Lapping and Using Arctic Silver 5 = NO TEMP DIFFERENCE?

BoxsterS258 · Post by **BoxsterS258** » Tue Sep 07, 2004 3:35 pm

I lapped my zalman 7000cu and my P4 2.8 Prescott HT, with a lapping kit from easypckits, applied arctic silver 5, following all directions, but there has been to temp change so far in 6 hours. I know there is curing time for the AS5, but i dont think it would do much since nothing has changed. I idle at 46 and I am at about 58 load, same as before.

infinitevalence · Post by **infinitevalence** » Tue Sep 07, 2004 7:41 pm

There is some times the risk that lapping will yield no performance improvment. That generaly happens when the heatsink is already flat and well polished. There is also the possiblity that in the process of lapping you did something that negated the improvment from the lapping. I hope that you do see some improvement over time, but i know from my experience that my temps only improved around 1-2c. Now take into account that when i am at my max overclock my cpu is giving off around 100w of heat which is a lot for even the best cooler.

A10Pilot3 · Post by **A10Pilot3** » Tue Sep 07, 2004 8:07 pm

infinitevalence wrote:There is some times the risk that lapping will yield no performance improvment. That generaly happens when the heatsink is already flat and well polished. There is also the possiblity that in the process of lapping you did something that negated the improvment from the lapping. I hope that you do see some improvement over time, but i know from my experience that my temps only improved around 1-2c. Now take into account that when i am at my max overclock my cpu is giving off around 100w of heat which is a lot for even the best cooler.

I just got a kit from them to review and I will post a link to my results...

and infinitevalence, how do you calculate how much heat your CPU is giving out? (watts)

I have my 3GHz Pressie at 3.6 and I'm pressing it farther, my poor Koolance can't keep up, still get to 45C according to the Koolance probe, probably 100C or so according to BIOS

but really the BIOS probably says 60-65C...

infinitevalence · Post by **infinitevalence** » Tue Sep 07, 2004 8:45 pm

Use this website it will give you your wattage.

http://www.benchtest.com/calc.html

Assuming no change in voltage it worked out like this

Stock speed: 3000
Stock Voltage: 1.525
Stock Wattage: 89
OC speed: 3600
OC voltage: 1.525

Max Watt:106.8
88% Watt: 94

A10Pilot3 · Post by **A10Pilot3** » Tue Sep 07, 2004 9:20 pm

I didn't see anything for a 3GHz Prescott, where is it?

Well I'm well over 100watts then, up to 1.5625V too.

infinitevalence · Post by **infinitevalence** » Tue Sep 07, 2004 9:29 pm

there is no option for that cpu, so you have to enter the info you have. i think they have not updated in a while. if you put in all the values your self you should find a more accurate result because i was guessing at your voltage.

Liquid3D · Post by **Liquid3D** » Thu Nov 11, 2004 10:19 am

There are many factor's involved in lapping. First, not all Pentium, or AMD IHS (Integrated Heat Spreader) are perfectly flat. In fact a good portion of the time the IHS will be convex. This is done primarily to defend against Heat sinks with imperfect bases (concave), this way where the core is centered under the IHS, contact is ensured. As you may or may not know, between two surfaces only a very small amount will make contact (even perfectly flat surfaces). This is due to the microscopic striations, pores, and pits inherent to many metals. This page from Koolance is an excellent source of info.

The empirical evidence for my claim, is Thermal Paste. This exists to ensure there's a "seal" or more accurately, medium between the two surfaces in question through which the kinetic energy (heat) will be conducted. If you want to Lap, then you should be using a mirror, or piece of glass upon which to place the grit paper. Then using water or as I use, isopropyl alcohol 70% (which still ensures there's water, but cleans away the metal as well) you would place the heatsink onto the paper, and move the heat sink in a figure eight. This is difficult for me to do rapidly, so I simply go in small circles, reversing direction. I would do this until each grain of paper seems to have exhausted it's use, meaning it's not removing anymore material.

A mirror image is not THE sign of a flat surface, simply a polished surface, this shouldn't necessarily be your goal. Flat is the goal. However; getting a mirrored surface does mean you've done a decent job.

You may want to lap your processor. I have done this twice, and it reduced temps a few celsuis degrees. But remember lapping isn't going to guarantee a large reduction in temp. Changing your cooling method is the best way to do that. If you lap the processor, remember your loosing the warranty, if it's OEM you've nothing to loose. Perhaps the most drastic reduction in temp, will result from removing your IHS. I wouldn't recommend this unless your watercooling, because better air-coolers use stand-offs due to their weight. Northwood's manufactured prior to 11/03 and A64's are safe. However Northwoods and Prescott's manufactured after 11/03 use thermal epoxy between IHS and core instead of thermal paste, this gives much better heat transfer. Trying to remove an epoxied IHS will destroy the processor, even though some claim there are tricks such as placing it in a freezer overnite. The reason Intel did this was for better thermal transfer, it's not as many thought, to dissaude people from removing the IHS. Out of a million CPU's sold maybe a hundred people might even try removing the IHS.

For Pentium's I've used the "string" method, where you find the corner of the IHS where Epoxy is absent, and use the string in a sawing motion slipping it underneath the IHS until the epoxy seal is gone. DO NOT FORCE THE IHS OFF! TWISTING it FROm SIDE to SIDE GENTLY is the safest method.

Thermal Paste application is also very important to temps. DO NOT SPREAD. Since the IHS is approximately 70% larger then the core beneath it, the heat is concentrated in the center. Yes the IHS is intended to "spread" heat by increasing surface area for that heat to dissipate over, just as your heatsnk is designed to spread heat. This is why the best heatsinks are behemoths weighing over 630-grams. More mass = more surface area, more mass = more area a fan can cool, drawing the heat into it.

PLACE a grain sized dollup of AS5 or Ceramique (or whatever you use) in the center of the IHS. Even if you've removed the IHS, place a smaller dollup of TIM on the center of the core. DO NOT SPREAD, simply mount the heatsink, and allow the natural pressure to spread the paste. WHY you ask? Because the heat transfer primarily occurs at the center of the IHS, and with a decent heatsink this heat will be rapidly dissipated. Spreading the paste over the IHS can actually defeat the purpose, as it it's akin to throwing a blanket over the surface. This is where the Convex IHS theory becomes doubly important. If (and many are) your IHS has a slight convex, then only that area will be making contact with the heatsink base, spreading paste over the entire surface of the IHS will be like "insulating" (throwing a blanket over) it holding heat in that would normally dissipate into the environment.

PS: by the way, I'm not trying to come off as a "know it all." I'm a very humble person, and I certainly don't know everything. These are things that others taught me, an that I learned form research. I wouldn't share anything with you I haven't done myself, and much of what I've learned I learned in long conversations with Nevin House of Arctic Silver. I've been lucky to call him a friend. I'm totally open to learning, and I still consider myself to be a noob in many areas.

infinitevalence · Post by **infinitevalence** » Thu Nov 11, 2004 4:57 pm

I think Capper knows all about removing the IHS from a prescott