Which block should i mill?

[dreamie]

Originally posted by bigben2k
If your trencg is goping to be 1mm deep, I'd go with a 2mm O-ring. 1.6 might be a better fit though, you'll have to try both.

It's also going to depend on the trench width.

hmm.. what trench width do u recommend? I think i will go 1.5mm trench width(will be a few 0.1mm off ), with 2 mm O-ring. what do u think?



Thanks!

[bigben2k]

(yikes, my spelling! Need more coffee! )

Here's an easier way to look at it:
The cross section of the o-ring should exceed the cross section of the trench.

If you're going to mill a trench that is 1mm deep, and x mm wide, then the cross section is 1*x (square mm).

If the o-ring's cross section is 2mm in diameter (radius=1mm), then the cross section is pi*1^2 = 3.14 (square mm). Your channel width would have to be less than 3mm.

Note: that means squishing the o-ring to half of it's thickness.

If the o-ring's cross section is 1.6mm in diameter (radius=0.8mm), then the cross section is pi*0.8^2 = 2.01 (square mm). Your channel width would have to be less than 2mm.

If your milling tolerance vary by as much as 0.3mm, then you'd be better off milling the channel 0.3 mm smaller.

Fixitt and Jaydee can probably tell you more.

[Paul_Vodrazka]

Do not over-squish your O-ring. They are designed to seal with very little squish. My water blocks have a 0.062" (1.575 mm) diameter o-ring. The channel is 0.05" (1.27 mm) deep and 0.08" (2.032 mm) wide. You need to make the channel wider so that the o-ring has somewhere to bulge. The nature of the rubber in the o-ring will keep it sealing. Squishing too much can cause the o-ring to split. It can also bulge out of the channel and become pinched causing it to tear. My blocks have been tested to 130 PSI of air pressure submerged with no leaks, and should seal to at least 500 PSI. 130 PSI is all I had to test with. Of course they will never see pressures anywhere close to that.

I noticed someone suggesting that one should stay away from clear lids because they eventually leak. I think this statement is ignorant. The reason these lids DO leak is because of the poor design. Most of the lids are secured to the copper by 4 screws (one in each corner) and have the 4 mounting holes beyond the copper. This is a recipe for disaster. All the pressure that is applied to the edges of the clear lid will cause it to flex in the middle where it is not secured to copper. This is why they leak. The other problem is people over tightening the lids and cracking the plastic. The lids only need to be tightened enough so that the screws don't back out (except in flawed designs like this one).



As you can see with my design, there is no stress applied to the clear lid like the other designs. Also I have 6 screws equally spaced to ensure a good seal.



This is just my opinion of course, feel free to add your $0.02 if you disagree.

[bigben2k]

Thanks Paul, your design makes more sense now!

[dreamie]

not sure whether what i'm saying is gonna make sense .

I will make the mounting of the plexitop as close to the center as possible.

As for the O-ring, i will ask th machinist to make a 1mm(Height) X 1.5mm(Width) trench.. i will order some 1.8mm O-rings.. what do u think about this planning? If no one objects, i will go into Solidworks and edit every single detail, making a trench for the O-ring and some holes for mounting.

[bigben2k]

Paul suggested (from experience) a channel that is 1.3 by 2.0 for a 1.6 mm o-ring. In terms of cross-section, that's 2.0 mm^2 for the o-ring, and 2.6 for the channel.

What you're proposing (1.8mm o-ring and 1.0 by 1.5 trench) would give 2.5 mm^2 cross section area for the o-ring, and 1.5 sq mm for the trench.

The obvious question is: how are you going to fit a 1.8 o-ring into a 1.5 channel?

I think you should make the channels 1.8 wide, and 1.5 deep. (what do you think, Paul?)

[Paul_Vodrazka]

I agree.

Too much squish is actually worse then not enough. Only 0.002" (0.05 mm) squish is actually adequate for our application. I opted for 0.012" squish because that is what an engineer at my work recommended as a maximum. I wanted the block to seal even under a lot of pressure even if it will never see it. To keep it from leaking under near zero pressure, all you need is for the rubber to be touching both surfaces, but of course a little squish is better for peace of mind.

If you over-stress the rubber, it is more likely to fail.

[dreamie]

Yo Paul and bb2k

Is 4mm mounting screw enuff for mounting? I want to use a mounting screw as small as possible, cos i want the block to be as small as possible.

[bigben2k]

Probably. Which threading?

What's the holding strenght for a 4mm bolt?

[dreamie]

I think my mouting of the lid would be similiar to where those 4 rubber pads on the Athlon XP are, just that they are screws to mount the lid. Since it can prevent a HSF from crushing the core, it can surely mount a lid quite nicely.

[dreamie]

Spent 15 minutes in Solidworks making a new block. Here it is:









Comments?

[utabintarbo]

Originally posted by dreamie
Spent 15 minutes in Solidworks making a new block. Here it is:

Images deleted

Comments?

Eliminate sharp angles - flow killers.

Bob

[dreamie]

Originally posted by utabintarbo
Eliminate sharp angles - flow killers.

Bob

If you paid enough attention to my diagrams, you can see that i placed a wall right after the entrance to promote turbulence and to guide the water down to the other side, rather than making there a dead spot.

BTW, my EHEIM 1250 seems to be pumping water with so much pressure that the water flowing into the reservoir is very strong. So i reckon by using 2 of these waterblocks, maybe I can cool it further and overclock it.

Anyway, anymore suggestions?

[bigben2k]

Well, it's an NB block, so it doesn't really matter now, does it? As long as it cools the chip, it'll be good.

[dreamie]

Originally posted by bigben2k
Well, it's an NB block, so it doesn't really matter now, does it? As long as it cools the chip, it'll be good.

hm.. what do u think of the block? if u give me the green light, i will send this to my machinist straight away

[bigben2k]

To be honest, you could purchase the same design from DangerDen: they call it the Z block.

[dreamie]

Originally posted by bigben2k
To be honest, you could purchase the same design from DangerDen: they call it the Z block.

really?? oh.. :shrug:

but i think i will stick to this, since my other block wasnt a good design. Do you think my design of the O-ring placement is good? or should i use round curves?

[bigben2k]

I'd use round curves.

[dreamie]

i just got an inspiration for a new block.. hope it isn't the same as anyone's else. Will render it out in Solidworks 2morrow after school.. gotta heat up my poor little gfx card again.

[dreamie]

here it is...



how is this one?

[Cathar]

Originally posted by dreamie
here it is...



how is this one?

'tis a pity you could never fit barbs on top of it that close together, let alone barbs with hosing of any substantial wall thickness.

[dreamie]

Originally posted by Cathar
'tis a pity you could never fit barbs on top of it that close together, let alone barbs with hosing of any substantial wall thickness.

i'm gonna drill some fancy slanted designs on the lexan top and fit in my barbs.. if worse comes to the worst, i will use some 90 degrees bending thing to fit in the barbs...

thanks for ur comments anyway

anymore comments?

[bigben2k]

You're stuck with the same barb problem. 90 deg fittings would be restrictive, but I guess it's not relevant, since this is only a chipset block.

45 degree fittings might be your solution here.

[dreamie]

Originally posted by bigben2k
You're stuck with the same barb problem. 90 deg fittings would be restrictive, but I guess it's not relevant, since this is only a chipset block.

45 degree fittings might be your solution here.

i have a new design which is simpler but wouldnt have any barbs inssue.. will show it to you guys by tonight.

[dreamie]

behold the "I" block (aka "O" block)