Check out our water cooling gear!

Water Cooling Project - Part 2

Created by Nick Curtiss, CrazyPC Date: 9/10/00

Description

Well, we can't leave well enough alone here at CrazyPC - meaning I'm determined to squeeze a degree or two more out of my new water cooling system. There were a couple of ideas that I (once again lifted from those overclocking enthusiasts over at overclockers.com) thought would work. I tried new improvements to my system based on the following two assumptions: I was also concerned about the corrosion effects of having a copper water block in my system so I also decided to switch to an aluminum water block.

In-Line Pump

I'm using the EHEIM 1046 in-line water pump in place of my little giant submersible pump. The EHEIM pump operates very quietly, but it is larger than my Little Giant pump. The size wasn't any issue though, since I've got plenty of room in my water case. Using the in-line pump means I have more water in the system, since I've taken the submersible pump out of the reservoir. I hope to get an added benefit in that the in-line pump shouldn't heat the water up as much since any heat generated by a submersible pump would have to go directly into the water. Here are the pics of the assembly for the new pump:

   

Notice, that I tapped into the bottom of the reservoir for the input line into the EHEIM pump. This eliminated any need for me to bleed the lines to get air out and it I didn't have to prime the pump either! I still have my trusty 3/4" plug for easy filling of the reservoir. There is one 3/8" barb that is on top of the reservoir to receive water. On the output from the pump, I've added a drain valve that is shown in the picture. This was done in-case I needed to prime the system, but I didn't need to use it.

Aluminum Water Block

As I mentioned above, I wanted to switch to an aluminum water block so that I didn't get the galvanic response from mixing copper and aluminum. I chose a 1.5"h x 1.25"w x 1/2"thick aluminum plate for this job. Once again my drill was whirring away - I drilled holes into the plate in order to increase surface area and I used a PVC 1-1/2" end cap to enclose the water block. I drilled two holes in the PVC cap for the 3/8" barbs. The holes were just big enough to get a nice snug fit around the screws. I then sealed the barbs using some Household Goop adhesive/sealant (this stuff is awesome). I also used the goop to glue the end cap onto the waterblock. Here's the result:

Here is the cap pre-application of Household Goop. I put a ton around it to seal it tight. "There will be no leaking", I said to my water block.

 
Here we have our aluminum block all shiny after a fresh lapping with the usual assortment of sand papers.

 
Ah there she is all put together

 
I glued a couple of L shaped pieces of plastic to the underside of the waterblock to add stability so that it wouldn't twist and wobble around. Worked pretty good!

 
At last - ready to go!
 

Conclusion  

Before I go on to the results, a note about getting those water blocks to lay flush with the CPU. What a pain! My original copper water block was a true champ and laid nice and flat on my lil'cumine but I had a bitch of a time getting the new block to lay flush. It took me several times to get the right amount of pressure using my spring loaded clip. I sure could use a copper spacer; that would help. The way I noticed that it wasn't completely flat is because my temps were coming out way to high (40C). I used four temperature readings room temp, CPU, System1, and System2. I monitored the temps with HWDoctor and took samples every 5 seconds for 30 seconds and averaged them. Room temperature was taken with a digital thermometer. System2 is a thermistor probe stuck behind the waterblock and up close to the CPU slug. All temps were taken after running Quake3 in a loop for an hour. All tests were run with my p3 500 @ 667MHz, 1.75 volts.

Setup Ambient Temp CPU System Temp (System 1) CPU Sensor (System 2)
Aluminum Block 667MHz using in-line pump 76F 30.9c 26.8c 27.2c
Copper Block 667MHz using submersible pump 74F 32.5c na 29.5c
Aluminum Block 667MHz with submersible pump 74.1F 32.3c 26.9c na
FKP32 Air Cooling with case covers off 76F 36.8c 27c 35c

I'm pretty happy with these results. First off, it looks like I didn't lose any performance by going from the copper block to the aluminum block when I was using the submersible pump. Second, the in-line pump was successful in lowering the temps by about 1.6C according to the CPU temp from HWDoctor or 2.3C if you look at my System2 readings. In any case, look at the water cooling temps compared to the FKP32 - Wow!  

I hope you found these articles useful. -Nick

Article by N.Curtiss, last updated: September 10, 2000.