Dissecting the MCW6000-A
 

Shortly after I finished testing the Swiftech MCW6000-A, I cut it open to see how it worked. That probably doesn't surprise many of you. What may be a surprise is that I was torn as to whether or not to post this. On the one hand, I know that there is a lot of "idea liberation" that goes on in our forums. It is also clear that it's hard to make money back on R&D when someone can just copy your design the day it's released. On the other hand, if I don't post this information then it can't benefit the DIY wb designers, it won't be clear to our community that the ideas and designs originated from Swiftech, and unscrupulous people will just cut wbs up and steal the ideas in private.

When you look at Swiftech's website, an important fact about the MCW6000-A's design is in fact already posted:

"The base is a one piece copper forging made of C110 copper, featuring 281 thin pins. "

Look now at Bill Adams' comments about the MCW6000-A design from our review:

"The MCW6000 is innovative in the following regards:
use of multiple temperature brazing/silver soldering processes
considered inlet nozzle design, placement, and internal flow control"

Let's keep the above in mind as we look at the MCW6000-A

I tried to cut the block open with a hand saw and tin snips; I mangled one corner of the wb pretty badly (and managed to gash my thumb to boot). Today I used my defective dremel (it only runs at maximum rpms) and did a much better job. You can ignore the pin design in the mangled area as that's my doing not Swiftech's :) Click on pictures for larger versions.

This is probably the most descriptive of all the pictures:


You can clearly see my handiwork with the handsaw and snips (Take note, Owen!) and you can also tell that the outlet of the block is on the bottom left. The pins are round, evenly spaced, and quite dense. The rationale for having a forged base is clear when you see the pin density; couldn't make something like that easily with a mill. The other important bit of design that is clearly visible is the "internal flow control, a "U" shaped piece of metal that forces water to go from the inlet out through the bulk of the pins rather than straight from inlet to outlet. You can get a better look in the picture:


Note the pins directly under the inlet are bent out some to lower resistance. It also is of importance that the inlet is centered in the wb rather than being centered over die of cpu. This means that water movement through wb must be very important to cooling rather than just the initial impingement at the inlet.

Just to show pin density here is a final picture:

LOL, you just saved me $40. Don't feel bad about posting this I was going to do the same thing anyway. :D This looks exactly as I pictured it in my mind from the animated example on their site. The little divider caught me off gaurd though. Makes me wonder why the Aqua Gold block has this divider milled in when all they need to do is bend a thin peice of metal and set it in the pins. :D

Good stuff pH.

I am thankful your not a surgeon. :)

Great work, I was tempted to to try that out, due to curiousity.

Don't feel bad both of my dremels are like this. Kinda sucks for doing anything slowly. Oh well, full on power is often the way to go. Except at 2am when your roomate is sleeping! :D

I was wondering about this

It is pretty obvious from the tests that there was more to in than what could be seen from the graphics on the site

I figured that no reviewer would want to do such a thing (tear it apart), but i figured wrong

The thing you would think would either be a "crescent" or a "maze"

remember when billa posted the link to that asetek R&D pdf?
there was a chapter analyzing the differences between a "crescent" flow control and a "maze"
they determined that crescent was better
just look at their waterchill block (no, not the antarcticrap or whatever it is)

very impressed

thanks so much for sacrificing the block for us Phaestus, but I am sure you were just as curious as we were, and someone needed to do it.
hey, they're only 40 bucks, right?

That's one sloppy cut :D.
Still, interesting. Expected the variation on Hoot's (effective) block.
I'm assuming the exit is located in the upper right corner (of the top picture), right?

Hehe. Been there, done that. I had to finish building a plane for a design project in one night, and i had to use the dremel at fuul power all night.

However, i didn't give a shit about waking up my roomate cus he destroyd my first plane in his drug induced tyrade the night before. :shrug:

I think I know where the inspiration for the 6000 came from... Once upon a TIME... I love this new block!

ya, everyone thought that when they first saw it

but the biggest diff. is instead of a maze, a crescent, er, half-moon flow pattern

this one seems more dense too

the trick was making it cheap too
hoot painstakingly trimmed each one of the pins by hand

I'm not really surprised by the design. I was expecting something to move the flow around, but I expected it to be a bit more complex. I wasn't thinking this because of the performance though, but because of the wasted pins... I don't think you'll ruin anything for Swiftech, as the trick is seemingly not to make the basic design (doesn't seem too hard), but to make it efficiently, and optimize it.

Question.. why use all those pins on the outside of the center pins if they aren't being used for cooling. The core is small. If they put this on a heat spreader, the heat isn't going out away from the center much. Couldn't they just use the inside 10X10 grids? To my knowledge, theres no way to mount a pelt to it is there? Wouldn't that make it less restrictive?

I'm guessing that the water will be forced through the outer pins due to the density and thickness of the pins.

The block isnt a very thin based block. 3/16th inch I beleive, and all those pins all across the block do help with performance. Thats where the some what thicker block base comes into play. If it was too thin then yeah the outer pins wouldnt help but its a balance of base thickness vs surface area and taking into account copper's thermal resistance. Yes, a very simplistic explanation but the mcw-6000 shows a thicker base waterblock can perform very well.

huh? I understand fluid flow to be like flow of electricity. It goes the path of least resistance. A huge flow may force it into the outer pins some, maybe this is why they dont use this for pelt cooling?

The pins are too dense/thick for all the water to go right around the inlet. The path of least resistance is fanning out so you don't get ~1LPM through a 5mm*2mm area, but rather 1LPM through 10 5mm*2mm areas. It's like doors and a crowd. The closest door is on the left, but 2/3 of the people are going there, so you'd naturally go to the door on the right, so everyone isn't crammed through together.

I don't get it.. where are you getting these measurements from? What is 5mm by 2mm?

Now I'm worried about sending a preview block to Phaestus....

hehe Cathar, if pH doesnt someone else will :) so heres the deal, just never let blocks leave your house ever and no problems! Infact destroy all blocks you make instantly to keep their secrets from sneaking out ! :p

Deleted.
Repetition of pH's observation.

I said preview. Completely understand the nature of people buying and pulling apart commercial blocks.

Their placement (the bent ones) and the overall discoloration pattern is something to notice also.

I am respectful of NDAs and of conditions placed upon testing of preview wbs Cathar. But precedence HAS been set re retail wbs :)

Way to go pHaestus! ;)

Giving it the ole can opener treatment!

I was also reminded about Hoot's pin fin block: it's been an inspiration to me too, but not to replicate/mimick it.

Can't wait to see Bill's comments on this... Gotta save the pics!

eh... u shoulda sent the block to me :D

I think the whole smplicty is quite deceptive here. It is not simple at all. The whole flow conrol uses two different mechanism there to my understanding. One is purely Newtonian (inertia) and second is derived from the Poiseuille's Law and pressure drop gradients.
Very clever due to elegant engineering employed at minimal cost!
As product it is just perfect!

BTW Nice hacking jobbie there pH! Did you enjoy doing it, huh? C'mon, we know you did!;)