[Build Log] M3G

I originally had a wolfsdale clocked @ 4.2ghz via convective cooling for about 2 years. After a week of 100+ degree ambient temperatures, it bit the dust. I quickly needed to replace my processor [ went with a yorksfield ] and find a sufficient way to cool in the summer.

This setup is the result of little to no research on liquid cooling and a complete rush for time. Feel free to point out things i did wrong. Input is greatly appreciated.

(no clock on the new chip at the moment, will start testing the LC system's efficiency this week in the heat)

Pretty uninspired case:


Ok, just put this\/ in there /\... shouldnt take too long..... -.-''''


Old system


Might as well throw this thing in too.


Cant believe i had these things in there the whole time.. must have been restricting airflow.


Continuously vacuuming between every step


Removed old cooler


Mounted the new Yorkfield


Thermal paste


New back plate.. old one broke from massive cooler


Mounted the block


Install fans on radiator (2000 rpm xspc fans and rad)


Fittings installed and tube cut to aprox length (was worried about getting crap in the radiator, thin channels n all)


Mounted in the case, old fan removed, 140mm clear casefan added


Installed pump into res (almost forgot the o-ring), the super glue is for the fittings. XSPC's ABS plasic is really soft, ripped the threads out installing the fittings finger tight..


Glue drying on the fittings, i put it on the threads aswell (would teflon tape have worked?)


Testing the pump for leaks and functionality. (didnt know what the supply side was)


Put into case and started filling with coolant


Had to add coolant 4-5 times before there were no longer gaps in the lines


Moved the res down one bay and booted to bios


tons of "microbubbles" at this point




This liquid level appears to be sufficient to avoid cavitating the pump


Door back on


Peripherals attached


Done.


BOOTED! and low idle temps; 2 cores around 35, 2 cores around 40.. is that normal?

Edited by vassskk - 8/4/12 at 9:24pm

Thank you for the input.

Blade:
I will definitely look into cable management. Had hard time fitting all the extra length in the back of the case. The sheath on the PS wires is extremely bulky.

Hanover:
Can definitely rotate the radiator. The tubes are a bit long to avoid pinching but i guess i overdid it lol. Will have to do some research on the effects of excessive tube length. Could cause reduced flow? The superglue probably wasn't the best idea, it does actually eat away at plastic over time; lock-tight next time
Edited by vassskk - 7/30/12 at 6:39am

Had some time this week to play around with the clock. Came up with 4.25ghz @1.6 vCore.
Running benchmarks, Core 0 seems to be substantially hotter than the rest.. Any idea whats going on here? (chip is from ebay for the record)
Could the issue be related to my install in some way?

^ahh makes sense thanks.

I am going to disassemble the system tomorrow and take care the following issues:
thermal paste
wire routing
tube routing/length
loop direction (currenlty cpu->rad->res)
cooling fluid (is not distilled water)

Just found this:

http://www.overclock.net/t/226970/updated-water-cooling-essential-threads

definitely would have been helpful to read before hand.

In order to make adjustments, had to drain the loop. Turned out to be less painful than initially expected.



for the record, purchased coolant smells like formaldehyde



Checked the thermal paste, waaaay too much



cleaned the block



and the cpu



added "pea sized" amount of thermal paste to cpu



mounted block and tightened as much as possible



front fan had red leds and did not match, swapping it out for an unlit one.



redid all of the wiring, was able to consolidate and organize alot



Had this fan laying around, decided to do some custom lighting.



HOW TO: CUSTOM LEDS
Drill some led sized holes.



Insert leds of choice and plan out your wire routing. I chose UV Leds.



once positions are finalized, super glue into place



now the fun part: Soldering. Cut all your intermediary wire to length and strip ends.



I am wiring the leds in paralell; this will maximize current pull, but minimize voltage requirements. All positive wires soldered here



Negative wires done,



Because of the paraelell wiring scheme, only need one resistor for the whole batch. Here i have tapped into ground and supply voltage to this fan. The yellow wire is rpm reference.



Test: 2 leds were duds, should have checked these before....



drilled out the bad leds and soldered 2 new ones in.



Good



Clip all the unnecessary wiring away.



Wrap in electrical tape. BEFORE YOU DO THIS: make sure you wrap each junction individually to prevent shorts.



zip tied it into place behind top 2 bays. should shine on tubing/mobo nicely.



output is pretty substantial, shallow viewing angle on these leds though.



back to liquid cooling. res mounted low to allow airflow to new fan.



rad mounted with ports towards the front, paper towels to prevent drips on mobo.



Tubing cut as short as possible. New loop: Res->Pump->Rad->CPU. (reverse of last time)



"super chill" distilled water from grocery store... - almost believe they intended it for liquid cooling.



Loop filled



probably not the best idea... but i maaay have put old glowstick fluid in the res.




Oddly, reversing the loop to the "correct" direction, and using distilled water actually increased temperatures slightly (same time during benchmark run, same 81 degree ambient temperature). Also, redoing the thermal paste, did not solve the 1-hot core problem =\