PDA

View Full Version : Applying Shin Etsu X23-7762 and G761



GBR
07-24-2003, 10:27 PM
I've been trying the G751 and X23-7762 for 6 weeks now.

I still think that Jah's application guide is the best technique"

http://amdforums.com/showthread.php?s=&threadid=230649&highlight=Application


And,suprisingly, Artic Silver's application guide also works...

but in reverse......

...with the Shin Etsu it works best applied to the heatsink and rubbing some into the core with a finger covered in plastic wrap.


--------------


Today I tried my new Barton 2500 so I cleaned it with isopropyl alcohol, and rubbed a thin coating into the core with Saran Wrap on my finger. The layer is so thin its almost like staining the core...just trying to rub the compound into the surface.

I started with a warmed syringe of X23 and a warmed heatsink.

I cleaned the heatsink and marked the contact area and taped off the area with Post-Its.

Then I rubbed some X23-7762 into the heatsink. Again, almost like staining.

Then I dabbed many small dots like in Jah's guide and patted them down with the cut down smooth plastic card and used a straight edge on another plastic card to trim the surface even with the thickness of the Post Its.

I also cheated and dabbed a bit more with my plastic wrapped finger.

This definitely produced the thinnest most consistent application I have achieved.

I do think that a cycle of high temp burn-in and cool operation makes most thermal compounds work their best...especially true with Shin Etsu.

Started the Barton at 2.2 Ghz and quickly went to 2.4 Ghz after some Prime 95 runs and vCore increases. Quickly achieved 32 Idle and 41 Prime 95 according to MBM 5.3 Compu Nurse thingey reads 5 degrees C lower than that at load.


======================================
I've started to form opinions on when to use X23 or g751.

There are lots of variables, but here are my rules of thumb.

Intel cpu: X23-7762

AMD cpu: G751 or X23-7762

Rough surface: G751

Lapped or polished surface: X23-7762

GPU, Northbridge, Southbridge, MOS -FET: G751

Not a very skiiled person applying the compound: X23-7762

X23 is newer and easier to apply, and with higher termal conductivity theoretically should work slightly better.

My personal experience and recent reviews tells me that X23-7762 is the first choice for Pentium cpu's since it is a larger chip and application is an issue.. G751 equal or better for AMD cpu's.

But this only refers to ease of application and how it affects getting a thin even 3 mil coating on a large die.

G751 is clearly better for Northbridge, Southbridge and GPU chips, since they are rarely flat chips and the heatsinks for them are also rougher and have more runout....so the filler quality of G751 is more important than ease of smooth application..

eva2000
07-25-2003, 12:22 AM
nice info :)

GBR
07-25-2003, 12:24 AM
Originally posted by eva2000
nice info :)

Thnaks! A work in progress.

I think there's as much art as science in this part of cooling.

Maybe that's why we see a wide range of results.

felix88
07-28-2003, 07:40 PM
Originally posted by GBR
Rough surface: G751

Lapped or polished surface: X23-7762

guess i should probably get some sandpaper and lap this SLK900. too bad they aren't smooth like they used to be.

GBR
08-09-2003, 10:52 PM
Originally posted by felix88
guess i should probably get some sandpaper and lap this SLK900. too bad they aren't smooth like they used to be.

I sort of have a love / hate relationship with the SLK-900... Its great, but the quality varies and the hardware still needs improvement.

At least Thermalright is super nice about sending replacement hardware!

I'll have to lap my two SLK-900u's also...even a 1 degree C improvement may make a difference in that last OC for benchmarks.

GBR
08-09-2003, 10:54 PM
My gripe has been when I change CPU's and the standoffs unscrew from thje X-bracket...when I only want the spring-loaded screws to unscrew.

GBR
08-09-2003, 10:55 PM
Oh, and I hate those clear washers! Too hard to find if they go flying!:slobber:

felix88
08-12-2003, 09:59 AM
you have problems with the standoffs unscrewing? they were such a tight fit through the holes in my NF7-S, i'm probably going to have a hard time getting them out. :rolleyes:

by the way, shouldn't your sig say "2.44GHz (204 x 12)"?

GBR
08-12-2003, 11:42 AM
Originally posted by felix88
you have problems with the standoffs unscrewing? they were such a tight fit through the holes in my NF7-S, i'm probably going to have a hard time getting them out. :rolleyes:

by the way, shouldn't your sig say "2.44GHz (204 x 12)"?

You're right! :toast:

typo :slobber: I'll go fix it!

I wish I could hit 200 x 14!:banana:

Actually a 200 x 14 that ran Prime95 and both 3D marks would get a
:banana: :banana: :banana: :banana: :banana: :banana: :banana: :banana: :banana: :banana: :banana: :banana:


As regards the ThermalRight hardware...I've heard of parts stripping out, breaking, or like one of my SLK-900u ...the wrong part unscrews.

I could reassemble and use a very low strength thread lock like Locktite makes.

But since this problem is on only one of my SLK-900u sinks, I think its easier to swap out all the parts since Thermalright was nice enough to send me new parts.

I really like the heatsink...but I do see room for improvement, and that will probably come in time!

felix88
08-12-2003, 01:51 PM
yeah, i really like this SLK900 except it's not preforming like it should. i just need to lap it.

GBR
08-12-2003, 02:35 PM
Originally posted by felix88
yeah, i really like this SLK900 except it's not preforming like it should. i just need to lap it.

Also on my 'to do' list!:D

Susquehannock
08-13-2003, 11:18 PM
Nice work GBR. :up:


Over the last couple months (in between too may other things)
I have been conducting a similar set of experiments.
What you presented in respects to usage and advantages
of G-751 over X23 in particular situations is right with what
I have been able to observe thus far.


The "finger bag" method doesn't work as well for me as
"feathering" it on with a plastic card though.
Every time there is almost as much stuck to the bag as the core.
Using a quick "feathering" motion with just the corner of
the card has served me quite well so far.
And there is very little waste.:) What little compound that does
stick to the card is just enough to rub into the contact patch
on the heatsink.

Something else I've noticed is that a perfect mirror finish
is no better than an even 800 grit finish. At least the difference
is less than one degree.;)

Again, nice work GBR.
Surely many will benifit from your efforts. :toast: