Fish's First WC Build - Log

[willardcw4]

Ok, so my nickname is Fish, but I tend to use 'willardcw4' as my forum handle... you'll get used to it.
_____________________________________________________________________________________
07/22/2013

HERE is the link to my previous forum post discussing the components for the build. I'm using a Monsoon res instead of the XSPC res for THIS build - I'll be liquid cooling my current PC (just the CPU) in a couple months and will use the XSPC res. Pump is a D5 Vario.

I also have demciflex fan filters on the 200mm intake on the front of the case, and the rear 140mm fan I'm using as an intake. I'll have more input when I actually get the rig running, but so far I'm REALLY impressed with these fan filters.

I'm going to be doing this build log in several pieces. I've been so busy with work and 'life' that it has taken me a lot longer to get everything together! All the parts are in, just slowly working through the build.


A few pics of the monsoon res all setup (D5 Vario inside) - I have a temperature sensor in the spare inlet port on the back connected to a 90 degree monsoon compression fitting:




Here is a pic of the stripped case with mobo / ram / psu / SSD:


_________________________________________________________________
07/26/2013

So I've spent some more quality time with the build... having some issues with the tubing which I bring up below. Not sure whether I should re-orient the CPU block to make things cleaner / not touch.

Anyway, here are some of the pics of my build in progress. First is the radiator sitting with some warm (distilled) water prior to my several 'radiator dances':



Next is the result of my first dance:



Lot's of particles in there!!! I did the dance a total of about 6 times. The next pic is of my 2nd to last dance (it looked good, but I did one more for good measure!):



Here is a pic of my CPU block about to be installed:



My TIM application (I had some issues getting the block on because I had idiotically installed the mobo in the case prior to installed the CPU block. There was a decent amount of sliding and potentially a corner of the block that lifted off of the CPU die, so if temps are bad then I know why. I may re-do the application due to the tubing concerns I posted later on in the thread anyway.):



Here's a pic of the build while leak testing:



There was a small leak that I go into more detail about later in the thread. For now I'll leave that info there until I make a decision about what to do with the loop, whether or not to re-do the CPU portion to make it cleaner.

I still have it chugging away - the D5 Vario on setting 3 is insanely quiet. Since I have two quick disconnects there is a chance I may bump it up to setting 4 if I'm not getting desirable temps (since there is some pressure loss over the QDs) but hopefully will be fine.

_________________________________________________________________
7/29/2013

Well, I got the PC completely built and up and running. I left the way the tubing is (see post below) for now. It's not perfectly ideal but my temps are what I would consider to be 'awesome' and everything else just fit into place.

Here's a pic with the extra 2 x GT AP-15's I put in the bottom of the case. I bought 6 total, expecting to do push-pull... but my temps looked good under initial stress testing with just push I decided to put them down there. If I get another 780GTX Hydro Copper down the line I'll probably add in another 240mm radiator at the bottom of the case:



Here is a view of the reservoir with the 'fixed' temperature probe. I had issues with it leaking THROUGH the fitting (they didn't do a good job siliconing the wire in the rear of the probe). The fitting is on the 'inlet' side (right) on the bottom port attached to a 90 degree Monsoon G 1/4 fitting. I got some clear GE Silicone II and it did the trick! Fluid temp @ 28C, room temp ~29C at the time of this picture:



Pic of all the cables during 'Project Cable Management'... luckily I was able to tame that beast prior to putting the sides back on!! Note this was before I added the additional 2 AP-15s on the bottom - I am controlling the 5 x AP-15s with my BitFenix controller and the 3 x case fans (front, rear, side) with the built in 3-speed controller):



Pain in the a** but got the CPU Power, somehow, plugged in with the radiator already installed... also (after this pic was taken) installed the D5 pump PWM header to the CPU Fan header on the Mobo (also a pain in the a**!):




Here's a side view with the case put back together:



Top view:



I had to take the Demciflex filter off of the 200mm fan in the front of the case. All of the sudden, after removing the filter and putting it back on, it would constantly be 'sucked' onto the center bearing of the fan, creating a very annoying scraping sound. I adjusted fan RPM, tried putting a spacer in, toyed with it in other ways, but couldn't get it resolved. What bugs me is that it was fine for 2 days and then started acting up. I replaced the Demciflex with the stock NZXT fan filter. I still have the 140mm square on the rear of the case for that intake.

I'll get some benchmarks screenshots soon, but I ran stability testing with AIDA 64 for wrangling in my CPU (also added GPU loading after I tuned my CPU) and HyperPi to get my memory timings down. I used the Unigine Valley Benchmark for GPU stability. Here are my current frequencies and voltages where applicable:

CPU:
Haswell i7 4770K (3.5GHz stock)
4.5 GHz @ 1.25V, peak temp 85C with ambient 26C on AIDA 64 after 25 minutes

Memory:
16GB (2x8GB) Corsair Vengeance Pro 1866 MHz (9-10-9-27 2T stock)
1866 MHz, 9-9-8-25 1T

GPU:
EVGA 780GTX Hydro Copper (Core 980MHz / Boost 1033MHz, Memory 3004MHz stock)
Using EVGA Precision X software
+102% Power Target (NOT linked to temp target)
+65 MHz GPU Clock offset
+130 MHz Mem Clock offset
Variable Voltage
Ultimate frequencies under load (Unigine Benchmark):
Core 1385MHz, Mem 3134 MHz --- 36C with ambient 26C


-------
Benchmarks / screenshots to come!

 

[willardcw4]

I ran a few benchmarks earlier with CPU temp monitoring going (Core Temp)... here are my GPU settings and memory timings:


The AS SSD Benchmark:


ATTO Benchmark:


Unigine Valley 1.0 Benchmark:


Unigine Heaven 4.0 Benchmark:


I'll get a 3dMark in at some point and some other benchmarks! My 2 x Dell U2711 27" Ultrasharps arrive this week so I'll likely re-do all of the graphics benchmarks with 2560 x 1440 resolution

 

[willardcw4]

Loop Rebuild

08/20/2013

With the concern my CPU mount was bad and expectation I'll be throwing in another 780GTX Hydro Copper in SLI (probably December time frame), I went ahead and drained my loop, reseated the Raystorm CPU block (with aluminum bracket & GC Extreme), and added a RX240mm rad in the bottom of my case in push. The following are some pics from the re-build with some comments from what I learned along the way.

First things first... how was my old TIM application? I was experiencing temps around 85C under AIDA 64 stress tests - seems high but it's Haswell. I also did some research prior to this re-build and found out that for both Ivy bridge and Haswell the paste in between the die and the lid of the CPU is basically crap for thermal conductivity. Many have used de-lidding with their Haswells and have seen drops of up to 18C in their temps with decent TIM. I don't plan on voiding my warranty by doing this, but after talking with a friend who works at Intel I discovered they actually changed their paste between ivy bridge and haswell CPUs, but it still sucks. I was also told every Intel CPU has part of a crushed engineers' soul inside...




Thoughts on the TIM application? The acryllic bracket on my raystorm wasn't permanently bowed when I removed it, but know it's a common problem with this CPU block.

Here is something interesting... I found this goodness inside my CPU block after I took the compression fittings out:


Lesson learned: after screwing in compression fittings, remove them and clean out any material that was removed from the block in the process. The block I used for the remount is actually a brand new Raystorm (I have two since working on my second build... using the original on the new build). So when I made ready my new Raystorm I screwed in my compression fittings and then removed them. I promptly cleaned out the scraps of plastic.

I used the two step Arcti-Clean product and, as stated a million times, it's simple and works. I decided to use them instead of just isopropyl in order to keep them in my 'bag of tricks':


Here is my TIM application of GC Extreme (it's messy and I'm a bit embarrassed, but oh well - decided to go with it).


I used the aluminum bracket for the Raystorm this time to eliminate any potential 'bending' issues that are commonly reported. Also flipped the block 180 degrees from my previous setup to make for a more 'logical' loop config:


Here is the RX240 rad at the bottom of the case:


A pic of the entire loop during leak testing:


Here is the order of the loop --
RES > RX240 rad (push) > GPU > EX360 rad (push) > CPU > RES

All fans are GT AP-15s at max rpm (noise no concern). I've been playing games for several hours (Payday 2 & Dota 2) and have noticed a couple differences in temps from my previous build.

First - my CPU temps are lower. No higher than 62C for the CPU with an ambient of 27C playing Payday 2 at 2560 x 1440 resolution with the clocks in my sig. My CPU peak temps after about 20 minutes of AIDA 64 hits about 80C where it used to hit 85C. I know the remount was a contributor, but not so convinced the extra raddage is contributing - here's why..

My GPU temps are higher by about 2-3C. Payday 2 temps around 42C. Unfortunately I don't have temps before the rebuild to compare to, but highest I would ever see in Dota 2 is 39C with same ambient. I'll play some more Dota 2 for an accurate comparison (Dota 2 isn't as graphically demanding as Payday 2).

So ultimately my CPU re-mount was a success. Still not sure on the extra raddage - but I would imagine it shouldn't make things worse. I'm moving more hot air through the case since I have the bottom rad in Push and am exhausting the rest of the air through the top rad (also push), but I'm not even close to needing this much raddage in the first place. I figure I'd notice a difference between the two setups if I actually had my other GPU installed for SLI.

Based on the feedback I get for my old thermal paste mount I may end up remounting this current block after a month or two. Slowly learning. Please comment / critique my 2nd application for the rebuild with the GC Extreme. I know a bit messy but at least showing an improvement with no burn-in.

 

[Conumdrum]

Veddy pretty res! Veddy! Looking forward to it.

Folks, I met this guy in RL. Smart, pleasant and I expect he'll build a nice rig the first time.

Notice he put a quarter down for size comparison....... Few do that. Or he has holy pockets.

 

[Ajay57]

SUB i am in to watch, good luck! AJ.

 

[EarthDog]

Excellent! I would love to see it!!!

 

[willardcw4]

Veddy pretty res! Veddy! Looking forward to it.

Folks, I met this guy in RL. Smart, pleasant and I expect he'll build a nice rig the first time.

Notice he put a quarter down for size comparison....... Few do that. Or he has holy pockets.

Thanks Condumdrum! ... and quarters should be necessary for every pic! My pockets are nice and sewn

Excellent! I would love to see it!!!

I'll adjust how the images are displayed in the post later today when I'm on a better computer. Thanks for the info!!!

 

[GTXJackBauer]

Welcome to OCF!

Sub'd to see your nice project take its course and along with it enjoy the ride. Can't wait to see what else is in store. Man that reservoir looks really great. I want one!

 

[willardcw4]

Soo.. question - with my XSPC Raystorm CPU Block... one of the ports has an "In' label on it. Does the flow HAVE to go in through this port? Is this block 1 direction only? Or can I bring the flow in from the other side of the block and it will work just fine??

 

[EarthDog]

The block is made to accept IN, in that port only if it says "IN". I would imagine it would still work, however, performance will likely be not as good (best case) or possibly terrible (worst case).

 

[Conumdrum]

Temps will be worse. It will decrease impingement, it's meant to be just one way.

 

[willardcw4]

Here is my dilemma... I set this up a little wonky - so basically my tubing isn't as 'elegant' as I'd like it, and the input and output lines from my CPU block are touching. Also, the line from my RES > RAD is touching the out line from the CPU into the GPU. Pic below:



The loop has been running for 8 hours - after my initial few minutes I noticed the Monsoon G1/4 temperature sensor was leaking water. The connection was tight in the 90 degree compression fitting, but the water was literally leaking from the temperature probe, THROUGH the connector, and down the cable used to connect the probe to a monitoring device.

Anyway, I quick-disconnected the reservoir (SO AWESOME) and took it out... capped the 90 degree compression fitting, put it back in, and let it run. Most of the air bubbles are out.

Here is a pic right after I got the pump running prior to discovering that first leak:



So my option is to take out the res (don't have to drain it), drain the rest of the loop, flip the CPU block 180 degrees (which requires I reapply the paste and block, kinda a pain since it's already in the computer) and then have the IN on the left side of the block, closer to where the radiator out is located. This would get rid of the inlet and outlet tubing touching each other. The long piece of tubing going from the RES > RAD I can probably strap to a side panel or something so it's not drooping near the GPU line.

Thought? Worth the effort?

 

[wagex]

the attachements are invalid. press the down arrow by the paperclip and insert the images that way.

 

[willardcw4]

the attachements are invalid. press the down arrow by the paperclip and insert the images that way.

Can you still not see them??? That's how I inserted them... but I swapped them out about a minute after I posted so maybe you caught a glimpse of no image attached?

I still have the loop running and occasionally move the case around to free up some air bubbles (and tap on some of the compression fittings and the rad). I don't think tubing touching is a 'bad' thing, so I might press on with the build and start connecting all my cables after I top off the res. Or I still might re-do the CPU block and flip it around for a cleaner setup... will be a huge pain with everything installed though. Leaning towards just pressing forward

 

[willardcw4]

So another issue... installing windows and everytime the PC posts it says no CPU fan detected... Probably a way to disable it in BIOS - is that the correct way to do it?? Or is there a header I missed with a jumper..

EDIT: Disabled the 'need to press key' on system start-up error. So I see the error when it Posts but it continues going.. so that one is resolved

 

[wagex]

yea theres an option in the bios that ignores the cpu fan error on most board or do like you did and continue on all errors

 

[Supertrucker]

If your pump supports rpm monitoring, and it looks like it does by the little blue wire next to the black and yellow, then you want that plugged into the cpu mb header. That way you get that warning if your pump doesn't start with the computer. Failing that, yes it can be disabled on most boards.

 

[Conumdrum]

Disable fan CPU monitoring, set an aggressive auto shutdown temp for the CPU. Or figure out if the pump has a RPM output for a 4 pin connector on the mobo.

 

[TerranBrackiatt]

For looks, personally, i'd swap the hoses at the rad, so the rad-cpu hose has less bending, and then route the rad-res hose under/behind the gpu hoses. And leave the cpu block alone.

 

[willardcw4]

For looks, personally, i'd swap the hoses at the rad, so the rad-cpu hose has less bending, and then route the rad-res hose under/behind the gpu hoses. And leave the cpu block alone.

I'll probably do that on my next drain. It should be cleaner - although the bend from the Rad to the CPU isn't as bad as it 'looks' as far as torsion on the tubing.

I added some more pics & clocks / voltages where applicable.

Benchmarks to follow!