Identifying cause of room nulls

[Greg Dunn]

I'm having some difficulties trying to figure out just what is causing a couple of room nulls in my listening area. Hopefully the attached .mdat contains the room simulation data as well as a sample of the RTA (using MMM) at a typical listening location; if not, I can attach a sketch of the room and positions of the speakers. This is a raw capture - no EQ at all.

I've tried moving the speakers (not easy, as they're gigantic - a pair of 70s Magneplanars), the listening chair, etc. but the suckouts at 144 and 280 Hz are stubborn. So I'm pretty sure they're room rather than position related but I can't identify from the REW room sim exactly which dimensions are at fault. My simple-minded interpretation of the data is that I have a couple of tangential modes, but if anyone wants to weigh in on other interpretations of the data, I'd be glad to hear any thoughts.

One thing that is important is that the room ends in a stairway at the rear; the stairs cut diagonally across the room and add a few feet to the depth in that area. I have no idea how to deal with the strange dimensions in REW's room simulation. The speakers, which are each 4' x 6' and 1" thick, pretty much span the room width (don't know how to show that in REW either), so moving them side to side is not an option - plus, limited repositioning tests indicate that's not the major problem.

The sound is really pretty good after I EQ the low end, but of course I would just be wasting power to try and fill in the nulls, so they are largely unchanged (though audible). I hope there's a way to reduce them without major carpentry. Or am I out of luck?

 

[sm52]

Greg Dunn
according to your graph nothing can be said. REW cannot do analysis with only a frequency response curve. Need an impulse response. The distribution of energy will be visible there. Dips of 100-200 Hertz are common in ordinary living rooms. To find out if the room is making them or not, take a measurement close to the speakers - 1m - 1.5m. Only the right side, then only the left.

 

[DanDan]

You file is not working with my 5.19 REW. If you can see a Waterfall, fire up the Signal Generator. Tick Frequency follows cursor. Slide the cursor over the mode in the Waterfall. Walk the room looking for a pattern of peaks and nulls.

 

[Greg Dunn]

Well, the intent wasn't to characterize the speaker, only to demonstrate that the averaged MMM measurement showed the audible nulls. Here's a pair of full range sweeps at 1m without any EQ applied. I placed the mic as near to equidistant from the panels as possible (measured with a laser), considering how large they are.

 

[sm52]

Now there is IR. From this distance from the microphone to the speaker, the pit at 300 Hz disappeared. Remained at 135 Hz. It is available in both channels. What caused it - it's hard to say. There is a reflection point somewhere in the 0.6m from both speakers. Maybe a wall. There is something else. Your speakers play up to 12.5 kHz. It's not very good.

 

[Greg Dunn]

Without a mirror or some additional measuring, I can't be certain, but I suppose it's possible that there's a path to the wall at 0.6m which reflects toward the listening position. The width of the LF/mid panels are about 0.8m though, so hard to tell where the acoustic center is for a measurement. A lot of that area has already been covered with acoustic foam to minimize HF reflections. I will see if that number makes sense in the speaker/room setup.

 

[sm52]

REW determines the distance with an accuracy of 5 mm. It happens up to 1 mm. You need to look at the distance determined by REW, take a tape measure, and measure from the end of the microphone towards the speaker. There will be a center. To get the correct measurement, take a measurement on one side. It is possible from the listening position, or 2-2.5 m. Then click on the gear in the upper right in the SPL & Phase window. Then 'Estimate IR delay'. Then 'Shift and update timing offset'. Then open the 'Measure' window, and on the right there will be a 'Timing offset' in ms and m. You can see this distance to the left of the graph in the measurement information, but for this you need to make 2 measurements as I described.