Fan Test System, 2010

[compet_pl]

Very good initiative!!!
Can you measure the fans at 5V, 7V and 12V also because many of us are using the fans at 5V or 7V without a fan controller?

Yes!
It's VERY important, to have such data.

[MikeC]

Wow, that looks nice. Could you fit a northbridge/chipset cooler in there?

No, it's the little brown resistor & black transistor pair that need to be cooled, the fan is blowing sideways on to them. There's another pair of the resistor & black transistor on the right, above which is a space for a similar little blower fan. It has to be some kind of clamp heatsink that will press against both the transistor and the resistor, big enough to dissipate the heat. I just haven't had time to scour through my parts bins (not likely to have anything quite right) or visit the local electronics parts stores.

[KayDat]

Oh, I see it. Next to the "WARNING HOT!!" message huh? Damn, that's tough.

[brilong]

btw, we need some help with a bit of coding for our fan articles. Please see this:

That type of interactive chart would be wonderful for us to use for measured fan data. Can anyone help us develop a similar function?

I was just looking at the code behind this interactive graph and it's not as slick as it looks. The line graphs were statically defined. For example, here are two of the lines which get layered on top of the background image (prepend the website name since I cannot post links):

/graphbits/wb1.gif
/graphbits/wb2.gif

I imagine you would like something truly interactive and generated on-the-fly. I'm not sure which Javascript library would be best to accomplish what you need, but I will look into it.

How does one ask you to review other fans? For example, I have a MiniBox M350 case with a Zotac IONITX-C m/b and I'd like a slow quiet fan to sit on top of the CPU/GPU heatsink. I found the Scythe Slim fans (100mm & 120mm). They are only 12mm thick and could fit between the heatsink and the top of the case. I have not yet purchased either of them. Any chance you could review these slim fans? Does Scythe provide samples or would an end user like myself need to donate the fans?

Thanks!

[MikeC]

brilong --

Thanks for your input re- the graphs.

As for fans, I think those Scythes are in the review roster already. Please, don't anyone send any fans right now -- we are drowning in samples!

[widdlecat]

[...]
Please, don't anyone send any fans right now -- we are drowning in samples!

That's actually a good sign of a strong... errr... fan base!?! LOL!

[fwiffo]

Maybe gRaphaël for the graphs?

[guerby]

Hi,

Do you have by any chance measured the amp taken by the fans? We have voltage but no way to get to the watt consumed by the fans .

Thanks!

[MikeC]

Hi,

Do you have by any chance measured the amp taken by the fans? We have voltage but no way to get to the watt consumed by the fans .

Thanks!

We did report that in previous reviews but not sure whether we'll bother in future -- the power is minuscule esp. for slow fans. Typically no more than a watt.

[trident]

Great article Mike, very informative.

I have a question.

As you stated in the article, thermal systems (heat source, heat-sink and fan) are very difficult to model. As an example: heat-sink 'X' may out perform heat-sink 'Y' with a 137W input, but the reverse can happen at a different heat input. This is because the thermal resistance of a system is not a constant, it varies with the amount of heat flowing through the system as well as the temperature at the input and the temperature at the exit point. Rather than a measure of general performance, won't the fan ratings be more reflective of which is the best fan to use on a Thermalright Ultra 120 Extreme with a 137W input?

[MikeC]

Great article Mike, very informative.

I have a question.

As you stated in the article, thermal systems (heat source, heat-sink and fan) are very difficult to model. As an example: heat-sink 'X' may out perform heat-sink 'Y' with a 137W input, but the reverse can happen at a different heat input. This is because the thermal resistance of a system is not a constant, it varies with the amount of heat flowing through the system as well as the temperature at the input and the temperature at the exit point.

Remember that we are not comparing heatsinks, however.

Rather than a measure of general performance, won't the fan ratings be more reflective of which is the best fan to use on a Thermalright Ultra 120 Extreme with a 137W input?

Well, yes it is that.... but it's also generally indicative of how each fan would do on many heatsinks. The 137W input is not equal to a 137W processor, i7 or otherwise, for reasons discussed in the article. I'd guess it's closer to a <100W TDP processor.

I'm not going to argue that this is a truly accurate way to assess fan performance in the PC context. I would venture to say that much more comprehensive empirical testing is needed for that -- ie, performance of each fan at various speeds on a variety of hs w/ different heat levels, as well in a couple of case cooling simulations. But such testing is simply too time consuming for us to pursue. Maybe if we had a fat contract....

I do argue that it will be way better than relying on mfg's airflow and noise specs -- if you seek quiet cooling performance.

[guerby]

Hi,

Do you have by any chance measured the amp taken by the fans? We have voltage but no way to get to the watt consumed by the fans .

Thanks!

We did report that in previous reviews but not sure whether we'll bother in future -- the power is minuscule esp. for slow fans. Typically no more than a watt.

Thanks, I found watt values here:

http://www.silentpcreview.com/article695-page2.html

Below 1W for < 1000RPM, around 1W at 1000 RPM and 3.3W for 2000 RPM so non linear in RPM.

For some noisy monster, 15W at 15000RPM : http://www.airperformancetech.com/get-p ... 0&Depth=56

[NeilBlanchard]

Why is there no link to this article on the front page anymore?

[MikeC]

Why is there no link to this article on the front page anymore?

It was there for 11 days, and read >12,000 times, enough, I thought. It's at the top of the fan section and also in the recommended/reference section. If you think it should go back to the front page for longer, that's easily done.

[NeilBlanchard]

Just checking, Mike -- I thought that the entries on the front page just got "pushed down" chronologically.

[MikeC]

"pushed down" chronologically -- normally, yes, unless it's marked to be a sticky at the top of all the lists it is in -- which this article is, so it would stay on top of other newer articles, which is why it was removed from the front page. It's a sticky in both Reference and Fan sections... and it will get read lots more when the fan reviews start coming.

[gojira]

Where did you get that slick die simulator? Did you buy it or make it? Me want.

Also, I love your new vane anemometer. Truly a thing of beauty. +/- 1% fs. Very nice.

As an ex silence enthusiast I do appreciate the fact that you review fans at higher speeds than would be relevant for silent or even quiet computing. I'm sure the rest of the enthusiast community who may not care so much about silence but do care about excessive noise do appreciate it. I know I certainly do. For me anything up to just above the noise level of a Zalman ZM-F3 at 12 volts is acceptable. Obviously that is not even quiet computing let alone silent computing.

As far as relying on temps to judge fans, I do agree that in the end it is the only thing that matters. Obviously the whole point of fans, even case fans, is to reduce temps. But I do see some problems (mitigated by the anemometer readings) having to do with the exponential drop in cooling performance as the cooling ability of the heat exchanger as a function of airflow goes pear shaped. As the fans get closer to that point (which will differ on each cooler depending on fin density and overall size and effectiveness as well as heat load) you are no longer able to compare the fans. This was demonstrated quite well in Ehume's fan review at overclock.net. The temperature differences between fans as the "temp rise" approached 50c became so small that it was hard to tell how much more "work" a fan was really doing. The fans were obviously well into the start of the exponential curve. You really need very accurate thermometers to measure differences of 1 deg c and smaller which you will surely see as you approach the drop off in cooling efficiency of the TRUE. Of course that part of the curve is probably not relevant to silent computing. I see that you are using a Mannix DT8852 thermometer. It looks like that has an accuracy rating of +/- 0.7 deg c. That means as much as a 1.4 degree difference could just be the result of thermometer error alone. The anemometer has an accuracy of +/- 1% (and I don't know of any more accurate ones). That would translate to +/- 3 fpm at 300 fpm, but the difference between fans in terms of fpm is much larger so the 6 fpm error interval is not a problem. OTOH, the fpm readings will only be good for a single cooling device. It is unclear how they will scale as fin density and thickness of the cooler change.

[Scoop]

This is great, looking forward to the results.

I'm especially interested in the difference between a Slip Stream 1600 RPM SY1225SL12H and a Gentle Typhoon 1850 RPM D1225C12B5AP-15. What is the difference in SPL and how did the fans sound subjectively, and of course how they performed temp-wise. These are quite high speed and definitely not quiet at full speed but I'm still crossing my fingers! For some reason unknown to man, the Gentle Typhoon series is almost twice the price of the Slip Streams where I live. They should be better right?

[Erelyes]

Great setup for testing HS fans, I likely.

I would add to the comment that testing @ 12v, 7v, 5v has its advantages. (Even 6v, for those that wire two fans in series). To me it seems easier for a home user to set a fan to run at a specific voltage, than to get it to run at a specific RPM.

I think you're on the right track with ditching CFM (or FPM) ratings altogether for measuring fans for HS performance.

Here's an "outside the box" idea for measuring a fan's performance as a case fan - set up a computer case as a test rig, with a single 120mm intake fan in the lower front, and measure the FPM at the exhaust in the top rear of the case. Seems to me that'd give a fairly accurate idea of the fan's performance in what we actually want it to do (move air in a large area).

[dhanson865]

To me it seems easier for a home user to set a fan to run at a specific voltage, than to get it to run at a specific RPM.

You must not be familiar with speedfan. Try reading viewtopic.php?p=513736 and see if you still feel that it's easier to control voltage than a software controlled percentage.

[EsaT]

I think it might be worthwhile to measure open-air FPM, because that would probably be pretty relevant for people who want to use them as case fans or in other situations involving low static pressure.

Here's an "outside the box" idea for measuring a fan's performance as a case fan - set up a computer case as a test rig, with a single 120mm intake fan in the lower front, and measure the FPM at the exhaust in the top rear of the case. Seems to me that'd give a fairly accurate idea of the fan's performance in what we actually want it to do (move air in a large area).

Original testing method used impeller anemometer and it was proven unreliable method for measuring fan's airflow. This excellent article discussed that:
http://www.silentpcreview.com/article734-page1.html
Sure that improved method wasn't perfect and didn't tell everything about fan but at least it put fans into even playing field with little of unknowns, little impedance and was immune to design peculiarities. (almost like Noctua S12 is designed for fooling impeller anemometer)

Here in Finland we have a saying fitting for our geography: From the ditch to the duck pond.
For overall usefulness this new method pretty much takes results deeper into mud of unknown. It introduces more unknowns, we have no idea about pressure curves of fans and wouldn't have any idea where in it we are because we don't know back pressure.
As TRUE's fin spacing is among densest in heatsinks its impedance is also among highest definitely favouring high pressure fans making generalizing results for lower impedance heatsinks and especially case fan use uncertain:
The 800 RPM Slip Stream L is also formidable, matching the performance of the Nexus fan in our fan test setup, and beating it for airflow and/or cooling when used with the Scythe Ninja Copper.
http://www.silentpcreview.com/article832-page6.html
And then there's this fan comparison in Noctua NH-C12P:
http://www.silentpcreview.com/article846-page6.html


I'm afraid that with all unknowns this is really just fan performance in TRUE and more unrealiable for "guesstimating" performance in various situations.

[MikeC]

EsaT --

The reality is that for our purposes, a fan that has a good acoustic signature is always preferable to one that doesn't; virtually any 120mm fan out there spins fast enough and moves enough air to adequately cool virtually any CPU with the right heatsink. To use your analogy, we're getting out of the water altogether -- I really don't give a hoot about airflow measurements because they're just distractions.

To cite the article you like ( http://www.silentpcreview.com/article734-page1.html ) --

So what practical implication does this have on the quiet-seeking PC builder or modder? What we've been saying all along for years:

* Choose the quietest fan with the best noise signature and adjust its speed to give a balance of cooling and noise you can live with.
* Don't worry about CFM. Just go for the quietest fan and the best low-airflow performance heatsink, and minimize airflow impedances in your case.

It's funny, but true: We went through this entire rigamarole of establishing an accurate system of measuring axial fan CFM only to tell you it doesn't really matter, and not to worry about it.

[RoGuE]

I really don't give a hoot about airflow measurements because they're just distractions.

That's really what it comes down to. When PC enthusiasts start trying to fully understand and model fluid mechanics (yes, air is a fluid), the discussion starts to get a little out of hand. For silent computing, choosing a fan with desirable sound characteristics is far more important than trying to classify fans based on airflow and pressure: it's just too complex!

[Olle P]

Good initiative Mike. The only thing I'm concerned is the TRUE choice for the reference heatsink. ...

That's my concern too, but for a completely different reason!

If I were to create a similar test rig one definite no-no would be to have heatpipes involved in any way. The thermal performance becomes highly unpredictable at "high" and "low" temperatures since you don't know the thermal properties of the substance in the pipes.
Too much heat and it's all gas, resulting in abysmal cooling. Too little heat and it's all liquid.

Case in point:

... the exponential drop in cooling performance as the cooling ability of the heat exchanger as a function of airflow goes pear shaped. As the fans get closer to that point ... you are no longer able to compare the fans. This was demonstrated quite well in Ehume's fan review at overclock.net. The temperature differences between fans as the "temp rise" approached 50c became so small that it was hard to tell how much more "work" a fan was really doing. ...

The solution is to use a solid metal heatsink where the thermal distribution will be consistent no matter what temperature range you're operating at.

I wish ... Fan manufacturers would provide fan curves (rpm/power/cfm/pressure drop) ...

Some of them most definitely do provide such curves. The data sheets from EBM-Papst are for example very informative.

Cheers
Olle

[dhanson865]

Does anybody make a high end heatsink without heat pipes anymore?

[MikeC]

Does anybody make a high end heatsink without heat pipes anymore?

nope.

[Olle P]

This is a rig for testing fans, not heatsinks, so there's no need to bother about what's available on the market.
With a solid heatsink you don't need that much power to reach the same temps.

Cheers
Olle

[datapappan]

(Hope you don't mind me reviving this thread.)

I have two questions about these measurements:

1. Have you done any comparative measurement on noise for the full setup, with heatsink and all? I would think that air passing throught heat sink fins could add noise. If so, a quieter fan (i.e. a fan giving mora air flow at the same noise level) may get added noise when attached to the heat sink, maybe cancelling, or at least diminishing the difference between fans. (Then again, my own testing as well as other reports state that it's disturbances BEFORE the fan that gives extra noise).

2. Are you saying heat sink design is not an issue? I would think that there might be situations when a fan gives better pressure/flow ratio on a more (or less) impeded heat sink than other fans - meaning under different conditions, an other fan might be quieter. This would also be the case when using as a case fan, granted Mikes testing is on the A/B points as per above - otherwise a good fan is still a good fan.

-----
I'll try to explain point 2 with this figure, showing two fans with equal noise, but different pressure/flow ratio.
Point A/B shows a sample operating point for the heat sink Mike uses - the red fan (A) gives better air flow, i.e. better cooling.
In point C/D, with a less restrictive heat sink, the green (C) fan gives better cooling.

-----

PS. Mike, can you now measure as low as 10 dB? Must be eerie being in that anechoic chamber.

/d

[dhanson865]

This is a rig for testing fans, not heatsinks, so there's no need to bother about what's available on the market.
With a solid heatsink you don't need that much power to reach the same temps.

Cheers
Olle

Well you could just take the heaviest heatsink you want and take side cutters to the heat pipes to separate them from the base.

It'd sure make the test synthetic though. I like that his test has at least some comparison to a real world usage.

[Rekonn]

I enjoyed the article very much and am looking forward to seeing more fans compared. Are you planning one massive article or multiple as you get to testing? If given the choice between waiting for months for a big article, or reading a short article every week with a new tested fan, I prefer the latter.

As to methodology - I like how the article compared fan performance at different decibel levels. Since everyone's tolerance for noise is different, I like how you can start at a desired decibel level, and see which fan performs best at that.