2008 SPCR facility/equipment upgrades: hemi-anechoic chamber

Site News

This is a blog in the forum about facility and equipment upgrades currently being implemented at SPCR. The big project is the building of a hemi-anechoic chamber on the ground floor of my house. The blog was started at the end of May.


July 20: There's usually less traffic on Sunday, so I tried more SPL measurements of the room, which has changed little since my last post.

On the advice of the creator of the SpectraPLUS spectrum analyzer software I'm using, the FFT (Fast Fourier Transform) size was increased to improve resolution at lower frequencies. The FFT size in previous measurements had been set to 4096 PTS, which resulted in a sampling resolution of 21.5Hz. Increasing the FFT size to 65,536 gave a much higher resolution of 1.35 Hz. The results changed the curves below ~500Hz.



The above image was captured at 4pm Sunday June 20 with the FFT at 4096, flat frequency response, without weighting. Note the SPL of just 16 dB, the lowest unweighted measurement yet. Obviously the blue fill and double doors are having an effect.




This was captured a minute or two later with the FFT at 65,536, flat frequency response, without weighting.

Comparing the above images, it's clear that the ~60Hz peak in the second higher resolution curve comprises much of the elevated response in the sub-100Hz range of the first image. That 60Hz peak may be coming from the hum of one or more AC motors in the fridges in the house... or it could be a room resonance... or it could be both. Again, more experimentation is needed to get at exact causes.

A similar phenomenon can be observed in the range between 300~400Hz. The first image shows a broad rise; the second image shows that it's really a very narrow spike at ~370Hz. Moving the mic and/or table in the room, and also finishing the blue fill lining in the room could affect such spikes.




This was captured a minute or two later with the FFT at 65536, with A-weighting applied.


In conclusion, it's obvious that the high FFT setting is much more useful in examining details of the room's acoustics, and more work will be done with this setting in the near future.




July 18: Over 90% of the rooms internal surfaces have been lined with the blue fill, with the exception of the floor. A second door, a solid core, has been hung, and weatherstripping has been applied all around the perimeter of the door. With both door closed, the sound proofing is pretty good, although the wall with the door faces the liveliest (acoustically) adjacent room, and still remains the most "porous" of the room's walls. A slot to run the mic cable into the room had to be cut in a new footer for the door, which is currently too tight; the doors have to be force-closed.



View from the entryway. The section to the right of the lamp is where the window used to be. Thus far, a little under 600 lbs of the blue fill is in the room.





View from the opposite side. The door and portion of that wall is still unlined.

On this Friday, at 8:30 am with typical traffic morning rush hour traffic on Main St 1.5 blocks away, the A-weighted SPL in the room was at a steady 12 dB. From the middle of the front yard, without any audible planes or immediate traffic, the SPL is around 26~28 dBA, with some 2-6 dBA jumps in level when large vehicles pass the intersection of my avenue and Main St., to which there is a line of sight. Such jumps in level are not registered by the mic in the test chamber.

Traffic directly in front of the house (no big trucks so far) causes the SPL to rise 1-2 dBA, and traffic at the nearest intersection half a block away, where cars accelerate from a stop, seems to cause only about a 1 dBA increase from the baseline 12 dBA in the lab.

Airplane noise seems to affect the overall level by no more than 1 dBA, but there have not been any very close planes so far today.

All the rises in noise are audible through the Grado SR60 headphone used for monitoring, but only when recording / playback gain is at maximum. With most of the sounds, it's just a bit of added low freq rumbling; with some sounds, a bit of the midband comes through -- but it's at a very low level. A single quiet fan in the room is probably enough to mask all this.

Will the noise level be improved when the room is 100% done?

Yes.

A portion of the door-wall is unlined, and as mentioned, this is the weakest wall -- along with the doors, which are also not lined. I'm sure some sounds come straight through the doors. Once these are treated, I expect the overall noise will drop at least 1 dBA, and spurious noise rejection will improve several dBA. The ceiling is currently lined with just one 8" layer of blue fill; it can be doubled up later if deemed necessary.

Keep in mind our objectives:

1) lower ambient than the quietest gear we measure (already achieved)

2) elimination of sonic reflections in the room (mostly achieved to below 300Hz)

3) freedom from external noise interference -- this is really the trickiest part. Do we have enough freedom? So far, it seems borderline. Once everything outlined above is done, I think it will be good enough. Most of the noise that comes through is low freq (below 200Hz) and even the room-in-room construction will not help much without a floating mass loaded floor, which is not an option I'm willing to pursue at this time.

More later.


July 9: Image captures of the ambient frequency spectrum in the room, about half-lined with about 300 lbs of acoustic damping, shows -10dB or better at most frequencies. A-weight SPL is now ~10dBA.



At 7:30pm on a weekday evening, flat, without any weighting. Note the vertical scale. The SPL at the moment of capture was 25 dB despite the fact that through most of the freq range, it's well below -5 dB. It's the low freq noise below 200Hz that's responsible for the SPL. Most of this is street traffic from 1.5 blocks away on Main St — which I cannot hear just standing still in the room — but the mic picks it up.





The A-weighting filter cuts all the noise below 200Hz but doesn't change much above that frequency, and it and is much closer to what I actually hear in the room — very little. The hollow core interior door with a big gap at the bottom is an obvious weakness that has to be dealt with. Noise from the adjacent room (not that there is any) easily comes through that door, as well as some loud noise from outside the house. Note that the top of the vertical scale is now 15 dBA, and the level at most freq remains below -10 dBA.





For full details and discussion, please go to the forum thread.