Lapping your IHS to increase cooling? Hah, wrong!!!

Cooling Processors quietly

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
Ralf Hutter
SPCR Reviewer
Posts: 8636
Joined: Sat Nov 23, 2002 6:33 am
Location: Sunny SoCal

Lapping your IHS to increase cooling? Hah, wrong!!!

Post by Ralf Hutter » Mon Apr 30, 2007 5:24 am


jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Mon Apr 30, 2007 5:39 am

Lapping your IHS to increase cooling? Hah, wrong!!!
um, the press release says nothing about lapping, either positive or negative; so what's going on w/ the title of this thread? :?

jhhoffma
Posts: 2131
Joined: Mon Apr 25, 2005 10:00 am
Location: Grand Rapids, MI

Post by jhhoffma » Mon Apr 30, 2007 6:03 am

I think Ralf is simply saying that lapping is not the answer to better Core-to-H/S heat transfer. Using superior materials and applying them properly is the key to get the best cooling from your hardware.

Operandi
Posts: 243
Joined: Mon Nov 08, 2004 5:20 pm
Location: Wausau, WI
Contact:

Post by Operandi » Mon Apr 30, 2007 7:28 am

If I read it correctly this is about the interface between the CPU core and heat spreader, not the heat spreader to the heatsink so this has nothing to do with lapping at all.

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Mon Apr 30, 2007 8:58 am

Operandi wrote:If I read it correctly this is about the interface between the CPU core and heat spreader, not the heat spreader to the heatsink so this has nothing to do with lapping at all.
precisely.

Rusty075
SPCR Reviewer
Posts: 4000
Joined: Sun Aug 11, 2002 3:26 pm
Location: Phoenix, AZ
Contact:

Post by Rusty075 » Mon Apr 30, 2007 10:27 am

I think you've missed the point...

Yes, the paper is about the CPU/IHS interface, but its concept, that making the interface flatter actually reduces the efficiency of the thermal connection would logically apply to the IHS/HSF interface as well. If the "hierarchical branching microchannels" double the efficiency of one of the two interfaces, it makes sense that it would do the same for the other. That's what Ralf was getting at.

I wonder how long until we start to see this in the aftermarket heatsink market.

charonme
Posts: 131
Joined: Tue Apr 11, 2006 4:17 am
Location: Slovakia

mambo peRhaps, jumbo peRhaps not!

Post by charonme » Fri May 25, 2007 8:51 am

It's possible that something like that can increase the effectiveness of cpu cooling but nevertheless, lapping has been reported to put down 5+ degrees with some heatsinks.

try this

Bluefront
*Lifetime Patron*
Posts: 5316
Joined: Sat Jan 18, 2003 2:19 pm
Location: St Louis (county) Missouri USA

Post by Bluefront » Fri May 25, 2007 3:34 pm

IMHO.....Lapping to get a base mirror-finished is obviously a waste of time, but lapping can turn a slightly concave base into a flat surface that will transfer heat much better. I'm still lapping if it measures anything other than 100% flat. :lol:

Das_Saunamies
*Lifetime Patron*
Posts: 2000
Joined: Tue May 15, 2007 1:39 am
Location: Finland

Post by Das_Saunamies » Sat May 26, 2007 7:12 am

I agree that lapping will definitely improve cooling. Results will depend on how curved or uneven the original surface is.

However, I also understand that this find suggests the main problem is not uneven external dissipation surfaces, but rather the internal construction; ergo no amount of lapping will match the efficiency improvement available through adjustment in the internal mechanics.

The topic just puts it a bit harshly, but the point is there. My loyalties are shifting to Intel, they're being a lot more innovative and are doing more to advance the field right now. Looking forward to seeing this applied for some real life test results.

Spanki
Posts: 68
Joined: Thu May 03, 2007 3:09 pm

Post by Spanki » Sat May 26, 2007 7:49 am

Das_Saunamies wrote:My loyalties are shifting to Intel...
You mean, IBM? :)

Das_Saunamies
*Lifetime Patron*
Posts: 2000
Joined: Tue May 15, 2007 1:39 am
Location: Finland

Post by Das_Saunamies » Sat May 26, 2007 7:57 am

D'oh! :?

I knew I should not have added that last, irrelevant comment, they always backfire! :lol:

Thanks for the correction though, in this case it was indeed IBM who made headway, not Intel.

murtoz
Posts: 122
Joined: Tue Mar 13, 2007 12:24 pm
Location: Wiltshire, UK

Post by murtoz » Mon May 28, 2007 12:04 pm

Seems like Noctua has found this out too... I got their NH-U12F and noticed grooves at the bottom of the heatsink. I checked their web site and it turns out they're actually there on purpose. Looks like the same principle as IBM's finding.

Mr Evil
Posts: 566
Joined: Fri Jan 20, 2006 10:12 am
Location: UK
Contact:

Post by Mr Evil » Mon May 28, 2007 12:53 pm

They aren't doing this because a ridged surface provides better contact in itself, but because it makes it easier to ensure that the paste is spread evenly. For people like us, who can spend the extra effort to seat a heatsink perfectly, a smooth finish is better.

Note how they talk about reducing the thickness of the paste, but we know that ideally there should be no thickness, with most of the contact area being metal to metal, with the paste only filling in the gaps. That's too difficult for mass production, so they just apply a dollop and push the heatsink down on top. The ridges help with that.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Mon May 28, 2007 1:46 pm

As Rusty05 already wrote, IBM's research was specifically on the heatspeader and its interface w/the core; obviously, their findings should apply to all heat transfer mating surfaces, such as between the heatspreader & HS base.

Note the 2nd last paragraph in IBM's press release:
Together with other industry-leading suppliers tools are developed to define the surface channels through the same copper stamping process currently used to fabricate high volume chip lids. This will define a full supply chain of low-cost parts to quickly integrate the new technique into products.
Looks like IBM is pretty sure AMD & Intel will buy in.

cpemma
Posts: 351
Joined: Sun May 11, 2003 2:31 pm
Location: S Yorks, OK
Contact:

Post by cpemma » Thu May 31, 2007 3:10 pm

Bluefront wrote:IMHO.....Lapping to get a base mirror-finished is obviously a waste of time, but lapping can turn a slightly concave base into a flat surface that will transfer heat much better. I'm still lapping if it measures anything other than 100% flat. :lol:
Back in 486 days the heat-sink was often just a length cut off from a continuous extrusion and the bottom could indeed be quite markedly concave or convex. Now, especially with copper inserts, they all seem to be finally machined to a decent standard. A ham-fisted lapper can easily make the original worse.

I remember lapping my Taisol 760 sink, which had a quite marked pattern of circular machining marks on the copper bottom - maybe Taisol pre-discovered this "new" channel system? :wink:

Post Reply