Fan PQ curve testing - Part 1

[Martinm210]

Ok, I rebuilding this thread and making it easy on myself this time around.

INTENT

The purpose of this thread and intent is to have fun and try doing some fan testing a little differently than has been done in the past. I've done enough pump testing to know that a pump's maximum flow rate and maximum pressure alone are NOT enough to tell the whole story on how well a pump performs. So my intent here is to attempt testing a full PQ curve of the fan to more fully understand how it performs under more than one condition. To do so I've built my own little flow chamber and it's not at all perfect as I've tested with it I see several areas where I would improve upon it if I had planned to build another one. Regardless, I plan to test everything with the exact same setup, so regardless of errors...the data should at least be relative to other tests as I've learned that's generally much more important than the numbers themselves.

TEST RIG

http://img3.imageshack.us/img3/1310/fantestingpqrig.jpg

http://img3.imageshack.us/img3/8294/...onradiator.png

The test rig consists of an anemometer permanently fixed to the inlet side of a split flow chamber. I've then arbitrarily calibrated it to one fan the yate loon slow speed. To calibrate it, I can adjust the anemometer's flow chamber area and it gives me a direct CFM reading. In addition I've installed a Dwyer Magnehelic 0-.25" pressure guage and installed two pressure tubes aimed and nearly touching the hub of both sides of the fan to measure a pressure difference. This gives me the ability to measure both pressure difference and air flow. To adjust restriction, I simply move a box closer or further from the chamber outlet. In essence this gives me a means to measure pressure difference and flow rate while adjusting restriction to plot out the full PQ or Pressure vs. Flow rate curve like is done with water pumps.

I'm performing three tests:
All tests are done after the fan is allowed to warm up at least 10 minutes.

1a) Low Voltage Startup Test - A quick little test to see what the minimum voltage to start the fan is to the nearest 1.0V.

1) Pressure Vs. Flow rate test - In this test I plot out the full curve, usually at a fixed 12.00V +-.03V. I first start with the chamber flowing freely. I zero out my guage at startup of each test. Then start recording, first at maximum flow (no restriction) and read RPM from my crystalfontz. Then I slowly start adding restriction by placing a box at the outlet and strive for reducing flow rate by about 5CFM increments. I then make another recording of CFM, Pressure, and RPM. I continue this until I begin having CFM read errors. This occurs when the CFM starts to climb as I increase restriction and it due to my flow chamber being too small, but thankfully it's generally off the curve enough that it's data that's not as important. I skip over this area, stop flow completely and get the last reading which is completely blocking the air flow and measuring static pressure. This is the scientific big picture test.

2) Test 2 is done to see affects of voltage on an open air condition. I simply leave the end wide open, adjust voltage and make measurements of amperage, voltage, RPM, and CFM. This test just gives you an idea how voltages changes would affect the fan in an installation like a case fan and it's more of a practical results test.

3) Same as test #2 except I place a heatercore radiator against the outlet. This is also more of a practical test seeing actual CFM on an dense heatercore radiator giving you some idea how voltage affect results of CFM on an actual radiator type of application.


WHAT DOES IT PRESENT?

It does not present noise AT ALL. Let me be clear than this is strictly performance only testing. CFM is however extremely important to radiator performance and this testing shows just how much difference there is between performance levels. In the testing done so far, I've seen some fans produce as little as 10CFM on the radiator and some as much as 80CFM. That's 8X the performance difference.

Is 8X the CFM going to mean 8X the radiator performance?...NO...but it can mean up to 5X the performance difference on a dense radiator. I would estimate the radiator gain is somewhere between 50%-70% of the CFM difference.

WHAT ABOUT TEMPERATURE, HOW DOES THIS TRANSLATE TO DEGREES?

In order to make any sort of "Degrees" estimate, you absolutely have to know what your water vs air difference is, I'll call this the Water/Air Delta. If the air entering your radiator is 20C and your water temperature is 30C, your water/air delta is 10C. On average I suspect 10C is about average, 5C is extreme performance levels, etc.

OK, so now lets use an example. Lets say your water/air delta is in fact 10C with the fans you currently have. Now lets assume that the fans you currently have are producing around 20CFM through a radiator. What would happen if you were able to double the CFM amount you might ask and produce 40CFM. If you did gain 100% performance your water delta would be exactly 10C/1+1 or 5C. But assuming it's 70% gain, it would be 10C/1+.7 or 6C for a 4C gain.

Lets do it again...say you had 20CFM and increased that by 10CFM. Your result is 10C/1+.7(10CFM/20CFM) or 10/1.35 = 7.4C, so that's about a 2.6C gain.

Now do it again, say you have a really high performance setup (ie extra large radiator or already higher speed fans). Assume it's a 5C water/air delta. Assume you have 30CFM (Rad producing) fans and increase fans to produce 40CFM on a radiator. Your gain would be 5C/1+.7(10cfm/30cfm) = 5/1.233 = 4.05C for a net gain of only 1C.

So as you can see, if you're already very strong in the radiator and/or fan department, extra fan power does little for you....BUT if you're on the other end, you can really see some decent gains by using stronger fans.

WHAT I WOULD LIKE TO CONVEY

More than anything I'd like people to better understand the power differences between fans. For example, a 900RPM 25mm fan may produce as little as 10CFM on a radiator and a 3000RPM 38mm fan can produce over 80CFM. So that's only triple the RPM for 8X the performance. It is an unfortunate reality that fan performance falls on it's face HARD as RPM go into silent territory. Fans are many times more efficient running at obnoxious RPM levels and you really need to understand the give and take. People that want to run sub 1000RPM fan need to face reality and install MANY TIMES the radiator frontal area than someone willing to even put up with moderate noise level. The fan and the radiator are a package performer. If you go on assuming a radiator will perform the same with any fan, you are sorely mistaken, the differences are many fold.

I personally don't like much noise, but I also run a dedicated quad radiator for my CPU all by itself. If you want high performance water cooling and silence, you simply need to understand that you need alot more radiator. At the same time, someone that has no issue with noise, can be happy knowing that a single 120mm sized radiator with a 3000RPM 38mm fan can have the capability of dissipating more heat than a quad radiator 800 RPM fans.

Bottom line, radiator performance differences are HUGE depending on the fans used, do not overlook the fan component. The FAN USED IS FAR MORE IMPORTANT THAN THE RADIATOR BRAND AND EVEN THE SIZE ITSELF!

Now don't take this as a "Go buy the strongest fan available". You won't like that, just understand how the fan is part of the dissipation package, and it's going to be a noise give and take. Unfortunately it's a pretty major give away performance for a take in silence.

SUMMARY RESULTS



THINGS LEARNED

• 38mm fan's extra thickness doesn't seem to help until about 2000RPM +
• Curved fan blades generally produce more CFM per RPM than straight blades
• Fans with a higher number of fan blades produce more CFM per RPM
• 38mm fans in general seem to have trouble with undervolting motor noise more than 25mm fans
• Fan RPM VARIES with restriction and it's not at all linear
• Some fans are a fair amount OFF on the specified RPM, hard to know what you're buying.
• Fan efficiency significantly increases at higher RPM on all fans.
• Fan performance is NOT proportional to cost.
• There are alot of GOOD fans that work well, many times very hard quantify how one is necessarily better than the next. Many times it will simply come down to personal preference on things you just can't quantify.

Here is the same charts expanded to see the higher speed stuff:

[Hondacity]

welcome back to testing..the itch i presume

can we have the toys list?

they look soo expensive

[Sadasius]

Well alright.....Glad to see you testing again and in a very interesting area ta boot. The information from this line of testing will be worth gold when your done. Your the man!

[Martinm210]

welcome back to testing..the itch i presume

can we have the toys list?

they look soo expensive

Hehe, a little itch, but part guilt from not using the toys I bought so long ago.

The anemometer is just an off brand I bought on ebay, I think I paid something like $40 for it and the pressure gauge was also in that same price range, bought new on ebay a while ago. The Mastech 0-30V, 5AMP PSU was something I bought a while back, I think they run around 100 bucks. Oh and then my $3 HF multimeter to check voltage right at the plug.

The box was scrap from my kitchen cabinet project....still in progress.

[Martinm210]

Martin you forgot to moniter fan noise with a dB meter

Great to see you came out of the closet......again.....J/K

I knew the tinker in you wouldn't stay to dominant

welcome back

Thanks!
I'm still trying to figure out how to do that one. These 1000RPM fans are soo quiet, I think I need to run some sort of separate test mounted on a radiator and just measure voltage again. I struggle with having a sound testing compatible environment too. Somehow the 5 year old and 2 year old don't seem to understand the importance of science just yet...

Sound is always one of those subject things too that db alone doesn't do it justice. My thought is something like mount the fan to a radiator, then video record the ambient level, and change in sound level while recording both the sound itself an video recoding the varied voltage. This way you'd get the sound level numbers and the sound quality itself measured. In addition, you'd capture noise generated by vibration/etc on the radiator.

To me, open air flow sound level measurements as we get in the specs is pretty meaningless and too hard to capture in a noisy environment. On a radiator though, I think I can measure something there...maybe.

But, yes...I will do something along those lines...that's "Part 2", part 1 is this P/Q curve thing.

[Vapor]

This is awesome!

[Hondacity]

Thanks!
I'm still trying to figure out how to do that one. These 1000RPM fans are soo quiet, I think I need to run some sort of separate test mounted on a radiator and just measure voltage again. I struggle with having a sound testing compatible environment too. Somehow the 5 year old and 2 year old don't seem to understand the importance of science just yet...

Sound is always one of those subject things too that db alone doesn't do it justice. My thought is something like mount the fan to a radiator, then video record the ambient level, and change in sound level while recording both the sound itself an video recoding the varied voltage. This way you'd get the sound level numbers and the sound quality itself measured. In addition, you'd capture noise generated by vibration/etc on the radiator.

To me, open air flow sound level measurements as we get in the specs is pretty meaningless and too hard to capture in a noisy environment. On a radiator though, I think I can measure something there...maybe.

But, yes...I will do something along those lines...that's "Part 2", part 1 is this P/Q curve thing.

i'm recording the sounds too. i love the gentle typhoons. cant wait to see your recordings...wish i can send you my 10different fans

[CyberDruid]

Thanks Martin

[skinnee]

Now this is exactly what is needed!

Thanks Martin!!!

[Martinm210]

Good stuff Martin, but this kind of confuses me. How good something have 0 pressure when it's moving the maximum amount air?

andyc

I didn't show you in the pictures, but I am adding restriction to get at the middle points and the static pressure point.

I should probably add one more test at the bottom on a radiator. I'm just finding that my anemometer looses accuracy below 10cfm. So on fans like this noisblocker, I start hitting that minimum 10cfm wall at around 9Volts. It'll probably work ok for the stronger fans though.

[Martinm210]

Martin
yes you are correct on sound levels in open setting i agree with ambiance testing with a mic sortta like seeing is believing eerrr hearing

Yeah, I've always struggled with how to do sound testing. I figure a video and sound clip to go with that is about the best we can do. Then you should at least pick up on any odd clicking or other annoying type of noise.

[Martinm210]

So when you cause restriction, that causes sort of a back pressure and that's what you're measuring, how much RPM given a certain restriction or back pressure? So you can measure the fans ability to push thru the fins of a rad for example.

Or am I still off?

andyc

I just finished retesting the Noiseblocker fan since I made some minor tweaks to my setup. I also added one more test of voltage and RPM while testing on the Bonnie Heatercore.

Here is the result of that, you do have a minor drop in RPM, but it's not much.


I'll grab a picture of the actual added test in a second.

This fan is extremely silent and very high quality sound. it doesn't turn as high an RPM as the Yate D12SL12's which will produce more pressure/flow, but if silence is key priority and you want 1000RPM and silence, these are very good. RPM per RPM, these fans were a little better than the YL D12SL12s

[Jiangxue]

glad to see u back testing again martin! and thanks for the info as well.

[Martinm210]

And this is what that very bottom test looked like running it against the heatercore. The heatercore is very restrictive for airflow, so most rads will probably have noticeably less pressure drop and result in more air flow than this, but it give you an idea.

I'm measuring the pressure difference on each side of the fan with the gauge and the air flow on the inlet side.

It's alot like a pump test, measuring the pressure difference relative to flow rate. It seems my last test is showing more of a prominent step in the curve. That's the sort of thing I'm interested in seeing and should show up alot better with the stronger fans.

[Pedalmonkey]

The Master Is Back!!!!! Woot!!

[xpatar]

Martinm210 awesome once again, just waiting for more results

[HESmelaugh]

Wow, amazing testing!

Thanks, Martin.

[Martinm210]

I'll start collecting each of these in their own post.

Here is the Yate Loon D12SL12 Straight blade version:

Photo of the fan tested:


Testing Results:


Overall a good fan, a bit more powerful than the NB XL-1 but mainly due to more RPM and they aren't quite as silent as the NB XL-1, still simply a good fan for the money.

[Boogerlad]

Thanks martin! Do you have a san ace?

[Martinm210]

And here is probably my most used fan. The Petra curved D12SL12, I also like the medium speeds I'll try next, but these yates with the curved blades are a little better than the straight blade types.





Overall very strong performance and better than the straight blade types.

I bought this one from Petra's a while back. Pay attention to the RPM numbers on the radiator, these are turning quite a bit faster than 1350. Even on the heatercore, it was turning around 1485RPM at 12V... Not a big deal if you're adjusting with a fan controller, but something to note.

[Martinm210]

Thanks All!

I've updated the OP here
http://www.xtremesystems.org/forums/...d.php?t=221899
with a summary chart of all the fans tested so far. This includes the performance PQ curve comparison as well as an RPM vs Flow on the radiator comparison for a look at the fan blades performance per RPM.

I'll being doing the noise tests in part 2 later on

[ecat]

All good info. Nice work, thank you

[Martinm210]

Thanks martin! Do you have a san ace?

Sure,

Yes, Alex sent me a San Ace 109R1212H1011 a while back.

I will be going on the bench before too long

[Hondacity]

i can't wait to see the results...i hope you still have the kaze3000s

[Martinm210]

Yes, I have the Kaze 3000, also a 1000 that will be interesting to see.

Here is the noiseblocker XL-2. It's a very nice fan, it won't show in this round, but the sound quality is better I suspect than the curved yate, it also undervolts down to nothing with a very low startup voltage.

First the photo:


And the results, this time I recorded RPM while doing the first PQ curve test. Not sure what's going on with the RPMs, but they did a roller coaster in this test. I'll see if that sort of trend continues.


This may be my new favorite fan (I just like the performance/noise level, it has a very smooth bearing/motor). I just wish I could buy them at yate prices...