Watercooing from China

[jaydee]

Originally Posted by killernoodle

Brass is generally used in heatercores for the tubes. If you look at the half tube on the edge of every heatercore, it is easy to see that the thickness is less than .5mm.

Not every heater core has half tubes on the side. Also the one heater core I have with half tubes is well over .5mm. More like 1 to 1.5mm. But that's just the cores I have. Many different designs.

[Kobuchi]

Originally Posted by killernoodle

brass which is stronger than aluminum.

So we'll always have the copper-zinc team vs. aluminum.

[killernoodle]

hmm, maybe it is just my cheap one?

[jaydee]

Originally Posted by killernoodle

hmm, maybe it is just my cheap one?

Here are some pics of the cores and an oil cooler I got laying around. Sorry for the blurry pics. My Digital Cam seems to be screwed up or something.

#1



#2



#3


#4



#5


hard to make it out but that is a turbulator in that last pic inside the oil cooler.

[joeyek]

Originally Posted by jaydee116

Ummm, I would think the tube material is very important. If it cannot transfer heat well then it isn't going to move the heat to the fins well either making the fins useless even if they were silver. By your logic we could use plastic for the tubes and copper for the fins with no loss in performance compared to a copper tubed rad. :shrug:

On the subject of material let's look at this page http://www.users.globalnet.co.uk/~dm.../materials.htm and tell me which has better thermal properties of the materials used in HC's?

From my interpretation it is
1) Copper
2) Aluminum
3) Brass

Jaydee, don't make ur decision by logic, lets do it in mathmatics. Referring to this equation in this famous article: http://www.amdmb.com/article-display...D=105&PageID=3

lets fill the same equation with the parameter of a typical watercooling sys:

Delta T= Q *(L/K), L means the thickness of the tube, and K means the conductive coefficient of the material of the tube, copper is 401 and aluminum is 237 and brass is only 109;
given the Q is up to 150Watt and thickness is .5mm;

then Delta T will be .000187 for copper and .000316 for aluminum, and .000688 for brass,
And then we must divide our initial answer by the tube area, which is up to hundreds times comparing to the die area of the Athlon 0.000117 m^2(=117 mm^2), given the total tube area for a single HC(15*15*5cm) is at least 13(tubes' No.)*3.5cm(atual tubes' width not 5cm)*15cm(tubes' length), result of the tube area is .0682m^2, up to over 600 times of the die area of the Athlon, so check the Delta T now, .00274c for copper and .00463c for aluminum and .001c for brass.

all of above is only math calculating, we still need to count some other factors, such as the temp diff of the liquid inlet and outlet of the rad, but as we know from BillA's rad test and others' article like Joe Citarella of overclockers, the temp diff between in and outlet is minors. so basically we can safely drive a conclusion , brass tube will not be beaten by alot.
Forgive me if I miscalculate above.

[jaydee]

Originally Posted by joeyek

Jaydee, don't make ur decision by logic, lets do it in mathmatics. Referring to this equation in this famous article: http://www.amdmb.com/article-display...D=105&PageID=3

lets fill the same equation with the parameter of a typical watercooling sys:

Delta T= Q *(L/K), L means the thickness of the tube, and K means the conductive coefficient of the material of the tube, copper is 401 and aluminum is 237 and brass is only 109;
given the Q is up to 150Watt and thickness is .5mm;

then Delta T will be .000187 for copper and .000316 for aluminum, and .000688 for brass,
And then we must divide our initial answer by the tube area, which is up to hundreds times comparing to the die area of the Athlon 0.000117 m^2(=117 mm^2), given the total tube area for a single HC(15*15*5cm) is at least 13(tubes' No.)*3.5cm(atual tubes' width not 5cm)*15cm(tubes' length), result of the tube area is .0682m^2, up to over 600 times of the die area of the Athlon, so check the Delta T now, .00274c for copper and .00463c for aluminum and .001c for brass.

all of above is only math calculating, we still need to count some other factors, such as the temp diff of the liquid inlet and outlet of the rad, but as we know from BillA's rad test and others' article like Joe Citarella of overclockers, the temp diff between in and outlet is minors. so basically we can safely drive a conclusion , brass tube will not be beaten by alot.
Forgive me if I miscalculate above.

Ok, so why wouldn't the tube material be important? Or were you just refering to what is already being used (aluminum/brass/copper). It maybe nit picking but if we want the best you have to use the best no?

One thing wrong with this calculation is it ignores the soldering of the tubes to the fins on brass/copper cores. I would assume the solder joint would act something like the TIM joint on CPU's making the heat transfer less efficent. It would seem to me the Aluminum cores may have a more efficent joint being it is an all aluminum and bonded joint (I think anyway).

[aaronspink]

Originally Posted by jaydee116

Aluminum is cheaper but is it cheaper to manufacture a all AL heater core? AL seems harder to bond (even AL to AL) than copper to copper or brass to copper. Wonder how they bond the fins to the tubes in all AL heater cores. Interesting about the dimples and dividers.

Don't know if it is cheaper to make, but it is cheaper to buy. Recent model heater cores for GMC K series pickup trucks look attractive (8.25x7.9x1 @ $30) and have impressive price performance. But they are Alu.


Aaron Spink

[joeyek]

Originally Posted by jaydee116

Ok, so why wouldn't the tube material be important? Or were you just refering to what is already being used (aluminum/brass/copper). It maybe nit picking but if we want the best you have to use the best no?

One thing wrong with this calculation is it ignores the soldering of the tubes to the fins on brass/copper cores. I would assume the solder joint would act something like the TIM joint on CPU's making the heat transfer less efficent. It would seem to me the Aluminum cores may have a more efficent joint being it is an all aluminum and bonded joint (I think anyway).

What is concerned in my calculation above focus on the heat transfer through the tube to outside, just like the base of a aircool heatsink, yes like u said, the soldering or some other joint technology are absolutely important, but no matter which thchnology u choose the heat transfer through the base is the same, when ignoring the second path for heat loss.
Jaydee, I said that is not important only wanna prove brass is not a loser when acting as the tube in a rad comparing to copper and aluminum, even it has a 1/4 conductive K of copper, that is all. However there is still some space to choose another metal with a lower conductive K according to my calculation, of cource at first make sure all we talk about is the big or huge rad.

[SysCrusher]

Originally Posted by unregistered

unfortunately copper and aluminum don't bond easily (that I'm aware of)
brass tubes are for strength, the modern aluminum tubes have internal dividers or dimples increasing both the surface area and/or turbulence
and aluminum is cheaper

always compromises

There is a big difference between the two I think everyone hasn't looked at. Aluminum will allow for greater surface area and is stronger than you could do with copper tubing. Aluminum allows for a thinner flat tube than copper which makes use of the available surface area better and increased turbulance inside the flat tube. The pic that JD posted of one cut in half will show this. The normal/old copper heater core tubes will not be as flat of a tube. Not to forget the weight between the two. Aluminum rad of the same size of the copper one will perform just the same as the copper and be lighter, cheaper. This is why auto manufactures tend to use Aluminum rads these days.
If you look at the pro-performance auto rad manufactures, The aluminum rads are bigger, thicker and have tons more surface area than the copper one of the same weight. They also cool alot better just based on that but you pay an arm and a leg for one.
But, we're talking about two different markets. I wouldn't want to buy an Aluminum rad for my pc and run 50/50 or more just for corrosion protection.

Me personally, would like to see a pc rad of copper based on Cathar's and a few other's post.

http://forums.procooling.com/vbb/sho...light=radiator

I would go as far as saying introducing turbulance inside the copper tubing - although it's an expense of pressure drop.

More food for the gander.

http://www.thermal-management-testin...issipation.htm

.05 in.H2O is about as good as you will get with an axial fan - if your lucky.

[BillA]

the current mfgn method for both copper/brass and alu is furnace brazing
I have been led to understand (no data !) that the thermall resistance of the brazed joint is lower than the soldered one - a lot to due with thickness of the solder 'TIM'

[jaydee]

Originally Posted by joeyek

What is concerned in my calculation above focus on the heat transfer through the tube to outside, just like the base of a aircool heatsink, yes like u said, the soldering or some other joint technology are absolutely important, but no matter which thchnology u choose the heat transfer through the base is the same, when ignoring the second path for heat loss.
Jaydee, I said that is not important only wanna prove brass is not a loser when acting as the tube in a rad comparing to copper and aluminum, even it has a 1/4 conductive K of copper, that is all. However there is still some space to choose another metal with a lower conductive K according to my calculation, of cource at first make sure all we talk about is the big or huge rad.

Ok, I misinterpreted your comments some. I thought you were saying it didn't matter much what the material was for the tubes not just copper brass or AL.

[jaydee]

Originally Posted by unregistered

the current mfgn method for both copper/brass and alu is furnace brazing
I have been led to understand (no data !) that the thermall resistance of the brazed joint is lower than the soldered one - a lot to due with thickness of the solder 'TIM'

Check this out. Adds some substance to your comments. http://www.copper.org/innovations/20...uprobraze.html

That article says the AL radiators are alloy. That kinds throws the calculations to hell. Wonder what the thermal properties of that alloy is..

[joeyek]

Originally Posted by jaydee116

Ok, I misinterpreted your comments some. I thought you were saying it didn't matter much what the material was for the tubes not just copper brass or AL.

thats it! JD
but I still wonder which rad is the ideal rad of Cathar, I can't find that thread.

[BillA]

thin, low fin density

[jaydee]

I think I finally found the thread. http://forums.procooling.com/vbb/showthread.php?t=10000

[jaydee]

I found this extremely interesting. I like the sound of a single pass thin rad that the fan can be utilized more efficently to make up for the lack of thickness. Pretty sure that's what he is getting at. That would be much closer to my opinion of a rad designed spacifically for PC water cooling. Should get less noise with more airflow aswell!

Originally Posted by Cathar

Where Bill is getting at here is matching the PQ curve of the fan to the air-flow vs efficiency curve of a radiator.

If people looked at the thread earlier where someone asked about putting two radiators in air-flow series and I munched the math, it was pretty clear that this was a bad idea. Two radiators in air-flow series is essentially the same as a single double-thickness radiator.

The issue was that the fan was unable to sustain adequate air-flow through the thicker radiators to the point where the extra core rows weren't doing much of anything at all (in fact they were contributing almost nothing).

With a piss-weak fan (and most axial fans are piss-weak) something like a large orificed area single-row cored heater-core, or a tranmission oil style cooler, will be a better match.

2-row heater-cores are designed to typically be matched up with their blower-fans that are pushing 1+ inH2O of air-pressure, which is an easy thing to do for an automotive heater-core blower where at full chop they will easily do ~3inH2O of air-pressure at quite staggering flow rates (in comparison to similarly sized axial fans). Heater-core blowers also tend to sit somewhere where the bulk of their real noise is muffled very heavily, so they don't sound quite as annoying as they really are.

Given a certain sized orifice area and a good bit of nouse, one could easily design a more efficient solution than what we presently have, and still be using fairly quiet fans and getting very good cooling performance. Of course brute-force will always win, but we're talking quiet and efficient here.

Current heater-cores, by my estimate, are perhaps 3-4x more restrictive than what is ideal for a quiet low-pressure fan to work really well with. Ideally we want our fan to be pushing something like 75-80% of its rated air-flow, instead of the 30-40% that we see today with today's dual-row cores. Of course with a more "open" core, the efficiency for air-flow will drop away, but when air-flows are fairly low, this is less significant than one would think.

Something like a 15x15cm cored area heater-core, that was single-row (~ 1/2" to 5/8" thick), with a nice 10-15 degree angled shroud leading into the 12cm low-speed fan that is pulling, and had a fin density around 10-12FPI with ~8mm or so spacing between the tubes would likely perform quite a deal better than a BIX when matched with the same fan.

As you stick on more and more powerful fans, the BIX will come into its own over our low-speed-fan optimised core, but then you're paying for it with noise.

The lessons are there given all the testing that Bill has published in the past for those who are prepared to munch the math.

[BillA]

and this is the intent of the BI Pro series, a la HW Labs

[Groth]

Originally Posted by joeyek

check the Delta T now, .00274c for copper and .00463c for aluminum and .001c for brass.

You've got a typo there, should be .010 for the brass given your conditions. Out of curiosity, where'd you get the rad with tubes that have surface area on one side only?

[jaydee]

Originally Posted by unregistered

and this is the intent of the BI Pro series, a la HW Labs

I seen the new BI's in that thread, well at least the 3 fan version. I now understand better why you don't want to help them make it better aswell and I assume your are already working on, if not having a protoype in hand, of a better version spacifically for Swiftech.

The main problem I have with current rads is they are not computer friendly as far as mounting. Swiftechs Rad Box helps to some degree but having that thing stick out the back of the case as demonstrated on Swiftechs website isn't a desirable thing for me. I would much rather it be easily fitted inside the case somehow. Can't always get what you want though.

[BillA]

6#s of sh*t in a 5# bucket
if the rad cannot be mounted via the fan, it will always be a chore

custom cases are the answer, but the bling bling makes it impossible to 'select' a case for all

[jaydee]

Originally Posted by unregistered

6#s of sh*t in a 5# bucket
if the rad cannot be mounted via the fan, it will always be a chore

custom cases are the answer, but the bling bling makes it impossible to 'select' a case for all

I hear ya. Let's also hope BTX never gets off it's feet!

[BillA]

Originally Posted by jaydee116

I hear ya. Let's also hope BTX never gets off it's feet!

amen
but the elephant walks where it will

[Cathar]

Yes - I would love to see some custom water-cooling cases.

A cavity at the bottom of the case with 3 x 12cm fan mounts along each side. Can stick in a BI I/II/III, along with 1/2/3 12cm fans as suits the user's budget/needs. A real opportunity there. Make it so the rads can be mounted on either side so the case can be positioned with the fans pointing away from the user to muffle fan noise, and rads facing the user so the user can always see if the rads are getting clogged with dust, and easily cleaned.

Could happily fit all that inside a mid-tower sized case, being effectively a midi-tower chamber atop the radiator enclosure in the bottom. Would have room for the pump in the bottom too.

As usual, the custom case crowd does make things somewhat difficult though.

[joeyek]

Originally Posted by Groth

You've got a typo there, should be .010 for the brass given your conditions. Out of curiosity, where'd you get the rad with tubes that have surface area on one side only?

yes u r right, tube area should be double , my calculation result is not accurate just some round numbers to show the diff of these 3 metal.

[aaronspink]

Originally Posted by Cathar

Yes - I would love to see some custom water-cooling cases.

A cavity at the bottom of the case with 3 x 12cm fan mounts along each side. Can stick in a BI I/II/III, along with 1/2/3 12cm fans as suits the user's budget/needs. A real opportunity there. Make it so the rads can be mounted on either side so the case can be positioned with the fans pointing away from the user to muffle fan noise, and rads facing the user so the user can always see if the rads are getting clogged with dust, and easily cleaned.

As usual, the custom case crowd does make things somewhat difficult though.

Though the CM Stacker is pretty close to that right now. Just need about an extra 40mm on the bottom compartment.