new watercooling setup; best of the best

[lisch]

ok, im trying to piece together the best of the best components... what do you guys think
http://www.leufkentechnologies.com/page5.html
for the pump and reservor
http://www.swiftnets.com/cgi/shop.pl/page=FrameSet.htm
for the mcw462 block ---- hey joe, what do you think of this one??? looks sweet
http://www.3dcool.com/radiator.shtml
for the 5x5 radiator. this would mount great in the top of my inwin Q500 full tower

btw, from earlier post, do you all think it would be better to make my own reservoir and go w/ a mag supreme 250 pump ?? again, i need to mount the radiator to the top of the case like this
http://www.overclockers.com/tips196/

man i need some advice!! i want this thing to last, and cool my o/c ready 1gig tbird down.

[Joe]

I think all of that is good, but some things to think about:

OCWC's blocks ( the Cu one and more so the Ag one) will blow the swiftecs away for now.

the 4"x4" radiator that OCWC sells smoke the forced path lil 5x5 ones!

and the metal res tank may only help conduct heat back into your case, but it would be a mintor thing to worry about compared to the others. But the block and radiator that you picked are not so good.

------------------
C-ya
Joe - Owner/Editor
www.ProCooling.com
Where the Completely Addicted Come to Cool Off

Somebody set up us the bomb.

[lisch]

thanks so much for your response! btw, i loved your DH articles

hmm, the main reason i liked the swiftech block, was because of the mounting and the 3x3 design, more room for water=more cooling no? also it got a good review from overclockers.com....

is the cooling just better on the other Cu blocks or is it the design thats bad...?

thanks again.

ps.. about pumps, do you think the leufkin pump ( Rio200 is what they use) would push water enough to the top back of my case? (InWin Q500 full tower)

[BillA]

I'm curious ....

Joe, do you have (or have you had) a Swiftech MCW462 ?
(I've paid for one but do not yet have possession.)

I've seen no DATA (NUMBERS eh) to support your statement: "OCWC's blocks ( the Cu one and more so the Ag one) will blow the swiftecs away for now."

Not on this site, nor any other.
Would you please substantiate your claim.

be cool

[Joe]

Funny enough the data I got was FROM swiftec stating that currently Chips blocks will surpass theirs, but in the future they think that will change.

Swiftec thinks the volume of coolant you keep in a cpu cooler affects cooling more then its movement over a surface. I mean Chips blocks have half the surface area of some blocks, but just remove the heat so well.

this weekend I will be doing some Numbers on his blocks. But Swift was the one who said that their current generation of blocks are behind chips design, and their new one will be up to OCWC's cooling specs.

I still doubt it though. And we have been offered one for ProCooling... just waiting for it.

The base to my conclusion to chips blocks being better is that there hasn't been a block thats been as light, and efficient at what it does. Also this can get into business experiences people have had with swiftec.

Chips stuff is tried and true. Also my statement was based on the design of swiftecs block Vs Chips. Theres a few flaws in it on how the coolant flows.

( Hint: Swiftecs treatment of its customers was a reason OCWC was formed.)

------------------
C-ya
Joe - Owner/Editor www.ProCooling.com
Where the Completely Addicted Come to Cool Off
Somebody set up us the bomb.

[This message has been edited by Joe (edited 01-05-2001).]

[This message has been edited by Joe (edited 01-05-2001).]

[Joe]

Before we break out some federal grand jury about real #'s and such. I would like to also clear up that what swiftec said about OCWC was told to me by a ProCooling reader, so it IS hearsay, but take it for what its worth.

------------------
C-ya
Joe - Owner/Editor
www.ProCooling.com
Where the Completely Addicted Come to Cool Off

Somebody set up us the bomb.

[lisch]

ok, thanks that clears some questions up for me. geeze, why hasnt someone just made an all copper block w/ the retension mech that is offered on the mcw462... seems like a no brainer.

[BillA]

For Joe:

Ahhh... Its like a breath of fresh spring air to hear a manufacturer 'fess up to the inadequacies of their product.

However I've had conversations with Gabe Rouchon and can assure you that he does not consider Swiftech's products to be inferior. So I thought I would discuss this with him and opened this thread to be able to quote you accurately.

Fortunately, because I saw your retraction before I repeated some trashy, unattributed nonsense.
You shot from the hip again Joe, and missed, again.

I asked you for data, and this was your response:
"Funny enough the data I got was FROM swiftec stating that currently Chips blocks will surpass theirs, but in the future they think that will change.
Swiftec thinks the volume of coolant you keep in a cpu cooler affects cooling more then its movement over a surface."
"But Swift was the one who said that their current generation of blocks are behind chips design, and their new one will be up to OCWC's cooling specs."

You are confused Joe, that's not data, that's a (now secondhand) opinion. So not having any data, you based your OPINION on ....
But no facts.

The "flaws in it on how the coolant flows", remember that until you have facts (numbers) its only a (not very well informed) opinion.

Yes, I too know Chip Eckert and I consider him to be one of the more clever designers in the watercooling "business". (In fact I have a small project going with him now that you are probably even aware of.)
Here are some recent thermal test results http://www.hwlabs.com/products and I trust you can see the difficulties in drawing absolute (or even firm) conclusions.

I look forward to your test (results), and trust you will tell us how the experiment was designed, steps taken to limit and control variables, and how the data was analyzed to attribute cause and effect.
(I think that super-monster-muther-waterbox of yours will have more variables than 10 gorillas could get their arms around. Just my opinion, eh.)

Rigor Joe, testing requires rigor.

be cool

[Joe]

DAMNIT
I had a whole nice big post all typed up and freak'n IE died...

OK this is the very very short version:

1. DH3 had 13 temp sensors, with those all you can gauge and temp variables that are introduced in a test set. ( as well as 2 ambient temp readings out side of the machine at 5 In from the floor and one at 4 ft from the floor.) in any test all the temps will be posted for the people ( like you) to work out efficiencies and and variables to discount the final temp reading)

2. I have built a machine just for water block testing and as soon as the gear gets in it will have 8 temp sensors itself. ( its a small 1 loop system deigned just for this)

3. Swiftec's flow problem is all about where the coolant is coming in and leaving. the Tidal Pool design is VERY effective IF you have the coolant moving correctly, hes squirting the coolant into one area in the cooler, and letting it fill to the point it flows out the exhaust port. Theres very lil actual movement of the coolant ( unlike the tidal pool design). ( example : Blow on your hand in one concentrated area. That area gets cool ( or warm) while the res is hardly affected. Now Blow across your hand, you notice you get the same effect across the entire area you are blowing, not just a tight center contact patch. Now this isnt a scientific test by any means, just an example)

4. This is all based on opinions and my experience with how coolant flow and design directly effects the final product. The final product maybe perform better then I think it will or worse. Its beautiful block, and the mounting mechanism is awesome, But for cooling I don't think it will surpass much of the higher end blocks around. His design IS better then the forced path blocks though, I will give him that, and its NOT anodized, which is another big plus in my book.

In all reality the difference between his block and OCWC's Cu poly block will be slim ( +/- 3 - 4Deg I would expect) But I think OCWC's Ag blocks will walk away from even the most advanced Cu/Al blocks.

Another thing I have a problem with is the fittings included on the swift block. I like the normal brass hose barbs more, just don't trust the ones hes using, I have had too many "other" hose connects like those either leak, damage the hose, or just come off. Its more a personal preference. ( also with the Dye Lite you can see if its creeping down the hose barb long before it actually gets to the point of leaking.

God I wish I didn't loose that long post..... although this one turned out pretty damn long

------------------
C-ya
Joe - Owner/Editor www.ProCooling.com
Where the Completely Addicted Come to Cool Off

Somebody set up us the bomb.

[This message has been edited by Joe (edited 01-05-2001).]

[BillA]

Much more informative explanation.
I look forward to the results.

I presume you're aware of thermal test dies
http://www.electronics-cooling.com/h...00_sep_a3.html

be cool

[Joe]

Yep! I was actually thinking of fab'n my own test setup ( like the boys at Trainwreck did.. that was a BITCH'n setup! with pelts used as the temp producers.)

I really want to get a constant heat source to run some threshold tests and such to see where the sweet spot is in some cooling setups.


------------------
C-ya
Joe - Owner/Editor
www.ProCooling.com
Where the Completely Addicted Come to Cool Off

Somebody set up us the bomb.

[BillA]

I don't think so, Joe
The purpose of the thermal test die is to replicate the SIZE, and to control the heat.
Even if you have a TEC the correct size, controlling the backside would be a bear.

On an older topic, since we've had to hear what Gabe Rouchon supposedly said, let's let him say what he DID say (quoted with his permission):

"I had a conversation with a customer a couple of days ago, and we were discussing the MCW371 (an all aluminum design), and I DID tell this customer to go an buy the silver base product because silver offers a vastly superior thermal conductivity than Aluminum. But that was ONLY referring to the MCW371. The new MCW372 (copper base) is not due out until next month, so I thought it was only fair to send the customer to the best candidate - for now :-)

As to the MCW462, we primarily designed it to accomodate a new generation of TEC's (larger sizes) for AMD processors. I do not believe that any of our competitors have looked into this.. But I am sure they will soon :-)

Performance-wise, I suspect that it will perform significantly better than a smaller block because it has a much larger interface with the cooling fluid. Heat EXCHANGE is the key word here. We are exchanging calories between the processor and the fluid, and the interface efficiency results from a combination of thermal conductivity and surface area. The MCW462 offers 100% more surface area than any other block I am aware of. So for now, we should lead the pack easily.

As to using silver for a base instead of Copper, let's look at numbers :

Absolutely Pure Silver has a thermal conductivity of 419 W/m-K
The high conductivity copper C1010 we use is at 391 W/m-K
The best aluminum alloys hover around 220

Whereas there is an enormous difference between aluminum and copper, the difference between copper and Silver is so small that it absolutely doesn't justify the cost.

There is a real problem in this industry with thermal management. I seriously doubt that silver is the solution."

And I have a recent article on overclockers.com to the same effect: http://www.overclockers.com/articles305/

be cool

[This message has been edited by BillA (edited 01-05-2001).]

[Joe]

will reply once I get home...

your comments on the pelt to produce heat accurately, you are exactly right. and thats the main drawback to it.

more on this later.

------------------
C-ya
Joe - Owner/Editor
www.ProCooling.com
Where the Completely Addicted Come to Cool Off

Somebody set up us the bomb.

[lisch]

( crap, what have i started now )

[BillA]

Hang in there lisch.

When this is all said and done I'm SURE you'll see that the MCW462 is tops.
(I'm going to cut a window in the copper base plate and solder in a Cusil plate.)

And I think we'll see that flow rates need to be "tuned" for the waterblock-radiator combination, something no one else is addressing.

be cool

[lisch]

wow! now thats interesting... never thought of that... man, i need to learn to be patient i was at hardware store today, piecing together what im gonna need..

only thing holding me up realy is the waterblock; one thing that really impressed me is the way you can bolt the mcw462 down to the mb distributing the load evenly. being this is my first watercooling attempt im afraid i cant offer much opinion, instead only ask questions, hah..

[BillA]

I'm waiting for a Palomino, and thats got me IMpatient.

be cool

[lisch]

omg yes! can u imagine the o/c'ing were gonna see out of those chips.. i think the first ones will be 1.333's cant wait

[Joe]

Originally posted by BillA:
Hang in there lisch.

When this is all said and done I'm SURE you'll see that the MCW462 is tops.
(I'm going to cut a window in the copper base plate and solder in a Cusil plate.)

And I think we'll see that flow rates need to be "tuned" for the waterblock-radiator combination, something no one else is addressing.

be cool

hehehehe You have had a bitch with me for some time haven't you?? I happened to get some links to other forums that you were basically slamming the work I was doing and TOTALLY missed the point on much of it.

You are SURE the MCW462 will be tops?? hows that?? if I cant say its going blow compared to others, how can you claim it will be tops?

Flow rate issue?? There is a point where you move the coolant TOO fast, but at that speed you would need a very powerful pump, and a good deal of pressure to move the coolant like that. The real key is keep the coolant moving on the plate face, and the entire face, not just the center. The "Tuning" that needs to be done is the Block and Radiator with the heat source.

Chips blocks are SO easily made bigger to cover a bigger pelt, as most can be.

CuSil is obscenely expensive and you were the person discounting the performance difference between Cu and Ag. Ag and CuSil has a SLIMMER advantage! So wouldn't it make sense to use an entire silver bottom and get overall performance within 1 or 2 Deg of CuSil at a fraction of the cost? You were the one bitching about the cost of Ag....

Swifts blocks are pretty, but there is a big difference between pretty, and functional. ( CpuFX is a good example of that, pretty blue blocks, but real poor performance.

I realize you are tight with swift, but I wont believe that the gods gift to cooling MCW462 will perform as good as you think.


------------------
C-ya
Joe - Owner/Editor www.ProCooling.com
Where the Completely Addicted Come to Cool Off

Somebody set up us the bomb.

[This message has been edited by Joe (edited 01-05-2001).]

[ck42]

This is great discussion guys! Let's keep it going.
Having an engineering background in test and process control, this is exactly the type of discussion I like to see.

I too would like to give a big thumbs up to the MCW462's mounting configuration. I sure wish other vendors of water blocks would implement this type of design.

In case neither of you are aware of this water block (if you want to call it that), Lytron makes a VERY interesting device called the CP20. There is someone named Mota who is putting together a system using the CP20 and is posting in the [H]OCP forum.
Visit here for pix and discussion: http://www.hardforum.com/showthread.php?threadid=82173

In any case, I'm still SLOWLY putting my first water cooled system together but still am not satisfied with the type of tests conducted out there...at least enough to make a sound decision on one part over another. If I had the $$ (student...again), I'd LOVE LOVE LOVE to put together a test rig and perform rigorous tests of all the water cooling equipment out there.

Happy trails!

[Joe]

Hey Ck

That CP20 is nice, I am getting me a few of those. Many ideas to use them on. Theres some SERIOUS surface area there. If only it was made from Cu

------------------
C-ya
Joe - Owner/Editor
www.ProCooling.com
Where the Completely Addicted Come to Cool Off

Somebody set up us the bomb.

[BillA]

That is a hellava thread on hardforum, lots of different (to me) stuff. OK, back to work.

No Joe, no bitch. I was stupefied with your super-monster-mutha-waterbox, did (do) not understand the purpose of that degree of complexity, and have said that. It was explained to me that you did it to receive free equipment to test. ok, that's your deal, not mine.

I observed that as how you had spent $1200, for $400 you could have bought a recirculating chiller (+ or - 1C to -15C), plus say another say $100 for incendentals, and had some control over your tests.

The basis of my critical comments has to do with the "design" of the experiment. The more variables, the harder to control, and the harder to understand the results. This is particularly the case when the increment being measured is very small. (If it is below the coefficient of variation for the equipment and procedure, it is invisible.)
Good testing equipment is simple, as are the procedures.

Why is it my opinion that the MCW462 will "be tops" ?
Glad you asked (and so politely, too).

The heat removal capability of the coolant is (among other things) a function of the temperature difference between it and the object being cooled. By initially applying the coolant, when it is at its coldest, to the center of the waterblock (which is presumed to be the hottest area), the heat transfer is optimized for that specific area of the waterblock.

The heat "removal" capability of the waterblock is (again among other things), a function of/dependant on the heated surface area exposed to the coolant; and the MCW462 has far more than the OCWC designs.

I am not convinced of the benefits of swirling coolant. The transfer of heat to the coolant takes time, note the very real design flow rate limits for flat tube (and fin) radiators. I suspect that your "sweet spot" (if I understood your reference correctly) may well be at lower flow rates. This testing can easily demonstrate - or deny.

While I like the use of polycarbonate for sub-zero cooling, I think a metal waterblock might be incrementally better at ambient. (I think we are now picking invisible nits, very hard to test.)

The best waterblock of Chip's was probably the Hedgehog in a box. Perhaps I should cut one down and solder it into a cutout in the MCW462's baseplate.

And the choice of materials, Joe: Go to http://www.overclockers.com/articles305/ and READ the GD article. If you wish to use that DATA, GET IT RIGHT. You waste everyone's time, and obscure the issues, by scrambling the numbers. Here now, I'll lead you by the hand .......

Aluminum: 220
Copper: 388
Silver: 418
Silver/Copper (Cusil) Alloy: 515

The difference between copper and silver is ... 30

The difference between silver and Cusil is ... 97

Now I will let you explain how that is LESS.
I will return to this issue (but why ?) after you straighten out your comments.

Am I tight with Gabe at Swiftech ?
No more so than with any technically competent individual with shared interests.
(But he did like the article on cold plate materials.)

be cool

[This message has been edited by BillA (edited 01-05-2001).]

[Kevin]

Originally posted by BillA:

No Joe, no bitch.

The only person bitching in here is you...

Originally posted by BillA:

It was explained to me that you did it to receive free equipment to test. ok, that's your deal, not mine.

I am SURE you did not here that from Joe. Joe's heart and soul are in his work. Free equipment? He spent SO MUCH money and time on this machine for his personal pleasure that it ain't even funny. The one thing Joe is NOT is a sell-out.

Originally posted by BillA:
The basis of my critical comments has to do with the "design" of the experiment. The more variables, the harder to control, and the harder to understand the results. This is particularly the case when the increment being measured is very small. (If it is below the coefficient of variation for the equipment and procedure, it is invisible.)
Good testing equipment is simple, as are the procedures.

DH3 was not an "experiment." It was a watercooling system that all of Joe's future computers will be built into. You get all mathematical and stuff. Joe, myself, and most others around don't build cooling rigs to make our coefficients greater than our increment or whatever... We do it because we find it interesting and fun. I always thought complicated things were cool, but I guess people like yourself need simpler things to understand them.
:/
Look... if you wanted to have a big argument with Joe about which block you think is going to be the best, you should have e-mailed him or done it in another thread.
.
Back to the topic at hand... Lish, I like your choice of components. I would recommend against the Swiftech block. Even IF that block slightly outperforms OCWC's (which I am unsure of... and it will DEFINITELY not beat his silver blocks...), it does not justify the exorbitant price. In agreement w/ Joe, the radiator isn't that great. I have it. It simply cannot keep the water near room temp w/ a 120w pelt.

[This message has been edited by Kevin (edited 01-06-2001).]

[lisch]

Kevin, thanks for your opinion too..
im so glad i found this site and forum, i've
been educated big time here...

guys; keep up the good work, we all have our
opinions on what is best ('cept me ) ... both blocks look great, and would appear to perform great too. i'll definitly get a different radiator; it will be hard to wait a month for the little black 4x4's to come in stock though...

thanks again for all the great information. without sites like this one, us "newbies" to watercooling would be lost

[Kevin]

No prob for the help... you might wanna grab a dangerden cube radiator. It should perform similarly to chip's 4x4 radiator.