glass

[Ugmore Baggage]

I'm often surprised by the assumption that all glass is perfectly flat.

Why is this? Take a pane of glass and apply pressure to a corner, you will see it flex. It's clear, so it is almost impossible to check for flatness.

Perhaps a mirror would be more useful, because you could look into it and correct for distortion somewhat by moving it to a flat area of whatever you're using to resist pressure.

[schismspeak]

Actually we dont assume it is perfectly flat.(Most of us anyway) We do assume it is the flattest thing we can get our hands on easily and we have recommended a mirror also.

[shiyan]

I think glass is also quite hard, so it provides a good base to lap on.

[Big_Baller]

Its just a lot more flat usually than things like tables and benches. Most of us would like some nasa grade tools but glass and mirrors work just fine.

[stan03]

and it also depends how thick the glass is and what it rests on... i think most of use glass that is sitting on another hard surface.... so the flexing isn't much of a problem

[slater3333uk]

Using normal window glass is a mistake it is NOT flat.

However for our purposes a mirror would be acceptible.

[Ugmore Baggage]

if not normal window glass, what is it you would look for?

[schismspeak]

A mirror, as he stated..

[9mmCensor]

Why wont cooling manufactures discover that people would be willing to pay a bit more for a good lapping job that is done with pro tools and is really good?

[Hookem]

Well Swiftech does...Their stuff is about as flat as you can get...I have never seen one of their heatsinks that needs to be lapped...And I have bought a bunch of em...Sure not like the "best" heatsink manufacturer's finish...I went to a glass company and they gave me a 10X12 piece of 1/2" tempered glass...Gotta give em props...Lonestar Glass on Bissonett here in Houston...That is what I use to lap anyting but Swiftechs...;)

[squeakygeek]

Originally posted by 9mmCensor
Why wont cooling manufactures discover that people would be willing to pay a bit more for a good lapping job that is done with pro tools and is really good?

Would they? Just because you would, or just because many people on these forums would, doesn't mean most people would.

[hannas]

Wouldnt most people buying these high performance heat sinks be the kind that are on these forums? They are an enthusiast product...im sure he meant 'most people' who would ordinarily buy them, to mean us - the enthusiasts.

[ZaP37]

Originally posted by 9mmCensor
Why wont cooling manufactures discover that people would be willing to pay a bit more for a good lapping job that is done with pro tools and is really good?
But you'd still have the problem of chip makers producing convex and concave chips...

[huneycutt]

The surface of the CPU die is intentionally made uneven with peaks and valleys to increase the heat sink compund efficacy in transferring heat to the HS.

I bought a 2500 Barton retail boxed and the HS now has a copper base and it is rough as a cob. I took the pink pad off and used Arctic Silver and left the HS as is and it cools just fine OC'd 2.0GHz. Wonder if the rough surface doesn't offer more places for the coupound to penetrate therby increasing heat transferrence efficacy die to HS, HS from die.

You want to see a HS with a surface so slick you can shave in it?
http://www.nexfan.com/evcucoacofrs1.html

Maybe that's whay it got such good reviews for it's time.

Once you use the pink pad it penetrates the pits in the die and you can never get it all out so go with white or Arctic Silver compound.

[shiyan]

uhh... actually the best contact would be with no thermal paste, no peaks and valies, and a perfect contact between the 2 surfaces. Of course this is very hard to achieve, which is why we have thermal paste.

I do remember seeing a picture of someone who lapped a P4 IHS and Alpha PAL 8942 so well that the P4 basically stuck to the heatsink without any thermal paste, purely from the perfect contact.

doesn't make any sense to rough up your contacting surfaces to fill it up with thermal paste.

edit:

found it:

http://www3.sympatico.ca/keith_lee/lapped.jpg

that's as close to perfect as you'll get probably.

[omaticrail]

Originally posted by 9mmCensor
Why wont cooling manufactures discover that people would be willing to pay a bit more for a good lapping job that is done with pro tools and is really good?

We would all love a specfriggintacular (TM) finish, and some would pay more for it, but how much more? Even knowing that a product has such a desireable quality, many people usually go for the--otherwise identical--cheaper sale, knowing that (a) it may be good enough for us, and (b) that if it's not, we can lap it ourselves for less total $. The company selling the superior finish at a higher $ usually loses out.

Think about how much time you spend lapping, and how many $ that would add to the price of the final product. Figure qualified/trained labor at $15/hour, all the right tools and workspace (overhead), producing a fine finish (that would still be debated in forums) in 45 minutes per block. Now add all that up, plus $.50 - $1.50 for consumable materials like abrasives, etc., then add those mark-up % that come at every step. It could add as much as $25 to the price of a block. Whether or not that's a "good deal" depends on how valuable your time is, and whether or not you're satisfied with the results you paid extra for.

[huneycutt]

"doesn't make any sense to rough up your contacting surfaces to fill it up with thermal paste."

I remember reading the reason they have the "pits" is becaue it allows the compound to penetrate and stick into the die to give the heat a better way out. (Don't know if I said that right, but you catch my drifit.) I do know they intentianally make the insides of engine blocks rough is to allow better heat dissipation into the cooling oil & liquid. I'm not a thermodynamics engineer but I understand the theory.

I did my lapping homework long ago and read all about it and IHMO it's more trouble than it's worth. There are millions of PC's out there that have been running fine for years with the stock cooling rigs and never have had a heat problem. Of course, most of those aren't OC'd.

Hey, a 2500 Barton retail boxed costs $92. bucks and to me if it lasts five years it's a throw away and by then I'll want to upgrade anyway.

I build 'em, fire 'em off and run 'em and if they OC cool and stable I don't worry about it. Hats off to those who use their time to make this or that run better or faster or cooler, but me, I just ain't got the time nor the desire. Different strokes for different folks. . . . .

[huneycutt]

Exactly my point and we're on the same page because as a chemical consultant I get $50. bucks an hour and I can work four hours and buy a new MB and CPU, so why worry about it?

Like I said, different strokes for different folks because I'd rather be swimming, fishing or playing golf than lapping a heat sink.

My other hobby is cooking so guess a lot of people think I'm crazy spending half a day cooking a gourmet prime rib dinner and in the same respect they enjoy lapping their heat sink and tinkering with their computer then going out for a Burger or Pizza.

I'll say this then wash up for chow. I am very thankful for computer tinkerer-perfectionists because one of the will come up with something better and put it on the market and we can all enjoy and take advantage of their fruits of their labor. Our computers will work better and they will be rich and everybody will be happy which is what's it's all about. BE HAPPY AND JAM ON IT!

[schismspeak]

Originally posted by huneycutt
"doesn't make any sense to rough up your contacting surfaces to fill it up with thermal paste."

I do know they intentianally make the insides of engine blocks rough is to allow better heat dissipation into the cooling oil & liquid.


Yes, but when you have die to metal contact, it is better to not have pits, so the heat can be tranfered into the hs more efficiently, the hs then has all the fins and does basicaly what the roughness of an engine block does.

[shiyan]

Originally posted by huneycutt
I do know they intentianally make the insides of engine blocks rough is to allow better heat dissipation into the cooling oil & liquid. I'm not a thermodynamics engineer but I understand the theory.

that's true, but they make it rough for a different reason - turbulence.

if the cooling channels in the engine block are extremely smooth like the inside of a cylinder, the flow inside those channels would be laminar and not turbulent, which is bad for heat transfer between the metal and the cooling fluid. if the channels were really smooth, what you'll see is that the fluid touching the sides move the slowest, and insulate the fast moving fluid in the center of the channel from the metal.

with the rough channels in the engine block, the fluid flow becomes turbulent. there is no boundary layer clinging to the metal surface, and pretty much the entire cross section of the fluid travels at the same speed. heat transfer from the engine block to the fluid is much more efficient this way.

we are not trying to do this with the interface between the CPU and HSF or IHS and HSF.

we want to maximise the metal/metal contact area, because the heat transfer for metal/metal is a lot better than metal/TIM/metal.

we use TIM because the heat transfer for metal/TIM/metal is much better than metal/air/metal.

so the perfect cenario would be for there to be no air gaps and no TIM between the IHS or CPU and the HSF. This is only possible if the two surfaces touching match each other, and it is a lot easier to have 2 flat surfaces match each other than 2 ridged surfaces match each other.

that's why I said it made no sense to roughen up the contact areas to fill them with TIM.

[huneycutt]

Understood. However. when using primitive metal clips and no exact calibration pressure measuring device between the die and the HS obtaining a perfect "meeting of the two" surfaces is impossible, so like the man said, "there ain't no perfect world."

That's why I use Arctic Silver and mash them two babies together and let 'em rock n' roll. :cool:

[squeakygeek]

Originally posted by huneycutt
Understood. However. when using primitive metal clips and no exact calibration pressure measuring device between the die and the HS obtaining a perfect "meeting of the two" surfaces is impossible, so like the man said, "there ain't no perfect world."

That's why I use Arctic Silver and mash them two babies together and let 'em rock n' roll. :cool:

You can't get it perfect, but you can get it pretty darn good by lapping, and it will work better too, even though it's not perfect.

[CrystalMethod]

Originally posted by shiyan
uhh... actually the best contact would be with no thermal paste, no peaks and valies, and a perfect contact between the 2 surfaces. Of course this is very hard to achieve, which is why we have thermal paste.

I do remember seeing a picture of someone who lapped a P4 IHS and Alpha PAL 8942 so well that the P4 basically stuck to the heatsink without any thermal paste, purely from the perfect contact.

doesn't make any sense to rough up your contacting surfaces to fill it up with thermal paste.

edit:

found it:

http://www3.sympatico.ca/keith_lee/lapped.jpg



That's actually not the inital post. I couldn't find the original one but I damn well saved a pic of that lapping job. As for the glass being flat and all. http://www.glassresource.com/sneakpeek/sample15.htm
I once read somewhere that 85% of the glass being manufactured today is "float glass". It's plenty flat enough for what we use it for.


that's as close to perfect as you'll get probably.

[BigSmokey]

All this when the man asked about glass...I find beer is the best conductive material. I pour beer into my convex face and I end up getting much more contact with conducive women, often only to wake up the next morning wondering why her face looks like its been laped with 50grit paper.

[squeakygeek]

Originally posted by BigSmokey
All this when the man asked about glass...I find beer is the best conductive material. I pour beer into my convex face and I end up getting much more contact with conducive women, often only to wake up the next morning wondering why her face looks like its been laped with 50grit paper.

haha that's great

[huneycutt]

"often only to wake up the next morning wondering why her face looks like its been laped with 50grit paper."

Like Jimmy Buffett said, "I never went to bed with an ugly woman but I sure woke up with a lot of 'em . . . . "

[huneycutt]

You can't get it perfect, but you can get it pretty darn good by lapping, and it will work better too, even though it's not perfect. [/B][/QUOTE]

Got a question for ya'. Let's say I have a good, stock, copper cooler (forget what brand) and I'm seeing 40C idle and if I lap it how much will the idle temp go down?

Forget all the particulars like how good a lap job I do and what compound I use and all that. Just an average of what cooler temps one could expect. Two degrees cooler, five degrees. ten degrees?

[CrystalMethod]

It varies. Some people don't get any decrease in temps. I once lapped a stock Intel CPU and got a 7°C drop in temps. But when I lapped one of my Alpha's, I didn't get any drop. Keep in mind the Alpha had a good finish to start with, so I wasn't really surprized to not see any difference.

[huneycutt]

See, that's what I'm 'fraid of doing all that work and accomplishing nothing. I have two of these EverCools and, I do declare, the finish on it is almost mirror like.
http://www.nexfan.com/evcucoacofrs1.html

Like most of us geeks, I'm always looking for better so I orderd a SLK900A from SVC at the ten bucks discount and it should be here today so we'll hook it up and see what if any improvments in temps it shows.

Thanx for the feedback and have a great day. OldBird

[BigSmokey]

Its a labor of love. If its that perfect, it probbly wont help, but the only way to know for sure is to bust out the sandpaper and waxon/waxoff grasshoppa

[PachManP]

it's like tricking out your car. You can pay to have somebody do it or you can do it yourself because you enjoy doing it. And you do hours and hours of work to shave off tenths of a second or tenths of degrees and it all comes down to whether you have something else you'd rather do.

[insulglass]

Originally posted by huneycutt
Got a question for ya'. Let's say I have a good, stock, copper cooler (forget what brand) and I'm seeing 40C idle and if I lap it how much will the idle temp go down?

Forget all the particulars like how good a lap job I do and what compound I use and all that. Just an average of what cooler temps one could expect. Two degrees cooler, five degrees. ten degrees? [/B]

I sell a lot of heat sink lapping kits, so I have quite a bit of feedback regarding this subject. This isn't an advertisement, I'm just establishing that I am in a position to hear the results of lots of lapping adventures.

Lapping typically provides a few degrees of cooling. I have actually plotted this on a bell curve with self-reported feedback, and 80% of the people who lap achieve between 2.5 to 3.5 degrees Celsius improvement. Note that this is NOT a scientifically controlled data set; however, my personal experiences actually follow the results very closely. I have seen as low as 1 degree improvement, and as high as 9, but these are pretty exceptional cases outside the norm I should think. I have never had someone tell me that lapping did not provide SOME improvement to their overall CPU cooling. More importantly, no one has ever asked for a refund because they were dissatisfied! :D

Lapping probably isn't as difficult as most might think. IMHO, it's more messy than hard to do. You should be able to put a mirror finish on a moderately sized heat sink in about an hour or maybe less if you are concentrating instead of watching TV the way I do when I am lapping. I will generally go through 7 to 9 grades of sandpaper to achieve my results, ending up at 2000 grit or 10 micron depending on what I lap with.

What surprises me is this... if a new TIM compound comes out that improves cooling by a degree or two, the overclocking community jumps all over it and the manufacturers sell tons of it. But, lapping also will typically provide the same results and most are reluctant to try it. :rolleyes:

Dave
insulglass@easypckits.com

[insulglass]

Originally posted by PachManP
it's like tricking out your car. You can pay to have somebody do it or you can do it yourself because you enjoy doing it. And you do hours and hours of work to shave off tenths of a second or tenths of degrees and it all comes down to whether you have something else you'd rather do.

True. My time is cheap to me. Other people get charged quite a bit more.

Dave
insulglass@easypckits.com

[Susquehannock]

Originally posted by Big_Baller
Its just a lot more flat usually than things like tables and benches. Most of us would like some nasa grade tools but glass and mirrors work just fine.

Yeah, if I went back to CAT to use their Granite stone
to lapp my heatsink they'd think I was crazy.
A piece of glass is good enough for home use. We aren't
doing rocket surgery afterall. ;)

If you need a lapping kit I'd humbly recommend buying
from EasyPCKits (http://www.easypckits.com/) . Everything you need at a reasonable price. :)