Tom's Hardware tries (and fails) again to address silent PC.

[derekva]

They have a new articleon building quiet PCs that can also perform well, comparing a quad-core Intel system to a Mobile Core 2 Duo system. Of course, being Tom's Hardware, they totally sabotage themselves by using the heatsinks that come with the platforms (in the case of the C2D, they use the 70mm heatsink that came with the Asus board), instead of doing something sensible like going aftermarket on the cooling. So the end result is that the C2D is only ~ 2dB quieter (in the mid-30s) than the quad-core.

<sigh>

Why do I even bother reading that damn site?

-D

p.s. At least this means that quiet PCs are becoming more mainstream, however frakked-up the mainstream may be in their attempt.

[merlin]

The article is definitely junk in terms of advice, but perhaps it'll help persuade people to take a greater interest in silencing. That's not a bad thing as long as people end up getting good information from better sites. Personally I gave up on the article the moment I saw them say a 7200.10 is "quiet", I use 7200.10's and they're louder than any of my current fans.

[seemingly.random]

It's amusing that someone can state with authority that 35dBA qualifies as quiet. Unfortunately, the uninitiated might actually believe and act on it and wonder why their shiny new $2000+ system isn't quiet.

[Konnetikut]

Table 1: Sound levels as humans perceive them
Source: "What is Sound? And Noise?" by Mike Chin, www.silentpcreview.com; reproduced by permission.

Interesting. I thought 45db was already very loud? It's apparently just "somewhat" audible.

[kittle]

yeah very interesting.
45db can be "somewhat" audible if the noise floor in your room is too high -- in their case 31.7 dBA.
so yeah i can see where they get that from.

I think the spcr testing room noise floor is about 19-21 dBA - which will endup sounding 1.5 times quieter than Toms hardware room. (based on volume that appears to double every 6 DB)

[jessekopelman]

yeah very interesting.
45db can be "somewhat" audible if the noise floor in your room is too high -- in their case 31.7 dBA.
so yeah i can see where they get that from.

I think the spcr testing room noise floor is about 19-21 dBA - which will endup sounding 1.5 times quieter than Toms hardware room. (based on volume that appears to double every 6 DB)

Actually it's every 10dB, which would make SPCR's room twice as quiet. Also, your math is a bit weird. If it were 6dB, SPCR's room would be about four times as quiet (if 6dB is double, then 12dB is double-double).

[Felger Carbon]

About testing room noise floors: the way you determine this is to take yer best SLM and measure it. What you're actually measuring is the sum of the acoustic room noise and the electrical SLM noise.

I have an Extech type II ~$300 SLM, their newest. It is advertised as having 26dBA self-noise. When I first got it, it was still cold here in KF. I measured as low as 26.9dBA (actually I seem to remember slightly lower than that!). What this means is what I suspected: most of that 26.9dBA is the SLM self-noise.

MikeC has a B&K type I (the best kind) SLM, with (I think) a 1" microphone. If so, then most of the 18dBA he measures is the room, not the SLM. If his microphone is smaller that wouldn't be true.

[The problem with my Extech is the small microphone: good for high frequency response, bad for noise floor.]

If anandtech specifies the SLM they're using, use google to find the specs and check the self-noise rating.

Full disclosure: long ago, in a different life, I made frequent use of a type I SLM with a General Radio 1" electret microphone. I'm familiar with low noise levels, and with the measurement of low noise levels. With the 1" electret, the SLM had a self-noise of 14dBA. My bedroom, at 3AM, measured 23dBA back then. I think its quieter here in KF in the dead of winter at 3AM - I didn't have double-glazing back then, and I do now.

edit: corrected "12dBa" to "14dBA"

[Fayd]

yeah very interesting.
45db can be "somewhat" audible if the noise floor in your room is too high -- in their case 31.7 dBA.
so yeah i can see where they get that from.

I think the spcr testing room noise floor is about 19-21 dBA - which will endup sounding 1.5 times quieter than Toms hardware room. (based on volume that appears to double every 6 DB)

Actually it's every 10dB, which would make SPCR's room twice as quiet. Also, your math is a bit weird. If it were 6dB, SPCR's room would be about four times as quiet (if 6dB is double, then 12dB is double-double).

uh... unless i'm mistaken, every 10 decibals is 10x higher.

[jaganath]

yes, the sound pressure is 10x higher, but the perceived loudness is only double.

[jessekopelman]

To further elaborate: 10 decibels is definitely 10X, as we convert from decibel to decimal notation by taking the value in decibels dividing it by 10 and then taking 10 to that power. However, the point, as Jaganath pointed out, is that we are talking about the relationship between loudness and sound pressure. This is not a linear relationship and it turns out that it takes approximately a 10X change (10 dB) in sound pressure to produce a 2X change in loudness for most listeners. What's interesting about this is if we know that a 10 dB sound pressure difference translates into a 2X difference in loudness we can easily translate any other sound pressure decibel difference the following way -- take the decibel value and divide it by 10 and then take 2 to that power. Using this method, we see that 10 dB = 2X loudness, 20 dB =4X loudness, 30 dB = 8X loudness, and so on. Pretty amazing to realize that increasing sound pressure 1000X only results in an 8X increase in loudness!