Best #Rotor block design corner

[g.l.amour]

I hope this thread will turn into something worthwile. I've been brewing on this for some time. the rotor method has been around for as long as I can remember, as a sort of open source variant of WB design. It can be done with the most affordable means. It is my opinion that this design can be made as performant as any expensive (CNC) milled WB. We are limited by the diameters of our drill holes and our dremels only ;-)

The purpose of creating this thread is to create a 'corner' where we can post WB blueprints and eventual results. This way we can hopefully learn from eachothers mistakes and from them maybe even refine an 'uber'rotor block.

I have learned alot from #rotor , i hope i can return something this way and maybe some ideas will pop up that are usefull for everyone using this method.

[g.l.amour]

this is the first evolution of a block i am going to make. the first was very high flow.

-this one has been designed to have high flow resistance at a 1" diameter around the inlet point.
-at the outlets i have tried my best to keep things low flow resistance.
-i have kept the atlantis block in mind when designing this. it wouldn't show off the drawing, but atlantis is 2 barb, this is 3 barb so the spiral doesn't have to point one way but two way. (i haven't forgotten , BillA)

i hope the drawings make some sense. anyone who wants to have the source DWG file, pls ask.

the baseplate thickness isn't too important in the drawing, i will try make it as thin as possible around the die area.

[g.l.amour]

drawing bottom half

[g.l.amour]

drawing top half

[Puzzdre]

I don't have the drawings, but have the finished block. Its also made mostly with drill press, drilling and cutting, but this mk2 involved cutting with one 2 mm mill bit for wider channels than cutting with dremel discs (approx 1 mm channels).

The block is drilled only on the bottom side, the top part is flat, only barbs.

The barbs are 10 mm inlet ID, 12 mm ID dual outlet.

The bottom part is made of 10 mm thick chunk of copper, drilled with 3.3 mm drill bit, with variable depth in every circle of the holes, starting in center:

center of the block is 3 mm base, first circle is 4.5 mm, second 3.5 mm, third 2.5 mm and last (outer circle) is 2 mm base. Toward the outlets I raised the base thickness for 1.5 mm, from 2 mm to 3.5 mm.

The far out circle is cutted between the holes with 2 mm mill bit, also the perpendicular line dividing the block into two halves, through the center is also cutted with mill bit to 2 mm wide, trying to equalise the flow through smaller radial channels (1 mm wide) and the center of the block...

The block is sealed with four Allen bolts and bathroom silicone, bottom sanded to 1000 grit.

This is the bottom half of the block:

[Puzzdre]

Here are the results for this block and my current system:

25 C air on the rad intake, idle 32 C and Cpu burn load for one hour, 38 C full load...

some temps on my setup:

fans @ 7V:

idle cpu 33 C, water 30.0 C, rad air intake 25.3 C
full load cpu 38 C, water 30.8 C, rad air intake 25.6 C

fans @ 12 V

idle cpu 32 C, water 28 C, rad air intake 24.8 C
full load cpu 36 C, water 28.8, rad air intake 24.8 C

and, after cooling the room a bit, to the 22 C, idle temp (@7 V fans!):

[Puzzdre]

and finished block:

[Arcturius]

Why is your inlet so much smaller than the outlets?

[g.l.amour]

arcturius; it doesn't sound logical. but i'm restating what I have learned from BillA.

puzzdre's mark2 and my rev2 correct some errors we both made in our first design.

1) you are not after the block with the smalles pressure drop. if your block restricts flow less than another, it doesn't automatically mean you have a better performing block.

2) the best is to have a lot of flow resistance(= high flow velocity = low flowrate) at the inlet, so the block is very turbulant at the die area. it will restrict the flowrate some more, but at the same time it will extract much more heat. (watch how puzzdre's water temp is relatively high, and his die temperature is barely higher than the water = means his block is extracting heat like a champ). hence the 3/8" at intake.

3) what we can achieve as home block makers is to make a block SPECIFICALLY made for pelt cooling or die area cooling. this should be our weapon to make our relatively amateuristicly crafted blocks , better performing than commercially available blocks that have to have best of both worlds. the 2 designs above can be considered die area, since they aim all convection at the die(chip) area.

i like additional questions, but lets try our best to keep this thread slightly on topic so future readers don't give up because of the sidejumps.

designs and comments?

[bigben2k]

G.L.amour: speaking of sidejumps... why would you want to direct the flow through the outer holes? As far as I know, it's quite pointless to try to get a high velocity flow that's so far away from the CPU core (if AMD Athlon). I've limited my fin pattern not to exceed the core area by more than 2 mm, following Cathar's findings.

It also looks like you've got many dead spots, where flow will be next to nothing.

[Puzzdre]

@arcturius: I tried to reverse the ID's of the barbs, 12 mm ID inlet, two 10 mm ID outlets and was idling on 1 C higher...

g.l.amour: nicely said!!! Thx!

[g.l.amour]

thk u bigben for commenting. your kind words are not a sidejump as they specifically serve the purpose of laying some weaknesses of my +/- n00b design out on the table. thk u for addressing those, i will make new drawings where there is one horizontal row less at north and south of inlet, so water doesn't travel as far out doing practically nothing. its like being a good gitarist doesn't necessitate the use of all notes in a solo. i make the error of wanting to use too much of my copper ;-)

edit: tomorrow i'm gonna sift out the whole radius thread to see if there are some more usefull ideas to be picked up.

[Puzzdre]

g.l., I know it's only preliminary drawing, but you might forget it later, will you move the bolt holes a little bit further from the mounting holes?

[g.l.amour]

@ bigben ; next drawing should incorporate some of your thinking ; i can however, sence your next comment (i think )

@puzzdre ; thx for reminding me, next drawing ...

[g.l.amour]

oops, there it is

[bigben2k]

Red marks a deadspot outside the core
Green marks an unecessary restriction
Blue marks a deadspot within the core area (very bad!)

[g.l.amour]

-so the green canals need widening
-the red ones won't do too much good, but less restrictions is still less restriction. so thats not too harmfull
-the blue has got me puzzled, my brain can't think of an immediate way to solve that one. ideas?

[bigben2k]

Just close it.

[Puzzdre]

I would add that row of holes north and south of the center, like in your first drawing. Than, open the squared area in the center (4 x 4 holes, designated with yellow square in bb2k's pic) for ALL directions, like the original #Rotor design is, and then try to open more outer channels for less restriction and water drainage. You can also try to do the center square with smaller drill bit and one or two more rows of holes in same area, get the thicker pattern with more posts and still lots of turbulance, than go with larger drill bit and open toward the outlets...

[Albigger]

This may sound somewhat off topic, but it pertains explicitly to this block, so I'm wondering what some thoughts are on this:

Concerning the drill tip used to do this block, I am considering adding half-sphere shaped dimples (with a ball mill) into the base of my next block. Also, by sharpening the drill (for home-made drilled rotor designs) to a very flat cone vs. a very sharp cone, different geometrically shaped "dimples" are created in the base.

Thoughts on sharpening the drill or using a ball mill to maximize surface area/heat transfer/turbulence and also minimize dead spots (water sitting in the dimples and not moving)?

Or does this even matter at all?




Other comments - I agree with bigben, just leave copper where he marked the blue dead spot.

[nicozeg]

Mi suggestions:

1- Green: elliminate that, they only contribute to dremel wheel spending.

2- Blue: widening that channel you facilitate flow spreading in the four rows.

3- White: Keep the center pin, maybe shape it like pyramid

[g.l.amour]

ok, tonight, back to the drawing tables.

thx puzzdre, nicozeg, and bigben

nicozeg; nice comments, the pyramid like pin i have thought about also. i am wondering if it would perform better than the swiftech like valley under the inlet. i was planning on taking the whole marker and make numerous pits so the inlet area has pitted surface allover. if the center pit could be made sharp enough with a dremel it would be hard to imagine it not to perform very good.

[#Rotor]

"It is my opinion that this design can be made as performant as any expensive (CNC) milled WB"



think about it this way.... the fact that a CNC is being used to manufacture a block, does in no way mean that block will outperform others not made by CNC. As more tests will be conducted on these designs, it will become evident that the secret to performance lies within the Design itself, not the manufacturing method.

though True that I'm highly envious of those with CNC...

but in no way is CNC required for the best performance available.

now onto some designs...

Some tips....

think molecular.......
if you where a water-molecule, try and make life as hard as possible for yourself to get through the grid, while staying as far away from the surface of the block, also refrain from going with long and winding roads.... have the water into and out-of the block as quick as possible.

don't think volume, think velocity. (as mentioned)

[Volenti]

I'd like to add another vote for high volecity over high volume water flow.

While I've effectively given up on copper water block designs (in favour of direct die) the results I've gathered from the changes in intake nozzles (low flow+high volecity being significantly better than high flow+low volecity) are just as valid in a copper water blocks as they are in direct die.( I even made a high restriction block, a conceptional clone of cathar's block, before I went to direct die, that easily beat all my previous blocks)

In other words, make the water work to get through the critical heat zone above the core, and then let it get the hell out of the block with as little further restriction as possible.

[dax]

just an old drawing i have lying around here... maybe i should work on it a bit more? It isn't refined at all, but maybe it's an interesting idea, if i find the time i'll try to build one



it uses that jet thinggy just like cathar's block, that would help to spread the water over the different rows more easily imo... (this is a small block though, inteded to be 50mm by 50mm, no idea on the height)