TFC MONSTA testing complete

[Navanod]

Wow....I took a few month's break from the hobby (and this forum) and come back to find Martin back in action!
Great job

Hmmm...seems there had been alot of improvements over the months....only wished my wallet had improved enough to keep up with all these new toys

[NaeKuh]

martin didnt test the san ace guys.

But if he had and put that sucker on a shroud....

Game Overs. San Ace wins @ all voltage...

also guys keep note, martin used probably only a push or pull config on the kaze ultra, 38mm fan... not 25mm fan... so if he did a push pull yate system, id probably do a LOT better on that thick rad.

Martin care to do a quick run with push / pull yates?

[Martinm210]

Martin why did the 122 score worse then the 121 @ higher RPMs?

I'm curious myself, the more fan testing I do, the more I think RPM sensors are simply not that accurate.

You generally see specification with +- 10%, so some of that may just be the RPM sensor itself.

I didn't capture noise in these tests, so it's simply a measurement of cooling per RPM.

The Triebwerk fans have a very gentle fan blade. This is very good for noise, they do seem to have a very good sound quality. I need to get some video going to show you that though. I had actually planned to test the fans first, then do the rad thermals but everyone wanted me to do the rad first..

The interesting thing about the fans is how they perform at very very slow speeds. The other fans seem to fall off in performance at very slow RPM more. Anyhow, I didn't mean for this to be a fan test, I just tested all the fans that I had in triplicate so you could get an idea how the radiator would react.

[Martinm210]

martin didnt test the san ace guys.

But if he had and put that sucker on a shroud....

Game Overs. San Ace wins @ all voltage...

also guys keep note, martin used probably only a push or pull config on the kaze ultra, 38mm fan... not 25mm fan... so if he did a push pull yate system, id probably do a LOT better on that thick rad.

Martin care to do a quick run with push / pull yates?

Yeah, I always used the UK3's and medium yates on all my previous testing, so that's why I used them so I'd have something to compare too. I'm also fan poor when it comes to more than one.

The only fans I have three of are:
Yate Loon D12SM12
Yate Loon D12SL12
Scythe UK3
Triebwerk TK-121s and 122s

IMHO, when it comes to radiator testing, it's not really as important what fans you use, as long as they are all the same when making comparisons. I chose the yates and UK3's a LONG time ago for my radiator testing because they were cheap and popular.

They are not my favorite fans though by any means. The UK3s are extremely powerful at 3000RPM, but so is any 38mm fan spinning that high. They are also obnoxiously loud though.

The key to fans and radiators is understanding the performance difference between very low RPM and High RPM. In this test, you can see that fans at 3000RPM have over 7X the performance as fans at 400RPM. The interesting thing about fan efficiency is that it goes up nearly exponentially with RPM. 3000RPM is not 3 time the performance of 1000 RPM, it's more like 10X, fans are just that way...much much more efficient at extremely high speeds.

I find it interesting to see that you would have to run 7 of these with 21 fans at 400 RPM to match just one with 3 fans at 3000RPM. That's not out of the ordinary though, most radiators are that way, but that's also why I prefer looking at heat dissipated instead of C/W. You may get the false impression that performance is good down low with c/w ratios. But it becomes fairly clear quickly how EXTREMELY dependant radiators are on fan power. This is not our typical 10-20% differences, this is many many X the difference. Fans are EVERYTHING when it comes to radiator performance.

This is the point I can't seem to get across to the sub 1000RPM folks. You take it in the shorts big time when you run RPMs down under 1000RPM. You may as well lean over and blow on it now and then...lol!

Anyhow sorry if I turned this into a fan comparison, that wasn't my intent. You can't compare fans using CFM per RPM without even considering noise or noise quality.

I'll work on the fan testing next..

[nzone]

This review just put a on my pre-plan NB/SB:CPU:GTX295:3-34CFM-750RPM:Monsta

[Martinm210]

Thanks, Martin! I'm always looking forward to your test results!

Anyway:
The more numbers we get on the monsta/treibwerk, the less impressive they look, imo. Sure the monsta beats the PA, but a 420 vs 360 rad isn't exactly a fair fight. Even with the built in shroud the $35 (?) triebwerk fans could only perform about as well $5 YL. I really like the idea of 140.x rads, but I think the monsta handles this poorly as far as price/performance/size goes. I also like the idea of fans optimized for radiators, but the numbers speak for themselves as far as triebwerk goes.

Thanks!

You could say the same thing about size difference between any of the slim and double thickness fans also, that it's not fair comparing a slim radiator to a double thickness etc. You could even say it's not fair comparing the 120mm rads that are 120mm wide vs. the ones that are 140mm wide, or the ones with 15mm fan spacing vs larger. Size is a factor, but I think noise and fan power being the same is always a worthy comparison.

My intent was to compare the same exact fan power, number of fans, and corresponding noise.

Price will always be something everyone has to consider for their own budgets. I was fortunate enough not to have to do that, but considering the time I've spent testing (probably 50+ hours on this one alone), I could have easily bought two or three of these had I chosen to work overtime instead of testing. As always, I do the testing because I enjoy learning how stuff works, and frankly can't stand installing something that I don't understand...a sickness of mine if you will...

I've grown bored with the everyday ordinary radiators, there's generally not a whole lot to distinguish some. Sure they all work fine, but so does a $20 bonnie heatercore. Do I really need two 480's and a Monsta to cool my rig. No...but I like it....it's different, and they are all fun for me.

If someone asks me what triple 120mm fan radiator has the best performance/price ratio for 1350RPM fans (Of the rads I've tested), I'd have to say the Swiftech MCR, HWlabs BIP, or XSPC RS series. If someone asks me which triple 120mm fan radiator has the ultimate performance in the 1350RPM range regardless of price, that would be the TFC Monsta. But there's also many good options in between and ones that are optimized for different areas and alot of different options that will fit size and budget needs.

[NaeKuh]

errr martin u dont have 2 more san aces?

lemme fix that for ya...

[Martinm210]

awesome possum . . . now get ur rear to PDXLAN . . .

I should go, it would be fun and it's only 5 hours away..

[antiacid]

Anyhow sorry if I turned this into a fan comparison, that wasn't my intent. You can't compare fans using CFM per RPM without even considering noise or noise quality.

I'll work on the fan testing next..

Now that we're talking fans and clearly blaming you for bringing up another can of worms, let's jump into it!

On a side note, before I start elaborating: Regarding your thick fans, how noisy were they?

Now, on to the real thing...

You know, I have been thinking that there needs to be a way to compare all of those different components in a way that makes sense to the adept. At this time, either there's too much information on the graphs or then the information isn't so clear to pick-up from that graph (which doesn't justify the actual graph). This is a general observation.

My line of thought was to try and come up with some kind of dimensionless number to discuss those radiators in a decent way. I mean, there's already a truck-ton of different dimensionless numbers for general heat transfer theory so there's no reason why we don't use them / recycle them / make up our own.

Concerning the testing of fans, specifically, there seems to be about three or four values of interest: CFM, static pressure, noise and RPM. One could also argue that the values for "free flow test" for those fans don't work on rads because of the added restriction. I'd argue that the "static pressure" doesn't change if you put a wall in front of it. It'll push the wall at exactly the same value as in free flow. I know that the noise doesn't change, the RPM shouldn't and the CFM change in relation to the "impedance" of the radiator tested.

In the classical restrictive pipe example, the pressure diminish in the smallest part of the pipe because the pipe takes momentum away from the fluid and the speed of the fluid increases in that section. In a radiator setup, that tube is very short: From the fan intake up to the radiator exit. In a typical scenario, that's about 5cm or about there. Does Bernouilli's equation works here or is it too short? The extra air would just NOT be pushed through the radiator if it was too restrictive, reducing CFM?

This brings us to an interesting value: This so called "impedance" or "restrictiveness" should be measured at different preset points (I propose low/med/high CFM, so about 35/55/75CFM) to give an idea of how air-restrictive is that radiator. In a way, this is already done by showing graphs of performance vs fan rpm vs fan type but it isn't clear how those results would vary for a fan of slightly different characteristics that didn't get tested. I'd like to get some kind of empirical rule developed for a given radiator and then expand it... One can dream

The reason I am bringing this up because I find it totally retarded that predictions cannot accurately be made if you were to test a radiator/fan combo with fan A and I were to use fan B, even while knowing the characteristics of both fans. The difference is not small, as shown in your graph 10c delta vs rpm: at 2000rpm, there's a 100watts+ difference (which accounts for different values of CFM/mmH20 @ given RPM @ particular fan).

BTW, you really should start labeling those graphs for easier commenting on them

The goal of developing a general "impedance to air" figure would be that one dude could shop for a fan, check it's cfm/dba/rpm/static pressure and then shop for a radiator with a given "ITA" (not only am I developing a new unit/concept but I'm also shortening it!) and then plug this in a simple equation (or online calculator, what do I care) and figure out an approx. value of efficiency that he could then relate to the price paid for then fan!

I'm thinking that all fans aren't made equal. xyz CFM @ abc mmH20 is really the more "correct" way of thinking about fans and it shouldn't be reduced to an RPM figure. It is about as useful as the label when comes time to make a decision on what to buy. That's like buying a pack of gum by the color it is and hoping that you'll get the taste you want.

I am suggesting that this ITA value for radiators would work as a coefficient to moderate the values given by the vendors (tested in free flow setups). That way, you'd get straight up an idea of how much static pressure and CFM is actually going towards cooling your computer! I have no idea how a curve of ITA vs RPM would work for a given fan though, so I can't really guess at the form of the equation. I think that once a stable relationship is established, it could be possible to develop an analytical solution that would solve efficiency of forced flow convective heat transfer against flow parameters.

Maybe I'm re-doing what's been already done though. My heat transfer knowledge isn't as good as it used to be (and that already started not that good) so maybe I'm just really coming up with another way to talk about Péclet number or something. I'll do some reading tomorrow morning and see if there's some theory that could be applied in this scenario and that would satisfy my need to quantify everything :P

My end goal would be to develop a program in which you input your fans, rads, cpu, gpu, avg length of tubing, etc, and then automatically get the values around the loop Sure, testing needs to be done for every single parts as they come out but once it has been tested and validated, that's all there is to it! I wonder if the watercooling scene is ready for such a unifying vision :P

Anyway, that's my wall of text for tonight and I'm not 100% certain of everything I said in there so I'm really hoping that you guys pick it apart and we can figure out if this "impedance to air" concept is really worth developing

edit:
I found a reviewer trying to ghetto-measure the static pressure values of noctua nf-p12 fans:
http://www.overclock3d.net/reviews.p...re_performance

He sorts of "get" it but doesn't really push or question his reasoning. I get the feeling that he just wanted to challenge noctua's claim that a re-design could bring more static pressure to their already excellent fans.

[skinnee]

Excellent work as always Martin!

[Martinm210]

Now that we're talking fans and clearly blaming you for bringing up another can of worms, let's jump into it!

On a side note, before I start elaborating: Regarding your thick fans, how noisy were they?

Now, on to the real thing...

You know, I have been thinking that there needs to be a way to compare all of those different components in a way that makes sense to the adept. At this time, either there's too much information on the graphs or then the information isn't so clear to pick-up from that graph (which doesn't justify the actual graph). This is a general observation.

My line of thought was to try and come up with some kind of dimensionless number to discuss those radiators in a decent way. I mean, there's already a truck-ton of different dimensionless numbers for general heat transfer theory so there's no reason why we don't use them / recycle them / make up our own.

Concerning the testing of fans, specifically, there seems to be about three or four values of interest: CFM, static pressure, noise and RPM. One could also argue that the values for "free flow test" for those fans don't work on rads because of the added restriction. I'd argue that the "static pressure" doesn't change if you put a wall in front of it. It'll push the wall at exactly the same value as in free flow. I know that the noise doesn't change, the RPM shouldn't and the CFM change in relation to the "impedance" of the radiator tested.

In the classical restrictive pipe example, the pressure diminish in the smallest part of the pipe because the pipe takes momentum away from the fluid and the speed of the fluid increases in that section. In a radiator setup, that tube is very short: From the fan intake up to the radiator exit. In a typical scenario, that's about 5cm or about there. Does Bernouilli's equation works here or is it too short? The extra air would just NOT be pushed through the radiator if it was too restrictive, reducing CFM?

This brings us to an interesting value: This so called "impedance" or "restrictiveness" should be measured at different preset points (I propose low/med/high CFM, so about 35/55/75CFM) to give an idea of how air-restrictive is that radiator. In a way, this is already done by showing graphs of performance vs fan rpm vs fan type but it isn't clear how those results would vary for a fan of slightly different characteristics that didn't get tested. I'd like to get some kind of empirical rule developed for a given radiator and then expand it... One can dream

The reason I am bringing this up because I find it totally retarded that predictions cannot accurately be made if you were to test a radiator/fan combo with fan A and I were to use fan B, even while knowing the characteristics of both fans. The difference is not small, as shown in your graph 10c delta vs rpm: at 2000rpm, there's a 100watts+ difference (which accounts for different values of CFM/mmH20 @ given RPM @ particular fan).

BTW, you really should start labeling those graphs for easier commenting on them

The goal of developing a general "impedance to air" figure would be that one dude could shop for a fan, check it's cfm/dba/rpm/static pressure and then shop for a radiator with a given "ITA" (not only am I developing a new unit/concept but I'm also shortening it!) and then plug this in a simple equation (or online calculator, what do I care) and figure out an approx. value of efficiency that he could then relate to the price paid for then fan!

I'm thinking that all fans aren't made equal. xyz CFM @ abc mmH20 is really the more "correct" way of thinking about fans and it shouldn't be reduced to an RPM figure. It is about as useful as the label when comes time to make a decision on what to buy. That's like buying a pack of gum by the color it is and hoping that you'll get the taste you want.

I am suggesting that this ITA value for radiators would work as a coefficient to moderate the values given by the vendors (tested in free flow setups). That way, you'd get straight up an idea of how much static pressure and CFM is actually going towards cooling your computer! I have no idea how a curve of ITA vs RPM would work for a given fan though, so I can't really guess at the form of the equation. I think that once a stable relationship is established, it could be possible to develop an analytical solution that would solve efficiency of forced flow convective heat transfer against flow parameters.

Maybe I'm re-doing what's been already done though. My heat transfer knowledge isn't as good as it used to be (and that already started not that good) so maybe I'm just really coming up with another way to talk about Péclet number or something. I'll do some reading tomorrow morning and see if there's some theory that could be applied in this scenario and that would satisfy my need to quantify everything :P

My end goal would be to develop a program in which you input your fans, rads, cpu, gpu, avg length of tubing, etc, and then automatically get the values around the loop Sure, testing needs to be done for every single parts as they come out but once it has been tested and validated, that's all there is to it! I wonder if the watercooling scene is ready for such a unifying vision :P

Anyway, that's my wall of text for tonight and I'm not 100% certain of everything I said in there so I'm really hoping that you guys pick it apart and we can figure out if this "impedance to air" concept is really worth developing

Yeah, that makes sense, the problem I see is measuring CFM accurately enough.

It seems as though you could come up with something similar to "Water Horsepower" that you do for water in an equivalent air form...perhaps that's already some sort of variable. Seem as though you could come up with some sort of "Air horsepower" variable that equates to fan power that's a combination of CFM and pressure.

The problem I see is the inherent errors in measuring both CFM and air pressure. I really don't believe anemometers are accurate enough to get absolute CFM numbers (it's ok for relative measurements, but it sucks for absolute measurements and really only a ballpark), I think to do CFM right...you'd have to measure with an array of pitot tubes and the extremely small pressures they could create over a duct. The same thing would have to occur on the inlet/outlet side of the radiator for pressure drop. Do do all this, you would simply have to build a flow chamber, and I mean BIG flow chamber. Something that would fill an entire room and permanently place you in the dog house for good!..

But even if you DID do all of that, you'd then have to come up with a way to equate each fan's performance to that testing. Sure it would be a very elegant way to do direct comparisons and removes all the fan type and RPM inconsistency stuff, but it also removes the ability to convey the information to something folks can understand and it also removes the deadspot and scrubbing/tubulence effects that are created by actual fans.

I've thought this through in detail when I very first started radiator testing. I fell back to something I had confidence in measuring which is fan RPM. Definately not perfect by any means though, just something that's easy to do, easy to log over the course of the test, and something that every single computer user out there has the ability to measure to some degree. As always, in any sort of comparison testing...there's absolute accuracy which is very difficult to do and needed to compare cross platform, and there's relative accuracy that's pretty easy to do and good for comparison purposes on that same test bed only.

Being hobby/recreational testers...relative is about all we can muster..

And I think everyone has had some sort of dream to package every little watercooling part into a box of science we can use to calculate and estimate from. Flow rate and radiator performance is probably doable, but I completely raise the white flag when it comes to understanding processors. How do you account for all those freaking variables like mounting pressure, TIM compound, TIM spread methods, lapping, processor sample variation, efficiency relative to heat, heat relative to Vcore and speed,....the list goes on and on. You'd have to spend years testing waterblocks with all the different variables controlled and by the time you were done...there would be a new processors out you hadn't even started on yet.

It's all good though, keeps things fresh..

[Martinm210]

Excellent work as always Martin!

Thanks!

I don't know how you're able to continue testing all those rads. It's getting way to hot outside to be cranking 600 watts into my laboratory (*cough* spare bedroom) for that many hours.

[skinnee]

I actually set aside the rads for now, CPU block testing takes all my attention. I have a quad radiator round-up to do, room stays quite cool with being in the basement and a dedicated HVAC zone. Only about a 2º natural swing...wife said no to a window shaker, so I put in a damper and actuator.

[Martinm210]

I actually set aside the rads for now, CPU block testing takes all my attention. I have a quad radiator round-up to do, room stays quite cool with being in the basement and a dedicated HVAC zone. Only about a 2º natural swing...wife said no to a window shaker, so I put in a damper and actuator.

Nice!

That's on the short list of home upgrades. The old 1978 force air furnace has to kick the bucket sooner or later. That's when I'll seize the chance at a combined a/c and heater unit.

I think it's time for some dirt bikes and dunes testing for me...need to run a sieve and moisture content analysis and determine what's optimal for those quick sand conditions that nearly pitch me over the handlebars..

[faster3200]

Wow, the Monsta is way more impressive than I thought. I could cool my entire loop with one of those with the fans I use .

[Monkeypillow]

How do you set your fans to run at specific voltages?

[HESmelaugh]

Great work as always, Martin!

Looking forward to what you'll see with 140 mm fans.

[WaterFlex]

Great job like always, Martin! Thanks

[G-UNIT91]

Great job like always, Martin! Thanks

I agree

but I preferred this type:

[Martinm210]

I agree

but I preferred this type:

You can calculate that with the watts dissipated easy enough.

Just take the watts for 10C delta and do this

Delta for X Watts = (10/ (10C Delta))*X watts

For example
For a 570 watt load
UK3 @3000 Monsta = (10/1080)*570 = 5.28C
UK3 @3000 PA120.3 = (10/820)*570 = 6.95C
UK3 @3000 RS360 = (10/580)*570 = 9.82C

Or at 1000 RPM
D12SM12 @1000 Monsta = (10/325)*570 = 17.53C
D12SM12 @1000 PA120.3 = (10/315)*570 = 18.09C
D12SM12 @1000 RS360 = (10/260)*570 = 21.92C

Etc. If 10C is average, 5C High performance...this gives you an idea. With that sort of heat load, they all do average to high performance with the high speeds fans and none of them are doing well enough at low speed fans.

Now break out those calculators, it won't hurt you I promise..I'll force some math and science on you if it's the last thing I do....

[Kayso]

@Martinm210 - good job man, awesome !!!

I just want to ask, will be MONSTA enough for a single loop of Q9650@4.32 Ghz, GTX280 and NBMAX and 5-7 Degrees Delta ?? I'm beginning a new project with a ATCS840 and I'm very very interested to have a monsta inside my case :P

[MomijiTMO]

Wow the most unexpected place is selling this in Aus. Just $419 for the white edition . . .. .

Hmm should I buy it?

[affiliate13]

Hi Martin,
Great info as always, thanks.

Did TFC mention 140mm Triebwerk models to you at all?
I think there was talk of them appearing shortly.

I would be interested to see how they might perform on this radiator.

[shazza]

Excellent review - thanks for the photos too. They really do show the size of this Monsta in comparison to your other rads. (You'll need steel supports when you load your Torture Rack with this guy).

[MomijiTMO]

It could probably stand by itself Shazza