room modes and treatments

[buggyboy]

Even I am sick of me, I hope you all can put up with me a little bit longer.

So,

My church sanctuary has dimensions of
L 71'2"

W either 23'4" or 46'8"
(this is due to a partition wall being opened or closed. The wall is closed during services, but at low frequencies I'd expect the partition to be basically acoustically invisible.) http://i.imgur.com/2vsMd.jpg

H 11'1" (this isn't marked on the blue prints, but I measured it to be close to that value.)


My first step is to calculate the room modes. I've downloaded several spreadsheets and used some online calculators. such as http://www.bobgolds.com/Mode/RoomModes.htm

My first trouble is making sense of the data. Would some kind soul take a look at the output from a mode calculator and tell me what's up?

The frequencies I have trouble with in this room are not emphasized, sometimes not even listed on these calculators.




Next I moved on to a practical test.
I downloaded rta software that does fft and so on. Injecting pink noise into my room let me watch the waterfall chart. The water fall (and other views, spectroscope) agree with my ears that my feedback and ringing frequencies are always the same. the biggest one is between 231hz and 233hz. just a massive hump there.
Here is one pic, (the pink is the unused right channel). the blue shows the feedback hump http://imgur.com/Bq3gw
here's a pic of the beginning of a frequency sweep. Notice the lack of harmonics. the yellow blips are a bit resonant, but I include this pic only to give context to the next pic.http://i.imgur.com/diZse.png
And last, http://i.imgur.com/lnrQl.png. boom, this is around 232. the harmonics go wild.

Further, I injected 232hz into the room and walked around and as could be expected, I walked into and out of numerous nodes. The particularly evil nodes are located on the centerline of the length. There is even a node right at the pulpit, which probably explains why I've been so vexed with ringing. (I believe this is probably due to my 2 speakers on stage left and right. this almost makes me want to just use one, hung from the ceiling in the center, but the low ceiling makes me pause, that could be really ugly...perhaps if i wrapped it in white cloth...).

The room could stand some haphazard absorption, especially behind the pulpit (where the pulpit mic picks up best). But that wouldn't address the 232hz problem. If my calculations are correct, 232hz has a wavelength of almost 5 feet (58 inches) and a 1/4 wavelength of 14 inches. so for porous absorption to work, it would have to be 14 inches away from the wall, where the particle velocity is highest.

Resonant treatments, on the other hand, should work right on the wall.

The panel absorbers i've seen equations for seem to indicate that cardboard (density .43 lb/ft^2) with an airspace of 1.25 inches would be tuned to absorb 231.87hz. but this equation doesn't take any measurement for total cavity volume. the equation is Fhz=170/(sqrt(depth* density)).

I've started to look at other resonators like perf, slat, and helmholtz.

for slats the numbers say a slat resonator should have
slat width = 7.6
slat thickness = .75
slot width = .5
depth from wall = 5.9

to target 233hz

All resonator type absorbers should be placed in a pressure high for the target frequency. This is easily done by injecting the frequency and measuring levels with a meter.

So I guess my question is, what room dimension is generating my problem frequency, and what can I build to absorb that frequency. (and where should I place it).

I am committed to making a change in the next 4 weeks. Without your advice I'll just have to guess on my own...perish the thought. Please opine on the subject.

I fervently hope I'm near the end of this saga.

Thanks again,
David

[cmchamp]

David
Are there any flutter echo's in the room, or slap-back that would need to be addressed as well?
If so, knowing what frequencies those were at, it would be possible to design a diffusion/slat absorber that could diffuse the offending flutters AND absorb the LF energy needed.
Though you may have noticed a "hump" at 233 Hz, what was your playback source, and was the playback source voiced to provide a flat response (true representation) of the pink noise, or is the hump due to an inappropriately voiced system?
I'm just getting home from a very long day, so I will revisit this tomorrow at some point.
C.

[cmchamp]

After looking things over David, I'm concerned that your @230Hz issue may be more of a mic placement, speaker proximity and speaker voicing issue.
I have a couple of mode calculators on my laptop with me this morning, however, the more complete mode calculator is on an HD at home somewhere, or I'll have to make the spreadsheet again.
These calculators I've used, don't indicate any issues with modes above 115Hz, but these only calculate through mode 2,2,2.
What speakers do you have, where are they mounted, and how close to the front lip of the stage?

[Brad Weber]

Originally Posted by buggyboy

The frequencies I have trouble with in this room are not emphasized, sometimes not even listed on these calculators.

That's because in a room of the size of yours the room modes of any concern are usually much lower in frequency. At higher frequencies the modes tend to be higher order and denser, thus of much less concern.

While is useful for assessing what is happening in actual use, using the sound system for your testing and looking at just the overall resulting response means that the results reflect the effects of everything in the reproduction chain including the source, the response of the electronics, the response (both frequency and polar) of the speakers and the response of the measurement system as well as the effects of the room. You also seemed to use both the left and right speakers, which means the measurements would also be reflecting any interactions resulting from having multiple sources.

Because of this, the results you noted could be the effect of specific reflections or the interactions caused by the same sound coming from two sources.

[buggyboy]

I used the old radio shack spl meter as a mic for the test, the rta software provided a calibration file for just that use. I also duplicated the tests with the pulpit mic to see what changes appeared. The results in all tests showed similar features.

I have not other speakers than the ones I used to project the pink noise, impulse tests, etc. (Bose 802 series 3) As far as speaker voicing, the speakers may indeed be poorly suited to the room measurement task, but I'm certain that this ringing at 232hz must be dealt with. Any errors are insignificant in relationship to the magnitude of the jump in room response. the impulse spl readout comparing Db at various frequencies goes from -85 at 100hz, to an average of about 54db, to 35db at 232hz.

I'd much prefer to have a partial solution to one at all. Just killing this 232 ring would be a massive improvement.

I detect no flutter echo, nor any discrete echos in the main sanctuary. the over flow is a different story, the flutter echo there is noticable. A low toned clap generates a ring/buzz that lasts about 1/4 second. But no such artifact is present in the main sanctuary.



Pulpit mic is at edge of stage, center, 6.5 feet above main floor. (stage is 1 foot high.)
Speakers are ~4 feet above floor, ~4 feet from side walls, ~3 feet from stage edge.
(this is the bet sound I've been able to get without hanging them. I am tiring to put together the needed materials for hanging them for a test run)
[[qualified professionals will do the rigging, liability, safety, etc,etc]]


I'm open to trying a better testing procedure, provided I can do such procedures with the gear I have. If there is not better option with this gear, I will have to proceed as best I can given that limitation.


Thanks

[cmchamp]

David
Are we re-visiting this? Sounds familiar.
Those 802's don't have any good forward pattern control till above 500 Hz.
Do you have the 802 controller in use?
Next question, can you hum or sing about middle "C", or play a middle "C" on the piano and get the same response? If not, the Bose need to be voiced properly.
Your test microphone isn't being played back through the system causing ringing, is it?
C.

[buggyboy]

Indeed we are, that's what happens when no solutions are found. But this time, 4 weeks hence, I WILL make changes, using my best calculations. I refuse to let this situation go untreated any longer for lack of a professional and a budget.

I have the 802 controller on hand, I've used it before. With or without it the issues are the same. That is, the main problem is so large that the changes made by the controller are subtle by comparison (which is saying something, I believe).

Yes, playing a note on the piano can generate the same effect. I forget the exact note...I think is was about a c#...I can check.

For the main tests I did not create feedback.

The frequency response tests I did were suggestive, but I'm not smart enough to really get a handle on them properly. I personally found the waterfall charts most useful. As a personal curiosity I did watch as I created and maintained feedback. Those charts were very readable to me. I could see the frequency bumps in a quiet room, then watch as I injected the pink noise, then watch as I turned the mic on.

The "resting" waterfall and the "pink noise" waterfall were quite alike...no large changes. But when the mic went live and the ringing started...Katie bar the door.

This graph showed the basic band involved in the ringing, but it wasn't precise enough to satisfy me. So I started some frequency sweeps and went out into the room to listen. The problem frequencies were obvious. I eventually narrowed then down using sine wave tones to just a few hz.

Scientific, not really... proper, I'm sure not. But I am pretty confident that the 232hz issue is real, and must be solved.

I'm generally a slow mover, I make changes rarely, and only if I'm sure it will be an improvement. But in this case, I've done all the homework I know to do, and this situation can't get worse. I am trying to train some soundboard operators, and they haven't got the finesse to tweak things all the time. (tweaking things mid-service is just a recipe for trouble).

So, if I am ever to have several backup tech guys, I need to get the sound to a point where we aren't riding the redline all the time.

(Also, in the back of my head I'm a bit concerned that I may get a job offer in another state...I don't want to leave the system like this.)

I await your wisdom.
Thanks

[Brad Weber]

Originally Posted by buggyboy

I used the old radio shack spl meter as a mic for the test, the rta software provided a calibration file for just that use.

An actual mic calibration file would be associated with a single microphone, as in for one specific mic. From some of the examples I've seen of calibration files for the RS SPL meter and what I know of its accuracy, it at least looks as though the output has the A or C weighting of the meter applied and the corrections are primarily compensating for the weighting rather than for the actual response of the mic.

[WaynePflughaupt]

Originally Posted by buggyboy

Next I moved on to a practical test.
I downloaded rta software that does fft and so on. Injecting pink noise into my room let me watch the waterfall chart. The water fall (and other views, spectroscope) agree with my ears that my feedback and ringing frequencies are always the same. the biggest one is between 231hz and 233hz. just a massive hump there.

<snip>

But that wouldn't address the 232hz problem. If my calculations are correct, 232hz has a wavelength of almost 5 feet (58 inches) and a 1/4 wavelength of 14 inches. so for porous absorption to work, it would have to be 14 inches away from the wall, where the particle velocity is highest.

Resonant treatments, on the other hand, should work right on the wall.

For a fraction of the cost of treatments, you could get a parametric EQ and notch out the hot frequencies. If you don’t have a reverberation problem, you probably don’t need any treatments anyway.

Regards,
Wayne A. Pflughaupt



My Bass Rig: Stage and FOH Friendly

[Brad Weber]

Originally Posted by WaynePflughaupt

For a fraction of the cost of treatments, you could get a parametric EQ and notch out the hot frequencies. If you don’t have a reverberation problem, you probably don’t need any treatments anyway.

If it is a system or device direct frequency response issue then a parametric EQ could indeed be a very effective approach. If the problem is a result of interactions between multiple devices, between devices and the room or of the room itself then equalization may be a less effective solution.

A simple example, the direct sound of a frequency and a reflection of that sound from a nearby surface would arrive at different points in the listener area at slightly different times and thus with varying phase relationships, resulting in summation of the two signals at some points and cancellation at others. If you equalized the source and reduced the level at that frequency, you might reduce the level of the sound at all listener locations but the summation and cancellation resulting from the combination of direct and reflected sound at each location would remain the same.

This is where both the difference between addressing reflections and reverberation and between using an RTA versus analysis tools that can provide information such as Transfer Functions and Impulse Responses come into play. For example, an Impulse Response and/or ETC curve could potentially show the reverberation 'tail' and also whether there are any discrete arrivals that could be a factor.

[buggyboy]

The calibration file for the radio shack spl meter is simply a list of frequencies matched with decibel level. Whether that is good enough, I don't know.

If a parametric would fix my problems, that would be great. But I doubt that it would. Even if I managed to purge the main problem frequency, the harmonics would still be excited in the room. Isn't that correct?

That is, if we notch out 232hz completely, but then the piano or something plays a 232hz note, that will still have a "long tail" and then also all the secondary problem frequencies would be exactly the same since nothing was done to them.


So, I have the software, I can attempt some tests for you all. What should I do?

As far as the treatments go, I'm leaning in two directions.

First, turn the stage into a much less reflective environment. On the side walls especially hang some broadband absorbers. then work on the area directly behind the pulpit, where the directional mic is pointed.

second, do the best I can to nail that 232hz. I'm looking into poly-cylindricals at the moment. They could easily be disguised as decorative columns. Sadly Everest doesn't give the formulas for calculating the dimensions of these. One example listed in the book shows good absorption at my target frequency using a 45" by 16" poly filled with fiberglass.

Perforated panels also look hopeful.



finally, I'm also trying to test flying the speakers. I can't imagine that will reduce the modal issues, but it will get the speakers further from the mic.


Best wishes

[buggyboy]

This thread died a little bit, but I thought I'd post here rather than open a new one.

Simple question:

If a mic is placed in a node, how can that node be reduced?

My instinct is that you'd need to find other nodes located on a wall and place treatment (appropriate to the target freq.) there. The effect would be indirect, but it should reduce energy at that frequency in the whole room...including the troublesome node.

correct?


I was reminded to ask about this question because I came across another post by Cory...

Here's another option - move the pulpit from center. If 315Hz is a resonant tone in the room due to geometry, and the pulpit mic is in a mode - then there's little one could do even if the speakers were moved.

the hypercardioid will reduce the sensitivity from the sides. This will not help, however, if the pulpit is in a mode.

So moving the speakers and changing the mic both seem to be ineffective in dealing with a pulpit mic placed in a hot node. Leaving...? some sort of treatment, presumably as described above.

thanks