Waterfall analysis....is this a room issue OR misunderstanding?

[lookstoomuch]

Hi all, I've been doing lots of work optimizing my theater room and one thing which stuck out was the waterfall graph. My setup is as follows:

1. PC Optical Out -> Denon Receiver
2. Adcom GFA-5500 amp
3. JBL HDI-3800 speakers
5. Dayton UMM6 measurement mic

Room is a basement with dropped ceiling tiles, tile floor (with padded carpet covering 50% of the listening area) and drywall walls, however it is very "wide" as in no sidewalls for 7 feet after the speakers, there is a back wall / corners. If a pic would help I will attach.

If you see below, above 1K, the waterfall "energy" doesn't fall off until 1750ms. Most others graphs I see typically fall in the 300-500ms range. Am I interpreting this incorrectly or should I start looking into some room treatments here. TIA

 

[ThatSoundsGood]

It looks as though most of it is falling off around 300ms. Is there some other noise in the room that could be 55-65db loud? That noise above 1k is only around 55-65db loud which could be background noise in a lot of situations. Have you done multiple measurements and gotten the same result?

 

[lookstoomuch]

It looks as though most of it is falling off around 300ms. Is there some other noise in the room that could be 55-65db loud? That noise above 1k is only around 55-65db loud which could be background noise in a lot of situations. Have you done multiple measurements and gotten the same result?

I don't think so however would not say for sure. My reasoning:

1. Duration and volume of the "tail" somewhat closely related to original amplitude.
2. It does eventually completely fall away

Of course, these are just my educated guesses after staring at this long enough, if I was an expert I would not be posting the question here so any and all input is appreciated.

On your second question, yes I can repeat this reliably from both left and right speakers getting a similar result.

 

[NTK]

You are possibly seeing the effects of the 9 -20 kHz resonance. However, I'd doubt that it is very audible since there isn't much energy in music in that frequency range, and with age our ears are quite insensitive at those frequencies.

 

[radix]

You could get a heavy blanket or comforter and have someone hold it near a wall or floor (if no carpet) or back wall to see if you can localize where the reflections are.

How is the intelligibility of voice? Removing some of the echo in the 500 - 4k range might help if you have problems.

 

[alex-z]

Prime example of how waterfall plots can be misleading just by changing the scaling. A resonance would manifest itself almost immediately, then drop off with the natural decay and damping of the room. The fact that the energy is 25-30dB down before it even appears points to that simply being the noise floor of your room or microphone.

Take some more measurements, including one of the ambient room noise, and one for each listening position. It is important to deal with a good sample size when approaching acoustic treatment and EQ, especially for a home theatre where you have 5+ channels and multiple seats.

 

[amirm]

Take some more measurements, including one of the ambient room noise, and one for each listening position.

Indeed. Just run the measurement without the tones playing and see what the "floor" of the measurement looks like. Then set that as the minimum.

 

[lookstoomuch]

Indeed. Just run the measurement without the tones playing and see what the "floor" of the measurement looks like. Then set that as the minimum.

OK, I went ahead and did a measurement with the stereo equipment shut off, below is the result. The few blobs around 20-30hz are my kids walking around upstairs. Does this tell us anything, i.e. the noise floor is well below the graph above?

 

[lookstoomuch]

Went ahead and ran a sweep with an old Polk I had sitting around and got the same result. Strange.

 

[radix]

It would be interesting to figure out where the noise is being introduced. Do you have an A/D you could hook up directly to the Denon to see if that's where the noise is? Or you could try driving the speakers straight from the Denon to make sure the Adcom is not the problem. Just trying to eliminate sources.

It is an odd waterfall, usually I see the problems with the LF, not the HF. Not that I've seen a lot of them.

 

[lookstoomuch]

Looks like I may have answered the question. Just so happens I had a UMIK-1 on loan from a friend sitting around which I never used having received the UMM6 before I got around to playing with it. Below is the waterfall using the UMIK-1:

 

[lookstoomuch]

And for more research, I expanded the parameters to see what was going on below 45 db and compared both mic's results. Looks like we can chalk this up to ambient noise and a perhaps a faulty mic / mic calibration?

UMIK-1

UMM6

 

[ThatSoundsGood]

That looks much better now. Nice work.

 

[ozzy9832001]

Your overall SPL levels are higher on the UMM6 mic. I'm willing to bet the gain is set higher or it's more sensitive and needs less gain to achieve the same result.

 

[lookstoomuch]

Your overall SPL levels are higher on the UMM6 mic. I'm willing to bet the gain is set higher or it's more sensitive and needs less gain to achieve the same result.

Fair, thought someone might raise that however, If you look at the Polk graph using the UMM6 it is at a similar DB level to the UMIK-1 measurements and it still has the odd upward slope above 1K-20K during the measurement window. To not unfairly throw the UMM6 under the bus I will do another sample tonight but pretty sure the results will be similar.

 

[lookstoomuch]

Fair, thought someone might raise that however, If you look at the Polk graph using the UMM6 it is at a similar DB level to the UMIK-1 measurements and it still has the odd upward slope above 1K-20K during the measurement window. To not unfairly throw the UMM6 under the bus I will do another sample tonight but pretty sure the results will be similar.

Put my Denon in "Pure" mode to turn off any EQ's did a few sweeps, observations:

1. The ALL SPL graphs between the UMIK1 and UMM6 are "close enough for government work" relatively
2. The waterfall graphs are still off, samples below with Mic and Receiver volume settings:

UMIK1 -15 DB

UMM6 -15DB

UMM6 -25DB