effect of poor seal quality

[Brian Dayton]

Before i enter this data, and subsequent data in later posts:  SEALING YOUR ROOM IS ONE OF THE MOST IMPORTANT SOUND ISOLATION THINGS YOU CAN DO.  POOR SEALS CAN RUIN THE BEST OF WALLS.  DON'T TAKE ANY OF THE BELOW AS JUSTIFICATION FOR SLAGGARDLY SEAL QUALITY.


i commented recently that one perfect seal was as good as 4 perfect seals all back to back to back to back

and, i also commented that the effect of seals was a mid/high frequency effect unless massive cracks are present in the wall.


over at RO i think my comments, originally on AVS, somehow wound up being misunderstood, and i don't think that my attempts to follow up on those comments faired much better.


so, last week at a certified NVLAP lab i decided to experiment with the effect of changing the seal quality.  In a certified lab i:

-tested a steel stud wall with layers of thin drywall (the actual arrangement of this wall is not up for discussion), and tested it.  normal 3.625" 25 gauge steel studs with R13 insulation

-drilled a 2" hole in one side of the wall, and another hole across from this hole and up about 3 feet (so the sound would have to pass through the fiberglass insulation to reach the other side)

-after that i removed 100% of the seal from the "source" side (the side with the noise) and tested again.  This leaves large (1/4"+) cracks at the walls edges that exposed the studs

-after that i removed 100% of the seal altogether.  this leaves a wall that you could plainly see light through from one side to the other



the results are shown in the graph below:
                                                   
lab seal:  STC=53, OITC=34, tennekes:  26.2
holes:  STC=53, OITC=34, tennekes:  26.2
holes + no seal one side:  STC=50, OITC=34, tennekes:  25.5
holes + no seal at all:  STC=29, OITC=29, tennekes:  24.3

and there you have it.  Eric seems to like the Tennekes rating, completely removing all seals so that you can see through the wall causes this rating to go down 2dB


STC goes down dramatically, but look at the graphs:  low frequency performance, even as high as 200hz is hardly affected...


and if you think that's shocking, wait til my next post.

[Brian Dayton]

it should be noted that the 3 non-sealed tests were not 100% official in that:

1)  absorption was not re-taken in the room prior to testing (absorption for the sealed wall was used)
2) background noise was not re-taken (Same, background noise from 5 minutes earlier was used)


these 2 deviations would not likely have changed performance by more than 1dB in any band



below is the change in TL for a very heavy wall with and without sealing of a large crack (details not available to the public)

for SEALED:  
STC: 58
OITC: 52
Tennekes: 49


for NO-SEAL
STC: 45
OITC: 45
Tennekes:  44



in talking with the lab manager at Orfield, his comment was "well, those wavelenghts are so big, they see that little crack and just ignore it"


i will add some more commentary later, i had some, then i edited it, then bob quoted me, now i must slumber i think

[Brian Dayton]

now obviously a big enough crack could really muck things up, really reduce performance and lower in frequency and all that, but without a doubt my comment that one perfect seal is as good as 6 perfect seals all in a row stands up under lab scrutiny.



this is also why, when people ask what sealant to use, i tell them USG Acoustic Sealant.  it's cheap, it's waterborne, it's cheap, it's not eternally miserably sticky, it's cheap, and a seal is a seal, more than less.

[Bob]

Brian:

cracks aren't open pathways to sound, they are resonators (unless VERY large, in which case they are open pathways) and blah blah blippity blunk

That sounds familiar.
For example I remember Eric talking about the edges around doors and if the seals are on the main face of the door, that the edges (the 1.5" wide spots) should have felt on them to absorb the resonance between the door and the jamb.

imagine a 20' long 55hz sound wave...  how's a 20' long wave gonna fit through a 2mm crack?  

That doesn't sound right. I'm thinking of perforated helmholtz absorbers which have a very small hole ratio to surface area but the low frequency sound goes through the few little holes.
Nevertheless, whatever the reason, nature is shown in your graph effectofseals.gif.

Test opening 8'x12' = 96 ft^2.
2" circle is 0.17 ft^2
If the 8'x12' wall is 53dB-TL, and the hole is 0dB-TL, then the wall system with the hole is predicted to be 27dB-TL -- if the holes were across from each other without insulation.
What you've got is more like a plenum full of insulation with the sound bouncing around -- apparently very well absorbed.

[Howler]

Brian.. this has been my gut feeling for a long while.

[Brian Dayton]

Brian:

cracks aren't open pathways to sound, they are resonators (unless VERY large, in which case they are open pathways) and blah blah blippity blunk

That sounds familiar.
For example I remember Eric talking about the edges around doors and if the seals are on the main face of the door, that the edges (the 1.5" wide spots) should have felt on them to absorb the resonance between the door and the jamb.

imagine a 20' long 55hz sound wave...  how's a 20' long wave gonna fit through a 2mm crack?  

That doesn't sound right. I'm thinking of perforated helmholtz absorbers which have a very small hole ratio to surface area but the low frequency sound goes through the few little holes.
Nevertheless, whatever the reason, nature is shown in your graph effectofseals.gif.

Test opening 8'x12' = 96 ft^2.
2" circle is 0.17 ft^2
If the 8'x12' wall is 53dB-TL, and the hole is 0dB-TL, then the wall system with the hole is predicted to be 27dB-TL.
You got STC=29, seems close enough.

hi bob, the 2" hole was offset to the 2" hole on the other side, so the sound had to go through some insulation.  it did nothing, zero, nada, couple tenths of a dB here and there.

Even the 2" holes plus complete de-sealing of one side did very little.

look at the graph, the hole caused ZERO change in any band.  i edited my commentary on cracks a moment before seeing your post.  


next time i'm back in one of those labs i'll drill a hole straight through a wall, all lined-up, still won't do anything, wait,  you'll see

[Bob]

Hi Brian:

look at the graph, the hole caused ZERO change in any band

You're too fast for me. I've been editing. See above.  :)

[Bob]

Hi Brian:

Trying again with the 1/4" gap.

Test opening 8'x12' = 96 ft^2.
1/4" gap on 4 sides is = .8333 ft^2 = ((8+12+8+12) / 4 / 12)
If the 8'x12' wall is 53dB-TL, and the hole is 0dB-TL, then the wall system with the hole is predicted to be 21 dB-TL.
That's much lower than your STC-29.

Guessing: Perhaps sound traveling through the steel into the fiberglass and being absorbed -- particularly LF ?

[Brian Dayton]

Hi Brian:

look at the graph, the hole caused ZERO change in any band

You're too fast for me. I've been editing. See above.  :)

you're too fast for me too, lol

we're both quick, yet slow

perhaps we'll wind up delerious and drooling

Brian.. this has been my gut feeling for a long while.

well you're one step ahead of me, i was pretty surprised when i first learned this some time ago...

but THANK CRIPES there is at least one thing that is BETTER at low frequencies than high.  

mass = declining TL as frequency falls (mass law)

decoupling:  failure as frequency falls, maybe badly, anti-effective at some low-frequency range

insulation/absorption:  fails as frequency falls



but seals.... finally something that is easy to get right at low frequencies!

[Brian Dayton]

Hi Brian:

Trying again with the 1/4" gap.

Test opening 8'x12' = 96 ft^2.
1/4" gap on 4 sides is = .8333 ft^2 = ((8+12+8+12) / 4 / 12)
If the 8'x12' wall is 53dB-TL, and the hole is 0dB-TL, then the wall system with the hole is predicted to be 21 dB-TL.
That's much lower than your STC-29.

Guessing: Perhaps sound traveling through the steel into the fiberglass and being absorbed -- particularly LF ?

but for the latter tests the cracks were aroudn the edges, no absorption there to help us out

and absorption fails most at low frequencies

AND, in the heavy example in my second post, that was a straight-open air path, and not a small one.  Yet 40dB of TL remained at low frequencies (i'm not at liberty to give the details of where when or what that test is all about).


the point being that low frequencies simply WON'T be affected by small cracks.  Really big cracks, like open doors, will probably still exhibit some low frequency TL.  seriously, i bet an open door would have measurable TL at 50hz and below, i'm going to tinker with that someday...  that sounds fun.

[Bob]

Brian:

but for the latter tests the cracks were aroudn the edges, no absorption there to help us out

The steel joists with fiberglass behind them would absorb bass, have some minor effect in the mid range, but reflect highs.
Think of the steel as the membrane absorber with a really really deap cavety behind it full of fiberglass.

Isn't that exactly what your effectofseals.gif shows?
Little change in the bass, because it's being absorbed.
Huge change in the HF, because it's not being absorbed.

[Bob]

If you don't have steel studs at the edge of this 1/4" gap, then please see Master Handbook Of Acoustics 4th pg 177 Figure 8-10 -- they have a door with a perforated hardboard miner fiber absorber around the edge to absorb sound traveling through the gap.

[Bob]

Anyone remember off hand any NRC test with multi-frequency TL data that talks about recepticles.

[Paul Woodlock]

Brian.. this has been my gut feeling for a long while.

Me too.

The seals leak on one of my studio doors. Because the door has slightly warped a bit. I'd was on the list to sort out, but I haven't bothered as the TL is fine as it is.

btw - great tests Brian. Thanks for posting them


Paul

[bert stoltenborg]

Haha, Brian.
This is great!!!!  Thanks for this work.

Is there ever stated here that small cracks influence low freq behaviour?

In your famous car hifi thread I remember Philip did measurements on a car. It was feared that small holes in the car like a vent or something would influence LF. Philip showed that even opening the windows doesn't affect the situation much, only under 20-30 Hz, if I remember right.



Thanks again,

Bert

[Brian Dayton]

Brian:

but for the latter tests the cracks were aroudn the edges, no absorption there to help us out

The steel joists with fiberglass behind them would absorb bass, have some minor effect in the mid range, but reflect highs.
Think of the steel as the membrane absorber with a really really deap cavety behind it full of fiberglass.

Isn't that exactly what your effectofseals.gif shows?
Little change in the bass, because it's being absorbed.
Huge change in the HF, because it's not being absorbed.

i don't think so.  if a 3.5" long steel stud over fiberglass could absorb that much bass we'd have one bleeding brilliant new bass trap design...

it's a function of wavelength, crack size, resonance, and things i don't fully grasp...

[Brian Dayton]

Brian.. this has been my gut feeling for a long while.

Me too.

The seals leak on one of my studio doors. Because the door has slightly warped a bit. I'd was on the list to sort out, but I haven't bothered as the TL is fine as it is.

btw - great tests Brian. Thanks for posting them


Paul

hey, no prob.  your door story, i remember reading it in some thread or other, about not perfect seal but screw it isolation was great anyway...

it was interesting to read that.  i tinkered with this in our chicken-scratch non-certified lab some time back and found basically no low-freq effect... that shocked me a bit.  

also interesting, making the wall non-sealed and leaky DOES NOT releive the MSM.  presumably you'd have to have an ENORMOUS leak to eliminate the force of this air spring...

[Bob]

Hi Brian:

i don't think so.  if a 3.5" long steel stud over fiberglass could absorb that much bass we'd have one bleeding brilliant new bass trap design...

I don't know about that.
The 'room' in this case is 3.5" x 8' x 1/4" = 0.05 ft^2
The bass trap is 3.5" x 8' x 12' = 28 ft^2
The bass trap is 560 times larger than the 'room'.

[Brian Dayton]

Haha, Brian.
This is great!!!!  Thanks for this work.

Is there ever stated here that small cracks influence low freq behaviour?

In your famous car hifi thread I remember Philip did measurements on a car. It was feared that small holes in the car like a vent or something would influence LF. Philip showed that even opening the windows doesn't affect the situation much, only under 20-30 Hz, if I remember right.



Thanks again,

Bert

i don't remember seeing anything about low-freq's and seals.  

but thanks for the compliment

got any other ideas for things to test?  i may have blown my wad on testing for the immediate future, but this experiment was fun.  also, i tested the 8' x 8' open hole, data below

i figured that the hole size (8' x 8') would give some TL at very low freqencies, and modal activity would give something or other, and just the fact that the rooms were different sizes (source = 4150 ft^3 recieve about twice that big) would do something...

the TL was pretty much zero, check it out

the test report contains a comments section from the lab manager... his comment for this test was "why?"

lol

[Brian Dayton]

this is not a case of the sound being absorbed at low frequencies on its way through.

note the same basic effect on the massive wall in my second post.  that is a zero-absorption structure, still minimal low frequency effect.  that structure won't absorb anything...

small cracks aren't open pathways to bass wavelengths