• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

FEATURED Vertical Benching Station

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

sadeter

Registered
Joined
Apr 11, 2013
Location
Atlanta, GA
I think if I was going to have a bench, I would want it flat. But this is awesome! I really love some of your design elements. What do you use to bend the acrylic like that?
 
OP
Navig

Navig

Senior Case Master
Joined
Dec 7, 2003
Nice Ajay, saw that on the Leet builds mega thread.


To do list for my system:

1) Digital control of pumps
2) SLI baby
3) Fluids and tube changes
4) De-lid 3770k
 
OP
Navig

Navig

Senior Case Master
Joined
Dec 7, 2003
Update Fall 2013



At the the time I started my updates in July, here was what my system looked like:


Current%2Bstate.jpg





Update #1

Added hard drives. My current hard drive list is now:

OS drive: Samsung 840 500gb SSD
Storage: WD Black 1 TB (drive for backups), WD Black 1 TB, WD black 2 TB, WD black 2 TB.

My 2 optical drives, Blu-ray burner and DVD burner, are now connected to e-SATA add-on PCI card.






Update #2

Digital control over pumps.


I decided I wanted digital control of my pumps. The reason is fairly simple--get the system quieter. With voltage control of my pumps via my Lamptron Touch controller, the slowest I can control my pumps to is about 2900rpms. Below this, and the pumps may not start up.





To take digital control of my pumps, first I had to cull out the cabling from the pumps:


Digital%2Bpumps1.jpg






Then I had to build a custom cable because I still wanted to monitor my rpms from my Lamptron controller, but I wanted digital PWM cables to run to my motherboard (Asus Maximus V Extreme):

Digital%2Bpumps2.jpg







And here is the end result:


Digital%2Bpumps3.jpg





After tuning AI suite, at idle the pumps run at a silent 1575rpms. At full load (my system is currently running 4.6ghz at 1.3v) the pump will cycle to full speed (about 3900rpms), putting full load at roughly 70c.

The downside to digital control of my pumps is that the Maximus V only has ONE digital-pwm controlled channel (which to me is utter laziness). Therefore my GPU-dedicated pump is slaved to the cpu. So when my cpu moves to 100% usage, I've got 2 pumps running at 3900rpms, which is loud. Fortunately, its rare that I push those kinds of temps.

Ideally, Asus should have provided multiple PWM controlled ports (I know the motherboard has multiple 4 pin fan plugs, but they are all actually analog voltage controlled. The only digital-PWM controlled port is the CPU_fan and its slave CPU_Opt). If had another PWM contoller, I'd stick my GPU pump at 1600rpms (even under 100% load 2xSLI GTX680s only put their max temps to 45c, more on that later....)
 
Last edited:
OP
Navig

Navig

Senior Case Master
Joined
Dec 7, 2003
Update #3 -- Let’s go SLI!





Drained and free’d up my GPU loop:


Bleed%2Bgpu%2Bloop.jpg









GPU%2Bdisassemble.jpg
















Added the second GTX 680:


GPU%2Bsecond%2Bcard.jpg













Added the EK SLI bridge:


GPU%2Bbridge.jpg












Ran into an issue with fittings and the EK bridge:


GPU%2Bbridge%2Bissue.jpg


It’s not really designed to have both lines coming from this direction, so this second port is actually countersunk (for a cap).

So it won’t fit a compression fitting.













Not a big deal to me, went with a standard danger den fitting, with standard clamps.


GPU%2Binstalled.jpg



Also going with a trial of Mayhem’s Pastel white as fluid.









Before SLI GTX 680:


Initial%2B3dmark%2Bscore.jpg








After SLI:


After%2B3dmark%2Bscore%2B4.6.jpg













After some furry testing:


GPU%2Bheat%2Brun.jpg


After 10 minutes of load, the GPUs are 46c, the pump is a MCP35-x running at 1600 rpms.

The rest of the equipment is: Swiftech Maelstrom reservoir, 3x120 alphacool monsta radiator, 3 38mm yate loons in push, 3 25mm yate loons in pull, all fans running 800rpms, custom radiator shrouds, ¾” OD x ½” ID tubing.

With this equipment and settings, my gpu loop is pretty much silent.
 
Last edited:

Ajay57

Member
Joined
Aug 19, 2012
Very nice up grade, should be OK with gaming on multi-screens and a nice score with 3DMark 11. I did this to help get rid of the Air in the Vertical Rads and it works great, as all the traped air is now gone!!
P1000647.JPG

P1000648.JPG

P1000649.JPG

Regards,

AJ.
 
Last edited:
OP
Navig

Navig

Senior Case Master
Joined
Dec 7, 2003
De-Lidding my 3770k





Well then let’s go to it!

Instructions followed here:

Razor-less Delidding

http://www.overclock.net/t/1376206/how-to-delid-your-ivy-bridge-cpu-with-out-a-razor-blade






Here’s my cpu socket area as it was prior to my current tinkering.


CPU%2Binitial.jpg












Taking down the cpu, you can see pretty good contact, with my choice of TIM, Phobya Hegrease.


CPU%2Bopen.jpg














Here is the cpu.


3770%2Bspecs.jpg













This was the rig I used for my hammer-delidding.


Rig.jpg



Secured standard benchtop vise. I’ve got some tape to soften the jaws. Lots of foam padding in case the core goes flying, as well as a tape tether.

Hammer, piece of 2x4.












Mount the cpu up:



CPU%2Bon%2Bvice%2Bcloseup.jpg










And here you go: Video of me razorless-delidding my i7 3770k.




It was really as easy as you see in the video. If you look carefully, I did have one extra hit after the IHS had already separated. That gives you some idea of the force I was using (not much). No flying cores, which was good, but having both padding and a tape tether is probably a good idea (like belt and suspenders, eh?).
 
Last edited:
OP
Navig

Navig

Senior Case Master
Joined
Dec 7, 2003
Here is the core newly separated from the IHS:


IHS%2Boff.jpg










After some cleaning with a credit card, then isopropyl with a microfiber cloth.


Core%2Bis%2Bclean.jpg ]













Next as recommended on the linked site above, I used Coollabs Liquid Pro as my TIM between the core and the IHS.

This is the first time I’ve used this TIM, so I thought I’d share my experiences.

I injected a dollop onto the core, trying to put as little as possible. Made a bit of a mess (see a couple of tiny blobs on the PCB and the paper towel), but no big deal.


Dallop%2Bof%2BCOLpro.jpg








With this specific sized dollop I began spreading the liquid metal with a cheap paintbrush. It spread thinly without any difficulty.


Here is a video of me working it:








After about 30 seconds of spreading, I was able to generate a very thin even surface:


COLpro%2Bspread.jpg













Next I reassembled the cpu to the IHS. They still stick together a little bit, enough to move to my motherboard without difficulties.

If you look closely, I did create some jaw-bites on the edge of the IHS. Not really concerning to me, I just filed them down flat.


Reassemble%2Bbites.jpg








I reapplied my TIM (once again Phobya HeGrease). This is how I do it, making a very thin square about ⅔ size of the IHS. Admittedly not my best spread here, but I find the Phobya TIM very tolerant in its application.


Phobya%2Bspread.jpg









Ready to start testing:

Ready%2Bto%2Bgo.jpg





Next post:

The numbers!
 
Last edited:

Ajay57

Member
Joined
Aug 19, 2012
I noticed you did not remove the Black rubber like glue around the edge of the IHS as some others have done!! Any reason why you decided to do your De-Lidd that way?

Respect,

AJ.
 
OP
Navig

Navig

Senior Case Master
Joined
Dec 7, 2003
Didn't really feel a need to.

The IHS fits right back into the glue impression perfectly, and actually keeps it in place, as I mentioned. I think its actually a good thing, so that the IHS isn't moving around on top of the core, like a gasket.
 
OP
Navig

Navig

Senior Case Master
Joined
Dec 7, 2003
3770k Lidded vs De-lidded

In case you are just joining, I have just finished taking the stock CPU (Lidded), removed the IHS and replaced the Intel stock TIM between the core and the IHS with CoolLabs Liquid Pro (De-Lidded).

Rules of Engagement:

Hardware is 3770k @ 4.6ghz with a core voltage of 1.3v.

Cooling is an EK Supremacy cpu block, on a loop with a Alphacool Monsta 3x120, with fans in push-pull (all fans set to 800rpms), with a Swiftech MCP35x in a Swiftech Maelstrom, with the pump RPMs set to 3000.

Asus Maximus IV Extreme, 2x8gb G.skill Ripjaws, Corsair HX850.

Load was generated by Prime95 in-place large ffts.
Temperatures logged by CoreTemp from minute 5 to minute 6 every 5 seconds.
Temperatures were the average of 3 separate runs for Lidded and De-Lidded.
Delta was calculated as the difference of the Load Temp minus the ambient temp (measured by a craftsman digital multimeter).
Core differential was calculated by averaging the the difference of the hottest core – coolest core.



Data

Lidded, Run #1
Average Load Temps 68.92 (Core#0), 74.67 (Core #1), 70.08 (Core#2), 65.83 (Core#3)
Ambient: 22.5

Lidded, Run #2
Average Load Temps 66.08 (Core#0), 71.08 (Core #1), 68.92 (Core#2), 65.58 (Core#3)
Ambient: 21.5

Lidded, Run #3
Average Load Temps 66.50 (Core#0), 72.08 (Core #1), 69.33 (Core#2), 66.00 (Core#3)
Ambient: 22


De-Lidded, Run #1
Average Load Temps 63.17 (Core#0), 64.58 (Core #1), 62.08 (Core#2), 62.5 (Core#3)
Ambient: 23.0

De-Lidded, Run #2
Average Load Temps 62.75 (Core#0), 65.67 (Core #1), 62.92 (Core#2), 62.25 (Core#3)
Ambient: 23.0

De-Lidded, Run #3
Average Load Temps 57.58 (Core#0), 60.00 (Core #1), 55.25 (Core#2), 58.17 (Core#3)
Ambient: 19




Summary


Lidded
Average Delta 46.88
Average Core Differential 7.06

De-Lidded
Average Delta 39.84
Average Core Differential 3.50







Graphs


Graph 1: Delta Load Temps (degrees Celsius), Average (yellow) and by each Run (blue)

Delta%2Bgraph.jpg


In the foreground are the Average delta load temps (yellow cylinders), Lidded on the left, De-Lidded on the right.
In the background stacked behind the averages are the values for each run (blue rectangles).






Graph 2: Maximal core differential temps, Average (yellow) and by each Run (blue)

Distribution%2Bgraph.jpg


This graph shows the inequality of heat distribution across the cores.






Conclusions

De-Lidding brought me:

-7 degrees on load temps!

Twice improved temperature distribution across cores.


:attn: FTW
 
Last edited:

Ajay57

Member
Joined
Aug 19, 2012
So looking at your results then a Overclock of 4.8 should now be achievable with the decrease in overall Temps!! Thanks for sharing and showing the difference your side.

Respect,

AJ.
 

SeeThruHead

Member
Joined
Feb 23, 2011
Location
Toronto, Canada
I think people remove the glue because it allows the IHS to get just that little bit closer to the die for better contact. (Apparently its not just shoddy tim but too much glue that caused horrible temps on ivybridge.) I don't know if you'd see any better temps by removing all the glue but you never know.
 

Ajay57

Member
Joined
Aug 19, 2012
This correct ^^^ it is the Space between IHS and the Chip that can make a different in temps as well. But each to there own way of doing things as they say!!

Respect,

AJ.
 
OP
Navig

Navig

Senior Case Master
Joined
Dec 7, 2003
I got it. I think next for my agenda is to push my overclock. It should be revealing.

Issues with non-delidding were obvious to me as I overclocked this cpu prior to its de-lidding. If you take your time and move your overclock slowly, typically there is a fairly linear or slightly parabolic curve between increasing voltage/core speed and temperatures.

As I did it with this cpu, it just suddenly hit a temperature wall, no matter I did--changing TIMs, changing heatsinks, air vs water (I think everyone would agree my current loop is overkill)--nothing would affect the load temps. It was obvious to me that something was up with whatever was going on underneath the IHS.

So as I push my overclock now that it is de-lidded we'll see what sort of behavior it shows.

Plus, at some point I will dissect the system again, and maybe at that point I'll look into removing the glue.

Solid -7c is pretty awesome, and on par with data I've seen from other de-lids.