Rear HT View showing the Vicoustic DC2 Diffusers a Four Section QRD

Let’s recap where I was at this point.

Overall I was sort of pleased with the rooms performance, but definitely not thrilled, and several issues had to be resolved.

1. There was too much HVAC air and it cooled the room far too fast. This was a result of installing an incorrect 1.5 ton 450 cfm system with too high an airflow and cooling capacity. It should have been 1 ton with 250cfm, as calculated. Fixing this was not very easy as I discovered that 1 ton units are no longer made for ducted designs due to recent efficiency requirements.
2. The two floor joists that pass under the rack room flex, and when walking on the floor caused the rack to move very slightly. The rack room joists should have been isolated from the main floor. Fortunately this flexing is only at a few Hz and higher frequency vibrations are not transmitted to the rack via the floor. This was partially rectified by adding additional supporting posts immediately under the joists at the rack room opening.
3. I should have designed height absorption into the ceiling when it was being constructed, as there was plenty of space above. Cutting the ceiling open now is far too big a job so I am stuck with things as they are and the use of EQ. It tuned out that I needed to add ceiling absorption for the first point of reflection. I thought I could get away without it but it turned out that it was required.
4. While the room doesn’t feel or even sound too “dead”, the RT time is slightly too short at 0.17secs. I would have preferred 0.20secs. Raising it would mean adding reflective materials behind the trap covers that are passive to LF, so I am probably going to leave it where it is.
5. While the room measures “flat” the upper mid range, particularly for female vocals, is a little “hard”. The standard Audyssey dip and roll off do not seem to tame this problem enough. I will need to run a number of voicing/EQ runs to see exactly where the problem lies.

Many months passed at this point while I listened to the room in order to determine what I needed to do to improve it. I upgraded to REW, ran a series of room sweeps, and decided to try to rectify the above issues and several others that I found from the sweeps, plus preparing to add two more subs.

The next two posts review those changes made and move me to the point of adding the two additional subs.

Get a cup of tea or coffee and be prepared, the last post is long, and is available only as a  PDF download.

My REW Initiation

There were a lot of changes to the room during the preceeding months, and it became apparent that I needed to join the Room Equalizer Wizard (REW) ‘club’. So I purchased a calibrated Dayton 622 microphone from Cross Spectrum Labs and installed REW on all my Apple computers. My last post shows some of the early REW measurements made, some of which are duplicated below with additional explanations.

NOTE: I am not going to try to teach anybody how to use REW. There are several web sites and lots of online documentation that will help any newcomer with that, and of course I am always happy to answer any questions. I will ,however, describe how to interpret my results.

I ran literally hundreds of REW sweeps, only a few of which I kept, and even fewer I will post here based on their relevance.

After the initial REW sweep tests, which were based on the original design and acoustic treatments, I had several issues to resolve:

1. The HVAC was too noisy and cooled the room far too fast.
2. No amount of mid range EQ tweaking using Audyssey Pro seemed to satisfactorily fix the ‘hard’ vocal upper range.
3. The LF decays were worse than I had expected.
4. The bass just didn’t ‘dig’ deep enough and was too ‘full’.
5. I had way too many early strong primary reflections from surfaces.

HVAC UPGRADES

The HVAC problems were relatively easy to rectify. After making several posts on a number of HVAC Forums and getting a couple of professional quotes I managed to get the noise and cooling rate to where I needed it to be. This was achieved by:

1. Adding a condenser bypass to divert over 150cfm, which left me with 100 cfm for each of the three 1sqft inlet registers. This provided me with 100cfm per inlet at a vent discharge velocity of 1.7 ft/sec. As this discharged into large plenum chambers, before being released into the room through natural diffusion, the resulting blower/air noise was inaudible and is only apparent below 40Hz as shown below.
2. Adding both a freeze stat and an anti-short cycle timer, to prevent the high volume of returned air from freezing the coil, and rapid recycling of the compressor.
3. Adding a flexible silencer in the main feed to the rack. This was originally a lined metal duct and provided insufficient attenuation of the blower noise due to its short length.
4. Adding dampers to all ducts to balance the system. These were actually not required except for the bypass and were left fully open so as to minimize any noise from air turbulence.

System Noise Floor and HVAC Noise Level

As can be seen from the graph, the rooms’ noise floor is little changed above 40Hz. The level increase below 40 Hz being virtually inaudible, well, at least to me. Based on the two common noise ratings the rooms LF noise performance exceeds NC15 and NR0.

As previously mentioned I am unable to check the rooms true NR/NC rating as I have no equipment available to me with a low enough noise floor.

Happy at last, well at least with the HVAC.

CHANGES TO THE ROOM ACOUSTICS

The room acoustic treatment changes are summarized in the revised elevations below:

Revised Room Elevations with First Reflections Shown Together with the Accompanying Diffuser or Absorption Treatments.

The room acoustic treatment changes are summarized below:

1. Added first point of reflection absorption for the height, left, center and right speakers to the ceiling and door area. (The floor absorbers were installed later, see below)
2. Changed the absorption thickness on the left and right walls next to the speakers to an average of 5”.
3. Replaced the domed diffuser opposite the rack with a simple 3″ thick slab of Roxsul. (Sorry the above drawing was not updated)
4. Modified all the speaker grills to remove/reduce edge refraction issues by rounding all edges.
5. Added nine Skyline 2D diffusers to the ceiling at the rear of the room. (Sorry the above drawing never got updated and only shows 6)
6. Added front sub isolators. These are composed of a high density MDF panel, with rubber backed carpet bonded to the top, and five high density foam supports bonded to the bottom, with the space between these supports filled with rigid fiberglass.
7. Added support for two more rear subs;  sub isolators, switched power and signal cabling. The subs were added at a later date and put me into a whole world of hurt. The good news is that after a labor of love the bass results are now outstanding. I finally reached bass nirvana! Much, much more on this in the last post in this series.

First Point of Reflection Ceiling Absorbers for the Height and Front L, C & R Speakers

Wall and Door First Point of Reflection Absorbers

As explained in an earlier post, the room is loosely based upon a LEDE design. I realize that it does not meet several criteria for such a room, in particular the Kicker Delay. The original delay was designed to be just over 10mS and I wanted at least a 20dB level drop by 1mS extending to 10mS. I originally had this, but found I needed to move my MLP backwards by over 6”which caused the primary reflection from the rear QRD to move to 9.5mS. The good news was that after exhaustive testing this earlier arrival has absolutely no effect on the sound stage and its removal makes the room feel too dead.

The biggest issue by far was the over bright upper mid range, especially on female vocals. The Genelecs are known for being a little bright but what I had was not fixable with simple EQ via Audyssey Pro.

I need to point out here that the original Audyssey XT32 Sub EQ and AVP XT32 EQ were never changed during the above acoustic changes and associated listening tests. Why? Because it takes over 2 hours to run and two hours to re EQ with REW and Pro. This was only accomplished after I had made all the acoustic changes described above.

So why were the above treatments necessary?

Energy Time Curve Measurements – ETC

The initial problems that I discovered were based on the Energy Time Curves (ETC) measurements. These allow you to quickly examine the time domain showing any severe reflections. Ideally, there should be no reflections that are greater than -20dB. Clearly this was not the case, with considerable energy above -15dB and a severe reflection at 4mS. It turned out that the primary reflections from the ceiling, walls and the non-linear reflections from the 1” thin absorption next to the left and right speakers were severally affecting this response, which Audyssey seemed to either ignore or incorrectly try to fix. I also had reflections from the door and from the center and RHS speaker that needed to be removed. The increase in reflection density at 10ms (referred to as the kicker delay) is a direct result of the rear diffuser. Ideally, this should have occurred after at least 18mS, but as explained before, achieving such a long delay is not practical for my room.

Original Energy Time Curve – ETC

ETC after Ceiling and Wall Treatments. Note the Significant Reduction in Reflection Levels in the first 8mS.

Once the reflection points were determined and the absorption was installed there was a dramatic improvement to what was already a reasonable sound stage and the upper mid range improved significantly, but was still not where I wanted it to be. I specifically used lots of absorption as I could not predict what diffusers would do and I wanted to keep the front of the room dead with minimal reflections. As I have “gobs” of SPL and headroom from the Genelecs, driving them a little harder is of no consequence.

After the installation of the absorbers there was still a very strong primary reflection at 1.25mS with a lot of “clutter” around it. This L/R symmetrical reflection, believe it or not, is from my suspended wood floor, which is covered with 5/16” felt and 5/16” carpet. I have to assume that as these products are both made from synthetic fiber they have very little acoustic absorption and will eventually have to be replaced with pure wool ½” acoustic felt underlay and a ½” wool looped carpet.

The lack of absorption was confirmed by placing a small piece of 1” thick fiberglass batting at the primary floor reflection point, and as can be seen below, after the installation of the floor absorbers the reflection is greatly improved. My wife at this point thought that I had finally lost my mind……and she might have been right!

There is no way for me to modify the room layout in any way to remove these floor reflections short of adding a REALLY thick rug across the front of my room! So I fixed the issue by adding two 3″ deep floor absorbers, see below.

Front floor absorber for the L, C & R first point of reflections.

The Current ETC With The Floor Absorbers in Place. Note the missing reflection at 1.25mS.

Once I reduced the level of these L/R reflections, I again noted a reduction in the upper mid range vocal levels. So much for Audyssey dealing with the impulse time domain.

The graphs below show the rooms’ current overall frequency response.

The Rooms Original Frequency Response. 1/6th Octave Smoothing.

I introduced the “suck out” between 500Hz and 3.5KHz using Pro as a result of trying to compensate for the overly bright response. The HF roll off above 12KHz is created by Audyssey and maybe removed by selecting a flat response in the AVP receiver.  The increased HF roll off above 5KHz, for the combined response, is normal and is a function of the microphone not being EXACTLY the same distance from each speaker.

The LF response 15Hz to 300Hz is more than flat enough and I have no desire to improve it. However, a flat frequency response does not necessarily equate to a good bass response. I also hoped that when I re-ran Audyssey and removed my Pro EQ compensation the remainder of the response would match it. Unfortunately, it basically stayed the same!

Waterfall Graphs

Wanting to understand how the sound energy in the room was decaying, particularly at low frequencies, I reviewed the Waterfall graphs. They show how the energy at each frequency decays with time. Ideally, as shown by the overlaid decays, all energy should decay at the same rate for each speaker. This, however, is not practical below about 60Hz for the average HT. It will generally rise, but should have decayed by at least 20dB in the first 160mS for all reproducible bass frequencies. That is clearly not the case here as it takes approximately 300 mS to reach that decay level. It turns out that the two long decays at 27Hz and 38Hz were what I was perceiving as making the bass sound full. The 15Hz decay is a function of the ceiling resonating (more on reducing this in the last post), the 27Hz decay is a function of the rooms longest length for both the internal and external shells. I was never able to determine what was causing the 38Hz decay, as it didn’t seem to correlate with any of the rooms dimensions. It may, of course, have been a function of the Audyssey Sub EQ filters, as all filters exhibit a decay time.

Waterfall Response 15Hz to 20KHz. 1/6th Octave Smoothing.

All decays above about 45HZ are regular and there is nothing shown that needed to be improved upon, except maybe increasing the decay slighlty, something I never attempted.

Waterfall Response 15Hz to 200Hz. 1/6th Octave Smoothing.

Close examination of this Waterfall shows the various room modes decaying but none of them above 40Hz are causing an issue that needs to be dealt with. Clearly, the decays  below 40Hz need to be improved on if I am to tighten up the bass. It was these extended decays that were making my bass too thick and heavy.

Note : I do not use data below 60dB as the electronics LF noise floor is too high there and below 32Hz most people cannot hear that level.

Spectrogram Response

Spectrogram Showing Time Taken for the Signal to Decay Between 15Hz and 200Hz.

The Spectrogram is a different way to view the room’s acoustic delays and confirms what can be seen from in the Waterfall response. It shows a maximum decay time of 450 mS for a 40dB drop in level for frequencies below 40Hz. There are no irregular decay patterns above this frequency. If the group delay, shown below, for the subs is taken into account then most of the LF decay relative to its peak is within 400mS. Not really acceptable to me, due to the predominant extended decays at 27Hz and 38Hz, and needed to be improved.

Group Delay Response

The Group Delay is a measure of how fast the acoustic phase of a system changes. Or put simply, the graph below shows how the low frequencies take longer and longer to reach your ears. The delay is a function of both the speaker acoustic and elctronic equalization designs. It is quite normal for any system, and varies considerablly from speaker to speaker. Ideally this should be a flat line, but below 100 Hz provided the group delay doesn’t exceed the time taken for 1.5x the period of the frequency, then it generally has no audible impact e.g. at 20Hz the period for one cycle is 50mS so 1.5×50 = 75mS maximum delay. The graph below shows a delay of 62mS at 20Hz, this is well within the 75mS limit. Ideally, the group delay should not rise above 1.0x the period of the frequency (50mS at 20Hz) as this cannot be detected. Fortunately, my group delay meets or exceeds the x1.5  criteria at all sub frequencies. Excessively long group delays are associated with poor transient accuracy or the ‘tightness’ of the bass.

Group Delays for the Left Right and Both Channels Driven. 1/6th Octave Smoothing.

There is nothing to improve here as the results are governed by the speakers and all the signal chain electronics, none of which I can do anything about.

Wrap Up

While I am sort of happy with the LF response in terms of extension and clarity etc., I would still like to improve the decay below 40Hz. I have, therefore, updated the room to support a pair of SVS SB2000 sealed rear subs in the hope that if I reduce/cancel the rooms primary length mode I may see an improvement in this bottom end decay. However, if this decay is strictly limited by the amount of LF absorption in the room, then I may see either no improvement or it may get worse depending on how the additional subs react with the rooms acoustics. I also wanted the bass response to go at least an octave lower and the only way to achieve that was to add more (sealed) subs.

The nine rear ceiling Vicoustic DC2 ceiling diffusers. These provide diffusion for the first point of reflection for all three pairs of rear surround speakers.

Finally, as stated earlier, I installed nine Skyline diffusers to the rear ceiling of the room. These are to breakup the primary reflections from the three pairs of rear surround speakers. While I am not aware of any problems caused by these reflections, nor did I see anything of consequence in the ETC curves, I want the rear of the room to maintain a highly diffuse sound field and any primary reflections needed to be removed. Also, based on my current experiences, I am no longer convinced that Audyssey ‘plays well’ with strong reflections.

Next post. Adding the two subs and creating the rooms final equalization.

Footnote:

Adding the additional two rear subs became a nightmare, and it took several months of work in order for me to optimize them. The good news was eventual bass nirvana!