Home Theater Forum and Systems banner

1 - 7 of 7 Posts

·
Registered
Joined
·
17 Posts
Discussion Starter #1
Room is 13 ft wide by 40 ft deep. The listening area occupies 1/3 of this room and is separated from the remainder of the space by a screen. The ceiling is canted, 7 ft high on the right side and 15 feet high on the left side. The listening area is carpeted and there are curtains over all the walls in this area. The speakers are set up according to Cardas recommendations. 72" from the back wall, 44" from the side walls and 72" apart. Listening position is 72" from each speaker. Image below
174329



The subs are disconnected as this is a new setup.
The frequency and waterfall graphs are below
174330



174331


The waterfall tells me that I need some bass traps to tame those <50hz waves. But what puzzles me is the periodic dips in the frequency graph.. The weirdness starts at 2.5k and has drops every 4.1 k all the way up to the high frequencies.
When I measure each speaker separately (and superimpose both graphs) I get the image below
174333


So clearly each speaker is pretty flat. I think one of the speakers is toed in a little less than the other so the high frequency falls off a little faster with it. But when both speakers generate the measurement signal i get that weird roller coaster. Clearly there is interference. And when I look at the phase trace it shows weirdness as well
174334



These results are reproducible. If I simply move the microphone to the left so that it is pointing to the left speaker and in line with it perpendicular to the back wall I get this
174335


and likewise if I move the mic in front of the right speaker
174336


The interference pattern is still there, but because the mic is closer to one speaker than the other in the two graphs above, the interference intensity is less.
 

·
Registered
Joined
·
17 Posts
Discussion Starter #2
If I simply move the Mic further away, the dips change. At a distance of 86" from each speaker (instead of 72" in the original graphs above) I get
174337


and at 108"

174338



So what am I to gather from all this - the Cardas recommendations are inappropriate for my weird room?
 

·
Registered
Joined
·
17 Posts
Discussion Starter #3
And here are the relevant mdat files. I found 1/6 smoothing is easier on the eyes
mdat files.zip

The attachment function in the reply menubar did not allow zip files or mdat files so I put a dropbox link
 

·
Registered
Joined
·
17 Posts
Discussion Starter #4
The first graph shows regions of interference every 4100 hertz with the first node at 2200 hertz. 4100 hertz corresponds to a wavelength of about 3.3 inches at my room's ambient temp. When I move the microphone back so its distance from each speaker is 86 inches (instead of 72) the interference waves are at 7000 hertz intervals. (1.9" wavelength) And when the mic is moved even further back to 108" away from each speaker, the interference is at 11,000 hertz interval(1.2" wavelength)

I am confused because I thought that as I move further away from the wave sources, I thought that the nodes of interference should be at longer and longer wavelengths (lower and lower frequency differences)
 

·
Registered
Joined
·
17 Posts
Discussion Starter #5
After some more testing it turns out that using a tripod for the microphone does not allow accurate testing. In picking up and moving the microphone, sometimes the legs are not fully extended. This results in microphone positions being off by as much as an inch or two. I redid all the measurements with the microphone taped to a table.
174339


so here are the frequency graphs with accurate measurements
72 inches from speakers without smoothing
174340

with smoothing
174341


So the graph I posted at the top of this thread (the first one with 4 humps) where I thought I was 72 inches from each speaker was likely inaccurate and probably off by a couple of inches. I could reproduce that graph by making an intentional shift.

microphone shifted 2 inches to the right without smoothing
174342

with smoothing
174343

microphone 18 inches to the right without smoothing
174344

with smoothing
174345


microphone moved 24 inches back
174346

with smoothing
174347

24 inches back and 18 inches to the right also looks similar

My conclusion is that my speakers are quite accurate and deviations of an inch or two in the measurement location results in significant positive and negative interference because the wavelengths we are dealing with are on the order of the distance the microphone is moved. For example, a 4000 hertz tone has a wavelength of 3 inches so when right speaker to microphone distance is 1.5 inches longer than left speaker to microphone distance, I should see significant negative interference. What I do not understand is why I see interference at regularly spaced intervals up the frequency graph.
 

·
Registered
Joined
·
17 Posts
Discussion Starter #6
To rule out that computer or signal processing artifacts unique to my equipment was responsible, I also reproduced the above findings using a mostly different signal path. I used a mac laptop instead of a pc and sent the signal to a different day and then to the same preamp.
 

·
Registered
Joined
·
17 Posts
Discussion Starter #7
Ok, with the help of the smart people on stackexchange, they instantly recognized this as comb filtering. This is the result of two copies of the same sound arriving at a specific location at different times. Basically, it's the interference I suspected that was the cause. I found this video that shows how comb filtering can muddy the sound
 
1 - 7 of 7 Posts
Top