Question: At what point is a better rad needed?

[Psyche911]

So, I've been wondering something.

If you have a heatercore/radiator and decent fans, able to dissipate much more heat than you're adding to the loop, what difference does an even better rad/fan combo make?

Let's say you're just cooling a Venice CPU, which really don't put out a lot of heat. Even heavily overclocked (for water) they probably only approach maybe 100 watts?

Now, let's say you have a very competant radiator. Let's say the new Swiftech MCR120QP with medium CFM fans. It is totally able to dissipate that CPUs heat just fine (could dissipate nearly 150watts with 1GPM flow rate and 55CFM from the fan).

But what if you go with an even better option. Let's say a Weapon+Shroud combo with some Sanyo Denki fans at 12 volts.

Both radiators could take the heat and more, but would there be a temp difference in the water exiting the radiators?

If so, how much?


I'm really liking the look of the new MCR220QP. Right now I'd just have my CPU in the loop (don't care much for GPU overclocking). But I'd like to have the option of a hotter CPU and adding a GPU later. I'm sure this radiator would be plenty, but I want to know if I could see an improvement with something even better without having that much heat in the loop.

[One_Hertz]

So, I've been wondering something.

If you have a heatercore/radiator and decent fans, able to dissipate much more heat than you're adding to the loop, what difference does an even better rad/fan combo make?

Let's say you're just cooling a Venice CPU, which really don't put out a lot of heat. Even heavily overclocked (for water) they probably only approach maybe 100 watts?

Now, let's say you have a very competant radiator. Let's say the new Swiftech MCR120QP with medium CFM fans. It is totally able to dissipate that CPUs heat just fine (could dissipate nearly 150watts with 1GPM flow rate and 55CFM from the fan).

But what if you go with an even better option. Let's say a Weapon+Shroud combo with some Sanyo Denki fans at 12 volts.

Both radiators could take the heat and more, but would there be a temp difference in the water exiting the radiators?

If so, how much?


I'm really liking the look of the new MCR220QP. Right now I'd just have my CPU in the loop (don't care much for GPU overclocking). But I'd like to have the option of a hotter CPU and adding a GPU later. I'm sure this radiator would be plenty, but I want to know if I could see an improvement with something even better without having that much heat in the loop.

A heavily oced 3800+ venice will be 150 watts or so. Use the formula thingie for calculating watts if u need to find out an approximate number

Stock wattage*((oced vcore/stock vcore)2 + Oced freq/Stock Freq)

Any radiator can dissipate hundreds of watts the question is at what temp will it keep the water while doing so? Your rad can take 500 watts load with water temp at like 50C, now do you really want that? The higher the delta T of water temp to ambient the more watts the rad can dissipate. Having a more powerful rad can dissipate a lot of wattage without too high of a delta T.

to tell whether you will beneft from a rad upgrade all you have to do is measure the temp of the water. A weapon monstercore would give u water temps of about 1-2C over ambient with just your venice and if you know your water temp u can right away tell how much of a difference it will make since lowering water temp by 1C also lowers processor temp by the same amount. I'd recommend a monster core instead of the swiftech rad... itll give u room to add more things to the loop and help your temps by quite a bit.

[Cathar]

Hmmm, where are these tests that suggest that heater-cores are better than special purpose designed water-cooling radiators?

I keep hearing this over and over. After much testing with radiators and development of such at my end, I can't say that I agree with the statement that heater-cores are better than purposefully designed water-cooling radiators.

[Psyche911]

If I knew enough about this to answer that, I probably wouldn't have this question. I don't mean any offense to you, or anyone from Swiftech, or Thermochill, etc. I have a ton of respect for you, Cathar. As do most people who know the first thing about watercooling.

That aside, I honestly haven't seen a lot of detailed reviews comparing different radiators on the market today. I've read thousands of forum posts regarding watercooling, and that's about the extent of my knowledge on the subject. I realize that doesn't mean I know much of anything, which is why I ask questions such as this.

You can't deny, many, many, many recommendations on this forum and others are for heatercores. From what people say, one would be lead to believe they perform at least as good as, if not better than a comperable Black Ice series (which seemed to be the standard for comparison for a long time), and did so at a much better price until recently. Maybe the performance is overrated. I sure don't know.

But I'm sure you'd agree, with the two radiators I used in my example, the Weapon would outperform the (single) MCR120 from Swiftech. It would do so at a much higher level of noise, but that's another topic. If I had used the MCR220 for the example, that might have brought up topic for debate. But that's not what I want to know right now.

My question just comes down to this:

If two radiators are both able to dissipate the heat in the loop, but one is capable of much more than the other, how much difference will the user experience with actual temps?

[WaltS_2]

Hmmm, where are these tests that suggest that heater-cores are better than special purpose designed water-cooling radiators?

I keep hearing this over and over. After much testing with radiators and development of such at my end, I can't say that I agree with the statement that heater-cores are better than purposefully designed water-cooling radiators.

I have been wondering the same thing for a while now.
I just thought I was stupid and overlooking the anwser.:slap:

I'm glad someone with a lot of knowledge in watercooling brought this up.

[Psyche911]

One_Hertz:
Thanks for the response. I don't have my setup yet, so I can't compare temps. ;)

I also realize the potential of the monstercore, but that's not feasable in my situation for a number of reasons (cost, noise, size). And I'll plug some numbers into that forumla at some point. Thanks.

[Bloody_Sorcerer]

the only reason i like heatercores is that 22 dollars for a 2x120mm fan radiator is not bad at all :)

[WaltS_2]

the only reason i like heatercores is that 22 dollars for a 2x120mm fan radiator is not bad at all :)

So the reason people like the heatercores so much is their price? Not necessarily their performance.

[masturdebat3rr]

its comparable performance for a lot less

[Quazi]

Now I am just thinking about things, and thinking out loud at that. So lets say in any room where the ambient temps are 20C. You have a radiator built for PC cooling, let's say a BIP 2-120 and a heatcore the same size and using 2-120's. You are using identical fans and shroud setups on each radiator. And you are cooling two identical rigs, one with the BIP 2-120 and the other with the heatercore with the 2-120's. If the cool ambient air moves more freely through the heatercore, the metal fins should be cooler than the fins of the BIP because there are more FIP, which in turn doesn't allow the same amount of cool air to flow through the radiator. I know that when more air freely blows on a thin piece of metal, that piece of metal should be cooler than the ambient air temps. This is why I would think that a heatercore, given that it is large enough to handle the heat produced by what it is cooling, would cool better. Cooler metal, cooler water.

Then you have the single pass of the heatercore and the double pass of the PC radiator. I am thinking that since the single pass heatercore allows the water to flow with more force, which I have read in this forum produces cooler temps on whatever you are cooling, and this should cancel out any benefit or extra cooling that the double pass of the PC radiator garners. Because even though the water is passed through the PC radiator twice, the fact remains that the air flow is still restricted because of the more FPI of the PC radiator, therefore the water is passing over metal that isn't any cooler than it was the first time around.

Now, this is my reasoning, be it flawed or not. Plus given I am new to water cooling and don't know the intracasies or the science of, I may be way off in my thinking. Feel free to disagree or even flame the noob.

[One_Hertz]

Hmmm, where are these tests that suggest that heater-cores are better than special purpose designed water-cooling radiators?

I keep hearing this over and over. After much testing with radiators and development of such at my end, I can't say that I agree with the statement that heater-cores are better than purposefully designed water-cooling radiators.

Well maxx is like a walking, talking, AND posting watercooling dictionary so if he says it, it must be true. :fact:

[Weapon]

Hmmm, where are these tests that suggest that heater-cores are better than special purpose designed water-cooling radiators?

I keep hearing this over and over. After much testing with radiators and development of such at my end, I can't say that I agree with the statement that heater-cores are better than purposefully designed water-cooling radiators.

just for clarification, which radiator are you testing against which heatercore and which fans are you using for the testing?

[HiJon89]

Well maxx is like a walking, talking, AND posting watercooling dictionary so if he says it, it must be true. :fact:
Yeah and Cathar doesn't know anything about water cooling :stick:

[Holst]

Fundamentaly it all comes down to water temp and pressure drop.

Most decent sized rads and heatercores are more than capable of keeping your water at around ambient, or just a few degrees above.

All specially designed watercooling rads SHOULD be low pressure drop .. the lower the better.

Allot of heater cores are also low in pressure drop (maybe max can give us some numbers soon, as he is testing rads at the moment) but some may not be.
Some heatercores WILL be better than some commercial rads, some will not.

So in answer to the original question.
You wont benefit from a new rad unless
1- your water is more than 3*c above ambient.
2- You will gain performance from higher flowrates.

[Cathar]

just for clarification, which radiator are you testing against which heatercore and which fans are you using for the testing?

I'm talking about comparing an arbitrary heater-core of the same facial area size to an equivalent properly designed water-cooling radiator using fans of 30dBA/70cfm rating or less, at full speed.

Given push-pull fannage upwards of 40dBA/100cfm fannage on most heater-cores and they start to deliver in comparison, but that's an awful lot of noise. Now for push-pull, can replace that statement with either pull- or push- only with stronger fans.

Heater-cores are optimised for use with squirrel cage radial blower sorts of fan pressures, which typically push from 4-10x the fan pressures that axial fans that us computer users are familiar with, unless we start doing push-pull configs of fairly high powered powered (and noisy) fans.

This is not to say that heater-cores are bad, not at all, but the flow/pressure optimisation point is significantly different to that which occurs in computer use. Of course, given a big enough heater-core facial surface area and the pressure drop concerns are lessened but now one is using a significantly larger heater-core just to compensate for the different pressure optimisation point that heater-cores have.

Raw unprepped heater-cores have their price going for them though, but in terms of space and performance-per-noise level they are far from being as optimised for computing use as a radiator that is specifically tuned for the characteristics that occur in computing use. In general can achieve better performance with less noise and less space with a properly tuned radiator over a generic car heater-core.

[Cathar]

Fundamentaly it all comes down to water temp and pressure drop.

Most decent sized rads and heatercores are more than capable of keeping your water at around ambient, or just a few degrees above.

All specially designed watercooling rads SHOULD be low pressure drop .. the lower the better.

Allot of heater cores are also low in pressure drop (maybe max can give us some numbers soon, as he is testing rads at the moment) but some may not be.
Some heatercores WILL be better than some commercial rads, some will not.


Agree with all your points, but air pressure drop is more important than water pressure-drop for the most part here. Most radiators and heater-cores have really quite low water-flow resistance with the fittings tending to offer by far the bulk of any flow resistance.

Revising your statement below:


So in answer to the original question.
You wont significantly benefit from a new rad unless
1- your water is much more than 2*c above ambient.

[Weapon]

I'm talking about comparing an arbitrary heater-core of the same facial area size to an equivalent properly designed water-cooling radiator using fans of 30dBA/70cfm rating or less, at full speed.

Given push-pull fannage upwards of 40dBA/100cfm fannage on most heater-cores and they start to deliver in comparison, but that's an awful lot of noise. Now for push-pull, can replace that statement with either pull- or push- only with stronger fans.

Heater-cores are optimised for use with squirrel cage radial blower sorts of fan pressures, which typically push from 4-10x the fan pressures that axial fans that us computer users are familiar with, unless we start doing push-pull configs of fairly high powered powered (and noisy) fans.

This is not to say that heater-cores are bad, not at all, but the flow/pressure optimisation point is significantly different to that which occurs in computer use. Of course, given a big enough heater-core facial surface area and the pressure drop concerns are lessened but now one is using a significantly larger heater-core just to compensate for the different pressure optimisation point that heater-cores have.

Raw unprepped heater-cores have their price going for them though, but in terms of space and performance-per-noise level they are far from being as optimised for computing use as a radiator that is specifically tuned for the characteristics that occur in computing use. In general can achieve better performance with less noise and less space with a properly tuned radiator over a generic car heater-core.

hmmm. the reason I was asking is because I do not believe that an arbitrary heatercore would provide an accurate representation of the true potential of heatercores even if it were based upon facial area of the two heat exchangers. From the differing models of cores that I have tested (which is getting to be a downright silly number of them) facial area on a heatercore has a definite impact on the core's performance but there are many other variables that can add up to an extreme difference in performance between two similarly sized heatercore. Here is a short and by no means comprehensive list of the differences between the cores I have on the bench at the moment that have produced noticable differences in performance:

1. Fins per inch - until I really started digging into this one, I did not realize there was such a wide range with regard to fin density in the core market -- I have some that are as sparse as 10 FPI which have very minimal air resistance. Even a quiet fan (if silence is the key) will move a lot of air through 10FPI. On the other hand I have one that is somewhere around the 20FPI range.

2. Fin design - some have the internal fins on the fins, others just have straight fins like most of the computer-specific rads. From my initial run at this one, there is a trade off between air resistance caused by the additional internal fins and the amount of extra surface area they provide but I haven't tried to hone it down to the point in air flow where one is better than the other.

3. waterchannel/flow design - straight flat tube or corrugated sheet. I'm still playing around with all the different ones here as the flow characteristics between them are so different. Also of interest is the number of flow channels or tubes going each direction -- single pass cores obviously have all the channels going one way but double pass cores mix it up. Some double pass cores have more channels on the exit side than the inlet side where others make it an even split.

4. tank design - tall and narrow vs. short and wide. seems like minimal impact but the tall & narrow tanks leave some room for interesting additions inside the tanks... ;)

5. shroud design - I doubt it is even necessary to do anything other than mention this one as it is widely accepted that it can be a "make or break" with regards to heat exchanger performance.

6. construction method - old fashioned soldered seams or one of the new high-speed brazed models? copper fabrication methods for cores are taking massive leaps forward and it is allowing for more effective copper/brass designs. It is also prompting several manufacturers to move away from aluminum and back to brass or copper for cores in the auto industry. thinner heatercores from copper and brass are on the horizon as 1.25 and 1" models that are currently made from aluminum will likely be available in copper/brass soon.


Given push-pull fannage upwards of 40dBA/100cfm fannage on most heater-cores and they start to deliver in comparison, but that's an awful lot of noise. Now for push-pull, can replace that statement with either pull- or push- only with stronger fans.

I cannot agree with the "upwards of 40dBA/100cfm" statement for cores -- modified san aces can run well under 40dBA at 7-10v (stock 109R1212H101s are rated a 38dBA at 12v) while still moving enough air through a 302 to take out most, if not all of the dual 120 commercial computer rads I have tested them against. With San Aces running at 102.4cfm, a 302 has a definite advantage over a dual 120 commercial rad (and at least a few of the triple 120 commercial rads...) with the same fans.

If you change the rules of the game to where the primary concern is absolute silence, then commercial rads might come out on top. If you make it entirely about maximum overall cooling performance, cores win.

[jinu117]

Well maxx is like a walking, talking, AND posting watercooling dictionary so if he says it, it must be true. :fact:

Lord.... Cathar you liking this? :)
If we are talking about reputation... Maxx do have long way to go. Cathar is well acknowledged from just about any H/W Forum I drop by. He is always step ahead of competition (I don't think he think of competition but all those H2O companies are trying to catch up to him it seems). The fact best blocks in it's times are designed by him with rigorous testing and thinking and applying (white water, storm)... And manufacturing outlets such as D-tek, Swiftech bought his design alone is proof enough.
We all have different opinion of what would be optimized watercooling system... What some of us thinking are decently noisey might be very noisey to just about everyone else... thus the optimization of airflow vs temperature discussion.
I would think general acceptance of noise level would be around 4 120mm fans running 1200-1400RPM (25mm). This is about 40cfm or so fans that doesn't have too much static pressure. From my experiment with BIP3 (commercial not optimized in my view) with 3 Panaflo fans running about 1000RPM (I am guessing about 40cfm as this fan is 38mm fan) on BIP3 did perform better than Weapons's heater core with 2 Panaflo fans running about 1300 RPM (guessing about 55cfm each). The noise level are very similar at this point between 3 fans and 2 fans and airflow is in favor of BIP3 by little. I lost 1-3c water temperature using BIP3.... There was thread about this I commented on while back when I had H2O frenzy :") - BIP3 had no shroud... weaponcore did.
What I found was that unless you have about 80cfm + airflow, the heatercore doesn't really start gaining the lead (around 80cfm + airflow... it was exceptional but noise -_- -> about 160cfm+ since dual fan)
I think BIP3 do have lot of room for improvement obviously in it's design but what I find constantly is that for low air pressure, it seems to be best solution on market other than PA160... (which I would love to see review done here).
Weapon did awesome job designing beautiful shroud, finding correct part and modifiying it, but bottom line is... the cores itself are not designed for computer water cooling. I believe PA160 might be one design where it was derived from computer cooling in mind and would love to see review of this with about 50cfm fan mounted (it is single fan solution) vs BIP, BIP2, and single heatercore... :)

PS) 40db sounds a bit of high but keep in mind, most of fans above 30db... for number of fans... you start adding 2db for effective noise level. Of course, perception of it is not necessarily same as we all know fans do make different noise based on restriction, etc.

[Cathar]

Weapon, there are a whole mass of variables as you point out.

What I'm getting at with a water-cooling radiator is that a conscious attempt has been made to carefully control all of those variables such that they are favorable for computer water-cooling use, as opposed to merely selecting a heater-core that happens to approximate something that would be nice for computer water-cooling.

All a computer water-cooling radiator is, is a heater-core that has all those variables carefully controlled.

This is what I mean when I don't buy into the whole "heater-core's are better than a properly designed water-cooling radiator" statements that get casually thrown about.

I have done masses of radiator research and development in the last 6 months, which builds on top of work than I had done somewhat sparingly over the last 4 years. Bill Adams at Swiftech has quite independently been conducting the same sort of research and controlling the same sorts of variable at his end too.

So after all the work that both Bill and I put into developing the best possible radiators tuned for water-cooling use in controlling all those variable, to then come along and have people blithely say "a heater-core is better", well that just smacks of an uninformed opinion.

[One_Hertz]

So after all the work that both Bill and I put into developing the best possible radiators tuned for water-cooling use in controlling all those variable, to then come along and have people blithely say "a heater-core is better", well that just smacks of an uninformed opinion.

With a few delta fans the heatercroe WOULD outperform the black ice rads of a similar size and since this is xtreme systems I would consider performance to be more important then noise.

[Weapon]

Weapon, there are a whole mass of variables as you point out.

What I'm getting at with a water-cooling radiator is that a conscious attempt has been made to carefully control all of those variables such that they are favorable for computer water-cooling use, as opposed to merely selecting a heater-core that happens to approximate something that would be nice for computer water-cooling.

All a computer water-cooling radiator is, is a heater-core that has all those variables carefully controlled.

This is what I mean when I don't buy into the whole "heater-core's are better than a properly designed water-cooling radiator" statements that get casually thrown about.

I have done masses of radiator research and development in the last 6 months, which builds on top of work than I had done somewhat sparingly over the last 4 years. Bill Adams at Swiftech has quite independently been conducting the same sort of research and controlling the same sorts of variable at his end too.

So after all the work that both Bill and I put into developing the best possible radiators tuned for water-cooling use in controlling all those variable, to then come along and have people blithely say "a heater-core is better", well that just smacks of an uninformed opinion.

I can't say I blame you for that one -- nothing like a mass generalization to produce a lot of misconception and I agree - some heatercores are utterly worthless for computer cooling. Also, digging through all the different core models can be a daunting task (especially if new to H2O cooling) and the dirt cheap ones always require some kind of mod before they will hook into the loop nicely. On the other hand, if you already have the right model number for the core and some modding skills, it is fairly simple.

While I don't plan on dropping shrouds or cores anytime soon I have a prototype weapon radiator that is entirely my design (i.e. started with a blank white page -- ok, it was really a blank white screen on a cad prog...). I am sure the prototype will require substantial modification (due to material-related construction costs with the current design) but I have the broad strokes of the design where I want them. Some fine tuning such as tweaking fin thickness, tank design and ultra-fine tuning FPI remains but that is all about trying to squeeze that last 1C out of it. Controlling the all of variables can make for interesting performance. ;)

With a few delta fans the heatercroe WOULD outperform the black ice rads of a similar size and since this is xtreme systems I would consider performance to be more important then noise.
true provided you have the right core -- of couse, most in here know which model numbers to buy so that is not much of an issue. If noise is no concern and you want absolute max cooling, the right core will kick some serious arse in the heat exchanger arena.

if you want down right silly levels of cooling, a four-120-per-side heatercore with 80+ cfm fans will nuke anything that is readily available in the current computer market. In the not-so-available segment of the cooling market, there is currently a Weapon OctaCore on my bench. lol -- yeah, OctaCore = eight 120s per side...

[jinu117]

Weapon OctaCore....
I say....
Just dig the darned ground and shove pipe 6ft deep into groupd :P
Seriously, is there any PA160 testing with comparible noise testing done somewhere? I noticed Maxx hasn't been reviewing things for a while... (or is he doing something big?)

[Cathar]

With a few delta fans the heatercroe WOULD outperform the black ice rads of a similar size and since this is xtreme systems I would consider performance to be more important then noise.

...and I would counter with saying that a really well designed tri-fan radiator could achieve ~0.02C/W (2C water temp rise per 100W of heat load) with a total of 25dBA of fannage on board.

This may be XtremeSystems, but it doesn't have to be DeafAndXtremeSystems.

I take Xtreme to meaning extremely optimised. If you're happy to sit next to a pair of screaming Delta's at 45dBA total, and I can achieve the same cooling performance from 25dBA of fannage, then which solution is more Xtreme, and which solution would you rather sit next to?

[Mr. Tinker]

I've read this whole thread, and it's good. Very good.

[chunkylover77]

With a few delta fans the heatercroe WOULD outperform the black ice rads of a similar size and since this is xtreme systems I would consider performance to be more important then noise.
One_hertz your seriously not arguing with Cathar about anything related to water cooling are you? That just smacks of ignorance.

[Weapon]

Weapon OctaCore....
I say....
Just dig the darned ground and shove pipe 6ft deep into groupd :P
Seriously, is there any PA160 testing with comparible noise testing done somewhere? I noticed Maxx hasn't been reviewing things for a while... (or is he doing something big?)

lol - it was a one-off made for a very specific system with one helluva lot of heat. I have no plans for making them in mass (or making another one for that matter)...it was a nice challenge for the torch though.

one other thing I forgot to mention regarding cores - it is possible to mod cores so that the majority of the fins are much thinner hence the air resistance is much less. Some cores have 2 less FPI or so than the PA 160. If the majority of the fins were thinned, even with thicker waterchannels, the air resistance should be about on par with the PA160. It is a time consuming job if you want it to look good but it might be worthwhile for a low airflow DIY heat exchanger with a $20 price tag...


I think BIP3 do have lot of room for improvement obviously in it's design?

true. I have modded a bix2 and bix3 (the mods would work on a bip2 or 3) & I have been tempted to send to them for comparison testing. lol - it's not like I ever intended to make a profit off of watercooling (building H2O gear doesn't pay nearly as much per hour as being an attorney so I lose money from the get-go)...however, making things work better is tons of fun. :D

[Cathar]

One_hertz your seriously not arguing with Cathar about anything related to water cooling are you? That just smacks of ignorance.


No, no. People should question me. I've been wrong before. Please don't blindly take anything I say as fact, but I will challenge you to provide evidence to the contrary if you disagree, that's all.

[chunkylover77]

Cathar I'm not saying you cant be questioned. Its just that when a person has done as many tests as you have done and designed the best waterblocks on the planet it should at least be known that you probably know what you are talking about. If his argument is bigger is always better nananana it is just kind of silly. He also is talking to you as if he does not know who you are. I mean I know Maxx knows alot about watercooling but come on. Making a statement that he will take Maxx's word over yours is somewhat ignorant. If a person did just a little bit of research into watercooling they would see your name almost immediately. That is all I meant by that. And reading it over I think my comment might be a little harsh. So I do apologize.

[MaxxxRacer]

Stew.. have you ever considered moving to los angeles? Watercooling advances would move along a tad more quickly if you did i think ;)

In respone to the few out there that seem to think I'm some heatercore peddler.. I never said heatercores were the end all. Just that if you dont mind the noise of more powerful fans that heatercores are a very good solution with some kickass performance.

If you want to hear a pin drop in your room.. well grab a pa160 and a nexus fan at 5v and be happy.

As a generalization here.

Personally I am tired of noise. dam tired of it. i run denkis right now but they never go above 7 volts (usually around 5) except for testing. so whatever advances can be made to use silent fans.. I'm all for it.
personally i think the ultimate radiator will be one that will use 2 nexus fans at 12v (with shroud of course) and cool as well as a 2-302 heatercore with 120CFM fans blasting away at 12v..

Now I know this isnt possible in the current state of physics, but its a nice dream.

I suppose a slightly more fair challenge is this. Design a radiator that works optimumly with two fans like the nexus (for silence). I say two fans as the 3 fan radiators are fun and all.. but where the hell am i gonna put the thing. So with that said, keep it to two fans. Now for the acutal performance. It would be nice if it would perfom the same as a BIX2/120.2/2-302 with sanaces at 7v. In other words do the same thing as I am doing right now with a BIX2 and san aces, but with thinner, quieter fans. Oh yes.. and one more thing. 2-302 flow restriction.. or lack there of.

[Cathar]

I suppose a slightly more fair challenge is this. Design a radiator that works optimumly with two fans like the nexus (for silence). I say two fans as the 3 fan radiators are fun and all.. but where the hell am i gonna put the thing. So with that said, keep it to two fans. Now for the acutal performance. It would be nice if it would perfom the same as a BIX2/120.2/2-302 with sanaces at 7v. In other words do the same thing as I am doing right now with a BIX2 and san aces, but with thinner, quieter fans. Oh yes.. and one more thing. 2-302 flow restriction.. or lack there of.

Believe me - I'm working on it - and it (almost) exists. ;)

I already have a working prototype that is extremely close to doing just what you describe.

[One_Hertz]

If you're happy to sit next to a pair of screaming Delta's at 45dBA total, and I can achieve the same cooling performance from 25dBA of fannage, then which solution is more Xtreme, and which solution would you rather sit next to?

The cooling performance of the heatercore with the deltas would be better then the triple rads. And if its beter even by one degree that would make the heatercore outperform the triple fan rad. Ofcourse it is not practical in any way or form but if you want the absolute maximum performance, which some people indeed do want, you'll go with the heatercore. Btw this thread started with one person asking if one 120mm rad was better then a much bigger in size heatercore and like I said its not, cause the heatercore would have many many more fins.

[Quazi]

Call me nuts for spending that much on this, but this is what I will be using with 4 Panaflos, each able to move 115CFM, and a 1" shroud:

http://www.voyeurmods.com/index.php?action=item&id=990&prevaction=category&previd=featured&prevstart=

I will mount it on the left side of my rig using a setup similar to the RadBox design in the H2O APEX Extreme, except the dead space will be doubled between the rad and the side of the case.

[BRiT]

...and I would counter with saying that a really well designed tri-fan radiator could achieve ~0.02C/W (2C water temp rise per 100W of heat load) with a total of 25dBA of fannage on board.

So when will we be able to purchase such a radiator?

[jinu117]

I guess we better give up on convincing one person on all of logistics that we all seems to agree on... -_-;
That aside, I love how the topic is evolving here.
I do vaguely remember you mentioning somewhere Stew for your radiator endeavor. Is it in conjunction with Swiftech or Marci? Is this what's holding you off from making impringement block for GPU cooling? (or is dual impringement block setup with storm and GPU is too much for pumps out there ... )
That sat aside, have you got chance to test PA160 at all Maxx? :)

[Psyche911]

I think he's still working on some for Thermochill, like he did with the PA160.

[MaxxxRacer]

jinu, i need to do some comparitive tests with the pa160, but any real, high quality scientific tests will have to wait till i get some money in my pocket and can buy test gear.


Cathar, what kind of fins does it use? Bill mentioned testing different fins a few days ago and I am insatiably curious as to what design is the most effective.

[Cathar]

Maxxx, assessment of fin type variations are limited by budgetary constraints. We're not dealing with car-company sized budgets here. I'm working with the standard louvered folded fin style designs, ala what exists in modern heater-cores and radiators in cars. I'd love to be able to vary fin thickness, but once you go non-standard things get expensive.

Still, within the limitations of dealing with commonly available parts there are very good optimisations to be achieved within the form factors suitable for PC water-cooling, and when dealing with quiet/weak axial fans.

My radiator research focuses heavily on optimising around 30-60cfm axial fans (25-38mm thickness), but with one careful eye on seeing that performance relative to other radiators/heater-cores is not lost at up to 100cfm fannage levels. Above 100cfm I deliberately do not optimise for because I strongly believe that such is a step backwards for the attractiveness of water-cooling for the wider market. Still, even though I don't optimise for >100cfm the PA160.1 still holds its own against other similarly sized cores at up to 120cfm fannage levels.

The other physical limitation we're fighting here is the limitation of the heat capacity of air itself.

Generally speaking, an actual 100cfm of 20-30C air-flow has an inherent thermal resistance of almost bang on 0.0170 C/W. So an actual 50cfm of air-flow (actual meaning flow actually occurring, rather than what a fan is rated for at peak flow) has an at-best C/W of 0.0340.

So the trick with weak fans is to restrict them as little as possible, and the other trick is to pass as much heat from the water into the air as possible, but then we have to do that in a 120x120mm facial area. The PA160.1 does so well 'cos it cheats by not limiting itself to a 120x120mm facial area.

Have made very promising progress but still a lot of work to do. Don't hold your breath.

[MaxxxRacer]

ahh ic..

i know what you mean regarding the budgetary constraits.. bill doesnt have as much in the way of budget issues, so hopefully he will/has come up with some good results that can be spread throughout the market regarding fin style and thickness.

i have no imperical testing on this, but i would assume that thinner fins would most certainly dump their heat better. as the surface area is closer to the center of the metal. the downside is just like with cpu coolers, the heat capacity with the fans off will be much less, but that isnt much of an issue as you should always have the fans on.

Thinner fins would also present less airflow resistance.. granted the difference will be small but its something to think about.

[jinu117]

What do you think about thermal syphon (basically like heatpipe solutuions)? What I noticed on high end air cooling now a day is that it is doing almost 90% of watercooling's duty considering same fan configuration... (probably has to do with better designed fin arrangement).... Thermal syphon do take out the problem with current heatsink size limitation some what while doing what water cooling does basically sans pump using vapor... (now, that means bit less heat contributed to loop as well) And phase changing nature means it might carry the heat in more effective fashion... http://www.overclockers.com/articles1248/
Don't think the thing is optimized much in sense of condensor much either but... oh well :P (notice no shroud, etc... thats just... funny :P)

[MaxxxRacer]

you mean for the radiator or for the waterblock itself??

for the radiator its entirely pointless. all the heatpipes are for is to MOVE the heat away from the cpu, to a different location. gee. sounds just like a watercooling loop.

[EnJoY]

Exciting stuff, I'll happily toss my 120.3 and M1A's out the window once I see a radiator coming from you Stew.

Btw, got upgraded to a Panworld 30PI-Z-D 24v pump for no extra charge over the 20 today, I thought that was nice. ;)

[jinu117]

you mean for the radiator or for the waterblock itself??

for the radiator its entirely pointless. all the heatpipes are for is to MOVE the heat away from the cpu, to a different location. gee. sounds just like a watercooling loop.

Combination of course.
The reason why I am interested in it is because heat load absorption is greater than say straight watercooling. Basically liquid gets in touches hot copper bottom of block, evaporates (taking heat with it) than goes up to condensor (radiator I guess) where it gets cooled in bigger surface than becomes liquid again. Same as heatpipe solutions really but just having much bigger heat radiation component if you think about it. To be honest, I don't think waterloop with single BIP with the same fan running in thermaltake typhoon will get better temp than Thermal take typhoon from my observation. Now, consider this effect with bigger cooling surface area. It should peform very close to most watercooling... :)

[MaxxxRacer]

jinu you'd be surprised. very surprised.

its now merely about surface area. its how you use that surface area, and how you transfer heat to it.

radiators are MUCH more effecient at moving the heat to the fins than the thermalright XP series or the typhoon. because of this they get much better temps.

an example is.. I was using a BIX to cool my 3000+ venice at 2.6ghz and x700pro. the temps were fine, but if i had an air solution with the same total surface area, the temps wouldnt be nearly as good.

[kemist]

Regarding airflow blockage, am i correct in assuming that given the current design of common popular car heater cores (2" thick style) that the BIP style cores (1" thick)should perform just as well if not better (due to less air blockage) than the car cores with low flow quiet fans (e.g. panaflow nexus etc). I know that it has somewhat less frontal area but also should be much less restrictive due to being much thinner and should get much better air flow correct?

Also would the BIP style cores perform even better with some of the fins cut out to further improve flow? If this is possible would 2 fans or 3 be suggested?

I was going to build a new system soon and was hoping i could have it be quiet and still perform near the max possible for water cooling. I was considering using a BIP triple 120 core for this.

If this is wrong please correct me, thanks.

[MaxxxRacer]

yes the bip style rads perofrm better with low speed fans (not bix though) .

cutting out some of the fins is NOT suggested as your chances of dammaging the rad are very high.

if you want max performance with quiet fans, wait a litle while.. thermochill is coming out with 3 new radiators designed perfectly for this.

[Psyche911]

Hasn't it been a couple months since they were "soon to be releasing new radiators?"

Is there any news or is this like DooM 3's release date? :p

[MaxxxRacer]

stew said in a thread a few days ago that it would be about a week.

[phextwin]

http://www.thermochill.com/news3.php

out ~26th.

[MaxxxRacer]

lol racin... funny comment, but good idea on not keeping it there.