Lapping FTW

[ViN86]

Originally Posted by mailman2

I used to start with 400. Now I start with 180 and get it to copper in minutes. Then I jump to 400 and finish with 600. The mirror finish makes it hard for the TIM to actually work. It has been recomended among veterns of OC and lapping that 600 or 800 is better cause there are actually slight imperfections for the TIM to settle in.

ideally, the TIM is supposed to be used in a very thin layer to create a contact between the imperfections in the heat spreader and the heat sink.

so, either higher grit with less grease, or lower grit with a little more grease.

but, the grease has a thermal resistance, so higher grit should create a better contact which should create better conduction. so, higher grit with a very small amount of grease should be the best.

Its not how mirror smooth the finish is, its the fact that you are levelling the surface making it true so that you get better contact across the IHS.

technically both should factor into how well it conducts heat.

[Bman212121]

Originally Posted by Zapablast05

Is that what's recommended? I'm really thinking about lapping my CPU and HS when I get home now that I found a guide on lapping my Xigmatek HDT-S1283. I found a grit guide and is it better to start rough with a low grit sheet then move up to smooth grit sand paper to give it a final polish?

Well there are a lot of different guides but they generally follow the same rule. Yes, you would start with a lower grit sandpaper to work out the bigger gaps and move to a higher grit to finish the polish. A visual image is think of your HSF like a mountain range. When you first start to sand it you need to do a lot of work to make the highest points even with the lowest points. So starting with a 150 or 200 will require a lot less strokes than say a piece of 400 grit to get the surface smooth. The reason for going up gradually is that you will reach a point where 150 grit will make more gouges into the surface than fix them. If you just jumped up to a 1500 grit sandpaper then you would sand the tops of your peaks but wouldn't really get the surface smooth. Using a few steps in between will bring the overall surface closer to smoothness before you make the jump to the final polish.

Originally Posted by mailman2

I used to start with 400. Now I start with 180 and get it to copper in minutes. Then I jump to 400 and finish with 600. The mirror finish makes it hard for the TIM to actually work. It has been recomended among veterns of OC and lapping that 600 or 800 is better cause there are actually slight imperfections for the TIM to settle in.

Its not how mirror smooth the finish is, its the fact that you are levelling the surface making it true so that you get better contact across the IHS.

Actually, the smoother the better. The only reason for needed TIM at all is because there are holes to fill so there isn't just air. (TIM conducts heat better than air) The ideal situation would be having both surfaces so flat that when you combine them together it makes both pieces of copper bond to each other making one solid piece of copper. Now I don't have physical data to back up that claim so maybe they know some special case where that may help, but from a science standpoint it makes a lot more sense to have one solid piece of copper for the best possible heat exchange. The only thing I could see changing that is if TIM conducts heat better than copper. Given that too much TIM makes it act like an insulator would make me think otherwise.

[lee63]

This guy had some good results with a i7 http://forums.guru3d.com/showpost.ph...04&postcount=1

I might have to give my CPU a shot also

[ViN86]

Originally Posted by Bman212121

Actually, the smoother the better. The only reason for needed TIM at all is because there are holes to fill so there isn't just air. (TIM conducts heat better than air) The ideal situation would be having both surfaces so flat that when you combine them together it makes both pieces of copper bond to each other making one solid piece of copper. Now I don't have physical data to back up that claim so maybe they know some special case where that may help, but from a science standpoint it makes a lot more sense to have one solid piece of copper for the best possible heat exchange. The only thing I could see changing that is if TIM conducts heat better than copper. Given that too much TIM makes it act like an insulator would make me think otherwise.

see my post above

[bob saget]

Originally Posted by lee63

This guy had some good results with a i7 http://forums.guru3d.com/showpost.ph...04&postcount=1

I might have to give my CPU a shot also

wow.....

[nekrosoft13]

Originally Posted by methimpikehoses

The interesting thing for me was that the idle temps are similar, about 40C, but the stress temp is 15-20C cooler.

speaking of temps, what temp are you talking about.

CPU temp or core temps?

haven't OC'ed yet, but my CPU temp is at 26c, core temps are about 38-40c

Originally Posted by Zapablast05

Is that what's recommended? I'm really thinking about lapping my CPU and HS when I get home now that I found a guide on lapping my Xigmatek HDT-S1283. I found a grit guide and is it better to start rough with a low grit sheet then move up to smooth grit sand paper to give it a final polish?

I went 600-1000-2000, and got a very good polish on mine. It wasn't mirror finish, but close. Used just a little water and soap on the 600 grit, and then did the 1000 and 2000 grit dry.

Originally Posted by mailman2

The wolfdales had issues with the sensors getting stuck and not going lower than 40c. Sounds like its another sensor issue. Even though its dropping under load I'd bet the sensors on the I7s are just busted like wolfdales were.

Mine aren't stuck on 40C, they usually stay there though during normal use. The start at about 36-38C and warm up when the machine is on for a while.

Originally Posted by lee63

This guy had some good results with a i7 http://forums.guru3d.com/showpost.ph...04&postcount=1

I might have to give my CPU a shot also

Interesante... I'm scared to lap the cpu... seems so easy to get the other side messed up.

[lee63]

Originally Posted by methimpikehoses

Interesante... I'm scared to lap the cpu... seems so easy to get the other side messed up.

You gotta put the black cover on it... that's what I did anyway.

Originally Posted by lee63

You gotta put the black cover on it... that's what I did anyway.

It's not like that thing is airtight though... dust... water... yikes.

You might have steadier hands than me.

[lee63]

Originally Posted by methimpikehoses

It's not like that thing is airtight though... dust... water... yikes.

You might have steadier hands than me.

I never used water.. I see your point though.

[mailman2]

Originally Posted by Bman212121

Actually, the smoother the better. The only reason for needed TIM at all is because there are holes to fill so there isn't just air. (TIM conducts heat better than air) The ideal situation would be having both surfaces so flat that when you combine them together it makes both pieces of copper bond to each other making one solid piece of copper. Now I don't have physical data to back up that claim so maybe they know some special case where that may help, but from a science standpoint it makes a lot more sense to have one solid piece of copper for the best possible heat exchange. The only thing I could see changing that is if TIM conducts heat better than copper. Given that too much TIM makes it act like an insulator would make me think otherwise.

The flatter the better. If you finish it too smooth the TIM will just run off with high heat. I've had this happen. You dont want a mirror finish on the chip, trust me. Doesn't sound like you've been doing this too long, but you want a flat finish with 600 or 800 for the best results. I've tested with multiple water blocks and many many chips over the years.

[ViN86]

Originally Posted by mailman2

The flatter the better. If you finish it too smooth the TIM will just run off with high heat. I've had this happen. You dont want a mirror finish on the chip, trust me. Doesn't sound like you've been doing this too long, but you want a flat finish with 600 or 800 for the best results. I've tested with multiple water blocks and many many chips over the years.

if this is the case then you were using too much TIM.

transport fundamentals says a higher grit (smoother) flat finish will have better contact and better heat transfer.