Swiftech Apogee GT vs the D-Tek FuZion CPU blocks @ CES

Quote:

---

Originally Posted by Graystar

Just as it would be silly to say that hot water placed in a freezer will freeze faster than cold water placed in the freezer...yet under certain conditions that's exactly what happens.

We live in a complex world. Nothing should simply be dismissed outright.

oh...what if your pool of liquid was hotter than the rock? The the statement would be true!

---

I'm not dismissing anything- especially the laws of physics, heat transfer and thermodynamics. Again.... let's not negate the laws of physics, heat transfer, thermodynamics and so on here.

As for the freezer thing- that's a commonly asked Heat Transfer 101 trick question. The hotter water freezes faster because it induces a faster rate of heat transfer due to the higher DeltaT. Been there, done that- even got the T-shirt ;)

It doesn't matter if the liquid was hotter than the rock, or whether we're talkng about the temperature of the hot rock or the liquid...... my point is this: the more surface area in contact, the more heat transfer. Lets also remember, the hot rock problem assumes no further heat in or out by either source (unlike in the computing situation we are in here).

and.... If my pool of liquid was hotter than the rock we wouldn't be watercooling would we :D

I'm cool with it... we can agree to disagree;)

Please accept apologies for the off topic...

Gabe, can you please check your PMs? Want to discuss editing your title to show who you are so there will be no more mis-conceptions.

Thanks

IFMU ~ Marty

So, any chance one of you with the FuZion measured it for TEC use? Its geting hard to find 6002s....

Quote:

---

Originally Posted by Plywood99

I did that. Heated the block up on a gas stove till the solder joint melted. Then used a small screwdriver to lightly knock the cover from the base. I think the U is made of brass.

Tricky part was getting it all soldered back together. Clean surfaces then use flux to help wick solder into the joint...


Ply

---

I may look into making a new mid-plate and top out of derlin or something.. if possible. If I get around to it... if if if if :rolleyes:

Quote:

---

Originally Posted by gabe

place index and thumb of each hand at each corner, and slide the CPU back and forth

---

In other industries this technique is well known to give convex bottoms. Don’t use it.

The proper technique for “lapping” (we’re not actually lapping) is to push the item forward then lift and return to the starting point and push forward again. Never rotate or pull back.

Here’s a complete description of the process...
http://www.ocforums.com/showpost.php...4&postcount=11

It’s even endorsed by BillA...
http://www.ocforums.com/showpost.php...05&postcount=4

Gabe, I set your account to be able to set your own user title. Go into user CP and you should be able to set it.

Quote:

---

Originally Posted by Graystar

In other industries this technique is well known to give convex bottoms. Don’t use it.

The proper technique for “lapping” (we’re not actually lapping) is to push the item forward then lift and return to the starting point and push forward again. Never rotate or pull back.

Here’s a complete description of the process...
http://www.ocforums.com/showpost.php...4&postcount=11

---

I concede that your technique migh be safer to untrained hands. Particularly if there is a lot of material to remove. We use the technique I described for finishing/polishing after the lapping machine has ground the parts flat. Both techniques do require training and patience. We are able to make perfectly flat/mirror pieces using this technique (as in measured with optical flat).

I am sorry Iany. I come from hundreds of hours of lapping and polishing parts myself over the years. What may seem easy to me might not be to others. Follow Graystar's technique, it is long, but safe.

Quote:

---

Originally Posted by Graystar

In other industries this technique is well known to give convex bottoms. Don’t use it.

The proper technique for “lapping” (we’re not actually lapping) is to push the item forward then lift and return to the starting point and push forward again. Never rotate or pull back.

Here’s a complete description of the process...
http://www.ocforums.com/showpost.php...4&postcount=11

It’s even endorsed by BillA...
http://www.ocforums.com/showpost.php...05&postcount=4

---

Very interesting. This raises some observations.

In lapping the IHS, we are taking out surface irregularities that may affect performance, but in doing so we are introducing a convex IHS surface.

We are then artificially manipulating the waterblock baseplate to be concave to better match our artificially created convex CPU IHS.

I guess the question to raise is this. Are the results of such a fair indication of performance?

Does the performance of the concave base-plate still hold true with an unlapped CPU IHS?

In terms of seeking pure performance, the convex IHS and concave w/b do appear to be a good match, but is everyone prepared to lap their >$900 CPU and thereby invalidate any warranty.

Just asking questions, that's all. No attempt to knock, but rather to clarify.

I thought it was known that lapping was imperfect?

Any method used means you put more pressure on one area than another depending on how you are holding the item and which direction you are moving it while lapping.. doesnt matter if you go straight, circle, or figure-eight.

Heck, i've even got a machine meant for this stuff (vibrating bowl that you fill with abrasive material) but its still imperfect.

Quote:

---

Originally Posted by Cathar

In lapping the IHS, we are taking out surface irregularities that may affect performance, but in doing so we are introducing a convex IHS surface.

---

Not so. We MIGHT be introducing convex surface if not careful. Care and rigor in the process are important. Checking progress with straight edge against a light helps greatly.

Quote:

---

Originally Posted by Cathar

We are then artificially manipulating the waterblock baseplate to be concave to better match our artificially created convex CPU IHS.

---

Not so. We are manipulating the water-block base plate to be convex, not concave.

Quote:

---

Originally Posted by Cathar

I guess the question to raise is this. Are the results of such a fair indication of performance?

---

Performance of what? fairness to what? the water-block hydraulic design, the thermal joint, both?

As to fairness, from what point of view? We are not discussing the merits of such or such hydraulic technique here. We are discussing a possible improvement in the thermal joint between IHS and water-block

Quote:

---

Originally Posted by Cathar

Does the performance of the concave base-plate still hold true with an unlapped CPU IHS?

---

Again, not concave..Convex. and yes, we did measure positive results with an unlapped CPU. Does it hold true in all case? We don't know, until we test every single IHS config we can get our hands on: concave, convex, irregular.

Quote:

---

Originally Posted by Cathar

is everyone prepared to lap their >$900 CPU and thereby invalidate any warranty.

---

A good point. I am curious as to how many people in this forum uncapped their AMD IHS. I for one never cared (dared?) doing it until OP showed me.

I have lapped my Kent (paid $1k for it). I did it because I feel comfortable with lapping due to experience. The guys at Xtreme lapped both Kents without even the hint of a second thought...

sounds like we need waterblocks that include different bases for different cpu applications.. let the end user test and decide which is best for their application....

choice = power :D

My mistake. For some strange reason I thought that the bases were being made concave.

My apologies. I really should get more sleep.

By "fair", I had meant whether or not it represented the usual/general scenario (vanilla w/b on a vanilla CPU), that's all. The comments were not meant to be inflammatory, although upon re-reading, I guess they may have sounded as such.

Quote:

---

Originally Posted by eva2000

sounds like we need waterblocks that include different bases for different cpu applications.. let the end user test and decide which is best for their application....

choice = power :D

---

That would only be good if different bases can be done for a small price, which I doubt aint possible, because copper is not that cheap.

I for my part dont want to spend a fortune on a waterblock. I usually buy a good performing block with a price that is ok for me.
No way I am spending an additional 30$ on a block that is 2C better.

That is why I will never get a storm over here. 84euro is way to much for me, which is why I went with a Alphacool Nexxxos XP that is half the price and not that much worse.

And from what I have read on German forums, most people think the same way I do. There are only a few extreme enthusiasts who buy the best stuff available, no matter at what cost (for the record: I consider myself a enthusiast aswell).

If I want to break records, I go with dry ice or LN2.
For 24/7 usage I want to have good performing components for a good price, a few degrees mroe or less dont matter.

Quote:

---

Originally Posted by eva2000

sounds like we need waterblocks that include different bases for different cpu applications.. let the end user test and decide which is best for their application....

choice = power :D

---

Heheheee they should call it waterblock for masochists or Apogee WM. Drain and bleed with every cpu change.

Well again there's power in choice

3 different packages

1. waterblock + base #1
2. waterblock + base #2
3. waterblock + base #1 + #2

heh

Quote:

---

Originally Posted by STEvil

Heck, i've even got a machine meant for this stuff (vibrating bowl that you fill with abrasive material) but its still imperfect.

---

That sounds like a deburring/polishing rig and you're right...it would be "imperfect" at actually producing a flat surface.

Hmm, this is so confusing - how many times will I have to reseat these things for optimal setup. Watercooling is confusing enough :(

The guys over at Cooling-Masters have a thread on DIY lapping machines made from old record players, nice results on blocks and dies, might be interesting for people who lap more than the occasional block/IHS.

In looking at most of the lapping progress shown in various users pics here, it seems quite obvious most all IHS's start out convex as evidenced by material being removed at the center first, then towards the edges to become more flat after lapping.... yes??

Therefor, it seems to me that a flat base, or even a slightly convex base should work just fine for the average Joe with an unlapped IHS. If the WB base is tweaked to a concave state just slightly during installation (at this point, I'd say it would have to be slightly overtorqued) you'd have near perfect mating between the convex IHS and concave base.

So why lap??

people!

it does not matter how flat your base is.

due to how the simple mechanics of how it is mounted, it should benefit EVERY cpu. even those without IHS.

here is a simple example for those who having a bit of trouble

Take a waterglass, and a CD

place the CD on the bottom of the glass. it will sit flat.

push on the outer parts of cd, (replicating a block mount) and you will see the inner part of the CD rises up and away from the center of the glass


that is what is happening to the base of the block when you mount it.

if you are able to understand this, you will understand that this will work for any cpu. and then you will see all this talk about different IHS is irrelevant.

Quote:

---

Originally Posted by cdelong

In looking at most of the lapping progress shown in various users pics here, it seems quite obvious most all IHS's start out convex as evidenced by material being removed at the center first, then towards the edges to become more flat after lapping.... yes??

Therefor, it seems to me that a flat base, or even a slightly convex base should work just fine for the average Joe with an unlapped IHS. If the WB base is tweaked to a concave state just slightly during installation (at this point, I'd say it would have to be slightly overtorqued) you'd have near perfect mating between the convex IHS and concave base.

So why lap??

---

We checked 2 of the C2's that we have at work: the QX6700 was convex. The E6400 was concave. At CES Fuggers QX6700's were what I can only describe as "irregular" with high edges and high center section. We are researching the subject matter, trying to see if there are statistics available.

Quote:

---

Originally Posted by cdelong

In looking at most of the lapping progress shown in various users pics here, it seems quite obvious most all IHS's start out convex as evidenced by material being removed at the center first, then towards the edges to become more flat after lapping.... yes??

Therefor, it seems to me that a flat base, or even a slightly convex base should work just fine for the average Joe with an unlapped IHS. If the WB base is tweaked to a concave state just slightly during installation (at this point, I'd say it would have to be slightly overtorqued) you'd have near perfect mating between the convex IHS and concave base.

So why lap??

---

In theory, this is fine, but hard evidence and testing in many different systems yields only one conclusion...flat waterblocks produce better temps with flat IHSs. While WSP's theory does make sense, but I'm not
willing to base anything off of it until we get some hard evidence or math that proves that our waterblocks are contorting to a significant degree, or even that bowed bases under significant contortion becomes flat. Also, if his theory is correct, than wouldn't a severely-bowed base perform differently than a barely-bowed base or moderately-bowed base? Would there be a "sweet spot" for each waterblock that allowed a bowed base to become flat? If so, then would there not also be a "sweet spot" involved in how much pressure was mounting the block to the CPU?

IMO, the problem with watercooling forums is that too many posts are based off of theories that "make sense", but don't have any actual evidence to support them. A specific theory is not true simply because it is currently our only mechanism for strange, previously unexplainable phenomenon. (An example of these "phenomenon" is bowed bases yielding better temps on mirror-finished, almost-flat IHSs.) I know that many sciences work in this manner, but I still don't like it.

Also, was the Fuzion block only performing better because of its base shape? When the Apogee GT had the advantage of that base shape, it outperformed the Fuzion. Does this mean that the Apogee GT is a superior waterblock? What if the Fuzion had a flat base and a comparison was then made? What if the base shape affects block performance in some mysterious way that we have never guessed before? Should more waterblocks be tested with this bowed base? I'm curious to see how a Storm would perform with it.

Finally, if the bowed base is superior because of waterblock contortion, then is it possible to make waterblocks that will not contort? I have always wondered about very large waterblocks with raised bottoms to clear capacitors that allowed for huge amounts of water to flow through the block and cool it down. Sort-of like a Mega-Apogee. Would this huge waterblock design also contain enough copper in the base to resist contortion to a significant degree? Could bases be made thicker to resist contortion? Could we modify waterblock mounting so that more pressure is on the center of the block? The outside? The possibilities are seemingly endless.

In conclusion, we need more testing to figure out why a bowed base does what it does.

EDIT: Here is another possibility: Bowed bases force good contact on the center of the IHS. Maybe good contact on the center and bad contact on the rest of the IHS is better than mediocre contact on the entire thing. Keep in mind that this is just a possibility. I think that we need to brainstorm as many possibilities for why this is happening as we can.

If two surfaces are machined perfectly flat then the adhesion between them is like glue. The more contact area gives the best transfer on flat surfaces. We can not see the die to IHS configuration to determine exactly which shape will overcome the inadequacies in the CPU manufacture.

Even if we obtain a perfectly flat mating between the block and the IHS we are still at the mercy of the die to IHS mating. Couldn't we also get to the point where a convex block could crack the die from too much center pressure? Obviously there is a "sweet spot" when it comes to block fitting. An extra turn of the nut can yield better temperatures so are we reshaping the two surfaces with the additional pressure or just making sure that the two surfaces are as flat as possible?

Quote:

---

Originally Posted by Mr. Peanut

IMO, the problem with watercooling forums is that too many posts are based off of theories that "make sense", but don't have any actual evidence to support them.

---

Hear hear! :clap:

Quote:

---

Originally Posted by Baenwort

So, any chance one of you with the FuZion measured it for TEC use? Its geting hard to find 6002s....

---

Can't you use DangerDen Maze4? Is there anything bad about it?