Saturday, August 31, 2013

Door re-test with iPhone, shows 4dBA improvement only

Since I didn't measure the door beforehand with my wonderful Galaxy (not Samsung, Galaxy brand SPL meter from 6 years ago, recommended and sold by Home Theater Shack for use with their famous equalization program) microphone, but only the iPhone, I need to trot out truly comparable before-after measurements, with a huge disclaimer.

The disclaimer is that currently I measure background noise level of 39.0 dB (+/- 0.5 db).  So that sets a limit on even how low infinite sound blocking could achieve.  Actually, I now measure dBA under same conditions as before in the room as 42.0dBA, compared with 46.0dBa previously.  In other words , only a 4dB improvement (comparing apples and oranges by using the Galaxy for the after-measurement showed 9.5dB improvement).  But since the ultimate noise floor is 39.0 dB, that then suggests that the unbiased estimate of the true noise level would be about 39dB (summing 3dB higher to 42dB, assuming no correlation), and more like a 6dB improvement.  It sounds to me more like the larger 9.5dB improvement I estimated earlier.  Anyway, it is showing over 30dB of total noise reduction now (well, from the listening position) which is pretty good.  Before the new door but after the wall upgrade it wasn't seeming like much more than 20dB reduction, as indeed it was measuring (much of that coming from the preceding wall improvements; the door is off axis from the hall that leads to the living room, so actually this isn't a test of the door per se.  Instead, it is intended as a realistic relevant test from the listener to the non-listener.  This is what I seek to maximize here.)

Anyway, with the current best measurement showing 

Friday, August 30, 2013

Door still needs sealing

The IsoDoor presents numerous difficulties to the installer because the seals and even the threshold are not initially installed, as they are with most door you buy that are intended to have seals (interior doors don't).  But now that the door is installed without seals (after two days of hard work by a very experienced carpenter about my age) I did some measurements.  The iPhone couldn't be found so I used a read SPL meter this time, the Galaxy recommended by (and purchased through) the Home Theater Shack.  I should have been using a real SPL meter all along.

The iPhone app RTA showed almost identical A weighted and C weighted results.  Well that was wrong, I see now, with the Galaxy the A and C weighted results are quite different, and my guess is that all the iPhone results, even with C weighting supposedly selected, are essentially A weighted possibly by the limitations of the hardware itself.

I say that also because the Galaxy C weighted SPL's don't show any improvement in sound reduction due to the new door over the C weighted SPL's I got with iPhone on the old hollow door.  In fact, there is only 16 dB reduction between living room and bedroom NOW, according to the Galaxy.  And the iPhone C weighted measurement showed 21dB reduction...with the old original lightweight door.  That can't be right (and btw, it doesn't sound that way to my ears either).

Comparing A weighted measurements across the two devices, I'm seeing a useful 9dB reduction from the earlier door.  Exact comparison before and after will have to wait until I find the iPhone.  The new numbers A weighted are 71.3 dB in the living room and 40.9dB in the bedroom, a sound reduction of 30.4 dB.  The previous iPhone A weighted measurements showed 21.5 dB of A weighted reduction with the old door.

It sounds like much more reduction than that.  But when you close the new door, still lacking seals, what you hear is mostly the high frequencies.  The midrange and bass are obliterated, but the extreme highs are hardly changed.  So when I play pink noise it sounds like distant very high frequency hiss in the bedroom.  The sound seems to be leaking around the door (but not under the bottom, which still lacks threshold but does already have the sweep).

Strangely, the extreme highs may even be attenuated less than before.  That's how it sounds.  It seem like the extreme hardness of the door is providing a better path for high frequencies around the door than before.

So seals will have to be added.  But at the same time, the door fits very tight into the jamb and it's unclear if the seals can even be added.  This is very worrisome.

It would have been far far better had Sound Isolation Store pre-mounted the seals.  In fact, other sound blocking doors (typically modified metal doors) come with pre-attached seals.  Such doors are either somewhat more expensive or far less attractive or both.

Tuesday, August 27, 2013

Bedroom STC

Actually, I don't yet know exactly how to calculate STC, but here are the measurements I made prior to the installation of the new sound isolation door (IsoDoor from Sound Isolation Company).  The doors are being installed as I write this...

The measurement I make is intended to be a relevant measurement simulating a person sleeping on a bed just away from the door.  I measure the SPL in the room one foot inward of the AC outlet (about 18 inches inward of the door), between 3 and 12 inches from the wall, and at elbow height.  I move the microphone in and out to get the lowest measurement (reducing one frequency modes just a bit), and the optimal position varies slightly from test to test.

The source measurement is made at the listening position (either nose position, or right ear position).  Source is Stereophile Test Disc 2, pink noise track, both correlated and uncorrelated.  I set the level control on Tact to 80.0 (about as loud as I dare at 1am) with the Sonos level at maximum.  The IOS app RTA was used running on my iPhone 3G.

C Weighted SPL's

               Listening Postion        Sleeping Position
Corr        67.7                             46.5
UnCorr   66.8                             47.2

While the correlation decreases output at listening position (due to relative lack of bass augmentation) it actually increases it slightly at the listening position.  Correlated is probably the more relevant test, since recorded deep bass is typically monophonic (with central image) and that shows 21.2 dB of sound reduction (this is actually about 10dB more than what I would have expected based on casual listening--there doesn't seem to be much reduction at all).

A Weighted SPL's

               Listening Postion        Sleeping Position
Corr        67.7                             46.2
UnCorr   67.7                             48.2

The A weighted test surprising shows almost exactly the same results in the correlated test, about 21.5dB of sound reduction (the 0.3dB difference well within measurement uncertainty, btw).  The uncorrelated result is 0.1dB worse than with the C weighting.