finished lapping my Q6600 (with pics and results)

[KBTuning]

am i the only one who thinks the results werent that great? i mean yeah you did drop your temps a lot but i think you've got other problems as it is considering your end IHS/TRUE lapped temps were what my non lapped CNPT9700NT temps are...

im a n00b when it comes to applying thermal paste and i put some AS5 on my Q6600 and slapped my CNPS9700NT on it and i have yet to break 52* in prime on my hottest core... and im at 3.0GHz with 9x333 and 1.275v in bios... 1.232-1.240 in windows...

 

[norcimo]

The results are very true and worth it, (if you dont mind violating the CPU warrantee)

what seems to be happing is that many users are not lapping properly and are skipping vital procedures.

I have a Q6600 OC @ 3.2 GHZ, this CPU under load is exactly 92F. GRANTED, im using a chill tec ultra cooler, but never the less, the same cooler without lapping, was 105F Idle.

proper lapping takes time and patience.

this is material needed to do a perfect (NASA) mirror finish:

400 GRIT
600 GRIT
800 GRIT
1000 GRIT
1500 GRIT
2000 GRIT
Metal Polish
Polishing cloths.
High speed dremel. with metal polish attachment
If done properly, you can achieve perfect results.

And yes, it is messy, but the more black the metal polish becomes, the more glassy the surface will be. That's the key, I hand polished it well over 8 times, until it was perfect!

I used the Dremel to remove clearly visable polish lines lift by the sanding procedure, then again and again I polished using one cloth with paste, i rubbed the CPU against it (circular motions), and another cloth to clean up the black goo, or "copper" to buff it and get perfect results.

 

[flexy]

Originally posted by: Furen
Real men lap the die/dies/dice, not the IHS

roflmao !

edit: Impressive job, nice temps!

 

[bobjc]

Is the nickel coating on hsf to prevent oxidation and if so what happens after you lap both hsf/proc and have copper to copper connection? Will the copper on both continue to oxidize or does tim prevent oxidation from occurring?

 

[Idontcare]

Originally posted by: bobjc
Is the nickel coating on hsf to prevent oxidation and if so what happens after you lap both hsf/proc and have copper to copper connection? Will the copper on both continue to oxidize or does tim prevent oxidation from occurring?

The TIM does not prevent oxidation, but it significantly slows it down to the point that it is not an issue in practice.

 

[Syzygies]

Originally posted by: norcimo
2000 GRIT
Metal Polish
Polishing cloths.

One often sees the comment elsewhere that "you didn't need to go all the way to a mirror finish". What's the evidence on this, either way?

To make an analogy, people go nuts sharpening knives, yet there's a reason they sell serrated knives on late-night TV: For slicing, serrated helps, and a coarser sharpening job is also fine. For pushing a Japanese woodworking plane, pull out all the stops for the finest sharpening job one can manage.

The argument I believe is this:

It's human nature to always take things to the limit because "it sounds right" with no actual basis in fact for the behavior. We're using a bit of thermal paste, what "sounds right" to me is once the two pieces are flat, there's a limit to any advantage to finer grits. The holes in the bread don't matter when you're putting peanut butter in the sandwich. Lapping avoids serious gaps, but lapping to a mirror finish is just obsessive behavior.

I'm not asking for the hell of it, I plan to lap a Q6600 next month. Meanwhile, I'm all ears...

 

[norcimo]

Everyone is entitled to thier own opinion. Some are right, others are way off.. and you
are way off. The more you achieve a mirror finish, the better thermal conductivity. It's a known scientific fact.

Goto Tomshardware, take a look a some CPU water cooler blocks, and you will notice how nice and MIRROR FINISHED they are...what is the reason for this? you might ask.. do some research before you open your pie hole next time.

If you're going to do this work on your CPU and void your warrantee, at least do it properly and give your CPU all the thermal help it needs.

You will find in Tomshardware a table that shows CLEARLY thermal resistance and the ones with the least resistance are in FACT the MIRRORED FINISHED PRODUCTS.

 

[Syzygies]

Originally posted by: norcimodo some research before you open your pie hole next time.
...
You will find in Tomshardware a table that shows CLEARLY thermal resistance and the ones with the least resistance are in FACT the MIRRORED FINISHED PRODUCTS.

All I did was voice an alternate theory for why people might pointlessly go further than necessary, I never stated it as fact. I have no evidence either way. You would be more convincing if you sounded less defensive. The "Pie hole" comment reads to me as a translation of "Oh frig, have I been wasting an hour each lap?" after the denial circuits fire.

One often sees the unsupported comment on the web, "you didn't need to polish that far".

The best overclocking motherboards allow one to disable the protection offered by Vdroop, yet as Kris Boughton's writings make clear, it is idiotic to do so. The manufacturers are pandering us, "The customer is always right", by offering the BIOS setting.

Similarly, the best water cooling blocks had better sport a mirror finish, or they won't sell. But a peacock's feathers don't further reproduction, they just further seduction. It is a mistake to assign causality to the coincidence that the best blocks are sold with a mirror finish, if there are competing explanations.

There are no products sold lapped to 1200 grit. Either the manufacturer doesn't bother to lap at all, or laps to a mirror finish. We all expect products that have been lapped to perform better, but we can't infer from this data whether or not going to a mirror finish helps.

Who here has lapped to 1200 grit, done extensive testing, removed the cpu and cooler and lapped to a mirror finish, done further testing?

 

[child of wonder]

Awesome work on the lapping job. I tried lapping a heatsink once a few years ago and it didn't really seem to help and I'm sure I didn't do the best job.

 

[aigomorla]

Originally posted by: norcimo
The results are very true and worth it, (if you dont mind violating the CPU warrantee)

what seems to be happing is that many users are not lapping properly and are skipping vital procedures.

I have a Q6600 OC @ 3.2 GHZ, this CPU under load is exactly 92F. GRANTED, im using a chill tec ultra cooler, but never the less, the same cooler without lapping, was 105F Idle.

proper lapping takes time and patience.

this is material needed to do a perfect (NASA) mirror finish:

400 GRIT
600 GRIT
800 GRIT
1000 GRIT
1500 GRIT
2000 GRIT
Metal Polish
Polishing cloths.
High speed dremel. with metal polish attachment
If done properly, you can achieve perfect results.

And yes, it is messy, but the more black the metal polish becomes, the more glassy the surface will be. That's the key, I hand polished it well over 8 times, until it was perfect!

I used the Dremel to remove clearly visable polish lines lift by the sanding procedure, then again and again I polished using one cloth with paste, i rubbed the CPU against it (circular motions), and another cloth to clean up the black goo, or "copper" to buff it and get perfect results.

The things i bolded is a no no no.

You dont want to use polish because polish puts a coat ontop to preserve the shine. This will hurt performance.

Polish is a nono, after you lap, you wipe with iso, and then mount add tim and your ready to go.

 

[jtleon]

I would only ask that you include the ambient temperature in your charts. By measuring ambient with core temps, you may discover your charts may be under reporting your achievements.

Regards,
jtleon

 

[Syzygies]

Originally posted by: norcimo
Everyone is entitled to their own opinion. Some are right, others are way off.. and you
are way off. The more you achieve a mirror finish, the better thermal conductivity. It's a known scientific fact.

Here's a thread at thetechrepository by AnandTech's own Kris Boughton on lapping:

How to Lap the Intel LGA775 IHS Flat

He recommends stopping at 600 grit based on experiments both ways, and invites anyone to provide data contradicting this recommendation:

Now continue with the 600 grit just to get a nice, smooth top. We're not looking for a mirror here. You will quickly reach the point of diminishing returns where no amount of continued sanding will yield any further quantifiable result. This is why I consider sanding with the 1000 grit paper as optional. Some people like to, I personally believe its a waste of time (and I can prove it to you). Again, for those that think this look bad. Go ahead and lap yours to a mirror then show me your gain....nothing. We're looking for two things here: 1) copper, 2) it's flat.

 

[graysky]

Originally posted by: aigomorla

Originally posted by: norcimo400 GRIT
600 GRIT
800 GRIT
1000 GRIT
1500 GRIT
2000 GRIT
Metal Polish
Polishing cloths.
High speed dremel. with metal polish attachment
If done properly, you can achieve perfect results.

And yes, it is messy, but the more black the metal polish becomes, the more glassy the surface will be. That's the key, I hand polished it well over 8 times, until it was perfect!

I used the Dremel to remove clearly visable polish lines lift by the sanding procedure, then again and again I polished using one cloth with paste, i rubbed the CPU against it (circular motions), and another cloth to clean up the black goo, or "copper" to buff it and get perfect results.

The things i bolded is a no no no.

You dont want to use polish because polish puts a coat ontop to preserve the shine. This will hurt performance.

Polish is a nono, after you lap, you wipe with iso, and then mount add tim and your ready to go.

Agreed, avoid metal polish like the plague. Remember, you aren't waxing a car here, your goal is a flat surface, not necessarily a reflective one!

 

[Idontcare]

Originally posted by: norcimo
Everyone is entitled to thier own opinion. Some are right, others are way off.. and you
are way off. The more you achieve a mirror finish, the better thermal conductivity. It's a known scientific fact.

This is true on a theoretical level, but the use of TIM eliminates the delta between "perfectly flat and atomically smooth" versus "flat and smooth enough" results in practice.

In practice the pursuit of achieving a mirror surface comes at the price of "over handling" the IHS and causing the surface itself to not be as flat is it otherwise would be. (the corners end up getting rounded off, reducing the total IHS surface area available for mating to the HSF surface area, total heat transfer rates decline albeit only ever so slightly)

Remember the sole purpose of lapping is to minimize the maximum thickness of the TIM at any given spot across the IHS/HSF interface.

A perfectly flat and atomically smooth surface is only necessary if you intend to not use TIM and want to simply mate the two opposing metal surfaces together by simple compression alone.

And for what its worth I have done that (not used TIM after lapping) and I got the same results with 800 grit as I did with 2500 grit in comparison to TX2, AS5, TX1, and Ceramique.

Also there is a thread over at XS forums (can't find it at the moment) where the dude did a systematic study of assessing the temps of his system after every iteration of lapping grits from 220 grit to 2000 grit and he found the minimum temps came with 800 grit, after that his temps went up with subsequent lapping efforts with the higher grits. Obviously he lost his flat surface in his pursuit of making a mirror reflective surface.

 

[OfficeLinebacker]

Copper doesn't oxidize fast like aluminum. With Al a microcoating of Aluminum oxide forms instantly with contact with air. Copper will gradually form a patina but not instantaneously as it's not as reactive.

However nickel oxidizes the least of all, that's what galvanized/outdoor fasteners are coated with.

 

[norcimo]

Intel Q9300 Lapped OC 3.0G . 90F Constant underload with Ultra Chiltec cooler.
400,600,800,1000, 2000,metal polish..PERFECT


pic1
pic2
pic3
pic4
pic5
pic6
pic7
pic8
pic9
pic10
pic11
pic12

 

[Idontcare]

Originally posted by: graysky

Originally posted by: aigomorla

Originally posted by: norcimo400 GRIT
600 GRIT
800 GRIT
1000 GRIT
1500 GRIT
2000 GRIT
Metal Polish
Polishing cloths.
High speed dremel. with metal polish attachment
If done properly, you can achieve perfect results.

And yes, it is messy, but the more black the metal polish becomes, the more glassy the surface will be. That's the key, I hand polished it well over 8 times, until it was perfect!

I used the Dremel to remove clearly visable polish lines lift by the sanding procedure, then again and again I polished using one cloth with paste, i rubbed the CPU against it (circular motions), and another cloth to clean up the black goo, or "copper" to buff it and get perfect results.

The things i bolded is a no no no.

You dont want to use polish because polish puts a coat ontop to preserve the shine. This will hurt performance.

Polish is a nono, after you lap, you wipe with iso, and then mount add tim and your ready to go.

Agreed, avoid metal polish like the plague. Remember, you aren't waxing a car here, your goal is a flat surface, not necessarily a reflective one!

Looks like you guys advice went well heeded.

 

[bryanW1995]

lol. you beat me to the punch idc.

 

[aigomorla]

Originally posted by: norcimo
Intel Q9300 Lapped OC 3.0G . 90F Constant underload with Ultra Chiltec cooler.
400,600,800,1000, 2000,metal polish..PERFECT


pic1
pic2
pic3
pic4
pic5
pic6
pic7
pic8
pic9
pic10
pic11
pic12

*sigh* the thing i bolded is a very big no no. Read couple posts above this.

 

[norcimo]

These were my results. my CPu mirror finish works EXCELLENT. My surface is nearly perfect, down to razor sharp flat. I used my jobs mag scope AND visually check the surface, Its nearly perfect!, only one corner of the chip is bearly off..which infact has no bearing.

The polish used has a chemical that protects the surface....guess what..that chemical is also found in the composition of the heat sink compound im using.. called silica.

My cpu runs at 50C TJmax under load: 50, 50, 52, 51., and this chip for being 45nm, still runs hot, when people are getting 57-63C at the junction using the stock fan, I think my method proved itself.

 

[Syzygies]

Originally posted by: norcimo
do some research before you open your pie hole next time.

Originally posted by: norcimoI think my method proved itself.

Hey, me again.

You've demonstrated that your method beats not lapping. You haven't compared your steps at the end to, say, an unpolished 800 grit lap. Others have.

I mean no ill will, I'm just lazy. I'd like to save the second hour of lapping if there's no point to it. The research indicates it may not help, and it can hurt.

Imagine that this was a tax forum, and you were sharing a method to save $1000 on each of our taxes. We'd all love you for it. But if your last step was "...and burn a fifty dollar bill", you have to be ready for someone to post "You know, you could hang on to that fifty...". They'd mean well, and I'd sure want to hear it.

 

[Idontcare]

Originally posted by: Syzygies

Originally posted by: norcimo
do some research before you open your pie hole next time.

Originally posted by: norcimoI think my method proved itself.

Hey, me again.

You've demonstrated that your method beats not lapping. You haven't compared your steps at the end to, say, an unpolished 800 grit lap. Others have.

And that is the critical point that goes against norcimo's claim that he has proven much of anything here.

All you compared was finished product to untouched product. Without assessing the impact of any of the individual steps that go into creating the finished product you cannot arbitrarily claim that it was the very last steps that made all the difference.

For all you know (and as others have suggested) those very last steps actually did more harm than good.

 

[norcimo]

All I did was post what works for very well for me. I don't need to demonstrate results on 800 lapping, everyone knows what they are. That information is all over the internet on many forums.

A comment was made that mirror lapping is overkill and obsesive, my following post reflect that arguement as being vague. And that lapping to mirror finish is not without it's rewards. You can use this or leave it..I really don't care.. but there are those that might think otherwise. My results speak for themselves. Included are links with pics showing my setup. You can believe it or you can dismiss it.

Most top water cooling blocks all have mirrored finish, there is a reason for that, and I stated that too...so it is, what it is. Good luck to you all.

This is my OC CPU at 400.31 X 7.5, which is an FSB of 1600.

Pic 1
Pic 1