Some new test equipment

[zhangmaster12]

You know where to get them in USA? Or just $20 when converted? I was looking for them for batch 2 but found nothing

Now that your first impression of them is strong Cathar....well, I want to find some

What are the fans you have now (or enroute), Cathar?

+1, where can i get them in the US?

[Cathar]

Tried googling for the coolermaster aluminium fans in the USA, and can't find them. They're not even listed on the Coolermaster site anymore, so maybe they're discontinued? That'd be a real pity.

Back On Topic

Test Standard Fan/Radiator Orientation

After some messing about with push, pull, and so on, I've decided to measure everything in fan-pull mode. For every fan I tried, pulling through a radiator managed to give me very slightly higher air-flow than pushing (~3% more with pull/suck). As measured before with the anemometer, air-flow is more even in pull/suck mode too, so it would seem that pull/suck is the best option. In practice (i.e. actually measuring heat dissipation performance) the differences are so minor as to be negligible, but if one had to choose which way would be very slightly better, pull/suck would be it I guess. I think if it came down to a choice that dictated mounting orientation and push/blow was neater, then you'd still choose push/blow 'cos the differences are so minor as to not warrant giving yourself unnecessary grief about always trying to maintain a pull/suck orientation.

So now I've decided to standardise on a pull/suck fan orientation through the radiator for testing purposes.


Test Standard Noise Measurement?

I'm left with the final issue of what is the most relevant way to report noise levels? My thinking so far is this. My sound level meter has a rated noise floor of 30dBA, but in practice will read to about as low as 28dBA. This isn't exactly fantastic, but unless I decide to spend $500 on an audiophile quality sound meter, it's going to have to do.

The meter has a wind-sock on it, so it's reasonably immune to wind-buffeting. My thinking is this: Record the sound pressure level at 2" (5cm) from the center of the fan hub. This is close enough that even with a 28dBA noise floor that even undervolted fans are very unlikely to register anywhere near the noise floor level. A quick test on a Yate-Loon at 7.5v, which is at the minimal readable air-flow through a PA120.1 radiator for me (12cfm), the SPL at 2" from the fan hub with the fan isolated from vibrating on anything else is still ~15dBA above the ambient noise-floor. i.e. the SPL meter really is just measuring the fan noise level far more than anything else, even with the fan undervolted.

I'm thinking of providing 3 noise level measurements. One with the fan unmounted and isolated, the second with the fan mounted onto the radiator but the radiator not mounted and isolated, and the third with the fan mounted and the radiator mounted in the testbed. I believe that between these 3 values that will be enough to inform people of the potential nature of the fan to reverb and excite any enclosures that they're mounted to, allowing people to decide for themselves which measurement best suits their intended application for the fan (although I would guess that most would be primarily concerned with fans mounted onto radiators that are themselves mounted onto something else). We're targetting real-world applicability here, not isolated anechoic chamber noise-levels that provide no real-world indication of what happens when the fan is mounted to stuff.

I'll see if I can put together some sound recordings of the fans when fully mounted, so those who want to have a listen to the type of noise that the fan makes can determine if it's annoying to them or not. Some people are bothered by low-frequency noises, others aren't so much. Same applies for higher frequency noises. What one person can tolerate is somebody else's skull rattling headache.


Looking for some feedback now

Okay, so I'm looking for some feedback as it whether everything I've proposed is satisfactory, or if anyone can suggest any (low cost) improvements? Once I get this all finalised and some consensus happening here that people are happy with how I'm going to measure and report, then I'll commence testing.

[Jedda]

I'm no audiophile. So sorry if this is a dumb one.
Is it possible that the sound of one fan will carry further than another of matched volume but varied frequency?
I'm wondering if anything could be gained from readings at one and two meters distance, since we're looking at practical application? One meter would be a lot of folks distance in normal use. Two would speak to general room noise of an installation.
Or am I asking too much?

[Cathar]

I'm no audiophile. So sorry if this is a dumb one.
Is it possible that the sound of one fan will carry further than another of matched volume but varied frequency?

In air, no. Through walls and closed doors, and inside cases yes. Lower frequency sounds will transmit through obstacles more easily than higher pitched sounds.

I'm wondering if anything could be gained from readings at one and two meters distance, since we're looking at practical application? One meter would be a lot of folks distance in normal use. Two would speak to general room noise of an installation.
Or am I asking too much?

Okay, assuming that there's nothing between you and the fan (just air), then at 20x distance (5cm vs 1m), the dBA should be 10 x log(20) = 13.01dBA less.

I just ran a test just then in preparation to respond to your question.

Quiet room with a measured ambient noise floor of 31.5dBA. It's daytime. I live in a house in an outer-fringe suburb where forested sections are just 1km away, and there's a major 4-lane road about 150m from my house, but behind a small hill so road noise is fairly isolated. Birds chirping outside, and so on.

Now, at this point, I'd like for people who read this to recalibrate their concept of noise levels. Everyone's so used to thinking that 30dBA is noisy due to fan manufacturer ratings, but from reading what some people's comments are here, what you think of as the noise that a ~15-20dBA fan makes is really more like 25-35dBA @ 1m in actuality, and a 30dBA fan is really more like 40-45dBA @ 1m. Standing in the middle of a field in the middle of no-where with minimal breeze is still around a 10-15dBA ambient noise level. Don't ask me how fan manufacturers come up with their absurdly low noise ratings - I have no idea. I think the noise-absorbing anechoic chambers have a lot to do with it, but it's not real-world reality. Sound does bounce off stuff in the real world.

Turned on a Yate-Loon curved bladed fan with a fanless 12v power-supply, so the only audible noise in the room, apart from the rather quiet ambient noise, is the fan itself.

At exactly 5cm distance from the fan hub, I measured 50.4dBA. At exactly 1m distance from the fan hub, I measured 37.3dBA. Lo and behold, a 13.1dBA difference, which is about exactly what the maths predicted it would be.

So if you want to extrapolate noise levels taken at 5cm distance to a 1m distance, subtract 13dBA from the 5dBA reading.

You do have a valid point though when it comes to enclosure excitation. These sounds don't originate from any one point, but really from the entire enclosure, and bounce around the room and just raise the ambient noise floor in general.

My sound test room is pretty much an exact 4x4m square with a 2.5m ceiling, or fairly standard for a bedroom/office, so what I'll do is measure the difference in the ambient noise floor in the middle of the room as a result of the fan and radiator fully mounted in the test unit.

Would that be useful?

[santiagodraco]

Cathar, what sound frequency range is covered by your SLM? It seems to me that testing at 2" from the center of the hub might not give a true representative sample of the "perceived" sound levels to a listener at say 8 feet, especially if the the sampled frequencies are under say, 50-75hz.

Maybe a sample should be taken at another distance, say 12". I realize that some noise levels will be so low that the sound at that distance might not be audible but when testing higher CFMs it might be important to give a more accurate reading related to actual listener perceived sound levels.

If that makes sense

[Cathar]

Cathar, what sound frequency range is covered by your SLM?

31-8000Hz

It seems to me that testing at 2" from the center of the hub might not give a true representative sample of the "perceived" sound levels to a listener at say 8 feet, especially if the the sampled frequencies are under say, 50-75hz.

Well, that's what was being suggested above. Take an ambient room noise measurement, which would be around 6' away from the mounted fan. Am not so interested in fan noise levels when unmounted as fans don't get used like that.

Maybe a sample should be taken at another distance, say 12". I realize that some noise levels will be so low that the sound at that distance might not be audible but when testing higher CFMs it might be important to give a more accurate reading related to actual listener perceived sound levels.

If that makes sense

Well, here you're really just asking me the exact same question that Jedda did. I did the test, and the result at 1 meter (~40") was exactly as mathematically expected, being a reading of 13dBA lower than at 2".

I can take a sound recording at 12", or something like that, and provide that. That'll give a decent indication of "perceived sound".

[santiagodraco]

You should slap me harder than that for not reading other posts properly

But thanks for the response. Sounds like you have the bases covered pretty well. Of course you can always toss in the third measurement test, the Catharameter sound level checker.

[Hey_Its_Cole]

Actually I'm a racetrack regular. To date, my last 4 accidents have all been at the race-track (over a period of 11 years). I guess it comes with the territory when you're pushing the limits. This most recent accident has me seriously questioning the wisdom of it though (going to the racetrack). 8 weeks on and I still can't walk, and my right hand is at about 50% mobility and 25% strength, and I'm starting to think that maybe someone's trying to tell me something. Let me just say that you never really appreciate the ability to walk and get around freely until you don't have that ability. I can't use crutches 'cos my broken wrist can't support the weight, so I'm effectively near immobile, and extremely frustrated. Needless to say that the wife is pretty cranky with me and doesn't want me to go to the racetrack again. I'm my own person but even I'm starting to appreciate her position when I can't do most anything I want to. I look at my kids and think how bad it would be if the injuries were permanent. If I didn't have the responsibility of my children it might be different, but I think I'll be giving up the racetrack for a while.

My dad reached that decision recently after breaking his collar bone on the track, he used to road race in the 80's (he is 51) and started getting back into it after a 17 year hiatus. He always bragged about being the only road racer he knew to never break a bone but then karma reached up and bit him in the ass. My mom talked him out of racing again, he is a dentist and really can not afford the injuries, that and he is getting to old.


Back on topic: I want to she what the golf ball fans can do.

[Cathar]

Man, what I wouldn't give for only a broken collarbone at this moment.

On a side note, I am ready to throw this freaking anemometer out the window. It cannot get any consistency out of it. One minute it's reporting one air-speed consistently, then wait a couple of minutes and it reports something 20% higher and will do that for 5 minutes, and then it'll report something else for another few minutes.

I then go back and revisit stuff, only to find that the readings have all changed. So that's it. I'm going to have to scrap all the setup work I've done to date, 'cos I can't trust it and have no faith in it, and get myself a better anemometer, perhaps something like a hot wire anemometer instead.

[dinos22]

Man, what I wouldn't give for only a broken collarbone at this moment.

On a side note, I am ready to throw this freaking anemometer out the window. It cannot get any consistency out of it. One minute it's reporting one air-speed consistently, then wait a couple of minutes and it reports something 20% higher and will do that for 5 minutes, and then it'll report something else for another few minutes.

I then go back and revisit stuff, only to find that the readings have all changed. So that's it. I'm going to have to scrap all the setup work I've done to date, 'cos I can't trust it and have no faith in it, and get myself a better anemometer, perhaps something like a hot wire anemometer instead.

outch!!!!!!!

[v0dka]

Sent you the fans today

[Cathar]

Hmmm, I think I found the source of my grief. It's not the anemometer, it's the fan.

From when I first turn the fan on:
After 1 min => 1375rpm
After 2 mins => 1385rpm
After 5 mins => 1400rpm
After 10 mins => 1405rpm
After 20 mins => 1405rpm
After 30 mins => 1405rpm

Thing is, every 30rpm equates to around an extra 1CFM through the radiator, and I started the fan up in the above test only after a 10 minute cool-down, so if it had started off really cool, the differences may be even larger. When I'm measuring stuff, a 50rpm difference which might have been occurring is a big deal.

It seems that in order to take accurate readings, I need to allow the fans to "sit" for ~10 minutes at any new setting before trying to take a reading.

The calibration stuff is all still okay though, 'cos I was holding a fixed RPM by ranging the voltages, but for the tests I was just trying to do, I was just setting a voltage and taking a measurement, and then set another voltage, and then I went back to the first voltage and noticed that the fan was now performing better. Now I understand why. The fan bearings must warm up to a stable operating temperature first.

[Cathar]

Sent you the fans today

Off to check PM's! Thanks!

[Jedda]

Oh drat, all fanned up and no where to blow!


On a more serious note.
Good old math, eh? We can rely on your equation to calculate any installation distance we happen to have. So that's very useful.
As useful, for practical purpose, as an external case, 'case excitation' reading may be, it's an awefull lot of additional set up work. Probably best left to another day. All though I'm getting a glimmer of an idea for recording fans and displaying a histogram of each recorded sound shape over the audible spectra.
That could provide a visual predictor of likely case excitation sufficient to our purpose.
Again, another day?

Hmmm, hot wire anemomete. I see a Lustrom for $350 Oz money. Ouch!

Edit,
This post makes more sense read before Cathar's last. lol
Good to see thought win out again.
However it raises an interesting point. For years I've been oiling my fans pre installation, with a couple of drops into the bearing. It's been an attempt to improve longevity but it may effect bearing noise.

[aussie-revhead]

Cathar , I havent read right through these pages yet , but you ordered some fans , I am looking for fans for a 3x120 rad (not sure which yet). I am using a fan controller , so I want them quiet when turned right down and with decent air flow when turned up, what would you get ? I saw a pic back there with three push fans and three pull fans , is this needed ? Or is it beneficial ?

Any help would be great.

[Cathar]

Cathar , I havent read right through these pages yet , but you ordered some fans , I am looking for fans for a 3x120 rad (not sure which yet). I am using a fan controller , so I want them quiet when turned right down and with decent air flow when turned up, what would you get ? I saw a pic back there with three push fans and three pull fans , is this needed ? Or is it beneficial ?

Push-pull (6 fans) is not really needed. Will make more noise than it'll help in terms of extra performance.

I'm in the process of establishing just what is the best fan that would do what you're asking.

It all depends on what you're trying to achieve when you say "decent air flow". How decent? Do you want to try for suicide overclocking runs (high powered fans)? Want to turn it up on occasion during summer (medium powered fans)? Want decent performance but not at any noise (low powered fans)?

[aussie-revhead]

I would like quiet fans, but would put up with some noise while benchmarking if it makes a difference. With my last two water cooling rigs there was little temp difference between low fan speed and high fan speed , but I will test it again when the new rig is running.

[Cathar]

Back on the topic of allowing fans to warm up. Let it cool for 45 mins.

On startup: 1305rpm
After 1 min: 1315rpm
After 2 mins: 1325rpm
After 3 mins: 1335rpm
After 4 mins: 1350rpm
After 5 mins: 1355rpm
After 10 mins: 1370rpm
After 15 mins: 1400rpm
After 20 mins: 1415rpm

Voltage was monitored the entire time, and remained consistent. The fans just warm up and vary in speed. 110rpm is a very significant difference. That's 4-5cfm through the radiator, and it's no wonder I was pulling my hair out wondering what was going on.

This exactly explains why I wasn't seeing repeatable results before. Things were changing even over a period of 20 minutes.

[Vapor]

Looks like Yate Loon SL to me....

You should try a new vs. old fan as well

[Mad_Man]

its explainable, as the fans starts up, bearings are cold, so the lubricant have higher viscosity, as it warms up, viscosity goes down=friction goes down and so are the rpm going high...

sad is, to take consistent and acurate results you have to wait for each fan ~30min before taking any measurements...

... or warm it up with a hair-drier

[Cathar]

Well, just spent the last 90 minutes gathering data.

Measured the:

Air-flow through the radiator
The RPM's (held constant in 50rpm steps from 1000-1450rpm)
The voltages at the end of the run for the rpms
The SPL @ 5cm distance
The SPL @ 1m distance (meter pointed at center of fan)
The ambient room SPL

This is all with the radiator mounted in the test box, and so includes box reverberation/excitation.

The fan in question is a Tricod Science SPDL1225S. Yes - it's a sleeve bearing. This is actually my reference test fan from the batch of Tricods I picked up a while ago. I just picked it out at random. It seemed no noisier or quieter, or spun no faster/slower than the others. It hasn't seen a lot of use, maybe 50hrs maximum, but is about 2 years old.

Well, that's 1 fan down. Lots more to go.

I was able to get consistency by monitoring the RPM, and basing everything around that. I'll generate a flow/noise graph from the data, but the data points are all based off RPM settings, not voltages or specific air-flows. The air-flow range will still cover from 11cfm up to the maximum the fan supports at <13v. I allow for up to 13v 'cos a number of PSU's have strong 12v rails, so it'll cover those scenarios too.

[Jedda]

That oughta tell me what I want to know.
WTG Cathar.

[jinsean]

hey cathar,

I heard that the benefit of a push/pull setup (e.g. 6 fans on a mcr 320) was more beneficial @ lower RPMs. Any truth in this? I plan to keep my fans around 900-1000rpm.

[Cathar]

hey cathar,

I heard that the benefit of a push/pull setup (e.g. 6 fans on a mcr 320) was more beneficial @ lower RPMs. Any truth in this? I plan to keep my fans around 900-1000rpm.

It may very well be. Below ~1000rpm the pressure with most 12cm fans drops very rapidly. For example, the air-flow difference through the radiator between 1000rpm and 1450rpm for the fan I just tested was ~11cfm@1000rpm and ~33.5cfm at 1450rpm. Air-flow is roughly proportional to the square of the air-pressure, so that gives you an idea of how fast the pressure is dropping as rpm's are dropped.

Putting a pair of fan each side on will roughly double the pressure, but then again, simply running the fans at 50rpm higher will do that too. Two fans equals +3dBA, while in my testing an extra 50rpm = ~1.5dBA (at 5cm distance) at around the 900-1000rpm range.

i.e. You could 1 run fan at 1000rpm and it'd create more air-flow than at a comparable noise level of 2 fans at 900rpm.

It may work okay for some fan models. I don't know what they would be though.

[jinsean]

It may very well be. Below ~1000rpm the pressure with most 12cm fans drops very rapidly. For example, the air-flow difference through the radiator between 1000rpm and 1450rpm for the fan I just tested was ~11cfm@1000rpm and ~33.5cfm at 1450rpm. Air-flow is roughly proportional to the square of the air-pressure, so that gives you an idea of how fast the pressure is dropping as rpm's are dropped.

Putting a pair of fan each side on will roughly double the pressure, but then again, simply running the fans at 50rpm higher will do that too. Two fans equals +3dBA, while in my testing an extra 50rpm = ~1.5dBA (at 5cm distance) at around the 900-1000rpm range.

i.e. You could 1 run fan at 1000rpm and it'd create more air-flow than at a comparable noise level of 2 fans at 900rpm.

It may work okay for some fan models. I don't know what they would be though.

Wow. Great info. Thanks, Cathar.

33cfm @ 1450 RPM seems weak for a 120mm fan. I guess then the ~1000 RPM threshold could be lowered a little for say a fan that outputs 40+CFM @ 1450 RPM, assuming the CFM scales down somewhat linearly.

How much does the fan speed & CFM effect the chip (cpu / gpu) temperatures? Are we talking about 1-5C differences b/t 30cfm and 50cfm or more like 8C+ differences?