Custom Waterblock

Man of Honour
Joined
13 Nov 2009
Posts
11,666
Location
Northampton
Hi guys

I'm in the process of designing a custom waterblock as project so I could learn some CAD software.

Firstly does there appear to be any issues to any of you (Pictures in link below)

Secondly where can I find a company that offers CAD/CAM/CNC services based upon Solidworks designs

And Third is there anyone here that uses SolidWorks that could point me in the direction of thermal analysis or a plugin (Free prefereable) that allows theremal analysis of the waterflow through the block.

http://s563.photobucket.com/albums/ss76/phil21191/Custom Waterblock/
 
the restriction on that block would be horrorific.
It's a massively long maze - with a very small channel.

Have a look at the insides of the current top blocks if you are planning on using this.
if just an exercise or piece of coursework - still worth getting it half right.
they use parrallel channels as you need to move 1-2gpm through it.

Yellow pages - give a few local machine shops a ring
can't help with solidworks - forum search may help as can google
 
Last edited:
As above, that sudden aperture change, would bring any pumps flow rate crashing. You need to balance flow rate with surface area.
 
o i would be better with with of my earlier idea using 3mm channels and 2mm walls and running the parallel rather than a maze. My understanding was velocity was important so i have designed the block to increase this, so I actually need to design it for low velocity. Would a pin matrix be better?

Also does anyone have a waterblock to take some measurements for me, I need the overall height of the block and length of the thread on the mounting screws
 
I think the swiftech blocks and the HK3 have good tecnical drawings.
Thermalright did a block - and I'm sure their site has drawings...enzotech is also worth a look
 
Last edited:
At sufficient head pressure that would work. Main design issue is that it isn't optimised for most of the heat load being in the centre of the processor. The pressure required to maintain a sufficient flow rate through this would be mad.

As a quick fix, wrt this image, mill a slot bottom to middle left and another top to middle right. This will leave many, many parallel slots and considerably lower resistance.

Regarding manufacture. First you have to persuade a machinist he wants to do this. If he's you or a good friend this will be viable. If going to a professional company, expect to pay over £500 for this. A single piece cut on cnc is rarely financially viable as the man hours involved are considerable. If you wanted 50 of the things, you'd be looking at 5k or so, cost per unit comes down sharply with cnc as most of the time is spend on set up. So unless you have access to a lab, give up on cnc and accept you'll have to cut this by hand.

On which note. That looks like you're milling channels 1.5mm wide by 4mm deep. That wouldn't work in brass, it definitely wouldn't in copper. With small milling bits, below 4mm or so, the best depth you can hope to cut is about half the diameter of the bit. Obviously this has a negative impact on surface area. The xspc edge was an attempt at doing something like this, they broke so many milling bits that they had to cease production. The block was damned good though. Oh, when the milling bit breaks, it tends to write off the piece of copper as well.

Most blocks today have a copper base with a vaguely circular indent cut out, and then a lot of 1/2mm or shallower slits cut parallel across. That's done with something that looks a bit like a thin grinding disk mounted with the axle horizontal. It'll cut these happily, but doesn't plunge well.

The normal waterblock design calls for a piece of 2 to 5mm copper with some basic machining, and a thick piece of delrin with more intricate shaping as delrin is a hell of a lot easier to machine than copper. The more advanced ones are sandwiches of various layers.

Waterblocks are difficult. I've put a few sketches together and shown them to my local machinist, at which point I discovered milling machines don't work quite like I expected them to. Curves are difficult by hand too. Don't be dissuaded, but getting this manufactured either means you cutting it yourself or you persuading a friend to do it for beer and pizza.

edit: I haven't given up on my plan of making waterblocks, but I have revised it to start with chipset/hard drive/ram blocks first as they're much easier. Specifically they can be considerably lower performance and it doesn't matter. So make these first, then progress to more exciting projects. I think I'm going with 16mm acetal top, milled, with 3mm copper base. Joining them together is hard, solder has its own issues, I haven't worked out how to size O rings or how to cut curves on a mill, so currently planning on using epoxy. Should have these made by the summer.
 
Last edited:
Thanks for the info Jon, looks like I shall have a rethink. I have a couple of friends that work with CNC eqiuptment so getting this machined shouldnt be too much of a problem, I was also very friendly with the design tech department at my secondary school and believe they recently aquired some CNC equiptment so I may have a word and see if they would fancy donating it for a few hours.

As far as design the current one uses 2x5mm channels

I have modified it to use 2x3mm channels and have included your suggestions

modified.jpg


EDIT: I've completely revised my design using 2mm channels in an arrangment to direct the flow from the inlet onto a pin matrix that uses 3mm channels, this arrangement seems to achieve flow over a large surface area of the entire block when compared to a traditional pin matrix or parrallel channels as the flow in staying within the out 2-3 channels from the simulations in solidworks.

flowrev2.jpg
 
Last edited:
Back
Top Bottom