Tuning speaker / room response?


I finally did an experiment this afternoon to check out my speaker and room response. The graph below shows the results:

I got this using the Stereophile Test CD 2 tracks 15 through 18 using my system. The first one provides pink noise, the others give warble tones at the various center frequencies shown in the chart.

A Radio Shack SPL meter, in fast mode, C weighted, was used to to capture SPL levels. The meter was in the 80dB range. As C weighting rolls of above 10kHz, I did not show the rest of the spectrum.

Now this does not look all that flat to me, but I have never done this before. Can anyone give me an opinion on how good or bad this looks?

Also, it looks to me like a little room tuning might help. Anyone have any suggestions as to where to start with this?

Niels.
njonker
How ever simple it might be to put this sort of thing on a thread it is still very impressive. Kudos Njonker, maybe more will follow.
OK, take two of the experiment will thus involve pink noise and a spectrum analyser. I have done a bit of research on linear mics, PC Sound Cards et all, and as it turns out, it is not really economical to turn my PC in to a spectrum analyser. (See other thread on PC Spectrum Analyser).

What I found is that using an osciloscope, I got a nice 1kHz sine wave from a test CD to look as such. When my sound card was done with it, looking at the representation on the PC, there were tons of harmonics. So, I have a spectrum analyser coming to do this right.

So let me describe my room, here is a quick picture:


The black boxes are the speakers, they are actually toed in about 50 degrees. The blue box is roughly where I took the meassurements, and where I normally listen. The room (narrow part) is about 14' wide, about 20' long. Ceiling at about 9'.

The left wall is almost exclusively glass, about 90%. Glass is framed in hardwood. The floor is hand-made tile over concrete. Front and rear wall are drywall. Right hand side wall on narrow part is brick. Front part right side wall open to higher level floor. Ceiling is hardwood boards with beams hanging down at 8' and 16' from the front of the room; beams are hardwood, 4" wide, 8" high.

There is a couch against the back wall, two recliners (one where blue square is, one to the left). There are rugs inbetween the speakers and listening spot, from about 1' left of left to about 1' right of right speaker, another rug between recliners and couch.

The plan for this weekend is to repeat the experiment using pink noise and spectrum analyser. I will perform the test at several points in the room to see what results I get.

Finally, I suspect some people are going to suggest changing speaker placement radically. That is not really an option due to the traffic pattern in the room. I am sure it would be better to move the speakers 3' more out of the wall, but it aint gonna happen. *sigh*.

More data later...

Niels.
Neils can you place the sound system on the back wall where the couch and recliners are? I see you realize that where the speakers are placed now, that they should be around 60in off the wall behind your speakers. Looks like you might get better results reversing the setup. If your auctual room size is 14 wide X 20 long, placing the speakers (woofer center to walls and opposite speaker) 42in from side window wall and 82in to opposite speaker and 60in from wall behind speaker and seating position 10ft from same wall. This could be a starting point to take measurments from. This is if you can reverse the room if possibe. Also that little nook at the bottom right, if you can turn that into a continuous wall (closet maybe with a solid door) might help with the reversal. Just for fun try this arrangment and see if there is an improvment. Good-nite, i'll ck in the morning.
I have a similar question to Njonker's. I have plotted a graph using the test tones from Stereophile test cd 3. I used a Radio Shack spl from the listening position. My results are much flatter than Njonker's, staying within +-4 db of 80db at 1 kHz, with the exception of a 10 db spike at 6000-8000 kHz. What room characteristic would cause this? how does this compare to others who have tried this? Njonkers, I think you may have started something with your graphs. You have inspired me to get up to speed with my computer skills.
The methods described so far really only give rough approximations to the actual response at the sweet spot. This is because of the multitude of reinforcements and nulls that are caused by the multiple reflections. Your actual response is actually much worse than the curves shown. Impluse methods, such as MLSSA or those used by Sigtech, or specialized sweep tones should be used for room tuning. The asc tube trap site has an extensive explaination on their MATT test page (http://www.tubetrap.com/. I use both ASC tubtraps and a Sigtech DSP to tune my room. They make a huge difference and my room is good to begin with.