PDA

View Full Version : Noise level testing advise


Blackeagle
03-03-2004, 08:10 AM
I'm interested in doing some testing in regards to noise levels in computers.

I've looked at the Radio Shack meters, both start @ 50dba which is a bit high for best accuracy IMO.

Cost is a consideration so I've looked around and here's the best I've found so far.

2 questions. A) is it good enough? B) Does anyone know of a better one at under $200?

www.technika.com Model I'm interested in is the Sper mini 840014c

Specs of the above meter.
Resolution => .1 db
Accuracy => -+ 1db

cost => $187.00 Bill is right, accuracy costs a LOT.

Any other advice in regard to the testing of noise levels anyone can give will be very welcome.

Thanks

pHaestus
03-03-2004, 08:21 AM
Oh yea I keep meaning to reply to that PM. Sorry I know nothing about sound meters. I'd suggest talking to people at silentPCreview as a start; don't they do a fair amount of sound measurement? Also try bugging spiv at the arstechnica case forums; I seem to remember his hooking up some PC software to monitor the frequency range of fan noise a long long time ago.

pdf27
03-03-2004, 08:31 AM
50dBA is pretty loud. By comparison internal organs are 15dBA, and the scale is logarithmic (dB = 20 log(sound power level)). Unless you've got a huge amount of money and access to an anechoic chamber, sound level meters aren't a lot of use for computer sound purposes.
Most decent quiet/silent PC kit will easily get down to 20 dBA or below nowadays, so a lot of the stuff you are likely to meed will be below the 30dBA threshold for that meter.

Best place to get started is going over to silentpcreview and looking at their testing methods - that should help a lot as they've found most of the possible problems by now.

Blackeagle
03-03-2004, 08:46 AM
I've done some reading on this, need to do more.

I'll check out silent pc review and see what I can learn there.

While a meter with a low end of 30dba is not going to be able to test many of the individual low noise products, it has to be better than the many site reviews I see where no meter at all is used.
But perhaps Silent PC Review is the Pro Cooling of noise control in computers.

BillA
03-03-2004, 09:33 AM
look around for an older B&K
I use a Type 2226 which has a mic with a noise floor of 24dbA (and has an output for the computer - LabVIEW)
yes, more money
but 30 dbA will be found too high

note that what (most) others do is to set the mic much closer to the source, thereby boosting the apparent spl
- not the correct way to do it at all

pdf27
03-03-2004, 09:58 AM
But perhaps Silent PC Review is the Pro Cooling of noise control in computers.
They're pretty good - I haven't yet seen anyone who is even close!

Blackeagle
03-03-2004, 09:57 PM
Bill, thanks for the info on the B&K meter, I will start doing a search for one of those as well.

Not in a rush, I want to get the best I can.

And I've seen what you refer to as to how many set the meter @ 1ft. (or even less in some cases) instead of @ 3ft. to allow much lower quality meters to be used.

Gooserider
03-10-2004, 04:37 PM
lackeagle
I've done some reading on this, need to do more.

I'll check out silent pc review and see what I can learn there.

While a meter with a low end of 30dba is not going to be able to test many of the individual low noise products, it has to be better than the many site reviews I see where no meter at all is used.
But perhaps Silent PC Review is the Pro Cooling of noise control in computers.
I will add yet another vote in favour of SPCR, as you said, they *ARE* the Pro Cooling of noise control in computers. I pretty much split my time between here and there, each for their respective areas of expertise. I am not a super expert, but from all I've read about the challenges of good low noise level testing, it makes WB testing to the BillA level look easy and cheap by comparison :shock:

BTW there has been discussion of both the RS meter you mentioned (deemed grossly inadequate) and I believe that MikeC, the SPCR owner has one of the B&K units that BillA mentioned - it is of only limited use since nearly everything that SPCR reccomends is quieter than that. (SPCR defines a 30dB system as NOISY, ~25dB as marginally acceptable, and wants people to get under ~20dB!)

Challenge one - the noise floor, or background noise level needs to be well under the level of the item under test - this is HARD without a very expen$ive anechoic chamber.

Challenge two - moving the mike closer doesn't really help, as the sound properties change over distance.

Challenge three - 'dB' mostly measures average sound levels, but sound 'quality' matters as much as the total - for instance a low 'whoosh' from a fan might be louder but less annoying than a lower volume high pitch bearing whine. At least when BillA measures a WB, the numbers are well defined and (should be) non-subjective. In noise, subjective opinion about sound quality, which is different for each user, matters nearly as much as the absolute numbers.

Challenge four - test repeatability is far harder as minute changes in environment or setup will cause major changes in results.

Challenge five - Many items (especially fans) are not consistent in their noise levels - I have seen several reports where a person would get several fans in the same order from the same vendor and find some to be quiet and others (with no visible defects) to be noisy.

Given all this, the general SPCR concensus is that the preferred approach is for many different people to do A/B comparisons under different conditions, and report results - after a while a concensus develops of what products work and which don't, along with a long list of 'best practice' techniques on how to use them.

Good luck...

Gooserider

BillA
03-10-2004, 05:10 PM
just would add my concurrence with the above
(for my specific uses, and budget, I can live with a mic floor of 24dbA)
I'll have to spend some time at SPCR
20dbA seems a rather tough target; to not be able to hear what cannot be heard

Gooserider
03-12-2004, 08:41 PM
They don't call it "SILENT" PC Review for nothing Bill... :D There is actually a very common addiction / syndrome that appears in SPCR readers which makes it a somewhat dangerous place!

A new user will start wanting to silence just one or two really noisy items. As soon as he gets rid of those problems he discovers that the next loudest noise, previously unnoticed, is now intolerable and MUST be dealt with. This process repeats recursively until the user either runs out of money or has a system that makes less noise than Bladerunner's!

However it is amazing what folks seem to accomplish through appropriate mods, careful ductwork and fanatical attention to managing case airflow. The typical SPCR box will do with one or two undervolted fans what a Pro-Cooling 'fanboy' will do with six or seven, and get better temps to boot.

Gooserider

Blackeagle
03-12-2004, 10:17 PM
I'm sorry to say I've looked around for a 2226 but so far no luck. I've seen some sound meter systems that cost 8K or more, WOW. I've not found any better than the one I first posted at prices I can afford, so I've increasingly been checking the refurbished & for sale areas at some locations. No luck there yet.

I've also done some more reading at SPCR, a interesting place.

Picked up on one thing I hadn't thought of that I really should have. Sorbothane for vibration damping of the case & hard drives. I've used Sorbothane for a number of years, in recoil pads for shotguns and center fire rifles, as well as inserts for the recoil pad on a shooting vest. It's great stuff. When I saw it mentioned at SPCR I went, DUH! As soon as I saw it mentioned I knew of several ways it could be used to cut noise. I've also priced it at McMasters and it's not unreasonable, so I'll likely be getting some.


Gooserider,

I'm going to try really hard not to get caught up in the noise reduction addiction over there. But it is rather interesting.
My goal is to have as quiet a system as I can get, while still pushing a good O/C with water. So my system may well end up as one the guys a SPCR would consider a bit loud, yet very quiet compared to most overclocked rigs.

I screwed myself when I went to register over there. When I went to my e-mail I deleated the bulk "spam" file, without checking first to make sure my entry/activation link was in my e-mail instead of the "spam" file. You can guess what happened.
I can't start over as I'm in the data base as registered, but awaiting activation. So I've been reading a lot.

What I've read has caused me to rethink a couple things I had in mind to do, or how to do them better. Will be interesting to see how much noise reduction is possible with a O/Cers rig.

Gooserider
03-15-2004, 07:29 PM
Well, I thought the most impressive thread I've seen that says something about the level of attainable silence was one I just saw where the user was complaining about the noise caused by resonant vibration in the power filter coils near the CPU! I suspect his problem isn't that unusual, but that nobody who didn't have an SPCR machine would ever hear the coils so it doesn't get reported.

I also think that you should be able to get a very quiet machine with a good overclock, judging by the reported results of several SPCR members. My box will be heavily over fanned by SPCR standards, in that I'll have 4 x 120mm fans plus the PSU, but I suspect that I will end up with most of the fans being off most of the time.

Gooserider