Ultimate block? Theory.

[utabintarbo]

Originally posted by bigben2k
That looks rather nice, and it does address some issues. What size barbs did you have in mind? How exactly is this easier to manufacture? I mean, I always assumed that one starts from a rectangular block of copper...

How do you propose to mount it?

Hey man, I just like making pretty pictures!

Just for a starting point, I used 1/2 OD barbs in and 3/8 OD barbs out.

The bottom can be held in a vise or a jig of some kind (if necessary), and the top can be bolted down to a plate for machining. Would probably start with some round bar stock.

If someone wants to do a prototype, I can supply drawings (jpeg), CAD models (IGES), CNC code (apt or G-code) ..., conditions are that they have to post performance data. PM if interested.

Bob

[V12|V12]

Originally posted by hemophilic
If that's the case, why does the innovatek perform so well then? Even with the ribs for increased surface area, it's still a rather large hunk of copper. In the same vein, why do the best aircoolers have thick baseplates as well?
Also, there would be significantly more flow with the wider channels and 1/2" barbs.

maybe I should have posted this somewhere else since this is a theory thread. I'm going for a producible block, not a theoretical masterpiece which is impossible to construct. Oh well.

Good IDEA! I was thinking about something like that/invotech and actually it should work well. Here at the material science Dept, THICK base plates absorb monsterous amounts of heat and copper has a very large buffer when it comes to thermal-spread resistance limits, so that thick core would take a lot of heat, provide a large buffer between temp spikes (if any). Heat applied to copper travels rather fast, so the core is NOT going to heat up as one would think. The thinner the base plate the more rapidly the heat overload limit of the metal is reached and then the core is affected b/c the heat has no where else to go/be absorbed. The Innvotech with the vains is a great design to shed heat from the large center core... I've got an idea on how yer block could benefit from some drilling, but I don't have time to MS-paint one up, so I'll post it when I get back this weekend... Good simple designs USUALLY work the best. The most complex the path way and shit... that just leads to increased problems and factors which arise from the complexity etc....

Folks need to start thinkings K.I.S.S! Keep It Short & Simple!

[dream caster]

Consider wich would be the best heat scape medium: an indefinitely big copper medium and how would heat travel there: expanding in an spherical shape What does this mean?
that is the shape that gives the shortest average path from a point heat source to a given amount of surface through wich that heat is going away;
so that shape should have the least temperature delta inside copper. I feel that this could be treated mathematically in a not so complicated way and be clarified; as for instance wich is the the best size and exact shape depending on the heat transfer conditions at the boundary of copper.

[dream caster]

and this variation:
take a cube, a very small copper cube over a cpu core and consider distance and surface; a 20mm cube over a 10mm square CPU, you will have a surface 20 times that of core and the longest heat path will be around 20.6mm. I suppose the average heat path will be well below 10mm. This same kind of analysis applies to innovatek shape, (the advantage of 3D heat path being somewhat less in this case but the flow easiness much better).

If i was into try using pelts no doubt i would use 5 pelts over a small copper cube (and if you look at www.tetech.com it is not too much expensive)

[HMB]

Since theres been so much talk about the innovatek V3 waterblock (which I btw happen to own ) i want to know if there are any test between it and other commercial blocks.

[bigben2k]

Originally posted by dream caster
expanding in an spherical shape What does this mean?

I meant that the heat travels in an even radial pattern, from its source. Since the source is the CPU die, then it's not quite a sphere, but you should get the idea.

[BillA]

HMB
here you go



the article on a wb testing methodology is still 'in the works'
(there is also a graph on pressure drops around here someplace)

EDIT: hell, if I can't find it, how could you ?
here it is again



a note on the low flow resistance of the (original) 462 with barbs
- the barbs are modded, not as bought from the hardware store

[bigben2k]

V12V12: where's that drawing?

I see what you mean, about copper having a large buffer area, but I can't help thinking that a larger buffer would have a hot side, and a cold side, and so, if the baseplate thickness is very large, then the core temp would be allowed to get up there. That's what Dream Caster is explaining.

I think that it would be best to minimize the baseplate thickness. The problem with that, is figuring out what it is, and that's the essence of the Ultimate Waterblock design.

Just like a heatercore, it's fairly easy to reach the balance point, with coolant temps around 30 deg C, but a larger heatercore yields little improvement, because as the coolant temp drops, so does the rad efficiency. In other words, it's easy to drop the first few degrees, but much, much harder to drop the rest to ambient.

As for KISS, well, that's where I'm headed, with a center inlet, and a rather large outlet. It's only going to LOOK complicated as I add fins, and a turbulator/spinner.

My latest idea expansion would be to put a motorized turbulator... It would save me the time of designing a turbulator.

[Jessfm]

-If anyoine listens to bill ( ), then the swirll design woula already be in a block design ( Lasers ?)

ATLANTIS

Been around for a couple of months now.
WHy do some ppl under estimate the power of google ?


BTW Thatns Bill for the pointers in the right directions

[bigben2k]

That's it, as far as the copper base goes, but the outlet will be radically different.

Also, this swirl, although I'm sure actually works, contrary to many other suggested designs I've seen, is a little different: I had something faster in mind.

Thanks for the reminder though. I try to stay out of the news, so that I'm not influenced. Nice to know I'm getting there...

[Volenti]

Originally posted by bigben2k


My latest idea expansion would be to put a motorized turbulator... It would save me the time of designing a turbulator.

I'll be investigating this approach over the weekend when I revive my "pump-block" project, I have sourced a smaller pump than the one I was going to use, and it's design favours a far simpler block design than the other one needed, I'll post a thread about it when I finish it.

[Gnozo]

Jessfm

That sh*t is wicked.....really nice work there

[Jessfm]

Originally posted by Gnozo
Jessfm

That sh*t is wicked.....really nice work there

Thanks

we managed to go through 5 stages of design to end up with the current unit.
Sadly the pic showing does not realy do it justice.

[Gnozo]

Originally posted by Jessfm
Sadly the pic showing does not realy do it justice.

Then take some more shots of that beauty

[Jessfm]

Take a look in this thread http://www.coolhardware.co.uk/module...iewtopic&t=593
About haflway down the page

[bigben2k]

You know, come to think of it... this block has a flaw:

The flow will tend to go right to the outlet, while the rest of the block's water will be barely moving. This could have been corrected by restricting the flow gradually, at the exit of each fins, so that the flow evenly exits the swirly fins.

That's why I'm going with the 2 inch exhaust tube.


Volenti: looking forward to it!

[Jessfm]

Ben,

but flow rate v's directional flow.

I will say this, the flow pattern is not equal , but we experimented heavily with differnt amounts open, and checked the results. varied flow rates, even a 1046 was enough for the block.

Hence 5 stages of design before we ended up with final.

Its a high flow/volume, low mass.
the block is only 50x60mm

[bigben2k]

Originally posted by Jessfm
Ben,

but flow rate v's directional flow.

I will say this, the flow pattern is not equal , but we experimented heavily with differnt amounts open, and checked the results. varied flow rates, even a 1046 was enough for the block.

Hence 5 stages of design before we ended up with final.

Its a high flow/volume, low mass.
the block is only 50x60mm

Could you tell us more about the "experimenting"? Details man, details!

If I wasn't experimenting, I think I'd buy that one. Are there any reviews out yet?

[Jessfm]

Originally posted by bigben2k
Could you tell us more about the "experimenting"? Details man, details!

If I wasn't experimenting, I think I'd buy that one. Are there any reviews out yet?

Same old story Im afraid.
not many ppl can test/review it, and I won't release any figures.
I don't ship to USA ( Bad exchange rates) so my market is UK , and a few in EU. So worthless me going the whole hog to have a US site review my product - although Webmedic will include one in his batch test at some point.
Details of testing - lets just say uncle Bill keeps pointing ppl in the right direction , if you take time to read you can learn allot. There will never be such a thing a comercialy available Uber block, just not cost effective, but if you look at the design, you should be able to work out where we started, and ended up.
As you say, closing som of the vane exit points you would think help temps - nope.
Flow is important, the surface area if you look is allot higher then many block designs within the 150mm² around the die.

[bigben2k]

I agree, the surface area in the region above and around the core is important. This design is well thought out.

I'll have to out-fin that design...

[murray13]

Here is a graph of thermal resistance in copper as a function of thickness.

There has been some question of how thick the base of a waterblock should be between the core and the water.

We have to remember that some of what a waterblock does is spread out the heat of the core so that we have more surface area to work with. If the bottom of the wb is thin the advantage of using copper as a heat spreader is diminished.

I can't remember exactly where I got these results, but it was at a university site with a java tool.

Just some food for thought!

[BillA]

more than a few undefined parameters there murray13
rest assured that wbs do not perform as that graph suggests
(read this thread ?? better yet check out Cathar's many threads on wb bp thickness on OCAU)
waterloo the source ?

[bigben2k]

That's interesting info, but we'd need more data to get to the point where we could use it to design a waterblock.

The flow rate is a big factor, as is the fin design.

Also, the way that you present this info, it seems that thicker is better. Maybe you could explain why, or more specifically "How thick is too thick".

I've come to the conclusion that there is a rate of heat conduction for different flow rates, and that given a channel size, there should be a way to figure out what the baseplate thickness should be, based on the amount and pattern of heat that it exposes to the coolant. Balancing these two I believe is the key (along with the optimal fin pattern).

[murray13]

bigben2k: The numbers say that thicker is better at conducting heat away from a point source. That is all, this was not done on Icepak or Flotherm. Oh how I wish I had a copy of one of these pieces of software! It would make this whole process much easier.

I was looking at how thickness affected thermal resistance. Trying to maximise surface area with the highest delta T.

IMHO these two factors, surface area, and delta T, are the two most important factors in waterblock design. Yes I admit that flow rate and a host of other things are important considerations. But consider the other half of the system, the radiator. Surface area rules! Next is the delta T, and then the flow rate of air over the surface. Increase any one of those and you get better results period.

The trick is that we have lots of constrainsts in waterblock design. Trying to maximise results within these limits is the hard part.


By the way, Kudos to all those who have posted here. I am new around here. There is a wealth of knowledge here! Some GREAT work being done. Lets all keep it up and maybe we can design a WB that costs $5 to make, works with 50GPH flow and keeps temps 1-2C above ambiant! I'll wake up now and stop dreaming!

[bigben2k]

Thanks.

As much as we are all aware that the rad is an important component, I'd like to keep that out of the equation, for now. I think that we can concentrate on optimizing the block design without this parameter, although a range of values might help.