PDA

View Full Version : To lap or not to lap



diluzio91
10-03-2011, 05:07 PM
So the reason i've been having so many troubles that i cancelled the oc competition:

My Rig:
one gfx card died on me, for no reason whatso ever.. stock clocks, fans at 80%... its a sad day

Also, i sold/traded my i7 870 and p55 mobo to/with my friend for an i5 2500K and a gigabyte UD4 p67 mobo. And to my suprise, its running hot as he**...


Friends issues.
Blown PSU
Water block crack
p/o gf locked friend out


Thats 4 rigs out of the 6 I had interested in the competition here, and the other 2 lost interest when they found out the 4 of us were having so many issues.


So my question. i noticed that the i5 that i have now is convex, so should I lap this processor? Or should i live with the slightly higher temps and wait till ivy bridge comes out. Basicly would lapping the processor hurt it's value? because the plan and reason for getting this proc/mobo combo was to go to the new ivy bridge chip when it comes out. also to sell the i5 to pay for part of it.

Thoughts?

Twigsoffury
10-03-2011, 05:14 PM
So the reason i've been having so many troubles that i cancelled the oc competition:

My Rig:
one gfx card died on me, for no reason whatso ever.. stock clocks, fans at 80%... its a sad day

Also, i sold/traded my i7 870 and p55 mobo to/with my friend for an i5 2500K and a gigabyte UD4 p67 mobo. And to my suprise, its running hot as he**...


Friends issues.
Blown PSU
Water block crack
p/o gf locked friend out


Thats 4 rigs out of the 6 I had interested in the competition here, and the other 2 lost interest when they found out the 4 of us were having so many issues.


So my question. i noticed that the i5 that i have now is convex, so should I lap this processor? Or should i live with the slightly higher temps and wait till ivy bridge comes out. Basicly would lapping the processor hurt it's value? because the plan and reason for getting this proc/mobo combo was to go to the new ivy bridge chip when it comes out. also to sell the i5 to pay for part of it.

Thoughts?


Got a pic of the thermal paste square left on the heatsink and processor? That'll tell you a lot about the variations in the base and heatspreader.

I wouldn't think it would affect its value, but that varies person to person and its normal to assume some somebody did a terrible job flattening the base.

What i might do is look at the left overs of the thermal paste and see wheres its been pressed harder and sand that spot down on the heatsink instead of sanding the processor down if your looking to resell that particular processor down the line. The heatsink will probably be trashed for any other processors but people want the processor and usually have there own heatsink to use.

diluzio91
10-03-2011, 06:04 PM
i'm really against trashing that heatsink, the flatness rating is 4 microinches or better (no idea what that means, but my stepdad measured it and said it was good. lol

Kayin
10-03-2011, 06:18 PM
Diluzio, get me a pic of the thermal paste footprint and I'll tell you what you need to know. If you can do it on a piece of glass or plexi even better.

They say lapping is dead, I beg to differ. And I've got the science (and machining experience) to prove it.

We'll get your SB cooled properly-though you may just have a high leakage proc like mine-I did 4.4 on an 860. Just runs hot as ****.

diluzio91
10-03-2011, 06:28 PM
i'll get it up probably thursday or friday since i have to work tuesday and wednesday night

altec
10-03-2011, 07:17 PM
i'm really against trashing that heatsink, the flatness rating is 4 microinches or better (no idea what that means, but my stepdad measured it and said it was good. lol

Fancy way to say .000004". In other words, REALLY FLAT! :D If it is already that precision ground, I would stab myself in the hand before even considering touching it. MAYBE buff it by hand...

Don't know how lapping effects value, but you have to word it fancy...Like saying you lapped it to 3 microinches! :D But really, you get the idea there.

Cale_Hagan
10-03-2011, 07:31 PM
diluzio, check your pm's

TheGreatSatan
10-03-2011, 09:33 PM
What paste are you using? I switched to IC Diamond and saw a ten degree drop.

diluzio91
10-03-2011, 11:09 PM
Using AS5 Ceramique. Only paste i'll touch these days. Cheap as sin, no set time, non conductive, and i've never seen it beaten by more than 1*C

Kayin
10-04-2011, 12:31 AM
I hate to say it, but I can think of TIMs that beat it by 10 to 20 degrees. Look up Indigo Xtreme.

diluzio91
10-04-2011, 01:14 AM
i've heard of it, but for $20 a pop... little too rich for my blood

Twigsoffury
10-04-2011, 12:14 PM
I hate to say it, but I can think of TIMs that beat it by 10 to 20 degrees. Look up Indigo Xtreme.

10 to 20 degrees is a little towards the recidulous side.


i can show you a 20 degree difference with the same thermal paste just by applying it differently.

Kayin
10-04-2011, 12:48 PM
http://www.xtremesystems.org/forums/showthread.php?232141-Review-Indigo-Xtreme-vs.-AS5-MX-2-IC-Diamond-Shin-Etsu-X23-7783D

Independent testing by a professional. Indigo Xtreme is more like temporarily soldering your proc to your HS or block. It comes off easily, but it's only metal and nothing else. By logic it should beat the living daylights out of anything else.

Twigsoffury
10-04-2011, 06:32 PM
http://www.xtremesystems.org/forums/showthread.php?232141-Review-Indigo-Xtreme-vs.-AS5-MX-2-IC-Diamond-Shin-Etsu-X23-7783D

Independent testing by a professional. Indigo Xtreme is more like temporarily soldering your proc to your HS or block. It comes off easily, but it's only metal and nothing else. By logic it should beat the living daylights out of anything else.

http://www.indigo-xtreme.com/images/largetimcompare.jpg

but it's only 6~7 degrees and not 10~20.

Kayin
10-07-2011, 05:30 PM
Those are all also perfect mounts. Ever had an imperfect mount? That'll mess up anything but Indigo. But, we'll see.

SXRguyinMA
10-07-2011, 06:16 PM
I lapped my i7-960 and noticed a drop in temps just using the stock Intel HSF. I think I'm going to pick myself up some of that indigo stuff :D

diluzio91
10-11-2011, 12:50 AM
Pics are up

paste fresh off the hsf

http://a2.sphotos.ak.fbcdn.net/hphotos-ak-snc7/s720x720/307766_10150352497763426_601993425_8205275_4378123 52_n.jpg

hsf after removal
http://hphotos-iad1.fbcdn.net/hphotos-ash4/s720x720/321565_10150352497843426_601993425_8205276_1671170 698_n.jpg

New paste

http://a5.sphotos.ak.fbcdn.net/hphotos-ak-snc7/s720x720/313245_10150352497888426_601993425_8205278_9812340 74_n.jpg

after pressing

http://a3.sphotos.ak.fbcdn.net/hphotos-ak-ash4/s720x720/317076_10150352497913426_601993425_8205279_2134050 961_n.jpg

Fresh paste on hsf

http://a8.sphotos.ak.fbcdn.net/hphotos-ak-ash4/s720x720/299325_10150352497948426_601993425_8205281_1843743 124_n.jpg

After attaching

http://a6.sphotos.ak.fbcdn.net/hphotos-ak-ash4/s720x720/315958_10150352497983426_601993425_8205282_2622163 68_n.jpg

Matching to above

http://a1.sphotos.ak.fbcdn.net/hphotos-ak-ash4/s720x720/297886_10150352498023426_601993425_8205284_6825210 21_n.jpg

Side note, just hit 4.5 ghz, i think my temp monitoring program was having trouble, because i downloaded real temp and my highest temp is lower than my lowest temp with aida.

Edit edit: 2 cores seem to be running about 7*c hotter in my prime 95 testing.

Twigsoffury
10-11-2011, 12:57 PM
That's using way to much thermal paste bro.

Cut the amount you use by 1/5th and spread it out evenly with a old credit card or a playing card. and only apply it to the top of the processor, then clean and polish out ALL of the old thermal paste from the heatsink, polish that son of a b@tch until your white cloth stays white, if its still pulling out old grey thermal paste, you haven't gotten all of the old paste out of the pours of the metal, then you won't get a good strong bond between the two surfaces and it'll insulate instead of dissipate.


If you can't see through the layer of thermal paste... its to much. they say 0.04mm thick, thats got to be pushing 0.20mm


I use HWmonitor (from CPUZ) It doesn't have a fancy GUI. or lots of options or pizazz.

But its accurate, its always accurate.

Kayin
10-11-2011, 01:22 PM
The credit card trick has been pretty much obsoleted by testing. It makes small air pockets in the TIM, and putting a blob on there and letting the sink spread it out both wastes less TIM and makes a better bondline. Check Vapor's exhaustive article on TIM application methods testing for the numbers on that.

Your IHS is cupped. Lap it.

And nothing is "accurate" except for a K-type thermocouple milled into the surface of the IHS and affixed there with thermal epoxy, according to the Intel white papers. HWmonitor is often off, it's simply consistently off, which makes it a fine basis for internal comparisons but never tells the true temp of the CPU.

Twigsoffury
10-11-2011, 01:35 PM
And nothing is "accurate" except for a K-type thermocouple milled into the surface of the IHS and affixed there with thermal epoxy, according to the Intel white papers. HWmonitor is often off, it's simply consistently off, which makes it a fine basis for internal comparisons but never tells the true temp of the CPU.

link?


The credit card trick has been pretty much obsoleted by testing. It makes small air pockets in the TIM, and putting a blob on there and letting the sink spread it out both wastes less TIM and makes a better bondline. Check Vapor's exhaustive article on TIM application methods testing for the numbers on that.



Now maybe i'm just stupid, but putting a blob of paste on there then just smashing the **** out of it with the heatsink and hoping for the best sounds like it would make a hell of a lot larger air bubbles.


but anyways to each his own i guess.

diluzio91
10-11-2011, 01:47 PM
Now maybe i'm just stupid, but putting a blob of paste on there then just smashing the **** out of it with the heatsink and hoping for the best sounds like it would make a hell of a lot larger air bubbles.
but anyways to each his own i guess.

Not really, when you squeeze a tube of toothpaste the thinnest layer of paste will be between the points of pressure, the rest of the paste gets pushed to the outside. same thing going on here. The dot pushes outwards because of the pressure of the hsf and the excess goes to the outside. Theres no chance of air bubbles because the paste can only go where there is a gap. with the cc method you actually create air pockets, but there is nowhere for the excess paste to go.

is there a decent lapping kit anyone reccomends?

Kayin
10-11-2011, 02:11 PM
ftp://download.intel.com/design/processor/applnots/319980.pdf

6.1.2.1, I can provide other white papers as well.

Also, I just buy sandpaper and go at it.

SXRguyinMA
10-11-2011, 03:03 PM
Also, I just buy sandpaper and go at it.

:stupid: That's what I did. Get wet-sand paper as well. Tape it to a sheet of glass, put a little water on it and have at it. I started with 400 and finished up with 1500 when I did mine. Don't be afraid to get your proc wet either. (from Kayin's tutrial). I actually washed mine gently afterwards to make sure there was no particles left over then tap-dried it with a paper towel. Works 100% :D

SXRguyinMA
10-11-2011, 03:09 PM
ftp://download.intel.com/design/processor/applnots/319980.pdf

6.1.2.1, I can provide other white papers as well.

Hmm...got me thinking about milling out my EK block to fit in a temp probe...... :think:

Twigsoffury
10-11-2011, 06:59 PM
there is a gap. with the cc method you actually create air pockets, but there is nowhere for the excess paste to go.


But i use so little paste there is no "excess"?

Man i need to snap a picture just to show y'all just how little f@#kin paste i use.

Kayin
10-13-2011, 02:14 PM
It's still putting air in it. The manufacturers don't even tell you to do that, they mention the drop of paste and use the sink to press it out.

The only paste that says to push it about like that is coollaboratory liquid pro, because it actually wets the metal. But it's a liquid metal, and not subject to the bondline gaps that pastes are.

diluzio91
10-13-2011, 03:04 PM
if i'm gonna have to order up some sandpaper anyways, what do you reccomend kayin? Keeping in mind that i'm poor as S**T atm.

Kayin
10-14-2011, 11:20 AM
there are lapping kits with a piece of glass for 19 bucks or so on line, or I did 600, 800, 1000, 1500 for about 10.

Twigsoffury
10-14-2011, 11:24 AM
wal-mart sells automotive finishing sand paper packs for like 6$ It's got 400~3500 grit sandpapers all in one.

Aldersan
10-14-2011, 11:24 AM
ooohhh good find twigs , i might have to grab one :D

NightrainSrt4
10-14-2011, 11:46 AM
As for one of your original questions, I wouldn't buy a lapped processor unless it was priced decently below the same non-lapped used processor. This is going for a random internet person. If it was someone I knew directly, then that would be another thing altogether.

A lapped heatsink I wouldn't mind too much. It's just a heatsink.

diluzio91
10-15-2011, 02:42 PM
bump on TIM reccomendation?

Kayin
10-16-2011, 11:28 AM
Top is Indigo Xtreme. Next is Coollaboratory LM Pro, then if you want an actual paste something like PK-1 or MX-4.

SXRguyinMA
10-16-2011, 12:52 PM
I got my indigo extreme installed on my MS build, now I just need to wait until it's ready to fire up so I can do the heat cycling for it. The only tough part is that the instructions say to make sure the MB is horizontal, but with how my loop is setup I can't make my res sit horizontal. Unless I use a separate tube for the res to re-orient it and just use regular water for the heat cycling phase, then drain and re-assemble the proper way lol

Kayin
10-17-2011, 09:54 AM
I think it'll be fine on its side for the 5 or so minutes it'll take.