Mirror finish IHS

[996GT2]

With both the Freezer 64 Pro and the IHS lapped with 280, 400, 800, 1200, 2000, and 2500 grit sandpaper (Kit at SVC includes 24 sheets and a flat plexiglass surface to lap on), temps have dropped some 7-8C; idle is now in the low 30C range at 2.75 GHz with 55% fan speed on the freezer and all case fans.

The results speak for themselves, but if you're still not convinced, perhaps a pic will help

http://s78.photobucket.com/alb...w&current=DSC04020.jpg

 

[aigomorla]

Originally posted by: 996GT2
With both the Freezer 64 Pro and the IHS lapped with 280, 400, 800, 1200, 2000, and 2500 grit sandpaper (Kit at SVC includes 24 sheets and a flat plexiglass surface to lap on), temps have dropped some 7-8C; idle is now in the low 30C range at 2.75 GHz with 55% fan speed on the freezer and all case fans.

The results speak for themselves, but if you're still not convinced, perhaps a pic will help

http://s78.photobucket.com/alb...w¤t=DSC04020.jpg

Can i just say i told ya so??? :laugh:

 

[Sylvanas]

Sweet nice job any change with load temps?

 

[Navid]

Instead of looking at random objects through the lapped surface, look at a high-resolution grid.
Then, you can tell if the surface is flat or not.

Just because a surface is shiny does not mean that it is flat.

You can create a grid in any simple drawing tool or even word processor and print for this purpose.

This is an example.
http://i259.photobucket.com/al...id200/FlatnessGrid.jpg
If the surface is not flat, the lines will not show as straight lines.
You can see the lines are curved at the edges of the sink, showing the surface is curved at the edges, which happens with lapping.

 

[Comdrpopnfresh]

lapping in truth only aids the thermal interface. a good paste can compensate for no lapping, and machine lines. I read some report somewhere where they used a vrillo or something and scratched up the hsf and ihs and got the kind of temp drop people normally get from lapping...

 

[MadScientist]

OP,
Your room must be mighty toasty. My room is 20 C and my oc'ed 2.70 Ghz Opteron 170 is idling at, according to Core Temp, 24 C with a lapped to 800 grit AC Freezer 64 Pro HS and an unlapped CPU IHS. If your 2.75 Ghz X2 3600+ was idling at ~40C before you lapped it and the HS you most likely had it installed incorrectly. I haven't read of anyone complaining about the AC Freezer 64 Pro's base not being flat. I lapped mine because the base had deep machine lines, but didn't see any decrease in temps.
Lapping your HS or IHS to a mirror finish has been discussed here and at other forums many times. Do a search of the forum on lapping and you will find numerous threads.

 

[VinDSL]

Originally posted by: Comdrpopnfresh
a good paste can compensate for no lapping, and machine lines...

AKA Thermalright

 

[Elcs]

I am very much considering lapping a heatsink despite all this "yes it works, no it doesnt" debate.

My X2 6000+ undervolted by 0.05V at stock speeds is idling at ~50C (CoreTemp & Everest) with the stock heatpipe cooler, fan running at 25%, using the stock thermal paste. Its still bloody noisy but as I am currently not stressing system, I am not concerned.

I have an Ultima 90 with Tunic TIM and a new fan loadout coming soon. I am tempted to try lapping my stock heatsink and using AS5, reseating the heatsink and monitoring the drop. I know that theres two big flaws here: 1) the TIM change and 2) the way stock paste is pre-applied is more than likely inefficient. It is a possible little project in the middle of waiting.

I am not so sure that going through all the different grits like the OP did is the most effective way. I love shortcuts and whilst I am impressed with the finish they have achieved, surely you can reduce the amount of sandpapers you went through.

Are there any good lapping guides on the net? Is it easy to do? Looks relatively inexpensive. Does it take a long time to do?

I am sure that lapping to some degree should provide better thermal interfaces and provide some sort of temperature difference, surely tangible. Perhaps the IHS does not make great contact with the cores? Maybe that is our next problem when it comes to thermal interfaces and such, improving mating surfaces and reducing temperatures.

 

[aigomorla]

Originally posted by: Navid
Instead of looking at random objects through the lapped surface, look at a high-resolution grid.
Then, you can tell if the surface is flat or not.

Just because a surface is shiny does not mean that it is flat.

You can create a grid in any simple drawing tool or even word processor and print for this purpose.

This is an example.
http://pics.bbzzdd.com/users/Navid/FlatnessGrid.jpg
If the surface is not flat, the lines will not show as straight lines.
You can see the lines are curved at the edges of the sink, showing the surface is curved at the edges, which happens with lapping.

hehe what about this?

http://i125.photobucket.com/al...aigomorla/IMG_0703.jpg

Originally posted by: Comdrpopnfresh
lapping in truth only aids the thermal interface. a good paste can compensate for no lapping, and machine lines.

errrr...

1. a good lapping will allow less TIM to be used per application, and more even metal on metal contact. which is greater then tim.

2. If the ihs is concave or convex, how will tim fill a gap thats now become a macrobubble instead of a microbubble.

3. Sorry pal, i have yet to see 1 quadcore get = temps all across the cores that WASNT lapped.

I have done a = across all cores 3 times, and all 3 times my IHS were lapped.



What suprises me the most is how people get worried about losing there warrenty on the cpu. But overclocking the cpu itself invalidates warrenty, and i think when intel is pulling the cpu apart to find out why it died... if they see you fried it, they'll know if you overclocked it or not.

Also Intel wont honor warrenty unless you send them a dirty stock heat sink. so all you aftermarket sink owners, you got OWNED if your looking for warrenty.

 

[Navid]

Originally posted by: aigomorla

Originally posted by: Navid

hehe what about this?
http://i125.photobucket.com/al...aigomorla/IMG_0703.jpg

Have you ever seen those mirrors in entertainment parks like Disneyland that make you look really skinny and tall or very short and fat?
Those mirrors have very shiny surfaces. But, they distort the image because they are not perfectly flat.
If you hold a grid in front of one of those mirrors, you can see the distortion clearly in the reflected grid.

Your picture shows a very shiny surface. Is it flat? Use a grid to find out!

 

[Elcs]

Well, this was my effort for this afternoon.

Just picked up a spare heatsink lying around the house, slapped a small sheet of 2000 grit on the top of a hardback book (with cover on) and started lapping away using figure 8's and no pressure, letting the weight of the sink work its magic.

As you can see, my photography skills are about as evolved as a 2 year olds but the shine is coming out and it looks scratched to hell. I think from the lines that it looks rather flat despite the scratches.

Now, my method isnt one Id recommend but just to get a feel of what to do, its pretty fun.