Gathering Radiator Thermal Testing Tools, now which fans?

[Martinm210]

I just opened up a new parcel today containing the last of my water heaters. First I decided on a smaller variac, and then I had hoped I could use the variac with an aquarium heater. I just tried out the heater with the variac and got a nice 0-300 watts of finely tuned heat load. Now between the three of them I can add 0-900 watts of heat to the pump heat dump and tinker with some thermal testing..:up:
http://www.martinsliquidlab.com/img/RadiatorHeatLoad.jpg

The water heaters will go in my custom 4" black PVC reservoir you see to the left in this picture, it should work pretty well I think.
http://www.martinsliquidlab.com/img/XSPC-DualBayReservoirPump-3.jpg
I'm not going to bother with insulating lines and all that, I just want to record water/ambient deltas using my crystalfontz and Dallas probes. I figured that'll be good enough..

Now for the questions?


Which Fans do I test with and at what voltage?
What heat loads should I test at?


I was thinking about trying to match some common heat load values as I plot out the trend.
Maybe something like 100watts, 400, watts, 900 watts or something similar. Then if I plotted a trendline for that I could interpolate the 10C delta point. Maybe something less for single/double radiators??

Also not sure on the flow rate variable. Maybe I'll run all of the above at a fixed 1.5GPM, and then do one series of tests to see how the radiator reacts to varied flow rates..?

Open to suggestions, I havn't found any sort of standard for this type of testing, so I'm scratching my head as I go...

This should be fun..:up:

Better start saving up for a few fans. I have a few yate loons sitting here, and three of the 86cfm panaflos, but nothing else. Seems some of the thermochill and swiftech testing used Nexxus fans which run at 1000 RPM. With yate loons being more popular, I think those are a giveme, but what else is really popular for radiators??

[Yeknom]

Scythe S-Flex maybe? Also, if you could get some Slip Streams, it could confirm Vapor's thoughts on them with rads.

[FuriousSalesman]

Is it ok suggest Yate Loon fans? I would also like to suggest Sanyo Denki fan's. My thought is the added price and availability of the fan isn't enough for the performance you get.

[Nate P.]

Maybe the Zalman FM-03's? They are great fans, I'm surprised they haven't been accepted by more people. They scored great in Vapor's testing, and I have six in the computer next to me. Excellent fans.

[Martinm210]

Thanks, I'm looking.

Might be a while before I round some up...I have a couple of unused yate loons sitting here and 3 of the medium speed panaflos, but that's it..

[Pedalmonkey]

what rads are you planning on testing? it would also be very cool to see how the flow rate in a system for each type of rad effects the wattage taken out of the liquid. soo many variables you could test, limitless combonations.

[LinusTech]

Martin - shoot me a PM. I may be able to arrange to sponsor you for some fans. (no guarantees yet though).

fans I'd recommend using:

Noctua NF-P12-1300 (best low noise high static pressure fans bar none)
Yate Loon med. speed fans.
some kind of horrible noisy thing.

[Martinm210]

what rads are you planning on testing? it would also be very cool to see how the flow rate in a system for each type of rad effects the wattage taken out of the liquid. soo many variables you could test, limitless combonations.

I've got a pair of 480GTXs, a PA120.3, MCR320, MCR220, Bonnie Heatercore, XSPC RS 120.3, 120.2, 120.1, and I have a source for potentially many more if I put in the effort to test them.

Yeah, I had figured on that too.

One thing I want to do for each fan combination is determine the 10C delta heatload point. I think I can do that easy enough, but just testing several heat loads and plotting the trendline. I have no idea if that trend is going to be linear or some sort of curve, but I could figure that out. And I would do all of this at a fixed 1.5GPM since that's probably average.

Then I thought I might run another series of tests to see the effects of flow rate. Maybe after I determine the 10C delta point I can apply that heat load and then run the test at .5, 1.5, 2.0, 3.0 GPM and plot that out.

But yeah, the variables can be overwhelming, so I'm going to have to keep it pretty simple or I'd get bored after just the first test..:D

[Martinm210]

Martin - shoot me a PM. I may be able to arrange to sponsor you for some fans. (no guarantees yet though).

fans I'd recommend using:

Noctua NF-P12-1300 (best low noise high static pressure fans bar none)
Yate Loon med. speed fans.
some kind of horrible noisy thing.

You have PM..:clap:

[starlon]

I'm glad to see some love for the Noctuas :)

[twwen2]

Saw the 480GTX in the list. It will be good to get some definative comparison testing done on the HWLabs rads, there's really not much out there atm.
If it's possible could you include a magicool extreme 360 slim profile (the same as BIX 360 i think?). They look good, but i've heard mixed reports about performance.
As for fans, probably one in each speed category (like cathar did in his fan tests). E.g 800, 1200 and 1600rpm. Maybe YL's and S-Flex's. They are widely used/accepted as being class leaders (S-Flex) or good value for money (YL).

Cheers Martin, you da best. Looking forward to the results in due time :up:

[The[R]eaper]

+1 for the Scythe S-Flex series. i use the "F" model and they are very silent and push a good amount of air.

also maybe some high >100cfm fans, if you can. to see how much of a temp decrease there would be between the medium speed ones most of us use.

also if the noise/performance ratio is worth it.

by the way i greatly appreciate your testing martin. thanks

[ArtosDracon]

I'd say the listed noctuas, some scythe ultra kaze 2Ks, the ZM-F3 of course, some Delta TFB EH and some YL mediums. That should represent the best variety of fans. Of course som Sanaces if you can get your hands on some.

Seems to me that, to be completely thorough, these would have to be three part tests. Part 1 with fixed flow(air and water) but varying heat loads, Part 2 with fixed airflow and heatload but with varying water flow and Part 3 with fixed water flow and heatload but with varying fans. Once you did part one you could test parts 2 and 3 at the radiators most efficient heat load from part 1 so the deltas would be slightly exagerated but, also the most poinient.

EDIT: oh, and if you're not able to get a sponsor for fans, I have a few of the fans on my list that I'd be very happy to lend you as i'm not using most of them. I'd have to see how many I have full sets of 4 of though. I know I've got at least 4 ZM-F3s, I don't have any TFBs but, I do have atleast 4 of the Delta FFB 38mm, I don't think I have 4 of the same rpm ultra kazes but, I have three of the 3K rpms, no noctuas or YLs though.

[LogAn'sRun]

S-Flex seems logical, as it's a very popular fan.
If possible, I'd like to see the Minebea series in there as well, since it has been recommended over and over.

[rockqc]

+1 to the Ultra Kaze 2k.

[Xilikon]

IMHO, try to keep it close to other test beds so you will need 3 versions of fans. Use the popular ones so it would be easier for us to reference and my own pick would be those :

Noctua NF-P12 or Yate Loon D12SL or Scythe SFF12D/E for low speed
Yate Loon D12SM or Scythe SFF12F for medium speed
San Ace 1011 for high speed

See with NCIX what they can offer. I know they could provide the low and medium ones easily but not sure about the high one (must be over 100 CFM). Also, if you need some sponsor, count me in if others can't (I can give up to 50$) ;)

[Vapor]

Noctua NF-P12-1300 (best low noise high static pressure fans bar none)
Yate Loon med. speed fans.Unfortunately, testing is not gonna show much of a difference between these two....:-/ D12SM-12 is ~55CFM and the P12 is only a little behind. Both probably have similar radiator characteristics too.

I recommend the three S-Flexes at 12V. 800RPM, 1200RPM, and 1600RPM. Also throw in some Panaflo H1A's for ~100CFM.

You don't need the best of the best, just something that's easy to relate to for users.

[leo_bsb]

I would try the YL on slow, medium and high speed. These are cheaper and easy to compare.
I agree with a jet turbine like fan like the panaflo to compare the highest performance achievable.

[ArtosDracon]

Unfortunately, testing is not gonna show much of a difference between these two....:-/ D12SM-12 is ~55CFM and the P12 is only a little behind. Both probably have similar radiator characteristics too.

I recommend the three S-Flexes at 12V. 800RPM, 1200RPM, and 1600RPM. Also throw in some Panaflo H1A's for ~100CFM.

You don't need the best of the best, just something that's easy to relate to for users.

If anything this would make the most sense for actual testing. Everything else suggested is a pipe dream(my post included). If you wanted to throw some YLs at it they're pretty popular but, the three s-flexes listed should give a pretty good thermal variation depending on the radiator. I'm not a big fan of the H1As, don't even remember why though.

[Martinm210]

Thanks guys, lots of great recommendations, but I've been narrowing it down to these. Any thoughts between these?

Low Speed
Either
Yate Loon D12SL12s (http://www.ncix.com/products/index.php?sku=22644&vpn=D12SL%2D12&manufacture=Yate%20Loon)at 12V and 7V (47CFM + one setting lower)
or
Noctua NF P12-1300 at 12V and 7V (http://www.ncix.com/products/index.php?sku=27129&vpn=NF%2DP12%2D1300&manufacture=Noctua) (54CFM + one setting lower)

Medium Speed
Either
Yate Loon D12SH12s at 12V (http://www.ncix.com/products/index.php?sku=28870&vpn=D12SH%2D124UB&manufacture=Yate%20Loon) (88CFM type)
or
Scythe S-Flex SFF21F at 12V (http://www.ncix.com/products/index.php?sku=17749&vpn=SFF21F&manufacture=Scythe) (67CFM)

High Speed
Either
Scythe Ultra Kaze at 12V (http://www.ncix.com/products/index.php?sku=26744&vpn=DFS123812%2D3000&manufacture=Scythe) (133 CFM)
or
Panaflo H1s at 12V (http://www.ncix.com/products/index.php?sku=16960&vpn=FBA12G12H1BX&manufacture=Panasonic) (104 CFM)

Any thoughts?

I'm thinking the low speeds plus one undervolt for the silence guys, then the mediums and high speeds at 12V should cover the range pretty well?

[starlon]

I'd actually be interested in seeing a comparison between the silent fans.

[ArtosDracon]

Low Speed
Noctua NF P12-1300 at 12V and 7V (http://www.ncix.com/products/index.php?sku=27129&vpn=NF%2DP12%2D1300&manufacture=Noctua) (54CFM + one setting lower)

Medium Speed
Either
Scythe S-Flex SFF21F at 12V (http://www.ncix.com/products/index.php?sku=17749&vpn=SFF21F&manufacture=Scythe) (67CFM)

High Speed
Either
Scythe Ultra Kaze at 12V (http://www.ncix.com/products/index.php?sku=26744&vpn=DFS123812%2D3000&manufacture=Scythe) (133 CFM)

IMHO. Though some Delta FFB1212EHE or TFB1212GHE would be kinda cool to see.

[Martinm210]

Are you saying you won't test the Noctuas if you have them in hand?


No, I'm good with either one, I just need to narrow down the number tested.

I'm tempted to just calculate the air pressure drop of the radiator, then if we had PQ curves for each fan you could calculate the actual air flow. I would need to picked up an anemometer and measured deltas relative to actual air flow to do that though..

That would probably the the more scientific way, but that would also mean we would need PQ curves or some other way of estimating actual air flow through the radiator depending on the fan.

I was just thinking past radiator testing usually just included a range of fans and called it good. It probably makes the most sense then to test with the most popular fans.

I havn't made any decision yet, just feeling out what people would prefer.:up:

Whatever I end up doing, this is going to take some time to sort out a solid method and will take some experimenting.

[MorGo7h]

Make sure you test the Yate's because lots of ppl use them.

[Duh]

+1 noctua :)

[shabranigdo]

Make sure you test the Yate's because lots of ppl use them.

I agree the YL Low, Medium and High are some of the least expensive fans around I and would bet in use a lot out there in the real world.

[Tw1st3d]

+1 For noctua and yate loon - i would really like a compere..

[afX]

Martin - shoot me a PM. I may be able to arrange to sponsor you for some fans. (no guarantees yet though).

fans I'd recommend using:

Noctua NF-P12-1300 (best low noise high static pressure fans bar none)
Yate Loon med. speed fans.
some kind of horrible noisy thing.

OMG thank you Linus, Iv'e been waiting for some of those new Noctuas to be tossed onto a radiatior. PS, when are the new Lian-Li cases commin?

[chiguy]

I'm really interested in seeing how the Noctua holds up.

[rocket733]

I'm tempted to just calculate the air pressure drop of the radiator, then if we had PQ curves for each fan you could calculate the actual air flow. I would need to picked up an anemometer and measured deltas relative to actual air flow to do that though..

That would probably the the more scientific way, but that would also mean we would need PQ curves or some other way of estimating actual air flow through the radiator depending on the fan.


I agree this would be the most scientifically based method since then you're relating airflow through the radiator to performance instead of fan to performance. At the end of the day more people are going to understand fan to performance instead of having to read charts to get an idea about performance. Best of luck whichever way you go.

[Polizei]

Martin! Those cabinets!

/wife ;)


In order to minimize testing, I say just go with the most commonly used fans. Some Yate SL's, Yate SM's and then something more substantial, like some highspeed Delta AFB's. Not much testing, but would still allow people to estimate what their fans would do reasonably well.

On a side note - I know it's a lot of work you put in to the testing, but I would love to do something like you are. Tinkering and experimenting like that is fun. Well done my friend, well done. You're an asset to this communuty. :up: :toast:

[tw33ter]

Martin! Those cabinets!

/wife ;)

+1 lol

[Martinm210]

Thanks!

I have been working on the cabinets, I have all of them built that I need for stage one, just finishing up and should be done with what I need in the next couple of weekends..:D

The more I think about the whole fan issue, the more I think it would save more time to go the pressure drop route.

Then I could have a little fun calculating some fan P/Q curves as well. If I built a little estimator you could just pick the fans and the actual air flow would be calculated and the appropriate c/w value returned.

I have a feeling I might end up doing alot of these radiator tests, and in the end it would take less time to just mount one strong set of fans and measure performance relative to actual flow rate across several voltage points rather than pulling fans and mounting another set and always getting someone wanting yet another fan.

With just one fan, if you calculated the P/Q curve like you do on a pump, you would then have all the data you needed. Then just with the radiator air pressure drop, you could use any fan to get a result. I could still plot a couple of the more common types, but I wouldn't have to test the radiator with them all.

The only problem with this is the shroud variable. I'm guessing it's probably more common to NOT run a shroud than to run with one and the larger and stronger fans with larger motor hubs have an even bigger dead spot and that's going to vary depending on the fan a little.

So I guess I would just have to plan on testing with a shroud to keep the results consistent with air flow and not fan interaction. Probably should try both ways as a test case and see if I can estimate accurately what the actual air flow would be based on the fan used.

[bobo5195]

Dpn't care what the fans are as long as i know the flow rate is. If i don't know the flow rate (or an implied measure of flow rate) then radiator testing is massively less useful and you can't interpolate the results at all and anything you get out is much less useful.

You have watts disipated by the delta T across the Rad (assuming your thermocouples are accurate setup). So wattage at the wall is only useful as a sense check.

Note if you have watts and airflow then you have air out temperature (from first law, watts and air in) which gives you pretty much all the variables you need to do things properly (all temperatures and flow rates). With all the temps then you can do sexy engineering

Bill A advised using a lower temeprature limit 10 C is far to high and doesn't model people.

Suggest that you take lots of measurements near the inflection point and vary the temperature range. Been very busy at the moment so can't type up how radiators work but the inflection point is everything.

[Martinm210]

Thanks I'm still a bit unsure on fans vs. airflow. I like airflow, but it would require alot more work and more testing equipment. It appears most most fans peak around .25" H20 in pressure, so I don't think my smaller manometer will be sensetive enough with a maximum resolution of .1".

I'd need to sell the smaller manometer and pick up one of the smallest dwyers. They have one that's good for 4" with a resolution of .001", that's what I would need.

I also would need an anemometer. With the two of those I could start some fan testing to calculate the fan curves then the radiator could be tested with just one strong set of fans and voltage adjusted to get the necessary data points.

Bills testing for Thermochill included a nexus fan, a mid range delta, and panaflos, that's why I was looking along the same lines. Unfortunately that would also require more work with having to retest with each fan and each fan voltage change, where air flow and calculating the air pressure drop could be used to apply any fan that has a curve generated for it.

[Polizei]

So what's the problem? Seriously, I don't understand... It's late for me. :(

Why not just run fans at a given voltage instead of plotting airflow? Just measure the temperature as a function of fan voltage instead of temperature as a function of airflow.

[Martinm210]

So what's the problem? Seriously, I don't understand... It's late for me. :(

Why not just run fans at a given voltage instead of plotting airflow? Just measure the temperature as a function of fan voltage instead of temperature as a function of airflow.

You can do it either way, I may just start with the fan route since I don't have the equipment to do the airflow pressure drop route yet. I'll see if I can pick up a cheap anemometer so I can at least record actual flow rate from the fans used. The samples that I get to keep, I'll record pressure drop later when I get the tools to do that.

[Martinm210]

Ok, I've made my decision, I'll see what Linus can help me with:

Ultra Low, and Low Speed

Yate Loon D12SL12s (http://www.ncix.com/products/index.php?sku=22644&vpn=D12SL%2D12&manufacture=Yate%20Loon)at 12V and 7V (47CFM + one setting lower)

I use these myself, and feel they are probably one of the most popular even though the Noctuas are nice, they are higher CFM and I want to use these because they will be more different from the Medium speed range. I think one test at 12V, and one at 7V should do the trick.

Medium & High Speed

Scythe Ultra Kaze at 12V and 7V (http://www.ncix.com/products/index.php?sku=26744&vpn=DFS123812%2D3000&manufacture=Scythe) (133 CFM + one setting lower)

These have a little higher CFM than the Panaflos, so I figure they will give more spread in the results, testing at 12V. Testing again at 7V, should provide a good result for the medium speed test


This would provide a decent range, maybe something on the order of 25, 47, 67, 133 (open air CFM).

After I get the anemometer and tinker with a test, I may tweak my standard voltage rating to help spread out the results more evenly.

:up:

[disruptfam]

nice your getting the ultra kazes

[Polizei]

Sounds good. :)

[Jamestuk]

Noctua definately. Xilence?

[Martinm210]

I'll probably pick up a single noctua to throw on a single rad test. I did order an anemometer last night too, so I'm planning to record actual radiator cfm while I'm at it.

This should allow me to generate some cfm based curves that will be usefull in fitting other fan types.

[Shadowtester]

Yea the most accurate way would be cfm and air pressure it really does not matter what fan is creating the cfm and air pressure the the radiator does not care about fan model or brand all it cares about is the amount of air flowing and its pressure.

[bobo5195]

Yea the most accurate way would be cfm and air pressure it really does not matter what fan is creating the cfm and air pressure the the radiator does not care about fan model or brand all it cares about is the amount of air flowing and its pressure.

There maybe a slight case for difference as fans are mounted fairly close to rads so blade sweep and associate run off is a parameter but agreed overall air speed is what matters not the fan.

Volumetric flow rate is useful as it tells you air outlet temperature (which is needed to do anything proper), it tells you how restrictive a rad is which is a very important consideration.

If you do any rad analysis (which i need to write up when im not working 100hr weeks! darn product development) you will find that water / air flow will tell you exactly if you need more or less rad and is a (sorry, the) key factor in performance.

[bobo5195]

oh forgot my personal recommendation for fans, Pabst its what engineers use hence i trust the fan curves. The fan curves are on solidworks (COSMOS Floworks) as well so their believable.

[NaeKuh]

Martin from my own testing, im getting some pretty good results on ultra kaze's @ 1300-1500RPM's on my PA120.3. Noise isnt too bad, and i get great coolant temps. :]


Can you compare the PA and GTX @ that voltage on the kaze ultra when you get the chance. This might make me go back to the GTX series if it can spank the PA @ that setting.

[Martinm210]

Yeah, that's my plan. I currently only have a 480GTX, but I may have access to test a whole bunch more.

I just want to spend a little more time up front getting my methods down before I go after testing a whole bunch. I wasted most of last weekend retesting the DDC tops after I realized I needed to also capture the suction line pressures.

Going to take my time with this, but I plan to do some good and detailed testing on all the rads I currently have on hand. Everyone is CPU block testing, so I'm not nearly as interested in that as I am learning more about radiators, there's just so little information on them. I want to start including both thermal performance as well as flow rate performance in my reviews and develop a new radiator heat load estimator from the data I'm collecting.

I also kind of like using actual fans as well as cfm right on the raditator to take into account the dead air space and the interaction of the fins to the fan. So if I use a common Yate, that would capture performance and the fan effects of a smaller hub fan for lower air flows, and then the Ultra-Kaze to take into account the larger fan hub.

[Martinm210]

Here are the other testing tools I just ordered:

Anemometer for measuring CFM

http://i19.ebayimg.com/03/i/000/c0/4c/cbd3_1.JPG

My third manometer. Some people collect stamps, some collect water cooling gear, I collect motorcycles, water cooling gear, and testing tools..:ROTF:

http://i1.ebayimg.com/03/i/000/e5/63/d292_1.JPG

This one is good for -20 to 20" H2O with a resolution down to .01". That should have enough resolution to generate fan power curves and radiator air pressure drop curves.:up: Never done it before, but I'm going to try..

[phelan1777]

(didnt' read through the whole thread) So if these are mentioned, just ignore me.:D

San Ace 120X38mm Mdl# 109R1212H1011

I have been using them on my PA for almost a year now. Three together can be a bit loud, but move a lot of air, and I have to clean off my RAD screen at least once a week from dust.:rolleyes:

[Nuckin_Futs]

I'd like to see the HW Labs Black Ice GTX240 compared directly to our popular ThermoChill PA120.2 2x120mm High Performance Radiator. for each overal quality, silence with good performance and just god cooling at full fan with high cfm and good pupm. it would ba cool to see some of the best of single 120mm rads for compact mATX boxes. or just compact all-in-the-case water cooling systems for the quiet solutions, not just us overclockers. I have friends that wont overclock, but like the look or idea of water.