Water block mass...

[DeadEye]

I asked this in an earlier thread but didnt get an answer to it. Does anyone know if the mass of the copper is critical to the design of blocks.
From most of the designs I see here (ignoring TEC blocks), the area of the die + 20% maybe is concentrated on for the maximum cooling effect, to the die.

Obviously some size restrictions are put onto the block by having to get at least one more barb in for outlet, but apart from this is having greater mass of benefit or detriment to cooling. If there are any discussions on this in other threads then please post any links.

Secondly, is having the most active part (over the die), placed offset n the copper any disadvantage, talking AMD cpu here.

Thanks guys.

[myv65]

The only purpose of "mass" is to permit the thermal energy to travel from the CPU to the fluid. That's the simple version. The complexity comes about when you consider that the heat transfer involves getting that energy into the copper, then conducting it through the copper, and finally convecting it to the fluid. Simplifying again, conduction is linear with distance. So your best bet is making the heat travel the shortest distance. For convection, however, heat transfer is linear with area. This means having a lot of contact surface between the fluid and solid.

More practically, the task is to balance conduction efficiency with convection efficiency, which really are competing concerns. Designs that improve convection efficiency, ala Cathar's jets (raise velocity, decrease boundary layer thickness, increase convection efficiency) allow you to concentrate the convective heat transfer into a smaller area. This is a win-win as you don't need to conduct the heat as far.

I guess in summary it isn't so much about the mass as how you use it.

[DeadEye]

Thanks for the explanation, maybe you or others could comment on the following then please.

Thinking here then as a machinist and not in the physics, which I barely understand. Why are blocks then being designed and made the way they currently are, with large slabs of copper ?
For example a simple round copper design of almost no mass with an o-ring seal on circumference could easily be made and utilize a method like the Cascade to cool it efficiently. The bulk of the design could then be plastic based, putting in the ports etc.
Is it purely the makers/engineers in here do not have a lathe or am I missing something in my thinking, before I go off and have a go at a design.

Thanks

Kevin

[myv65]

By using a lathe I assume you are describing a design where flow entered at the center, over the die, and exited at the periphery? "Channels" would be radial emanating from the center rather than linear?

One of the problems with such an approach is that the ratio of channel cross sectional area to channel surface area may be optimized in a "linear" design. This isn't really possible with a "radial" design. eg, the channels in the Whitewater design have the same ratio of fin area to surface area along the entire length. In a radial design, the fins and/or the channels must get wider as you leave the center. This affects flow velocity and the ratio of channel cross section to channel surface area. It would work, but would not work as well as a "good" linear design.

I'm not sure what you're getting at with the comments about plastic, either. Many blocks use copper for the "bottom" and plastic or metal for the "top". The top material is inconsequential for heat transfer. It must be capable of sealing against the bottom, having inlet/outlet ports, accomodate a hold-down device, and not be a dissimilar metal that would cause galvanic corrosion. Lots of blocks offer plastic tops to "show off the internals". The combination of the bottom and top must be structurally sound to deal with the clamping load.

[DeadEye]

Originally Posted by myv65

By using a lathe I assume you are describing a design where flow entered at the center, over the die, and exited at the periphery? "Channels" would be radial emanating from the center rather than linear?

No, I am talking about a small disk of copper for the block, rather like a coin, (silver could be used maybe) incorporating for example the jet design of Cathar.

The plastic as used by most currently here are almost purely a top with the barbs in, what I am suggesting is if you put the block maybe 3/4" (20mm) thick and large enough for the hold downs, into a lathe, it is possible to bore to suit the disk/seal with a small sholder and then bore out (undercut) a reasonable cavity for flow out.
Barbs and jet tube can be added to this.

Am I still crazy to imagine this will not work. I know the seal will work fine, as I am a pro machinist and build remote submarine equipment that goes to many hundred metres of depth, with seals of this nature.

Maybe I could chuck a quick Acad drawing together to show the basic idea.
Thanks for reply.

[Cathar]

It'll work.

The issue seems to be more one of wasted material and time. How much material do you lose in cutting a piece of copper round to the correct thickness? How long does it take to make that cut?

For a one-off block, sure, it's a viable option. For a mass-production scenario, cutting through 2" of copper round and effectively discarding 30-40% of the total copper material due to the cutting, is significantly less desirable than just just cutting up rectangular bar segments.

Now your 2" x 5mm thick copper round will weigh in at around 90g, but wasted about 60g of material to cut it.. The typical 2" x 3" x 5mm copper bar weighs in at 170g but wasted about 10g of material to cut it.

You've saved about 80g on the final weight of the block, and about 30g in material overall, but spent about 5x longer cutting it.

So it'd be a material vs time trade-off, and really, 30g of copper costs next to nothing anyway, and much of your 60g win on the block's weight in terms of copper reduction gets lost again due to the extra plastic required to adequately re-inforce the block against typical clamping pressures due to the metal playing less of a structural role.

Basically, if it floats yer boat, do it, but there's no real particular benefit to doing it with copper round material.

[DeadEye]

Still confused here, this is what I was asking about mass of copper how important is it, when using a design similar to your Cascade. As mentioned in my last reply I am talking about a coin sized block maybe 25mm diameter, not a large diameter.
This thread was way above my head, so was why I asked the question here.

I appreciate the higher costs of round bar compared to flat, and understand why you would never do it for commercial reasons particularly for large diameters, but I am asking from a 1 off point of view as a development exercise and maybe give others here an alternative route.

To summarise will a block that is only 25mm diameter and maybe 5mm thick with the right design be able to work efficiently.

Kevin

[Cathar]

Originally Posted by DeadEye

To summarise will a block that is only 25mm diameter and maybe 5mm thick with the right design be able to work efficiently.

Yes - that'd be bare minimum. That wouldn't cover all of the IHS of various CPU's.

Assuming that you want your circular shape to cover the IHS of Intel P4 CPU's, you'd probably want to be looking at around a 40mm diameter at a minimum.

Q: How are you planning on holding the copper into the plastic surrounds? Threaded and just screw it in with an O-ring around the outside with a lip? Or bolt it into the middle plastic piece with some counter-sunk screws (outside of the O-ring groove of-course)?

[pauldenton]

Originally Posted by Cathar

It'll work.

The issue seems to be more one of wasted material and time. How much material do you lose in cutting a piece of copper round to the correct thickness? How long does it take to make that cut?

For a one-off block, sure, it's a viable option. For a mass-production scenario, cutting through 2" of copper round and effectively discarding 30-40% of the total copper material due to the cutting, is significantly less desirable than just just cutting up rectangular bar segments.

Now your 2" x 5mm thick copper round will weigh in at around 90g, but wasted about 60g of material to cut it.. The typical 2" x 3" x 5mm copper bar weighs in at 170g but wasted about 10g of material to cut it.

You've saved about 80g on the final weight of the block, and about 30g in material overall, but spent about 5x longer cutting it.

So it'd be a material vs time trade-off, and really, 30g of copper costs next to nothing anyway, and much of your 60g win on the block's weight in terms of copper reduction gets lost again due to the extra plastic required to adequately re-inforce the block against typical clamping pressures due to the metal playing less of a structural role.

Basically, if it floats yer boat, do it, but there's no real particular benefit to doing it with copper round material.

presumably it'd potentially be more advantageous with silver?

[Cathar]

Originally Posted by pauldenton

presumably it'd potentially be more advantageous with silver?

Problem is that silver doesn't usually come in round pieces...unless specifically cast into that format by the end user.

I guess you could lathe a square bit into a round, but that's kinda defeating the purpose...

[AngryAlpaca]

I'm not sure if covering the IHS is enough. The RBX probably covered the complete IHS (it doesn't matter near as much with an IHS, as you mount it with a LOT of pressure, and so it can't really move) but it still had mounting issues with the Athlon XP. Why do you want to do it round?

[Groth]

If you have the tools and the skill, sure a waterblock with only a smaller round metal insert will work.

But most of us don't have all those tools and skills. Sandwiched flat rectangular layers are easier to machine, align, and seal.

[DeadEye]

Originally Posted by AngryAlpaca

I'm not sure if covering the IHS is enough. The RBX probably covered the complete IHS (it doesn't matter near as much with an IHS, as you mount it with a LOT of pressure, and so it can't really move) but it still had mounting issues with the Athlon XP.

I was really talking about the Athlon XP range from my original thread though not stated. The issue with the RBX if I remember correctly was it not sitting on the Athlon pads, this is not an issue if the small copper disk is flush with the mounting block as the mounting block will be sitting on the pads.

Why do you want to do it round?

Two fold, firstly I am trying to establish if all the excess copper is really required, or if it helps or hinders cooling.
Secondly, I am a hobbyist like yourself, but mainly the difference is I come from a modelling/model engineer background. From my experience, many more people from this background, tend to have a lathe rather than a milling machine, a Myford or similar is common. Someone having a Bridgeport at home is much less likely.

[DeadEye]

Originally Posted by Groth

If you have the tools and the skill, sure a waterblock with only a smaller round metal insert will work.

But most of us don't have all those tools and skills. Sandwiched flat rectangular layers are easier to machine, align, and seal.

In my workshop at work I have access to every milling machine CNC or otherwise that I could possibly want, right up to a £250,000 Hitachi CNC. Most people do not, have these facilities.

The #Rotor design opened up an alternative to those with a pillar drill and Dremel. I would like to offer a similar to this but using a lathe.

Dont get my wrong, for example I modify my computer cases using a drill, fretsaw and files, not even a Dremel.
I could easily take a lexan panel in and whip around it with a CNC and it would be perfect, but I could hardly say I achieved much in self gratification for the mod.

To give you another point I would willingly put a patent on a design to stop it being commercially made if I thought it would help others in here, but to do this would stop me openly talking about a subject I know very little about, that is why I want help.

The innovation from all you guys I respect 100%.

[Groth]

Originally Posted by DeadEye

I would like to offer a similar to this but using a lathe.

Lookin' forward to seein' it.

[DeadEye]

Here is the basic idea:
A= The small copper block/disk
B= O-ring seal
C= The 'Slab' now mainly plastic
D= water way/cavity, no real shape is shown smoothing of the flow could be improved
E= optional 2nd barb if required

This is only roughly to scale. The disk could be held in place by either of the methods Cathar pointed out, The favoured method would be screw cutting a thread. If this method was used then I would probably redesign the o-ring to be a 'corner style' design.allowing more threads. However having hold down screws has other advantages, in that the actual clamping of the system to the processor increases the force applied.

I hope this gives a better 'view' to the basic idea. Please feel free to tear the idea apart, thats the purpose of the discussion.

[AngryAlpaca]

Now I see the purpose. It sounds like a good one. I think the threading would be the easiest way, and also the most reversible (for whatever reason, such as changing platforms, let's say) I don't think it is necessary that the copper part touches the entire IHS, but it is necessary to cover the entire core.

[Cathar]

I've got one item on my mind which is kinda obvious, and that is "alignment". With a rectangle there's pretty much only two ways it can go, and this makes it nice for blocks that need correct alignment between the plates. A circular insert makes it that much more difficult to ensure that you've got things aligned properly for the design in which this is important.

[Etacovda]

I guess he could just mount the bottom insert until it seats, and have the inlet rotating. As for height, as long as the thread was fine enough, he'd have enough room to work with. Depends on how fine he can get the thread... With a jet plate, id assume thats not jet tubes (obviously) so surely it would be alright?

[AngryAlpaca]

Couldn't he use a flared top, like a screw, so it could only go in a certain amount, and then stop, with it in the correct orientation?

[DeadEye]

Appologies to everyone for disappearing for a few days. My present ISP 'NTL' in UK decided to cut my broadband off AFTER I paid the bill and it has taken them days to get me connected again.
I will try and catch up tonight with any comments or suggestions.

Kevin

[DeadEye]

Originally Posted by AngryAlpaca

Now I see the purpose. It sounds like a good one. I think the threading would be the easiest way, and also the most reversible (for whatever reason, such as changing platforms, let's say) I don't think it is necessary that the copper part touches the entire IHS, but it is necessary to cover the entire core.

Threading it seems attractive, however it will require a lathe that has feeds or a screw cutting attachment, some may not have these facilities.

[DeadEye]

Originally Posted by Cathar

I've got one item on my mind which is kinda obvious, and that is "alignment". With a rectangle there's pretty much only two ways it can go, and this makes it nice for blocks that need correct alignment between the plates. A circular insert makes it that much more difficult to ensure that you've got things aligned properly for the design in which this is important.

Good point, something that was overlooked. It could easily be aligned using a pin somewhere, however we are now getting into the realms of accurately placing holes to do this. This would then require a milling machine and we are back where we started.

Maybe the design will need to be more of a radial design, move away from the original plan. Concentric circles that could have slots sawn or dremelled in ?

Anyone any ideas here?