PA120.3 vs HE120.3 vs PA160

[Marci]

PA120.2 Data

Air<: 22.5 | 22.0 | 22.5 | 22.0 | 22.5 | 22.5 | 22.0
Air>: 32.0 | 31.5 | 32.0 | 32.0 | 32.0 | 32.0 | 32.0
Liq<: 33.0 | 33.0 | 33.0 | 33.0 | 33.5 | 33.5 | 33.0
Liq>: 32.5 | 32.0 | 32.5 | 32.5 | 32.5 | 32.5 | 32.5
CPU: 18.0 | 17.0 | 18.0 | 18.0 | 18.0 | 18.0 | 17.0
GPU: 18.0 | 18.0 | 18.0 | 18.0 | 18.0 | 18.0 | 18.0

Avg Air <: 22.3
Avg Air >: 31.9
Avg Air Dt: 09.6
Avg Liq <: 33.1
Avg Liq >: 32.4
Avg Liq Dt: 00.7

Slotted into the summation:

PA160 with 1x Panaflo M1A @ 7v: CPU - 21 GPU - 22
HE120.3 with 3x ^^ ^^ ^^ ^^ : CPU - 18 GPU - 19
PA120.2 with 2x ^^ ^^ ^^ ^^ : CPU - 17.7 GPU - 18
PA120.3 with 3x ^^ ^^ ^^ ^^ : CPU - 15 GPU - 15

So in this scenario in terms of final CPU & GPU temps, the new PA120.2 beats (just) the old HE120.3

[Cathar]

Originally Posted by Marci

So in this scenario in terms of final CPU & GPU temps, the new PA120.2 beats (just) the old HE120.3

Gotta be happy with that.

[Long Haired Git]

Originally Posted by Marci

PA120.2 with 3x ^^ ^^ ^^ ^^ : CPU - 17.7 GPU - 18

So the PA120.2 had three fans, or a typo?

[ProHandyman]

Quote:
"So the PA120.2 had three fans, or a typo?"


No, Marci lists both the PA120.2 and the PA120.3 with the PA120.2 just beating the HE120.3 in performance. Cathar then mentions you gotta like the results of the PA120.2 with 2 m1a fans... infering the PA120.3 would be far superior to the HE120.3 with the m1as...

[Marci]

So the PA120.2 had three fans, or a typo?

Typo corrected

Gotta be happy with that.

Damn right! Less space consumed and less noise produced!

[Marci]

Just to shut some folks up on other sites, had to do this test...

2x PA160 with 1x Panaflo M1A@7v on each

AirIn <: 23.5 | 24.0 | 23.5 | 24.0 | 23.5 | 23.5 | 23.5 | 23.5
AirRad1>: 29.0 | 28.5 | 29.0 | 29.0 | 29.0 | 29.0 | 29.0 | 29.0
AirRad2>: 29.0 | 29.0 | 29.0 | 29.0 | 29.0 | 29.0 | 29.0 | 29.0
Liq <: 31.0 | 30.5 | 30.5 | 31.0 | 31.0 | 31.0 | 31.0 | 31.0
LiqMid : 30.5 | 30.0 | 30.0 | 30.0 | 30.0 | 30.5 | 30.0 | 30.0
Liq >: 30.5 | 30.0 | 30.0 | 30.0 | 30.0 | 30.0 | 30.0 | 30.0
CPU : 16.0 | 16.0 | 15.0 | 16.0 | 16.0 | 16.0 | 16.0 | 16.0
GPU : 15.0 | 15.0 | 15.0 | 15.0 | 15.0 | 16.0 | 15.0 | 15.0

Avg Air Into Rads <: 23.6
Avg Air Out of Rads>: 29.0
Avg Air Dt : 05.4
Avg Liq Into Rads <: 30.8
Avg Liq Out of Rads : 30.0
Avg Liq Dt : 00.8

2x PA160's produce final CPU / GPU Temps more or less identical to PA120.3

PA160 with 1x Panaflo M1A @ 7v: CPU - 21 GPU - 22
HE120.3 with 3x ^^ ^^ ^^ ^^ : CPU - 18 GPU - 19
PA120.2 with 2x ^^ ^^ ^^ ^^ : CPU - 17.7 GPU - 18
2x PA160 with 1x ^^ on each ^ : CPU - 15.8 GPU - 15.1
PA120.3 with 3x ^^ ^^ ^^ ^^ : CPU - 15 GPU - 15

[ricecrispi]

PA120.3 pwns them all!!

Man, WCing has gotten so sweet.
I wish i didn't waste money on stuff that is now obsolete......

[Long Haired Git]

Ahh, but only due to three fans.
C/W vs noise would thus be interesting.
Eg, PA120.3 with only 2 fans?
Dual PA160.1's with 3 fans?

I can use every second of Marci's spare time....

Would suggest the dual PA160's with three fans would rule roost: simple case of surface area.

Eeek, too much time reading CW thread, now I am typing like BillA.

[WarlordBB]

I just sent this email to Thermochill but thought it might be appropriate to post it here. Please excuse if the answer is common knowledge to all who post here, I am but a newb in awe on these forums.

Here's the body of the email:

I want to use my PA120’s (I now have both the 120.2 and 120.3, will be buying the 120.1 soon too) as they were meant to be used. My understanding of the situation is:

I know that the PA120 line benefit more from lower CFM fans than do the HE series.

Of course that’s not to say that higher performing fans won’t outperform lower CFM fans on the PA120, just that the efficiency of the lower CFM will be better.

What I don’t know is where this line is. You say lower than 100CFM but I’m sure there is a scale up to that figure such that diminishing returns start to happen much earlier.

Can you please elaborate on this?

My goal is to have the most efficient fans on my rads as possible. I don’t want to get into the situation where I’m adding CFM (and noise) for 10% (or less) gains.

Thanks for any help you can provide.

[BillA]

you are chasing your tail
efficiency ?
see other thread

[ricecrispi]

Why don't you read the FAN recommendations on thermochill website or maybe the one Marci is using here.

[WarlordBB]

Originally Posted by BillA

you are chasing your tail
efficiency ?
see other thread

@BillA

I respect you greatly but I don't understand any line of your comment. It's probably a case of "he doesn't know enough to know he doesn't know" and I certainly understand that you can't hand hold every newb that shows up here but I wanted to point out the fact that your post helped only to the point that I know BillA thinks I'm chasing my tail (in what way, I'll just have to guess).

@ricecrispi

The only recommendations I could find (I didn't read through their forums) was in their FAQ. I have taken those recommendations to heart but my question above wasn't answered.

----

I certainly hope that none of you take any offense by this post because I certainly mean none.

I wasn't so much asking "what's the best fan for the PA120 series" as I was asking does anyone know at what point below 100CFM you start getting diminishing returns?

I thought it was a valid question. Maybe I'll understand why it's not as soon as I figure out which of the thousands of posts is the "other thread".

EDIT: @ricecrispi - I have the fans that Marci is using as well as the fans that cathar mentions in this thread (among many others) but, like I said, I wasn't looking for specific fan recommendations.

[Cathar]

Define what you want when you say "efficiency".

The PA radiators will achieve higher levels of overall heat dissipation performance than the HE radiators when using the exact same low-moderate throughput fans.

That basically means that the PA rads are more efficient in such scenarios.

Now when asking what is the most efficient fan to use, then really that is a question which cannot be answered. You need to specify a set goal, whether that be noise or cooling performance, or both, and then decide upon a suitable solution for those stated goals.

I think that this is perhaps what you're after when you say "efficient", but until you define what your goals are, the question can't be answered. You need to define what your needs are, and then construct a solution that suits. If you want high cooling performance for lower noise, than that is what the PA rads are very, very good at. As for an exact end-solution though, need to know what you want to achieve.

[BillA]

spot on
it is the fan plus rad under defined conditions
fan efficiency is twaddle

[WarlordBB]

Originally Posted by BillA

spot on
it is the fan plus rad under defined conditions
fan efficiency is twaddle

I understand this. However, I thought I was only talking about the PA120.x (more specifically the PA120.3 for my purposes) so I thought we might be able to have a discussion specific to that.

@ cathar

I understand this too, and I thought I was bringing that into the equation by saying "I don’t want to get into the situation where I’m adding CFM (and noise) for 10% (or less) gains." I actually meant to say 1% gains which accentuates my point.

I guess I'm saying noise/cooling ratio but I thought this could be discussed without talking about specific fans. IOW, I thought there would be a "in general" sweet spot with the PA line.

Maybe I can put it best by illustration. I thought someone would post something along the lines of:

"Thermochill is right, you aren't going to get increased performance past 100CFM but I've found that you gain (on average) 20-35% from 0CFM up to 60CFM and only an increase of 5% more out of that entire other 40CFM. Therefore, I don't try to push any more than 60CFM through my PA."

I would understand that such a comment would be highly dependant upon the fan in question but it would give me a decent rule of thumb. Am I still chasing my tail here?

EDIT: My basis for even thinking there were some "in general" thoughts on this matter is Thermochill's site itself who obviously have some way of knowing that after 100CFM there is not substantial enough gains to warrant it.

[Cathar]

In my testing I continue to see steady improvements with the PA120.x radiators from weak fans right up to ~120CFM fans. Haven't tried stronger fans. I do have a 220CFM 12cm TFB Delta here, but it's noisier than a construction site when turned on.

It's not like the PA120.x rads suddenly run out of puff with stronger fans, nor do they really even taper off that much. They taper off slightly more steeply than the HE rads, but because they start out so superior to the HE rads with weak fans, the cross-over point is way up in the 110-120cfm fannage range.

The PA rads like fans with stronger pressure characteristics, meaning that one can't simply just look at a fan's CFM rating and predict that it'll necessarily be better than a fan with lesser CFM. One such example is the ThermalTake ThunderBlade fans, which at 85CFM you would think that they'd do pretty well, but are out-performed by the 69CFM Panaflo L1A's, and even equalled by the 55CFM Papst 4312L.

It really comes down to finding the fans with the PQ curves that work best with the rad when trying to balance performance vs noise.

In that respect, my recommendations from fans I've personally tried are:

Near Silent: Nexus, Noiseblocker SX1
Quiet: Yate-Loon/Tricod Science
Moderate: Papst 4312L
Noisy: Panaflo H1A

If you wanted a fan that does it all, and don't mind the bearing drone at lower speeds, the H1A is your best bet. At 12v it provides ~40% higher heat dissipation than the Yate Loon @ 12v. At 7v it equals the Papst 4312L @ 12v, but is quieter than the Papst by quiet a deal. At 5v it provides performance on par with a Noiseblocker SX1, but had a bearing tick that roughly puts it on par with the Tricod @ 12v for noise.

An H1A on a rheobus will give you the best of all worlds so long as you're not a silence nut. I am a bit of a near-silence nut myself though, so I don't use the Panaflo fans, and instead use the Yate-Loon/Tricod's, and then rheobus them down in speed with a lot of radiator capacity.

So I guess, if there's one fan you're after for that's best in terms of power, noise, performance, when used with the PA120.x, it would be the Yate-Loon's. When given sufficient radiator capacity (eg. 120.3, or more) then there's almost no PC based heat load short of cooling TEC's that it won't satisfactorily deal with (satisfactory meaning no more than a 5C rise in radiator outlet water temps above ambient) in relative quiet with the fans at 12v.

Oh, and thanks for the Acoustifan Dustproof to try out. Noise-wise there is no comparison with the Yate-Loon/Tricod. The Acoustifan is much noiser. I'm going to attempt to gather some performance results on the PA120.1 with the Acoustifan Dustproof, but my initial reaction to its noise-level, both blade and bearing, is such that I'm incredibly surprised that some would even dream of recommending it as a quiet fan. Only at 7V can I get it to be anything like as quiet as the Yate-Loon's at 12v.

[Long Haired Git]

Check out Graph 15 here:
http://thermal-management-testing.com/ThermoChill.htm

These are the older HE series. I'd assume the graph for a PA series would start with a lower C/W for the low end of CFM, but not reach the lows of the HE series with mega-CFM.

With the HE series, you can see the graph is really steep until about 1.25 M^3/min, which acording to Graph 14 is around the power of either dual Papst 4312L's @ 12v or dual Delta 1212VHE's at 7v.

So WarlordBB's question is reasonable: where is the bend expected to be for a PA series radiator?

Note that in Graph 15, the steepness of the curve is still reasonable even at 4.5 m^3/min. This is approx equiv to Dual delta 1212VHE's at 12v. Shame there aren't more points to the right as thermodynamic laws require the graph to level out and it would be nice to see if there is another kink further on.

Shame similar graphs are not in the new radiator review at Swiftech.

[BillA]

ahem LHG
merely a different plotting (visualization) of #14,
understand the data's limitations wrt airflow quantification

2 types of considerations:
one can discuss matching a fan to the core (I did so),
which begs the issue of how the core will be loaded in an actual system (Cather did so)

both a 120.1 and a 120.3 will cool a 400W TEC app, but the airflow needed from each fan is ~1/3ed with the 120.3
- no reason to put big fans on a triple w/o kW loads (only a slight exaggeration)

[Nugit]

Stacking low pressure fans should help too, any idea as to how stacked nexus's will perform in comparison with higher powered, noisier fans?
Empiric evidence at sprc indicates that noisewise two very lownoise fans are almost indistinguishable from one at the same voltage.

[learning curve]

Originally Posted by Marci

2x PA160's produce final CPU / GPU Temps more or less identical to PA120.3

PA160 with 1x Panaflo M1A @ 7v: CPU - 21 GPU - 22
HE120.3 with 3x ^^ ^^ ^^ ^^ : CPU - 18 GPU - 19
PA120.2 with 2x ^^ ^^ ^^ ^^ : CPU - 17.7 GPU - 18
2x PA160 with 1x ^^ on each ^ : CPU - 15.8 GPU - 15.1
PA120.3 with 3x ^^ ^^ ^^ ^^ : CPU - 15 GPU - 15

So in this scenario in terms of final CPU & GPU temps, the new PA120.2 beats (just) the old HE120.3

I was wondering . Is the data on the 2x PA160's run in parralel or series? Thanks in advance

[Marci]

Series

[learning curve]

Hello Marci I have a another question. I am going to add a second PA160 to my loop and would like yours and others opinions which would be the better way to plumb the second rad in series or parralel? It will be running through a G5, a silverprop Cyclone Nexus SX, and a silverprop fusion hl and the pump is a iwaki RLZT20. Thanks to all in advance.

[Long Haired Git]

Series is easier. Parallel flow halves flow rate through each rad, but flow vs C/W is basically flat. Parallel reduces PQ from them both but compared the blocks the rad resistance is basically irrelevant.
Did I mention series is easier?

Oh, we're talking about coolant flow series, not air series.

[learning curve]

Fisrt of all thanks for the response LHG. Yes I was meaning coolant flow. I am thinking that parallel might maximize the deltaT in both rads and thus lower coolant temps better than series, or Would the halved flow through parallel rads counteract the benifit of the higher deltaT? Thanks again for any and all info.

[learning curve]

Any other thoughts/opinions?