So how far away from the HSF do you think 5cm from the side of the case is?

Does that mean the distance from the fans varies based on the size/shape of the heatsink?

Sweet! Well, it seems Xigmatek HDT-S1283 is still one of the best coolers. It is pretty even with heavy contenders like TRUE and Prolimatech... but man... 3200 RPM Delta 120mm fan... *shivers at the hought* but luckily there are more fans too.

IFX-14 seems to be on league of its own and holding new contenders off pretty well. New Scythe's didn't do half bad either.

36DBa BASELINE WTF???

Did they do a scatterplot of the temperature/noise data?

don't fixate on the absolute values of the dBA readings, it's used a comparative tool between the heatsinks and fan's tested, nothing more, nothing less. Could have used abstract numbers, but preferred these.

our test lab is nowhere near as advanced as SPCR (we lack the $10000+ budget).

36db is probably the lower limit of the SPL meter, yet they certainly seem to get the relative diffrences between fans correct on anything abouve the 39-40 db (at 10 cm), but anything below is a mystery. Like a passive Ninja 2 is (i'm kinda sure) quieter than 36db

If they used same device at 1m most coolers would not register anything over the noise floor. So I guess using it close range is the only way to go if you want to make any kind of comparison.

wonder if the passive testing was done with the case cloosed? this would likely affect temps.


I like how the shrimp awards sum it up nicely, so you know what your looking at at a glance

You don't need $10,000 to drop your rated inaccuracy from 2.5 dBA to 1.4dBA and lower the noise floor by 10dBA.

You can get an Extech 407738 Sound Level Meter with Memory for around $250.

http://www.silentpcreview.com/forums/vi ... p?p=463534 has more on pricing of cheap sound meters.

As to your review I don't need to know if a fan is quieter than 20dBA. You don't need an anechoic chamber for every test known to man. But I am concerned about you having the meter so darn close to the source of the noise. I'd trust the numbers more if you used a better SLM and moved it further away from the source.

What is the average distance from the mic to the source in your tests? 5cm to side of case + some difference in height of heatsinks = ??? Would you say that ranges from 6 to 10 cm? I understand why you don't test from 1 meter but it'd be nice if you could get that distance up to 30cm.

Increasing the measuring distance will decrease issues unrelated to the rated accuracy of the SLM.

yes case was closed during all tests.
@Dhanson: not currently looking for an upgrade of the SLM, just spend $500+ earlier this year for test gear; but a higher quality dBA meter is definitely on the wish list:)

All tests is inaccurate. "All thermal tests" maybe but not "all tests".

yes:)

I check closest EU retailer, has the Extech 407738 for € 429,59

despite test wasn't SPCR level, it still good giving the guide line. SPCR has tested the NCB fan so we can predict somehow how coolers effectiveness is. Not too many SPCR'ians still use stock fans anyway.

jmke, thank you very much for the big task you've made ^^

Thanks jmke!

What stood out for me was that if I don't care about getting the very last 1C out of the CPU, if a couple of degrees C is ok, then there are a LOT of coolers that provide similar results at similar dBA. So, if this holds true, then really, doesn't it come down to ease of installation and cost?

I really appreciate all the work going into that. Esp. the thermal load tests and temps. One can always deduce something about the noise from the fan speeds knowing the type of fans used.

I can understand the lab noise floor and meter limitations. I wish I had a 18dBA noise floor too with no spectral spikes. But I don't :)

Only one minor plea, for the generator, could you give us a text only output for the names and fan speeds. Makes searching inside browser much faster for long lists. Now that it's full graphics, it's very pretty, but must be visually searched instead of using browser highlighting.

BTW, select all function for the generator would be a nice bonus.

Thanks!

Ouch, that is twice the cost I'd pay in the US for it. If I win the lottery I'll just mail you one.

You might try the Nady DSM-1 since it is the cheapest of the 30 dBA meters here in the US. It should be less than half the cost of the Extech 407738 and still shave 5dBA off your minimum.

Until then thanks for the effort.

I kinda liked how they put out massive reviews at a time. It makes things a lot more convenient and comparable. That goes to the CPU roundup and the fan roundup that they eariler did.

jmke: Good job on the roundup.

thanks EekTheCat. I'm nowhere near MikeC's ambient noise level or sound recording; I have a goal to work towards to! he has set the standard for noise level testing, no other site out there is matching his work;



you did make me bookmark the Extech 407738 page... I'm considering it, after a few more paychecks, so might be few months still. Maybe shipping it over might be cheaper; $250 + $100 shipping would be €262 ... interesting.

jmke -

You could try to find an old Brüel & Kjær model 2203 like the one we used here for years (shown in first photo here - http://www.silentpcreview.com/article875-page1.html). It's good to a bit below 20 dBA. With mic, you can find them on eBay for as low as $500. I'd sell you the backup unit but the 1" mic capsule stopped working a while ago tho. A new mic is very expensive (near a grand!) so the trick is to get a used one or another Brüel & Kjær SLM which uses a 1" mic capsule -- they are interchangeable.

One other thing you must have to be serious about sound measurements is a mic calibrator. Described in detail in the article linked earlier. Changes in temperature, humidity, batteries, even simple aging affect SLM sensitivity, so it's important to check this often. We typically check at least once a week, more often if there are big weather changes. The mic is left powered on all the time so there's no drift caused by internal temp changes. (Our setup never reads off more than 1 dB -- and more often it is accurate to better than 0.5 dB)

Of course, all this is moot if your ambient level isn't at least 20 dBA or lower, at least if you're trying to measure quiet stuff that's <25 dBA/1m. A 0.5m distance would be a decent compromise, tho. Easy to extrapolate roughly to 1m distance.

hey MikeC, thank you for your input, much appreciated. I'm afraid that in very near future I won't be able to make room which will hit that low noise level; while I'm living in quiet little town, I have not enough room at this point in time to dedicate a whole room to noise testing, as this will be required to hit that low dBA mark.

when the kids are out of the house (will have to wait approx 25 years for that;)) I can start preparing a silent room with mega isolation

the heatsinks I liked stayed a few degrees warmer (1-3). like the ninja2, thermalright 120, mugen. My favoirte (not in tests) is an xp90 looking design, base has fins permanent attached.The heatsinks with the open base into pipes only, the designs with the space directly above the solid part of heatsinks (the base). I don't even like to look at it. It looks not normal.
I have concluded the few degrees warmer with the fins attached to the base is the stronger one, and soapstoning into a longevity very important.No crazy temp variants, like opening an 80 degree rooms window to maines 15 below zero. Gimme the soapstoners any time as choice.

My 2 cents. And fantastic cpu choice for testing. The heatsinks with duct only testing and no fan were educational, it still dropped temps by 7 degrees. reverse cooling is a winner as well.

serious remark or ironic?
I want to increase heat load above 100W for CPU alone to be able stress those heatsinks build for Core i7, 6 heat pipes monsters are not properly stressed with a <100W CPU;

And by then you won't need any quiet components anyway Like when I shop for parts for my parents and then realize neither one can hear for beans

you are indeed quite right; my parents are not bothered by a PC which makes to my standards lots of racket