Sign in with
Sign up | Sign in
Your question

lapped my ultra-120 extreme (pics and temp results)

Last response: in Overclocking
Share
May 1, 2007 9:24:15 PM

When my Ultra-120 X and I have to say I'm a little puzzled. The base where it should contact the heat spreader is not smooth at all, it's actually grooved! You can see a scratch which is where I gently ran my thumb nail over the surface; I could feel the rough edges.

Have a look for yourself:


Anyway, others encouraged me to lap it which I've never done before. After wrestling with the idea for a couple of days as well as reading many articles/guides, I decided to give it a go. $20 worth of sandpaper, a $2 piece of flat glass, and 4 hours of careful work (and sweat) later, I was left with a pretty darn flat HS. You can see by the pictures that this particular one was quite concave instead of being flat which isn't good for keeping contact between the HS and IHS of the CPU.





Did it work you're probably wondering. The temp data as measured in speedfan.exe for a ~1 h x264 encode (uses all 4 cores with a CPU load of >99 %). I had speedfan log the temps (which it does every 3-4 seconds) and I averaged the whole data set per core for the 2nd pass of the 2-pass encode (the 2nd pass is the most CPU intensive). Room temp for both experiments was ~23 °C. By the way, I added a constant of 15 to each core in speedfan since it incorrectly displays temps for quads by 15 °C.

System specs: Q6600 @ 9x333=3.01 GHz (stock voltage), P5B-Deluxe in an Antec p182 case.

[code:1:7ddd965cc2]Before lapping the HS:

Core 0: 66.9
Core 1: 66.4
Core 2: 60.6
Core 3: 60.6

After lapping the HS:

Core 0: 64.9
Core 1: 64.4
Core 2: 59.0
Core 3: 59.4

Delta:

Core 0: 2.0
Core 1: 2.0
Core 2: 1.6
Core 3: 1.2[/code:1:7ddd965cc2]
May 1, 2007 9:58:28 PM

Here are some pics of the process (they're all in one 1.2 meg file - sorry for your analog modem folks). Anyone tell me what I did wrong?

May 2, 2007 1:56:43 AM

Everything is good, see the first post of the thread which I edited.
Related resources
a b K Overclocking
May 2, 2007 3:53:38 AM

glad u got it working...but your images are not working. please compress them with jpg, if you already did this crop and scale them to about 1200 x XXX thats more then enough.
May 10, 2007 3:20:55 PM

I started lapping my heatsinks when I got my 3.2GHz P4. I originally used rough(200-600 grit) off the shelf sand paper and then used a metal polish to smooth it down the rest of the way. That technique was awful compared to what I use now, check this site's products http://www.easypckits.com/products/premiumlk/. Their premium kit goes from 400 grit to 10 micron paper and then they have a diamond compound that's about equal to 10,000 grit paper. I go one step further using an optical grade red rouge that's about 0.7 microns, 60,000-80,000 grit. The finish is perfect, you probably won't see much gain beyond a 2000 grit finish but that mirror finish does look really nice. If you ever have the urge to you can also try out lapping the CPU. I've done it on my P4 but not yet on my C2D, I'll wait till their cheap enough to replace if I ruin it.
May 10, 2007 4:20:08 PM

Glad it worked out for you.

BTW, there is a minor problem with the number in your core 0 delta table.

It should be 2.0 not 4.1. Or the numbers for core 0 in the other two tables are bad. :D 

Anyway, it's always good when things work out in the end. Keep in mind that the temps hopefully will drop another couple of degrees over the next week. :) 
May 10, 2007 7:11:31 PM

@Gneisenau - thanks for catching that math error. I fixed the post. You can see the numbers after I lapped the CPU which have greatly improved over what's posted in this thread. Here is the thread about lapping the Q6600.
a c 131 K Overclocking
May 25, 2007 1:00:45 AM

Could your nice results have been due to the proper amount of as5 instead of the lapping?
May 25, 2007 4:51:48 AM

If you want a mirror finish then go as high of a grit as you want. If your strictly going for performance then you may as well stop at 600grit. It has been proven time and time again that after 600-800 grit there will be no measurable gains in performance from lapping.

And you can buy SINGLE sheets of sandpaper from ACE Hardware for about 50 cents a sheet. So for less then $2 you can lap your HS.

I only used 200, 400 and 600 grit and it looked like this:

I got an immediate 8C drop in load temp.
May 25, 2007 11:45:23 AM

Quote:
If you want a mirror finish then go as high of a grit as you want. If your strictly going for performance then you may as well stop at 600grit. It has been proven time and time again that after 600-800 grit there will be no measurable gains in performance from lapping.

And you can buy SINGLE sheets of sandpaper from ACE Hardware for about 50 cents a sheet. So for less then $2 you can lap your HS.

I only used 200, 400 and 600 grit and it looked like this:

I got an immediate 8C drop in load temp.


How long did it take to lap your HS using only 3 grades of sandpaper? Did you use oil, or just water? Nice finish btw :wink:
May 25, 2007 6:55:35 PM

From tearing down to rebuilt, I took 2.5 hours. This was my first lap job and I took my time, inbetween beers that is.
May 25, 2007 7:33:42 PM

I use to have to polish steel for microscopes. The one thing that most uses over look is a perfectly ground plane for the sandpaper and micro cloth to reside on. The second mistake most users make is trying to lap going in both directions. Doing so rocks the part and round edges. I had tables with variable low rpm wheels which made life easy. Most all were a min of 1 micron, most down to 1/4.

You also want to change directions with every grit change and you don't go to the next till all scratches are gone.

??? Did you lap Both the CPU and HFS. If not this may be the reason for needing more AG5.
May 25, 2007 7:43:02 PM

Quote:
my first lap job
How much did she charge you for that? Sorry I just had to repeat that, sounded too funny when I read it lol
May 26, 2007 8:00:01 PM

Quote:
Did you lap Both the CPU and HFS. If not this may be the reason for needing more AG5.


I did both the CPU and HS. They both were concave. Both the CPU and HS looked like this:
May 26, 2007 8:31:19 PM

Whooow, that was a huge concave HFS. Some of my HFS were concave, but nothing like that.

You had your work cut out for you. But copper be soft is easy to work.

Nice work.
May 27, 2007 1:40:49 AM

Thanx. Part of the 4 hours it took was me learning how to do it/developing a system/technique. If I had it to do over again, I'm sure it would be less time.
May 28, 2007 5:20:45 PM

:cry:  None of the pictures in the thread are working except the very first one by the OP... darned free image hosting...

-mcg
May 28, 2007 10:30:21 PM

Quote:
Error: server cannot be reached or image not available (timeout: 10 seconds).

Image URL: http://img515.imageshack.us/img515/111/img0004ty1.jpg


I believe if you paste the URL to the pic you'll be able to see it. It think the problem has something to do w/ tg's hardware and not imageshack.
May 28, 2007 11:14:34 PM

Quote:
From tearing down to rebuilt, I took 2.5 hours. This was my first lap job and I took my time, inbetween beers that is.


Beer is a fine multi-purpose lubricant.

For lapping, of course.
June 11, 2007 6:31:49 AM

Do you have to use oil/water to lubricate or do can you just do it dry? If you need to lubricate, what kind of lubricant should be used?
June 11, 2007 7:01:01 AM

I've read lots of different tutorials oh how to lap a heatsink; most of them use a little dish soap and water. A few did it dry. And a few used mineral oil. I'd recommend using some sort of lube. It seems to help make things go more smoothly. (obvious?)
June 11, 2007 8:39:24 AM

Soapy water is easy. You can also use low odor kerosene to do it; I'd caution you against using mineral oil just because it's so hard to get rid of afterwards.
June 11, 2007 9:04:16 AM

Thanks for the advice, so where and how much soapy water should be applied? I just ordered that lapping kit mentioned earlier in this thread so naturally I'd like to know the proper method to use it :wink:
June 11, 2007 9:46:39 AM

The basic principle behind the liquid is not to lubricate, it's to keep the sandpaper from clogging up so it sands faster and continues to do so evenly.

I like to use a good packing tape to tape down the sandpaper to the glass, then take a cup of detergent water and just splash some down. After it gets really dark you can rinse it all off in the sink or with a hose then dump a little bit of detergent water on it again. It doesn't take much detergent, just a tiny squirt into the glass.

Trust me when I tell you a mirror finish is overrated, it has no thermal benefit to get it that shiney, it just makes double the work. Perfection in this is highly overrated, you will get 90% of the benefit in the first 5 minutes so long as you don't start out with too fine a sandpaper, 220 grit is about right if it's concave but otherwise 400 is a good start, and finish - if you don't want to use 800 or higher you aren't missing out on but a tiny fraction of a degree if that.

Some places sell a mixed pack of 3M wet/dry sandpaper that is much cheaper than $20, maybe about $5. Once I made the mistake of buying generic Chinese wet/dry from Harbor Freight. It was horrible quality, brittle when dry and fell apart immediately when wet plus the sand didn't even seen to stick to it, was like they used water soluble glue!
June 11, 2007 10:58:22 PM

Quote:
Thanks for the advice, so where and how much soapy water should be applied? I just ordered that lapping kit mentioned earlier in this thread so naturally I'd like to know the proper method to use it :wink:


From what I've read, front-to-back/back-to-front is the better technique since circular or figure-eight motions can introduce uneven surfaces; you have more control when you minimize the angular motions. I have always used soapy water (like 1 drop of dish soap in 1 liter or so of water). You just want to wet/moisten the sandpaper. Be sure you clean the sandpaper off with water/papertowels between lapping runs. I wouldn't recommend that you do it without. The water lubricates as well as immobilizes the metal partials. Don't worry about using grits higher than 800 or so... I think the most benefit you'll get out of lapping is a flat surface; you can make it smooth beyond say 600 or 800 grit, but as I understand it through reading, the extra mile doesn't give you much if any additional benefit. It does make it reflective, but who cares since it can't be seen once it's installed.
June 12, 2007 6:32:30 AM

Thank you very much for the advice, I'll save myself some time by just going for a smoothe surface and not going down to the ~40 microns my package came with. Once again, thanks for all the help :p 
June 12, 2007 4:31:30 PM

You've made some good comments here but you've also overgeneralized. Lapping is just one of many methods in the science of metallography and it is a highly developed science. For example, when you say:

Quote:
The basic principle behind the liquid is not to lubricate, it's to keep the sandpaper from clogging up so it sands faster and continues to do so evenly.


That is true but it misses some important points. If a soft metal like copper is ground without sufficient lubricant, the sandpaper doesn't just clog up. The copper that is left on the paper acts to work and deform the sample that is being ground. That's not a good thing when you're trying to develop a high thermal conductivity surface. In addition, when there is inadequate lubrication, the sample can become embedded with tiny pieces of the abrasive. The thermal conductivity of SiC is pretty high (~120W/mK) but it's WAAYYY below that of copper, so you don't want to be embedding it into your IHS nor your HSF base.

Quote:
Trust me when I tell you a mirror finish is overrated, it has no thermal benefit to get it that shiney, it just makes double the work. Perfection in this is highly overrated, you will get 90% of the benefit in the first 5 minutes so long as you don't start out with too fine a sandpaper, 220 grit is about right if it's concave but otherwise 400 is a good start, and finish - if you don't want to use 800 or higher you aren't missing out on but a tiny fraction of a degree if that.


Once again, you're generalizing. Search around the literature and you'll find examples of thermal interfaces that use no TIM and get excellent heat transfer. How? By preparing the surfaces to be very flat and free of deformation. Certainly a mirror finish is required for TIM-free interfaces.

There are numerous resources online that discuss the fundamentals of metallography but for the purposes of this thread, this one should suffice. Check out the comments in that reference WRT lubricant cooling, etc. It all applies to HSF grinding (lapping is not the correct term for sandpaper grinding).
August 5, 2007 2:27:01 AM

... ummhmm... when i get myself a heatsink i am going to check this out. i never knew such a thing was normal. this lapping process stuff actually works.

pretty cool. in a few days ( hopefully not more than a week) i will be posting something about my cpu heatsink to see if it needs lapping.

thanks for the enlightenment.
August 6, 2007 9:53:12 AM

I just wanted to showoff my Tuniq Tower lap job. It took my about an hour. I didnt take "before" pictures. Check out that mirror finish!! I used grit 320, 400, 800, 1000, and 2000 for that mirror finish. All that on a wet flat glass.

August 6, 2007 7:28:46 PM

cherie2298 - that is one shiny hs base. The real question now is what sort of temp decrease did it translate into for you? Also, have you lapped your IHS on you chip?
August 7, 2007 11:24:12 AM

Its a hot summer here in GA! Amazingly enough, the cpu temperature stays cooler than the case temperature. I think decreased 12+/- degrees while the CPU is over cloaked at 3.2Ghz.
I did lapped my CPU, no pictures cause its a pain in the butt removing the motherboard to take off the Tower 120.*edit* The CPU has a mirror finish too. Here is a pic of my cpu-z and speedfan

August 16, 2007 4:33:03 AM

Whats so funny ?
August 30, 2007 7:41:42 PM

Maybe I'm looking at that pic wrong, but does that say your fan speeds are over 1 million rpm?
August 30, 2007 8:06:51 PM

graysky said:
cherie2298 - that is one shiny hs base. The real question now is what sort of temp decrease did it translate into for you? Also, have you lapped your IHS on you chip?



Doesn't speed fan report temps 15C too low in some cases? That is way too low of temp, seeing as its only 10C or so above your ambient. I think your temps should be offset by 15C.
August 30, 2007 8:10:49 PM

cherie22984 said:
Its a hot summer here in GA! Amazingly enough, the cpu temperature stays cooler than the case temperature. I think decreased 12+/- degrees while the CPU is over cloaked at 3.2Ghz.
I did lapped my CPU, no pictures cause its a pain in the butt removing the motherboard to take off the Tower 120.*edit* The CPU has a mirror finish too. Here is a pic of my cpu-z and speedfan
http://img.photobucket.com/albums/v281/cherie22984/Untitled-1-1.jpg



sry, I meant to quote this in my above post.
August 31, 2007 10:26:48 PM

can anyone tell me how 2 overclock a amd opteron 2.2ghz 148
October 24, 2007 9:44:40 PM

I was wonfering why they give a nickel plating to the copper heatsink. Is it be to protect the HS from oxidizing?

So maybe everyone who's been lapping their HS's down to bare copper will have some sort of majot failure down the road, when their HS's turn green and fuse to the CPU IHS.

Thoughts? Anyone checked thermal performance 6-12 months after the lap? Maybe you should check it out, graysky?
If someone was really ambitious, they could take the HS off again and have a look at it, just to see if it has turned green. Maybe it wouldn't matter.

Cheers
October 24, 2007 11:57:41 PM

@General - we'll, I did it to my Athlon 3200+ like 3 years ago and it's still fine :) 
October 24, 2007 11:58:09 PM

General_Disturbance said:
Doesn't speed fan report temps 15C too low in some cases? That is way too low of temp, seeing as its only 10C or so above your ambient. I think your temps should be offset by 15C.


This is true and accounted for in the tests.
October 25, 2007 12:13:06 AM

graysky said:
@General - we'll, I did it to my Athlon 3200+ like 3 years ago and it's still fine :) 


a'lappin I be goin'!
November 11, 2007 10:27:02 PM

General_Disturbance said:
I was wonfering why they give a nickel plating to the copper heatsink. Is it be to protect the HS from oxidizing?





I was actually wondering the same thing. The oxidation prevention sounds like the most likely case. Nickel is probably a cheaper metal so I was thinking that it is cheaper to use in place of copper...but just slapping it over the copper won't save any money. So it must be to prevent oxidation? The high temperatures will drive the copper to oxidize faster than normal.

What I don't get is the fact that the nickel is in between the heat sink copper and the CPU. There should be a decrease in the heat conductivity using nickel, making it less efficient at pulling the heat away from the CPU. See my thermal conductivity numbers I put below at ~75C. Nickel is about 4.5 fold worse than copper.

What I am really wondering is if the major gains from lapping (besides flattening the surface) are actually due to the removal of the nickel surface to expose the copper surface. People are showing that a highly polished copper surface doesn't improve temps that much compared to a less refined copper surface. It looks like as long as there is a flat copper surface it seems to max out the cooling potentials.


Copper 396 (W/mK)
Aluminum 240 (W/mK)
Nickel 85 (W/mK)
November 11, 2007 11:09:36 PM

That's exactly what I was thinking too when I asked about it. Copper we all know is a very good conductor; I didn't think Nickel was at all and your numbers show that. That's great.

I think you're right, and that lapping helps for 2 reasons: 1) Increase in surface area contact (obvious), but 2) increase in thermal conductivity from exposing the bare copper.

But greysky (I think) said his several year old lapped HS's haven't showed oxidation and still work well, so, why the heck are they putting the nickel layer on in the first place??
November 12, 2007 12:36:08 AM

Wait a minute. Pure copper should oxidize quickly right? Copper is number 2 in heat conduction, so if it doesn't oxidize there is no point in Nickel layer. So it has to be to prevent oxidizing. Just look at the motherboard circuit. It's all covered by something.

The one whoever lapped, maybe it's an alloy and not pure copper? That way oxidizing rate should decrease...the Thermalright ultra 120 extreme is all aluminum btw and that's what greysky lapped.

While we're on heat conductivity, wouldn't a copper Thermalright ultra 120 be a lot better than an aluminum one???

EDIT: The Thermalright ultra 120 is all aluminum except for the base and heatpipes(?)
November 13, 2007 12:48:47 PM

Evilonigiri said:
Wait a minute. Pure copper should oxidize quickly right? Copper is number 2 in heat conduction, so if it doesn't oxidize there is no point in Nickel layer. So it has to be to prevent oxidizing. Just look at the motherboard circuit. It's all covered by something.

The one whoever lapped, maybe it's an alloy and not pure copper? That way oxidizing rate should decrease...the Thermalright ultra 120 extreme is all aluminum btw and that's what greysky lapped.

While we're on heat conductivity, wouldn't a copper Thermalright ultra 120 be a lot better than an aluminum one???

EDIT: The Thermalright ultra 120 is all aluminum except for the base and heatpipes(?)


Pure metal should conduct heat better than impure metal. Some heat sink makers tout that their copper is 99.998% pure. If anyone knows about copper oxidation rates...I would like to hear about it, so far I haven't found much. Yeah if it was an alloy it probably could oxidize slower or not at all, but that reduces conductivity on the other hand.

I was trying to do a little research in that using http://www.frostytech.com/ They have a good comparison chart with all of the sinks. I forget if it was on that site or not, but I remember seeing that copper vs aluminum didn't make that much of a difference in overall CPU temp with the same model. The overall design probably dictates cooling ability to where the metal variation is negligible (like surface area and air flow properties).

I know that aluminum goes to aluminum oxide fairly rapidly however. The heat conduction of aluminum oxide is far below nickel…I can’t find the data for what the conductivity is around a CPU temp but at 20C it is 28-35 (W/mK) so…at higher temps it will be even lower. I see the benefit to having nickel coating in that case. Copper eludes me though….

Ah, here is a good explanation from an engineer. I just found this in the middle of typing this up. http://episteme.arstechnica.com/6/ubb.x?a=tpc&s=5000956...


Edit: I found the aluminum vs copper heat sink on frostytech. Look at the Apack Zerotherm BTF90 vs BTF80. The only diff is that one is copper and one is aluminum. This is getting way off the subject but it looks like cooling via air flow has reached a bottle neck despite the metal.
http://www.frostytech.com/articleview.cfm?articleid=198...
December 3, 2007 6:05:32 PM

I was planning to Lap my 120-Ultra and here i found everything i need to know :) 

Sorry for not adding anything to the post, just want so say thanks for the "indirect" help :) 
December 4, 2007 5:19:57 AM

Always glad to help! Please post your results and remember that you'll get the best results if you lap both surfaces of the system (i.e. the base of the HS and the IHS on the CPU).
December 6, 2007 5:09:15 PM

I went yesterday to RenoDepot(montreal) to get the sandpaper, but no luck.
There are no Specs like 200/400/600, the only specs i saw was 80/100/120/180/200 top.

I guess i went to the wrong place :( 

About lapping the CPU, i will receive the quad any day soon (already shipped), but i'm scare, to be honest! :) 
I guess is part of a “personal process”, lets see in the future! O:) 

Anyways, as soon as i do it i will post results and my experience :) 
December 6, 2007 5:47:46 PM

Holy shnikies. You guys need an education in chemistry. Nickel > Copper (in cost), Nickel < Copper (thermal conductivity). Nickel is in place because of its very stable pure nature (does not corrode - oxidize - easily or quickly).

Quote:
Pure metal should conduct heat better than impure metal.

HORRIBLE blanket statement. Some alloys conduct heat better than pure metals. Do not make comments like this with zero knowledge of the subject matter.

Edit:
Kyle, BS Chemical Engineering, University of Cincinnati
!