mATX Q6600 Water Cooled Build

Biffa

Biffa

Biffa

Man of Honour
Joined
30 Oct 2002
Posts
15,849
Location
Surrey
This is a follow on from my experiences with building a Quad Core Q6600 mATX rig over a month ago, the thread for which is here: Q6600 mATX build

To summarise, I had a X-Qpack mATX case with a 939 Opteron setup which is my Vmware box. I decided to swap out mobo, cpu, memory, and cooler for a C2D Quad core setup, and I was pretty happy with the result.

But something bugged me, temperature and having to back off on the overclocks because of it. I was hitting 85 degrees on coretemp on a very hot day :eek:

It wasn't a huge deal, I knew I had the older "hot" B3 version of the Q6600 and if I wanted to drop temps I could just get a G0 version and lose some temperature that way.

But I had this itch.. a watercooling itch, and it needed scratching.

Some history. A few years ago I had a AMD FX51 in a Coolermaster 201 case (I still have the case) and I again wasn't happy with temps that hot chip generated when overclocked, and the noise of the 80mm fans in the case to keep it cool. So I bought a Socket 940 watercooling WaterChill kit. And never installed it. I've still got it in fact and hope that my experience with this build will inspire me to get a new 775 block and install it on something else!

But back to this build. And what inspired me was the launch of the Swiftech H20 120 Compact Kit which looked ideal for my plans, vague and unformed as they were.

So I skrimped and saved, sold bits of kit in MM to fund the purchase of one, waited for them to come in stock somewhere and finally took the plunge.

OK lets get started.

Please Note: if you want to see larger versions of the images please click on them.


Packaging and contents

Box arrived just wrapped in brown paper but after unwapping was none the worse for wear in the post. Here you can see the box Front, side and inside the lift up flap there is overclocking info and reccomendations depending on how many cores you have an what extra blocks you use.


Inside it was well packed and no parts were missing or damaged. As well as the block, tubing, fluid, screws etc there are a few other parts of interest.



The kit fits socket 775 as standard, but comes with a clever adapter for AMD sockets (754, 939, 940, AM2 and Socket F) that seems fairly straightforward to change. Also comes with a 12v-7v 3pin converter for the fan if you need it.

I won't go into too much detail on the kit itself as there is lots of info out there on it. Suffice to say, the base of the block is protected with a plastic cover and has a very good finish, not highly polished but without machine marks. For full technical specs of the kit check the Swiftech site, and unlike me I would suggest studying hard the dimensions diagram
 
Last edited:
Q6600 mATX WaterCooled - Removing the old kit

The build - Removing the old kit

This is what I started with:

V.1 Aspire X-Qpack black windowed mATX case
intel b3 q6600 quad core cpu
gigabyte ga g33m-s2 matx motherboard
ocuk 2gb (2x1gb) ddr2 pc2-8500c5 1066mhz dual channel kit
zalman cnps8700-led cpu cooler

the zalman cooler in particular is pretty much as good as it gets for matx cases of this type, and can hold its own with the big boys in a cooling dust up as well which you can see from various reviews on the net e.g. X-Bit Labs which is why I bought it really.


Image from BCCHardware

I was able to get 3.0Ghz stable in the setup I had. I could get to 3.2Ghz stable but the temperatures were too high for my concience (80+C)

Here is a screenshot of the temperature before I started it was quite cool that night, and I have tried to maintain a stable temperature between tests which for the past few nights I have been able to do, if anything it was actually a couple of degrees cooler on the night I measured temps on the Air cooling:



So that will be my benchmark to measure against for the new setup.

Here is the old kit after removal from the case, using the X-Qpack's handy removeable motherboard tray. As you can see the Zalman is a bit of a monster!



I then removed the Zalman, and then the motherboard, retrieved the Zalman backplate from the back of the motherboard and fitted the Swiftech backplate.



Next came the planning of how I would fit the radiator. Oh dear!
 
Last edited:
Rad planning

The build - Rad planning

There are many variations on how you can fit this kit to your case, the main one from a cooling perspective to decide is which way the fan will blow, for a cooler CPU you would pull air in from outside, but this needs a well ventilated case as it can increase internal ambient temps. For a slightly warmer CPU temp you can push air out through the rad. I wasn't too worried about other components in the case overheating other than the CPU so chose to pull air into the case through the radiator.

In preparation I also removed the heatsinks from the chipsets on the motherboard and replaced the thermal pads with Arctic Silver 5, just to help a bit.

So in my mind when I ordered the kit I planned to do an internal setup with the radiator and fan inside the case, made sense to me, meant I didn't have to try and get the tubing outside the case and would have been quite quick to setup.



How wrong I was. As you may have gathered from my mention earlier, Swiftech have a nice dimensional diagram of the radiator on their site, I didn't check it and of course didn't realise how much larger than the 120mm fan the radiator actually was. There was no chance I would get it inside the case. So I then planned to have the fan inside the case and the rad outside, cutting a couple of small holes in the removeable motherboard tray to poke the radiators pipes through to the inside of the case where I could easily attach the tubing.

Here is the plan, the fan positioned so I could mark up the backplate with the radiator, and the backplate marked up



However the best laid plans etc..
 
Last edited:
The build - Continued

When I tested the location of the radiator, it was too tight a fit to the case, and the radiator fittings were positioned such that it was virtually impossible to fit the tubing over them.

So to give myself a bit more room, I modified my plan one last time and put the fan outside the case between the case and the radiator. I also decided that it would be a good idea to remove the fan grill from the case to improve airflow.

Here is the final plan, and the holes completed, I put some protective sleeving around the edge to protect the tubing when it is fitted to the radiator.



Next I slid the motherboard try into the case to start positioning the tubing and the best position for the connectors on the block to give me the shortest route least likely to kink the tubing. At this point I also fitted the fan and radiator loosely so I could guage the length of the tubing to cut.

So reapply AS5, fit block and route tubing (you can also see where I removed the fan grill.

 
Last edited:
The build - Continued

Here you can see the added space I gained by putting the fan between the radiator and the case, and the tubing connected to the radiator fittings, made things much easier to fit.



Its advised to remove the setup from the case to fill it, but obviously given the tight tolerances of the fitting and the way I had set it up that wasn't going to be possible, so I jury rigged the pump to start with a paper clip but didn't power on the system. Filled the radiator with distilled water and the coolant they provided, cleared the air from the system and topped it up, then left it running for 30 minutes to check for leaks. I know it could be longer, but we are talking two 4 inch lengths of tubing here!

No leaks and I was all set to connect everything back up, put the case sides and top on and put it back into action. Thats when I ran into my next problem.

As you can see from the pictures above the fan fits very close to the top and left side of the chassis. It was actually so close that I couldn't get the case back on. So it was out with the metal cutters and I cut away the case around where the radiator was. It wasn't pretty so I filed it all back nice and smooth. Its hidden by the radiator so didn't have to be a work of art.

Here you can see just after I cut the case back, then after it was filed, and how hidden it is from view

 
Last edited:
Finishing - temps and overclocking

So a bit of an epic journey really, and admittedly much of the errors could have been prevented by a bit of forward planning. But then I probably would have never started it in the first place if I had known what I would need to achieve to complete the project.

The finished result I am happy with. It looks pretty integrated and the machine itself actually doesn't need to look that good because its not out on show all the time. But if it was I would still be happy with the look I achieved. Its not very "Radical" and doesn't have lots of flashy lights. Perhaps my next one!



As for the main reason I started this project. Temperatures. Yes remember them? Well the first thing I did when I fired it up was load it up the same as it was when I took its temperature under air. The result? A whopping 10-15 degrees lower.

So then I spent some time clocking up a bit, pushed the FSB to 355 for an extra 200Mhz giving me 3.2Ghz not an amazing overclock but then I needed to get this build report done! When I have more time I'll update with more temps and other overclocks. Hopefully I'll also look at whats possible with the same rig with a G0 quad. But thats for the future, for now I leave you with some screenshots

3.2Ghz with lower temps:

32Ghz_screeny.jpg


Comparing temps between 3.2Ghz watercooled and 3.0Ghz Air cooled as you can see its doing the job and the internal of the case is a bit warmer because its pulling air across the radiator into the case.

Speedfan:
Temp2 (System) +3C
HD0 +7C
HD1 +2C
Core0 -15C
Core1 -12C
Core2 -13C
Core3 -15C

Coretemp:
Core0 -14C
Core1 -11C
Core2 -13C
Core2 -12C
compare_1.jpg


Conclusion

The entire build took 5 hours. My first time watercooling so I was particularly careful, my first time modding my case as well, so if you take of time for experience, head scratching, mistakes, and doing things multiple times I would think I could do it again in half the time. And if I didn't have to mod anything fitting it to a case that had the room would be around about an hour plus leak testing I reckon.

There have been different reports on this solution around the web, I think that compared to something like an Ultra120 in a large well cooled case with a single or dual core cpu the temperature difference might not be so great. But in this scenario it has worked wonders, can't wait for the AS5 to cure and remeasure in the future. Mind you hope to have a G0 before then.

Is it worth the cash? Again depends on your requirements. For me it did exactly what it said on the tin, so I was not dissapointed. I've now got the watercooling bug, and next stop will be a "proper" setup :)

Thank you for your time and patience.
 
Last edited:
Must say very good work there :), big difference in those temps, does a good job at keeping the cpu cool, Im quite surprised.
How noisey is the kit, might invest in one for my media center.

Rob
 
I've had Coretemp logging for a long time so I will put the data from before the watercooling and after the watercooling into a spreadsheet and generate a graph. Will do that tonight, I also have a screenshot somewhere of the watercooling with exactly the same settings as the aircooling. Will find that too and update the worklog.

As for noise, I have the fan running full speed, haven't had enough time to measure the difference with the fan at 7v or turned down using speedfan. But will do that when I get some time. Other than the fan the pump itself is very quiet and there seems to be little to no vibration on the motherboard.
 
The post is an excellent piece of work - thanks for making the effort.

Looking a bit more promising although if I get one I anticipate mounting it internally with the case fan sandwiched between the case and the radiator - perhaps.

I'd also want to keep the dust out so I'd be pushing the air out - none of which will help the temps no doubt.

Presumably the existing tubing reached to the radiator fitted externally?
 
Last edited:
Marse said:
The post is an excellent piece of work - thanks for making the effort.

Looking a bit more promising although if I get one I anticipate mounting it internally with the case fan sandwiched between the case and the radiator - perhaps.

I'd also want to keep the dust out so I'd be pushing the air out - none of which will help the temps no doubt.

Presumably the existing tubing reached to the radiator fitted externally?
Click to expand...

Radiator positioning is a toss up between increased internal case temps or cooler cpu temps. In to out gives you warmer case temps, cooler CPU (more dust) and out to in is the reverse. They reccomend if you have good case cooling already out to in can give you equivalent results.

I had to actually cut about 5 inches off the tubing :)
 
RJC said:
How noisey is the kit, might invest in one for my media center.
Click to expand...

Biffa said:
As for noise, I have the fan running full speed, haven't had enough time to measure the difference with the fan at 7v or turned down using speedfan. But will do that when I get some time. Other than the fan the pump itself is very quiet and there seems to be little to no vibration on the motherboard.
Click to expand...
 
Thanks for the review Biffa, been following this in the thread in the Overclocking forum (as you know).

This may just have resparked my want for this water cooling setup :)
 
On Swiftech site they 'suggest'

2wbzybd.png


with a Kentsfield – presumably not G0.

Biffa’s figures look better than this, in the low to mid 50’s, although I’m not familiar with all the terminology Swiftech use.

I think Swiftech’s figures are based on internal installation and they anticipate better performance with external installation.
 
Last edited:
I just found the screen grab I did straight after I rebuilt, and the temps are the same at 3.2Ghz as they were at 3.0Ghz.

Now I'm pretty sure that I didn't modify the voltages when going from 3.0 to 3.2Ghz, but these G33 mATX motherboards all show massive Vdroop when pushing the overclocks which is all I can assume is causing the different vcore readings.

So the jury is out, but its possible that the reason the water cooling temps are the same at the two speeds is just because there isn't that much difference, or of course it could be operator error. But I'm not going back to test everything again from scratch :)
 
You must log in or register to reply here.
Back
Top Bottom