Lapping the Northwood

[captain]

I saw I guy that lapped a intel pentium northwood until the copper was visible.

i would like to know if this is dangerous (will i destroy my cpu) and also how he did it. Did he use sandpaper or a dremmel?

[Kevin]

Well this was pretty commonplace w/ the heat spreaders on the older celerons. It is only really useful if the heat spreader isn't totally flat. This definitely can be a bit dangerous. I'm not sure how thick the heat spreader is though. I'm sure he used sand paper. The purpose was to flatten the surface mostly, not to just reveal the copper. Using a dremel wouldn't guaruntee a flat surface. If you are interested in lapping you northwood, I would reccommend looking for lapping guides in reference to the older celerons.
-Kevin

i saw the same thing... lapping down to the copper can increase the thermal transfer efficeincy. Lap like you would a heatsink, just MUCH MORE CAREFULLY! also start with a much higher grit sandpaper and be careful to only press down in the center of the CPU. wear a grounding strap at all times.

[GigaFrog]

You got a Northwood already? Wow.
Wich speed? Tried to overclock yet?

I lapped my PIII 700, and it didn't have a heat spreader. Lapping the core directly improves thermal transfer by reducing the thickness of the core substrate. I didn't see any significant improvment.

The Northwood will probably be my next system, particularly if it comes with DDR chipset. What mobo do you have?

[Brad]

It'd be interesting to do, but if you went too far, well, you'd have problems.

Just how much copper is there anyway? Also, how big is the IHS itself, you could just mount a pelt on there instead of needing another cold plate

[beav]

I lapped my PIII 700, and it didn't have a heat spreader. Lapping the core directly improves thermal transfer by reducing the thickness of the core substrate. I didn't see any significant improvment.

That is just stupid. The silicon on the p3 cores are almost perfectly flat, and only microns thin. Lapping would do worlds more harm than good.

However, i just bought some tualatins, and i was tempted to lap the heatspreader. I'll wait a little bit longer until i see some safe, worthwhile results and methods. Until then, my p3 runs cool enough

[Brad]

Just do it TM

Or just remove the entire IHS, then you can sand it down quite comfortably

[GigaFrog]

That is just stupid. The silicon on the p3 cores are almost perfectly flat, and only microns thin

You obviously don't know what you are talking about.
What you remove by lapping the core is the protective resine. You really think that the silicon is exposed? The substrate is more than micron thick.

Insults and unaccurate statement form a pro/staff, thats not something I was expecting here!

Just one simple search on Core+lapping would have tell you this:
http://www.ocworkbench.com/articles/.../lappingp1.htm

Your position of pro/staff does not make you that knowledgable that you don't need anymore research to insult something you never saw before. Lapping the core was actually quite common before the heat spreaders.

[beav]

Thanks for correcting me, as my information was slightly off, and i should ahve explained some more, and i learned something new after researching this.

What i meant was the the die/slug/core/silicon is extremely flat, thus the protective coating just thinly covers this, not really affecting it's flatness. This does not excuse my extreme carelesness of words and thought.

While researching i discovered several things. It's not a protective resin as you said, or a silicon substrate as i thought before, and it's a lot more than a few microns thick. It's a copper or sometimes alluminum alloy filmed over the cache/core that measure in at .051mm. This is still pretty small, but then again i said it was microns thin. This process is called OLGA or Organic Land Grid Array and was introduced by intel with the Pentium 3.

So, both of our information was wrong. I'm sorry to have upset you so, as I was just pulling rough details from my memory, and what i have picked up from other people.

Thanks for your understanding. I still hold strong behind my opnion that lapping a FC-PGA OLGA package is not worth the risk. But that's my personal thought.

[Brad]

quite a few people have done it, most find their temps are basically the same, if they've done a good job it might drop 1 or 2C, if they do a bad job it might go up 1 or 2C

[GigaFrog]

Cool.

My temp did not change that much either, 1 or 2 C.
The problem is removing enough thickness to get some effect, but not too much which would kill the die.

[Brad]

I'd have thought trying to keep it flat, and not round off the corners would be a big issue

[GigaFrog]

I didn't encounter problems with that. Since my temp did not increase (!) I didn't screw it too much at least.
The sand paper was attached by the corners to a thick flat piece of wood. The beginning is the most tricky, until the sealant goes away (the blue stuff).

[Brad]

I'd be way to scared to do that I think, I haven't seen anyone improve their temps hugely either

[GigaFrog]

Neither did I.

However, I just replace my water block, and there was some residus of Artic Silver that I could not take of the core. A little laping on 1500 sand paper took care of it.

Being scared? I have done a few mods, on the mobo Vcore regulation particularly, that scared me more that lapping the CPU core. My thought before restarting the computer was not "will it run faster?" but simply "will it run?". It always did, as for today.

My HIP6021 Vcore regulator has now 4 pins desoldered and lifted from the PCB (the VID pins). I still can't drive dynamically these pins as I would like too, but I'm working on it.

Scary, but fun.

[Brad]

I just did the wire wrap on my p3's. That will get me up to 2.4v, although it doesn't matter anyway, they won't post above 2.05v

[GigaFrog]

What, they don't post above 2.05V ???
Mine posts at 2.24V. I just tried yesterday.

They don't post because the coppermine has only 4 VID pins (VID0 to VID3) while the Vcore regulator actually supports 5 VID pins (I guess this was to be compatible to Katmais).
Motherboard designed for Coppermines (like my CUSL2) has the VID4 pin (the 5th one) soldered to the ground. You need this pin to be open to go above 2.05V.

I have the HIP6021 on my board, and it was not posting above 2.05V until I desoldered VID4. Now it runs great at 2.24V.
What Vcore regulator do you have?

Spec of the HIP6021:
http://www.intersil.com/data/fn/fn4/fn4684/FN4684.pdf


You need to lift pin3.
Be carefull, set your Vcore to 1.3V before doing so, otherwise, lifting pin3 while at 2.05V will give you ... 3.5V, and most likely a fried CPU (check the VID table on the above datasheet, page 2-304 and 2-305).

[Brad]

I have them both defaulting at 2.05v, by wire wrapping all of the pins, and in the vp6 bios, I can go to 1.3 - 2.4 in .05v steps at a stock 2.05v

[GigaFrog]

That's what I did too. I wired the 4 VID pins to ground so it would default at 2.05V. By doing so, I also got Vcore steps at 2.1, 2.2, 2.3 and 2.4V in the BIOS, but none of them would work (no post) except the 2.05V step.
I then looked at the Vcore regulator chip of the mobo: HIP6021 (the VP6 might use the same one). The datasheet says there are 5 VID pins while the CPU drives only 4 of them. The 5th one was grounded, and it is actually the one that enables voltage steps above 2.05V.

Your board behaves very much like mine. I finished the mod tonight. I actually had to lift VID0, VID2, VID3 and VID4, so that I would get 2.2V according to the HIP6021 datasheet (VID=11101).

I wanted to copy a screen shoot of MBM to show you. How can I attach a gif file to a post?

BTW, I updated my sig with my new voltage.

[Brad]

ok, COOOL.

as a step by step guide, what do I do exactly?

[GigaFrog]

A little more tweaking got me to 1092MHz (1099 would load windows and crash, arrrrrg).

I'll work on the step by step guide tonigh, I have to go to work now.

Bye.

[GigaFrog]

Brad, check your email.

[Brad]

hehe, I got it, but I have been really busy the last two days, I just haven't had the chance.

btw, it is an excellent article, very well done

[shroomer]

Be careful... I lapped my duron to get all the writing off so that only the core was touching the heatsink... I don't know what I did, but I took a chip that did 1024 to one that "only" does 980.

I'm sure that lapping the IHS of the P4 would have little affect and I agree that it is just not worth it. A duron is only $40, but a P4 is much more and not as easy to replace. Put your energy into fan placement and plans for upgrades.