Water Block to Die Sim Alignment project.

[jaydee]

Bill showed us THIS (http://www.overclockers.com/articles870/) project JoeC just came up with.

My mission here is to take that example and improve it.

What I have came up with in the last few hours is adding a way to raise the die sim after the block had been aligned. So basically you get it all aligned and then you crank a nut or whatever to a point of predetermined pressure. That way your not trying to align the die sim to the block WHILE trying to achieve that pressure which is what I think JoeC is doing now.

I think I have a rough idea how to do this, but it just isn't clicking all together in my head yet. In fact it might not be possible. I will try and explain it when I get home tonight.

As for now I would like to hear any suggestions on what you think might improve it. Remember it isn't just about pressure, but also (and more importantly) alignment.

One thing I will throw out there now. Would it be better to go with the 4 corner adjustment he has now (The lower one with the black knobs), or would it be better to go from the middle of each side such as a + sign. It almost seems to me it would be easier to go from the middle of each side instead of the 4 corners..

[iggiebee]

Think you can borrow some ideas here (http://www.longwin.com/PRODUCTS/9052.HTM) , which is same concept but commercially made.

[BillA]

ohhh
very interesting site - how did you encounter it ?
Thanks

[jaydee]

I have a plan, but there is no way in hell I will be able to convay it. So as I build it I will post pics as I go along. Nothing may come of it in the end, but I may learn something trying. :)

[Blackeagle]

Originally posted by unregistered
ohhh
very interesting site - how did you encounter it ?
Thanks

Agreed.

And a question arises, who is the target market as no one has heard of this device before?

AMD & Intel could sure make good use of this. As Bill didn't already know of it, it would seem the makers of heat sinks and water blocks have not been aproached by the maker.

Just to look at it and the posted stats this is a answer to accurate pressure & alinement for CPU & water block tests.:drool:

But what does it COST???????:eek: $$$$$$$$$:cry: :evilaugh:

Also makes you recall some drill presses, or Lee's suggestion of a cider press.

[BillA]

Blackeagle
'our little world' does not quite define the limits of "no one"
lol

I suspect that this co is a major supplier of such to the far east;
others here make similar setups, for big bucks
we had an Asian hsf mfgr visiting who remarked on the commonplace existance of $200,000 test benches

but I still wonder how the site was encountered ??

[Blackeagle]

Ahh, your prior post led me to believe the set up was unknown to you. Which did surprise me, as while I figured it would cost like crazy, I thought you'd know about it.

I see now, you did know of such, you were just surprised at the posters knowing of it as well.

Will still be interesting to see how JoeC, pH & JD all do with adding this to the testing rigs they are each working with. Can only lead to better and more accurate testing.

And if the community gets to where there are a number (2-4) testers giving quality results, it'll have to improve the situation we have now.

[UberBlue]

Instead of trying to achieve alingment from the corners, why not use a single ball joint directly under the center of the SIM die. It would be self alingning.

http://www.sc0rian.cjb.net/procooling/uberbluesfolder/mount.jpg

[BillA]

ideas on a source ?

a cv joint with flat hubs ?

[jaydee]

Uber Blue, I had that in mind actually. I was thinking about the kind on the end of a C clamp. Those big C clamps are not to expensive. You can find a big enough one and tear it apart. It is already threaded. I am sure there are other options aswell.

[jaydee]

Here is a pic. Even has a built in handle to turn. :D

[iggiebee]

Originally posted by unregistered
but I still wonder how the site was encountered ??

Okay..okay.. damit. The site's URL found its place in my favorites folder long time ago when the watercooling bug first bit me, and and was googling for information on the subject.

God only knows what search keywords a watercolling noobie is capable of using. ;)

[bigben2k]

No no no...

In order for that to work, you'd have to have the same pivoting head at both the bottom (under the block) and on top of the die. Otherwise, the block might end up being pressed up against the heat die with some lateral force. That lateral force is going to return different results, if there's any curvature (convex) to the block (or god forbid, the die).

Can ya'll picture it, or am I just blabbing away here?

Try it: use that big clamp, with handles on both sides, and press them together: they won't end up perfectly perpendicular to the axis of the screw arm, I guarantee it.

Right idea though... So, silicone tubing?

[jaydee]

Originally posted by bigben2k
No no no...

In order for that to work, you'd have to have the same pivoting head at both the bottom (under the block) and on top of the die. Otherwise, the block might end up being pressed up against the heat die with some lateral force. That lateral force is going to return different results, if there's any curvature (convex) to the block (or god forbid, the die).

Can ya'll picture it, or am I just blabbing away here?

Try it: use that big clamp, with handles on both sides, and press them together: they won't end up perfectly perpendicular to the axis of the screw arm, I guarantee it.

Right idea though... So, silicone tubing?
You don't understand a word I have said about the clamp. I was saying remove the C part of the clamp and utilize the threaded part and the swiveling end on it and incorporate it into the design. Not use the C clamp itself. Most of it would be scraped. Just want the threaded part with the swivel end.

[bigben2k]

LOL! No, I get it all right, I'm saying that it should have a swivel end at the top and bottom, whatever the contraption ends up looking like.


The example of using a double-headed clamp is just an example to prove my point, although in reverse.

[BillA]

if the vertical alignment is attainable,
why not the horizontal also ?

so why then the swivel(s) ?
way too complicated

[8-Ball]

I would suggest that some kind of pneumatic system for applying pressure might be suitable since it shouldn't be that hard to supply a consistent pressure resulting in the same mounting pressure time and again.

8-ball

[jaydee]

Originally posted by bigben2k
LOL! No, I get it all right, I'm saying that it should have a swivel end at the top and bottom, whatever the contraption ends up looking like.


The example of using a double-headed clamp is just an example to prove my point, although in reverse.
I guess I don't see why there would need to be one on both sides. :shrug:

[UberBlue]

With a single ball joint, load would be distributed evenly across the die regaurdless of horizontal alignment. With a heavy/robust construction, lateral forces caused by misalignment would be negated by the inability to move in that direction.

How much lateral force are we talking about? I bet you could center it up to within .1mm. With that and a reasonable horizontal alignment (no need for a Zig-Align thingy) lateral forces would be really tiny.

C-clamp guts are too tiny and would flex to one side due to said lateral forces. A better solution would be to use 1" - 1.5" .acme threaded rod (http://www.mcmaster.com/param/asp/psearch.asp?FAM=acmerodgen&FT_717=29754&FT_136=1669&FT_101=494&session=desc=Acme%20Threaded%20Rods;acmerodgen;717 =29754;136=1669;101=494) (same thread style a C-clamp uses) and round off one end in a lathe. Mill out a hemisphere in a piece of steel, heat it up cherry red, dunk it in water to temper it.

Another advantage to a robust build is repeatability. How many times is this thing going to get cycled? Alot. Do you really want a rig that is going to flex under a 100lbs of force? No. If it is built light, how many cycles is it going to handle before it is seriously jacked up.

As I understand it, what matters in a test bench is a reasonable degree of precision, accuracy, and repeatability. Can repeatability be achieved using screws in the corners without spending bloody ages doing it?

Once things are set up, I bet you could unmount and remount in fifteen minutes using a single ball joint rig, and have repeatability.

[bigben2k]

Originally posted by UberBlue
...
Once things are set up, I bet you could unmount and remount in fifteen minutes using a single ball joint rig, and have repeatability.
I bet: not.

Yes the lateral forces will be minimal, but it's enough to affect the mount, IMO.

Maybe there's a way to use the Zig-Align in another way...

[jaydee]

Good points UberBlue.

Ben you can just make the bottom adjustable with the 4 corner approch JoeC took and get it close. Then the ball joint should take care of the fine tuning by itself I would think. It should automatically move in order to flush mount the block and the die. Could be wrong..

[murray13]

So spring mount the ball joint. That will take care of the lateral force and ensure a parallel mount. With a scale to ensure the total mounting force the springs are just there to take care of lateral forces.

Like this?

Thanks to UberBlue for the origanal pic.

[UberBlue]

Originally posted by bigben2k
I bet: not.

Yes the lateral forces will be minimal, but it's enough to affect the mount, IMO.

Maybe there's a way to use the Zig-Align in another way...

Ben,

Just out of curiousity, how would one go about figuring lateral forces? Is it something like doubling the off axis angle quadruples the lateral force? Would an off axis angle less than 1/10th of a degree, under 100lbs of force, cause any appreciable lateral force?

Quantitatively, how much do lateral forces affect (or effect? I can never remember.) the mount?

What is an acceptible amount of lateral force? Ideally it would be zero, but in a situation where it was unavoidable, what would be the maximum allowable?

With just a set of calipers, would it be possible to achieve an alignment good enough to minimize the effect of lateral forces to a level set in the previous question?

If you or anybody else doesn't know, could you point me in the right direction to find the answers?

Again, just extremely curious.

murray13,

Unless the springs had exactly the same spring rate, wouldn't they themselves impart a rotational torque on the ball joint causing lateral forces?

jaydee116,

A rig like JeoC's is what I was imagining. Only made from 1/2" by X by X steel (long dimension on the load axis) and welded together instead of shelf bracket and plywood, with the ball joint and jacking assembly between the scale and die sim.

You'd be able to park a truck on the thing and it wouldn't flex.

-UberBlue

[murray13]

True.

If money wasn't an object... I would put four load cells, one on each side and adjustments at the corners and adjust for equal pressure.

[bigben2k]

I certainly didn't mean to imply that one should be measuring the lateral force, I'm just trying to point out that it's going to exist: glad you can "see" it! ;)

Now can you picture that a ball "head" at the top and bottom would result in zero lateral force?

I don't see the need for all the heavy steel: the weight could make things more complicated. Then again, one could achieve 100 lbs clamping force with barbells, I suppose...

Lateral forces are going to affect the effective contact area, between the heat die and the block, which is in turn going to affect the TIM joint.

Then again JoeC was able to get some pretty repeatable results, but I'd challenge him to turn the water block around (and in several other orientations), and see if he even gets results in the same range: if he mounted the block in relatively the same position, I'd have no doubt that the "repeatability" would be there..:rolleyes:


8-Ball: I didn't ignore your suggestion, it's actually a good one. We're just trying to sort out how it's all going to align together.


I gotta read that article again: I can't quite see the advantage of it over springs anymore...:shrug:

[bigben2k]

Originally posted by unregistered
if the vertical alignment is attainable,
why not the horizontal also ?

so why then the swivel(s) ?
way too complicated
This is the kind of comment (cryptic) that takes me a fews days to figure out.

Yes, if you align on an X axis, and align on a Y axis, you have a full alignment: no need for a swivel that covers all angles in between.

[RoboTech]

Good ideas... :)

Another option is to borrow design ideas from some of the commercial RCC devices (Remote Center of Compliance). They are used a lot in robotics (end of arm tooling). I thought I had one in my junk box to take a few pics but I can't find it. Here is a link that describes the working principle better than I can explain it anyway...

http://www.pfa-inc.com/rcc_hiw.htm

I have used several of these units in various designs - they are very expensive and have rather limited travel for use in waterblock mounting - but the basic idea seems applicable. They correct both angular and translational misalignment.

jayde116... you can build a poor-man's RCC out of two Aluminum plates and three springs. Co-bore holes in each plate to mount the springs in, evenly spaced around a mounting circle. Once assembled the two plates will be held apart (~1/4" depending on your design) by the springs. The two plates can move relative to each other with six degrees of freedom. The spring size, stiffness and mounting circle diameter control how stiff or "compliant" the device is.

I like murray13's approach of using a ball (or universal/CV joint) on top of a translational (X-Y) table... :)