[granroth]

I got to thinking the other day... is it possible to do your own STC measurements of a wall? There are many published ratings done by certified labs for specific wall assemblies and those do a fantastic job of giving you an idea of what kind of performance to expect. But lab measurements only go so far. Those are ideal settings for one wall. When you build it on-site and then add doors and such and connect it to slabs and other walls and ceilings and on and on, then you'll have different performance. By how much?

I have been unable to find anything online that describes how to actually perform an STC measurement of your own wall or theater at home. This seems to be the exclusive domain of labs. Indeed, the STC curves may well be proprietary.

When did a little obstacle like that ever stop me!

The Theory

ASTM E413 lists a 1/3 octave range of 125Hz to 4000Hz as being important for sound transmission and concentrates on testing in that range. A test tone spanning those frequencies (pink noise, perhaps) is played at a specific noise level and the frequency response is captured on one side of the tested partition. I believe that the dB value at 16 specific frequencies is captured. The microphone is then placed on the other side of the tested partition and the noise band replayed.

The dB values at each measured frequency range are compared and the difference is extracted. The differences are graphed into a curve. The curve is then compared against a set of published STC curves. Whatever curve matches the closest to the actual tested results is considered the STC value of that partition.

DIY Style

Given that, how can I do the testing myself on my own theater? Here's my thoughts:

1. Setup a pink noise generator in the theater to play at a consistent level. It doesn't matter what that level is.
2. Place an omnidirectional microphone 1 meter from the wall and 1 meter high. I'll use a calibrated Dayton Audio UMM-6 microphone
3. Play the pink noise and capture the result using an RTA measurement in Room EQ Wizard. Export the results into tab-delimited text file.
4. Move the microphone to the other side of the wall and place it 1 meter on the opposite side and 1 meter high.
5. Replay the same pink noise and capture it using an RTA measurement. Export the results
6. Insert the results into a spreadsheet and get the difference between them
7. Enter the deltas in the following online STC calculator: http://www.diracdelta.co.uk/science/source/s/o/sound transmission class/source.html

I don't believe that I can get the actual STC curves since they are hidden behind a firewall. If those are freely available (legally), then let me know.

Reality Check

What do you'all think? Is this a valid test to do? Or am I missing something very fundamental that makes it utterly pointless and invalidates any results I do get?

 

[JohnstownFlood]

I think a test like this would be valid without the curve...at least in the vein of DIY. I could set up a filter curve in Cool Edit Pro against the chart the the SoundProofingCompany provides here: https://www.soundproofingcompany.com/soundproofing101/understanding-stc/

 

[granroth]

Well, a test of the transmission loss alone is definitely valuable, but I'm fixating on STC in particular. I'm really curious how my own wall assemblies compare to the published equivalents.

I don't think the chart showed on the Soundproofing Company website helps because I'm pretty sure that's just one of many. That is, that's the STC 55 curve that they are showing. I'd need to get all of the other curves, as well. I also don't know that I could recognize what "closest" means without seeing a lot of examples.

That's why I'm liking that calculator I linked to above. If that's accurate, then I don't need the curves. IF that's accurate. No idea.

None of the docs I've seen have mentioned the initial volume of the noise. I wonder if maybe I would just need to put it loud enough that all of the measured frequencies are well above my microphone's threshold.

 

[granroth]

Calculating STC

I got ASTM E413-10 from the library and it turns out that my preconception on how calculating STC ratings is was wrong. It's not about matching it to a curve as much as it's a formula based on a calculation derived from values in a standardized table. The link shared above done by the Soundproofing Company (Understanding STC) actually does describe this method in the appendix -- I didn't realize that until I saw it spelled out in ASTM E413-10.

I'm going to summarize how it works.

The Table

This is the standard table that all STC values are derived from:

Those are the only frequencies that matter for the calculation and those particular values are for STC 0. To get the right values for any other given STC rating, just add that rating to each of those numbers. For instance, STC 55 has a value of 39dB for 125Hz (55 + (-16)) and 59dB for 4000Hz (55 + 4). It's worth nothing that since the offset for 500Hz is 0dB, that means that 500Hz will always equal the full STC rating.

The Measurements

The next set of values that is needed is the measured Transmission Loss (TL) for those 16 frequencies between the inside and outside of your tested partition. That is, you measure on the inside of a wall and then on an outside of a wall; subtract the values; and then you have the set of TL values for the calculation

The Algorithm

Now that you have the data, you just subtract your measured TL from the calculated STC value for your estimated rating. If the value is negative (your value is higher) then it's ignored. If the value is positive, then it's stored off.

Your estimated rating is the actual STC rating if the following is true:

1. The sum of all of the stored off (non-zero) values is not greater than 32
2. None of the stored off values are greater than 8

(ASTM E413 calls the stored off values "deficiencies"). If either of those cases aren't true, then try again with a lower STC value until they are true. If your first guess is true for both, then try increasing the guessed value until it's no longer true. The last one that was true is the STC rating.

Automating This

Yeah, this type of things screams out for a spreadsheet or some other calculator. If you just want the STC value, then enter your TL values into this calculator and it'll spit out the result:

STC Web Calculator

If you want the calculator to also include the TL calculation (i.e., just enter your before and after measurements) plus have graphs, then I created a spreadsheet to do that. Caution: I'm terrible at making Excel spreadsheets so while it works, it ain't pretty. I attached it as a zip file since AVS doesn't allow uploading Excel docs.

The spreadsheet has an example case of a wall that is STC 56. You will see that any STC value at or under 56 will show as STC while any value above that will be Not STC

 

[BIGmouthinDC]

Interesting exercise. I think one thing you need to keep in mind is that the biggest problem with theater sound isolation is sub-woofer rumble. I think most systems are set up with a 80-100 Hz sub-woofer crossover point which is below the lowest STC measurement frequency.

Kind of like measuring rainfall when the wind blew the barn over.

Having said that I will be interested in your results.

 

[granroth]

Interesting exercise. I think one thing you need to keep in mind is that the biggest problem with theater sound isolation is sub-woofer rumble. I think most systems are set up with a 80-100 Hz sub-woofer crossover point which is below the lowest STC measurement frequency.

Kind of like measuring rainfall when the wind blew the barn over.

Having said that I will be interested in your results.

Yeah, ASTM E413 even has the following paragraph:

4.1 These single-number ratings correlate in a general way with subjective impressions of sound transmission for speech, radio, television, and similar sources of noise in offices and buildings. This classification method is not appropriate for sound sources with spectra significantly different from those sources listed above. Such sources include machinery, industrial processes, bowling allies, power transformers, musical instruments, many music systems, and transportation noises such as motor vehicles, aircraft, and trains...

Hence the range from 125Hz to 4000Hz. I find the bolded part interesting because that's precisely what we are doing soundproofing for in home theaters -- to contain the equivalent of music systems and to block out road noise and the like. Yet STC explicitly doesn't cover those cases!

As the only single number rating system that's routinely published, though, it's all we have to work with.

I'm very interested if it would be possible to create an alternative, though. For instance, what if we prepended frequencies between 20 and 120 Hz to the list of measured frequencies? 20-120Hz the definition of LFE by the ITU. It would be a pointless number unless multiple sources picked it up and started using it, but it would be an interesting exercise.

 

[JRock3x8]

You're crazy granroth. And I mean that in the nicest possible way.

 

[granroth]

On Measurements

The procedure for measuring the noise level used to calculate STC is specified in ASTM E90. Unlike ASTM E413, in which most of the relevant info could be compressed into a half-page, the info in E90 is dense and pervasive over the entire 16 pages. It's intended to be detailed instructions for an accredited laboratory. No, it's absolutely not possible to follow the necessary elements of E90 in a home setting.

Still, I'm going to pick and choose some of the relevant bits.

The Sound

The standard does not specify an actual noise level of the sound source, but it does say to first measure the ambient or background noise and then ensure that the generated noise should be at least 10dB higher than background for all tested frequencies. That includes both sides of the tested partition.

On that note, they also don't specify an "inside" or "outside" of the partition since there should theoretically be no difference between tests done on either side. They acknowledge that there may actually be differences in reality and so suggest optionally running the test with the sound source on each side and averaging the result.

This doesn't apply to testing a built theater since there is clearly an inside and an outside and the test results may well be drastically different depending on which side the sound source is on.

The doc says that the sound signals "shall be random noise having a continuous spectrum within each test frequency band". I read this to mean white noise, rather than pink noise. White noise has the same volume for each frequency while pink noise does it at an octave level. Their description sounds more like white noise to me.

They recommend recording an average level for long enough to be with +- 0.5 dB at each frequency, which is done with the formula T = 310 / (f * e^2) where f is the frequency and 'e' is the dB confidence level. Therefore, you should measure for 12.4 seconds for 100Hz since 310 / (100 * 0.5 ^2) = 12.4s. Interestingly, if we tested down to 20Hz, then their recommendation would be to average out the measurement for 62s! That's quite a bit. Wonder if REW can do that.

Each test should be run multiple times and averaged together. They list a minimum of four times and a maximum of 12 times. The end result goes into a series of formulas to determine a 95% confidence level, but since this is all about laboratories, I'm not going to run those numbers. Suffice to say it's a good idea to run the tests multiple times.

The Microphone

The microphone should be omnidirectional and calibrated. They recommend either multiple (at least four) stationary microphones or one rotating microphone. If stationary, each microphone needs to be at least 1.5m apart and none should be at the same height in the room. A rotating microphone should travel in a circular path with a minimum radius of 1.2m

Oof. Already we're in dangerous territory. No way a DIYer is going to have multiple microphones. And rigging up a rotating jig would be even more problematic. This is one area where I'm going to have to deviate from the recommendations pretty sharply.

The microphones should also be greater than 1m from any surface and 2m from any sound source. That shouldn't be a problem.

The standard goes into far FAR more detail on the microphone, but it's hyper specific to a laboratory setting so I'll stop here.

DIY Conclusions

I need to generate white noise in the theater and measure the frequency range of (at least) 125Hz to 4000Hz both inside and outside the theater. The measurement should be able to average for 12.4 seconds or longer. The test must be run at least four times and the results averaged. Since I only have one microphone, I'll have to run the tests eight times.

The microphone will be on a tripod with the tip 1.2m from the wall and 1.2m from the floor.

The white noise shall be played at a level that the lowest reading (likely the 125Hz reading) is 10dB above the ambient noise on the outside of the theater. We'll see if my theater can get that loud.

I'm going to have to investigate REW some more. I know it can run multiple sweeps and average the results, but I don't know if it can do the same for RTA tests.

If anybody else has any other thoughts or conclusions, then let me know!

 

[Nyal Mellor]

Some complements to STC:


NR - noise rating http://www.engineeringtoolbox.com/nr-noise-rating-d_60.html
RC - room criterion http://www.engineeringtoolbox.com/rc-room-criteria-d_729.html


Note that for measurement of noise levels below about NC30 you need a quiet mic, a USB mic will not work, too much self noise.

 

[granroth]

Some complements to STC:

NR - noise rating http://www.engineeringtoolbox.com/nr-noise-rating-d_60.html
RC - room criterion http://www.engineeringtoolbox.com/rc-room-criteria-d_729.html

Note that for measurement of noise levels below about NC30 you need a quiet mic, a USB mic will not work, too much self noise.

My understanding of NR, NC, and RC are that they are all single number ratings of a static noise level and that they exist because a simple dBA or dBC number doesn't cover the range that the makers of those ratings cared about. I didn't think they were related to transmission loss through a partition, though.

But perhaps it could make sense to capture an RC rating on both sides of a partition and just subtract them to determine a transmission loss rating? That might make more sense than STC since it does go down to 16Hz.

I do, indeed, have a USB mic for my measurements but I'm assuming that it won't matter that my floor is 30dB if I just output my source sound loud enough that all frequencies are at least 10dB louder than that. Since I'm doing a difference and not an absolute value, the actual level shouldn't matter.

 

[granroth]

First Results

Intro

With this first set of testing, I'm expecting relatively poor results since my door is not properly sealed (there are visible gaps). I also got very odd results, though, which makes me wonder about the validity of the test at all. More on that later.

The Setup

Equipment: Dayton Audio UMM-6 Microphone (calibrated); MacBook Pro running on battery; Integra DTR-5.3 AVR; two iOS devices
Software: Room EQ Wizard 5.1; Audio Tools for iOS; Remote Generator for iOS

Generated Sound: Full spectrum 'balanced' white noise from Remote Generator (controlled by other iOS device); AVR at "90"

Microphone Position in Theater: 54" from ground; 54" from door; 8' 11-1/2" from front wall
Microphone Position in Hallway: 54" from ground; 54" from door; centered in hallway

Test Procedure: Run RTA in REW with 65535-point spectrum using Rectangular window and 8 averages on 1 second update intervals; Take four measurements and reset averages between them; Average the four measurements

The Pictures

Theater measurement station:

Microphone:

Hallway setup:

AVR setup:

The Results

Here's the two averages:

And the calculator:

Final STC rating: 22

My Thoughts

Those measurements must take the microphone calibration into account since the actual dB level in both the theater and the hallway was substantially higher than 50dB (around 80dB). I did do ambient testing and yes, the end results were more than 10dB higher than ambient for the measured values.

Still, I was expecting more of a gradual curve. If you look at the graph, you can see my measured TL just jumping all over the place. The actual measurements also often swung wildly within only a few Hz. For instance, at 630Hz, I saw a reduction of 40dB to 25dB (-15dB), but if I look at 640Hz, my reduction is 48dB to 8dB (-40dB). That's a gigantic difference for only 10Hz.

If I look at the STC test procedure and try and identify the huge differences, other than not having multiple microphones, I do see that my hallway position is much closer to the walls and ceiling than they should be. Could that result in wildly varying measurements?

All in all, I'm a bit at a loss as to determining if this is a valid test or not. Maybe STC doesn't make sense out of a lab after all?

 

[granroth]

Second Try

Intro

After the first test, I got to thinking that maybe the way I'm measuring the sound could be better. As such, I read up on the RTA measurements in REW in more detail and also experimented with the microphone positioning.

The Changes

Test Procedure: Run RTA 1/48 Octave in REW with 65535-point spectrum using Flat-Top window and 16 averages on 1 second update intervals using Max Overlap 93.75%; Take four measurements in multiple positions and heights and reset averages between them; Average the four measurements

Generated Sound: Full spectrum 'mono' white noise from Remote Generator on iPad at maximum volume

This resulted in a much louder input signal (100dB instead of 80dB) and a more stable measurement curve.

The Results

Here's the inside and outside averages. They still jump around, but are starting to have more of the expected curve:

And then the new STC calculations:

Final STC Rating: 32

My Thoughts

This seems like a far more accurate test than the first time around. The calculated STC value of 32 better "fits" the actual measured overall dB difference between the inside and outside (~100dB to ~70dB ~= 30dB transmission loss).

It's still a pretty jumpy curve, albeit not as jumpy as before. I'm not sure what to make of that.

As is, though, this feels more like a valid test and my earlier thoughts that maybe it's not possible to do STC testing at home are mostly mollified.

 

[JRock3x8]

I somehow lost the thread where you were measuring bass output on different stage types. Does the bass stay within the room better with a stage vs without a stage?

 

[Mfusick]

who cares about 125hz + when your bass rumbles the entire house at 20hz-80hz ?
Or am I missing something?

 

[granroth]

I somehow lost the thread where you were measuring bass output on different stage types. Does the bass stay within the room better with a stage vs without a stage?

The link to that experiment is in my signature: http://www.avsforum.com/forum/19-de...ing-stage-filler-sand-fiberglass-nothing.html

I didn't find any notable difference no matter what I did related to the stage.

who cares about 125hz + when your bass rumbles the entire house at 20hz-80hz ?
Or am I missing something?

I wouldn't say that STC has no value. First off, it's the only commonly quoted single number rating of the soundproofing capabilities of wall assemblies. So if you're wanting to compare walls, then STC is your huckleberry.

Also, if you don't care about LFE and primarily watch TV or movies that are strongly dialoged based (which rarely have any LFE at all) then STC does cover your frequency range pretty well.

But yeah, if you do care about LFE, then STC really only has value in its limited comparative properties against other walls. I'm not aware of any single value soundproofing rating that covers LFE and is widely researched and shared.

 

[BIGmouthinDC]

Interesting. Do you think you are measuring the room or the door STC ? Does the room share a wall with any other space? How about measuring directly above?

 

[JRock3x8]

The link to that experiment is in my signature: http://www.avsforum.com/forum/19-de...ing-stage-filler-sand-fiberglass-nothing.html

I didn't find any notable difference no matter what I did related to the stage.

ok that made me feel terribly lazy

so in your estimation, the stage is 100% cosmetic?

 

[Neurorad]

who cares about 125hz + when your bass rumbles the entire house at 20hz-80hz ?
Or am I missing something?

I wonder about the history of STC. Was it developed by the industry, because little can be done to affect lower frequencies, product-wise?

 

[BIGmouthinDC]

because little can be done to affect lower frequencies, product-wise?

respectively disagree, Isolated damped mass will tame the low frequencies.

 

[HopefulFred]

Interesting. Do you think you are measuring the room or the door STC ? Does the room share a wall with any other space? How about measuring directly above?

I hope you can do this. I would say it's important to get your procedure verified first. But especially since the traditional wisdom is that the door is the weakest link, an uninterrupted wall would be a more digestible test.