Some 220mm 13-blade fan noise data

Control: management of fans, temp/rpm monitoring via soft/hardware

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
Felger Carbon
Posts: 2049
Joined: Thu Dec 15, 2005 11:06 am
Location: Klamath Falls, OR

Some 220mm 13-blade fan noise data

Post by Felger Carbon » Mon Jun 04, 2007 5:13 pm

I have bb fans and sleeve bearings. "Cold" (slow) fans and warmer fans. Suspended fans, fans mounted outside (wart style), and fans mounted in stock fashion. I thought I'd do some noise comparisons.

I took an empty Xclio 3060 case, and positioned it on the edge of a table. I positioned the SLM's microphone 5" from the fan hub on a tripod.

First up was the slowest of my 220mm fans, the Thermaltake "Kandalf" side-panel fan I originally bought and mounted on the side of an Evercase ECE3055. It's now suspended using foam and Stretch Magic on an otherwise standard Xclio side-panel. At 12V exactly I measured 40.7dBA and 570.0RPM. 570 became the RPM at which I would measure all the other fans. Here's the data (all fans 13 blades):

ball-bearing "cold" suspended fan (Kandalf)
40.7dBA 12.00V 570.0RPM standard "grille" in place

ball-bearing "warm" suspended fan (my other suspension)
40.8dBA 7.95V 570.0RPM standard "grille" in place

sleeve-bearing "cold" new Kandalf
40.1dBA 10.17V 569.6RPM, outside "wart" soft mount no grille
41.0dBA 10.27V 569.6RPM, stock (hard) mount, grille present

2nd ball-bearing "warm" fan
41.5dBA 7.07V 569.6RPM, outside "wart" soft mount no grille
43.6dBA 7.65V 569.6RPM, stock (hard) mount, grille present

What the heck?? Those last two sets of numbers don't make sense! So I repeated the soft-mount measurements and I got 41.6dBA the second time around as the only change. Those differences are real!

About the 43.6dBA: this is the only data for a ball-bearing fan that's hard-mounted, and yep! the metal side panel is acting as a sounding board to the ball-bearing high-frequency vibration (which is why I originally went to suspension in the first place).

The sleeve-bearing fan didn't go up as much, but there must be a smaller sounding-board effect there too, but without the nasty high frequencies associated with ball bearings.

My hunch is, the resistance of the "grille" (only 55% open space) accounts for the higher voltages needed to maintain the RPM. Anybody who carefully read the most recent 120mm fan review data by Devonavar knows standard sleeve bearings can consume a lot of power, so the additional grille resistance didn't show up as much as compared to the more power-efficient ball-bearings. You may have a different explanation.

By "soft mount" I mean rubber screws, the ones sold by Jab-tech for open-corner fans.

In my original posting on suspended fans, I reported I could still hear a small amount of ball-bearing noise despite the suspension. I had gotten rid of the side-panel sounding board, but the large fan itself is a smaller "sounding board". So the soft-mounted ball-bearing fan was still 1.4dBA noisier than a soft-mounted sleeve-bearing fan.

Oh. While flashing the strobe, I adjusted the fan voltage until the fan "stood still". That's how I got such consistent RPMs. :)
-------------------------------------------

About steel perforated sheets aka "grilles":

Check McMaster, page 381 ("Find" "page 381"), where they have steel perforated sheets. The grilles for 220mm fans use the "staggered hole pattern", and the Kandalf grille uses what seems to be the .079" x 7/64 hole diameter and spacing, which McM sez is 45% open space. I get 46+%, so the ".079" must be a rounded figure. [McM has the correct (rounded) figure where S = 1.5 times D, for example D=.25", S=3/8"]

This pattern's open space is defined by the hole diameter (D) and spacing (S). In the topological limit, D = S and the open space is equal to pi divided by SQR(12); about 0.9069. A practical grille always has S greater than D.

The grilles on the Xclio, Mio, and Aplus side panels all seem to use 7/64" holes spaced 9/64". The open space is D/S squared times .9069; in this case 54.86% or about 55%.

Jaganath, if you want to check the derivation, remember that SIN(30deg) = SQR(3)/2. ;)

edit: corrected the speling of a name (blush)
Last edited by Felger Carbon on Tue Jun 05, 2007 1:58 am, edited 1 time in total.

McBanjo
Posts: 671
Joined: Thu Dec 08, 2005 1:40 pm
Location: Sweden
Contact:

Post by McBanjo » Mon Jun 04, 2007 8:30 pm

Nice info, great job Felger :)
Jaganath, if you want to check the derivation, remember that SIN(30deg) = SQR(3)/2.
:lol:

Bluefront
*Lifetime Patron*
Posts: 5316
Joined: Sat Jan 18, 2003 2:19 pm
Location: St Louis (county) Missouri USA

Post by Bluefront » Tue Jun 05, 2007 2:59 am

FC.....I've got a new device you might want to try out. It would be very easy to check out with your test rig. It's designed (sort-of) to reduce vibrations, which in turn reduce noise in a computer setup. I'm going to call it a "Dynamic Vibration Reduction Device". I'll try to post a few photos tonight. If it looks promising I'll try to get you some for testing. I'll test some myself, but your setup would yield more accurate results.....particularly with a case panel that apparently adds to vibration noise that you can measure.. I'll post in the cases forum....but it can also be used with fans or hard drives, because it's available in many sizes. Heh....no snake oil either. :lol:

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Tue Jun 05, 2007 9:33 am

was going to suggest measuring @ 1m to make it comparable w/ SPCR results,but the self-noise of the SLM is too high for that,right?

Felger Carbon
Posts: 2049
Joined: Thu Dec 15, 2005 11:06 am
Location: Klamath Falls, OR

Post by Felger Carbon » Tue Jun 05, 2007 1:47 pm

jaganath wrote:was going to suggest measuring @ 1m to make it comparable w/ SPCR results,but the self-noise of the SLM is too high for that,right?
The difficulty is somewhat more complex. The reason noise measures lower at 1 meter than at 5" is that as noise spreads out, its energy (a constant) has to cover more area and hence is less intense in any one area.

How fast the noise level drops off depends on how fast the noise can spread. If the fan under test is mounted in the center of an infinite wall, the sound can spread as a complete hemisphere, and the sound intensity will drop 6dB per doubling of the distance. But if the fan is at the junction of an infinite wall and an infinite floor, the sound intensity will drop 3dB per doubling of the distance. If the fan is in the corner of an infinite room, the sound intensity will drop 1.5dBA per doubling of the distance.

About the "infinity" bit: needed only for echoing rooms. An anechoic acoustic test facility does not need to be infinite, just larger than the distance you're measuring over. Think huge foam wedges. ;)

But SPCR does not test in an anechoic chamber, nor do they specify how quickly the sound is allowed to spread. I don't think they even know the answer! This is a serious problem, because if you don't know if the noise intensity is dropping 1.5dBA, 3dBA, or 6dBA per doubling of the distance, of what value is the 1m test distance except to compare fans under the same conditions (whatever those unknown conditions may be)?

By testing 220mm fans at 5", I believe I'm measuring the intrinsic noise intensity at the source. I can also compare same-sized fans at this distance. Also (as you noted) I stay as far as possible above the ambient noise level of the room+SLM combination. I claim 3 benefits from using a 5" test distance.

I can identify only one benefit from SPCR's use of the 1m test distance in their facility.

However, this should not be interpreted as a call for MikeC and SPCR to change their ways. MikeC has always been an advocate of testing at 1m, and it's entirely possible (even likely!) that there are benefits of which I'm unaware. Also, I'm a strong believer in "your money (and time), your choice".

I get to choose, too! And I'm satisfied with the results I'm getting, given the limitations of my testing room(s) and my measuring equipment. You, in turn, get to choose who to pay attention to and whom to ignore. :D

Or do you like both blondes and redheads? :D

McBanjo
Posts: 671
Joined: Thu Dec 08, 2005 1:40 pm
Location: Sweden
Contact:

Post by McBanjo » Wed Jun 06, 2007 3:14 am

I belive that the main reason MikeC measures at 1m is becourse it's the normal distant from a computer to the user. Everything is silent when it's far away enought and everything is noisy when it's close enough

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Wed Jun 06, 2007 10:35 am

Here's something Mike wrote a while ago:

VIA pdf
The most broadly accepted standards [ISO7779/ANSI/AMCA210-85?] for noise measurement use a 1 meter measuring distance between test object and microphone. Such a distance places high demands on microphpone sensitivity and on the S/N ratio of the preamp circuits when the noise source is low in level. If it is one of the quietest PCs, an SLM capable of measuring accurately down to 20 dB (or even lower) is ideal, but such highly sensitive devices are expensive. One workaround to the 1 meter measuring distance requirement is to measure at 0.5 meter. Because of the way sound decays over distance, if the measuring environment is a full anechoic chamber and the noise source is positioned in the center of the chamber, the difference in measured SPL between the 0.5 meter and 1 meter distances should be 6 dB. (In a normal room, it is more complex to work out, and may have to be verified experimentally, but the difference should be 3 dB.)
He also goes on to give tips about how to use an SLM in an ordinary room to get useful results.

Felger Carbon
Posts: 2049
Joined: Thu Dec 15, 2005 11:06 am
Location: Klamath Falls, OR

Post by Felger Carbon » Thu Jun 07, 2007 5:03 pm

All of the previous fans were made by the same unknown fan maker, whose 13-blade fans are unbiquitous in the market. Today I checked out the copy of that fan made under the Power Works label and sold separately from a PC case. All the measurement arrangements were the same as for the other fans.

Power Works fan:
570.3RPM 7.27V wart soft mount 40.0dBA 27.1dBA amb
570.3RPM 7.49V hard mount 40.5dBA 27.1dBA amb
777.8RPM & 49.1dBA @ 11.95V, starts at 3.24V (no check lower)

At that start voltage, this is obviously a ball-bearing fan. But those dBA numbers are very low, and the difference between a soft and hard mount is half that of a sleeve-bearing fan from the other brand (whatever the brand is), and vastly less than the difference of the other ball-bearing unit. There are different grades of ball bearings, and this particular Power Works fan obviously has good ones!

psiu
Posts: 1201
Joined: Tue Aug 23, 2005 1:53 pm
Location: SE MI

Post by psiu » Thu Jun 07, 2007 6:21 pm

Great news--since that's the one I bought! :D

Post Reply