edges serve to not only absorb sound but also to scatter or disperse sound hitting the ceiling. Above the acoustic panels are fiberglass insulation and unfinished attic space.
Good home theater acoustics require more than just placing a few sound-absorbing panels to kill a few critical reflections. It involves controlling the overall reverberant environment in the room with a balance of absorption and dispersion. If the room is too dead, with too much sound absorption it can sound very unnatural. We are used to hearing some reverberation from our voices and other noises in the room as that gives us clues to the space around us. If you have ever been inside an anechoic chamber when the doors are shut – well it can freak you out.
When a sound comes out of a loudspeaker, e.g. dialog from the center speaker, the first thing you hear is the direct sound from the speaker followed by all the reflections of that sound off of the floor, ceiling, and walls. How we perceive these reflections or echoes, depends on the timing, amplitude, and direction of the various reflections. A strong echo at the wrong time can interfere with intelligibility. On the other hand, if that echo is partially absorbed and broken up or dispersed into a train of lots of little echoes at different times and from different directions, then one gets a sense of spaciousness and intelligibility is fine.
The tricky part is making sure that the sound absorption and dispersion happen more or less equally across a critical range of frequencies. It is easy to absorb very high frequencies and much harder to knock down lower frequencies as their wavelengths are much longer and it requires thicker materials to absorb them. So, one can’t just put a thin piece of cloth on the wall at a critical sidewall reflection point and call it a day. If you are going to try and absorb critical reflections it must be done with fairly dense and thick, e.g. 2” material to absorb the frequencies relevant for dialog. Deep bass is a very different matter that will be dealt with in detail below.
If you look at various pictures in this blog so far, you might notice tapestries on the wall and think that is all I did to kill the critical reflections from the sidewalls. If you look carefully, the tapestries are at eye level, and below and partially behind those tapestries, at ear level, is a 2” thick and dense 2’x4’ ATS panel specifically designed to absorb the frequencies of interest. I added an extra layer of loose weave fabric for color. What you don’t see is that behind the tapestries on top of the ATS panels are the same expanded polypropylene plastic bead panels with balanced absorption and dispersion that are used on the ceiling. So, any sound that goes through the tapestry will be partially absorbed and dispersed before reflecting back into the room.
I should also comment on the foam columns in the front corners. While they are often sold as “bass traps”, they are really not thick enough or dense enough to do much bass trapping. They are largely there as a removable structure to hide the various cables that run from the equipment rack, up the corners, and through the soffits to the surround and Atmos speakers, as well as the projector. I am sure they do some good in the treble and midrange, but they are not bass traps.
Some readers may have noticed that the new final paneling on the soffits is more than just a layer of nice plywood but is broken up by darker pieces of Khaya with truncated pyramids from Lacewood on them. These not only add visual interest, but they also can be removed to gain access to inside the soffits if needed, and may also help diffuse the sound as every edge breaks up echoes and scatters the sound, which is a good thing.
No room has perfect acoustics, or at least no room in which I would want to sit or have dinner! Fortunately, most all of today’s receivers and pre-pros meant for home theater usage have what is called “room correction” software that will play some test tones through the various speakers and use a microphone to record the resultant response of the speakers and the room. The pre-pro will analyze that response and apply various filters to bring the sound as close to the desired response as is practical. Room correction should ideally be used as a final tweak of the sound in a well-conditioned room, rather than as a tool to correct a difficult or acoustically untreated room. My AVM 60 uses Anthem Room Correction or ARC, one of the first room correction systems available to consumers, and one that is still well-regarded today.
ARC has now been upgraded to ARC Genesis and while ARC required the use of a Windows PC, ARC Genesis blessedly can be run on a Mac, iPhone, or iPad. Since I am a Mac guy this was huge. ARC Genesis, or from now on simply ARC, when run on a Mac or PC, allows users to easily define their own “house curve” or customized frequency response, another plus.
ARC works by sending test sounds through each speaker, it then analyzes the results and calculates a correction curve that once uploaded to the AVM 60, corrects what it sees as deviations from the user-defined desired target frequency response or house curve. It also recommends a crossover frequency for that speaker to the subwoofer(s).
ARC tests and corrects all the speakers, including the subwoofer(s), but it does not play the speakers and subwoofer together, as happens when actually listening to music or watching a movie. Getting that interaction right is a bit tricker in that one needs to get the timing of the subwoofer right relative to the other speakers so that the sound from the various speaker and the subwoofer arrives at the listening position(s) at the proper times. This is typically done by tweaking the distance of the subwoofer in the pre-pro, the AVM 60 in my case. This process will be covered in some detail later on in this section using REW, or Room EQ Wizard, a software tool that we can use not only accurately set the subwoofer distance, it can also analyze the acoustics of a room. How well do my various acoustic treatments work in the critical midrange? How well did my choice of room dimensions and subwoofer placement deal with room modes? REW can answer these questions.
Room EQ Wizard, or REW, is one of the most amazing pieces of software I have used, and it is free! One does need to buy a USB calibrated microphone like the miniDSP UMIK-1 for about $100. REW is similar to ARC in that it plays test tones through the various speakers and analyses the results, but in a much more detailed way than ARC. I used to work in the scientific instrument business, and among other things, managed a software department that wrote software to do similar analysis, via a Fourier transform, from those instruments. I know what is involved in analyzing and displaying the results of the testing, and I am impressed with REW. Did I mention that it is freeware?
Above is a screenshot of the REW interface showing the frequency response of my center channel and subwoofer working together. This trace shown is unfiltered, i.e., it is the raw room response. If you are worried about the very jagged nature of this trace, just move your head or the microphone a few inches and it will be different.
That is why REW has a number of different filtering algorithms that allow one to see a trace that represents much more of what we actually hear. The trace above represents Psychoacoustic filtering of that same measurement and shows that the overall response is ± 3dB, but with an obvious slope, called a house curve, that has a bit of a bass boost and a smooth treble roll-off. More about the house curve later.
REW not only measures a speaker and room’s combined frequency response, but it also can look at how those frequencies decay in time, and how even that decay is with respect to frequency. This is how we answer the question about the effectiveness of the various acoustic treatments in the room as well as how to optimize subwoofer placement.
When I was working on the home theater in Redwood City, about the only acoustical measurement I could use was T60, the time that it takes for a sound impulse in the room to decay 60 dB or a factor of 1000 times the measured Sound Pressure Level (SPL). It is very hard to measure T60 directly as the decaying signal typically hits the noise floor of the room before 60 dB of decay. However, T60 can be extrapolated from measuring the time the sound level drops by 20 dB and 30 dB (T20 and T30). For home theaters, a good value of T60 is between 200 and 500 milliseconds or a few tenths of a second, and ideally, the values of T60 will be within ±25% for both the T20 and T30 extrapolated T60 values between 250 Hz and 4 kHz.
I used to do this sort of thing with a microphone and a digital oscilloscope and I would get one number for the overall T60 of the room. But REW uses the same measurements used to determine the frequency response of the system to also derive the decay rate. The RT60 tab in REW provides a graph of T60 as extrapolated from T20 and T30 as a function of frequency. From the graph on the right, one would conclude that the room response in the critical frequency range in question meets the suggested criteria and therefore the acoustic treatments in the room are doing their job, at least in the midrange. Certainly, more absorption or sound deadening is not needed. We will look at the bass frequencies next. You can read more about T60 here: http://www.acousticfrontiers.com/20111013acoustic-measurement-standards-for-stereo-listening-rooms-pu-html/
When it comes to bass frequencies, one is also concerned with decay rates, as if the bass at a particular frequency builds up in the room, perhaps because of excitation of a room mode, the bass can sound bloated or boomy, rather than tight or accurate. Low-level sounds will be obscured by any ringing or the slow decay of bass frequencies. Read more about room modes here. http://www.acousticfrontiers.com/room-modes-101/
In REW, one normally looks at bass decay via a waterfall plot where hopefully one can see all frequencies decaying to the noise floor in less than 300 or 400 ms. What one is looking for here is a ridge, or hopefully the lack thereof, with a long decay rate relative to the other frequencies. Such a ridge would likely indicate that a room mode has been excited. The waterfall plot shown above is from an early measurement with a single subwoofer in the front corner of the room. It shows a ridge at ~52 Hz that does not decay as quickly as other nearby frequencies, causing that particular bass frequency to be overemphasized as the sound decays or to “ring”. Not good.
Room modes occur because certain bass frequencies fit well in the distance between two parallel surfaces in a room, and that frequency resonates or is reinforced by that fit, relative to frequencies around it. I carefully chose room dimensions with magic ratios to have the frequencies associated with the various room well-spaced with minimal overlap. However well-spaced, room modes still exist and can affect the smoothness of the bass response of the room if strongly excited.
One very useful tool for understanding room modes is the calculator by amcoustics.
The figure above is from that calculator; and with my room dimensions, one can see a reasonable spacing of room modes as indicated both from the room mode frequency chart and the Bonello chart. This Bonello chart shows the number of modes per third octave beginning with the lowest mode. According to the “Bonello-criteria”, this function should be increasing with each third octave to indicate a good distribution of modes.
Based on my room dimensions, this simulation indicates that I should have a length axial room mode (1-0-0) at ~30 Hz, a width axial mode (0-1-0) at 40 Hz, and a height axial mode (0-0-1) at 53 Hz. There will also be a higher-order axial mode, e.g. 2-0-0, tangential modes involving two pairs of parallel surfaces (1-1-0), and oblique modes (1-1-1) involving all three pairs of surfaces. All are shown in the frequency chart above with the various axial modes taller than the tangential that are taller than the oblique modes.
In the figure above, I have placed my cursor on the mode at 40 Hz and one can see from the Room 3D illustration that the 0-1-0 width mode has high pressure on the side walls. As can be seen from the smaller figure (left) from Acoustic Frontiers, there is also a null or node in the center of the room. If a subwoofer is placed on the sidewall this mode will be strongly excited. As this mode minimizes the bass response on the centerline of the room it caused me grief early in my experience with room correction via ARC as will be discussed below. Hint, you don’t want to sit or measure in a null or node.
So, what room mode is responsible for the ringing at 52 Hz seen above? There are two candidates close to that frequency in the simulation, a ceiling mode at 53 Hz and a tangential mode (1-1-0) at 50 Hz involving both the front/back wall pair, and the side walls. With the (1-1-0) width tangential mode shown above, one can see that the high-pressure areas are in the corners of the room and a subwoofer placed there will likely excite that mode. While tangential modes are often less of a problem than axial modes because they involve more surfaces and are 3 dB less in pressure, they do exist and can be excited if a speaker is placed in one of the corners.
From my testing with a single subwoofer, the ringing of the 52 Hz mode was excited when the subwoofer was placed in the corner but not when the subwoofer was pulled forward in the room 3 or 4 feet. More importantly, that ringing was essentially canceled once I put the second subwoofer near the other front corner. How does that work?
In the graph from Acoustic Frontiers above there are + and – signs associated with the two high-pressure areas on the sidewalls, and the figures from Amcoustics have red and blue areas, both indicating that the various high-pressure areas, + and -, and red and blue, are opposite in phase. If two subwoofers of the same phase are placed, one in a + area and one in a – area, then the room mode in question will be essentially canceled. Does that mean that there is no sound at that frequency? Not at all. Both subwoofers are still playing in phase so they won’t cancel each other at the listening position, but since they are in opposite phase areas of the room mode, the amplification of that modal resonance by the room is canceled. This is very well explained on the Acoustics Frontiers website, and it certainly worked for me.
But, you say, the simulation says the room mode is at 50 Hz and the ringing you measured is at 52 Hz. Is that a problem? I don’t think so. The room mode calculations are based on the actual sheetrock to sheetrock dimensions of the room but don’t include the effects of the soffits, cabinet, furniture, and especially the columns in the back corners that no doubt make the room sonically feel smaller for this mode than the actual distance between sheetrock walls indicate. It is also pretty clear that it is not the height mode, as moving the subwoofer out of the corner and/or use of a second subwoofer would not cancel that floor/ceiling mode. And I would suspect, or hope, that this mode would be pretty well eliminated anyway by the nature of the ceiling panels and space above.
With the dual subwoofers about three feet out from the corner (and equalization), the results are much better in terms of room modes ringing.
The spectrogram tab in REW shows the same data as the above waterfall, but in this tab, the color represents the intensity of the sound as it decays. In both the spectrogram and waterfall graphs above, the noise floor is a bit higher than it should be as I did not bother to turn off the AC, refrigerator, kegerator, etcetera, for those particular runs.
Above is an earlier run with the AC and refrigerators off and the noise floor is much lower, especially above 40 Hz.
I am impressed that without any actual bass traps in the room, the bass decay seems acceptable. Part of this is thanks to good room dimensions, good subwoofer placement and perhaps the ceiling helps with any modes that involve a high-pressure area at the ceiling.
Another useful graph in REW for looking at the time decay of the sound is the impulse response graph that shows the overall decay envelope of the signal. The graph above shows the first 50 ms of decay from the center channel speaker and subwoofer. What one is looking for here is any obvious spikes from reflections or echoes that are well above the “hair” or what seems to be the noise floor of the measurement. I was looking at the impulse response to see if I could see any obvious reflections off of the front window that is right 38 inches behind the center speaker, but such a reflection would show up as a spike at ~5.5 ms, and there is no obvious spike in that timeframe, so the heavy drapes are doing their job. What looks like “hair” however is not noise but rather is the reverberant field or a train of small echoes from all the surfaces in the room. This can be seen in the 300 ms plot below.
Yes, the decay does eventually reach the room’s noise floor, but only after 200 ms. Since the signal amplitude is shown in dB, or on a logarithmic scale, the apparent linear decay indicates that sound in the room is decaying as a single exponential function after the initial drop of ~ 20 to 30 dB. Perhaps this initial drop would be greater with more absorption and less dispersion in the room, but a single exponential decay after the initial drop is considered a sign of good acoustics according to F. Alton Everest’s Master Handbook of Acoustics, so I am not complaining.
I think all of the above measurements and discussion also illustrates how much the acoustics of a small room influence the sound we here from our speakers and how little of the sound is the direct first arrival sound from the speaker. While it is of interest to look at a speaker’s on-axis response in an anacoeic chamber, a speaker’s on and off-axis frequency reponse interacting with the various room surfaces and acoustic treatments also play a critical role in what we hear. That is also why speakers with the same on-axis frequency response, but different dispersion characteristics or off-axis response, e.g. vertical left and right front speakers and a horizontal center channel, might not create as seamless a front sound stage as 3 identical vertically orientated speakers.
Our goal was to achieve good acoustics in a room that didn’t look like a recording studio or the crew quarters on Babylon 5 with obvious hi-tech looking acoustic panels on the wall, and it seems as if we have achieved that, if you don’t look at the ceiling, that is. I am certainly not an acoustics expert, but I like how the room looks, sounds, and measures.
The goal of all these acoustic treatments and measurements is for the system to accurately reproduce the source material ideally without the equipment or room adding or subtracting anything. One would then assume that to achieve this, we want a ruler flat frequency response for 20 Hz to 20 kHz and beyond. Since I am in my late 70s, I don’t worry too much about anything above 10 kHz, but I do like to use a bit of what is called a house curve to add additional emphasis to the bass frequencies and tone down the treble a bit. I am certainly not alone in this; and that is why some room correction capabilities like ARC allow the user to specify a house curve.
So, let’s look at how one might use REW and ARC along with equalization in the two subwoofers to achieve this. While what I’ve outlined above in the section on bass decay rates and room modes, along with the section below hopefully seem like a logical progression of steps that I took, in reality, the path I wandered down was actually much more chaotic as I went back and forth between ARC and REW over a period of one year learning how to use each tool, and how they played together, not to mention, how to understand and tame room modes.
A logical first step in frequency response optimization and creation of a house curve is to use REW to get a sense of the response of the subwoofer in the room. The next step, in my case, was to use the Velodyne provided mic and test tones generated by the DD-15 and the DD-15’s parametric equalization window to get a rough correction of the subwoofer response. The DD-15 supposedly can do this eq automatically, but I chose to do it manually.
At this point, one can use ARC to take it most of the rest of the way. With ARC you start by defining one or more set of Measurements and which speakers are to be used for that measurement. In my case I wanted at least two measurement sets to cover different seating situations. I wanted a set of measurements that would use all speakers and optimize the response for the two front seats. I wanted another set of measurements that also included the back-row sofa seats. Thus, I could have two sets of room correction, one for just the front seats when it was just me and my wife watching TV and another when we have a movie night and the back row is occupied. (I assumed that if it was “just” me and my wife watching TV or a movie, the dogs on the sofa would not be concerned with sound quality.)
One then performs a series of measurements of a given Measurement category, e.g. the Front Row, using the calibrated microphone provided by Anthem with the AVM 60. And while a measurement run in REW uses a single microphone position, ARC uses multiple locations, seven in my case around the chosen listening area.
Once the measurement run is complete, one can create up to four profiles in ARC that define the overall frequency response Target i.e. the house curve for that profile, and the Measurement set to use. In this case, my profile 1 was for Front Row optimized movie viewing and I selected a Deep Bass Boost of 3.5 dB centered at 40 Hz and a treble Tilt of -3 dB starting at 800 Hz, creating a rising slope in the bass and a downward slope in the treble. Some people prefer a more extreme house curve, but this one sounds about right to me. In this target figure, one can also get some idea of the overall system response and the crossover frequency. When done specifying the various targets for the various profiles and calculating correction based on the specified Measurement, the ARC corrections are uploaded to the AV 60 for use.
I also created another profile for listening to records that did not use the height speakers or the subwoofers. Want to try several different house curves and see what is best? Or different crossover frequencies for the Atmos height speakers and switch quickly while listening to reference materials? No problem; just create a different profile for each case (up to 4) and then assign those profiles to different “Inputs” on the AVM 60. On the AVM 60, one can define a number of different “Inputs” that are the same physical input, e.g. HDMI 1, but have different characteristics, including the ARC profile. One can then quickly switch inputs while viewing and easily compare different profiles. That is how I found out that trying to cheat by using an artificially low crossover for my height or ceiling Atmos speakers, was not a good idea.
Above are the frequency response graphs for all the speakers in my system when finished with the ARC process. The Red trace is the measured response, the Black trace the target for that speaker, and the green line the predicted system response after ARC correction. Although everything looks great, the keyword in the last sentence is “predicted” or calculated. ARC does not measure the corrected response. That is why it is worth a final run or two with REW to confirm that the response is what it should be, AND to optimize the main speaker’s interaction with the subwoofer by tweaking the subwoofer distance setting.
With the AVM 60, one has to manually measure and enter the distances of all the various speakers into the AVM 60 setup menu. While some receivers and pre-pros do this automatically, Anthem obviously figures it is not hard for the user to use a tape measure to do so, and it is probably more accurate that way. The object of all this is to compensate for the time it takes for the sound coming from a certain speaker to reach the listening position. For instance, surround speakers are often closer to the listening position than the front speakers, and yet we don’t want echoes from the surround speakers to reach the listener before the direct sound from the front speakers. By knowing the distance of the various speakers, the pre-pro or receiver can adjust the timing of the sound coming from the different channels so that they arrive at the listening position with the proper timing.
The timing of the sound from subwoofer(s) is also very important in that at the listening position, the low bass frequencies from the subwoofer have to blend in with the higher frequencies from, e.g. the center speaker. The crossover between the subwoofer and the center and other speakers is not a “brick wall” or infinitely sharp filter, but both speakers will contribute ~ ½ octave above and below the crossover frequency. If the timing is not right and the signals are out of phase, the sound from the subwoofer and the speaker can cancel each other at the crossover frequency. Getting the subwoofer distance right is therefore critical and also a bit tricky as the nature of the crossover filters can change the timing of the sound from the subwoofer so that the physical distance might not be good enough to get this right.
The first REW screen frequency response measurements shown previously, filtered and unfiltered, is one of several measurements runs in this session in which I varied the distance setting for the subwoofer in the AVM 60 to see how that setting would affect the frequency response in the crossover region between the center channel and the subwoofers.
The REW traces shown above represent the post-calibration state of the system and the house curve after ARC correction and subwoofer distance optimization.
I believe that symmetry is important in a home theater. The screen, the front speakers, and the two subwoofers are symmetrically placed on the front wall. Columns, soffits, and acoustical treatments are symmetrically placed in the room, and the seating is symmetrical with respect to the long axis of the room.
ARC, however, does not respect symmetry as much as I do, it seems. When one does measurement in ARC, one uses multiple microphone positions at and around the Main Listening Position or MLP, to better characterize the effect of the room at the MLP. And it weights the first measurement position more heavily than the subsequent measurement positions, e.g., it sets the relative gain or volume control for the various speakers to be equal at that first measurement position. Since we have TWO MLPs, my wife’s and mine, I wanted to be fair to both positions and my first measurement in my first attempt at ARC placed the mic on the centerline of the room and just in front of our table between our two chairs. That was a disaster. ARC It told me it could not calibrate the system at the normal 75 dB SPL level, but had to drop it to 65 dB. I was confused as I had a gazillion watts of power available.
Then I looked at the individual traces from each of the measurement positions as fortunately, ARC allows you to see. The measurement in the first position on the centerline of the room was, perhaps predictably, in a node of the 0-1-0 width room mode, and the then single subwoofer response there was very weak there as explained above. ARC concluded that that was all the subwoofer was capable of and punted, lowering the target levels. Lesson learned. Don’t sit there, and if you are not going to sit there, don’t measure there. Or better yet, add a second subwoofer to cancel the 0-1-0 width room mode so that the centerline of the room is no longer a node or a problem. Using two symmetrically placed subwoofers not only cancels several important room modes, it assures that the sound distribution in the room is also symmetrical, just like the seating. And a single timing or distance setting is appropriate for the two subwoofers, simplifying the setup.
In subsequent ARC runs, I used my chair as the MLP as it better represented both seating positions better than the centerline of the room. The problem now was that ARC optimized the gains for the various speakers to equalize things at my chair on the left side of the room, i.e. turned up the gains for the right front and right surround speakers higher than those on the left. This meant that the speakers on the right side were now very “hot” at my wife’s chair on the right side of the room. So, I manually symmetrized the gain controls to the average of the left and right values. It is a simple fix, and It would be nice if ARC had this Symmetrize Gains as an option.
I did confirm the gains with an SPL meter and did use REW measurements around both my and my wife’s chair to confirm that both seats had the same level and quality of sound.
All of the measurements above were made more than a year ago now, and while there is room for improvement with continued tweaking of this, that, and the other thing; I declared that “good enough is”, and moved on. Those who have been counting may realize that this process used three different calibrated mics, one that came with the Velodyne DD-15, one that came with the AVM 60 for use with ARC, and one I bought for use with REW. This is not a bad thing in that when all the mics agree on the frequency response of the room, it is most likely an accurate representation. In summary, the Velodyne mic and DD-15 EQ firmware were used to measure and tweak the bass region only. REW was mainly used to analyze the acoustics in the room and confirm the frequency response/house curve optimization that the AVM 60’s ARC achieved, as well as finding the optimal subwoofer distance. REW has many other capabilities as is obvious from the tabs above the graphs. ARC does all the final tweaking of the frequency response of the system to achieve the user-defined house curve.
Thank you for reading my three part series documenting my home theater build. I hope you found it interesting and enjoyable. Please feel free to post any questions in the comments section below.