Fan Testing Round 5 (Working Thread)Are "NANO" Bearings the new sliced bread?

[meanmoe]

pitot tube + electronic pressure transducer (£50) + usb datalogger (£140) = nice automated system.

we use sensors by these people:
http://www.sensortechnics.com/index....a65aa2f14103b5

EDIT: best one in link gives ~0.05 inH2O

bit expensive but very reliable and easy to use,

also the construction of your fan test chamber is very much like an orrifice plate http://en.wikipedia.org/wiki/Orifice_plate
so you could probably get away without using the pitot probe

Hey another person with wind tunnel exp... I've never used the term before, but I've used the principle behind the orifice plate before for velocity measurements. There's no need for a pitot probe. I believe that the total pressure is constant in this regime (isentropic) so you can use continuity (vdot == vdot) and bernoulli to get velocity by varying area and taking multiple static measurements.

Thanks! I'm looking into it, I'm just not sure with these lower CFM levels if I can measure pressures that small. Even with a 3.5" diameter reduction, I'm still only measuring velocities from about 100-600 feet/min, and my particular meter seems to struggle below around 100 and make noise at 500.

If I understand this right, using a pitot tube would require measuring pressures down under .01" H2O. That's getting really small. I know dwyer has some meters that measure down to .001", but that seems like a challenge.
...

You may be able to get solid state transducers that are fairly accurate with that type of precision and accuracy. That is pretty small though... If you want to use a manometer you would have to incline it. This offers a large increase in precision. I've inclined down to 18.5 degrees before for precision. Technically you could go further if you wanted to stretch it across a wall. At this level you might have to correct for temperature though depending on materials used - ah, maybe not for differentials. I'd look for a couple of good cheap solid state pressure transducers... After I put the kids to bed, I'll think about it some more.


To answer your question about the hot film response... my experience matches SuperSilents. The response is excellent, but I do not remember ever using it for free stream velocity measurements. We backed that out using tunnel pressure taps - using the method referred to above.


EDIT: you need a good atmospheric pressure measurement too

EDIT: I just watched the videos... awesome setup - but now you have to watch the vids for the review. seems it'd be nice to use the serial bus for the anemometer and an a/d card for the rest, throw together a quick data acquisition program in labview and instant results - I miss my lab

[Metroid]

Third looks to be amps.

Amps?

Yes, third display is Amps. After frying my RPM monitor, I had to add something back in there..

Thanks . I was kind lost on that .

AP-15 uses less power than Noctua , AP-15 has shown to be really good on all areas. I ordered 5 of them already

[meanmoe]

I thought about the velocity measurement problem some more. Yes, 0.001 inH20 would definitely be challenging... (not impossible though, e.g. decrease exit area by 10)
There are solid state transducers accurate to 4e-5psi, but ...
with water, aggressively inclined, you're still measuring inches H2O in fractions of 32nds

BTW, very surprised by the YL push push... and the fact that you GT performs as good or better than YL push-pull matching cfm at what I think are tolerable noise levels. I'm disappointed by the gelid. The GT is really one to beat.

[Martinm210]

Thanks for the info.

Spent some time reviewing the videos and evaluating the new software's A-weighting ability. Decided to go ahead and plot out the A-weighted dB vs CFM, and this is what I got. I still don't think it captures noise quality, but it is better fitting of my ear anyhow.

I'm really surprised and still absorbing and reviewing. Figured I'd post and see what you all think.

This is a little different than my previous tests, but I did change several important things. This test is using the Zoom H1 in line with the fan hub, 1' away, my previous tests were off axis. This test uses the A-weighting software plus my Zoom H1 which should theoretically capture a greater and more accurate response (two high quality microphones). My previous R2 tests were done with the unweighted mastech meter up close and 90 degrees to the fan. This test also uses the new flow chamber which should be better I think at measuring more accurate air flow with the flow straightener and better sound dampening internally (Insulation).

Maybe it's time to toss the Mastech meter and either buy this software or buy a really good a-weighted meter..

What's more real world, measure the sound level in line with the fan, or off to a corner??

So many questions..

Per this chart, the king of 10CFM would be the noise blocker, 20CFM is shared by the Zalman and yate push/pull (GT is close), 30CFM would be a close tie by the Zalman, GT, Yate push/pull, and 40CFM is clearly the GT. I think the GT, Noctua, and Gelid all experienced difficulties with resonance on this RS120 radiator, that may or may not happen with a larger or different radiator. I may consider buying an MCR120 or other radiator in the future to see that changes with a different radiator.

A bit of a mixed bag considering 3db is considering barely perceptible.

My favorites of this round would be the Zalman for it's overall good performance,the GT for it's exceptional 12V high speed performance, and the noise blocker for it's very smooth low sounding low speed performance. I'm also fairly impressed with push pull as a good option to improve the yate. It's a bit of an apples to oranges comparison, but a good trial run. Push push also worked strangly enough, but I didn't find that when testing pullpull thermally on the radiator...not sure why.

[SuperSilent]

Thanks! I'm looking into it, I'm just not sure with these lower CFM levels if I can measure pressures that small. Even with a 3.5" diameter reduction, I'm still only measuring velocities from about 100-600 feet/min, and my particular meter seems to struggle below around 100 and make noise at 500.

If I understand this right, using a pitot tube would require measuring pressures down under .01" H2O. That's getting really small. I know dwyer has some meters that measure down to .001", but that seems like a challenge.

the velocity tappings on the wind tunnel i use give a repeatability of about 2 ft/s if i use a 30Hz low pass filter to get rid of electronic noise in the system and average for several thousand readings on a 16bit DAQ so its more than possible to get low enough to provide good data.

but i have to admit a good vane annemometer is a much simpler solution. can you not just take velocity and noise data at different times, removing the noisy annemometer when you are takeing the recordings ?

[meanmoe]

the velocity tappings on the wind tunnel i use give a repeatability of about 2 ft/s if i use a 30Hz low pass filter to get rid of electronic noise in the system and average for several thousand readings on a 16bit DAQ so its more than possible to get low enough to provide good data.

but i have to admit a good vane annemometer is a much simpler solution. can you not just take velocity and noise data at different times, removing the noisy annemometer when you are takeing the recordings ?

what transducers are you using?

[SuperSilent]

its a minature (1cm^3 ish) sensortechnics model with a built in amplifier and 12.5 mbar full scale deflection. im sorry i dont have a model number.

[Metroid]

That Zalman ZM-F3 is 150% cheaper than the AP-15 and yet performs really great. It is a really good alternative for us Europeans because we do not have the cheap Yates around here.

[Church]

Metroid: just buy some set of yates, eg. @aquatuning.de. Imho 5eur (set of 3 or 4) or 4.33eur (set of 9) is cheap enough.

[Martinm210]

Thanks..just noticed I missed the coolermaster. It's pretty hard picking the dbA number with the fluctuations, I may try to postprocess on aa slower averaging and redo the graphs anyhow. I may also try and do the same with the mastech unweighted numbers and compare the two measurement methods.

Sorry for so much change on the test methods, I'm having a hard time understanding what works.
Also looking for suggestions anyone has. I do think it would be worthwhile to change rads though. Seems like the resonance issues are really affecting the results more than I figured they would. Ill plan on buying something else and retesting some of these next time.

[Vinas]

Great work, Martin! What about hooking the fan up to a scope so we can see what the relationship between power, RPMs, etc are? That variable PSU you have outputs really clean power, but the scope may catch some idiosyncratic data that we can make sense of.

[Martinm210]

Great work, Martin! What about hooking the fan up to a scope so we can see what the relationship between power, RPMs, etc are? That variable PSU you have outputs really clean power, but the scope may catch some idiosyncratic data that we can make sense of.

Sorry, I'm a bit of a newb when it comes to electronics. What's a scope?

There is an oscilloscope in the noise analysis software, but I'm guessing you're talking about some sort of power related tool?

I did use the kaze master to control voltage again, perhaps there is something not so clean about the fan controller power? I fed the power through the PSU, but still used the controller so I didn't have the switching clicks disturbing the test.

I really don't know much about electronics, about all I know is how to measure voltage, amperage, and that I provide a good grounding source to household current...so I'm all ears..

[Martinm210]

Here is the updated set of software generated A-weighted dbA vs CFM including the coolermaster R4. There are a few other SPL loggings I can extract out too. One I can export out as an excel file over the entire test that gives me SPL all, SPL all-p, and SPL for each frequency over every second. I think I'll try this same thing with the Mastech readings and see how the two compare.




This shows the coolermaster noise level is fairly comparable to the yate and fairly powerful for a rad fan, and actually one of the most consistent of the fans, no resonance issues at all.

I still don't get why some of the fans like the GT and Noctua struggle with the resonance thing so much..

Perhaps I do just have a bad fan, or perhaps they do not like the RS120, but it's not all the fans doing this. Ahh...the fun of real world rad fan testing..

Oh, and it appears sleeve bearing fans from a noise perspective do pretty well. I've heard many folks say they don't last long, but I've had yates running two years straight and never had a failure. Personally I don't see anything wrong with sleeve bearing fans...they seem to work really well...?

[defect9]

I thought failure in sleeve bearing fans tended to be accelerated if they were mounted blowing air vertically (up or down). I've read that using them to pull air out of hot systems (or pull from hot rads/heatsinks) also acellerated their deaths. so pulling/pushing cool air while mounted blowing horizontally is theoretically the way to keep them alive longer.

but, that's probably not true across the board for all sleeves, so dunno if that's why your yates last or if yates are simply better made sleeve bearings.

[Johnny87au]

you martin just saved me alot of time, thanks for the effort and time mate!

[Martinm210]

Yeah, I've heard that too, just never experienced any problems. Mine are mounted vertically pulling on my rads, never a hint of a problem.

Anyhow, working my way through the videos to extract the mastec@h unweighted numbers as a comparison..will post when I get done to see how important acweighting is.

I'm also suspect of the fan controller, may try a retest with just the power supply.

[Martinm210]

Fyi, spent some time troubleshooting this resonance issue and I think I found the culprit with the GT. It's the metal bearing cage that is causing this tin can like harmonic at varied rpms. First I dismounted the fan and listened manually to see if I can recreate that effect, sure enough it was there. Its pretty hard to hear in a noisy room, but the same goes for many of the fan noises. I also tried bypassing the fan controller..no change. Then listened to the fan facing vs opposing and there is some difference. With the fan in pull, its a little less obvious because that metal cup is obscured by plastic frame.

Anyhow, this resonance thing I'm picking up and recording is nothing more than highly sensetive microphones at max gain finding flaws in what we all thought was the perfect fan. Sorry, its not perfect, it has a metal cup surrounding the bearing that resonates a bit as very specific rpms.

Not that the fan is bad, its still my favorite at the 1800-2000rpm range. I just think its finicky when undervolting..you want to avoid those resonance points if you can.


I'm also pretty we'll convince my old mastech meter is part of my past problem in not liking my number results. The unweighted spl is a real problem. It's definately not meeting my own ear perceptions. Time to buy a better sound level meter or this software that can also do the appropriate a weighting.

Oh well, live and learn. I'm going to throw out my round 2 charts, they are no good with this weighting issue. I didn't know any better when I bought the meter and started fan testing..but I know better now. A weighting isn't perfect either but is much more representative of perceived loudness of different frequencies.

Live and learn..

[Vinas]

Sorry, I'm a bit of a newb when it comes to electronics. What's a scope?

There is an oscilloscope in the noise analysis software, but I'm guessing you're talking about some sort of power related tool?

I apologize for not elaborating. Yes you are correct, I meant an oscilloscope test. Some software based scopes are pretty good, especially the 100MHz usb based scopes. Maybe hook up your software and see what kind of results you can get? An oscilloscope and audio oscillator are almost imperative. They do not have to be very good. They would be used to look for hi-frequency oscillations and to view the voltage waveform to see how each fan responds as the signal changes. Then again I think your testing methods are fine as is, so it might not be worth the effort.

[Altaram]

That Zalman ZM-F3 is 150% cheaper than the AP-15 and yet performs really great. It is a really good alternative for us Europeans because we do not have the cheap Yates around here.

YL true that are cheaper than the GT, but are not comparable in terms of quality.

While the YL are sleeve bearing, GT is Double Ball Bearing. That means that the life of the YL is much lower than that of the GT, plus sleeve bearing fans work very poorly in horizontal (and your life is shortened even more).

Overall we are comparing a Japanese brand such as Nidec leader with a Chinese brand of low-end fans like Yate Loon, sorry but there is no comparison.


The availability is the main problem of the GT, but in Europe it is difficult to get them. In UK you have them in QuietPC and Spain in ProSilentPC, for example.

http://www.quietpc.com/es-es-eur/pro...ns/gt-120-1850

http://www.prosilentpc.com/product.php?id_product=59

[Martinm210]

I apologize for not elaborating. Yes you are correct, I meant an oscilloscope test. Some software based scopes are pretty good, especially the 100MHz usb based scopes. Maybe hook up your software and see what kind of results you can get? An oscilloscope and audio oscillator are almost imperative. They do not have to be very good. They would be used to look for hi-frequency oscillations and to view the voltage waveform to see how each fan responds as the signal changes. Then again I think your testing methods are fine as is, so it might not be worth the effort.

Ok, I played around with the Correlation tool listening to the GT, and this is what I see. When I don't hear the odd harmonic, the waves are all irregular like the first picture, but when I do, they become very consistent like the second...

I could record a video of the audio wave playing if that's helpful. I'm just not quite sure what it means.

[Martinm210]

Alright, I ran a mockup scope test video of just the GT..Ill post a video after it gets done rendering and uploading. Seems like this could be a usefull tool in identifying problem areas, although I suspect its more rpm related than voltage...I don't have my new kaze master just yet. Just want to see if this is of any use.

The video does at least confirm the odd harmonic is not coming from my radiator. I tested this open air just so I can eliminate those variables.

[MagisD]

Its a wonky GT's, dammit. I'm going down fight'n for that one. LOL

Sent from my X10a using Tapatalk

[Martinm210]

Its a wonky GT's, dammit. I'm going down fight'n for that one. LOL

Sent from my X10a using Tapatalk

It could be, but it's not the radiator or my test chamber...here is proof! lol

SCOPE TEST VIDEO
http://www.youtube.com/watch?v=DmCEXLoVcek

And here are the noise analysis charts I get reading the wave file. I don't really know what they mean yet, but they show the spikes...there are at least 6 maybe 8 spots where it resonates like that.

[kroks]

thanks for the video
I prefer noiseblocker under 10cfm very quiet and nice design

[Waterlogged]

Do you put those in the chassis holes? Does not that cause the fan to suck air from the sides if you don't get a seal on the side? Or am I thinking all wrong now? Is there a picture with them mounted?

The screws that come supplied with the gaskets are not as thick as the "self drilling" ones you usually gets. So i think there is less risk of the screws touching the hole "edges" or how what you should call it. But having something like that is probably best

Sorry for the late reply.

Yep, actually have to drill the holes out a little. There is a bit of a gap but C strip fills most (if not all) of it rather nicely and makes the hole look a bit more "finished".

Here's a couple of pics of how it looks on my current build.





It looks a little rough here because I cut the hole as large as I could for max air flow, it actually looks a bit better on a regular fan hole.

For screws, I usually use 6-32 or 10-32 machine screws depending on whether I'm mounting a rad (6-32) or fan(10-32). The only exception to that rule is the GT, it has a molded area between the holes in the open area for a small 6-32 nut to be inserted and a little hot glue holds it nicely in place.