oscilloscope screen shots for nexus, tek-chain, yate loon,..

[frankgehry]

Listening to and comparing sound files is probably as difficult as listening to the fans themselves. Sometimes there's just not enough difference to make up you mind. In my advanced sound lab just off Broadway in quiet New York City, I have assembled the latest and greatest test gear to measure fan noise using a pc based oscilliscope, a C-Media sound card, and an Audio Technica ATR-55 23.5 trillion dollar directional microphone. My approach so far is just to hold the microphone in front of the fan and watch the oscilloscope. I hope to test and verify the accuracy of these results, and scale the output to db's, but for now I thought I would go ahead and post. I will do a CompUSA MadDog vs. D12SM-12 in a few days. - FG

NOTE: The present test method does not attempt to assign db levels. The waveforms simply compare the noise levels near the fan as recorded and displayed using the oscilloscope software.

3-15-05: I am updating these images.
3-18-05: The tests have been repeated with a line level pre amp added to boost the microphone signal. The microphone was placed on the intake side of the fan at an angle of 45 deg. and 3 in. from the edge of the fan frame.


Cheiftec/Yate Loon D12BM-12D @ 1000 rpm


Tekgems Yate Loon LED D12SM-12 @ 1000 rpm


Tek-Chain 120 @ top speed - 1650 rpm


Nexus 120 @ 650 rpm


Tek-Chain 120 @ 650 rpm


Coolermaster A12025-12CB @ 650 rpm


I'm going to retest the last three. I would have expected the Nexus to have the smallest waveform, followed by the Tek-Chain, and then the Coolermaster.

UPDATE: I retested with no change in order.

[ilh]

I think looking at the spectra (power vs. frequency) would be much more useful. I work in speech recognition, and it is pretty much impossible to tell anything by looking at waveforms since they can be so effected by phase, noise in particular bands, etc. A spectrogram (frequency on vertical axis, time on horizontal, log intensity as gray value) can be useful to since you can see how the spectrum changes over time. For example, if there is a time-varying whirring or clicking sound, that should come out.

Perhaps I could try some of my spectrogram tools with your waveforms?

[frankgehry]

Sure Lee,

I would be up for that. I can send you the waveforms that I've done and see what you can come up with. Do you want to send me you email address?

I saved everything as bitmaps so I'll have to do a few over, but it shouldn't take long. - FG

[MikeC]

Perhaps I could try some of my spectrogram tools with your waveforms?

This can be done with all the noise recording that have been posted in SPCR reviews for the last ~6 months or so. There are already 194 files, all recorded and processed into MP3s in the same way.

I am working to post a database of all sound recording links and descriptions in one place.

Listening to and comparing sound files is probably as difficult as listening to the fans themselves. Sometimes there's just not enough difference to make up you mind.

If this is so, then it would seem safe to conclude the difference is not meaningful & thus irrelevant. The main issue then is how much air a fan is blowing at a given noise level. Without this info, the noise measurements / recordings are not really useful. Hence the effort in all the fan recordings & measurements in the reviews to show CFM and noise together.

[ilh]

Right, Mike. I almost forgot. I'll play with those at some point over the next couple weeks when I get some time just to see if we can "see" any differences that somehow correlate with perceived differences. Of course, as you point out, all that really matters is if the listener perceives an important difference at a relevant airflow.

Frank, don't bother sending me anything right now. I'll play with the already existing MP3s for now.

[Gorsnak]

If this is so, then it would seem safe to conclude the difference is not meaningful & thus irrelevant.

Yes and no. Obviously what we're ultimately concerned with is the subjective impact of our computers on our auditory systems, and any difference I can't hear from my chair is irrelevant. This certainly should be a guiding principle.

However, that doesn't mean that a difference I can't hear when listening to two fans in free air is irrelevant, since the difference might be amplified when the fan is installed in the machine. Just for example, my case is extremely effective at damping high frequencies, but less effective at damping low frequencies. So, if I'm trying to decide on a cpu fan and the candidates sound just as loud as each other, then a power vs frequency chart would be very relevant in deciding which fan is going to be more audible once it is in place.

Another case where fine measurements might be relevant would be when a machine is flirting with being below ambient levels. While most people probably can't distinguish between volumes within 3dBA of one another when listening to different fans in turn, small differences may well be the difference between being audible over ambient noise or not.

It's also worth pointing out that it takes practice to hear differences. It took me quite a long time before I could pinpoint nulls due to phase interference when playing pink noise over sound systems, but now I hear them easily. The first several times? Just a lot of hissing. Hard measurements might uncover differences that people will hear over the long run, but can't pick out when listening to a couple samples.

[MikeC]

If this is so, then it would seem safe to conclude the difference is not meaningful & thus irrelevant.

!) Yes and no.... that doesn't mean that a difference I can't hear when listening to two fans in free air is irrelevant, since the difference might be amplified when the fan is installed in the machine. Just for example, my case is extremely effective at damping high frequencies, but less effective at damping low frequencies....

2) It's also worth pointing out that it takes practice to hear differences... Hard measurements might uncover differences that people will hear over the long run, but can't pick out when listening to a couple samples.

1) Invariably, this is related to vibration, which cannot be picked up directly with acoustic measurements. You CAN feel it and comment on it. You can also record / measure it as sound-- but only when the fan is mounted on something that will vibrate in sympathy with the fan and make noise.

The trouble with this technique is that a particular fan vibration may trigger resonances in the board (or whatever you've mounted the fan on). Now a different fan with equal vibration but a little further outside the resonances of the mounting board may sound a lot quieter. In practise, they may sound the same in a particular case -- and different in another.

This is not something you're going to be able to predict with measurements without getting into some really complex stuff with expensive vibration monitoring tools. But it's so easy to ddescribe subjectively -- "All 4 samples of this fan have moderately high vibrations compared to our reference fans (name them)." -- and I think this comment is actually a better indicator of real performance than complex measurements for 99% of SPCR readers.

2) I doubt very much that hard measurement can pick up anything relevant we can't perceive with our own senses -- not without really serious test gear and great measurement efforts. Again, if we can't hear it... With regard to hearing stuff better over the long term, this is probably true of most of us: If the reduction in PC noise we achieve lowers the ambient noise & enhances our ability to hear quieter sounds, the risk of hearing more clearly the things we could not hear before (which might bother us) is just the unavoidable fate of PC silencers -- anyone who obsesses about low noise...

[frankgehry]

Mike,

My guess is there is not really much difference in cfm at 600rpm among 120mm fans. Also as thorough as your test are, they will be out of date in a matter of months. Users will want to know about fans that use the 4 pin standard, fans that work well with the t-balancer, and what about the new m3 panaflos. There will always be a place for user reviews and tests. I am just sharing work that I have done for myself and the waveforms serve to confirm my earlier subjective tests posted several weeks ago. - FG

[Gorsnak]

1) Invariably, this is related to vibration, which cannot be picked up directly with acoustic measurements. You CAN feel it and comment on it. You can also record / measure it as sound-- but only when the fan is mounted on something that will vibrate in sympathy with the fan and make noise.

Invariably? I must disagree. The example I gave doesn't depend on vibration (well, except in the sense that sound is vibration). My case damps high frequencies better than lows, so for any two fans at the same dBA in open air, assuming they are soft-mounted to avoid resonance issues, the one which emits more of its sound at higher frequencies will seem quieter. That is a case of a difference between the fans being amplified. Perhaps you thought I was referring to the sound itself being amplified? I could have been more clear, I suppose.

[MikeC]

Mike,

My guess is there is not really much difference in cfm at 600rpm among 120mm fans. Also as thorough as your test are, they will be out of date in a matter of months. Users will want to know about fans that use the 4 pin standard, fans that work well with the t-balancer, and what about the new m3 panaflos. There will always be a place for user reviews and tests. I am just sharing work that I have done for myself and the waveforms serve to confirm my earlier subjective tests posted several weeks ago. - FG

You're right about cfm / rpm -- but your oscilloscope pics above weren't of the fans at the same rpm...

I dunno if the fan reviews will be out of date. 4-pin fans are not taking over the world, they are generally found as OEM gear on new Intel based systems. It's not like the fans we like are going to stop being useful or produced.

I applaud you sharing your info; no problem with that at all, that's what SPCR has always been about. I made my comments in the spirit of ensuring that the data posted is meaningful & useful.

Gorsnak -- You're right, invariably is too sweeping; I should have said most often. It is in my experience.

Your point was that while you heard no difference between the fans in free air, you did hear a difference when they were mounted in your case. And this was not due to vibration but the ability of your case to damp high freq noise. So what you are saying is that 2 fans with the same overall dBA reading can sound different depending on their fequency balance when put in a certain case.

Hmmm...

I would think that the freq balance difference would be audible in free air if it was big enough for the case to have that much of an effect. I can't say I have ever heard this. Usually, what I hear in free air is pretty much what I hear when the fan is mounted in the case, except for low frequency vibration effects, and sometimes, ticking / buzzing (if it is bad), which are also vibrations that can be amplified by the case. But in my experience, if two fans sound the same in free air and have the same low freq/vibration, then they sound the same in the case.

You must have a very unusual pair of fans & an unusual case. Either that or maybe I still don't fully understand what you are saying?

[Gorsnak]

I would think that the freq balance difference would be audible in free air if it was big enough for the case to have that much of an effect. I can't say I have ever heard this. Usually, what I hear in free air is pretty much what I hear when the fan is mounted in the case, except for low frequency vibration effects, and sometimes, ticking / buzzing (if it is bad), which are also vibrations that can be amplified by the case. But in my experience, if two fans sound the same in free air and have the same low freq/vibration, then they sound the same in the case.

You must have a very unusual pair of fans & an unusual case. Either that or maybe I still don't fully understand what you are saying?

No, I think you're largely right - if the frequency balance difference is great enough to have this effect, it will likely be audible. However, we will rapidly run into a situation where I cannot reliably predict which of two fans will seem louder once they're stuck onto my heatsink if the one is slightly louder but overall slightly higher in pitch. (Well, okay, to be fair this isn't really true, because I won't be able to hear most fans at all, but let's pretend a bit.) All I'm saying is that a nice graph charting voume against frequency would be very useful data, and at times more helpful than subjective descriptions of the quality of the sound.

And yeah, my case is a tad unusual.