WATERBLOCK DESIGN (With detailed technical drawings)

Detailed construction drawings to build your own waterblocks. — Petar Lazarevski.

After my last two articles,

Watercooling – Experiences and Tips and

Chipset Watercooling,

were made public on this site, many people have mailed me with requests for details regarding waterblock construction and, especially, exact dimensions and drawings. I have tried to help everybody, but since there were just too many questions, I have asked Joe to put a few words on the site and let everybody know that there is going to be an article, covering every detail about “home-made” waterblock design I could think of, available soon. Of course, he did so, and now I’m just keeping my promise.

There are not going to be too many words here, but there are going to be a lot of drafting and 3D pictures explaining everything one needs to know about making a good quality copper waterblock for CPU cooling. I will assume that you have read Watercooling – Experiences and Tips so I will not repeat the things that were written before.

Covered here, one will find the detailed technical drawings of four types of copper waterblocks and means for secure attachment and fixing against the CPU core. All the work is based on my experience with the latest AMD processors (Duron and T-Birds) and recent Socket 462 (Socket A) based motherboards with four holes surrounding the socket.

If you have some other CPU or motherboard type, just ignore the text and drawings regarding the clamps (fastening means) – you can still make one of the four waterblocks.

The article has two main “parts”, one comprising the text that you are reading now with pictures, and the other “invisible” one that consists of FOUR zipped folders with detailed technical drawings prepared for download. Joe and I had to do everything this way since the 450 pixel resolution is simply not nearly enough to make a normal technical drawing viewable. In each folder there are five subfolders:

  1. WB raw – enclosing one normal 3-view drafting and one or more section views of the bare waterblock, just drilled and threaded.
  2. WB assembled – enclosing one normal 3-view drafting and one or more section views of the assembled waterblock, with nipples and Allen bolts.
  3. Nipple – enclosing one normal 3-view drafting and one section view of the brass nipple that is to be used with the particular waterblock.
  4. Mounting plates – enclosing one normal 3-view drafting and one section view for each of two Plexiglas plates (base and upper) that serve as waterblock fastening means.
  5. Assembly – enclosing one normal 3-view drafting and two section views of the whole assembly: waterblock pressed against the CPU in the socket by means of Plexiglas clamps.

Probably the best waterblock design is of a “labyrinth” type, with the cooled water entering in the center, inlet nipple (right over the CPU core), flowing around and exiting through the outlet nipple at one of the corners. Such design is shown on the picture below.

Pic 1

However, to make such a device you have to use a good milling machine (preferably an NC one) and just have to assemble it out of two parts, the second one being hollowed and threaded for fitting the nipple, as seen on the following picture.

Pic 2

Two parts could be bonded together by means of a number of small bolts with a gasket inserted between, or simply welded. A final assembly looks like this:

Pic 3

Although this kind of waterblock usually works perfectly, it takes an access to expensive machines and a lot of skill to make it. All the designs that follow are much simpler to be carried out. In case you are able to buy nipples, the rest of the work is only cutting, drilling and making threads.

The first of four waterblocks is almost identical to the one I have working in my PC for a couple of months now. It cools down my T-Bird 1.13 running absolutely stable at 1.4 GHz (9.5×147 MHz) with Vcore of 2.1V on ABIT KT7A motherboard.

You can download the detailed drawings
HERE. (1.0 MB)

Here’s how this cross-drilled waterblock should look like after drilling:

Pic 4

I have made nipples myself, since I could not find suitable ones in the stores – but if you are able to, buying will save you a lot of trouble. If you make them yourselves, all the dimensions that you need are in the drawings. You can choose the length that suits you best. I have made a thin pitch metric thread M12x1.25, but M12x1 would be even better. In this case, the vertical hole for the nipple in the waterblock should be 10.92mm. The material is brass and it looks like this:

Pic 5

After screwing in the nipples and Allen bolts, the waterblock is ready for use. The following picture shows how the assembled waterblock should look like:

Pic 6

Just to let you have an idea how the inside of this simple assembly looks like, here is a 3D cross-sectional view of the unit:

Pic 7


Petar Lazarevski – Yugoslavia

There are two ways to attach your waterblock – using the plastic barbs on the socket or four holes in the motherboard. After overtightening of the central bolt on my old brass clip, I have broken the small barb and crushed the core of my old Duron 700, so I have decided to make a safer waterblock attachment device.

The material chosen is Plexiglas, being very easy to machine. I made the base plate, which fits under the motherboard, with the primary purpose to take the load off the motherboard’s PCB, and the upper plate with the threaded hole in the central portion for the tightening bolt. Of course, there are four symmetrical holes on both plates for the bolts protruding from the holes surrounding the socket and two large holes for the nipples. After attachment, everything should look something like this:

Pic 8

When you look under the motherboard attached to the case, the base plate with four bolts should be thin enough not to touch the metal of the casing.

Pic 9

For this purpose, the base plate has V-shaped (chamfered) holes for flat head bolts, as obvious from the picture that follows:

Pic 10

It is important to know that the base plate of the attachment unit is identical for all four waterblocks. I have used plain, flat head M5 bolts, but nylon bolts are highly recommended. The upper plates differ and the one used for this type of waterblock is shown here:

Pic 11

It is obvious that the large holes are for the hoses that are to be attached to the nipples.
Any waterblock having a length larger than 50mm (meaning it has to partially cover the cam box of the socket) has to be machined on the base side (side in contact with the CPU core) in a way not to get in contact with the mentioned cam box. The machined groove should look like this:

Pic 12

Although this was a picture of a second waterblock in the group, it does not matter since the grooves look the same.

This second waterblock has similar outer dimensions as the previous one, but is of a little more complicated design. The major difference is in the position of the nipples – they are not perpendicular, but parallel to the main surface of the unit, thus making it suitable for piggyback HSF mounting.

The detailed technical drawings of this waterblock can be downloaded HERE. (0.8 MB)

The 3D cross section view of the waterblock is on the picture that follows:

Pic 13

The nipples are particularly critical here. I suggest a thin pitch M8 thread to be used here since there is a danger of breaking a brass nipple while screwing it in the block. If you make a 6mm inner hole in the nipple, using a thin pitch threading is almost mandatory. The nipple should look like the one on the picture below:

Pic 14


Petar Lazarevski – Yugoslavia

After assembly you should have a good looking, slim unit similar to this one:

Pic 15

The second difference between this block and the previous one is in the water path. The following picture should give you a good idea about the flow of the water through this waterblock:

Pic 16

The cooled water enters the center (on the picture, upper one) nipple and flows directly to the portion of the block which resides above the CPU core, separates in six ways and then collects back in a single channel which leads to the outlet nipple. Allen bolts shown here are chosen to have the exact length to cover the unwanted holes completely. Of course, If you are not able to purchase the ideal ones, shorter will do fine.
The upper attachment plate is very simple, as obvious from the following picture:

Pic 17

The hole in the middle of the Plexiglas plate is used for the tightening bolt. I have used the plain M8 thread here, but thin pitch could be a better choice.
After attachment of the waterblock and the plates, the whole assembly should look like the one on the picture below:

Pic 18

There are two remarks that have to be made here:

Before you chose to make this type of waterblock, check your motherboard since there is a possibility that the position of the socket does not allow the clearance for the nipples and hoses.

Second, consider the attachment of a piggyback HSF – in this case, you will have to use different type of attachment means. If you have a cooler like the Swiftech 462, the only thing that you are going to need are longer bolts.

The next waterblock design is similar to the previous one, considering the water path inside it. However, there are two major differences – it is smaller (50x50mm instead of 63x50mm) and has nipples mounted perpendicular to the main surface of the unit.

The detailed technical drawings of this waterblock can be downloaded HERE. (1.0 MB)

Here’s how this cross-drilled waterblock should look like after drilling:

Pic 19

This type of waterblock is particularly suitable when the surrounding of the socket on your motherboard is covered with a lot of elements, e.g. capacitors.
The 3D cross section view of the waterblock is on the picture that follows:

Pic 20

The nipples that should be used for this unit are exactly the same like those from the first example. After assembly, the waterblock ready for combat with heat should look like the one on the picture below:

Pic 21


Petar Lazarevski – Yugoslavia

The 3D cross sectional view of the assembled unit is shown here:

Pic 22

The water path is exactly the same like in the previous example. Special care has to be taken with the short Allen bolts near the nipples, since there is so little space for them.
The upper fitting plate is pretty complicated due to the vertically protruding nipples, as shown on the picture that follows.

Pic 23

After mounting on the CPU and attaching the plates, the whole assembly looks similar to the one from the first example:

Pic 24

Finally, the last and probably the best performing waterblock in the group. This is a small 50x50x10mm unit with vertical nipples.

The detailed technical drawings of this waterblock can be downloaded HERE. (0.9 MB)

This waterblock is meant to simulate the “labyrinth” type waterblock mentioned at the beginning of the article, machined only by drilling, though. Here’s how it should look like after drilling the holes:

Pic 25

The following picture could give you an idea of the way the channels are made inside the block:

Pic 26

However, a 3D cross sectional view gives a much better picture of the way the channels are drilled inside:

Pic 27

After screwing in the nipples, being exactly the same as in the first and third example, and Allen bolts, the waterblock is ready for use. The inlet nipple is in the center of the unit. The following picture shows how the assembled waterblock should look like:

Pic 28


Petar Lazarevski – Yugoslavia

On the 3D cross sectional view of the assembled waterblock, it is obvious that the water entering through the inlet nipple in the center makes four 90° turns before leaving through the outlet nipple:

Pic 29

The following picture just shows how much better the waterblock with rounded edges looks like, but I don’t know if it is worth the trouble:

Pic 30

The attachment means are the most complicated of all. First, there is an upper plate with four 5mm and two 13mm holes in it, one being right over the center (inlet) nipple:

Pic 31

Since the inlet nipple is right over the center of the CPU core, it is obvious that there is no place for a tightening bolt. I thought about a couple of ways to solve this, but using a spring looks like the safest solution to me. The spring should have the inner diameter a little larger than the largest diameter of the nipple. This is how it’s position looks like on the side view:

Pic 32

The spring resides between the upper plate and the upper surface of the waterblock. It is not easy to find a suitable spring, since it has to be of a certain diameter and should apply sufficient pressure on the CPU core. If you wonder how high this load should be, here’s the exact data: between 12 and 24 lbs, preferably 16 lbs. Here’s how the spring should look like, through the semi-transparent Plexiglas, residing around the inlet nipple:

Pic 33

Since I have not built this waterblock yet (and I intend to), I can only assume that it would be a good performer. I thought of a little improvement one could make with a mill bit:

Pic 34

Instead of a V-shaped end of the hole for the inlet nipple made with a drill bit, a flat, deeper hole could be made by milling. This way, the distance between the water and the CPU core would be lesser, which could lead to some performance improvement. The following cross section drawing of the whole assembly with this kind of waterblock maybe makes the idea more obvious:

Pic 35

The CPU core is colored red and the water path blue. Since the incoming stream of the cooled water is separated from the CPU core with a copper wall just 1mm thick, I expect that the taking away of the heat produced by the CPU would be better than in the case of the V-shaped hole.

I hope that this pictures and detailed technical drawings (there are 60 of them for download) will be useful for many of people eager to make a watercooling system themselves.

Petar Lazarevski – Yugoslavia